Full text of "Journal of the American Chemical Society"

See other formats

Google

This is a digital copy of a book that was preserved for generations on Hbrary shelves before it was carefully scanned by Google as part of a project

to make the world's books discoverable online.

It has survived long enough for the copyright to expire and the book to enter the public domain. A public domain book is one that was never subject

to copyright or whose legal copyright term has expired. Whether a book is in the public domain may vary country to country. Public domain books

are our gateways to the past, representing a wealth of history, culture and knowledge that's often difficult to discover.

Marks, notations and other maiginalia present in the original volume will appear in this file - a reminder of this book's long journey from the

publisher to a library and finally to you.

Usage guidelines

Google is proud to partner with libraries to digitize public domain materials and make them widely accessible. Public domain books belong to the
public and we are merely their custodians. Nevertheless, this work is expensive, so in order to keep providing this resource, we liave taken steps to
prevent abuse by commercial parties, including placing technical restrictions on automated querying.
We also ask that you:

+ Make non-commercial use of the files We designed Google Book Search for use by individuals, and we request that you use these files for
personal, non-commercial purposes.

+ Refrain fivm automated querying Do not send automated queries of any sort to Google's system: If you are conducting research on machine
translation, optical character recognition or other areas where access to a large amount of text is helpful, please contact us. We encourage the
use of public domain materials for these purposes and may be able to help.

+ Maintain attributionTht GoogXt "watermark" you see on each file is essential for informing people about this project and helping them find
additional materials through Google Book Search. Please do not remove it.

+ Keep it legal Whatever your use, remember that you are responsible for ensuring that what you are doing is legal. Do not assume that just
because we believe a book is in the public domain for users in the United States, that the work is also in the public domain for users in other
countries. Whether a book is still in copyright varies from country to country, and we can't offer guidance on whether any specific use of
any specific book is allowed. Please do not assume that a book's appearance in Google Book Search means it can be used in any manner
anywhere in the world. Copyright infringement liabili^ can be quite severe.

About Google Book Search

Google's mission is to organize the world's information and to make it universally accessible and useful. Google Book Search helps readers
discover the world's books while helping authors and publishers reach new audiences. You can search through the full text of this book on the web

at|http : //books . google . com/|

♦■ .<• #

Boston
Medical Library

8 THE Fenway

r-u

>,—.).. ^af. -

J 4

■

> \

■

1
■

THE JOURNAL

OF THE \ »\ °

American Chemical Society

VOLUME XXX

1908

COMMITTEE ON PAPERS AND PUBLICATIONS:

William A. Noyes, Editor, Arthur A. Noyes,

Edward Hart, J. D. Pbnnock,

W. F. HlLLEBRAND, H. N. STOKES,

J. H. Long, H. P. Talbot,

Wm. McMurtrie, H. W. Wiley.

BASTON, PA.:
BSCaBMBACH PRINTING COMPANY

1908

JUN15 1921

UbraT^^

By WituAM A. NoYES, Edward Hart, W. F.
Hii,i«BBRAND, J. H. Long, Wm. McMurtrib,
Arthur A. Noy«s, J. D. Pbnnock, H. N.
Stokbs, H. p. Tai^bot, H. W. Wii,«y.

CommiiUe on Papers and Publications of the
American Chemical Society,

Papers.

January.

William A. Noyes: The Choice of the Most Probable Value for an Atomic Weight:

The Atomic Weight of Hydrogen 4

Theodore W. Richards and J. Howard Mathews: The Relation between Com-
pressibility, Surface Tension and Other Properties of Material 8

William A. Noyes and H. C. P. Weber: The Atomic Weight of Chlorine 13

H. C. P. Weber: Preparation of Chlorplatinic Add by Electrolysis of Platinum

Black 29

Edward W. Washburn: The Theory and Practice of the lodometric Determina-
tion of Arsenious Acid 31

Gregory Paul Baxter and Francis Newton Brink: The Specific Gravities of the
Iodides of Sodium, Potassium, Rubidium, Caesium, Calcium, Strontium and

Barium 46

A. W. Browne and F. F. Shetterly: On the Oxidation of Hydrazine, II 53

R. W. Thatcher: On the Reaction between I4me and Sulphur 63

Arthur E. Hill: The Relative Solubility of the Silver Halides and Silver Sul-

phocyanate 68

Walter E. Mathewson: On the Analytical Estimation of Gliadin 74

F. J. Alway and R. M. Pinckney: The Effect of Nitrogen Peroxide upon Wheat

Flour 81

Oswald Schreiner and Howard S. Reed: The Power of Sodium Nitrate and Calcium
Carbonate to Decrease Toxicity in Conjunction with Plants Growing in Solu-
tion Cultures 85

C. A. Crampton and L. M. Tolman: A Study of the Changes Taking Place in

Whiskey Stored in Wood 98

Alvin S. Wheeler: The Condensation of Chloral with Primary Aromatic Amines 136

Notes: The Boiling-point of Isobutane; The Stereochemistry of Indigo 142

Review: Researches on the Density of Gases 143

New Books: Annual Report on the Progress of Chemistry for 1906; Organic Chem-
istry, Including Certain Portions of Physical Chemistry, for Medical, Phar-
maceutical and Biological Students; A Text-Book of Organic Chemistry;

Poisons, Their Effects and Detection 156

Recent Publications 161

February,

Marston Taylor Bogert: American Chemical Societies 163

C. James: The Bromates o£ the Rare Earths 182

Gregory Paul Baxter and John Hunt Wilson: A Revision of the Atomic Weight

of Lead 187

E. B. Spear: Catalytic Decomposition of Hydrogen Peroxide under High Pres-

sures of Oxygen 195

F. C. Mathers: A Method for the Separation of Iron from Indium 209

F. C. Mathers and C. G. Schluederberg: Some New Compounds of Indium. ... 211

George Steiger : A New Form of Colorimeter 215

George Steiger: The Estimation of Small Amounts of Fluorine 219

Wm. Herbert Keen: Volumetric Method for the Determination of Zinc 225

L. M. Dennis and Ellen S. McCarthy: The Determination of Benzene in Illumi-
nating Gas 233

IV CONTENTS OF VOLUME XXX.

Wm. L. Dudley: The Effect of Coal Gas on the Corrosion of Wrought Iron

Pipe, Buried in the Earth 247

Frederick B. Power and Arthur H. Sal way: Chemical Examination of Micromeria

Chamissonis 251

H. M. Gordin: Marrubiin 265

Oswin W. Willcox: Decomposition Curves of Some Nitrocelluloses of American

Manufacture 271

E. B. Hart: Variations in the Amount of Caesin in Cow's Milk 281

Notes: The Use of the Centrifuge; Apparatus for the Centrifugal Drainage of

Small Quantities of Crystals 285

Review: Recent Progress in Physical Chemistry 288

Correction 303

New Books: A Text-Book of Electro-Chemistry; J. G. Gentle's Lehrbuch der

Farbenfabrikation; Der Nahrungsmittelchemiker als Sachverstslndiger;

Chapters on Paper Making; The Principles of Copper Smelting 303

Recent Publications 307

March.

F. W. Clarke: Fifteenth Annual Report of the Committee on Atomic Weights.

Determinations Published during 1907 309

Arthur A. Noyes and Yogoro Kato: The Equivalent Conductance of Hydrogen-
ion Derived from Transference Experiments with Nitric Add 318

Arthur A. Noyes: The Conductivity and Ionization of Salts, Acids, and Bases

in Aqueous Solutions at High Temperatures 335

Launcdot W. Andrews: The Rdractive Indices of Alcohol- Water Mixtures 353

J. Livingston R. Morgan and Reston Stevenson: The Wdght of a Falling Drop
and the Laws of Tate. The Determination of the Molecular Weights and

Critical Temperatures of Liquids by the Aid of Drop Wdghts 360

L. W. McCay: The Action of Hydrogen Sulphide on Alkaline Solutions of Zinc

Salts 376

George M. Howard: The Determination of Antimony and Arsenic in Lead- An-
timony Alloys 378

James R. Withrow: The Influence of Temperature on the Electrolytic Precipita-
tion of Copper from Nitric Add 381

Victor Lenher and A. W. Homberger: The Gravimetric Determination of Tellu-
rium 387

Sherman Leavitt and J. A. LeClerc: Loss of Phosphoric Add in Ashing of

Cereals 391

F. J. Moore and R. D. Gale: The Colored Salts of Schiff's Bases 394

William Lloyd Evans and Benjamin T. Brooks: On the Oxidation of Meta-

Nitrobenzoyl Carbinol 404

J. E. Teeple: Long Leaf PineOil 412

W. D. Richardson: Transparent Soap — A Supercooled Solution 414

Notes: Rapid Determination of Petroleum Naphtha in Turpentine; Determina-
tion of Sodium and Potassium in Silicates; The Determination of Total

Nitrogen Induding Nitrates in the presence of Chlorides 420

Reviews: Review of Analytical Work Done in 1906, 422; Inter-rdations of the

Elements 467

Correction 473

New Books: An Elementary Study of Chemistry; Die Kathodenstrahlen; The
Microscopy of Technical Products; Introduction to the Theory and Practice of

CX)NTENTS OF VOLUMIS XXX. V

Qualitative Analysis by Solution; Electro- Analysis; Church's Laboratory

Guide; Dairy Laboratory Guide; Annuaire pour TAn 1908 474

Recent Publications 479

April.

Arthur A. Noyes, William C. Bray, and EUwood B. Spear: A System of Qualita-
tive Analysis for the Common Elements 481

Gregory Paul Baxter and Herbert Wilkens Daudt: The Carrying Down of Soluble

Oxalates by Oxalates of the Rare Earths 563

Victor Lenher: Yttrium Earths 572

Gregory Paul Baxter: Modified Spectroscopic Apparatus 577

William M. Dehn: Simple Demonstrations of the Gas Laws 578

C. C. Tutwiler: An Improved Hygrometer for Determining the Minimum Tempera-
ture of Gas in Distribution Mains 582

A. H. Low: Technical Method for the Determination of Lead in Ores, Etc 587

F. J. Metzger and H. T. Beans: The Electrolytic Determination of Bismuth 589

Richard B. Moore and Ivy Miller: A Separation of Iron from Manganese 593

S. Avery and G. R. McDole: The Action of Sodium Benzyl Cyanide with Cinnamic

Ester 595

S. Avery and Fred W. Upson: The Synthesis of Certain Aromatic Succinic Acids 600

Harry Snyder: Influence of Fertilizers upon the Composition of Wheat 604

Charles D. Howard! The Precipitation Method for the Estimation of Oils in Fla-
voring Extracts and Pharmaceutical Preparations 608

Eugene E. Dunlap: A Comparison of Two Tests of Red Lead 611

Notes: Notes on the Separation of Silica and Alumina in Iron Ores; An Apparatus
for the Quantitative Electrolysis of Hydrochloric Acid; A Supposedly New

Compound from Wheat Oil ; Determination of Phosphorus in Ash Analysis .... 614
Reviews: Review of Inorganic Chemistry for 1907; On the Non-Equivalence of the

Four Valences of the Carbon Atom 618

New Books: Immunochemistry 650

Recent Publications 652

May.

Charles A. Kraus: Solution of Metals in Non-MetaUic Solvents: II. On the For-
mation of Compounds between Metals and Ammonia 653

Gilbert Newton Lewis: The Osmotic Pressure of Concentrated Solutions, and the

Laws of the Perfect Solution 668

Daniel F. Comstock: The Indestructibility of Matter and the Absence of Exact

Relations among the Atomic Weights 683

Herbert N. McCoy: Two New Methods for the Determination of the Secondary

Ionization Constants of Dibasic Adds 688

H. E. Chandler: The Ionization Constants of the Second Hydrogen Ion of Dibasic

Adds 694

M. deKay Thompson and M. W. Sage: On the Free Energy of Nickel Chloride. . . 714

Frederick H. Getman: A Study of the Solutions of Some Salts Exhibiting Negative

Viscosity 721

Victor Lenher: The Action of Various Anhydrous Chlorides on Tellurium and on

Tellurium Dioxide 737

Victor Lenher: The Homogendty of Tellurium 741

E. H. Archibald, W. G. Wilcox and B. G. Buckley: A Study of the Solubility of

Potassium Chloroplatinate 747

G. S. Jamieson, L. H. Levy and H. L. Wells: On a Volumetric Method for Copper . . 760

VI CONTENTS O^ VOLUME XXX.

S. W. Parr: Sodium Peroxide in Certain Quantitative Processes 764

D. F. Calhane: The Comparative Oxidizing Power of Sodium Peroxide and Its Use

in Qualitative Analysis 770

C. M. Johnson: The Determination of Carbon in Steel, Ferro- Alloys, and Plum-
bago by Means of an Electric Combustion Furnace 773

Allerton S. Cushman and Prevost Hubbard: The Extraction of Potash from Feld-

spathic Rock 779

W. h. Dubois: Flask for Fat Determination 797

William A. Johnson: A Proposed Method for the Routine Valuation of Diastase

Preparations 798

C. A. Mooers and H. H. Hampton: The Separation of Clay in the Estimation of

Humus 805

Marston Taylor Bogert and William Klaber: Researches on Quinazolones (Twen-
tieth Paper) on Certain 7-Nitro-2-Methyl-4-Quinazolones from 4-Nitroacetan-

thranil 805

William McPherson and Wilbur L. Dubois: On the Action of a-Benzoylphenylhy-

drazine on the Halogen Derivatives of Quinones 816

J. Bishop Tingle and H. F. Roelker: Studies in Nitration. II. — Melting Point
Curves of Binary Mixtures of Ortho- Meta- and Paranitranilines: a New

Method for Determining the Composition of Such Mixtures 822

W. J. Karslake and W. J. Morgan: Some Derivatives of i,3-Dimethyl-2,6-Dini-

trobenzene-4-Sulphonic Add 828

R. E. Lyons and G. C. Bush: Concerning a-Dinaphthyl Selenide and Telluride.... 831
Henry A. Torrey and H. B. Kipper: Hydrazones of Aromatic Hydroxyketones.

Alkali-insoluble Phenols 836

Henry A. Torrey and C. M. Brewster: The Action of Phthalic Anhydride on Resa-

cetophenone 862

Chas. H. Herty: The Optical Rotation of Spirits of Turpentine 863

Parker C. Mcllhiney : A Method of Analyzing Shellac 867

Chas. H. Herty and W. S. Dickson: The Volatile Oil of Pinus Serotina 872

Augustus H. Gill : On the Oxidation of Olive Oil 874

Fred W. Morse: The Effect of Temperature on the Respiration of Apples 876

J. H. Long: Observations on the Stability of Lecithin 881

J. H. Long and Frank Gephart: On the Behavior of Emulsions of Lecithin with

Metallic Salts and Certain Non- Electrolytes 895

J. T. Willard: On the Occurrence of Copper in Oysters 902

Notes: Notes on Mr. Keen's Paper on the Volumetric Determination of Zinc; The
Detection and Identification of Manganese and Chromium in the Presence of

Each Other 904

New Books: Organic Chemistry for Advanced Students; Kurzes Lehrbuch der
Organischen Chemie; Exercises in Elementary Quantitative Analysis for Stu-
dents of Agriculture; Testing Milk and Its Products; The Chemistry of Com-
merce; Modem Pigments and their Vehicles Technologic der Fette und Oele,
Bd. II, Gewinnung der Fette und Oele, Spezieller Teil; Traits Complet D'-

Analyse Chimique Appliqu^ Aux Essais Industriels 906

Recent Publications 913

June.

R. E. Swain and D. W. Harkins: Arsenic in Vegetation Exposed to Smelter

Smoke 915

W. D. Harkins and R. E. Swain: The Chronic Arsenical Poisoning of Herbivorous

Animals 928

I
I

I CONTENTS OF VOI^UME XXX. VU

Julius Stieglitz: Note on the Solubility Product 946.

Lawrence J. Henderson: A Diagrammatic Representation of Equilibria between

Adds and Bases in Solution 954

C. S. Hudson and F. C. Brown: The Heats of Solution of the Three Forms of Milk-

Sugar 960

Grajit T. Davis: A New Instrument for Reducing Gas Volumes to Standard Con-
ditions 971

Wm. L. Dudley: A Lecture Table Down-Draft 973

Lewis A. Youtz: Purity and Volatility of Precipitated Antimony Sulphide 975

C. James: A Scheme for the Separation of the Rare Earths 979

Charles F. Mabery and J. Howard Mathews: On Viscosity and Lubrication 992

F. J. Moore and R. G. Woodbridge, Jr. The Colored Salts of Schiff's Bases. II.

The Hydrochlorides of Bases Formed by Condensing /»-Aminodiphenylamine

with Aromatic Aldehydes looi

A. B. Vinson: The Endo- and Ektoinvertase of the Date 1005

Frank T. Shutt and A. T. Charron: Note on the Dyer Method for the Determina-
tion of Plant Food in Soils 1002

George Borrowman: Some Observations on the Assay of Telluride Ores 1023

S. W. Parr and W. F. Wheeler: The Deterioration of Coal 1027

George O. Adams and Alfred W. Kimball: Studies on Direct Nesslerization of

Kjeldahl Digestates in Sewage Analysis 1034

H. W. Clark and George O. Adams: Studies of Incubation Tests 1037

Note: The Quantitative Determination of Arsenic by the Gutzeit Method 104 1

New Books: Lehrbuch der Gerichtlichen Chemie; Benedikt-Ulzer, Analyse der
Fette und Wachsarten; Detection of the Common Food Adulterants; Medico-
Physical Works; Descriptive Biochemie mit besonderer Beriicksichtigung der
chemischen Arbeitsmethoden; Studies in Plant and Organic Chemistry and Lit-
erary Papers; Life and Scientific Activity of N. A. Menshutkin; Neue Capillar-
und Capillaranalytische Untersuchungen; A Course of Practical Organic
Chemistry; Practical Methods for the Iron and Steel Works Chemist; A Lab-
oratory Outline for Determinations in Quantitative Chemical Analysis;
Analysis of Mixed Paints, Color Pigments, and Varnishes; Commercial Organic

Analysis; Chemical Reagents 1042

Recent Publications 1053

July.

J. Livingston R. Morgan and Eric Higgins: The Weight of a Falling Drop and the
Laws of Tate. The Determination of the Molecular Weights and Critical Tem-
peratures of Liquids by the Aid of Drop Weights. II 1055

Eric Higgins: Some New Formulae Correlating the Various Constants for Non-
Associated Liquids 1069

Herbert N. McCoy: The Relation between the Ionizing Power and the Dielectric

Constants of Solvents 1074

Frederick H. Getman: The Viscosity of Non- Aqueous Solutions of Potassium

Iodide 1077

Chas. H. Herty and R. O. E. Davis: The Character of the Compound Formed by

the Addition of Ammonia to Ethyl-Phospho-Platino Chloride 1084

W- P. Bradley and C. F. Hale: Pure Carbon Dioxide 1090

D. Mcintosh: The Basic Properties of Oxygen 1097

Louis Kahlenberg and Francis C. Krauskopf : A New Method of Separating Lith-
ium Chloride from the Chlorides of the Other Alkalis, and from the Chloride of
Barium 1 104

Vni CONTENTS Ot VOlrUME XXX.

Edward DeMille Campbell and Walter Arthur: Determination of Nickel and

Chromium in Steel 1116

W. F. Hillebrand: The Influence of Fine Grinding on the Water and Ferrous-Jron

Content of Minerals and Rocks ^ 1 120

Philip Adolph Kober: A New Apparatus for the Quantitative Distillation of Am-
monia 1 131

Marston Taylor Bogert and Roemer Rex Renshaw: 4-Amino-o-Phthalic Acid and

Some of its Derivatives 1 135

Latham Clarke: Methyl Ethyl IsobutyJ Methane 1 144

Henry L. Wheeler and Leonard M. Liddle: Researches on P)rrimidines: S3mthesis

of Uracil-3-Acetic Add 1 152

Henry L. Wheeler and Leonard M. Liddle: Researches on Pyrimidines: Synthesis

of Uracil-4- Acetic Acid 1 156

C. S. Hudson: The Inversion of Cane Sugar by Invertase 1 160

Lucius L. Van Slyke: Conditions Affecting the Proportions of Fat and Proteins in

Cow's Milk 1166

J. H. Norton: Quantity and Composition of Drainage Water and a Comparison of

Temperature, Evaporation and Rainfall 1 186

W. D. Richardson and F. O. Farey : Lard from Oily Hogs 1 191

Notes: Preparation of a Solution for Making Standard Solutions of Sodium Hy-
droxide; The Action of Hydrochloric Acid on Manganese Dioxide 1192

New Books: Thermochemistry; Jahrbuch des Vereins der Spiritus-Fabrikantes in
Deutschland, des Vereins der Starke-Interessenten in Deutschland und des
Vereins Deutschen Kartoffeltrockner; Engine Room Chemistry; Decoration
of Metal, Wood, Glass, etc 1 193

Recent Publications 1 196

August.

Charles A. Kraus: Solutions of Metals in Non-Metallic Solvents; III. The Ap-
parent Molecular Weight of Sodium Dissolved in Liquid Ammonia 1197

O. F. Tower: The Determination of Vapor Pressures of Solutions with the Morley

Gauge 12 19

Andrew A. Blair: The Determination of Vanadium, Molybdenum, Chromium and

Nickel in Steel 1229

Edward DeMille Campbell and Edwin LeGrand Woodhams: A New Method for the

Determination of Vanadium in Iron and Steel 1 233

Helen Isham and Joseph Aumer: Direct Combustion of Steel for Carbon and
Sulphur 1236

Oscar S. Watkins: A Home-Made Drying Oven 1240

Henry A. Torrey and J. E. Zanetti: On Ethyl Pyromucylacetate. (Second Paper).

3-Furyl-5-Pyrazolone 1241

John E. Buchner: The Constitution of i -Phenyl- 2, 3-Naphthalene-Dicarboxylic

Acid 1244

Richard Sydney Curtiss and Paul T. Tarnowski: Methyl Mesoxalate and Some of

its Reactions 1264

William L. Dudley: Notes on the Roese Method for the Determination of Fusel

Oil, and a Comparison of Results by the AUen-Marquardt Method 1271

Wm. Antoni: The Estimation of Alcohol in Fermented Liquids 1276

Philip Adolph Kober: Ammonia Distillation in the Presence of Magnesium or Cal-
cium Salts 1279

Theodore W. Richards and J. Howard Mathews: Concerning the Use of Electrical

Heating in Fractional Distillation 1282

CONTENTS Olf VOI^UME XXX. IX

H. W. Cowles, Jr.: The Determination of Malic Acid in Food Products 1285

W. G. Whitman and H. C. Sherman: The Effect of Pasteurization upon the De-
velopment of Ammonia in Milk 1 288

Oswald Schreiner and Edmund C. Shorey: The Isolation of Picoline Carboxylic

Add from Soils and Its Relation to Soil Fertility 1295

R. E. Lyons and C. C. Carpenter: A Chemical Examination and Calorimetric

Test of Indiana Peats 1307

J. H. Long and Frank Gephart: On the Behavior of Lecithin with Bile Salts, and

the Occurrence of Lecithin in Bile » 1312

New Books: Roscoe and Schorlemmer*s Treatise on Chemistry; Thermodynamics

of Technical Gas Reactions; Stereochemistry; Book of Chemical Labels 1319

September.

Charles A. Elraus: Solutions of Metals in Non-Metallic Solvents; IV. Material
Effects Accompanying the Passage of an Electrical Current through Solutions
of Metals in Liquid Ammonia. Migration Experiments 1323

H. W. Poote and E. K. Smith: On the Dissociation Pressures of Certain Oxides of

Copper, Cobalt, Nickel and Antimony 1344

P. T. Walden: On the Dissociation Pressures of Ferric Oxides 1350

Gflbert N. Lewis: The Determination of Ionic Hydration from Electromotive

Force 1355

John Johnston: The Free Energy Changes Attending the Formation of Certain

Carbonates and Hydroxides 1357

E. E. Free: The Solubility of Precipitated Basic Copper Carbonate in Solutions of

Carbon Dioxide 1366

Frank Curry Mathers: The Electrolytic Formation of Selenic Add from Lead Sele-

nate 1374

Owen L. Shinn: The Electrolytic Determination of Nitric Add 1378

Svante Arrhenius: On Agglutination and Coagulation 1382

H. W. Foote: On the Nature of Predpitated Colloids 1388

J. Bishop Tingle and F. C. Blanck : Studies in Nitration, III. Nitration of Aniline

and of Certain of Its iV-Alkyl, N- Aryl and iV-Acyl Derivatives 1395

J. R. Bailey: Hydantoin Tetrazones 141 2

^^^lliam M. Dehn and Silas F. Scott: Some Characteristic Color Reactions Pro-
duced by Sodium Hypobromite 1418

S. Avery and G. R. McDole: The Oxidation and the Reduction of ^,pDiphenyl-

y-cyanbutyric Add 1423

S. Avery and Fred W. Upson: The Nitration of /9-/>-Tolylglutaric Add 1425

Albert P. Sy: Three New Prdiminary Tests for Maple Products 1429

H. Aug. Hunicke: Malt Analysis;^ Determination of Extract, II 143 1

A. Hugh Bryan: The Estimation of Dry Substance by the Refractometer in

Liquid Saccharine Food Products 1443

B. C. Kendall and H. C. Sherman: The Detection and Identification of Certain

Redudng Sugars by Condensation with /»-Brombenzylhydrazide 1451

F. Zerban and W. P. Naquin: On the Determination of Redudng Sugars 1456

A. Lowenstdn and W. P. Dunne: Determination of Sugar in Meats 1461

H. D. Gibbs: Methylsalicylate. The Analytical Separation and Determination

of Salicylic Add and Methylsalicylate, and the Hydrolysis of the Ester. . . 1465
W. A. Puckner and W. S. Hilpert: The Detection and Estimation of Hexamethyl-

enamine in Pharmaceutical Mixtures 1471

B. M. Chace: The Detection of Small Quantities of Turpentine in Lemon Oil. . 1475

X CONTENTS OF VOI^UMB XXX.

Elton Fulmer and Theo. C. Manchester: The Effect of Heat upon the Physical and

Chemical Constants of Cottonseed Oil 1477

R. W. Comelison: A Method for Detecting Synthetic Color in Butter 1478

R. E. Doolittle and A. W. Ogden: Composition of Known Samples of Paprika. . 1481
Horace C. Porter and F. K. Ovitz: The Nature of the Volatile Matter of Coal

as Evolved under Different Conditions i486

Notes: A Characteristic Test for Hippuric Acid; An Automatic Siphon Pipette;

Method for Determining Unsaponifiable Matter in Oils and Fats 1507

New Books: The Utilization of Wood Waste by Distillation; Handbook of Ameri-
can Gas Engineering Practice; Typhoid Fever — Its Causation, Transmission
and Prevention; Kolloides Silber und die Photohaloide; Sewage and Sewage
Bacteriology; Elements of Water Bacteriology; Les Nouveau Livres Sden-
tifiques et Industriels 151 1

October.

W. D. Richardson and Erwin Scherubel: The Deterioration and Commercial
Preservation of Flesh Foods. First Paper: General Introduction and Ex-
periments on Frozen Beef 1515

C. S. Hudson: The Inversion of Cane Sugar by Invertase, II 1564

Horace G. Byers: Behavior of Calcium and Sodium Amalgams as Electrodes

in Solutions of Neutral Salts 1584

J. Bishop Tingle and F. C. Blanck: Studies in Nitration, IV. Nitration of JV-Acyl
Compounds of Aniline Derived from Certain Polybasic, Aliphatic and Aro-
matic Adds 1587

Oswald Schreiner and Edmund C. Shorey: The Isolation of Dihydroxystearic

Acid from Soils 1599

A. G. Woodman and E. F. L3rford: The Colorimetric Estimation of Benzalde-

hyde in Almond Extracts 1607

Albert P. Sy : The Lead Value of Maple Products 161 1

A. M. Doyle: Typewriter Carbon Papers 1616

Edward Gudeman: Gluten Feeds — Artificially Colored 1623

H. C. Sherman and A. H. Kropff: The Calorific Power of Petroleum Oils and the

Relation of Density of Calorific Power 1626

Frank O. Taylor: Sodium Chloride, C. P .* 1631

New Books: The Chemical Basis of Pharmacology. An Introduction to Phar-
macodynamics Based on the Study of the Carbon Compounds; Determi-
nation of Radicles in Chemical Compounds 1634

November.

Clarence W. Balke and Edgar F. Smith: Observations on Columbium 1637

William M Barr: A Study of the Spectrum and the Bromides of Columbium ... . 1668

Joel H. Hildebrand: The Arc Spectrum of Columbium 1672

Edgar T. Wherry and Wni. H. Chapin: Occurrence of Boric Acid in Vesuvianite. 1684
Edgar T. Wherry and W. H. Chapin: Determination of Boric Acid in Insoluble

Silicates 1687

George I. Kemmerer: The Atomic Weight of Palladium 1701

Jacob S. Goldbaum and Edgar F. Smith: The Separation of Alkali Metals in the

Electrolytic Way 1705

Edward H. Keiser and Sherman Leavitt: On the Preparation and the Composition

of the Acid Carbonates of Calcium and Barium 171 1

Edward H. Keiser and LeRoy McMaster: On the Composition of the Acid Car-
bonates of Calcium and Barium 17 14

CONTENTS OF VOLUME XXX. XI

Horace G. Byers: The Passive State of Metals 1718

Irving Langmuir: The Velocity of Reactions in Gases Moving through Heated

Vessels and the Effect of Convection and DifTusion 1742

S. F. Acree: On Catalytic Reactions Induced by Enzymes 1755

Stanley R. Benedict and Frank Gephart: The Estimation of Urea in Urine. . . . 1760
J. Bishop Tingle and H. F. Rolker: Studies in Nitration, V. Melting Points

of Mixtures of Ortho- and Paranitranilines 1764

C. S. Hudson: Further Studies on the Forms of Milk-Sugar 1767

Wm. L. Dudley: The Filtration of Alcoholic Liquids through Wood Charcoal. . 1784
Notes: The Determination of Antimony and Arsenic in Lead-antimony Alloys;

Apparatus for Polarizing at 87°; Arrangement for Preventing Frothing

in Crude Fiber Determinations 1789

New Books: The Metallurgy of Iron and Steel; Inorganic Chemistry; Die Chem-
iscfae Affinitat und Ihre Messung; Laboratory Manual of Qualitative Analy-
sis; Outlines of Qualitative Chemical Analysis; Qualitative Analysis, vom
Stand punkte der lonenlehre; The Chemistry of Diazo-Compounds; An In-
troductory Course of Quantitative Chemical Analysis with Explanatory Notes
and Stoicliiometrical Problems; Electroanalytische Schnellmethoden; The
Data of Geochemistry; Report of the Eleventh Annual Convention of the
Association of State and National Food and Dairy Departments 1793

Recent Publications 1801

December.

G. A. Hulett and Ralph E. DeLury: The Reduction of Cadmium by Mercury

and the Electromotive Force of Cadmium Amalgams 1805

L. H. Duschak: The Mixed Barium-Strontium Chromate Precipitate 1827

D. M. Lichty: Absolute Sulphuric Acid: Its Preparation from Sulphur Trioxide

and Water; Its Specific Electric Conductivity and that of More Dilute Acid . 1834
Lloyd C. Daniels: Derivatives of Complex Inorganic Acids; Aluminico Tung-
states and Aluminico-Phosphotungstates 1846

William Blum: Derivatives of Complex Inorganic Acids; Phosphovanado-

Molybdates 1858

John A. Schaeffer: Double Fluorides of Titanium 1862

Mary E. Holmes: The Use of the Rotating Anode in Electrolytic Separations. . 1865
J. Bishop Tingle and Ernest E. Gorsline: Investigation of the Claisen Con-
densation. III. Further Contributions towards the Elucidation of the

Mechanism of the Reaction 1874

J. Bishop Tingle and H. F. Rolker: Intramolecular Rearrangement of Phthal-

amidic Adds, III 1882

M. A. Rosanoff and W. L. Prager: Studies in Esterification, I. Victor Meyer's

Esterification Law 1895

W. L. Prager: Studies in Esterification, II 1908

Note: On the "Color Demonstration of the Dissociating Action of Water" of

Jones and Allen 1914

New Books: Text-book of Physiological Chemistry; A Text Book of Inorganic
Chemistry; Die Lagerung der Atome im Raume; A Text-book of Experi-
mental Chemistry (with Descriptive Notes). For Students of General
Inorganic Chemistry; House Painting, Glazing, Paper Hanging, and White

Washing 1916

Recent Publications 1920

:oi,J

Vou XXX. January, 1908. No. 1.

OF THE \

JUN15 1921 ^

American ^Chefsat^' Society

REPORT OF THE INTERNATIONAL COMMITTEE ON ATOMIC

WEIGHTS. 1908.

Received November 6, 1907.

Since the preparation of our report for 1907, several important deter-
minations of atomic weights have been published. They are, briefly,
as follows:

Nitrogen. — Richards and Forbes^ have redetermined the ratio between
Ag and NO3, as shown in the composition of silver nitrate. The ratio
found, with all corrections applied, is AgtNOj = 100:57.479. Hence,
if Ag = 107.930, N = 14.037; and if N = 14.008, Ag = 107.880. In
short, the higher atomic weight hitherto assigned to silver is inconsistent
with the lower value for nitrogen as found in several recent investigations.

Sulphur. — Richards and Jones^ have measured the ratio between
Ag2S04 and AgCl. From the data obtained, if Ag = 107.930, S = 32.113,
a value much higher than that commonly accepted. If Ag = 107.880,
then S = 32.069, which is near the figure given in our former tables.
Additional evidence relative to this constant is much to be desired, for it
influences the determination of many other atomic weights, especially
those of the rare-earth metals.

Potassium. — From the ratios Ag:KCl and AgCl:KCl, Richards and
Staehler* find K = 39.114, when Ag = 107.930 and CI = 35.473. From
the corresponding bromide ratios, with Br = 79.953, "Richards and
Mueller* find K = 39.1143 and 39.1135. The final result of both re-
searches is K = 39.114, a distinct lowering of the constant in question.

Manganese. — Atomic weight redetermined by Baxter and Hines,^ from

' This Journal, 29, 808; and Z. anorg. Chem., 55, 34.

* This Journal, 39, 826; and Z. anorg. Chem., 55, 72.
■ This Journal, 29, 623; and Ber., 39, 361 1.

* This Journal, 29, 639; and Z. anorg. Chem.. 53, 423.

* This Journal, 28, 1560; and Z. anorg. Chem., 51, 202.

2 COMMITTBE ON ATOMIC WEIGHTS.

analyses of the chloride and bromide. The mean of their very concordant
determinations is Mn = 54.957, when Ag = 107.930.

Cobalt, — From new analyses of the chloride, Baxter and CoflGm^ find
Co = 58.997, or 59 in round numbers. This confirms the earlier deter-
minations by Richards and Baxter.

Indium. — Mathers,* from analyses of indium chloride, found In =
114.88. From the bromide. In = 114.86. The author recommends the
adoption of the rounded-oflf value, 114.9, when Ag = 107.93, CI = 35.473,

and Br = 79-953-

TeUurium. — Norris," by twelve concordant reductions of the basic
nitrate, 2Te02.HN03 to TeO,, found the atomic weight of tellurium to
be 127.48, when N = 14.01. With N = 14.04, Te = 127.64, which is
in better agreement with other recent determinations. The true cause
of the difference is not clear.

Neodymium. — Holmberg* has redetermined the atomic weight of this
element by synthesis of the sulphate from the oxide. In mean, when
S = 32.06, Nd = 144.08. This is higher by 0.48 than the value given
in our table.

Dysprosium, — In two series of determinations, based upon the ignition
to oxide of the octohydrated sulphate, Urbain and Demenitroux* find
for the atomic weight of dysprosium, values ranging between 162.29 and
162.75. Ill mean, Dy = 162.53.

Radium, — Madame Curie,* in a series of three new determinations,
has found a more precise value for the atomic weight of radium. Work-
ing with material more abundant and pure than that formerly analyzed,
she found Ra = 226.18 when Ag = 107.8 and CI = 35.4. With Ag =
107.93 8.nd CI « 35.45, Ra = 226.45. This number is higher than her
earlier determination by more than a unit.

From the data here given, and from those cited in previous reports,
it is evident that the entire table of atomic weights is in need of revision.
The values assigned to K and Na are too high; those given to CI and S
are too low; and these constants affect the determinations of many others.
They depend, however, upon the atomic weight of silver, which is prob-
ably, but not certainly, as low as 107.88. It is well known that work
upon these fundamental constants is now nearing completion in several
laboratories, notably under T. W. Richards, W. A. Noyes, and probably
other investigators also. Within a few months it should be possible to

^ This Journal, 28, 1580; and Z. anorg. Chem., 51, 171.
' This Journal, 39, 485; and Ber., 40, 1220.
■ This Journal, 28, 1675.

* Z. anorg. Chem., 53, 83.

* Compt. rend., 143, 598.

* Ibid., 145, 422. The atomic weight of radium is also under investigation by
Thorpe.

REPORT OP COMMITTBB ON ATOMIC WEIGHTS.

enter upon a satisfactory revision of the table, a task which would be
unsatisfactory, if tmdertaken now. It is true, as Brauner has suggested/
that the present table contains inconsistencies, but they are small in
amount, and are due to inconsistencies in the original data from which
the values are derived. In our next report we hope to recompute the
entire table; but meanwhile, awaiting the completion of the researches
which we know to be in progress, we prefer to leave the table practically

International Atomic Weights. 1908.

Aluminum Al 27 . i

Antimony Sb 120 . 2

Argon A 39. g

Arsenic As 75.0

Barimn Ba 137.4

Bismuth Bi 208.0

Boron B 11. o

Bromine Br 79 • 96

CarimiiiTn Cd 112. 4

Caesium Cs 132.9

Calcium Ca 40 . i

Carbon C 12.00

Cerium Ce 140.25

Chlorine CI 35-45

Chromium Cr 52 . i

Cobalt Co 59.0

Columbium Cb 94 .

Copper Cu 63.6

Dysprosium Dy 162 . 5

Erbium Er 166 .

Europium Eu 152 .

Fluorine F 19.0

Gadolinium Gd 156 .

Gallitun. Ga 70.0

Germanium Ge 72.5

Gludnum Gl 9.1

Gold Au 197 . 2

Helium He 4.0

Hydrogen H i .008

Indium In 115.

Iodine I 126.97

Iridium Ir 1930

Iron Fe 55.9

Krypton Kr 81.8

Lanthanum La 138.9

Lead Pb 206.9

Lithium Li 7 .03

Hagnesitun Mg 24 . 36

Manganese Mn 55.0

Mercury Hg 200.0

^ CheuL-Ztg., May 11, 1907.

Molybdenum Mo

Neodymium Nd

Neon Ne

Nickel Ni

Nitrogen N

Osmium Os

Oxygen O

Palladium Pd

Phosphorus P

Platinum Pt

Potassium K

Praseodymium Pr

Radium Ra

Rhodium Rh

Rubidium Rb

Ruthenium Ru

Samarium Sa

Scandium Sc

Selenium Se

Silicon Si

Silver Ag

Sodium Na

Strontium Sr

Sulphur S

Tantalum Ta

Tellurium Te

Terbium Tb

Thallium Tl

Thorium Th

Thulium Tm

Tin Sn

Titanium Ti

Tungsten W

Uranium U

Vanadium V

Xenon Xe

Ytterbium Yb

Yttrium Y

Zinc Zn

Zirconium Zr

96.0
143-6
20.

58.7
14.01

19X.0

16.00

106.5

310
194.8

39 15
140.5
225.
103.0

855
101.7

150.3
44.1

79.2

28.4

107 . 93
23 05
87.6
32.06
181.
127.6
159 -2
204.1

232.5
171.

119. o

48.1

184.0

238.5

51.2
128.

173.0

89.0

65.4

90.6

4 WILLIAM A. NOYBS.

unchanged. A conservative policy seems to be safer than one of haste,
and thje delay of another year will do no harm. One exception to the
rule may, however, be made. Dysprosium, with the atomic weight
162.5, n^ay now be properly added to the list of the chemical elements,
and we recommend its insertion in the table.

It is with the deepest regret that we record the loss, by death, in Febru-
ary last, of our distinguished colleague. Professor Moissan. The Chemical
Society of Paris has designated Monsieur G. Urbain as his successor ujDon
this Commission.

(Signed)

F. W. Clarke,

W. OSTWALD,

T. E. Thorpe,

G. Urbain.

THE CHOICE OF THE MOST PROBABLE VALUE FOR AN ATOMIC
WEIGHT: THE ATOMIC WEIGHT OF HYDROGEN,^

By William A. Noybs.
Received December 6, 1907.

A large amount of material has been accumulated from which the
atomic wieghts of the more important elements can be calculated. A
very superficial examination of this material reveals the fact that the
experimental results on which our knowledge of these constants is based,
vary ver>'^ greatly in their value and that many of the older determina-
tions have been rendered practically worthless by recent work, which
has been more careful and accurate.

As some of these new determinations affect the values for elements
of such fundamental importance that a recalculation of the whole table
of atomic weights will be necessary in the near future, it seems desirable
to formulate some general principles to aid in the elimination of results
which have little or no value and in the combination of the results which
remain. Such principles, if they meet with general acceptance, will
be of value, not only for the purpose stated but also as setting a certain
standard which must be attained by future workers in this field, if their
work is to be of permanent value.

The most important general principle which has been proposed for
the combination of the results of different observers, is the one based on
the mathematical discussion of accidental errors of observation. In ac-
cordance with the theory of probabilities, these results, if subject only
to accidental errors, should be weighted in inverse proportion to their

^ Presented in abstract at the N. Y. Meeting of the American Chemical Society,
Dec. 28, 1906.

THIS ATOMIC WEIGHT OF HYDROGEN. 5

probable errors.* A very serious objection to this method of treatment
lies in the fact that every determination of this kind is subject to con-
stant errors, and that the amount of these errors is not proportional to
their ** probable errors."^ Thus Stas obtained 132.8445 ±0.0008 parts
of silver chloride from 100 parts of silver, while Richards and Wells*
have obtained 132.8670 ±0.0005 parts. The most probable value
calculated by the mathematical rule would be 132.8607. If this value
is the true one, the real error of the value obtained by Richards and
Wells is 12 times its probable error, while the real error of Stas is 20 times
the probable error. And, whatever the true value may be, the real error
of one of the results, at least, is many times its ** probable error." An
examination of other cases shows that the relations here found are typi-
cal, and it seems evident that the question of constant errors requires
some other treatment than the simple mathematical one. The proper
treatment, which is an experimental one, has been clearly illustrated
in the case which we have under consideration. Richards and Wells
studied their method very carefully with especial reference to the elimina-
tion of constant errors and to secure evidence as to the amount of those
errors which could not be wholly excluded. They also pointed out
certain errors in the work of Stas and determined, approximately, the
magnitude of some of these. It is evident for this reason that very much
greater weight attaches to the value found by Richards and Wells than
to that found by Stas, and it is proposed as a general principle that when
a later observer has pointed out sources of error which are considerable
in comparison with the ** probable errors" and where the later observer
has succeeded in avoiding these sources of error, the earlier work must
be looked upon as having only confirmatory value and the result of the
later work should be accepted without modification. It has been ob-
jected to this that the later work is also subject to constant errors which
may be in the opposite direction from those of the earlier determina-
tion and that if we give a certain weight to the earlier work we may
eliminate these errors. But we certainly are not justified in using a value
that contains a known error in one direction merely for the chance that
we may compensate an unknown error.

» F. W. Qarke: "Constants of Nature," Part V., Edition of 1897, p, 7.

* Professor Clarke has, of course, recognized the importance of constant errors
and has often rejected values which he considers subject to such errors. In 1898
{Amer. Chem, /., 20, 543) Prof. T. W. Richards pointed out the importance of
selecting atomic weights on the basis of the methods employed in their determina-
tion and the probable freedom of those methods from constant errors. It is inter-
esting to notice that of the seven values in Professor Richards's table, which differed
sit that time decidedly from the values given by Professor Clarke, the numbers for four
of the atomic weights are nearer to the numbers now given in the International table
than were the values then given by Professor Clarke.

• This Journal, 27, 524.

4 WitUAM A. NOYEg.

unchanged. A conservative policy seems to be safer than one of haste,
and th,e delay of another year will do no harm. One exception to the
rule may, however, be made. Dysprosiinn, with the atomic weight
162.5, ^ay now be properly added to the list of the chemical elements,
and we recommend its insertion in the table.

It is with the deepest regret that we record the loss, by death, in Febru-
ary last, of our distinguished colleague, Professor Moissan. The Chemical
Society of Paris has designated Monsieur G. Urbain as his successor upon
this Commission.

(Signed)

F. W. C1.ARKE,
W. OSTWAI.D,
T. E. Thorpe,

G. Urbain.

THE CHOICE OF THE MOST PROBABLE VALUE FOR AN ATOMIC
WEIGHT: THE ATOMIC WEIGHT OF HYDROGEN.*

By William A. Noybs.
Received December 6, 1907.

A large amount of material has been accumulated from which the
atomic wieghts of the more important elements can be calculated. A
very superficial examination of this material reveals the fact that the
experimental results on which our knowledge of these constants is based,
vary very greatly in their value and that many of the older determina-
tions have been rendered practically worthless by recent work, which
has been more careful and accurate.

* Presented in abstract at the N. Y. Meeting of the American Chemical Society,
Dec. 28, 1906.

THE ATOMIC WEIGHT OF HYDROGEN. 5

probable errors.* A very serious objection to this method of treatment
lies in the fact that every determination of this kind is subject to con-
stant errors, and that the amount of these errors is not proportional to
their ** probable errors. "^ Thus Stas obtained 132.8445 ±0.0008 parts
of silver chloride from 100 parts of silver, while Richards and Wells*
have obtained 132.8670 ±0.0005 parts. The most probable value
calculated by the mathematical rule would be 132.8607. If this value
is the true one, the real error of the value obtained by Richards and
Wells is 12 times its probable error, while the real error of Stas is 20 times
the probable error. And, whatever the true value may be, the real error
of one of the results, at least, is many times its ** probable error." An
examination of other cases shows that the relations here found are typi-
• cal, and it seems evident that the question of constant errors requires
some other treatment than the simple mathematical one. The proper
treatment, which is an experimental one, has been clearly illustrated
in the case which we have under consideration. Richards and Wells
studied their method very carefully with especial reference to the elimina-
tion of constant errors and to secure evidence as to the amount of those
errors which could not be wholly excluded. They also pointed out
certain errors in the work of Stas and determined, approximately, the
magnitude of some of these. It is evident for this reason that very much
greater weight attaches to the value found by Richards and Wells than
to that found by Stas, and it is proposed as a general principle that when
a later observer has pointed out sources of error which are considerable
in comparison with the "probable errors" and where the later observer
has succeeded in avoiding these sources of error, the earlier work must
be looked upon as having only confirmatory value and the result of the
later work should be accepted without modification. It has been ob-
jected to this that the later work is also subject to constant errors which
may be in the opposite direction from those of the earlier determina-
tion and that if we give a certain weight to the earlier work we may
eliminate these errors. But we certainly are not justified in using a value
that contains a known error in one direction merely for the chance that
we may compensate an unknown error.

* F. W. Clarke: "Constants of Nature," Part V., Edition of 1897, p. 7.

■ Professor Clarke has, of course, recognized the importance of constant errors
and has often rejected values which he considers subject to such errors. In 1898
(Amer. Chem. J., 20, 543) Prof. T. W. Richards pointed out the importance of
selecting atomic weights on the basis of the methods employed in their determina-
tion and the probable freedom of those methods from constant errors. It is inter-
esting to notice that of the seven values in Professor Richards's table, which diflFered
sit that time decidedly from the values given by Professor Clarke, the numbers for four
of the atomic weights are nearer to the numbers now given in the International table
than were the values then given by Professor Clarke.

■ This Journal, 27, 524.

6 WILI^IAM A. NOYES.

The principle outlined above has been recently proposed, independ-
ently, by Professar Guye,* in his discussion of the selection of the most
probable value for the density of a gas. A second principle proposed by
Professor Guye is that when the values obtained by two observers agree
while that obtained by a third observer is discordant, the values which
are in agreement should be given much greater weight. As an exten-
sion of this principle, a value of an individual worker which differs ma-
terially from the values obtained by several others, should be rejected
entirely.

After eliminating the results which are excluded by the application
of the foregoing principles, it is proposed to arrange those which remain
in the order of their probable errors. Any result with a probable error
more than five times that of the smallest probable error may be excluded,
as such a result will have only one twenty-fifth the weight, according
to the theory of probabilities. In practical effect, this is the same as using
the mathematical rule which Professor Clarke has so long employed
in weighting the results of different workers. As at least five or six ob-
servations are necessary to give a probable error which has any sig-
nificance, results based on a smaller number of determinations may be
excluded unless other evidence warrants the belief that the work is of an
unusual degree of accuracy.

The values for any given ratio which remain after the elimination
of results which have little value, may well be combined by weighting
them inversely as the squares of their probable errors.

For further use, the ratios which are selected in this manner should be
weighted, not by. the probable error calculated by the mathematical
rule but by the deviation of the results of different observers from the
value selected. If the results of only one observer remain after elimina-
ting untrustworthy values (as in the case of the ratio of silver to silver
chloride), this result should be weighted in accordance with the average
deviation of the results of this observer from his mean. This will, I
think, give a much fairer basis than the "probable error" for weighting
the value in such cases. Thus the "mean error" of the value of Rich-
ards and Wells given above is 0.0018, while the " probable error" is 0.0005.
When we consider the certainty that some sources of constant error
will always remain, I think every one will agree that the real error is
much more likely to correspond to the former than to the latter value.

After selecting the most trustworthy experimental ratios as suggested,
we have still to combine them for the calculation of atomic weights.
This may usually be done in a variety of ways. In choosing among
these, the same general principles as before should be applied. For
a given atomic weight, only those ratios should be used for which the un-

' Arch. sd. phys. nat., 24, 44.

THE ATOMIC WEIGHT OF HYDROGEN. 7

certainties of the values will affect the atomic weight chosen less than
five times as much as any other combination of ratios which might be
used. In most cases this will probably lead to the selection of ratios
which furnish a direct comparison with oxygen, silver or one of the halo-
gens rather than of those in which the comparison is more indirect. Den-
sities of gases corrected to the condition of an ideal gas by the method
of D. Berthelot' may be considered as direct comparisons with oxygen,
and molecular and atomic weights calculated from these densities should
be included with those determined by chemical methods.

The Atomic Weight of Hydrogen.

The following is a summary of the determinations which have been made
of the atomic weight of hydrogen by the chemical method :

No. of Prob. Real Re»l error

Date, expts. Value. error, error. Prob. error'

Dtdong and Berzelius 1821 3 i .00667 35^ 108 0.3

I^educ 1892 2 1.00749 83 26 0.3

Hrdmann and Marchand 1842 8 i. 00160 71 615 8.7

Thomsen 1870 8 1.00570 71 205 2.9

Rayleigh 1889 5 i .00692 56 85 1.5

I^umas 1842 19 I 00250 44 525 12 .0

Kdser 1898 8 i. 00753 31 22 0.7

Dittmar and Henderson 1890 24 1.00840 29 65 2.2

Noyes (recalculated) 1890 24 1.00765 17 10 0.6

Thomsen 1895 21 1.00826 14 51 3.6

Cooke and Richards 1887 16 1.00826 13 51 3.9

Noyes (original) 1890 24 i .00654 ^^ i^^ ^^ -o

Keiser 1888 10 i .00306 7 469 67 .0

Noyes 1907 48 1.00787 2 12 6.0

Morley 1895 23 1.00762 2 12 6.0

The probable errors of the table are calculated from those assigned
by Professor Clarke.' For the results of Erdmann and Marchand and
Leduc, the values are arbitrary. For convenience these errors are given
in units corresponding to the last significant figure of the values for the
atomic weights.

On appljring the principles which have been outlined, we find that
the results of Dulong and Berzelius, Erdmann and Marchand, and of
Dumas, are excluded because the later work of Dittmar and Hender-
son by the same method, demonstrates that serious constant errors were
involved in the earlier work. Leduc's value is to be rejected because
the number of experiments was too small. Keiser' s earlier value is to
be rejected because it is not in accord with any of the later work and be-
cause he has himself given us a later and better value. My own original
value must be rejected because it was subject to a constant error and

* Compt. rend., 144, 76.

» "Constants of Nature." Part V.. p. 24 (1897).

8 THEODORE W. RICHARDS AND J. HOWARD MATHEWS.

the recalculated result may be considered as superseded by my later
and more careful work. Because the probable errors of all of the other
determinations are more than five times as great as those of Morley and
myself, they would be excluded by the third principle proposed. The
final value, if calculated from these two results, is 1.00775.

It is interesting to notice the relation between the real errors of the
various values (assuming this value as true) and the probable errors.
Only in those cases where we now know that there were serious constant er-
rors, is the real error more than six times th^ probable error,

Morley calculates a value corresponding to 1.00762 from his deter-
minations of the .densities of the gases and their combining volumes.
This value has not been considered here, partly because the probable
error of the density of hydrogen is about 3 in 100,000, instead of 2 for
the chemical method, but chiefly because of the imcertainty of the ratio
of the combining volumes. '

If a value is calculated by Professor Clarke's method, weighting each
result in inverse proportion to its probable error, only Keiser's older
value and my own original value would affect the value which I have se-
lected by more than about one part in 100,000. Reiser's older value would,
however, reduce it by about 40 parts and my own original value by about
4 parts in 100,000.

Univbrsity op Illinois.
Urbana, III.

THE RELATION BETWEEN COMPRESSIBILITY, SURFACE TENSION

AND OTHER PROPERTIES OF MATERIAL.

(preliminary PAPER.)

By Thbodorb W. Richards and J. Howard Mathews.
' Received October 30, 1907.

A recent paper by Albert RitzeP upon gas solubility, compressibility
and surface tension, seems to render important the brief publication of
some work carried on by us during the winter of 1905-06. This work
was presented to the Physico-chemical Club of Boston and Cambridge
on May 2, 1906, under the title *'The Relation of Compressibility to
Other Physical Properties, with Particular Reference to Surface Tension,"
and was discussed there. On account of the subsequent absence of one
of us in Germany the publication of this work has been delayed, the
amount of material being so large that a careful * study of the relations
demanded more time than was then available.

The present notice seems desirable because Ritzel has touched upon
one of the relations studied previously by us. He has used the method
of determining compressibility which we have used, and shows that this

* Morley: "Smithsonian Contribution to Knowledge," No. 980, p. no (1895).

* Z. physik. Chem., 60, 319 (1907).

THE RELATION BETWEEN COMPRESSIBILITY, ETC. 9

property of substance is a significant one. His paper is interesting and
valuable as far as it goes, and scarcety touches at all the immediate ground
which we covered, but in order to save the time of any one else who might
be thinking of continuing his work, it seems only fair that our results
should likewise be put into print.

As Ritzel has referred in his paper to the suggestions of van der Waals
and others concerning a possible relation between compressibility and
surface tension, it is not necessary for us to repeat this discussion. The
subject was studied by us from somewhat a different point of view, namely,
from the point of view of the theory of compressible atoms.* The logic
of our train of thought was this :

Premises, — (i) All bodies under high pressure have a smaller com-
pressibility than the same bodies under low pressure. (2) The physical
aflSnity which causes surface tension probably exerts pressure in its
action.

Conclusion, — ^Therefore, the greater this physical affinity of surface
tension, the less will probably be the compressibility under additional
outside pressure, at any rate in substances of similiar composition. This
probable relationship was suggested by Ostwald at a discussion con-
cerning atomic compressibility in the same Physico-chemical Club in the
autumn of 1905, and with his approval we immediately began the study
of it. The subject fitted very suitably into the schen-e upon which we
had already begtm work — a scheme which embraced many other physical
properties of substance.

In this preliminary notice it is not necessary to describe in detail the
methods which we used for determining the various quantities in question.
It is enough to say that all the substances were redistilled fractionally
until their purity was reasonably satisfactory, the substances of com-
merce having been shown to be altogether too impure to give significant
results. The method of Richards and StulP was used for determining
the compressibilities, and with some modifications the method of Ramsay
and Shields,* for determining the surface tensions; the boiling-points
were corrected for the projecting thread of the thermometer, and the
specific gravities were determined with great care by means of the Ostwald
pycnometer. All the data given below except the heats of vaporization
were made in the Harvard Laboratory, some of them as parts of another
investigation by one of us in collaboration with Dr. StuU, the rest by the
present authors. Each figure is suitably designated in order to show
who found it. The heats of vaporization are taken from the work of

* Proc. Amer. Acad. Arts and Sciences, 39, 581 (1904); also, Z. physik. Chem., 49,
15 (1904). This Journal, 26, 399.

* Z. physik. Chem., 49, i (1904); also, Pub. No. 7 Carnegie Institution of Washing-
ton.

* Z. physik. Chem., 12, 433 (1893).

THEODORB W. RICHARDS AND J. HOWARD MATHEWS.

others, as tabulated in the well-known tables of Landolt and Bomstein.
The compressibilities are expressed in terms of the kilograms per square
centimeter, which is 0.967 atmosphere; that is to say, a substance having
a compressibility of /? = 80 X lo"^ would be altered in voliune by the

Substance.

Methyl aniline

Dimethyl aniline

Ethyl aniline

Diethyl aniline

Toluidine (ortho)

Toluidine (meta)

Cresol (ortho)

Cresol (meta)

Cresol (para)

Benzyl alcohol

Ethyl acetate

Propyl formate

Ethyl benzene

Xylene (ortho)

Xylene (meta)

Xylene (para)

Isoamyl formate

Methyl iso valerianate.

Ethyl butyrate

Isobutyl acetate

Ethyl isobutyrate

Ethyl propionate

Methyl isobutyrate

Methyl butyrate

Valerianic acid

Methyl acetate

Ethyl acetate

Methyl alcohol

Ethyl alcohol

Propyl alcohol (norm).
Butyl alcohol (norm). .
Butyl alcohol (tert) . . .

Isoamyl alcohol

Ethyl bromide

Ethyl iodide

Ethylene chloride

Ethylene bromide

S2
|x

41.87
47.98
45 89

49.79

40.43
41.89

42.24

42.58

42.14

40.20

81.6

78.3
64.8

64.4

64.6

657
72.8

74.6

76.9

78.6

80.8

78.8

80.4

75-8

69.4

78.7
81. 6»

857

72.75
71.2

69.2

79.6

75.6

895

74.4
61.5

50.5

bo
e

195.7**

193.7
206.5

217.5

196.5
199.0

187.0

201.0

200.5

204.5

77*
8i»

136*
142'

139*
138'

I23»

ix6»
120*
116. 5'
no*

991

92*

IOi»

175

57
77*
66

97
117

131
38.0

72.2

837
131-7

c
Q

0.9865

0.9555
0.9625

0.9344
0.9986
0.9887
1.0482
I. 0341

1.0347
1.0463

0.8990

0.8982

0.8759

0.8633

0.8642

0.8612

0.8706

0.8808

0.8785

O.8711

0.8710

o . 8907

0.8906

0.8982

0.9301

0.9286
o . 8990*
o . 7940
0.8040
0.8044
0.8094
0.7887
0.8I2I

I • 4307
I • 9330
1.2569

2. 1823

a
o

Q
V

tr.
39.46
36 . 50
36.58

34.17
39.76
36.92

36.82

36.58

38."

23.87

24.45
28.90

28.40

28.48

28.24

24 58
24. 10

24.44
23.62

23.26

23.32

23.72

« •

25.23
24.58

23 87
22.39
22.68
24.23
24 .25
20.44

23 • 56
23 23
28.24
32.50
38.83

o.
>

72.8

63.9

8.2

5.8
6.7

7.5

9.4

14.15

II. 9
14. 1

18.8
28.4

42.0

24.3

72.8

(88 . ey

44.0

15.8

(5.o)«
31. 75

(387)"
(iio)»
62.0
9.2

Average value of K

^ 2

S
42.67
40.19

42.67

45.17

30.94
31.37
33.88

• *

34.72

• •

34.72
33.88

m m

33.05
32.21

33.05
43.92
29.07

30.94

35.14

39.74
41.41

44.34

40.37

52.94
27.61

30.54

• •

34 30
= 2.53

•I
U

2.6

2.8

2.6

2.5

2.4

• *

2.4

2.4
2.6

2.6

2.7

2.6

2.4

2.5

2.5
2.4

2.5

• «

2.4
2.6

2.6

2.5
2.2

2-3

2.3

2.1

2.4
2.8

3.0
30
3.0

* Richards aifd Stull.

* From I^ndolt and Bomstein,

THE RBl^ATiON BETWEEN COMPRESSIBILITY, ETC.

addition of 0.967 atmosphere, 0.08 milliliter in one liter. This particular
value is about doublfe that for water.

Having thus briefly stated the nature of the results, the table contain-
ing them may be given at once. In this table, the results are classified
according to the composition of the substances, isomeric bodies being
placed together. A glance at the table will show that in general the
compressibility is large when the surface tension is small, and vice versa.
Empirically it was found, especially among similar substances, that if
the four-thirds power of the surface tension is multiplied by the com-
pressibility, very nearly a constant value is found, especially among
similar substances. This approximately constant value is given in the
last column.

The results tabulated in the first and fourth columns of figures in this
table may well be plotted in a diagram, showing the relation between
compressibility and surface tension. As has been said, the general
tendency of this curve is expressed by the equation py^ — constant.

Comparison of the surface tensions and compressibilities of thirty-one compounds
of carbon, hydrogen and nitrogen or oxygen. Surface tension (y) is plotted in the
direction of ordinates, compressibility (fi) in the direction of abscissae. The curve
represents the equation py*^ ^ constant.

ITiis diagram contains all the results, excepting the four halogen com-
pounds, which have surface tensions too high to correspond with their
compressibilities — that is to say, where the value for the constant (3.0)
is considerably above the average, 2.5.

That the relationship should be affected by the specific nature of the
material need cause no surprise. Indeed it is surprising that the parallel-
ism should be so great as it is. The surface tension may be supposed
to be determined chiefly by the cohesive affinity of the substance, or
from the molecular point of view, by the attraction between one molecule
and another. We have shown that this attraction is probably one of the
factors entering into the compressibility of a substance, but it is not the

12 THEODORE W. RICHARDS AND J. HOWARD MATHBWS.

only one. Obviously, not all substances have the same compressibility
when subjected to the same increase of pressure, even when they were at
first under the same internal tension. Moreover, we must suppose that
the compressibility includes within its magnitude not only the change in
volume of the outside portions of the molecule, affected by the physical
pressure of cohesion, but also the internal alteration of the molecule as
well. Hence it is not at all surprising that the specific nature of the
substance exercises a distinct effect upon this latter property, and there-
fore upon the relationship between it and surface tension.

In endeavoring to connect the other properties of material we see in
the same way that the specific nature of the substance conceals in part
relationships which might otherwise become manifest. * For example,
the specific gravity would naturally be supposed to be greater in com-
pounds with great surface tension and small compressibility, than in
those with small surface tension and great compressibility. The latter
substances should be compressed into smaller bulk by the energy of their
own cohesive affinity. Such substances also should have high boiling-
points and high molecular heats of vaporization. To a certain extent
one may trace this connection of properties upon comparing the data
for a number of substances taken at random, but it is evident that the
matter is not quite so simple as would appear from the above statement.
Clearly the specific gravity is enormously affected by the nature of the
elements which build up the atom, and variation in the composition may
wholly conceal the effect due to surface tension or cohesive affinity. For
example, the average specific gravity of hydrogen in an organic com-
pound appears to be only 0.18, because the atomic volume of hydrogen
is 5.5, while the specific gravity of carbon in an organic compoimd is
1.09, and that of oxygen varies between 1.3 and 2.0 on the usually accepted
basis. These values are simply obtained by dividing the atomic weight
by the usually accepted atomic volume.

On the other hand, in cases of isomeric compotmds, regularities appear
in the expected direction, except indeed where methyl compounds are
concerned. Take for example the substances with the formula C^HuOj,
eth^d butyrate and its isomers. From the figures given in the table it
is clear that the order of magnitude in the case of all these properties
places ethyl butyrate on one extreme and ethyl isobutyrate on the other
extreme with isobutyl acetate in the middle; and that the direction is
always that demanded by the reasoning above.

It is not, however, our purpose in the present paper to attempt to sift
out and explain all the perplexities of these data. That, when sufficient
knowledge is obtained concerning them, the variations will be capable of
explanation, at least in a qualitative way, we have no doubt, and indeed
most of them are at present explicable. The object of the present paper

THE ATOMIC WEIGHT OI^ CHLORINE. 1 3

is to point out the particular relationship between surface tension and
compressibility, and to call attention to the fact that we are working
further upon this relationship and other relationships concerning other
allied properties of substance.

We are greatly indebted to the Carnegie Institution of Washington for
generous aid in this research.

Summary. — (i) In this paper are given a number of new results on the
compressibiUty, surface tension, boiling point, specific gravity and the
vapor pressure at 20° of a number of organic substances.

(2) It is shown that approximate relationships exist between some of
these quantities, particularly that as a rule substances with large surface
tension possess small compressibility.

(3) This relationship is discussed briefly from the point of view of the
theory of compressible atoms.

Harvard University,
Cambridge, Mass.

[CONTRIBUTIONT PROM THE CHEMICAL LABORATORY OP THE BUREAU OP STANDARDS

No. 4.]

THE ATOMIC WEIGHT OF CHLORINE.

By William A. Noybs and H. C. P. Webbr.
Received October 9, 1907.

The ratio of the atomic weights of oxygen and of chlorine is one of ex-
treme importance, on account of the number of atomic weights based
either directly or indirectly upon the atomic weight of chlorine.

During the last few years alone, since the determination of the ratio
silver to chlorine by Richards and Wells, ^ the atomic weights of a con-
aderable number of the common elements have been determined, basing
them on the value of chlorine. These values have been calculated on the
oxygen basis, assuming that the ratio silver : oxygen : : 107.93: 16 is correct.
Guye and Ter-Gazarian^ have called attention to a possible source of
error in the chlorate ratio of Stas, correction for which would bring the
value of silver down to 107.89. The newly accepted value, 14.01 for
nitrogen also points to the lower value of 107.89. Very nearly at the
close of this work conclusive evidence has been presented by Richards
and Forbes* and by Richards and Jones* that the value, 107.93 for silver
is too high. The only direct comparison between hydrogen and
chlorine which we have is that of Dixon and Edgar.* In this determina-

* This Journal, 27, 459.
'Compt. rend., 143, 411.
' This Journal, 29, 808.

* Ibid., 29, 826.

* Phil. Trans., 205, 169. Series A., Chem. News, 91, 263. The determinations by
Dcutsch (Dissertation, 1905) in the laboratory of Professor Guye, were scarcely of
saflicient accuracy to be considered as atomic weight determinations.

14 WILLIAM A. NO YES AND H. C. P. WEBER.

tion hydrogen was made to bum in an excess of chlorine. The hydrogen
was weighed absorbed in palladium and the chlorine in the liquid form
in a glass bulb. The hydrogen and chlorine were caused to unite in a
large globe of glass containing a small quantity of water to absorb the
hydrochloric acid formed. Corrections were applied for the quantity of
chlorine remaining uncombined by titrating the amount of iodine liberated
by it from a solution of potassium iodide. A further correction was
applied for the amotmt of chlorine used up in liberating oxygen from
water, by determining the amount of oxygen set free.

The fact that there was only one such direct determination of the
ratio between chlorine and hydrogen, together with the opportunity
afforded of carrying out the determination with hydrogen prepared in
the same apparatus used for generating the hydrogen in the recent deter-
mination of the ratio of hydrogen to oxygen, made a new determination
seem to be worth while.

The method we have used, besides being a direct comparison between
hydrogen and chlorine, involves the principle of complete synthesis with
the determination of the weights of all the substances reacting and of the
reaction products formed. Briefly stated, the method consists in weigh-
ing the hydrogen absorbed in palladium, and the chlorine in the form of
potassium chlorplatinate. The hydrogen is passed over the heated
potassium chlorplatinate from which it removes chlorine to form hydro-
chloric acid. The hydrochloric acid formed is condensed in a third
section of the apparatus and weighed. We have thus the weight of
hydrogen used, the weight of chlorine removed, and the weight of hydro-
chloric add formed. In this manner two series of ratios, each independent
of the other, are obtained.

Working in this manner and with hydrogen prepared under the same
conditions and at the same time as that used in the determination of the
ratio hydrogen to oxygen by one of us, we believe that we have very
favorable conditions for bridging the gap

Ag »-^ CI C H •^ O.

Purification of Materials and Weighing.

Hydrogen. — ^The hydrogen used in these experiments was prepared
and purified in the same manner as described in a previous paper on the
atomic weight of hydrogen.* The gas was taken from the generating
apparatus at intervals covered by the period of the work on hydrogen.
Consequently all remarks concerning its character and purity as used
in the hydrogen-oxygen ratio apply to these determinations. As in that
work so in this, two methods of generating the hydrogen were employed.
In the last series of determinations the hydrogen was obtained by the
electrolysis of a solution of barium hydroxide.
* Noyes: This Journal, 29, 1720.

THE ATOMIC WEIGHT OF CHLORINE. 15

Plaiinum, — ^The platinum used in the preparation of potassium chlor-
platinate was originally obtained in the form of platinum sponge. The
preliminary purification consisted in dissolving this in aqua regia and
evaporating to remove nitric acid. The separation from other platinum
metals was carried on according to the method of Schneider and Seubert
as described in Graham Otto's Lehrbuch.^

The solution of chlorplatinic acid was boiled for half an hour with ex-
cess of caustic soda, acidified and the platinum precipitated as potassium
chloiplatinate. The chlorplatinate so obtained was reduced with sodium
formate. The platinum black was then heated with dilute hydrochloric
acid to remove iron and washed until it commenced to go through as
colloidal platinum. It was then redissolved and the process repeatedly
gone through until the mother-liquors from the chlorplatinate precipita-
tion were practically colorless and free from other platinum metals. As
the same platinum was continually used and underwent a large ntunber
of successive solutions and reductions during the preliminary work, it
seems safe to assume that it was sufiidently pure.

Potassium Chlorplatinate. — During the preliminary work it was soon
discovered that the preparation of chlorplatinic acid by the use of aqua
regia was unsatisfactory. The removal of nitric acid by the process of
repeated evaporation was tedious and at best uncertain. To overcome
this difficulty and eliminate nitric acid entirely, a process of dissolving
the platinum electrol3rtically in purified hydrochloric add was devised.
This proved quite satisfactory and will be described in the following paper.
After solution of the platinum in hydrochloric acid had been effected,
the solution, which contained approximately 120 grams of platinum and
measured 500 cc, was evaporated to about one-half its volume in a glass
stoppered wash-bottle. At the same time a current of chlorine was
passed through the boiling solution. The chlorine used for this purpose
was prepared by the action of pure potassium permanganate upon chemi-
cally pure hydrochloric acid which had been previously boiled with a
small quantity of permanganate to insure its freedom from bromine
compounds. The solution of chlorplatinic acid thus obtained had a
beautiful bright color matching almost exactly that of a o.i per cent,
solution of methyl orange. It contained about 100 grams of hydrochloric
add in excess and after filtration and dilution to one liter was used directly
for the predpitation of potassium chlorplatinate. For this purpose a
solution of potassium chloride was prepared, using an excess of one-third
above the theoretical quantity dissolved in one liter of water. The ex-
cess of potassium chloride as well as the excess of hydrochloric add in
the chlorplatinic add were deemed necessary to check hydroljrtic de-
composition. For the same reason the predpitation of potassium chlor-

' Oraham Otto's Lehrbuch, 5th Ed., 4, 11 53.

14 WILLIAM A. NOYBS AND H. C. P. WEBER.

The method we have used, besides being a direct comparison between
hydrogen and chlorine, involves the principle of complete synthesis with
the determination of the weights of all the substances reacting and of the
reaction products formed. Briefly stated, the method consists in weigh-
ing the hydrogen absorbed in palladium, and the chlorine in the form of
potassium chlorplatinate. The hydrogen is passed over the heated
potassium chlorplatinate from which it removes chlorine to form hydro-
chloric acid. The hydrochloric acid formed is condensed in a third
section of the apparatus and weighed. We have thus the weight of
hj'^drogen used, the weight of chlorine removed, and the weight of hydro-
chloric acid fonned. In this manner two series of ratios, each independent
of the other, are obtained.

Ag »-^ CI O H •^ O.

Purification of Materials and Weighing.

■«.■ ^rnH

vTOMIC WEIGHT OF^CHU>RINE. 17

I i^m MK. hours, after which it was concentrated and the

^ i... - r : MB. ns- »v'ed to crjrstallize out. Following this the salt

^ ^^- -^i atar ^KL times from water, precipitated three times by

L - - '.^mkr^eL w ipitated from aqueous solution by purified hydro-

. crystals in these various processes were separated

^o: a ar^ • s by centrifugal drainage.

_^_^_^ ride certainly contained less bromine than i in

')romine the following process, a modification of
'rews,' was adopted.

ai — .- «

y^xtam

Pig. 1.

kd with a ground glass joint at B which ends in an

lary at C. The side tube D has a trap to prevent

I fried over. An ordinary distilling flask is affixed

le tube with a rubber stopper in such a manner as to

well into the bulb of the flask E. The flask A is

■ . of distilled water, 20 cc. of N/5 potassium iodate

itric acid. The flask E contains about 5 cc. of a 4

•f potassium iodide (which must not liberate iodine

1). The flasks are then connected by means of the

:icuum applied. The capillary tube which is groimd

'^ *E-* is connected with a carbon dioxide generator, and

*" — ')een applied, should yield a fairly steady stream of

e- .. liquid, so as to insure regular boiling and at the same

t -^:^_ y to diminish the vacuum. It is desirable to have a

^ li as carbon dioxide or nitrogen passing through the
V s of reducing substances in the air used, would vitiate
^as passing through the capillary, besides insuring
^ as a diluent for the steam formed and as such checks

HjO + Br, « 2HBr -f O.

., 275.

l6 WILLIAM A. NOYES AND H. C. P. WEBER.

platinate was carried out with as concentrated solutions as practicable.
The precipitation itself was carried out by pouring the platinum solution
into the potassium chloride in a fine stream, the precipitate meanwhile
being agitated thoroughly by a current of air. In some cases as much
as four hours were spent in precipitating 300 grams of potassium chlor-
platinate. The potassium chlorplatinate so obtained was of a pale yellow
color, resembling precipitated sulphur, and wa« microcrystalline. It was
filtered from the mother-liquor by means of a suction pump, washed
with water and finally with alcohol and ether. After having been re-
moved from the hardened filter, it was heated for some time in a large
platinum dish on an electric air bath until the greater part of the moisture
retained had been driven off. The quantity necessary for one deter-
mination was then transferred to a hard glass tube. This tube was
placed in a cylindrical air bath which could be raised to a temperature of
400°. The potassium chlorplatinate was gradually raised to this tem-
perature, a current of air, dried successively by sulphuric acid and phos-
phorus pentoxide passing over it continuously. The behavior of the
salt under these conditions served as a criterion of its purity. At the exit
of the current of gas, a wash-bottle was placed containmg a drop of methyl
orange. At first a small quantity of moisture and hydrochloric acid
passed off. Finally, however, a point was reached where no further
hydrochloric acid could be detected in the issuing gas stream. At this
point the heating was stopped, usually after the chlorplatinate had been
heated to 400 ^^ for about seven hours, or more. It seems that the chlor-
platinate, when pure, will stand the temperature of 400° indefinitely in a
non- reducing atmosphere, without suffering any change chemically.
No appreciable quantity of hydrochloric acid could be detected after the
hydrochloric acid and water held mechanically had once been driven off.
This all, provided the salt was pure to start with. In the first trials,
in which the chlorplatinate had been prepared by the use of aqua regia
its behavior was entirely different. In these cases decomposition set in
in the neighborhood of 250° and seemed to go on progressively throughout
the whole mass of the salt.

After having been dried in this manner, the chlorplatinate was trans-
ferred to the final apparatus with little or no exposure to the air, through
an opening which was sealed off after filling. In this it was subjected
to final drying at 350° and evacuation, as described under the manipula-
tions.

Potassium Chloride. — ^As a starting point for the preparation of pure
potassium chloride, the purest commercial article obtainable was taken.
This was first recrystallized from a solution made slightly alkaline with
potassium hydroxide to remove traces of ammonia which were present.
The salt was then redissolved in water and chlorine passed through the

TBB ATOMIC WBIGHT OP^CHLORINB. 1 7

bot solution for several hours, after which it was concentrated and the
potassjum chloride allowed to crystallize out. Following this the salt
was recrystallized five times from water, precipitated three times by-
alcohol and finally precipitated from aqueous solution by purified hydro-
chloric acid gas. The crystals in these various processes were separated
from the mother-liquors by centrifugal drainage.

This potasaum chloride certainly contained kss bromine than i in
50,000. To test for bromine the following process, a modification of
that described by Andrews,' was adopted.

'0^<

, yxcmm

Fir. 1

The flask A is fitted with a ground glass joint at B which ends in an
extremely fine capillary at C. The side tube D has a trap to prevent
spray from being carried over. An ordinary distilling flask is affixed
to the end of this side tube with a rubber stopper in such a manner as to
bring the side tube well into the bulb of the flask E. The flask A is
charged with 200 cc. of distilled water, ao cc. of N/5 potasMum iodate
and 20 cc of 2 N. nitric add. The flask E contains about 5 cc. of a 4
per cent, solution of potassium iodide (which must not liberate iodine
upon being acidified). The flasks are then connected by means of the
stopper at F and vacuum applied. The capillary tube which is ground
into the flask at 6 is connected with a carbon dioxide generator, and
after vacuum has been applied, should yield a fairly steady stream of
bubbles through the liquid, so as to insure regular boiling and at the same
time not appreciably to diminish the vacuum. It is desirable to have a
pure neutral gas such as carbon dioxide or nitrogen passing through the
capillary, since traces of reducing substances in the air used, would vitiate
the results. The gas passing through the capillary, besides insuring
regular boiling, acts as a diluent for the steam formed and as such checks
the reaction

H,0 + Br, =■ 2HBr + O.
■ This Jounia], 39, 373.

l8 WILLIAM A. NOYES AND H. C. P. WEBER.

The flask containing the potassium iodate and nitric acid mixture is
then heated on the water bath until loo cc. have distilled into the flask
containing the potassium iodide. If the resulting distillate is colored
by the presence of free iodine, the process is repeated after adding water
to make up for the quantity distilled off, until a satisfactory blank is
obtained. Then 3-5 grams of the potassium chloride to be tested are
dissolved in a little water and added to the flask containing the iodate
and the volume of the solution made up to 250 cc. again. The distillate
containing the free iodine corresponding to the amount of bromine distilled
over, is then titrated with a solution of thiosulphate corresponding to i
mg. of bromine per cubic centimeter. To 5 grams of potassium chloride
which had been treated until blank distillates were obtained the follow-
ing quantities of bromine were added and found :

Added, i mg. 0.5, 0.3, o.i.
Found, I mg. 0.52, 0.36, 0.12.

The potassium chloride used for precipitating potassium chlorplatinate
contained less than 0.1 mg. bromine in 5 grams or i : 50,000. This amoimt,
it is safe to assume, was further reduced in the preparation of potassium
chlorplatinate.

Hydrochloric Acid. — ^The hydrochloric acid used in the preparation of
pure hydrochloric acid was free from sulphuric and nitric acid. It was
treated by allowing chlorine to bubble through it for one day. Follow-
ing this, air was bubbled through the acid saturated with chlorine until
the chlorine was expelled and the acid was again colorless. Usually the
air was left passing through the acid over night, a reduction of about
one-fifth in the volume of the acid, due to evaporation, taking place
with the removal of the excess of chlorine. During the manipulations in
preparing chlorplatinic acid, it was further subjected to the action of chlo-
rine twice, namely, during electrolysis of the platinum and on evaporation
of the platinum solution.

Chlorine. — ^AU the chlorine used was generated by the action of pure
potassium permanganate on hydrochloric acid which had been previously
boiled with a small quantity of potassium permanganate.

Water. — ^The water used was obtained by redistilling distilled water
with alkaline permanganate, rejecting the first part of the distillate until
it no longer contained ammonia.

Balance and Weights. — ^The balance and weights were identical with
those described imder the atomic weight of hydrogen.^ The air of the
balance case was dried by means of a current of air as described in the
previous paper. Bach piece of apparatus was weighed with a corre-
sponding counterpoise approaching it within I cc. in volume and 30 grams

« Noyes, This Journal, 29, 1723.

l8 WILLIAM A. NOYES AND H. C. P. WBBER.

The flask containing the potassium iodate and nitric acid mixture is
then heated on the water bath until lOo cc. have distilled into the flask
containing the potassium iodide. If the resulting distillate is colored
by the presence of free iodine, the process is repeated after adding watei
to make up for the quantity distilled off, until a satisfactory blank i
obtained. Then 3-5 grams of the potassium chloride to be tested an
dissolved in a little water and added to the flask containing the iodat
and the volume of the solution made up to 250 cc. again. The distillat
containing the free iodine corresponding to the amount of bromine distiUe
over, is then titrated with a solution of thiosulphate corresponding to
mg. of bromine per cubic centimeter. To 5 grams of potassium chlori*
which had been treated until blank distillates were obtained the folk)
ing quantities of bromine were added and found :

Added, i mg. 0.5, 0.3, 0.1.
Foimd, I mg. 0.52, 0.36, 0.12.

The potassium chloride used for precipitating potassium chlorplatin
contained less than o. i mg. bromine in 5 grams or i : 50,000. This amoi
it is safe to assume, was further reduced in the preparation of potass
chlorplatinate.

Hydrochloric Acid, — ^The hydrochloric acid used in the preparatio
pure hydrochloric acid was free from sulphuric and nitric acid. It
treated by allowing chlorine to bubble through it for one day. Fo
ing this, air was bubbled through the acid saturated with chlorine
the chlorine was expelled and the acid was again colorless. Usuall
air was left passing through the acid over night, a reduction of :
one-fifth in the volume of the acid, due to evaporation, taking
with the removal of the excess of chlorine. During the manipulatic
preparing chlorplatinic acid, it was further subjected to the action o^
rine twice, namely, during electrolysis of the platinum and on evapc
of the platinum solution.

Chlorine. — All the chlorine used was generated by the action c
potassium permanganate on hydrochloric acid which had been pre
boiled with a small quantity of potassium permanganate.

Water. — ^The water used was obtained by redistilling distillec'
with alkaline permanganate, rejecting the first part of the distilla
it no longer contained ammonia.

Balance and Weights. — ^The balance and weights were identk
those described under the atomic weight of hydrogen.^ The ai
balance case was dried by means of a current of air as describe*
previous paper. Bach piece of apparatus was weighe'^ with
spondingcounterpoise approaching it within i cc, in v^
' Noyes, This Journal, 29, 1723.

the hard

was con-

>rplatiiiate

off without

is was done

acuated to a

'. This tern-

12 THKODORE W. RICHARDS AND J. HOWARD MATHEWS.

only one. Obviously, not all substances have the same compressibility
when subjected to the same increase of pressure, even when they were at
first imder the same internal tension. Moreover, we must suppose that
the compressibility includes within its magnitude not only the change in
volume of the outside portions of the molecule, aflfected by the physical
pressure of cohesion, but also the internal alteration of the molecule as
well. Hence it is not at all surprising that the specific nature of the
substance exercises a distinct effect upon this latter property, and there-
fore upon the relationship between it and surface tension.

In endeavoring to connect the other properties of material we see in
the same way that the specific nature of the substance conceals in part
relationships which might otherwise become manifest.* Fo^ example,
the specific gravity would naturally be supposed to be greater in com-
potmds with great surface tension and small compressibility, than in
those with small surface tension and great compressibility. The latter
substances should be compressed into smaller bulk by the energy of their
own cohesive affinity. Such substances also should have high boiling-
points and high molecular heats of vaporization. To a certain extent
one may trace this connection of properties upon comparing the data
for a number of substances taken at random, but it is evident that the
matter is not quite so simple as would appear from the above statement.
Clearly the specific gravity is enormously affected by the nature of the
elements which build up the atom, and variation in the composition may
wholly conceal the effect due to surface tension or cohesive affinity. For
example, the average specific gravity of hydrogen in an organic com-
pound appears to be only 0.18, because the atomic volume of hydrogen
is 5.5, while the specific gravity of carbon in an organic compotmd is
1.09, and that of oxygen varies between 1.3 and 2.0 on the usually accepted
basis. These values are simply obtained by dividing the atomic weight
by the usually accepted atomic volume.

On the other hand, in cases of isomeric compotmds, regularities appear
in the expected direction, except indeed where methyl compounds are
concerned. Take for example the substances with the formula C^HijOa,
ethyl butyrate and its isomers. From the figures given in the table it
is clear that the order of magnitude in the case of all these properties
places ethyl butyrate on one extreme and ethyl isobutyrate on the other
extreme with isobutyl acetate in the middle; and that the direction is
always that demanded by the reasoning above.

THE ATOMIC WEIGHT OF CHLORINE. 1 3

We are greatly indebted to the Carnegie Institution of Washington for
generous aid in this research.

Summary. — (i) In this paper are given a number of new results on the
compressibility, surface tension, boiling point, specific gravity and the
vapor pressure at 20° of a number of organic substances.

(2) It is shown that approximate relationships exist between some of
these quantities, particularly that as a rule substances with large surface
tension possess small compressibility.

(3) This relationship is discussed briefly from the point of view of the
theory of compressible atoms.

Harvard University,
Cambridge, Mass.

[Contribution from the Chemical Laboratory of the Bureau of Standards

No. 4.]

THE ATOMIC WEIGHT OF CHLORINE.

By William A. Noyes and H. C. P. Wbbbr.
Received October 9, 1907,

During the last few years alone, since the determination of the ratio
silver to chlorine by Richards and Wells, ^ the atomic weights of a con-
siderable number of the common elements have been determined, basing
them on the value of chlorine. These values have been calculated on the
oxygen basis, assuming that the ratio silver : oxygen : : 107.93: 16 is correct.
Guye and Ter-Gazarian^ have called attention to a possible source of
error in the chlorate ratio of Stas, correction for which would bring the
value of silver down to 107.89. The newly accepted value, 14.01 for
nitrogen also points to the lower value of 107.89. Very nearly at the
close of this work conclusive evidence has been presented by Richards
and Forbes' and by Richards and Jones* that the value, 107.93 ^^^ silver
is too high. The only direct comparison between hydrogen and
chlorine which we have is that of Dixon and Edgar.' In this determina-

* This Journal, 27, 459.
'Compt. rend., 143, 411.
' This Journal, 29, 808.

* Ibid., 29, 826.

* Phil. Trans., 205, 169. Series A., Chem. News, 91, 263. The determinations by
Deutsch (Dissertation, 1905) in the laboratory of Professor Guye, were scarcely of
sufficient acctu'acy to be considered as atomic weight determinations.

14 WILLIAM A. NOYES AND H. C. P. WEB^R.

tion hydrogen was made to bum in an excess of chlorine. The hydrogen
was weighed absorbed in palladium and the chlorine in the liquid form
in a glass bulb. The hydrogen and chlorine were caused to unite in a
large globe of glass containing a small quantity of water to absorb the
hydrochloric acid formed. Corrections were applied for the quantity of
chlorine remaining uncombined by titrating the amount of iodine liberated
by it from a solution of potassium iodide. A further correction was
applied for the amount of chlorine used up in liberating oxygen from
water, by determining the amount of oxygen set free.

The fact that there was only one such direct determination of the
ratio between chlorine and hydrogen, together with the opportunity
aflforded of carrying out the determination with hydrogen prepared in
the same apparatus used for generating the hydrogen in the recent deter-
mination of the ratio of hydrogen to oxygen, made a new determination
seem to be worth while.

The method we have used, besides being a direct comparison between
hydrogen and chlorine, involves the principle of complete synthesis with
the determination of the weights of all the substances reacting and of the
reaction products formed. Briefly stated, the method consists in weigh-
ing the hydrogen absorbed in palladium, and the chlorine in the form of
potassium chlorplatinate. The hydrogen is passed over the heated
potassium chlorplatinate from which it removes chlorine to form hydro-
chloric acid. The hydrochloric acid formed is condensed in a third
section of the apparatus and weighed. We have thus the weight of
h)^drogen used, the weight of chlorine removed, and the weight of hydro-
chloric acid formed. In this manner two series of ratios, each independent
of the other, are obtained.

Ag .-^ CI C H »^ O.

Purification of Materials and Weighing.

Hydrogen, — ^The hydrogen used in these experiments was prepared
and purified in the same manner as described in a previous paper on the
atomic weight of hydrogen.* The gas was taken from the generating
apparatus at intervals covered by the period of the work on hydrogen.
Consequently all remarks concerning its character and purity as used
in the hydrogen-oxygen ratio apply to these determinations. As in that
work so in this, two methods of generating the hydrogen were employed.
In the last series of determinations the hydrogen was obtained by the
electrolysis of a solution of barium hydroxide.
* Noyes: This Journal, 29, 1720,

THE ATOMIC WKIGHT OF CHLORINE* 15

Platinum. — ^The platinum used in the preparation of potassium chlor-
platinate was originally obtained in the form of platinum sponge. The
preliminary purification consisted in dissolving this in aqua regia and
evaporating to remove nitric add. The separation from other platinum
metals was carried on according to the method of Schneider and Seubert
as described in Graham Otto's Lehrbuch.^

The solution of chlorplatinic acid was boiled for half an hour with ex-
cess of caustic soda, acidified and the platinum precipitated as potassium
chlorplatinate. The chlorplatinate so obtained was reduced with sodium
formate. The platinum black was then heated with dilute hydrochloric
acid to remove iron and washed tmtil it commenced to go through as
colloidal platinum. It was then redissolved and the process repeatedly
gone through until the mother-liquors from the chlorplatinate precipita-
tion were practically colorless and free from other platinum metals. As
the same platinum was continually used and underwent a large number
of successive solutions and reductions during the preliminary work, it
seems safe to assume that it was sufficiently pure.

Potassium Chlorplatinate. — During the preliminary work it was soon
discovered that the preparation of chlorplatinic add by the use of aqua
regia was unsatisfactory. The removal of nitric acid by the process of
repeated evaporation was tedious and at best imcertain. To overcome
this difficulty and eliminate nitric acid entirely, a process of dissolving
the platinum electrolytically in purified hydrochloric add was devised.
This proved quite satisfactory and will be described in the following paper.
After solution of the platinum in hydrochloric acid had been effected,
the solution, which contained approximately 120 grams of platinum and
measured 500 cc, was evaporated to about one-half its volume in a glass
stoppered wash-bottle. At the same time a current of chlorine was
passed through the boiling solution. The chlorine used for this purpose
was prepared by the action of pure potassium permanganate upon chemi-
cally pure hydrochloric acid which had been previously boiled with a
small quantity of permanganate to insure its freedom from bromine
compounds. The solution of chlorplatinic acid thus obtained had a
beautiful bright color matching almost exactly that of a 0.1 per cent.
solution of methyl orange. It contained about 100 grams of hydrochloric
add in excess and after filtration and dilution to one liter was used directly
for the precipitation of potassium chlorplatinate. For this purpose a
solution of potassium chloride was prepared, using an excess of one-third
above the theoretical quantity dissolved in one liter of water. The ex-
cess of potassium chloride as well as the excess of hydrochloric acid in
the chlorplatinic add were deemed necessary to check hydrolytic de-
composition. For the same reason the predpitation of potassium chlor-

^ Graham Otto's Lehrbuch, 5th Ed., 4, 11 53.

24 WILLIAM A. NOYBS AND H. C. P. WEB^R.

was kept fully submerged in liquid air. After the supply of hydrogen
had been turned off, the pressure in the apparatus gradually sank until
all hydrogen had been used up and all hydrochloric add condensed. The
conditions then were: excess of the chlorplatinate at 350° to 400°; the
hydrochloric add at about — 180° and a small fraction of a millimeter
residual pressure. The stopcocks of the condensation tube were then
closed and the residual gases in the remaining parts of the apparatus
pumped out.

The condensation tube was then separated from the remaining parts
of the apparatus and the transfer of the hydrochloric add to the water
commenced. The water bulb L was plunged into ice water and the
condensation bulb K removed from direct contact with the liquid air
but not entirely beyond the cold vapors. As soon as the pressure of the
hydrochloric acid had risen to a small part of an atmosphere, communica-
tion was established with the cold water in L. After some hydrochloric
acid had been absorbed by the water the bulb L was cooled by ice and
dilute sulphuric acid instead of by ice alone. With the aid of occasional
shaking of the absorption bulb the last of the condensed hydrochloric
acid was finally absorbed by the water. With cautious manipulation
the pressure in the apparatus did not rise above one-half atmosphere
during the transfer and remained at a few millimeters at the close. The
stopcock / between K and L was then closed, and the apparatus was
cleaned, rinsed, wiped and set aside to be weighed. The determinations
of the second series required about ten hours from beginning to end.
About two hours were consumed in transferring the solidified hydrochloric
add from the condensation to the absorption bulb. The solidified hydro-
chloric acid resembled snow in appearance and the greater part of it was
absorbed by the water without passing into the liquid state. Towards
the end of the transfer, when the absorption became less rapid, the re-
maining hydrochloric acid would liquefy attended by a sudden rise in
pressure. Shaking of the absorption tube immediately reduced the
pressure and caused the liquid hydrochloric acid to solidify to a glassy
solid.

Errors and Corrections.

Hydrogen, — If there were any errors due to a contamination of the
hydrogen, they are not apparent. The hydrogen was prepared and
purified with all possible precaution. It was repeatedly tested for im-
purities in the work on hydrogen and oxygen and none or only negligible
quantities found.'

Two corrections on the weight of hydrogen were found necessar>'.
The first of these was the amount of hydrogen remaining in the chlor-
platinate tube at the end of the determination and pumped out with the

* Noves, This Jotunal, 29, 1724.

THE ATOMIC WEIGHT OF CHU>RINE. 25

residual gases to be analyzed. In eight cases out of the twelve this cor-
rection was applied, the maximum being o.ii mg., the minimum 0.0 1 mg.
and the average 0.047 ^^K- The determinations affected are i, 2, 4, 5,
6, 7, II and 12. The second correction was due to hydrochloric acid
in the palladium tube. Under certain conditions, when the current of
hydrogen from the palladium tube was not sufficiently rapid, hydro-
chbric add found its way back into the palladium tube, either by diffu-
sion or by being drawn back. Correction for this was readily applied.
After having been weighed, the hydrogen apparatus was charged for the
succeeding determination. The tube was fitted with a stopcock at both
ends, as may be seen in the photograph. After the palladium had been
saturated with hydrogen the stopcock farthest from the hydrogen genera-
tor was connected to a Liebig flask containing water colored by a drop of
methyl orange. The stopcock at A was opened while the hydrogen
generator was kept going and the hydrogen bubbled through this water
before escaping. If the presence of hydrochloric acid made itself known,
the palladium was heated to 160° to expel the larger part of the hydrogen
it had absorbed and with it any hydrochloric acid present. It was then
aUowed to cool with a current of hydrogen passing through it and finally
filled again at normal temperatures. The hydrochloric acid absorbed
by the water was then titrated by N/io sodium or barium hydroxide.
The real weight of the hydrogen was greater than the apparent loss of
the tube by the weight of hydrochloric acid found. This correction was
applied in determinations i, 3, 4, 10 and 11 varying between 0.35 and
9.26 rag. and averaging 4.8 mg.

Chlorine from the Loss of the Platinum Tube. — ^The purity of the chlorine
entering into the composition of the chlorplatinate has been spoken of.
The maximum amount of bromine in the ingredients used seems to have
been i in 50,000. The resultant error in the weight would be half of
this, since 35.5 grams are replaced by 80, or i: 100,000. There seems to
be ample justification for considering this point beyond question.

The only other impurities which could possibly affect the results were
volatile products in the platinum or elements capable of producing volatile
products such as ammonia, water, hydroxyl or oxygen. The method of
preparation seems to preclude the possibility of the presence of ammonium
salts. The potassium chloride was especially treated to free it from
these. Following this there was repeated treatment with chlorine in hot
solution. The final heating of the chlorplatinate to 400^ over an ex-
tended period of time must have caused the destruction and removal of
any ammonium compounds which had escaped previous treatment.

The three following impurities, water, hydroxyl or oxygen in the platinum
salt, were somewhat nTore difficult of treatment. It is known that potas-
sium chlorplatinate hydrolyzes in aqueous solution. With this point

26 WILlrlAM A. NOYES AND H. C. P. WEBSR.

in view the precipitation of potassium chlorplatinate was carried out in
concentrated solutions with both an excess of potassium chloride and
hydrochloric acid, and the mother-liquor was removed as soon as practi-
cable. It was shown that by heating sufficiently long at 400°, all occluded
hydrochloric acid could be removed, the chlorplatinate being perfectly
neutral after this treatment. This served as indirect e^ddence that the
water was also removed. The agreement between Series i and 2 may
be considered as additional evidence on this point. In the first series
the chlorplatinate had opportunity to become saturated with water
vapor at the temperature of 350** during the course of the experiment.
In the second series it was in equilibrium with solid hydrochloric acid at
— 180^. Now hydrochloric acid with its well-known affinity for water
vapor may be considered as a very perfect drying agent. Yet the differ-
ence between the two series is i in 10,000. One hundred and twenty
grams of potassium chlorplatinate were used. The presence of 0.05
per cent, moisture in this salt would have raised the atomic weight of
chlorine found from 35.184 to 35.244.

Further, in the second series it was noticed that if the condensation
tube for the hydrochloric add was allowed to contain a trace of moisture
before the determination was begun this would remain as a trace of aqueous
hydrochloric acid after the gas had been transferred from this part of the
apparatus to the absorption bulb, the temperature during this transfer
remaining, of course, below zero. If, however, the condensation tube
was perfectly dry to start with, no such trace of aqueous hydrochloric
acid remained, that is, no water had been carried over from the chlor-
platinate.

A final test of the chlorplatinate was made for water, hydroxyl or
oxygen. One hundred grams of potassium chlorplatinate were treated
exactly as for a determination* The chlorplatinate was heated to 400*^
in a current of dry air, transferred to the apparatus used in the deter-
minations and then evacuated at 350°. Next, pure hydrogen was led
over the chlorplatinate until it was completely reduced. The hydro-
chloric add formed was led through a narrow U-tube surrounded by a
mixture of solid carbon dioxide and alcohol ( — 78°). Finally, the whole
apparatus including the U-tube was evacuated, the final conditions
being, 350° in the chlorplatinate tube, — 78° in the U-tube and several
thousandths millimeters pressure. Upon weighing, the U-tube had
shown an increase in weight of 0.9 mg. or i in 30,000 on the hydrochloric
acid formed.

A number of corrections on the weight of the chlorplatinate tube were
necessary. The first of these was necessary on account of the traces
of air or nitrogen fotmd in the chlorplatinate tube at the end of the deter-
jnination, These corrections were found necessary in Experiments 4,

THE ATOMIC wmCHT OP CHLORINE. 27

5, 6, and 7, their magnitude being 0.35, 0.76, 0.25 and 2 rag. This air may
have been occluded by the chlorplatinate, or it may have come from the
water of the absorption tube. It was immaterial how this correction
was applied as it lay within the limits of the experimental errors, the
maximtun correction being i : 10,000.

In the second series it sometimes happened that with the most cautious
manipulation, chlorplatinate was blown over into the condensation bulb
at the beginning of the experiment. After the h3'drochloric acid had
been absorbed by the water and the apparatus weighed, this chlorplatinate
was removed by dissolving in water and weighing first as potassium
chlorplatinate and then again as potassium chloride + platinum, after
reduction. This correction was applied in Experiments 8, 9 and 11, the
amounts being 2.03, 15.0 and 36.5 nig. The apparent loss of the chlor-
platinate tube was of course diminished by these quantities.

Hydrochloric Acid. — From the agreement of Series I and II it seems
reasonable to assume that no errors were introduced by using water as
the absorbing agent for the hydrochloric acid.

The corrections on the apparent gain of the absorption tube were of the
following nature: First, the apparent gain was increased by the hydro-
chloric acid found in the palladium tube and by the hydrochloric add
pumped out from the chlorplatinate tube. The former has been spoken
of under hydrogen. In Experiments i, 6, and 7, there were found 1.27, 0.8
and 0.15 mg. of hydrochloric acid, respectively, in the gases pumped out.

In Experiments 8, 9, and 11, the gain of the absorption tube was dimin-
ished by the amount of the chlorplatinate blown over. The corrections
were 2.03, 15.0 and 36.5 mg.

The final check on correctness of manipulation, freedom from leaks
and other losses was found in the checking of the sum of the weights of
the hydrogen and chlorine with the weight of the hydrochloric acid. In
the preliminary experiments of both series, discrepancies were found until
the details of the determination had been mastered. On this account
and on account of other known errors the preliminary determinations
were rejected entirely.

Results.

The values obtained are given in the following tables. The weights rep-
resent those found after all corrections had been applied. The second
part of Experiment 5 was lost.

The value as found for the atomic weight of chlorine is 35.184 with a
probable error of ±10.0013. The value obtained for the molecular weight
of hydrochloric acid is 36.184 with a probable error of ±0.0012. The
combined average of both sections of the two series is 35.184 with a prob-
able error of ±0.0008. This is, of course, on the basis of hydrogen = i.

28 WILUAM A. NOYES AND H. C. P. WEBER.

Hydrochloric
Hydrogen. Chlorine. acid. At. wt.Cl. Mol.wt.HC1.

1 0.25394 8.93293 918695 35177 36.178

2 0.28004 985590 10.13259 35195 36.183

3 0.51821 18.23468 18.75359 35188 36.189

4 0.67631 23.79587 24.47123 35186 36.185

5 0.58225 20.48158 35177

6 0.47989 16.88423 17.36310 35.184 36.182

7 0.64132 22.55816 23.20054 35.175 36.176

Average of Scries I, 35 . 183 and 36 . 181

8 0.81608 28.71691 29.53167 35.188 36.187

9 0.83194 29.28055 30.11207 35.195 36.195

10 0.39074 13.74926 14.14078 35.187 36.188

II 0.75560 26.58427 27.33926 35183 36.182

12 0.77518 27.26746 28.04110 35.177 36.175

Average of Series II, 35. 186 and 36. 185

Total average, 35 184(3) 36 . 183(7)

On the oxygen basis this value becomes 35.452 if H = 1.00762* and 35.461
if H = 1.00787.* The values for silver calculated from Richards* ratio
and these two values are, respectively, 107.87 and 107.89.*

The mean values must be considered as most probable at the present
time. These are 35,457 for chlorine and 107,88 for silver.

In eleven experiments 6.41925 grams of hydrogen were united with
225.86017 grams of chlorine and yielded 232.27288 grams of hydrochloric
acid. The values obtained from these figures are 35.1846 and 36.1838.

H 4- Cl. HCl. Difference.

1 9.18687 9.18695 +0.00008

2 10.13594 10.13259 --0.00335

3 18.75289 18.75359 4-0.00070

4 24.47218 24.47123 — 0.00095

6 17.36412 17.36310 — 0.00102

7 23.19948 23.20054 +0.00106

8 29.53299 29 53167 — 0.00132

9 30.11249 30.11207 — 0.00042

10 14.14000 14.14078 +0.00078

II 27.33982 27.33926 — 0.00056

12 28.04264 28.04110 —0.00154

232.27942 232.27288 — 0.00654

* Morlejr's value.

• Noyes' recent value.

'#The value calculated from the results of Dixon and Edgar in the same manner
are 35.463 or 35.472 for chlorine and 107.90 or 107.93 for silver. The mean values,
35.467 and 107.91 must be considered, for the present as the most probable which can
be calculated from their work. The values for silver are calculated from the results of
Richards and Wells, who found that 100 parts of silver gave 132.867 parts of silver
chloride, This Journal, 27, 525.

PRBPARATION OF CHIX>RPLATINIC ACID. 29

It may be interesting to note the discrepancies between the weights of
hydrogen and chlorine and the hydrochloric acid formed in the individual
experiments.

In these eleven experiments there were seven with an apparent loss of
weight and four with an apparent gain, the total loss being i in 35,000.

Bureau of Standards,
Washington, D. C.

[Contribution prom ths Chbmical Laboratory op thb Bureau op Standards

No. 5.]

PREPARATION OF CHLORPLATINIC ACID BY ELECTROLYSIS OF

PLATINUM BLACK.

By H. C. p. Wbbbr.
Received October 9, 1907.

In the work* on the atomic weight of chlorine it was necessary to pre-
pare considerable quantities of chlorplatinic acid free from nitric acid.
When using aqua regia to dissolve platinum, considerable diflSculty was
experienced in removing the last traces of nitric add by evaporation.
When working with as much as 100 grams of platinum, the oft-repeated
evaporation to dryness of the solution becomes exceedingly tedious and
even at best, yields uncertain results. If the evaporation is carried on
with strong hydrochloric acid considerable quantities of material become
necessary, while with the use of water there is danger of hydrolytic de-
composition of the chlorplatinic acid and consequent contamination of
the chlorplatinate with hydroxychlorplatinates.

The process here described overcomes these difficulties and yields a
pure solution of chlorplatinic acid. The platinum is prepared for elec-
trolysis by dissolving platinum scraps or platinum sponge in aqua regia.
The excess of acid is removed either by neutralization or evaporation, and
the platinum solution is reduced by zinc or an alkaline formate, prefer-
ably the latter. The solution is decanted from the precipitated platinum,
which is then warmed with a little dilute hydrochloric acid to remove
iron. The platinum is then transferred to the electrolytic apparatus,
the washing of the precipitated platinum being completed in this ap-
paratus, which is constructed as follows:

It consists of a cylindrical tube about 4 cm. in diameter and 35 cm.
long, which ends in a narrow glass tube, about 4 mm. bore, which is
given the form of a siphon. The anode is a thin disk of sheet platinum
which just fits into the tube and is perforated with numerous small pin-
holes. A small piece of platinum wire is welded to the disk and carried
through the glass tube by means of sealing glass. The other end of the
platinum wire ends in a glass tube which is carried to the top of the ap-
* Noyes and Weber: See preceding article.

paratus and is filled with mercury to make connection for the current.
The platinum disk should be about 30 cm. from the top of the apparatus
at that point where the tube
commences to narrow. After
the anode has been sealed
^ ^jj into the tube the space below

it is filled with glass beads to
support the platinum disk,
which should rest firmly and
evenly upon them. About 5
cm. from the top of the tube
three notches are pressed into
the glass. From these the ca-
thode chamber is suspended.
This consists of a porous por-
celain filter about 18 cm. long
and 25 mm. in diameter. It
CAitOM is well that the rim of the

filter fit snugly in the glass
tube so that the filter cap
hangs in the tube fairly
rigidly.

The cathode consists of a
sheet of platinum 4 to 5 cm.
long and z to 3 cm. wide. To
it is connected a platinum
wire passing through a glass
tube. It is suspended from
A«vm' a perforated watch glass,

which serves as a cover for
the apparatus.

The whole apparatus is sus-
pended in a long cylinder by means of a large cork for the purpose of
cooling it by running water when necessary. This is not shown in the
diagram and may be dispensed with when low currents are used. The
apparatus has been used with a current up to 10 amperes. With a cur-
rent of this strength the cooling jacket is essential, as the apparatus grad-
ually becomes hot.

The platinum is transferred to the tube, being dropped on the anode
plate, and is here washed with dilute hydrochloric acid until clean. The
wash waters are drawn off by gentle suction at the siphon end S, The
platinum is then covered with concentrated hydrochloric acid. There
should be such a quantity of hydrochloric acid that the liquid stands on

lODOMETRIC DETEI^MINATION OP ARSENIOUS ACID. 3 1

a level with 5 when the porous cylinder is inserted. The porous cylinder
is then inserted and filled to the top with hydrochloric add. After the
cathode is inserted the apparatus is ready for electrolysis.

The current may be taken from a 120 volt direct current, lighting circuit
with a number of incandescent lamps in parallel with each other and in
series with the cell. The cell may be run continuously on 8-10 amperes.
The current is used quantitatively in dissolving platintun. During a
run of four and a half hours at 8 amperes, 64 grams of platinum were
dissolved. The theoretical quantity for 36 ampere hours is 65 grams.
While the apparatus is in operation the hydrochloric add travels from the
cathode cell to the anode tmder the influence both of gravity and electric
endosmoses. With the proper adjustments of height of hydrochloric
add in the anode cell, the heavy layer of chlorplatinic add solution is
delivered at the tip of the siphon S, drop by drop. If the flow of con-
centrated solution ceases for any reason it may again be started by gentle
suction at 5. For this purpose it is best to have the siphon tip 5 con-
nected with a receiving flask by means of a double perforated stopper.
The acid in the cathode chamber is replenished from time to time as it
becomes necessary.

K towards the end of the operation, when the amotmt of platinum
remaining upon the perforated disk becomes small, bubbles of chlorine
commence to rise through the liquid, it is an indication that the current
density is becoming too great. In this case, bringing fresh acid into the
ndghborhood of the platinum black and decreasing the current will
remedy the chlorine evolution.

In concentrating the solution of chlorplatinic acid after it is so pre-
pared, chlorine is passed through it for a short while. This insures free-
dom from platinous compounds in case any have been formed during the
electrolysis.

[Contributions prom the Rbsbarch Laboratory op Physicai^ Chemistry op the

Massachusetts Institute op Technology, No. 20.]

THE THEORY AND PRACTICE OF THE lODOMETRIC DETER-
MINATION OF ARSENIOUS ACID.

By Edward W. Washburn.
Received November 2, 1907.

I. IfUrodticiion. — ^The first application of arsenious acid solution in
titrametiic analysis was made by Gay Lussac,* who used it in chlorimetry
with an indigo indicator. Penot' later improved the method by the use
of strips of starch iodide paper as the indicator and the use of caustic
soda as the solvent for the arsenious acid, instead of hydrochloric acid,

' Gay Lussac, Ann., x8, 18.

' Penot, Dingl. Pol. J., 12% Z34 and zag, 286.

32 BDWARD W. WASHBURN.

which had been used by Gay Lussac. Friedrich Mohr* still further im-
proved the process and brought it into its present form by using standard
iodine solution instead of chlorine water as the titrating agent and adding
the starch indicator directly to the solution. In the earlier editions of
his "Titrirmethode" Mohr directs that the arsenious add solutions be
made up with sodium carbonate and that this substance be present in
excess during the titration. In the later editions, however, he recom-
mends the substitution of ammonium carbonate as the neutralizing
agent, since he finds it to give "a more permanent end-point." Most
of the modem text books on analytical chemistry, however, have adopted
sodium bicarbonate for this purpose and this is the custom among chemists
in this country.

A critical study of this analytical method particularly from the stand-
point of the equilibria involved, seems not to have been made, although
all of the data necessary for this purpose are now available in the Uterature
of physical chemistry. In fact many of the latest editions of the standard
text books on analytical chemistry including Fresenius, Classen, Sutton
and Mohr contain incomplete or misleading statements in regard to the
theory of the method and the precautions to be observed in applying it.
For example to quote from Sutton's "Volumetric Analysis," (1904):
"The principle upon which this method of analysis is based is the fact
that when arsenious acid is brought in contact with iodine in the presence
of water and free alkali it is converted into arsenic acid. The alkali
must be in sufficient quantity to combine with the hydriodic acid set
free, and it is necessary that it should exist in the state of bicarbonate
since monocarbonated alkalies interfere with the color of the bltie iodide of
starch used as an indicator.'* It might be inferred that the normal car-
bonate or even the hydroxide would not be objectionable if the use of
starch as an indicator were dispensed with.

The development of this method seems to have been purely on em-
pirical grotmds and the conditions existing in the solution at the end-
point, particularly the conditions necessary for securing a "permanent
end-point" are not clearly understood. Owing to this rather unsatis-
factory condition of the literature on the subject and in view of the fact
that under proper conditions this method is one of the most accurate
in the field of analytical chemistry, the following treatment of the sub-
ject will be made somewhat detailed. The equilibria involved in the
method will first be considered and the calculation of the proper con-
ditions to be observed at the end-point as well as the methods for securing
these conditions will be given in detaiL The preparation and preserva-
tion of standard solutions and the operations and precautions to be ob-
served in applying the method will then be described.

^ Friedrich Mohr, Lehrbuch der Chemisch-Analytischen Titrirmethode, Ed. i, 1859.

lODOMOTRIC DETBRMINATION OF ARSENIOUS ACID. 33

2. Theory of the Method. — When a solution of iodine in potassium
iodide is added to a solution containing arsenious acid the reaction which
takes place may perhaps best be expressed as follows:

H,As03 + I3- + H,0 = H3ASO, + 2H+ -f 3I- (I)

or in words, arsenious acid reacts with triiodide ion and water to produce
arsenic acid, hydrogen ion and iodide ion (the "free iodine" in a o.i N
iodine solution being practically all in the state of triiodide). This re-
action is reversible and can be made to go completely^ in either direction
according to conditions. The reaction from left to right is made use of
in determining arsenious acid, while that from right to left is the basis
of a method for determining arsenic add. We shall consider here only
the former case and will proceed to determine the conditions necessary
to make the reaction go quantitatively from left to right.

In general an equilibrium can be displaced from left to right by in-
creasing the concentration of the substances on the left-hand side of the
equation, or decreasing the concentration of those on the right, or by com-
bining both. In the present instance it is obvious that the concentration of
the arsenious add cannot be increased since the object of the titration is to
secure its complete oxidation, and for a similar reason the concentration
of the arsenic add cannot be decreased. The concentration of the Ij-ion
must not exceed the first recognizable amount since this determines the
end-point. The concentration of the water is, of course, a constant. We
are therefore reduced to regulating the hydrogen ion or iodide ion con-
centration, or both, in order to effect our purpose. Since considerable
amounts of iodide are introduced with the free iodine, to keep the
latter in solution, and since moreover the presence of iodide ion is neces-
sary to give the proper colored end-point, if starch is used as the indicator, ^
it would be inconvenient to attempt to reduce the iodide ion concentra-
tion to a small value. It is obvious, therefore, that the success of the
whole titration will depend upon the possibility of maintaining the hy-
drogen ion concentration at a sufl&dently small value.

The most obvious method of reducing the hydrogen ion concentration
of a solution is to add an alkali and, since* (H+)-(OH"") = K, by adding
suffident alkali the hydrogen ion concentration can be made as small
as desired, the hydroxyl ion concentration, of course, increasing pro-
portionally. In the present case we are, however, limited to a very
small value of the hydroxyl ion, since iodine reacts in alkaline solution
to form iodide and hypoiodous add and eventually iodate, as expressed
by the following reactions,

* By "completely" or "quantitatively" is meant to such a degree that the amotmt
of the substance to be determined, remaining unchanged shall be negligibly small.

• Sec Section 13, Note 6.

' Expresaons of the form pC) will be used to indicate formula weights of the sub
stance per 1000 ca of solution.

34 EDWARD W. WASHBURN.

OH- + Is"- - HIO + 2l- (2)

and

60H- + 3ls" « lOs" + 81- 4- 3H,0. (3)

It is thus evident that the hydrogen ion concentration must be kept
small enough to allow reaction (i) to proceed quantitatively and yet
large enough so that reactions (2) and (3) shall not take place to an ap-
preciable extent. These two conditions might, of course, be mutually
contradictory, depending upon the values of the equilibrium constants
of the reactions in question.

J. Calculation of the Upper Limit of Hydrogen Ion Concentration, — ^The
equilibrium equation for reaction (i) is

(H^O,)-(H+)»-(I-)»

(H,AsO0-(I,-) "^^ ^^'

The value of this constant is not known very accurately but a value
sufficiently exact for the present purpose can be calculated from the data
of J. R. Roebuck.* From the data in Roebuck's second paper for the
experiments made without sulphuric acid (Table 38), the value K =
3 , io~* for o® has been calculated. From the experiments made in the
presence of sulphuric add (which greatly complicates the calculation)
Luther* has recently computed the value K = 3.3 • io~^ For 25° the
value would be K = 7 • 10"^.

Assuming a desired accuracy of o.ooi per cent, in the titration, which,
as will be shown later, it is possible to approach, the calculation of the
upper limit of the hydrogen ion concentration is made as follows: 100 cc.
of a 0.1 N arsenious add solution and 100 cc. of a 0.1 N iodine solution
are used in the titration, the total volume when the end-point is reached
being 230 cc. The standard iodine solution contains also 0.12 moL of
potassium iodide per liter. The following relations, therefore, exist:
100 cc. of 0.1 N iodine solution give 0.005 ^o\ of triiodide and 0.012 mol.
of potassium. 100 cc. of 0.1 N arsenious add solution give 0.005 mol. of
HjAsOj. Together they produce 0.01 mol. of iodide to which should be
added 0.012 moL, giving 0.022 mol. of iodide present at the end of the
titration. There are also produced 0.005 mol. of arsenic acid (H5As04-f
H2ASO4— ), and 0.01 mol. of hydrogen ion, the latter, however, being neutral-
ized as far as necessary by the presence of some neutralizing agent. These
amounts are present in a volume of 250 cc. Assuming the iodide to be
98 per cent, ionized we have

(I—) = 0.022 • 0.98 • 4 = 8.6 • IO""^
Since only 0.001 per cent, of the arsenious acid is to remain at the end
of the titration, we have

(HjAsO,) = 0.00001 • 0.005 • 4 -■ 2.0 • io~"^

* Roebuck, J. Phys. Chem., 6, 395, and 9, 756.

* Luther, Z. Blektiochem., ij, 289.

lODOMETRIC DBTQRMINATION OP ARSENIOUS ACID. 35

For the concentration of the triiodide we shall assume the amount of
free iodine necessary to give the color of the end-point in a volume of
250 cc. As will be explained later,* this is found by experiment to be
(If) = 2.0 • lO""^

To obtain the value of (HjAsOJ it will be necessary to know the ioniza-
tion constant for arsenic acid. This can be readily calculated from the
conductivity data of Walden' in the usual manner and is found to be
4.8 • io""3.» We have therefore the two equations

(H+) • (H^sO -) ^ . 10-3

(H,As05 ^-4.8 10

and

(HjAsO^") 4- (HjAsOJ = 0.005 • 4^=^ 2.0* io~*,

from which (H^AsOJ can be readily obtained in terms of (H+). Sub-
stituting the foregoing values in the equilibrium equation

(H^sO,) ' (H+)' • (!-)« .

(H,AsO,) • (I.-) ^ '° ' ^^^

and solving for (H+) we obtain the result (H"*") = i.o • io"~* as the upper
limit of the hydrogen ion concentration.*

4. Calculation of the Lower Limit of the Hydrogen Ion Concentration. —
To find the concentration of hydroxyl ion which will produce an error
of 0.001 per cent, due to the formation of iodate according to the reaction

60H- + 3ls~ = 10,- + 81- + 3H,0,
it is necessary to know the equilibrium constant for this reaction. This
constant can be obtained from the following data :

(H.)«.(y(I-)»^^3 . ^^^, ^^^^^^^, ^^^

(H+) • (0H-) = 1.0 . 10-'^ (6)

fj\ ■'"1.3 • 10-3 (Noyes and Seidensticker).* (7)

IMviding equation (5) by equation (6) raised to the sixth power and then
multiplying by equation (7) cubed we get

(lOr) ' (I-)' „fi . ,„^ (o.

which is the constant desired. As before, we have total triiodide used

' See Section 13, Note 6.

* Walden, Z. physik. Chem., a, 49.

* See also Luther, Z. Elektrochem., 13, 297.

* It is worth noting that if an accuracy of only o.i per cent, is required, this value
is still 5 . ID""*.

* Sammet, Z. physik. Chem., 53, 640.

* Noyes and Seidensticker, Z. physik. Chem., 37, 357

and

36 EDWARD W. WASHBURN.

0.005 niol' ^^ ^his o.ooi per cent. = 5.0 • 10"^ mol. will produce
1/3 • 5.0 • 10"^ = 1.67 • 10"^ mol. of lO,"".
We have therefore

(IOs"~) = 4 • 1.67 . io~® = 6.7 . 10"®,
(I-) = 8.6 . IO-^
(I3-) = 2 . 10-7,
(0H-) - X.
Substituting in the equilibriiun equation (7) above and solving for (OH""),
we find (OH"") = io~^, a value which, if exceeded, will cause an error of
0.00 1 per cent, due to the formation of iodate.
The equilibrium constant for reaction (2)

OH- + I3- = HIO + 2l~,
is obtained from the following data :

(H+) • (0H-) = i.o • 10-'*, (9)

^^ - ..3 • .0- . <.o,

Multiplying (8) by (10) and dividing by (9) gives

(HIO)(I-)« , , ,

r^. = 1.3 • 10^ (11)

(I3-) • (0H-)

for the value of this constant. Solving as before for (0H"~) we find (0H~") =
io"5 which is a smaller value than that obtained for reaction (2).

Therefore, since (H"*") • (OH^ ) = i.o • lo""'^, we get for the lower limit of
the hydrogen ion concentration the value (H"*") = 10^ We therefore con-
clude that if an accuracy of o.ooi per cent, is desired in this titration,
the hydrogen ion concentration in the solution when the titration is
finished, must lie between the limits, (H+) = io~^ and (H+)= io~^.
The best value will evidently be the geometrical mean of these two which
is 3 • 10"^. This differs very little from the concentration of hydrogen
ion in pure water so that we may state as the final conclusion of the pre-
ceding calculations, that, at the completion of the titration of an arsenious
acid solution with standard iodine, the solution should be neutrcU.

5. Solutions of Constant Hydrogen Ion Concentration. — A solution will
automatically keep itself at any desired hydrogen ion concentration even
though small quantities of acid or base be added to it, provided it con-
tains something which will remove both hydrogen and hydroxyl ions
when acid or alkali are added to the solution. A solution which con-
tains the salt of a weak acid (or base) together with an excess of the acid
(or base) has this property of automatically maintaining itself at a prac-
tically constant hydrogen ion concentration. If for example acid be
* Sammet, Loc. cit.

lODOMBTRIC DETERMINATION OP ARSENIOUS ACID. 37

added to a solution containing su£5cient amounts of the weak acid HA
and its salt MA in the proper proportions, hydrogen ion will be immediately
removed by the reaction,

H+ + A- = HA,
and in a similar manner alkali will be neutralized according to the re-
action,

HA + OH- = Hfi + A-.
The proper porportions of acid (or base) and salt to use in order to keep
the solution at any desired hydrogen ion concentration are readily calcu-
lable from the ionization constant of the acid (or base). If the mixture
is to be equally efficient in the removal of hydrogen and hydroxyl ions,
it is evident that the concentration of the acid (HA), should be equal to
that of the ionized salt (A" ), i. e.

(HA) = (A-) = r(S) (a)

where f is the degree of ionization and (S) the total molar concentration
of the salt. Combining this relation with the equilibrium equation for
the iqnization of the acid,

(H^) - (A-) ,

(HA) - ^' ^^^

we obtain the result K = (H+).

That is, the ionization constant of the acid should be numerically equal to
the desired hydrogen ion concentration. In case no suitable acid can be
found which exactly fulfils this condition, one is chosen which most
nearly does so. Suppose its constant is

K = n(H+), (c)

then if n is not too large, this acid can be used, provided the ratio of
salt (S), to acid (HA), is changed. By combining equations (a), \b) and
(c), it is evident that this ratio must have the value,

-^=- (d)

(HA) y' ^"^^

In applying these considerations to the problem in hand, since the

solution is to be kept neutral, i. e., (H"^) = (0H~) = lo""^, it is evident

that the add chosen should have an ionization constant not far from

10"^. Of the numerous acids which might be used for this purpose we

shall consider here only the following acids and their salts.

Add. Salt.

NaH^O* Na,HPO.

H^O, NaHCO,

H,BO, NagBO,

Phosphoric Acid, — The ionization constants of the three hydrogens
of this acid are given by the following equations :*

' The values of these constants have been recently determined in this laboratory
by Mr. G. A. Abbott in an investigation which will be published shortly.

38 EDWARD W. WASHBURN.

(H+)(H,PO-)^

(H,PO,) ' '° • :

(H < ) (HPO, )

(H,PO,-) 2.1 lo .

(H+)(PO, )

(HPO, ) ^'^ ^ •

The first hydrogen is a strong add but the salt of this acid, NaH,P04,
acts as a weak acid with the ionization constant 2.10"^ and is therefore
suitable for the purpose in hand. The ionization constant of the third
hydrogen is so small that the ionization of HPO4 will be entirely negligi-
ble. We have, therefore, (S) -Ci,.,hpo,» and (HA) = (H,PO -) = riCNaH,po,
where Cj^^jj^po^ is the total molar concentration of NaH2P04 and /•, its
degree of ionization as a salt. Substituting these values in equation
(d) above we obtain the result

CNatHP04 wyi
CNaHjP04 y

and since /'j can be assumed equal to y, we conclude that the solution
at the end of the titration should contain about two mols of Na,HP04
for every mol of NaH^PO, in order to preserve neutrality.

Calcidation for Carbonic Acid, — ^This acid is in general not adapted
for the purpose of maintaining a constant hydrogen ion concentration
because, being a gas, it escapes gradually from the solution so that its
concentration and consequently that of the hydrogen ion is an uncertain
quantity. Only when the solution is kept saturated with carbon dioxide
at a definite temperature and pressure is the concentration of the carbonic
acid in tbe solution of a definite known quantity. For our present purpose
therefore it is necessary to know what concentration of sodium bicarbonate
must prevail in a solution saturated with carbon dioxide if the concentra-
tion of the hydrogen ion is to be io~^. We have the following ionization
equations for the acid.^

(H+) • (HCO3-)

(H^)>(C03")^

(HCO,-) ^-^^ ^° •

Here also the dissociation of the second hydrogen in a neutral solution is
negligible. In a solution saturated with carbon dioxide at 25^ and i
atmosphere (HjCO,) = 0.0338. Substituting in equation (d) again we
obtain

(S) 3.04 , . , . rs
-— ^===^-^ which gives CNaHco, = 0.12.

O'033o y
Therefore in a 0.12 molar solution of sodium bicarbonate, saturated with
* McCoy, Am. Chem. J., 39, 437.

lODOMHTRIC DETERMINATION OF ARSENIOUS ACID.

carbon dioxide the hydrogen ion concentration will be about lo"'^. The
ratio between acid and salt (1:4) is not so favorable as in the case of the
phosphate. Owing to the relatively small concentration of the add, the
hydrogen ion concentration will be rather sensitive to the escape of CO,
from the solution, since to offset this loss, carbonic acid is being con-
tinually produced at the expense of the hydrogen ions according to the
reaction,

H+ + HCO,- = H,CO, - H,0 + CO3.

Boric Acid, — Sammet* by the measurement of the E. M. F. of a hy-
drogen electrode, has determined the hydrogen ion concentration in a
solution saturated with respect to both borax and boric acid and found
it to be (H+) = 6 • io-7.

We have therefore three combinations which theoretically should be
equally suitable for keeping the solution neutral during the titration.

6. Test of the Conclusions, — ^The conclusions of the foregoing con-
siderations were tested by standardizing an approximately 0.1 normal
iodine solution against a 0.1 normal arsenious add solution. About
100 cc. of the iodine solution were used in each case. The solutions were
weighed instead of measured, the end-point being reached by using a
very dilute arsenious acid solution, the procedure being that described
in the following sections. It will be noticed that the conclusions of the
foregoing calculations are completely justified. These results will be
discussed later when the procedure has been described.

Value of 100 grams of iodine solu-
tion in terms of Aafir

NaHCO, +
HjCO,

Na«BOs +
H,BO,

NasHP04 +
NaH,P04

KI alone.

0.48737

0.48733

0.48735

—

0.48736

0.48733

0.48736

—

0.48738

—

0.48736

—

0.48735

"^—

—

0.000078

0.000073

Blanks in same unit 0.000078

Mean, 0.487354.

a. d., 0.000012 — 0.0025 per cent.
A. D., 0.0000040 «■ 0.00082 per cent.
Max. d., 0.000056 ■- o.oi per cent.

7. Titration by Weight. — For the most accurate work in titrametric
analysis it is necessary to weigh all the solutions used instead of measur-
ing them, and the solutions are conveniently made up on the weight
standard system, that is, so that 1000 grams of a normal solution contain
one equivalent weight of the substance. Indeed even for work in which
only 0.1 per cent, acciu'acy is required, this method possesses several
advantages over the volumetric method and need consume very littk

' Loc. cit.

BDWARD W. WASHBURN.

more time. Its principal advantages are: (i) since graduated vessels
are not used, no time need be spent in calibration and no errors are in-
troduced from this source; (2) frequent cleaning of the burettes is not
necessary as drops clinging to the sides of the burette do not affect the
result; (3) the result is independent of the temperature of the solutions;

(4) no time is consumed in allowing the burette to
drain, in fact the weight-burette can be weighed in
the time usually consumed in drainage.

A convenient form of a weight-burette, de-
vised by Mr. C. A. Kraus, of this laboratory, is
shown in the figure. The stopcock and glass cap
for the tip are very carefully ground so as to in-
sure a tight joint. The opening for filling the
burette is placed at the side so that none of the
solution can come in contact with the stopper
which is hollow and open at the bottom. In order
to permit the entrance of air during the titration,
the stopper and the side of the socket are pro-
vided with two small holes which can be made to
coincide by turning the stopper in its socket. In
weighing, the burette is suspended from the bal-
ance arm by the hook, a similar burette suspended
from the other arm serving as a counterpoise.
For most work a suitable weight-burette can also
be readily made from a separatory funnel by cut-
ting off the stem, drawing it down to a point,
and grinding on a glass cap. Previous to filling
the burette, a small quantity of the solution is
shaken in it and then allowed to run out. This
insures the saturation of the air in the burette
by the vapor of the solution.
8, Preparation of Standard Iodine SoltUions. — ^To prepare a liter of ap-
proximately o.i N iodine solution, 12.7 grams of resublimed iodine are
weighed into a small beaker, about 20 grams of pure potassium iodide
added, and the whole covered with water and allowed to stand with oc-
casional stirring until solution is complete. The solution is then filtered
through an asbestos filter into the stock bottle and sufficient water (best
distilled) added to make the volume about i liter. The statement is
frequently made that iodine solutions cannot be kept very long unchanged
and require frequent standardization. In the experience of the author,
however, an iodine solution properly prepared and used is one of the most
stable of standard solutions. The chief causes of change in titer are
losses from evaporation, from particles of dust which may get into the

lODOMETRIC DETERMINATION OP ARSENIOUS ACID. 4 1

solution, and from impurities in the water used in preparing the solution.
With proper precautions these losses are, however, inappreciable. The
neck of the bottle should be kept covered with a beaker to exclude dust,
and when the solution is shaken to mix it, care should be taken to avoid
wetting the stopper. The solution should always be removed with a
pipette, never by pouring. It is advisable to allow a newly-prepared
solution to stand a few days previous to standardizing and the stock
bottle should be kept in a dark place. The following results were ob-
tained for the standardization of an iodine solution at an interval of
two months, the solution being in frequent use meanwhile.

November 5th, i gram I solution » 0.0048504 gram Aafi^,
January 8th, z gram I solution » 0.0048505 gram As,0,.

9. Preparation of Tenth-Normal Arsenious Acid Solution, — ^The arsenious
oxide is purified by recrystallization from hot water and by sublimation.
After drying in a vacuum over sulphuric acid, about 5 grams of the crystals
are placed in a small weighing tube and about 4.95 grams are accurately
weighed out into a glass stoppered liter flask which has been previously
cleaned, dried and weighed to centigrams. Ten to 12 grams of pure
caustic soda are then dissolved in about 30 cc. of water, which has been
freshly distilled from an alkaline permanganate solution to insure absence
of organic matter and dissolved oxygen. After filtering through asbestos
it is added to the flask which is then allowed to stand until solution is
complete. The chief impurity which it is necessary to guard against
in the caustic soda is iron. The grade known as "purified by alcohol"
is usually satisfactory for this purpose as it gives a very small blank.

When solution of the arsenious acid is complete (about half an hour),
100 cc. of water are added and a delivery tube is inserted in the flask
below the level of the liquid. The delivery tube is connected with a
generator supplying carbon dioxide and the gas is allowed to bubble slowly
through the solution imtil saturation is reached. After removing and
washing the delivery tube the solution is diluted to nearly a liter with the
freshly distilled water. The flask is then placed on the balance pan and
water added from a dropper until the weight of the solution is 206.73
times the weight of the AsjOg used. After thorough mixing, the solu-
tion is ready for use and will preserve its titer almost indefinitely. One
thousand grams of the solution correspond to 0.1 equivalent of arsenious
acid.

Instead of the above method of preparing the solution with sodium
bicarbonate, the following method in which sodium phosphate is used
gives fully as satisfactory results and possesses certain advantages as will
be pointed out later. In this method the arsenious acid is weighed out
as before but is dissolved by adding a strong solution of sodium hydroxide
of known strength, the amount added containing 12 grams of NaQH*

42 EDWARD W. WASHBURN.

When solution of the arsenious acid is complete a solution of pure phos-
phoric add containing 0.15 mol of H3PO4 is added and the whole made up
to the final weight as before.

10. Preparation of the Starch Solution. — This solution, prepared as
directed in Treadwell's "Analytical Chemistry,"* and preserved in small
bottles, is eminently satisfactory. A large supply should be prepared
at one time, as solutions prepared at different times are liable to give
somewhat different colored end-points. When properly prepared, the
first color obtained with iodine should be pink. Starch solution which
gives a blue or greenish-blue color as the first shade should be rejected.
The ** soluble starch" of commerce is not to be recommended for the
finest work. Two cc. of the starch solution are used in each titration.

11. Standardization of the Iodine Solution against the Arsemous Acid
SoltUion. — About 100 cc. of the arsenious acid solution are removed with
a pipette and transferred to a carefully tared Erlenmeyer flask of 500 cc.
capacity provided with a rubber (or better, glass) stopper.^ The weight
of the solution to the nearest milligram is then determined, and the flask
is placed upon a large sheet of white paper in a good light (preferably
north). The weight pipette is then filled with the iodine solution as
previously described and is weighed to the nearest milligram. The
pipette is placed in its support and the flask containing the arsenious
acid solution is held in the hand so that the tip of the pipette touches
the inside of the neck of the flask. About 100 cc. of the iodine solution
are added and the flask is stoppered lightly and rotated until the yellow
color of the iodine disappears. If the arsenious acid solution was prepared
with bicarbonate the solution will effervesce, and during the effervescence
the flask should be held in a slanting position so that the walls of the
flask and not the stopx)er receive the fine spray thrown up by the effer-
vescence. The titration is continued slowly until finally one drop of the
iodine solution produces a permanent yellow color. The flask is then
stoppered and allowed to stand while the weight burette is weighed again.

The titration is completed by using a dilute arsenious acid solution.
This is prepared by diluting 25 grams of the 0.1 normal solution to a
volume of a liter. This 0.0025 N arsenious acid solution is kept in a
bottle with a burette permanently attached to it. A titration is finished
by adding this dilute solution drop by drop until the yellow color has al-
most disappeared. Two cc. of the starch solution are then added and the
addition of the arsenious acid solution continued until the blue color

* Analytical Chemistry, Treadwell-Hall, Vol. 2, p. 513 (1904).

' It is convenient to have a number of these flasks ready for use. The lightest
flask in the set is used as a partial counterpoise, and the weights of the others are marked
on their stoppers. The flasks should be kept in the balance room so as to be under
the same conditions of temperature and humidity as the balance.

lODOMETRIC D^ERMINATION OP ARSENIOUS ACID. 43

gives way to a rose color which just matches the color of flask No. 2, of
a set of color standards whose preparation will be described below. Be-
fore adding the last two or three drops it is best to wait ten or fifteen
minutes to insure the completion of the reaction and the attainment of
equilibrium. The flask should be protected from the light during this
interval. This end-point when reached {if the flask is kept stoppered and
protected from the light) is permanent both in shade and intensity of color
*or several weeks, and the end- point is sensitive to two drops of the dilute
arsenious acid solution , that is, to less than 0.002 per cetU,^

12. Preparation of Permanent Color Standards. — A flask containing a
solution, the titration of which has just been completed, is made just
colorless by the addition of a few drops of the dilute arsenious add solu-
tioiL. It is then placed under a burette containing a 0.005 N iodine
solution and iodine added until a faint permanent yellowish-pink color is
produced. In a second flask a solution is made up which exactly matches
this color. This is color standard No. i. It can be made by combining
solutions of ferric chloride, copper nitrate and cobaltous nitrate in the
proper proportions. Another drop of the dilute iodine solution is then
added and a second color standard made up to match this color, which
is more of a pink shade. A third drop of iodine solution is then added
and the third color standard made to match this color. In titrating a
solution the end-point is always taken as the color corresponding to the
middle flask in the set of color standards. This shade of pink,, which is
close to what is known as the '* sensitive tint" in polarimetry, is much
more sensitive to small changes than are the blue tints obtained by the
further addition of iodine.

J J. Notes on the Process. — (t) In titrations with the use of sodium bicar-
bonate as the neutralizing agent it is absolutely essential that the solution
be kept saturated with carbon dioxide as the calculations indicate;
consequently the titration should never be made in a beaker but always
in a stoppered flask.' This is a precaution which is not mentioned in the
text books^ but its importance is readily shown by the following experi-
ment. The titration as described above was made in a beaker. When
the end-point was reached the solution was stirred for thirty seconds. At
the end of this time the solution had become completely colorless. That
this was not due to the slowness of the reaction was made evident by

^ It is best to avoid unnecessary exposure to strong light, especially direct sun-
light, since light catalyzes the reaction

O, + 4H+ + 4I— - 2I, + 2H,0
which may take place to a slight extent, even in neutral solution, in case dissolved
oxygen is present. See Plotnikow, Z. physik. Chem., 58, 214.

' The author has found nothing superior to Erlenmeyer flasks for ail titrametric
analyses. They are much more reliable and convenient than the beaker and stirring-
rod so frequently used.

44 BDWARD W. WASHBURN.

adding a few drops of hydrochloric add to the solution. This caused the
evolution of carbon dioxide and the color immediately returned to fade
away again on further stirring and consequent escape of carbon dioxide.
Unless sufficient free carbon dioxide be present the concentration of
hydroxyl ion becomes sufficiently large to cause the formation of an
appreciable amount of hypoiodous acid and a consequent fading of the
end-point.

Some authors recommend cooling the solution to o° with cracked ice
as giving a more delicate end-point. This is not to be recommended,
however, as the reaction is too slow at this temperature. The observed
increase in the delicacy of the end-point in the cold solution is simply
due to the greater solubility of carbon dioxide at this temperature, so
that a solution open to the air does not become alkaline so rapidly as at a
higher temperature. At room temperature and in a closed flask the delicacy
and permanency of the end-point are all that could he desired.

(2) The use of the sodium phosphate has the advantage that the acid
is not a gas and its concentration in the solution can be controlled by the
operator. Since there is no effervescence when it is used, mechanical
loss from this source is impossible. It is to be recommended in preference
to the bicarbonate.

(3) The use of the borax-boric acid mixture, while it gives accurate
results, is not convenient and has no advantage over the other two. The
use of ammonium carbonate as directed by Mohr in his last edition, is
still less to be recommended, since owing to the volatility of both the
acid and base, the concentration of hydrogen ion is still less imder the
control of the operator.

(4) In titrating an unknown solution of arsenious acid, the solution,

if alkaline, is made neutral with hydrochloric acid; if acid, it is neutralized

with sodium hydroxide, using phenolphthalein as the indicator. The

neutralizing agent, bicarbonate or phosphate, is then added from time

to time during the titration and when the titration is nearly completed

more of the neutralizing agent is added until the total quantity added

amounts to about 5 grams of NaHCOj or u grams of Na,HP04.i2H,0

for every 100 cc. of o.i N iodine solution required in the titration. The

volume at the end of the titration should be about 250 cc. for every 100

cc. of the 0.1 N iodine solution used. Under these conditions the add

produced by the reaction is enough to saturate the solution with carbon

dioxide or to produce NaHjPO^ in the proper amotmt to give the molar

Na«HPO
ratio XT ^ pQ^ = 2, at the end-point. The Na2HP04 should be added,

in solution, from a burette.

(5) For the determination of small quantities of arsenious acid a 0.01
N iodine solution is used and in this case it is necessary to add the proper

lODOMETRIC DETERMINATION OF ARSENIOUS ACID. 45

amount of hydrochloric acid to insure saturation with carbon dioxide
or to produce the proper amount of NaH2P04, since the add produced
by the reaction is not suflScient. The amount of hydrochloric acid to
use in the case of the phosphate has to be obtained by calculation for
each case if the highest accuracy is required, so that the use of the
bicarbonate is more convenient in this instance. Potassium iodide must
also be added to give the proper color for the end-point.

(6) The necessity of the presence of sufficient iodide in the solution
to develop the proper color with the starch indicator has been emphasized
by Treadwell. If a dilute iodine solution be added to water containing
starch, a considerable amount of solution is required before the blue color
is produced. The presence of an iodide in the solution causes the color
to appear when only a small amount of the iodine has been added. Other
salts have a similar effect in developing the color. In order to com-
pare the effect of different salts, the following experiment was made.
To a flask containing 250 cc. of water, 3 grams of potassium iodide, and
2 cc. of starch solution, the dilute iodine solution was added until the
end-point color was reached. A set of flasks containing 250 cc of 0.1
molal solutions of the substances named below was then prepared and
after the addition of 2 cc. of starch solution, the dilute iodine solution
was added to each until the same depth of color was produced as in the
first flask. The amounts of iodine solution required in each case is shown
in the table.

o . I molal solution of cc. of I aolutlon used.

KI 5.5

Ka 17.0

Naa 17.0

MgSO« 16.0

Ba(NO,), 13.0

(NHJaSO* 14.0

(Pure water) 45.0

The color of the end-point in the case of the iodide is of a different
shade than the others and is probably due to a different effect. It is
usually attributed to the formation of a compound, the statement being
made that its presence is essential to the production of the "blue com-
pound" with starch. That this may not be the whole explanation is
shown by the effect of the other salts.

14. Determination of Blanks. — In making a blank the solution is made
up to correspond exactly to the solution to be titrated but with the omis-
sion of the arsenious acid. In addition, it is necessary to add pure potas-
sium iodide (3 grams per 250 cc.) for the reason mentioned in the pre-
ceding note. Pure hydrochloric acid is used, if necessary, to insure
saturation with CO, or to produce the proper amount of NaH2P04. The
blanks are made by using an 0.005 N iodine solution whose ratio to the

46 GREGORY P. BAXTER AND FRANCIS N. BRINK.

dilute arsenious acid solution has been determined. With pure materials
the blanks, which are purely volume corrections, should be very small, never
exceeding a value equivalent to o.oooi gram AsjOg. The author has ob-
tained blanks as low as 0.00004 gram AsjO,, which in an analysis requiring
100 cc. of 0.1 N iodine solution amoimts to only 0.008 per cent. The
blank is always subtracted in calculating an analysis.

13. Applications of the Method. — It is evident from the preceding calcula-
tions and experiments that under the proper conditions iodine can be
quantitatively reduced to iodide by arsenious acid and that a definite,
permanent and exceedingly delicate end-point is obtained. The accuracy
and definiteness of the end-point makes this method a valuable and con-
venient one for determining the atomic ratio between arsenious acid and
iodine since the materials necessary for the determination can be readily
purified and the method is free from any sources of error due to side
reactions, adsorption, presence of water, etc., which have to be corrected
for in many precipitation methods in atomic weight work. The accurate
knowledge of this ratio is important to the analytical chemist since
arsenious acid is the most convenient and accurate standard for iodimetry.

Boston, November i. 1907.

[Contribution prom thb Chemical Laboratory op Harvard Collbgb.]

THE SPECIFIC GRAVITIES OF THE IODIDES OF SODIUM, POTAS-
SIUM, RUBIDIUM, CAESIUM, CALCIUM, STRON-
TIUM AND BARIUM.

By Gregory Paul Baxtbr and Francis Newton Brink.

Received November 7, 1907.

Attention has been recently called by various authors to large dis-
crepancies and inaccuracies in the present tables of specific gravities.^
If for no other reason, the interest attached to molecular volume* makes
the accurate knowledge of specific gravities of considerable importance.
Since the specific gravities of the chlorides and bromides of the alkali
and alkaline earth metals have already been determined with care,* the
present research is concerned with the iodides of these elements. Lithium
iodide was not investigated because its density has recently been found
by one of us.*

Among the chief sources of error in specific gravity determinations
may be mentioned imperfect dr)dng of the substances and inclusion of
mother-liquor by crystals. Both these difficulties were avoided in the

* Proc. Am. Acad., 31, 163; Am. Chem. J., 31, 220, 229, 558; Trans. Chem. Soc.,

9i| 56.

' See especially Richards, Proc. Am. Acad., 37, 3, 399; 38, 293; 39, 581.

* Landolt-Bornstein-Meyerhoflfer.

* Baxter, Am. Chem. J., 31, 558.

IODIDES OF SODIUM, POTASSIUM, RUBIDIUM, CAESIUM, ETC. 47

present research by .fusion of the salts, in an atmosphere of nitrogen to
prevent decomposition of the iodides.

The purification of the salts from saline impurities and silica is usually
of less importance than the elimination of water, since the densities of
probable impurities in each case were not far from those of the substances
under investigation. Nevertheless, care was taken to remove at least
the major part of these impurities.

Of course the usual precautions were taken in setting the pycnometers
and in dislodging and boiling out air entangled in the solid material.

The procedure was as follows: The pycnometer employed was de-
vised some years ago for the determination of the specific gravity of very
hygroscopic substances,^ and is a modification of a pycnometer devised
by T. W. Richards for the determination of the specific gravity of solids.
A weighing bottle was provided with two glass stoppers, one of which
was of ordinary shape and was used during the weighing of the sub-
stance. Into the other were sealed two capillary tubes which served
to fill the bottle with liquid. The weighing bottle and the pycnometer
stopper were both made of thick glass in order to avoid distortion when
the stopper was inserted. A metallic carriage was used in all the weigh-
ings of the bottle.

The salt, contained in a platinum boat, was fused in a current of dry
nitrogen in a hard glass tube connected by a grotmd joint with a bottling
apparatus by means of which the boat, after being heated, could be trans-
ferred to the weighing bottle without exposure to moist air.^ Heat was
applied gently at first, till the greater portion of the water had been ex-
pelled, then the temperature was increased until fusion had taken place
and the fused salt was limpid and free from bubbles of gas. The boat
was then allowed to cool in a current of nitrogen and, after the nitrogen
had been displaced by dry air, the boat was transferred to the weighing
bottle and weighed.

The nitrogen was made by passing air through concentrated ammonia
and then over hot copper gauze, and was freed from excess of ammonia,
as well as from carbon dioxide and moisture, in the usual way by means
of dilute sulphuric acid, potassium hydroxide and concentrated sul-
phuric acid. In the experiments with barium, strontium and calcium
iodides the last traces of moisture were removed from the nitrogen by phos-
phorus pentoxide. Air was purified and dried with the same reagents as in
the case of nitrogen.

After the weighing of the salt, the ordinary stopper was removed,
enough toluene to cover the boat and salt was quickly poured into the
bottle, and the pycnometer stopper, which had been weighed with a small

' Baxter and Hines, Am. Chem. J., 31, 220 (1904).

* Richards and Parker^ Proc. Am. Acad., 32, 59 (1897)

48 GREGORY P. BAXTER AND FRANCIS N. BRINK.

quantity of sirupy phosphoric acid to make the joint tight, was inserted.
The pycnometer was placed in a vacuum desiccator, which was then ex-
hausted, and the toluene was allowed to boil gently for some time with
frequent jarring to expel the air contained in the crevices of the salt.
By means of the capillary tubes the bottle was completely filled with
toluene, and, while the pycnometer was immersed as far as possible in a
water thermostat at 25®, the toluene was adjusted to a mark etched
upon one of the capillaries. The weight of the system was then deter-
mined, after the pycnometer had been wiped with a clean, slightly moist
cloth and had been allowed to stand in the balance case for a few moments.
Prolonged standing in the balance case produced no difference in weight.
After a second adjustment of the toluene the system was again weighed.
In every case the two weights agreed within a very few tenths of a milli-
gram. The weight of the pycnometer containing the empty boat and
filled with toluene was determined six times. From the average of these
six weights,* the weight of the salt, and the weight of the system including
the salt and filled with toluene, was calculated the specific gravity of the
salt.

The toluene was dried over metallic sodium and distilled, the first and
last portions being discarded. Its specific gravity was determined by
means of an Ostwald pycnometer, the capillaries of which were provided
with grotmd glass caps.

Weight Weight

Weight of of pycnometer Weight of of pycnometer Weight of

pycnometer. filled with water. water, filled with toluene, toluene.

Grama. Grams. Grams. Grams. Grams.

10.6915 19.4087 8.7172 18.2231 7 5316

19.4092 8.7177 X8.2237 75322

zo. 591 1 (without caps) 19.3091 8.7180 18.1237 7.5326

.... .... 18.1240 7 5329

• • .

Average, 8.7176 Average, 7.5323

Volume of water 8 . 7522 cc.

Weight of toluene in vacuum 7 . 5418 grams

Density of toltiene 25 V4* 0.8617

The specific gravity of the toluene was also determined with the
pycnometer used in the experiments with the salts.

The close agreement of the results for the density of toluene by the
two methods shows that the special pycnometer was yielding satisfactory
results.

The weights were carefully standardized to tenths of a milligram, and
the corrections of the thermometer at 0° and 32.38° were determined.
In the foregoing and in the following tables vacuum corrections were ap-

^ These weights are given below. Since the platinum boat gradually lost in
weight, corrections for this loss and for the toluene displaced by the platinum which
disappeared, were applied.

IODIDES OP SODIUM, POTASSIUM, RUBIDIUM, CAESIUM, ETC. 49

plied as follows: toluene, +0.00126; Nal, +0.00019; KI, +0.00024;
Rbl, +0.00021; Csl, +0.00012; Calj, +0.00016; Srij, +0.00012;
Balj, + 0.00009.

Weight of preiioin. Weight of pycaomefcer and Weight of oyenometer and
eter and boat. boat, filled with water. boat, fiUea with toluene.

Grams. Grams. Grams.

22.5227 33-3865 31-9073

33-3865 31-9066

31.9068
31.9066
31.9069
31.9070

• m

• • •

• •

• ••• ••••

Average, 33 3865 Average, 31.9069

Weight of water 10. 8638 grams

Volume of water 10. 9072 cc.

Weight of toluene in vacuttm 9 . 3960 grams

Density of toluene 25V4® 0.8615

Average density of toluene by the two methods o. 8616

By evaporating toluene which had been in contact with the fused salts
it was shown that all the salts examined are insoluble in toluene.

All the substances were prepared by acting on the carbonates or hy-
droxides of the metals with pure hydriodic acid and crystallizing the
iodides, in most cases twice. This hydriodic acid was made from iodine
which had been once distilled from aqueous potassium iodide and once
reduced to hydriodic acid with hydrogen sulphide and set free by dis-
tillmg the hydriodic add with pure potassium permanganate, with inter-
mediate boiling of the hydriodic acid for some time to remove cyanogen.*
The final product was thus twice distilled from an iodide, the iodide in
the second case being already of considerable purity. Hence chlorine
and bromine must have been eliminated.^ The iodine was then once
distilled from pure water, and again reduced with hydrogen sulphide.
The resulting hydriodic acid was filtered to remove sulphur and finally
distilled shortly before use, with a quartz condenser. After distillation
the acid was always colored brown with free iodine, and the iodides made
from it, even after crystallization, showed traces of free iodine. During
the fusion of the salts, however, this iodine must have been expelled and
any iodates must have been decomposed. The hydriodic acid was free
from sulphate.

The salts were all mixed with pure ammonium iodide before fusion,
in order to prevent the production of basicity. In the case of the alkalies
a very small quantity of ammonium iodide was sufficient to produce the
desired effect, but in the case of the alkaline earths even a considerable
amount did not completely prevent basicity.

> Richards and Singer, Am. Chem. J., 2% 205.
' Baxter, Proc. Am. Acad., 40, 421 (1904).

.'ION /v/e^y

50 G^GORY P. 3&T1^ -AND FRANCIS N. BRINK.

Sodium lodideS^oiLttffisfl^ fffi|ium oarbonate was three times cr3rstal-
lized from aqueous soltlttuu i» u platinum dish. The carbonate was next
dissolved in a slight excess of hydriodic acid and the solution of sodium
iodide was evaporated to crystallization in a glass dish. The crystals
were freed from mother-liquor by whirling in a platinum centrifugal
machine.^ In many cases the fused material, when dissolved in water
and tested with phenolphthalein, gave a slight alkaline reaction. In no
case, however, did this alkalinity, which was determined by titration
against hundredth-normal acid, correspond to more than five hundredths
of a milligram of sodium oxide, hence no measurable effect could have
been exerted upon the result.

Weight of salt Weight of displaced Density

in vacuum. toluene in vacuum. of Nal.

Grams. Grams. 25^/4°.

2.9209 0.6879 3-658

1.4635 0.3436 3670

3.I4IO 0.7376 3669

2.5324 0-5953 3 665

4.8574 I. 1425 3663

Average, 3.665*

Potassium Iodide. — ^Acid potassium carbonate was recrystallized, once
in glass and twice in platinum. The final product was then essentially
free from sodium. The pure bicarbonate was converted into iodide,
as in the case of sodium. The fused material when dissolved in water
was neutral to phenolphthalein in every case.

Weight of salt Weight of displaced Density

in vacuum. toluene in vacuum. of KI.

Grams. Gram. 33^/4°.

3.2078 0.8859 3.120

3.4772 0.9660 3.IOI

32367 0.8948 3. 117

2.0803 0.5740 3- 123

3.2059 0.8861 3-ZI7

3.1652 0.8759 3- "4

3-5595 0.9862 3-"o

Average, 3 •"5*

^Richards, This Journal, 27, no (1905). Baxter and Coffin, Ibid., 28, 1582
(1906).

' Previous determinations of the density of this salt are as follows:

Pilhol, Ann. chim. phys. [3], 21, 415 (1847) 3.450

Favre and Valson, Compt. rend., 77, 579 (1873) 18®, 3.654

' Previous determinations of the density of this salt are as follows:

Boullay, Ann. chim. phys. [2], 43, 266 (1830) 3-078 and 3 . 104

Karsten, Schw. J., 65, 394 (1832) 2 .908

Playfair and Joule, Man. Chem. Soc. 2, 401 (1845) 13**, 3.048 and 3.070

Filhol, Ann. chim. phys. [3], 21, 415 (1847) 3056

SchiflF, Ann., xo8, 21 (1858) 2 .850

lODIDBS OF SODIUM, POTASSIUM, RUBIDIUM, CAESIUM, ETC. 5 1

Rubidium Iodide. — Rubidium material was purified by five crystal-
lizations as dichloriodide from dilute hydrochloric add solution. The
dichloriodide was dried and converted to chloride by heating to 150°.
The chloride was changed to sulphate with sulphuric acid, the sulphate to
hydroxide with barium hydroxide, and the excess of barium hydroxide
was removed by carbon dioxide. The rubidium hydroxide was neutralized
with an excess of hydriodic add and the iodide crystallized. The fused
salt, when dissolved in water, was in every case neutral to phenolphthalein.

Weight of fialt Weight of displaced Density oi

in Tacnnm. toluene in vacuum. Rbl.

Grama. Gram. as^/4°.

3.0856 O.774X 3-434

34762 0.8719 3.435

34942 0.8735 3.447

3.2558 0.8156 3.439

3.3482 0.8398 3.435

Average, 3-438*

Caesium Iodide. — Caesium iodide was made from caesium dichloriodide
which had been five times recry^tallized.' The dichloriodide was con-
verted to iodide exactly as in the case of rubidium. The fused salt when
dissolved in water was in both experiments neutral to phenolphthalein.

Weight of salt Weight of displaced Density

in Tacnum. toluene in vacuum. of Csl.

Grams. Gram. ^s^Ia'*,

4-0805 0.7795 4-510

4-2525 0.8124 4-510

Average, 4.510*
Calcium Iodide. — Marble was dissolved in nitric acid and the solution

Buignet, Jahresb., 14, 15 2 .970

Schroder, Ann., 192, 298 (1878) 3-077, 3.081, and 3.083

Spring, Ber., x6, 2724 (1883):

fused* 20®, 3.012

fused, subjected to 20,000 atmospheres pressure 22 ^ '3. no

fused, twice subjected to pressure 20^, 3 . 1 12

Buchanan, Am. J. Sci. [4], 21, 25 (1906), by displacement of mother-
liquor 24. 3°, 3.043

^ Previous determinations of the density of this salt are as foUows:

Clarke, Am. J. Sci. [3], 13, 293 (1877) 3.023

Sctterberg, Oefvers. Stockh. Akad. F6rh., 6, 23 (1882) 3.567

Buchanan, Am. J. Sd. [4], 21, 25 (1906), by displacement of
mother-liquor 24.3°, 3.428

* Wells, Am. J. Sd. [3], 43, 17 (1892).

'Previous determinations of the density of this salt are as follows:

Setterberg, Oefvers. Stockh. Akad. F6rh., 6, 23 (1882) 4-537

B€ketoff, Landolt-Bdmstein-Meyerhoffer 4 . 523

Buchanan, Am. J. Sd. [4], ax, 25 (1906), by displacement of
mother-liquor 22 .8^ 4.508

52 GREGORY P. BAXTER AND FRANCIS N. BRINK.

was heated with an excess of lime. The filtered solution was acidified
and evaporated to crystallization and the product was three times re-
crystallized.^ Calcium carbonate was precipitated from the nitrate by
means of ammonium carbonate and was thoroughly washed by decanta-
tion. The carbonate was then dissolved in hydriodic acid and the iodide
crystallized. The salt when fused alone in a current of nitrogen, after
solution in water is strongly basic. This basicity was partially remedied
by mixing the calcium iodide before fusion with a considerable quantity
of ammonium iodide, and keeping the temperature below the sublimation
point of ammonium iodide until practically all the water was expelled.
The salt was then heated to fusion long enough to eliminate all of the
ammonium iodide. The alkalinity of each sample is indicated in terms
of calcium oxide in the third column of the table. The density of calcium
oxide as given in Landolt-Bdmstein-MeyerhoflFer is 3.3; hence, the follow-
ing determinations could not have been appreciably affected by this
impurity.

Weight of salt Weight of displaced Density of

in vacuum. toluene in vacuum. Percent. ^"^^l-

Grams. Gram. ofCaO 35°/4 .

2.1287 0.4636 0.09 3*956

2.1919 0.4777 0-23 3-953

2.5370 0.5523 0.05 3958

Average, 3-956"

Stroniium Iodide. — A solution of strontium chloride was heated first
with hydrogen sulphide and then, after filtration, with strontium hy-
droxide and sulphate.* From the filtered solution the strontium was
precipitated as carbonate with ammonium carbonate, the carbonate was
washed and repredpitated as carbonate. The product, after being
washed, was dissolved in hydriodic acid and the iodide was cr)rstallized.

height of salt
in vacuum.
Grams.

Weight of displaced

toluene in vacuum .

Gram.

Per cent.
ofSrO.

Density of
Srii.

1.9305

0.3657

O.IO.

4-549

2.2813

0.4323

0.03

4-547

I. 8901

0.3573

0.30

4-558

2.2628

0.4284

4-551

2 . 1664

0.4105

0.00

4.547

2 . 7070

0.5135

O.OI

4.542

2.6167

0.4199

0.16

4.549

Average. 4. 549*

* Richards, This Journal, 24, 374 (1902).

' Only one previous determination of the density of this salt exists:

Ruff and Plato, Ber., 35, 3612 (1902) 20°, 4.9

■ Richards, Proc. Am. Acad., 30, 375 (1894).

* Bodeker found the value 4.415 at 10^. Die Beziehtmg zwischen Dichte tmd
Zusammensetztmg bei festen tmd liquiden Stoffen. Leipzig, i860.

ON THE OXIDATION OF HYDRAZINE. 53

In spite of the addition of ammonium iodide to the salt before fusion,
the fused salt was slightly basic in all but one experiment. Since the
density of strontium oxide is 4.6,* the basicity could not have affected the
results.

Barium Iodide. — Barium nitrate was five times crystalUzed and con-
verted into carbonate by precipitation with ammonium carbonate. Prom
the carbonate the iodide was formed as in the case of calcium and
strontium. Here also ammonium iodide failed to prevent basicity com-
pletely, but as in the cases of strontium and calcium, the error is too
small to consider, since the density of barium oxide is about 5.3.*

Weight of salt Weight of displaced Density of

in vacuum. toluene in vacunm. Percent. Bal*.

Grams. Gram. of BaO. ^5^14*^'

4.8746 0.8110 o.io 5- 179

3.9124 0.6551 0.18 5- 146

5.0750 0.8525 0.21 5- 129

4.3577 0.7292 0.07 5.149

3.9504 0.6615 014 5 146

4.3432 0.7270 0.12 5- 147

50332 0.8414 0.07 5- 154

Average, 5. 150*

It is to be noted that the values obtained by earlier experimenters
are in most cases lower than ours, owing probably to the dangers
mentioned at the beginning of this paper.

As a result of this investigation the densities of the following salts at
25° referred to water at 4® were found to be :

Sodium iodide 3 665

Potassium iodide 3 • 1 15

Rubidium iodide 3 . 438

Cacsinm iodide 4- 510

Calcitun iodide 3-956

Strontium iodide 4 . 549

Barium iodide 5 . 150

Cambudgb, Mass.,

November 5, 1907.

ON THE OXIDATION OF HYDRAZINE. U.

By a. W. Brownb akd p. P. Shbttbrly.
Received November 6, 1907.

Curtius and Schulz* have shown that alcoholic solutions of hydrazine
hydrate and iodine react quantitatively in accordance with the equation

5N^,.H,0 + 41 = 4N2H,.HI + sHfi + Nj,

' Landolt-Bdmstein-Meyerhoffer.

'Filhol obtained the value 4.917. Ami. chim. phys. I3], 21, 415 (1847).

* J. pr. Chem. [2], 42, 521-49 (1890).

54 A. W. BROWNE AND F. F. SHETTERLY.

and that this reaction may be used as the basis of a method for deter-
mining the strength of solutions of hydrazine hydrate.

Rimini* found that the reaction between hydrazine sulphate and potas-
sium iodate could be expressed by the equation

5N2H,.H,SO, + 4KIO3 = 5N2 + I2H,0 + 2K,S0, + 3H2SO, + 4L

For the determination of hydrazine he recommended the following pro-
cedure : A weighed sample of the material to be examined is treated with
a measured excess of potassium iodate. After the solution has been
heated (in order to expel the liberated iodine) and then cooled, it is acidified
with dilute sulphuric acid, and the excess of potassium iodate is deter-
mined in the usual way with a standard solution of sodium thiosulphate.
Rimini stated that bromates react with hydrazine, when heated, but
that chlorates do not react at all.

Riegler* devised a method for the determination of formaldehyde,
which consists in the measurement of the volume of nitrogen liberated
from a known amount of hydrazine sulphate (in presence of the formal-
dehyde) by the action of a solution of iodic acid. From the difference
between the volume of nitrogen formed under these conditions, and the
volume liberated from the same amount of hydrazine sulphate (in absence
of formaldehyde), may be calculated the amount of formaldehyde present.
In a subsequent article* Riegler suggested the use of a gasometric method
of alkalimetry, based upon the reaction between hydrazine sulphate
and sodium iodate. Still later* he recommended for the determination
of iodides, a method consisting in the measurement of the nitrogen liberated
by the addition of an excess of hydrazine sulphate to the iodate solution
obtained by oxidizing the iodide in alkaline solution with potassium
permanganate.

Stoll6' described an iodometric method for the determination of hy-
drazine. The titration is carried out in the presence of primary potas-
sium or sodium carbonate. Rupp* maintained that the presence of
sodium potassitun tartrate or sodium acetate leads to more satisfactory
results, but that the necessary delay of 15 minutes in awaiting the end
of the reaction detracts from the practical value of the iodometric method.

Schlotter^ found that the reduction of the alkali bromates with hydrazine
sulphate proceeds quantitatively, and made use of this fact in his gas

* Gazz. chim. ital., 29, 1, 265-69 (1899); Atti. accad. Lincei Roma [5], 15, II, 320;
Chem. Zentr., 1906, II, 1662.

* Z. anal. Chem., 40, 92-4 (1901).
» Ibid., 41, 413-19 (1902).

* Ibid., 46, 315-18 (1907).

* J pr. Chem. [2], 66, 332-38 (1902).

* Ibid. [2], 67, 140-42 (1903).

^ Z. anorg. Chem., 37, 164-71 (1903).

ON THE OXIDATION OF HYDRAZINE. 55

volumetric method for the determination of bromates.^ He also as-
certained that iodic acid may be reduced quantitatively to hydriodic
acid, by the action of hydrazine sulphate, but that potassium chlorate
undergoes quantitative reduction only after prolonged boiling.'

Roberto and Roncali' expressed the reaction between chlorine water
and hydrazine sulphate by means of the equation

N^H^.H^sd, + 2CI2 = Na + 4HCI + H2SO,.
They recommended the use of hydrazine sulphate in the analysis of
chloride of lime and of Javelle solution. In a later article* they have
described the action of oxidizing agents upon hydrazine sulphate with
the aid of the following general equation:

NjH^.HaSO, 4- 2O = H2SO4 + 2H2O + Nj.
After expressing the reaction between potassium permanganate and
hydrazine sulphate in acid solution by means of an erroneous equation,*
they predicted the probable usefulness of hydrazine sulphate in the
analysis of oxidizing agents in general, including peroxides and chlorates.
Jannasch and Jahn* found that in alkaline solutions potassium iodate
and potassitun bromate are easily reduced by hydrazine sulphate, but
that potassium chlorate is decomposed only to a limited extent. When
potassium chlorate is treated with hydrazine sulphate in the presence of
hydrogen peroxide and nitric acid, considerable quantities of hydronitric
add are formed. This result was tacitly attributed by Jannasch and
Jahn to the action of nitric acid^ upon the hydrazine sulphate, in accord-
ance with the observations of Sabanejeff and Dengin.*

The present article contains the description of a series of experiments
illustrating the respective action of potassium chlorate, potassium bromate,
potassium iodate, chlorine, bromine, and iodine upon hydrazine sulphate
in add solution. These experiments have been performed primarily for
the purpose of investigating the possible formation of hydronitric acid
as one product of the above reactions, and of determining the yield of the
add under certain specified conditions.

* Z. anorg. Chera., 37, 172-76 (1903).

* Ibid., 38, 184-90 (1904).

• Llndustria chimica, 6, 93-5 (1904); Chem. Zentr., 1904, 1, 1294.

* L'Industria chimica, 6, 178-79 (1904); Chem. Zentr., 1904, II, 616.

' Compare Petersen, Z. anorg. Chem., 5, 1-7 (1893); also Medri, Gazz. chim. ital.,
3<i» I, 373 (1906); Chem. Zentr., 1906, II, 459.

• Ber., 38, 1576-89 (1905).

^ In the light of certain experiments performed in this laboratory (Browne, This
Jonmal, 27, 551-55 (1905) ; see also the results shown in Table I of the present artide)
it is apparent that the formation of hydronitric acid under the conditions prevailing
in the experiments of Jannasch and Jahn might also be attributed to the action of
either one or both of the other oxidizing agents present: hydrogen peroxide and potas-
sium chlorate.

• Z. anorg. Chem , ao, 21-29 (1899).

56 A. W. BROWNE AND V, F. SHETTBRLY.

Preparation of SoltUions. — The solution of hydrazine sulphate con-
tained lo.ooo grams of the salt per liter. The solutions of potassium
chlorate, bromate, and iodate contained 9.4226, 12.8418 and 16.4575
grams of the respective salts per liter. All four solutions were therefore
almost exactly equimolecular.

General Procedure Followed in the Experiments. — ^A measured volume
of the hydrazine sulphate solution was placed in a one liter round bottom
flask provided with a f9ur-hole rubber stopper through which passed
(i) the stem of a separatory funnel, (2) a thermometer, (3) a glass tube
serving as the air inlet, and reaching to the bottom of the flask, and (4)
a glass elbow tube communicating with the condenser. After the de-
sired amount of concentrated sulphuric acid had been added, the solu-
tion was heated to about 80®, and a measured amount of the oxidizing
agent was added drop by drop through the separatory funnel. Through-
out the entire experiment a current of air was in general drawn through
the apparatus, in order to carry the hydronitric acid from the flask through
the condenser into the absorption apparatus (which contained 5 ca of a
10 per cent, solution of silver nitrate, 2 cc. of a 10 per cent, solution of
sodium acetate and 35 cc. of distilled water), for a description of which
reference may be made to the first article of the present series.^ After
the entire quantity of the oxidizing agent had been added, the solution
was boiled until all of the hydronitric add had been distilled from the
flask. In the quantitative experiments the amount of hydronitric acid
formed was determined as usual by the method of Dennis and Isham.^
In certain experiments the yields of ammonia were determined by making
the residual solution (after the hydronitric acid had been completely
expelled) alkaline with sodium hydroxide and distilling with usual pre-
cautions into standard hydrochloric add. The yields of both hydronitric
acid and ammonia have been calculated on the basis of the equation*

2NaH, -f 2O = HN3 -f NH3 + 2HjO.

Results Obtained in the Experiments. — ^The results are tabulated under
five dififerent heads: (i) action of potassium chlorate upon hydrazine
sulphate; (2) action of potassium bromate upon hydrazine sulphate;
(3) action of potassium iodate upon hydrazine sulphate; (4) action of
chlorine, bromine and iodine, respectively, upon hydrazine sulphate;

' Browne and Shetterly, This Journal, 29, 1305-12 (1907).

* Ibid., 29, 18 (1907).

» This equation recalls the work of E. Fischer (Ber., 10, 1336 (1877)), who obtained
diazobenzenitnide and aniline (analogous to hydronitric add and ammonia) by the
action of iodine upon phenylhydrazine. This analogy in the behavior of the organic
and the inorganic derivatives of hydrazine, as well as the interesting question of the
possible formation of certain new inorganic compounds of hydrogen and nitrogen as
intermediate products of the oxidation of hydrazine, is now under investigation in this
laboratory, and will be discussed in detail in a future communication.

ON THE OXIDATION OF HYDRAZINE.

(5) action of potassium chlorate, bromate and iodate, respectively, upon
hydrazine sulphate in presence of silver sulphate.

(j) Action of Potassium Chlorate upon Hydrazine Sulphate, — ^When
heated together in presence of free sulphuric add, aqueous solutions of
these substances react with the formation of considerable quantities of
hydronitiic add and ammonia. The add was identified by the series of
qualitative tests described in an earlier article. * The details of a number
of experiments are given in Table I.

Tabl« I.

N,H4.HsS04
Number of solution,
xperiment. cc.

KClOa
solution,
cc.

Concentrated AgCl
HsSOi obtained,
cc. Gram.

Yield

HN3.

Per cent.

Portion of dis-
tillate containing
bulkofHNs.

100

28.6

10. 0 0.0738

1339

• •

100

28.6

10. 0 0.0840

15-24

• •

100

33.3

10. 0 0.0814

14-77

• •

100

33-3

10. 0 0.0874

15-86

• •

100

50.0

0.5 0.0204

370

100

50.0

5.0 0.1207

21.90

I€X>

50.0

10. 0 0.0822

14.91

4 and 5

100

50.0

25.0 0.0942

17.09

land 2

100

50.0

50.0 ....

• • • •

100

55-6

0.5 0.0299

5-42

too

55-6

5.0 0.0426

7-73

100

55-6

10. 0 0.0947

17.18

4 and 5

100

55.6

25.0 0.0513

9.30

I and 2

100

55.6

50.0 ....

• • ■ •

100

66.7

10. 0 O.IO91

19-79

• «

100

66.7

10. 0 0.1237

22.44

• •

From these experiments it is apparent that the yield of hydronitric
add does not vary in any regular way with the concentration of the
oxidizing agent or of the sulphuric add. In fact, the magnitude of the
yield seems to depend fully as much upon the method adopted in bring-
ing the two substances together, in heating the mixture, and in distilling
off the hydronitric add, as upon any other factor. The liberation of
dilorine during the experiment militates against the production of a large
}'ield of the acid. That the presence of a considerable amount of sul-
phuric add is necessary seems to be shown by the low yields obtained in
Experiments 5 and 10. A separate series of experiments (each performed
in duplicate), the results of which are given in column 7 of Table I, has
shown, moreover, that the formation of hydronitric add takes place most
readily within a certain range of concentration of sulphuric add. In
each of these experiments the distillate was divided into six 25 cc. frac-
tions. To each fraction was added i cc. of a 9 per cent, ferric chloride
solution. From the depth of color shown by the various fractions the
relative amounts of hydronitric add present were then roughly estimated.

» Browne, This Journal, 27, 5.SI-55 (1905)-

58 A. W. BROWNE AND F. F. SHETTERI^Y.

In every case the bulk of the acid was found in one or at most two of the
fractions; moreover, as the concentration of ^sulphuric acid was increased,
the formation of the hydronitric acid took place earlier in the expetknent.
Thus in Experiments 5 and 10, in which but 0.5 cc. of sulphtuic add was
present, the hydronitric add came over in the sixth fraction, while in
Experiments 9 and 14, in which 50 cc. of sulphuric add were present,
the hydronitric add appeared in the first fraction. These facts cannot
be explained on the ground that hydronitric add might be expected to
distil over more rapidly from solutions strongly addified than from solu-
tions but slightly addified with sulphuric add. Curtius and Rissom^
have shown that when an aqueous solution of hydronitric add is distilled,
almost the entire amount of the add passes over with the first fourth of
the liquid. It is consequently legitimate to infer from the presence of
the greater part of the hydronitric add in the sixth fraction obtained in
Experiments 5 and 10, for example, that the formation of the acid does
not readily take place until toward the close of the experiment, when the
concentration of the sulphuric add has become sufficiently high.

For the purpose of determining the yield of ammonia formed tmder
conditions substantially similar to those under which the largest yield of
hydronitric acid had been obtained, four additional experiments were
performed. Care was of course taken to use suffident potassium chlorate
to insure the complete oxidation of the hydrazine, since any hydrazine
remaining unoxidized would have been at least in part distilled over with
the ammonia when the residual solution was heated after having been
made alkaline. In each of the experiments, 100 cc. of the hydrazine
sulphate solution and 10 cc. of concentrated sulphuric add were em-
ployed. In the first two, 66.7 cc. of the potassium chlorate solution were
used in each case ; in the last two, 77.8 cc. The yield of ammonia in the
four experiments amounted respectively to 48.76, 40.99, 31.64 and 30.06
per cent.

(2) Action of Potassium Bromate upon Hydrazine Sulphate, — ^Aqueous
solutions of these substances react when heated together in presence of
free sulphuric add, forming appreciable quantities of hydronitric add
and ammonia. The results obtained in a series of experiments (parallel
in the main with the experiments in which potassium chlorate was used)
are shown in Table II.

The best yields of hydronitric add were obtained when 5 cc. of sul-
phuric acid were taken. The decrease in yield observed as the concentra-
tion of sulphuric add was increased beyond this limit is at least partially
attributable to the greater diflSculty of preventing, in strongly acid solu-
tion, the liberation of free bromine during the course of the reaction.

* J. pr. Chcm. [2], 58, 261-309 (1898).

ON THB OXIDATION OI^ HYDRAZINE. 59

TablS II.

Number of
experiment.

N,H4.H,S04

solution.

cc.

KBrOs
solution,
cc.

Concentrated

HSSO4.

cc.

AffCl

obtained.

Gram.

Yield
HNt

Per cent.

lOO

28.6

5.0

0.0298

5-41

lOO

28.6

10. 0

• • • •

lOO

33-3

0.0368

6.68

ICX>

33-3 .

10. 0

0.02I7

3-94

lOO

50.0

0.0

0.0058

1.05

lOO

50.0

0.5

0.0104

1.88

lOO

50.0

5.0

0.0341

6.x8

lOO

50.0

lO.O

0.0252

4-57

lOO

55.6

0.5

0.0179

3.25

lOO

55.6

5.0

0.0346

6.28

lOO

55.6

10. 0

0.0095

1.72

lOO

55-6

25.0

0.0017

0.31

The yield of ammonia was determined in four additional experiments,
in each of which 100 cc. of hydrazine sulphate solution were taken. The
other details were respectively as follows: potassium bromate solution,
66.7, 66.7, 77.8, 77.8 cc. ; concentrated sulphuric add, 5, lo, 5, 10 cc. ;
)deld of ammonia, 9.77, 3.91, 1.21, 1.21 per cent.

(j) Action of Potassium lodate upon Hydrazine Sulphate, — Preliminary
experiments seemed to indicate that no hydronitric acid was formed by
the interaction of these substances. In the four experiments subsequently
performed, distilled water was consequently substituted for the usual
absorbing solution (containing silver nitrate and sodium acetate), in
order to facilitate the detection of minute amotmts of hydronitric acid.
In each of the experiments 100 cc. of hydrazine sulphate solution were
employed. The amounts of the other substances were respectively as
follows: potassium iodate solution, 28.5, 33.3, 50, 50 cc; concentrated
sulphuric add, 5, 5, 5, 10 cc. Some difficulty was experienced in keep-
ing the iodine that was liberated during the reaction from passing over
into the distillate, especially in the fourth experiment, in which 10 cc. of
sulphuric add were employed. In certain cases it was found necessary
to shake a portion of the distillate (to be tested for hydronitric acid)
with metallic mercury before adding the ferric chloride solution, in order
to remove the small quantities of free iodine that were imavoidably
distilled over. In no case was the slightest indication of the presence of
hydronitric add observed.

In the fifth experiment 200 cc. of the hydrazine sulphate solution, to
which had been added 20 cc. of sulphuric add, were treated with 133.3
oc. of the potassium iodate solution. The solution was heated imtil the
evolution of gas had ceased, and was then made alkaline with sodium
hydroxide. No indication of the presence of ammonia was observed
when a moistened piece of red litmus paper was suspended above the
liquid in the flask. The subsequent addition of 2 mg. of solid ammonium

6o A. W. BROWNE AND F. ^. SHETTERLY.

chloride (corresponding to 0.5 per cent, yield of ammonia) to the same
solution caused the litmus paper to turn distinctly blue.

(4) Action of Chlorine, Bromine, and Iodine, Respectively, upon Hy-
drazine StUphate. — In the foregoing experiments it was observed (as
might have been expected) that the tendency for the free halogen to appear
during the course of the reaction varied directly with the atomic weight
of the halogen. In the experiments with potassium chlorate, for ex-
ample, it was comparatively easy to keep the free chlorine from passing
over into the absorption apparatus, while in the experiments with potas-
sium iodate it was found necessary to employ some caution in the addi-
tion of the oxidizing agent, and in the regulation of the temperature at
which the reaction was permitted to take place, in order to prevent the
carrying over of considerable quantities of free iodine. In almost every
case it was found possible, however, by the exercise of proper precautions,
to effect the reduction of the free halogen by the hydrazine sulphate still
remaining in the solution. From these facts it is obvious that the hy-
drazine sulphate is in each case virtually subject to the action of two
different oxidizing agents during the course of the experiment, either one
or both of which might be responsible for the production of the hydronitric
acid.

In order to investigate qualitatively the behavior of the free halogens
toward hydrazine sulphate, with reference to the formation of hydronitric
acid, a series of experiments was performed in which hydrazine sulphate
was treated in both acid and alkaline solution with chlorine, bromine
and iodine solutions, respectively. The chlorine solution was prepared
(immediately before the experiments were performed) by saturating
distilled water at room temperature with chlorine gas. The bromine
solution contained about 6.9 grams of bromine per liter. The iodine
solution contained about ii.o grams of iodine, and 22 grams of potassium
iodide per liter. The details of the experiments are given in Table III,

Tablb III.

Number
of experi-
ment.

NSH4.H.SO4
solution,
cc.

Oxidizing?
solution,
cc.

Cone.
HSSO4.
cc.

NaOH.
Grams.

Result of
tests for HNs.

100

(chlorine)

Small amount oi

100

•

i- fl CI

100

•

i( ir <i

100

•

(1 *t t<

100

(bromin«)

■

NoHN,

100

•

it it

100

•

Trace of HN,

100

•

fl CI

100

(iodine)

•

NoHN,

100

•

:c n

100

•

tt cc

100

•

.1 «<

cc
tl
tt

ON THK OXTOATION OI^ HYDRAZINE. 6 1

For the experiments in which sulphuric acid was used the general
procedure was briefly as follows: the oxidizing solution was first added
slowly to the acidified hydrazine sulphate solution, with frequent shaking
in order to hasten the reaction. When the reduction of the halogen was
complete (in the experiments with iodine it was found necessary to heat
the solution at intervals), the solution was distilled, and the first fractions
were carefully tested for hydronitric acid. For the experiments in which
sodium hydroxide was used, the general procedure was as follows : The
oxidizing solution was first brought into contact with the sodium hy-
droxide (in the form of a 20 per cent, solution). The hydrazine sulphate
was then added, and the solution was shaken until the evolution of gas
had nearly ceased. After having been heated for a short time, the solu-
tion was slightly acidified with sulphuric acid and was distilled. Careful
tests for hydronitric acid were made, as before, upon the first fractions
of the distillate. In no case were indications given of the formation of
any very considerable quantity of hydronitric acid. When chlorine
was used it was fotmd that a small amount of hydronitric acid was formed
in both acid and alkaline solution. With bromine, traces of the acid
were formed in alkaline solution, but none in add solution. With iodine
there was no formation of hydronitric acid in either add or alkaline solu-
tion. From these results the conclusion may be drawn that the tendency
of the free halogens to form hydronitric add as one of the oxidation
products of hydrazine sulphate is but slight in any case, and that this
tendency varies inversely with the atomic weight of the halogen.

It seems reasonable also to infer that the liberation of the halogens
in the chlorate, bromate and iodate experiments must be unfavorable
to the production of large yields of hydronitric add, even though pre-
cautions are taken to keep the free halogen in any case from passing over
into the absorption apparatus. Since the liberation of a given amount of
iodine would presumably militate more strongly against the production
of hydronitric add than would the liberation of a corresponding amount
of bromine or of chlorine, and since the liberation of iodine takes place
more extensively than that of the other halogens, it is by no means sur-
prising that hydronitric add was not formed in the experiments with
potassium iodate. It has moreover been shown by experiment that even
potassium chlorate, under conditions otherwise very favorable to the
formation of hydronitric add, does not produce appreciable yields of the
add in presence of potassium iodide. Two experiments were performed
in which the procedure adopted in. Experiments 6 and i6, Table I, was
carefully duplicated. In these cases, however, 0.64 gram and 0.85 gram
of potassium iodide were dissolved in the acidified hydrazine sulphate
solution before the (gradual) addition of the potassium chlorate solu-
tion. In both cases the tests for hydronitric add gave negative results.

62 A. W. BROWNS AND F. P. SHBTTl^RLY.

(3) Action of Potassium Chlorate, Bromaie and lodate, Respectively,
upon Hydrazine Sulphate in Presence of Silver Sulphate. — ^A natural sup-
position to be made on the basis of the facts discussed in the preceding
paragraphs is that if the liberation of halogen in the chlorate, bromate,
and iodate experiments could be entirely prevented, the yield of hy-
dronitric add might be appreciably increased. In the experiments re-
corded in Table IV this condition has been realized by the addition in
each case, of i gram of silver sulphate, in solid form, to the acidified
solution of hydrazine sulphate, prior to the introduction of the oxidizing
solution. The procedure was in all other particulars the same as in the
earlier experiments with these oxidizing agents.

Table IV

Number
of experi-
ment.

NSH4.H9SO4

solution.

cc.

Oxidizinz Concentrated
solution. HtS04.
cc. cc.

AgCl
obtained.
Gram.

Yield
HN,.

Per cent.

Yield

NHg.

Per cent.

100

(Kao,) 78.0 10

00734

13 32

15.64

100

" 78.0 10

0.0707

12.83

16.96

100

(KBrO,) 66 . 7 10

0.0374

6.79

25.25

100

* 66 . 7 10

0.0642

11.65

27.42

100

(KIO,) 66.7 10

0.0520

9-43

12.68

100

66 . 7 10

0.0615

II. 16

16.16

These experiments show clearly that while in the case of the chlorate
solution the presence of silver sulphate does not increase the yield of
hydronitric acid, with the bromate solution a slight increase, and with
the iodate solution a very marked increase is to be noted. In other
words, the influence of the silver sulphate in augmenting the tendency
of the chlorate, bromate, and iodate solutions to form hydronitric acid
from hydrazine sulphate in acid solution, varies directly with the atomic
weight of the halogen.

Summary.

When potassium chlorate or bromate is brought into contact with
hydrazine sulphate in the presence of sulphuric acid, the hydrazine is not
completely oxidized to nitrogen and water. A secondary reaction takes
place which may be expressed by the equation

2NjH, + 20 = HN, -f NH, + HjO.

With potassium chlorate the highest yields obtained were 22.44 P^r
cent. HN3, 48.76 per cent. NH,; with potassium bromate, 6.68 per cent.
HNj, 9.77 per cent. NH3. With potassium iodate under similar con-
ditions no formation of HN3 or of NH, was observed. The amount of
hydronitric acid formed, consequently, decreases with increase of the atomic
weight of the halogen.

By the action of the free halogens upon hydrazine sulphate but little
hydronitric acid was obtained in any case. When chlorine was used, a
small amount of the acid was formed in both acid and alkaline solution.

k^ACtlON Bli^tWtBli iStlC^ AKD SUtPHUK. 63

With bromine, traces were formed in alkaline solution, but none in acid
solution. With iodine there was no formation of hydronitric add in
either add or alkaline solution. The slight tendency of the free halogens
to produce hydronitric acid from hydrazine sulphate, consequently de-
creases with increase of the atomic weight of the halogen.

The maximum yields of hydronitric add and ammonia obtained in a
series of experiments with potassium chjorate, bromate, and iodate, in
the presence of sulphuric add and silver sulphate, were as follows: with
potassium chlorate, 13.32 per cent. HN3, 16.96 per cent. NHj; with potas-
sium bromate, 11.65 per cent. HN3, 27.42 per cent. NH3; with potassium
iodate, 11. 16 per cent. HNg, 16.16 per cent. NHj. The influence of the
silver sulphate in augmenting the yield of hydronitric add consequently
varies directly with the atomic weight of the halogen.

The behavior of a number of other oxidizing agents toward hydrazine
sulphate is now under investigation in this laboratory.

COENBLX, UNIVESSITT,

October, 1907..

Oir THE REACnOir BETWEEN LIME AND SULPHUR.

Bt R. W. Thatchbr.
Received August 15, 1907.

The compounds which may be formed by the tmion of calcium and
sulphur, dther with or without oxygen, are quite numerous and varied
in their properties. They have been extensively studied in connection
with the theoretical prindples involved in the replacement of oxygen
by its analogous element sulphur. Recently, the subject has been given
a very important economic bearing by the very extensive use of solu-
tions prepared by boiling together in water, lime, sulphur, and sometimes
other ingredients, and applied as insectiddes for scab on animals and for
soft-bodied scale insects on fruit trees. Some idea of the extent to which
this wash is being used on the Pacific Coast States may be obtained from
the fact that a single firm has recently installed in California a plant
which is producing at each single boiling fourteen carloads of a concen-
trated lime-sulphur solution of twelve times the strength in which it is
used in orchard practice, while at least two other firms are maniifacturing
similar concentrated solutions on a large scale, and a very much larger
amount of the wash is produced by home-boiling in orchards and sheep
camps.

The formulae which have been used for the preparation of the in-
sectidde wash have differed widely. The various modifications of the
original formula have been based on a great variety of conceptions as
to the nature of the compotmds formed in the wash an^ their insecticidal
properties, none of which, however, were based on any accurate knowl-

64 tU W. tHAtCHKl^.

edge of the reactions involved. Because of this, we began in the labora-
tory, in 1902, a study of the nature of the reactions involved and the
compotmds formed when lime, sulphur, and other ingredients, are boiled
together in water. This study has been continued, as opportimity has
permitted, up to the present time. We have now suflScient data from
which to draw certain conclusions, which are presented herewith. Re-
ports of the progress of these investigations have been issued as Bulletins
Nos. 56 and 76 of this Station,' and complete details of our analytical
data up to the date of their issuance may be found therein. The earlier
investigations showed clearly that, whereas it had been supposed that the
wash was a very complex mixture containing many different compounds,
in reality it contains only two characteristic compounds, calcium penta-
sulphide and calcium thiosulphate, small amounts of sulphite or sulphate
being occasionally found as a result of oxidation subsequent to the forma-
tion of the true products of the reaction.* Later, studies were carried
on with a view of ascertaining whether the proportion of these two in-
gredients, or of lime and sulphur in the solution can be varied by chang-
ing the conditions of boiling.

The results of the analyses of a few typical solutions, made according
to formulae representing the extremes of those which have been suggested
for the preparation of the wash, are included in Table I. Many others,
of varying formulae and conditions of boiling, have been prepared and
analyzed, but the general relationships found were the same as in those
here recorded, and the analytical data are omitted in order to economize

Tablb I.

COBiPOSITION OF LiMS-SULPHUR SOI^UTIONS PRBPARBD ACCORDING TO DiPPSRBNT

Formulas.

Sulphur in loo cc. of solution.
# * s

As I«ime Ratio Ratio

Formula. As As sulphite in lime penta-

penta- thiosul- and sul- loo cc. of to sulphide to

I«ime. Sulphur. Salt. Water. Total, sulphide, phate. phate. solution, sulphur thiosulphate
Parts. Parts. Parts. Parts. Grams. Grams. Gram. Gram. Grams, in solution. sulphur.

2 I I 25 3.94 3.19 0.72 0.03 2.06 1:1.91 1:4-43

I I I 25 3.94 3.18 0.73 0.03 2.03 1:1.94 1:4-36

I I 25 3.94 3.19 0.72 0.03 2.01 1:1.96 1:4-43

I ij .. 25 3.97 3.25 0.69 0.03 2.04 1:1.96 1:453

1 • 2i 25 3.95 3.16 0.71 0.05 1.32 1:2.24 1:4-43

space. Our results show that when freshly slaked lime and sulphur,
either with or without the addition of salt, are boiled together in an ex-
cess of water, in open vessels, they dissolve in the average proportions
of I part calcium oxide to 1.94 parts sulphur, if the lime is in excess, and
of I part calcium oxide to 2.24 parts sulphur if the latter is present in
excess. The solubility of lime, as calcium hydroxide, in water in the

' Haywood has independently arrived at similar conclusions, see This Journal,
a8y 245, and Bureau of Chem. Bull. No. zox, 10.

REACTION BETWBBN I^IME AND SUIrPHUR. 65

proportions used, is sufficient to account for nearly two-thirds of the
difference between these ratios, and its solubility in the lime-sulphur
mixture may be somewhat greater. It is very probable, therefore, that
the proportion of calcium in combination with sulphur in the solution
is the same regardless of which of the two components is used in excess,
the increased amount of lime in solution when the latter is in excess being
simply dissolved calcium hydroxide. This excess of calcium hydroxide
may after a time unite with some of the calcium pentasulphide of the
solution, since if solutions prepared with an excess of lime are allowed
to stand for several days, or longer, they frequently deposit bright red
or yellow crystals of oxysulphides of calcium. These latter are probably
of variable composition, since they have been variously stated to be
3CaO.CaS4.i2HaO, 2CaO.CaS8.10 or iiHjO, 5CaO.CaS5.2oHjO.
4CaO.CaS4.i8H20, or 3CaO.CaS8.14 or 15H3O.1
The comparatively slight variations in the proportions of the ingredients
fomid in the several solutions are easily accounted for, either by analjrtical
errors due to the fact that the methods used have been very recently
devised and may not be quite perfect,^ or by some chemical changes
produced by too prolonged boiling of the solutions, resulting in subse-
quent rearrangements according to reactions which have been pointed
out by Haywood.*

The possible reactions between calcium hydroxide and sulphur are
represented by the following equations:

(i) 3Ca(0H), + 12S = CaSA + 2CaS5 -1- 3H,0 or CaO: S:: i : 2.286

(2) 3Ca(OH)3 + 8S = CaS208 + 2CaS8 + 3H3O or CaO: S:: i : 1.524

(3) 3CaOH, -f 4S = CaSA + 2CaS -\- 3HjO or CaO: S:: i : 0.762
and perhaps intermediate ones, resulting in the formation of CaS4 or
CaSy Schdne* states that when calcium sulphide and sulphur are boiled
together in water, they invariably dissolve in the proportions to form
CaS^ and CaS^. Since our analyses have shown, however, that after
allowing for the lime in solution as calcium hydroxide the ratio of lime
to sulphur in combination is always about i : 2.24, it appears that equation
(i) above represents the reaction by which the chemical tmion takes
place when these two substances are boiled together in water in open
vessels. Our results show, furthermore, that any excess of either lime
or sulphur above these proportions is left uncombined in the mixture,
and has no more insectiddal value than it would have if boiled with
water alone. Hence the most economical use of these ingredients for
the preparation of lime-sulphur compounds for insecticide purposes re-

* See Watts' Dictionary, Vol. i, 667.

* The methods used were practically identical with those described by Haywood ;
This Journal, 28, 247--248, and Bureau of Chem. Bull. No. zox, 9.

' This Journal, aS, 249.

* Pogg- Ann., XX7, 58.

66 R» W. THATCHER.

quires that they be taken in the proportions on i part lime to 2.24 parts
sulphur. Since in practice, lime is rarely absolutely pure, a somewhat
larger proportion of it, say i part to 2 parts sulphur, had best be used.
Further, it is sometimes desirable to have an excess of imdissolved lime
in the mixture, as ** whitewash" in order to make it more easily visible
when sprayed on the tree. In such cases, the Ume must be used in larger
proportions than i part to 1.94 parts sulphur or none will remain un-
dissolved.

It will be noticed in the above table of analyses that the ratio of sulphur
in sulphide form to that in thiosulphate form is always slightly less than
would be produced by the reaction represented by equation (i), i. e,,
I to 5. This is imdoubtedly due to secondary oxidation of pentasulphide
to thiosulphate during the boiling in open vessels, according to the re-
action suggested by Haywood,* since both Haywood's experience and
our own have demonstrated that the longer the mixture is boiled after
the sulphur is dissolved, the greater the proportion of thiosulphate sul-
phur becomes. Since the chief, if not the only, insectiddal value of the
mixture lies in the pentasulphide sulphur which it contains, care should
always be observed in preparing solutions for use as insecticides to prevent
this oxidation as far as possible. This can be done by shortening the
boiling and by covering the cooking vats so as to exclude the air as much
as possible.

Concentrated Lime-StUphur SoltUions, — ^Within the past two years
several firms have been experimenting with the view of producing highly
concentrated solutions of lime-sulphur compounds to be shipped and
sold for insecticide use. A considerable number of these have been
submitted to us for analysis. Some of the results of these analyses are
presented in Table II. In each of the three different brands represented,
or solutions manufactured by three different firms (represented by A,
B, and C respectively), the samples which were first produced are re-
corded first, and the following analyses show the results of later attempts
to increase the concentration and proportion of sulphur in solution.
These solutions are made by boiling by means of steam heat, a thick
** milk of lime" with the proper amount of very finely divided sulphur,
in tanks from which the air is excluded as completely as possible by
means of heavy wooden covers. The boiling is continued until the maxi-
mum amount of sulphur which will stay in solution after cooling is dis-
solved. The manufacturers state that further concentration of the solu-
tions beyond the point represented by the final sample of each brand
results in the separation of crystals on cooling. All these samples have
now been standing in our laboratory for six months or more, and show

* This Journal, 28, 249.

REACTION BETWBBN LIMB AND SULPHUR.

no deposition of crystals; hence, they probably represent approximately
saturated solutions.

Table II.
Composition op Concentrated Limb-Suu>hur Solutions.

Sulphur in loo cc. of solution.

Sample
Mo.

Total.
Grams.

penta«

sulphide.

Grams.

thio-
sulphate.
Grams.

sulphite

and
sulphate.
Gram.

Lime in
100 cc.
of

solution.
Grams.

Ratio
lime
to
sulphur
in solution.

Ratio
pentasulphide

to
thiosulphate
sulphur.

20.30

17.68

2.05

0.57

8.17

1 : 2.48

i: 8.62

32.38

29.79

2.02

0.57

12.49

I : 2.72

1:14-75

35.63

34.18

1.23

0.22

13.74

1:2.49

I : 27.78

2364

19.00

4.10

0.54

10.50

1 : 2 . 29

i: 463

25.47

22.58

2.08

0.81

10.66

1:2.39

i: 10.85

30.21

27.80

1-74

0.67

H.54

i: 2.61

i: 15.92

35.89

34.29

1.48

0.12

14.28

1 : 2 . 52

1:23.18

17.65

14.00

332

0.33

8.67

i: 2.04

i: 4.19

26.98

24.82

1.88

0.28

10.47

1:2.58

i: 13.20

34.07

32.03

1.80

0.24

13-30

1:2.56

i: 17.78

It will at once be noticed from these results that in the highly con-
centrated solutions the proportion of sulphur in sulphide form is very
largely increased over that formed in the solutions prepared in open
vessels, as shown in Table I. It is a well-established principle that the
pentasulphide is the highest possible poly sulphide, and the reaction by
which calcium pentasulphide is produced by the union of lime and sul-
phur (see Equation i above) yields only 80 per cent, of the total sulphur
in sulphide form, whereas in solutions A3, B^, and Cg, the proportion of
the total sulphur which is in this form is 95.95 per cent., 95.54 per cent,
and 94.00 per cent, respectively. Hence in the preparation of these
solutions some secondary reaction must take place, resulting in the change
of a large part of the thiosulphate into polysulphide sulphur. Just what
this secondary reaction may be is very difficult to determine. The fol-
lowing possible equations have suggested themselves:

(5) CaSA + Ca(OH)a -1-48 = CaSg + CaSO^ + H^O

(6) 2CaS308 -h 3Ca(SH)j = 2CaS5 + 3Ca(OH)a

(7) CaSA + 3H2S = CaSg -f- 3H2O

(8) 4CaSjO, - CaSj + sCaSO^.

A reaction according to (8) appears to be very improbable, both from
theoretical reasons and from the fact that it would produce much more
calcium sulphate than we have ever found in such a mixture. A re-
action according to (7) or (6) requires the previous formation of hydrogen
sulphide or calcium sulphhydrate respectively, the production of either
of which under the conditions of boilidg these solutions seems highly
improbable. Equation (5) seems, therefore, to represent the most prob-
able reaction of those which have suggested themselves. The presence

68 ARTHUR B. HILL.

of a larger proportion of sulphate sulphur in these concentrated solutions
than was found in the dilute solutions boiled in open air also favors the
supposition that the reaction by which pentasulphide is produced at the
expense of thiosulphate is accompanied by the formation of calcium
sulphate. The smaller proportion of sulphate sulphur in some of the
more concentrated solutions seems to oppose this view, however. On
the other hand, the analytical methods for distinguishing quantitatively
between these several forms of sulphur compounds are not yet thoroughly
perfected and it may be that this apparent objection to the reaction
would be removed if more exact methods of analysis were available.
In the absence of better analytical methods, these results are presented
as the best obtainable, and the conclusions suggested as a possible step
toward a better knowledge of the reactions between lime and sulphur
under varying conditions.

Laboratory of the
Washington Aoriculturax, Experiment Station,

Pullman, Wash.

[Contribution prom thb Havbmbybr Chemical Laboratory, Nbw York Unfvkr-

SITY.]

THE RELATIVE SOLUBILITY OF THE SILVER HALIDES AND

SILVER SULPHOCYAWATE.

By Arthur £. Hill.
Received October 22, 1907.

The theory of electrolytic dissociation teaches that if saturated solu-
tions of two very insoluble salts which have an ion in common could be
mixed without increase of volume, precipitation of both salts would
occur. For example, saturated solutions of AgCNS and AgCl would,
upon mixing, precipitate both compounds in part, since by the addition
of silver ion the solubility product of each salt would be exceeded. The
quantities of chloride and sulphocyanate precipitated would be such as
to leave the solution saturated in respect to both salts, and the equilibrium
finally reached would be expressed for the respective compounds by the
equations C^g X Cqss ^ ^i ^^^ ^Ag X C^i = K,, where C stands for
concentration in equivalents per unit volume, K^ and K^ are the products
of the free ions (solubility products), and the subscripts denote the re-
spective ions. By division,

Aff ^ CNS ^^1 vt

The common term C^ may be cancelled out, the relation becoming

CNS ^^1 «-»

^_^ = ^ (Equation 2).

'CI

SILVER HALIDES AND SILVER SULPHOCYANATE. 69

The same equilibrium will be attained if to a solution of AgCl, in pres-
ence of excess of the salt, a soluble sulphocyanate, such as KCNS, be added
in sufficient quantity to exceed the solubility product of AgCNS. Pre-
cipitation of AgCNS will result, reducing the concentration of silver
ions; the solution will therefore become temporarily undersaturated with
respect to AgCl, and the latter salt will dissolve, thereby increasing the
concentration of chloride ions and also of silver ions, so that the cycle of
reactions will again be set in operation. Simultaneous precipitation of
AgCNS and solution of AgCl will continue until the mixture contains the

ions CI and CNS in such quantities as are in equilibrium with the two
silver salts, as expressed in equations i and 2. . The solution thus ob-
tained differs from the hypothetical solution discussed in the preceding
paragraph in having present an additional cathion (potassium ion) and
in having the chloride and sulphocyanate ions present in amounts measur-
able by ordinary anal3^ical methods. The theory of solution equilibria
of this character was first stated by Nernst.^

By the application of equation 2 to reactions such as the foregoing
it becomes possible to determine the relative solubility of two difficultly
soluble salts whenever (i) the concentration of the free ions (Cj^ and
C^ in the example given) can be calculated from the total concentration
of the substances in the solution as analytically determined, and (2)
the degree of dissociation of the two insoluble salts in saturated solution
is known, so that the total solubility can be calculated from the solubility
products (Ki and Kg). Following this method a number of such equilibria
have been investigated, and the solubility of the salts determined. Guld-
berg and Waage's classic results on the reaction BaCO, + K3SO4 =
BaS04 + KjCO, have been shown by Nemst' and by Meyerhoffer* to
refer to an equilibrium of this class, although originally interpreted other-
wise. Other investigators have studied the following reactions of this
typei

TlCl 4- KCNS "^ TICNS + KCl (Kniippfer*).

Pbl, -h Na^O^ !^ PbSO^ + 2NaI (Findlay^).
CaCOg -f K^CA ^ CaCA + K2CO3 (Foote«).
2 AgCl -f 2KOH ^ AgjO -f- 2KCI -f- HjO (Noyes and Kohr').

BaCjO^ -f CaCla ^ CaCjO^ + BaClz (Foote and Menge«).

* "Theoretische Chemie," 5th edition, p. 535.

• Z. physik. Chem., 53» 5i3 (iQOS)-

* Ibid., 26, 255 (1898).
» Ibid., 34, 407 (1900).

• Ibid., 33, 740 (1900).

^ Ibid., 42, 336 (1902). This Journal, 24, 1141 (1902).

• Am. Chem. J., 35, 432 (1906).

70 ARTHUR B. HILL.

BaCO, + CaClj ^ CaCO, + BaCl, (Foote and Menge).
BaFa + CaCl, ^ CaF, + BaCl, (Foote and Menge).

The investigation here described has as its object the study of the
relative solubility of the silver halides and silver sulphocyanate by means
of the solution equilibrium discussed above. The reactions studied were
the following:

AgCl 4- KCNS ^ AgCNS + KCl.

AgCNS + KBr ^ AgBr + KCNS.

AgBr + KI ^ Agl + KBr.

The series seemed worthy of investigation both because of the value
of additional data on the solubility of these important silver salts and
further because the reactions appeared to be particularly well adapted
to the method. The degree of dissociation of the four potassium salts
was calculated from electrical conductivity data;* the values were found
to agree within 1.8 per cent, in fifth-nonnal, 0.6 per cent, in twentieth-
normal, and o.i per cent, in hundredth-noniial solutions, and the salts
may therefore be regarded as equal in dissociation. (As will be seen
from what follows, the small error introduced by this assumption of
equality is reduced by the extraction of the square roots in equation 4.)
Since salts of equal dissociation constant having a common ion become
equally dissociated when mixed in any proportions,' it follows that the
ratio of the free ions (Equation 2) will be the same as that of the total
chloride and sulphocyanate present, as analytically determined. Equa-
tion 2 then becomes

^Ul^PHOCYANATE ^1 /ta ..

P =• ^^ (Equation 3) ,

^CHLORIDE ^

in which the subscripts denote total concentration. Furthermore, the
four silver salts have been shown by Bottger* to be completely dissociated
in saturated solutions, so that the ratio of the products of the ionic con-

centrations, ^\ becomes the ratio of the squares of the actual solubilities

of the salts, ^ — . By substitution of this value for — in equation 3,
and extraction of th6 square roots, the relation becomes

^AgCNS __ VCsULPHOCYANATB /p .. .

^AgCl VCcHLORIDK

that is, the ratio of the solubilities of the two salts is equal to the ratio

* Landolt-B6mstein-Meyerhoffer, Tabdlen, 3rd edition, p. 744.

* Nerast's " Theoretische Chemie," 5th edition, p. 509.
^ Z. physik. Chem., 46, 602 (1903).

SILVER HALIDES AND SILVER SULPHOCYANATE. 71

of the square roots of the total concentrations of anions, free and com-
bined. This simple relation renders unnecessary the corrections for
inequalities in dissociation which have been applied in all the previously
noted studies on solution equilibria except that of Kniippfer, thereby
freeing the results from that source of possible error. In one particular
only does the series fail of the ideal for an investigation of this character —
the solubilities are in two cases widely divergent, so that small errors in
analysis would cause noticeable variations in the ratios found. It has
been possible, however, to reduce these errors to such an extent that they
do not greatly affect the constancy of the ratios obtained.

The silver salts used in the experiments were prepared by precipitating
pure silver nitrate in hot dilute solution with excess of the haloid salt,
and washing the precipitates by decantation until the wash waters were
free from halogen. The silver chloride was dissolved in ammonia, re-
precipitated by nitric acid and again washed by decantation. All the
silver salts were kept in a moist condition and protected from the light.
The potassium chloride and sulphocyanate were Kahlbaum's C. P. prep-
arations, and the latter was foimd by analysis free from chloride; the
potassium iodide and bromide were Baker Ae Adamson's analyzed prep-
arations, the iodide containing 0.005 P^r cent, and the bromide a mere
trace of chloride.

The reactions were carried out in half-liter flasks suspended in an
Ostwald thennostat regulated to 25®, and the reaction mixtures stirred
by glass paddles operated by a gas engine. In each experiment a solu-
tion of the potassium salt of approximately the desired concentration was
taken, and about twice its equivalent of the moist silver salt added to it.
The mixture was stirred for about two hours, after which a further quantity
of the silver salt was added and the stirring continued two to four hours
longer. The time thus allowed was considerably greater than was neces-
sary, as the reactions are apparently very rapid in all cases; equivalents
of AgCl and KCNS have previously been shown^ to react to the extent
of about 43 per cent, in two minutes in solution of hundredth-normal
concentration. Further experiment on this reaction showed that the
equilibrium was reached in less than one hour; the velocity of the other
reactions was not investigated. After equilibrium had been reached,
the solutions were allowed to stand in the thermostat until the precipitates
had settled and samples were then pipetted out and analyzed. Each
equilibrium was approached from two directions. The concentration
of the potassium salt was varied through a considerable range, so that
any irregularity in the ratios, due to hydrolysis or the formation of com-
plex salts, might be noticed. The constancy of the ratios shows that no

' Rosaaofif and Hill, This Journal, 29, 272 (1907).

ARTHUR E. HII.L.

measurable irregularities occur within the range of concentrations selected,
and that the assumptions leading to equations 3 and 4 are justified.

Table I shows the results obtained for the equilibrium AgCl + KCNS
*^ AgCNS 4 KCl. The solutions were analyzed for total chloride and
sulphocyanate by the method of Volhard/ and for sulphocyanate by the
colprimetric method, the chloride being determined by difference. For
the Volhard determination 25 cc. samples were taken in experiments
la and 16, 50 cc. samples in 2a and 26, and 100 cc. samples in 3a and 36.
For the colorimetric tests the standards were made in every way identical
with the analyzed solutions by mixing solutions of potassium chloride
and potassium sulphocyanate in such quantities that the total salt in the
standard was equal to that in the sample tested, and by adding equal
quantities of iron-ammonium alum to the two solutions. In the fol-
lowing table, column i gives the number of the experiment, column 2
the approximate concentration of the soluble salt used, column 3 the
salts taken for the reaction, and columns 4 and 5 the final concentration
of chloride and sulphocyanate as analytically determined. Column 6
gives the ratio of the square roots of these concentrations, which, accord-
ing to equation 4, is equal to the ratio of the solubility of the two salts.

TABLE I-
EquUibrium AgO + KCNS !^ AgCNS + KCl at 25'

No.

la
lb
2a
2b

3»
3&

Approx.
concen.

N/5
N/5
N/20

N/20

N/ioo

Reacting salts.

AgCNS -I- KCl
AgCl + KCNS
AgCNS + KCl
AgCl + KCNS
AgCNS + KCl
AgCl + KCNS

Cone. Cl.

0.197

0.193

0.0501

0.0489

O.OIO

Cone. CNS.

0.00104
O.OOIOO

0.000293
0.000280
0.000060
0.000057

CCNS

Cci

Mean ratio,

SAgCNS

TABLE II.
* Equilibrium AgCNS + KBr !^ AgBr + KCNS at 25''.

0.0726
0.0719
0.0764
0.0759
0.0774
0.0750

=- 0.0748

No.
la
lb
2a
2b

Approx.
Concen.

N/5

N/20

Reacting salts.
AgCNS + KBr
AgBr + KCNS
AgCNS + KBr
AgBr -I- KCNS

Cone. Be.
0.0647
0.0665
0.0165
0.0176

Cone. CNS.
0.1205

o. 1207

0.0312
0.0324

SAgBi

Cbi

CCNS

0.732
0.742
0.727

0.737

Mean ratio ^ ^ ^ "0. 735
SAgCNS '•'*'

* J. pr. Chem., 9 (N. F.), 217 (1874). See also this Journal, 29, 269 (1907).

SILVER HALIDES AND SILVER SULPHOCYANATE. 73

Table II shows the results obtained by comparing AgBr and AgCNS.
The solutions were analyzed for bromides by the method of Rosanoif
and Hill,* the total bromide and sulphocyanate being determined by the
method of Volhard. The columns have the same significance as in the
previous table.

Attempts were made to carry out experiments in solutions of hundredth-
normal concentration, but fine suspensions of the silver salts resulted,
which would not settle in reasonable time and could not be filtered out.
The solutions were finally analyzed in the presence of the suspended solid,
but under those conditions the results obtained by approaching the equi-
librium from the two directions differed by 4 to 5 per cent., the amount
of bromide in solution being larger when silver bromide had been taken
than when its original source was the soluble potassium bromide. The
values for the ratio were 0.0709 and 0.0700 when silver sulphocyanate
was the original solid phase, and 0.0787 and 0.0763 when silver bromide
was taken; the mean is almost identical with that of Table II. Whether
this peculiarity is referable simply to analytical errors due to the presence
of the finely suspended silver salts or, as seems to be indicated by Table
II, to some change in the nature of the solid phase (mixed crystals,
solid-solution, or the like) muSt remain for the present unanswered.

Analytical difficulties prevent the determination of the relative solu-
bility of silver iodide and bromide from being as accurate as might be
desired. The iodide in solution when equilibrium has been attained
amounts to only 0.0 1 to 0.02 per cent, of the bromide present, expressed
in equivalents, and no method is known by which such minute quantities
of iodides can be exactly determined in the presence of such large amounts
of bromides. Of the various methods proposed for this separation,
that of Fresenius,* which depends upon the oxidation of the iodides by
nitrous acid in sulphuric acid solution, extraction of the free iodine by
means of an organic solvent and titration with Na^SjOj, was thought
to be best suited to the estimation of small amounts of iodine. Test
analyses were made of solutions containing 0.1 to 0.5 mg. of iodide in
presence of 2 to 3 grams of bromide dissolved in 150 cc. of water. The
iodine freed by five drops of a solution of nitrous acid in concentrated
sulphuric acid was collected by extraction of the aqueous solution with
10 cc. portions of chloroform until no further coloration of the solvent
could be detected; the iodine was then titrated with N/400 NajSjOj.
The quantity found was always less than that taken, and by amoimts
varying between 12 and 40 per cent. If this maximum error be assumed
to have occurred in the analysis of the mixtures recorded in Table III,
the error in the ratios calculated will be about 27 per cent., and the ratios

' This Journal, 29, 1461 (1907).

' Quan. Chem. Anal. (Braunschweig, 1898), p. 482.

74 WALTER E. MATHEWSON.

given below may therefore diflfer from the true ratios by that amount.
In the experiments tabulated below, somewhat greater concentrations
were used than in the previous experiments, in order that the iodide
present might become a measurable quantity.

TABLE III.
Equilibrium AgBr + KI ^ Agl + KBr at 25**.

-ycBr

No.

Approz.
Concen.

Reacting salts.

Cone. 3r.

Cone. I.

Agl + KBr

0.955

24.4 X (lo)-5

0.016

AgBr + KI

0.945

13.9 X (l0)-5

0.012

N/5

Agl + KBr

0.185

4.83 X (lors

0.016

N/5

AgBr + KI

0.187

2.2 X (lo)-5

O.OXI

Mean ratio ^ -«o.oi4

OAgBr

From the ratios recorded in Tables I, II and III the relative solubility
of the whole series may be calculated. The values thus obtained are
given in Table IV. Column i designates the salt, column 2 its relative
solubility referred to that of silver chloride as unity, and column 3 the
absolute solubility ' in gram-molecules per liter, taking Kohlrausch and
Rose's* figure for the silver chloride as the standard. Column 4 gives
the maximum and minimum values obtained for the solubility of these
salts by other methods; the figures are taken from the table complied
by Abegg and Cox.*

TABLE IV.

Absolute solubility. Extreme values.

1.6 X (io)-5 1 . 25—1 . 64 X (io)-5

1.2 X (lo)~^ 1.08— 1.25 X (10)"^

8.8 X (lo)-7 6.6 —8.1 X (10)"^

1.23 X (10)"® 0.97 — 1.05 X (io)~*

The foregoing investigation is one of a series planned in collaboration
with Professor M. A. Rosanoff. A change of residence has made it advis-
able to carry out the separate parts of the work independently. Credit
is gladlv given to Professor Rosanoff for many of the ideas contained
in this paper.

New York University,
October, 1907.

Salt.

Relative solubility.

AgCl

I. 00000

AgCNS

0.07480

AgBr

0.05500

Agl

0.00077

ON THE ANALYTICAL ESTIMATION OF 6LIADIN.

By Walter H. Mathbwson.
Received October ai. 1907.

That the common methods for the determination of gliadin are un-
satisfactory is generally recognized. The amount of nitrogenous material

' Abegg and Cox, Z. physik. Chem., 46, 1 1 (1903).
' Loc. cit.

ON THE ANAI.YTICAL ESTIMATION OF GUADIN. 75

extracted by dilute alcohol varies with the strength of alcohol used,*
the relative proportion of alcohol to the flour,* and with the relative
amounts of certain non-nitrogenous constituents of the latter, as adds*
and salts. The experiments described here were carried out in the hope
that a better procedure might be devised than that ordinarily followed.
Five of the flour samples used were patent flours milled in the Experi-
ment Station with the Experimental Roller Reduction Mill by Mr. C.
0. Swanson. They had been bolted through No. lo and No. 12 cloth.
The sixth sample (No. 6) was a patent flour from the Manhattan Milling
Company.

All alcohol used had been redistilled before use and was perfectly
neutral to sensitive litmus paper. The concentrations given are by
weight, not by volume. The alcohol, phenol, etc., used in the nitrogen
determination were tested by blanks and the phenol was also examined
for optical activity. All extractions were made in duplicate, the maxi-
mum difference allowed on the nitrogen determinations being 0.03 per
cent. The results are expressed in terms of crude gliadin, obtained by
multiplying the nitrogen found in the extract by 5.7. No attempt was
made to correct for the amides present, since the values obtained in this
determination are so largely dependent upon the protein precipitant
used.* They could have been present only in very small amount, as
the flours were sound and fresh. All percentages given are based upon
the air-dry substance. The validity of the views of Osborne and Voor-
hees regarding the wheat proteins is assumed throughout the discussion.*

Below are given the percentages of moisture, crude protein (nitrogen
X 5.7), and crude gliadin of the samples as determined in one of the
usual wa>^; namely, by extracting 4 grams of the charge with 100 cc.
cold 70 pQT cent, alcohol in a tight bottle, filtering, and determining
nitrogen in 50 cc. of the filtrate. In the last column are given the re-
sults obtained when 16 grams of flour to 100 cc. of alcohol were taken.

Total Crude g^Uadin. Crude gUadin.

No. of crude 4 grams flour, 16 grams flour,

Sample Moisture. protein. 100 cc. alcohol. 100 cc. alcohol.

I ZI.20 11.42 6.00 5.08

N 2 11.70 7.21 3.72 3.19

3 11.23 11.20 5.56 4.90

4 io-9<^ 1117 592 5-34

5 11.52 12. 6i 6.75 5.93

To guard as far as possible against incomplete extraction the mixtures

* Tdler, Bull. Ark. Expt. Sta., No. 53, p. 63.

* Chamberlain, Bull. Bur. Chem., U. S. Dept. Agr., p. 125, No. 90. Chamber-
lain, This Journal, 28 (1906), 11, p. 1660.

* Snyder, Annual Rept. Minn. Expt. Station, 1904, p. 206.

* Schulze, Landw. Vers. -Sta., 26, 213 (1881).

' Osborne and Voorhees, Am. Chem. J., 15, 392 (1893).

76 WALTER n. MATHS WSON.

were allowed to digest about forty-eight hours before filtration, being
begun in the afternoon and frequently shaken until evening and during
the next day. The fear that any gliadin would be coagulated by pro-
longed contact with 70 per cent, alcohol seems to the writer to be en-
tirely unfounded, as pure gliadin prepared by Osborne's method, dis-
solves completely in alcohol of this concentration and the solutions remain
clear for weeks, perhaps indefinitely. Nor would one anticipate that
the error due to solution of nitrogenous matter not gliadin would be in-
creased much relatively by the standing during the second day.

Sample No. 6 when treated with varying amounts of solvents gave the
following results:

Grams flour per Per cent,

leo cc. alcohol. crude gliadin.

1 4-57

2 4-39

4 4.37

8 4.29

16 4.22

The lower results obtained when the larger quantities of flour are used
may be due both to less complete extraction and to the solution of non-
gliadin nitrogenous substances, which dissolve but slightly and in more
or less constant amount.

Gluten appears to be a solid colloid solution containing essentially the
two gluten proteins with water. It would seem possible that a certain
amount of gliadin might be held in solution in the glutenin, tending to
divide itself between the two phases according to the distribution law.
To test this a solution of pure gliadin in alcohol of about 60 per cent, vras
prepared. It gave a rotation of 21.6° Ventzke in a 200 mm. tube in a
triple shadow saccharimeter. Twenty-five cc. of this solution were added
to 5 grams of gliadin-free, dried flour prepared from sample No. 6 by re-
peated extraction with dilute alcohol, washing with concentrated alcohol
and drying. The mixture was shaken frequently for three or four hours,
allowed to stand over night, filtered and the filtrate polarized. A sample
of the gliadin solution was also filtered under the same conditions, as a
check to determine possible loss by evaporation under these conditions,
and another sample of the flour was digested in the same way with 70
per cent, alcohol, to make sure that optically-active, alcohol-soluble
substances had been removed. The gliadin solution was fotmd to have
suffered no change in concentration that could be detected with the
saccharimeter, by contact with the flour. A second experiment was
carried out in exactly the same way except that instead of flour, 5 grams
of air-dry, pulverized crude glutenin were used, this being prepared by
treating thoroughly washed gluten, cut into small pieces in a meat cutter,
with successive portions of dilute alcohol for some time. It was then

ON THE ANALYTICAL ESTIMATION OF GLIADIN. 77

washed with strong alcohol, allowed to dry at room temperature, grotmd
to a powder that would pass through a i mm. mesh sieve, and this ex-
tracted in an extraction apparatus with ether and with absolute alcohol.
After repeatedly extracting again with dilute alcohol, it was rinsed with
strong alcohol and dried at room temperature.

The crude glutenin swelled up in the gliadin solution, but after applyilig
the small correction for increase in concentration on filtration, the rota-
tion was found to have increased from 10.8 ° V. to 12.1° V. Instead
of removing gliadin from the solution the glutenin had evidently taken
up water or dilute alcohol, thus increasing the rotation of the liquid.
If glutenin had a marked tendency to hold gliadin in solution one would
hardly expect this result.

As has been shown conclusively by Chamberlain^ dilute alcohol dis-
solves other protein substances from flour beside Osborne's gliadin.
Gliadin dissolves most readily in alcohol of about 70 per cent, (by volume),
but a weaker alcohol dissolves a considerably larger percentage of nitrog-
enous matter from flour. Had we some method of rendering this foreign
protein matter insoluble before the extraction, without affecting the
gliadin, it would remove one of the most important sources of error. It
was thought by the writer that possibly this could be effected by heat.
A sample of pure gliadin was heated for six hours in an ordinary drying
oven surrounded by boiling water. It did not seem to be changed physi-
cally and dissolved to a perfectly clear solution on warming somewhat
with dilute alcohol, the solution remaining clear on cooling. Nor did
gluten made from flour heated in the same way appear to differ much
either in properties or amount from that made from similar flour which
had not been heated. According to Osborne, leucosin, the albumen of
wheat, is coagulated at temperatures of about 55-65°. The globulin is
partially coagulated at 100°. Proteoses and amides are present in very
small amount.

Below are given the results obtained by extraction with 70 per cent.
alcohol after previous heating for five hours in an oven surrounded by
boiling water. In the first column are given the results obtained by
treating with cold alcohol (100 cc. for 4 grams) in the ordinary way.'
In the second set of determinations the same proportion of flour and
solvent were used, but the bottles were maintained at a temperature of
60-70** for about four hours during the digestion, the hot mixture being
frequently shaken. As in the other determinations the flour remained
in contact with the alcohol about forty-eight hours, during 12 to 14 of
which the mixture was frequently shaken.

^ Chamberlain, This Journal, aS, 1661 (1906).

* The much lower results obtained by Chamberlain are doubtless due in large
measure to the difference in time allowed for extraction.

78 WALTER B. MATHBWSON.

No. of Crude gliadin. Crude gliadin.

sample. cold extraction. hot eztraction.

1 4-97 5-62

2 3.24 3.60

3 478 5.46

4 5.30 585

5 5-94 6.66

As there seemed some danger that the glass-stoppered digestion bottles
might not altogether prevent the loss of alcohol during the heating, they
were weighed at the beginning and the close of the digestion. The highest
loss was 0.3 gram, most of them not having lost half this amotmt.

In the opinion of the writer it is impossible to obtain satisfactory re-
sults by washing a flour repeatedly on a filter with hot alcohol of the
usual concentration. A filter that will retain the fine suspended matter
cannot act rapidly and in the washing, it is practically impossible to
prevent losses of alcohol from the solvent, as a result of which it is enabled
to dissolve more protein.

The percentages of gliadin found by the cold extraction of the dried
flour are likely too low, the strongly dried protein going into solution
very slowly in the cold solvent. The amounts dissolved in the hot ex-
traction are almost as high as those given by cold alcohol on air-dry flour.
If a purer gliadin was dissolved after the heating, the tendency to higher
results by the ordinary method must be more or less offset by incomplete
extraction.

It was thought that perhaps dilute fi-propyl alcohol might prove a
suitable solvent for the separation of gliadin. Seventy per cent, propyl
alcohol is a mixture of constant boiling-point (about 86°) and hence can
be used in an extraction apparatus. Seventy per cent, propyl alcohol dis-
solves pure gliadin with some difficulty when cold, but readily when hot.
The results given below were obtained by using the solvent in a percolating
extractor, the 2-gram charge of flour being held in a S. and S. filter paper
shell. The liquid in the flask was kept rapidly boiling to bring about as
rapid an extraction as possible.

No. of sample Duration of extraction, Percentage

hours. crude gliadin.

1 10 5.99

2 5 3.66

3 10 5.51

4 10 6.27

5 10 6.18

The figures indicate no special advantage over the preceding methods.

Anhydrous phenol dissolves gliadin readily and apparently without
affecting it chemically.^ Further, pure phenol has a comparatively
limited power as a solvent, and gliadin has a high specific rotation in this

* This Journal, 2$, 1483 (1906).

ON THE ANALYTICAL ESTIMATION OF GLIADIN. 79

liquid. Mixttires of flour with phenol are difficult to filter and in making
the experiments described below this was accomplished by using a Gooch
crucible with asbestos felt. The filtering tube containing the crucible
was connected with a loo cc. distilling flask which served as the filtering
flask. The pump was turned on and the mixture poured on the felt.
A good vacutun was almost at once produced in the little filtering flask.
A pinchcock was then placed on the rubber tube leading to the pump
and, the apparatus being tight, it was not usually necessary to exhaust
it again. The Gooch crucible was covered with a watch-glass and sur-
rounded by a block-tin coil, through which steam could be passed to keep
its contents from congealing. With this arrangement the filtrations
could be made without any perceptible loss of solvent. Ten-gram charges
of the flour were weighed out, dried and extracted with accurately
measured volumes of phenol (96 cc. for sample one, 100 cc. for the others),
the mixtures being kept at about 40^ and frequently shaken for the
first three or four hours. After about twentv-four hours thev were
filtered. Nitrogen was determined in an aliquot part by placing it in a
Kjeldahl flask, adding about 300-400 cc. of water, 2 cc. of concentrated
sulphuric acid and a little granulated zinc. It was then boiled until
only a few cubic centimeters remained, the water vapor carrying off
practically all the phenol, but the sulphuric acid preventing any loss of
nitrogen as was indicated by the concordance of duplicates. Sulphuric
add was then added and the determination carried out in the usual wav.
In calculating the percentage of crude gliadin from the polariscope read-
ings the optical rotation of gliadin was taken [a]g =132^.

Crudb Guadin Extracted by Anhydrous Phbnol.

Crude gliadin Crude gliadin

calculated from calculated from

No. of aample nitrogen determination. polariacope reading.

I 9.86 8.08

2 6.19 4.83

3 9-43 7-93

4 7.12 6.39

The high results and differences between the percentage of crude gliadin
as determined by polarization and by the nitrogen determinations show
the presence of another protein or proteins largely soluble in phenol.
The alternative is that dextro-rotatory substances were also present
in the extract. Portions of the latter from samples 2 and 3 were taken,
treated with one-fifteenth their volume concentrated sulphuric add and
the gelatinous precipitates filtered off. Both filtrates seemed to con-
tain a trace of dextro-rotatory substance, but in so small amotmt that its
presence could not be demonstrated with certainty — certainly less than
would correspond to 0.4 per cent, protein in the flour. Duplicate nitrogen
determinations were made on the filtrate from No. 4; it was free from

8o WAI^TER E. MATHEWSON.

nitrogen. It is unlikely that any carbohydrate-like substances that might
have been dissolved in the phenol would have been precipitated by the
sulphuric acid. Gliadin solutions in phenol give no precipitate with
phenol solutions of mercuric iodide or of iodine.

The above facts taken in connection with the fact that the gliadin
apparently remains unchanged in phenol are of some interest because
the view has been advanced that gliadin is formed by the action of dilute
alcohol on the wheat, being split off from some other protein substance.
The high percentage of protein removed by the phenol evidently in-
cludes another protein or other proteins in considerable quantity. An
attempt to gain more data regarding this was made by extracting flours
with dilute alcohol until the washings were practically free from protein.
By this washing any dextro-rotatory substances soluble in dilute alcohol
would also be removed. These samples were then dried and extracted
with anhydrous phenol, the nitrogen content and the rotation of the
extract determined. The amounts of protein dissolved were too small
to enable accurate estimations to be made, but in both cases the crude
gliadin, as calculated from the rotation, was not 60 per cent, of that
obtained by multiplying the nitrogen by 5.7.

What other substances beside gliadin are dissolved by the phenol has
not been determined. No careful experiments in this direction have
been made, but the common animal proteins do not seem to be dissolved
by it. Witte's peptone dissolves readily. The extremely weak acid
character of phenol (its dissociation constant being not more than one
two-hundredth that of carbonic acid) hardly leads one to suspect that
it acts similarly to the strong acids on the other wheat proteins. In
fact an alcoholic solution of phenol has been advocated as a quantitative
protein precipitant.*

It might seem likely that besides the gliadin but one other wheat protein
dissolves in phenol. Amides and proteoses are present in very small
amount and it is doubtful if the former are soluble. If x represents the
true percentage of gliadin in a flour and but two proteins are present in
the phenol solution, then the specific rotation of the second protein in
phenol woidd be :

(crude gliadin calculated from rotation of extract — 7^
LajD " (crude gliadin calculated from nitrogen of extract) — ^*

regardless of whether the extraction of the second protein were com-
plete or not. The great influence of small experimental errors on this ratio
renders it of comparatively little value, but in no case do the percentages
of gliadin by a given method, when substituted for x, give concordant
values for [a] for the four samples.

' Jago, Science and Art of Bread-making, p. 585.

NITROGEN PEROXIDE UPON WHEAT FLOUR. 8 1

Air-dry flour treated with phenol yields mixtures so difficult to filter
that they were not investigated. Single extractions made with samples
2 and 3 gave the following values which, though not accurate, show
that the protein dissolved from the air-dry flour is not pure gliadin.

Crude gliadin Crude gliadin

calculated from calculated from

No. of sample. nitrogen determination. polariscope reading.

2 5-24 4.15

3 780 6.43

The filtration of the extract was so slow that it may have changed more
or less in concentration from evaporation or absorption of water.

Summary.

With these flours 8 to 17 per cent, more nitrogenous matter was ex-
tracted when 4 grams per 100 cc. of the solvent was used than when four
times as much flour was taken.

After drying six hours in the water oven, 10 to 20 per cent, less gliadin
was obtained by extracting with cold solvent. With the hot solvent the
figures were nearly the same, being slightly lower. Pure gliadin remains
soluble in dilute alcohol after the same treatment.

No tendency for glutenin to remove gliadin from its alcoholic solu-
tions by absorption or with the production of a solid solution could be
demonstrated.

Propyl alcohol of constant boiling-piont (70 per cent, by weight) used
in an extraction apparatus gave results probably no more accurate than
the others.

Anhydrous phenol dissolves a high percentage of protein matter from
the flotu'. The dissolved matter is not pure gliadin, however, nor does
it seem to consist of gliadin with but one other protein.

I acknowledge with pleasure and gratitude the encouragement I have
enjoyed from Prof. J. T. Wiilard in making these experiments, also my
indebtedness to Mr. C. O. Swanson for much valuable data concerning
the samples.

Kansas Statb Aoricultural Collbob,
October 18, 1907.

THE EFFECT OF nTROGEN PEROXIDE UPON WHEAT FLOUR.

BT F. J. ALWAT and R. M. PllfCKNBY.

Received September 17, 1907.

Studies on the bleaching of flour by means of the oxides of nitrogen
have been published by Avery,* Ladd,* and Snyder* in this coimtry, by

> This Journal, 29, 571 (1907).

' BoU. 72, N D. Agr. Bxp. Sta. (1906).

' Report on bleaching of floor.

82 F. J. ALWAY AND R. M. PINCKNEY.

Balland^ and Fleurent* in France and by Brahm* in Germany. All these
have recognized that there is no appreciable change in the chemical com-
position, but as to the effect of the nitrogen peroxide upon the acidity,
the color, the absorption, the taste, the odor and the baking qualities
there is little agreement.

The flours referred to in this article are the same as those described in a
previous publication.* The unbleached flours of high grade had a more
or less yellow tint, the intensity of coloration varying greatly. The lower
grades of unbleached flour had a gray tint ; the amount of yellow coloring
matter in these, appeared to be about the same as that contained in the
higher grades obtained from the same wheat. These unbleached flours
when treated with liquids that are able to dissolve the fat, such as ether,
chloroform, benzene and petroleum ether, gave yellow solutions which
lost their color when exposed to the sunshine. The more yellow a floiu:
was, the more yellow was the solution obtained from it. The bleached
flours when treated with the same solvents gave more faintly yellow col-
ored solutions or even colorless solutions, if the bleaching had been carried
far enough. If the amount of nitrogen peroxide used had been excessive,
the solution was yellow. The fat from unbleached flours obtained by
evaporating the ethereal solutions was yellow, that from thoroughly
bleached flours colorless and that from pvertreated flours yellow or yel-
lowish brown. The lower grades of flour did not have their gray tints
weakened by bleaching, and many of the samples of bakers' grade ap-
peared more imattractive in the bleached, than in the unbleached condi-
tion, the yellow tint of the latter partly obscuring the gray color.

The Effect of Bleaching upon the Acidity of Flour, — Forty-nine pairs of
flours sent by mills having bleachers were tested, and of these, thirty-
nine showed no difference in acidity between bleached and unbleached.
In three pairs, the bleached flour was the less acid, the differences being
o.oi, 0.02 and 0.04 per cent. In seven pairs, the bleached flour was the
more acid, the difference being o.oi per cent, in five cases and 0.03 per cent,
in two cases. In the two bleached flours showing 0.03 per cent, higher
acidity than the imbleached, the amount of nitrites, expressed as sodium
nitrite, amotmted to 4.4 and 10.0 parts per million of flour. In the five
other cases the nitrites amounted to 18.8, 6.2, 6.2, 3.8, 6.2 and 3.1 parts
per million of flour.

An unbleached flour was treated with different amounts of nitrogen
peroxide. On the following day, the acidity was determined with the
following results:

^ Compt. rend., 139, 822 (1904).

* Ibid,, 142, 180 (1906).

' Versuchs-Anstalt des Verb. Deut. MuUer (1904).

* Alway and Gortner, This Journal, 29, 1503 (1907).

NItROGSN P^ROXID^ UPON WH^At l^LOUR. 83

Experiment No.

a . ■' III*. ^

1.2 3 4 56

Volume of nitric oxide used to each 1000 grams

of flour (in cubic centimeters) 0.0 25 25 25 50 50

Acidity, per cent 0.05 0.05 0.05 0.05 0.05 0.05

Portions of another flour, treated with different amounts of nitrogen
peioxide on October 16, 1906, were tested for acidity on February 18,
1907.

Number of Volnme of nitric oxide used to Acidity,

experiment. each kilogram of flour (in cc.) Per cent.

o o 0.07

•I 10 0.07

2 20 0.07

3 30 007

4 50 0-07

5 75 008

6 . 100 0.08

7 125 o.io

8 150 o.io

9 175 0.09

I ZO 200 O.IO

11 300 O.II

12 400 O.II

13 500 O-II

14 1000 0.15

Same flour bleached at mill 0^07

When the amoimt of nitric oxide used, did not exceed 50 cc. per kilo-
gram of flour, there was no appreciable change in acidity. Larger amounts
of the oxide caused an increase in the acidity.

Those investigators who have reported an increase in acidity due to
bleaching, have probably experimented with " overtreated " flours.

No difference in absorption or in the strength of the gluten was foimd
between bleached and unbleached flours.

Baking Tests of Bleached and Unbleaced Flours. — Loaves of bread
were made from 23 samples of imbleached flours as well as from
the corresponding 23 mill bleached flours. The samples came from
12 different mills and represented three grades, seven pairs belong-
ing to the patent, ten to the straight, and six to the bakers* grades. The
loaves were baked on four different days, along with a bakers' commercial
product, in an oven heated with wood. The color, texture, odor, and
! taste of the two members of each of the 23 pairs were compared.

In all cases, the bleached flour gave the whiter loaf. No difference
was detected in the texture, odor, or taste of any pair. The 23 loaves
from the unbleached flours weighed 12,857 grams, while the 23 loaves
from the bleached flours weighed 12,904. The volume of the 23 loaves
fiom the unbleached flours was 42,851 cc, and from the other 23 loaves,
42>735 cc. In eight pairs the tmbleached loaf was the heavier, in 14 pairs

84 ^' J. ALWAY AND R. M. PtNCKNSY.

the lighter, and in one of the same weight as the bleached. In twelve
pairs the unbleached loaf was the larger, in ten pairs the smaller, and
in one pair of the same size. The diflFerences in weight and volume of
the two members of any pair were small, and only such as might occur in
the case of two loaves made from the same flour. In all cases, the two
loaves were of practically the same size and weight.

Baking tests were made with samples of a flour that had been treated
in the laboratory with different amounts of nitrogen peroxide.

Number of flour.
I 2 3 4 5

Vol. of nitric oxide used per

kilogram of flour (in cc.) o 50 100 150 250

Weight of loaf (in grams) 545 541 533 544 534

Volume of loaf (in cc) 1919 1827 1981 2105 1981

Color of crumb Almost Pure Pure Ptu-e White

white white white white

All five loaves were baked at the same time. All were of fine even texture*
and, with the exception of the last, were of agreeable odor and flavor.
No. 5 had a musty odor and taste, but was eatable.

In the case of two bakings, i and 2, all the loaves made from bleached
flours contained nitrites, while in the cases of two other bakings, 3 and 4,
none of the 14 loaves contained nitrites. In the case of the first two bak-
ings, the average amount of nitrites was 0.8, while the average amount
of nitrites in the flour from which the loaves were made, was 6.0 parts
per million. There was no relation between the amount of nitrite in the
flour and that in the resulting bread (as shown in the following table)

•

Amount op NrrRiTBs in Bread Madb Prom Blsachbd Flours.

Number Series Series Partsper million of nitrite

of number number « * >

baking. of flour. of loaf. in flour. in bread.

I 5 4 3-4 II

16 6 4.4 0.6

I 10 7 3.1 1.6

I II 10 18.8 0.6

I 16 12 2.5 0.8

I 24 14 3.1 0.9

I 25 16 2.8 1.6

I 26 18 8.8 0.2

II 35 20 4.4 1.9

II 40 22 6.2 0.8

II 41 24 10. o 0.2

II 46 26 3.7 0.8

II 47 28 12.5 0.9

II 50 30 3-4 0.2

II 51 32 31 0.4

II 55 34 6.2 0.8

SODIUM NItRAtit AND CALCIUM CARBONATE. 85

Conclusions.

(i) The yeDow color of flours is due to a very minute quantity of a
ookred substance which is contained in the fat. When the fat is re-
moved, high-grade flours become white. Exposure to the stmlight or
treatment with nitrogen peroxide changes the colored compound into
one or more colorless compounds. Both the fat and solutions of the fat
from thoroughly bleached flours are practically colorless. Overtreated
(so-called "overbleached") flours have a yellow to brownish-yellow
cobr, and the fat, as well as the solutions of the fat, from overtreated
flours are also colored.

(2) Bleaching with nitrogen peroxide does not increase the acidity of
flours, while overtreating them with the same agent does.

(3) Neither the absorption of a flour nor the expansion of its gluten
is affected by bleaching.

(4) Bread made from bleached flours does not differ in weight, light-
ness, texture, odor or taste from that made from tmbleached flours; it
is, however, in all cases whiter, where high-grade flours are used. Low-
grade flours, when bleached, produce bread with an uninviting color.

(5) Bleached flours sometimes yield bread containing nitrites and at
other times bread free of nitrites. In all cases the amotmt of nitrites in
the bread is much smaller than that in the flour.

(6) The quantity of peroxide may be so increased as to seriously injure
the quality of the flour, but such a quantity at the same time unfavorably
affects the color.

(7) Low-grade flours when bleached do not resemble patent flours in
appearance.

(8) Many of the conflicting opinions in regard to the effect nitrogen
peioxide has on wheat flour are to be attributed to the investigation of
flours that had been "overtreated."

tabcmlatokt of agricultural chemistry,
Urivbrsity op Nebraska,
l4NcoLir, Nebraska.

THE POWER OF SODIUM NITRATE Aim CALCIUM CARBONATE

TO DECREASE TOXICITY IN CONJUNCTION WITH PLANTS

GROWING IN SOLUTION CULTURES.'

By Oswald* Scbrbinbr and Howard S. Reed.
Received October 23, 1907.

Investigations upon the nature and action of toxic agents upon organisms
have shown that there is not always a simple relation between them.
Although the harmful effect of the toxic agent upon the organism is the
main factor in the problem, it is no less true that the organism exerts an

^ Published by permission of the Secretary of Agriculttire.

&6 OSWALD SCHREIN^R AND HOWARD S. RE^D.

influence upon the toxic agent which may modify its action to a greater
or less extent. The present paper reports the results of a study of the
action of living plants upon solutions of toxic organic compounds with
and without the addition of sodium nitrate, calcium carbonate and other
substances.

It has been known for some time that the addition of a second solute
to a toxic solution often decreases the toxicity of the solution. Krdnig
and PauP fotmd that the addition of hydrochloric add and of halogen
salts to solutions of mercuric bichloride decreased their toxicity. Kearney
and Cameron* showed that the toxicity of sodium carbonate and of
magnesium salts is greatly lowered or overcome by the addition of calcium
salts to the solution. True and Gies* have demonstrated the same thing
for mixtures of the salts of heavy metals with salts of the light metals.
Pigorini* has shown that the toxicity of silver nitrate may be remarkably
lowered by the addition of the poisonous sodium thiosulphate, although
in this case there would be an actual chemical interaction between the
toxic agents.

The question arises whether the presence and activity of the or-
ganism does not play a part in ameliorating the toxic conditions.
Reed has ]x>inted out in a recent article' that such an action appears to
play some part in the observed antagonism between caldtmi and mag-
nesium. In the present study particular attention was paid to the effect
of the plant upon the toxidty of organic compounds with and without
certain inorganic salts.

Wheat seedlings of uniform age and size were employed in all the ex-
periments described in this paper. The seeds were germinated on per-
forated cork plates, which floated on the surface of a pan of water.* The
water cultures were made by placing the solution to be used into salt mouth
glass bottles having a capadty of 250 cc. Ten seedlings were inserted in
the same number of notches cut in the edge of each cork in the manner
described by Whitney and Cameron^ and by Livingston.*

The water used in making solutions was taken from the laboratory
still and shaken with washed carbon black. At the end of thirty minutes
it was filtered and was then ready for use. This method of treating
ordinary distilled water with an insoluble absorbent agent as described

> Z. Hyg., 25, I (1897).

* Report 71, U. S. Dept. Agr. (1902).

• Bull. Torr. Bot. Qub, 30, 390 (1903).

* Atti. r. accad. Lin., Gasse sd. fis. mat. nat. (5) 16 (i), 359 (1907).

• Ann. Bot., 2X, 565 (1907).

' For details of the method by which the seedlings were grown, the reader should
see Plant World, 9, 13 (1906), and Bull. 40, Bureau of Soils, IT. S. Dept. Agr. (1907).
' Bull. 23, U. S. Dept. Agr. (1904).
» Plant World, 9, 13 (1906).

SODIUM NITRATE AND CALCIUM CARBONATE. 87

by Livingston' is found to give physiologically pure water, and appears
to be generally applicable to physiological work.

In estimating the growth of the plants in different solutions, records
were kept of the weight of the green tops and of the amount of water
transpired. It was foimd, however, that for measuring the effects of the
different compounds employed in these experiments the transpiration
record was more useful than the green weight. This is undoubtedly
due to the fact that in many cases the root growth was more affected
than the top growth. This would seem to be a necessary consequence
of the more intimate contact with the toxic agent and doubtless for this
reason, the root has been a standard indicator of toxicity studies in plant
physiology. It has been shown that transpiration is more nearly propor-
tional to the growth of both roots and tops of wheat and is therefore
a better indicator of the effect on the plant than the weight of the green
tops. In Table IV both green weight and transpiration are given to-
gether with a photograph of the plants themselves, and a comparison of
the three records will serve to illustrate this point.

The experiments which are to be described were designed to study the
effect of various treatments in overcoming the action of some organic
substances which the writers have shown to be toxic to plant growth.

The Effect of Root Oxidation and of Adding Pyrogallol. — It was shown
that tyrosine lost its toxic properties as a result of oxidation incident
to continued exposure to the air. The aqueous solution of tyrosine,
which was perfectly colorless when first prepared, became dark brown
upon standing in contact with the air for several months. The exact
nature of the products of oxidation is somewhat in doubt, although
it is probable that homogentisinic acid and other oxidation products
of tyrosine are present. These oxidation products were favorable to
plant growth. The same relation was shown by the action of the
compounds neurine, choline, and betaine. Increased state of oxida-
tion was accompanied by decreased toxicity. It is not to be presumed
that increased oxidation of an organic cotnpotmd always produces a
decrease in toxic effects, but such a result undoubtedly follows in certain
cases.

Experiments were performed in which certain organic compounds
possessing known toxic properties were subjected to mild oxidation.
It seems possible that beneficial changes might be brought about by
oxidation. It has been shown by the work of* Raciborski' that the roots
of plants possess very definite powers of oxidation. The writers' have
shown, furthermore, that the roots of wheat plants grown in soil extracts

* BuM. 36, Bureau of Soils, U. S. Dept. Agr. (1907).

*BiiU. Acad. Sd. Cracovie. Math-nat. CI., 1905, 338; Ibid., 668.

•J. Biol. Cbem., 3, Proc., 24 (1907).

88 OSWALD SCHREINER AND HOWARD S. REED.

and synthetic nutrient solutions are capable of quite energetic oxidation.

Experiments were accordingly planned to show whether the solutions
of organic compounds would be less toxic to a second set of plants by
reason of the oxidation performed by the first set. Solutions of five
different compotmds were prepared in concentrations of i,ooo, 500, 250,
100, 50, 25 and I part per million. Wheat seedlings were installed in
these solutions and allowed to grow twelve to fourteen days and the
toxic effects noted. The first set of plants were then removed, the water
lost by transpiration restored, and second sets of wheat seedlings in-
stalled.

Table I shows that for three of the five substances the lowest concen-
trations causing injury to plants had been altered during the growth of the
plants. The concentrations causing the death of the plants were the same
for both the first and second sets of plants.

TabItB I. — Showing thb Effect of Plants in Altering the Concentrations at
wmcH Injury Was First Shown in Various Toxic Solutions.

(p.p.tn. ^ parts per million).

Lowest concentrations causing injury.
Solutions. First crop. Second crop.

Arbutin 25 p.p.m. Originally 500 p.p.m.

Cumarin i p.p.m. " 100 p.p.m.

Cinnamic add 25 p.p.m. " 25 p.p.m.

Sodium cinnamate 100 p.p.m. " 100 p.p.m.

Vanillin 50 p.p.m. " 500 p.p.m.

It will be seen that the arbutin solution which originally contained
500 p.p.m. was so reduced in toxicity by the growth of the first set of
plants that it was no more toxic than a fresh solution containing 25 p.p.m.
of arbutin. When cumarin solutions were replanted, the solution originally
containing 100 p.p.m. was no more toxic than a freshly prepared solu-
tion of I p.p.m. The cinnamic acid solution showed no apparent im-
provement when used the second time; it was thought that this might
have been due to the acid properties of the compound. The experiment
was accordingly repeated with sodium cinnamate, which was much less
toxic to seedlings than was cinnamic acid, as was shown by True,^ but the
growth of the first set of wheat plants did not perceptibly alter .the point
of injury for the second crop. It would seem from this that the activities
of the roots were not able to alter sufiiciently the properties of cinnamic
acid to change its toxicity to seedlings. The question whether the activi-
ties of the roots were not modified by the properties of the cinnamic acid
in such a way that the ameliorating powers were lost, seems worthy of
more careful study than we have been able to give it.

The toxicity of the vanillin solutions was likewise greatly reduced by
the growth of one set of plants. A solution which had originally oon-

» Am. J. Sci. (IV), 9, 183 (1900).

SODITJM NITRATE AND CALCIUM CARBONATE. 89

tained 500 p.p.m. of vanillin appeared to be no more toxic to the second
set of plants than a freshly prepared solution containing 50 p.p.m. had
been to the first set of plants.

The roots growing in the stronger solutions of vanillin oxidized some
of the vanillin to a dark purple dye, which was deposited on the root,
similar to the oxidation effects noticed when roots grow in solutions of
a-naphthylamine, benzidine, etc. The point may be raised, and with
some propriety, that the first set of plants absorbed and removed some
of the toxic substances from the solutions. While this may be true,
it is not probable that the amounts absorbed were sufficient to account
for such differences as were noted in the case of arbutin, the concen-
tration of which was so altered that the action of a solution originally
containing 500 p.p.m. was no more injurious than that of a freshly pre-
pared solution containing 25 p.p.m. That the concentration at which
injury was manifested was altered, while that at which death occurred
was not, is in accord with the assumption upon which the experiment
was made. The roots which grew in the stronger solutions were soon
seriously injured, hence any power they may have possessed to oxidize,
or otherwise to ameliorate toxic conditions had a very limited time in
which to act.

Without speculating too much upon the nature of the changes produced
by the action of the growing plants, it may be pointed out that they ap-
peared to alter materially the toxicity of three of the five solutions used.
Although the plants undoubtedly absorbed some of the dissolved matter
from the solutions through their roots, they accomplished a still greater re-
duction in toxicity by their activities in the solution. It would seem
that the oxidizing powers of the roots are able to change some of the
organic compounds into other compounds less toxic to plants.

The beneficial action of certain organic substances has been pointed
out in various experiments described by Livingston* and by the writers*
which showed, for example, that the addition of 500 p.p.m. of pyrogallol
to an improductive soil increased the growth of wheat plants loi per
cent The addition of small amotmts (2 to 10 p.p.m.) of pyrogallol or
.T-naphthylamine to the extract of an unproductive soil also has a very
beneficial effect. Organic substances, like the ones mentioned, cannot be
considered as plant nutrients; even if they were, the small amounts added
would be insufi&cient to account for the results produced. Their action
may be upon the plant itself, causing it to resist the toxic action of the
soil extract, or else these substances, being quite active chemically, act
directly upon the toxic substances in the solution.

It became of interest, therefore, to test the action of pyrogallol upon

• Bulls. 28 and 36, Bureau of Soils, U. S. Dept. Agr.
' Boll. 40, Bureau of Soils, U. S. Dept. Agr.

90 OSWALD SCHREINBR AND HOWARD S. READ.

some substance which was known to have a decidedly toxic action upon
plants. Cumarin was selected as the substance to be tried because it
had a nearly neutral reaction in aqueous solution. Four cultures of
wheat plants were set up in each of four solutions containing 75, 50, 25
and 5 p.p.m. of cumarin respectively, together with four cultures in dis-
tilled water. Half of the cultures received pyrogallol at the rate of 2
p.p.m. and the other cultures served as controls for comparison. The
experiment ran ten days, and growth was measured by transpiration.

The plants grown in solutions of cumarin containing 75 p.p.m. were
killed alike whether they contained pyrogallol or not. In 50 p.p.m. of
cumarin the growth was only 36 per cent, of the control plants, but the
addition of pyrogallol allowed a growth which was 82 per cent, of the
controls. The solutions containing 5 p.p.m. of cumarin supported a
growth which was 93 per cent, of the controls, and when pyrogallol was
added the plants made a growth which exceeded the controls by 16 per
cent. These results make it almost certain that pyrogallol had affected
the toxic action of the cumarin.

Pyrogallol has no direct value as a nutrient substance for the higher
plants, so far as known. Nevertheless, the growth of plants in the cumarin
solutions was benefited in each case, and in the solution containing 5
p.p.m. of cumarin the growth of plants was distinctly better than in
distilled water. This result would seem to indicate that the presence of
the pyrogallol had not only been able to overcome the deleterious effect
of the cumarin but also to bring about a slightly greater growth. From
this it would appear that the presence of an organic compound possessing
no value as a nutrient may aid the plants in overcoming toxic conditions.

The Effect of Adding Sodium Nitrate and Calcium Carbonate. — ^To study
this matter further, experiments were made in which substances com-
monly used as fertilizers were added to solutions of organic substances
which were known to have a marked toxic action upon wheat plants.
Forty plants in four cultures were used for each concentration of the
toxic agent employed. The fertilizer salts were added to two of the
cultures and comparisons were made with the two which received no
fertilizer material. Sodium nitrate and calcium carbonate werex the two
substances experimented with. The former is especially efficient in
producing improved growth when added to toxic soil extracts, in fact it
almost invariably produces increased plant growth when applied to
soils or water cultures. Calcium carbonate is likewise efficient in im-
proving growth and is found to be quite generally beneficial when applied
with green manures which are known to contain a variety of organic
compounds.

Solutions were prepared containing arbutin, cumarin and vanillin,
making them up in the concentrations shown in Tables II and III, Om

Fig. I. EfTect of calcium carbonate in overcoming the toxicity of (
Plants grown in: (i) Cuiiarin solution 25 p.p.m.; (2) Cumarin solution plus 200c.
p.p.m. CaCOj; (3) Cum^iin solution 10 p.p.m.; (4) Cumarin solutionplus 2000 p.p.m.
CaCO,; (s) Cumarin solution 1 p.p.m.; (6) Cumarin solution plus 2000 p.p.m. CaCO,;
(7) DislUled water; (8) Distilled water plus 2000 p.p.m. CaCO,.

Kg. 2. Effect of sodium nitrate in overcoming the toxicity of vanillin for Ihc
first and second crop of wheat plants, (i) First crop in vanillin solution; (i) Second
crop in vanillin solution; (3) First crop in vanillin solution plus 100 p.p.m. NaNOg
(0 Second crop in vanillin solution plus 100 p.p.m. NaNO,; (5) First crop in vanillin
MMBcin plus 2000 p.p.m. CaCO,; (6) Second crop in vanillin solution plus 2000 p.p.m.
Cfltt^: (7) Control in distilled water.

\ ,-

SODIUM NITRATE AND CALCIUM CARBONATE. 9I

series of solutions served as controls, to another sodium nitrate was
added at the rate of 100 p.p.m. of NO3, to another series calcium carbonate
was added at the rate of 2000 p.p.m., except that arbutin was omitted
from this series. The arbutin cultures were allowed to grow for nine
days, the cumarin cultures for eleven days, and the vanillin cultures for
ten days.

Tables II and III show the comparative growth of wheat plants in
various concentrations of the toxic substances and the effect of the in-
organic salts upon them. The growth of control plants in pure distilled
water is used as a basis for the comparison and represented as 100 in each
case.

Tabls II. — Epfbct op Adding Sodium Nitrate to Toxic Solutions. Rklativs

Growth Measured by Transpiration.

(p.p.m. — parts per million).

Relative growth.
Without 100 p p.m.
No. Solutions. nitrate. NO« added.

1 Control in distilled water 100 289

2 Vanillin, 500 p.p.m 25 34

3 '* 100 " 53 184

4 '* 25 " 80 238

5 " ID " 126 246

6 " I " 132 271

1 Control in distilled water 100

2 Arbutin, 500 p.p.m 23 27

3 '* 100 " 41 78

1 Control in distilled water 100

2 Cumarin, 100 p.p.m dead dead

3 " 50 " 47 53

4 " lo " 66 105

Table III. — Effect op Adding Calcium Carbonate to Toxic Solutions. Relative

Growth Measured by Transpiration.

(p.p.m. « parts per million).

Relative growth.
Without with aooo p.p.m.
No. Solutions. carbonate. CaCOs added.

1 Controls in distilled water •. . 100 209

2 Vanillin, 500 p.p.m 25 107

3 " 100 " 53 127

.4 " 25 " 80 183

5 " 10 " 126 184

6 " I "' 132 201

i Controls in distilled water 100

2 Cumarin, 100 p.p.m dead dead

3 " 50 " dead dead

50 "

4 25 " 36 58

5 *' 10 " 74 127

6 " I " 97 166

92 OSWALD SCHREINER AND HOWARD S. REBD.

It is quite evident from a survey of the results that these inorganic
salts had a beneficial action when added to the toxic solutions, the effect
being more marked in the case of the weaker solutions. The plants
which grew in the cumarin solutions containing 25 p.p.m. and less are
represented in Fig. i.

It may be noted that one of the toxic substances (vanillin) produced,
in the lower concentrations, what is ordinarily interpreted as stimulation.
The phenomenon of increased growth in the presence of weak poisons
is quite general and has been worked out in much detail by Raulin,^
Richards,' Ono,* and others. There can be little question but that the
increased growth of the plants in the vanillin solutions was caused by
some stimulating action of the vanillin upon the functions of the plants.
Turning now to the column expressing the effect of the inorganic salts
upon the vanillin, no stimulating effect will be noticed. None of the
solutions containing vanillin plus an inorganic salt produced an effect
upon growth greater than did distilled water containing only the inorganic
salts. In other words, the stimulating effect of the toxic agent totally
disappeared.

It would appear that the fertilizer salts either had an action upon the toxic
organic substances ameliorating the conditions for the growth of plants,
or that they acted upon the plants in such a way that increased growth
was possible in spite of the presence of the toxic compounds. Further
evidence on these points was gained from the results of replanting certain
of the solutions. The vanillin solutions enumerated in Tables II and III
received a second set of wheat seedlings after the first had been removed.
The water lost by transpiration was restored by adding distilled water,
and the plants were allowed to grow nine days. The behavior of the
second set of plants indicated that the conditions for growth were on
the whole, even better than those existing during the growth of the first
set, except in the solution originally containing 500 p.p.m. vanillin, in
which the plants were again killed. In the first crop the injurious effect
of vanillin itself was primarily on the root development and this condi-
tion was largely ameUorated by the calcium carbonate and sodium nitrate
when added to the vanillin solutions. In the second crop the root system
was, generally speaking, much improved in all three series, including those
without fertilizer salts, again showing that activities of the living roots
were able in part to overcome the toxic agents.

In order to ascertain whether the growth of the plants was a correct
indication of the presence of vanillin, the solution was submitted to a

^ Ann. Sd. Nat Bot. [V] zz, 91 (1869).
* Jahrb. wiss. Bot., 30, 665 (Z897).
' Jfour. Coll. Sd. Tokyo, Z3, 14Z (Z900).

SODIUM NITRATB AND CALCIUM CARBONATE. 93

chemical examination.^ Fairly large amotmts of vanillin were shown
by this test to be present in the solutions originally containing 500 p.p.ni.
vanillin, while only traces could be found in the solutions originally
containing 25 or 10 p.p.m. No vanillin could be demonstrated in any
of the solutions in which calcium carbonate or sodium nitrate had been
present. There had been a diminution and even a total disappearance
of the toxic substance.

An additional experiment will be described for the purpose of giving
a direct comparison between the growth of the first and the second set
of plants in toxic solutions containing inorganic salts. Three different
solutions were employed in this experiment: the first contained 100
p.p.m. vanillin; the second, 100 p.p.m. vanillin plus 100 p.p.m. sodium
nitrate; the third, 100 p.p.m. vanillin plus 2000 p.p.m. calcium carbonate.
A set of wheat plants was allowed to grow in each of these solutions for
eight days. The growth of the plants was of the same general character
as represented in corresponding treatments in Tables II and III. After
remo^ang the plants from these solutions the original volumes were re-
stored by the addition of distilled water to replenish that transpired by
the first set of plants, and a second set of seedlings was installed. Noth-
ing was added to the solutions except the distilled water. At the same
time a set of plants was installed in a new set of solutions, precisely similar
in composition to the original set. Accordingly, the plants in the new
solution represented a first crop in toxic solutions containing fertilizers,
and the replanted set represented a second crop in the originally similar
solutions. The plants in these solutions together with controls in pure
distilled water were allowed to grow ten days. The relative growth of
the plants in the various solutions is given in Table IV and the plants are
shown in Fig. 2.

The numbers and order of the solutions in the table correspond to
those of the plants shown in the figure.

It will be seen that the results of this experiment confirm those of
preceding experiments in showing that the toxic properties were ame-
liorated both by the action of plant roots and by the presence of inorganic
salts. Where the two agencies worked in conjunction, the growth of the
plants was the best. The second crop showed the better growth in each

* The method used seems to have been first described by Moerk, Amer. J. Phar.,
^St 572 (1892), and dted in Z. anal. Chem., 32, 242 (1893). It consists in decolorizing
tbe vanillin solution (if necessary) with freshly precipitated lead hydroxide, then
adding bromine water, drop by drop, until a slight excess is present. Ferrous sulphate
is finally added until the maximum blue-green color is reached. The test was slightly
impaired in solutions to which calcium carbonate had been added, by the yellow color
formed with the reagents, but the presence of sodium nitrate does not interfere with
the test. It was not found to be strictly quantitative although the amotmt of color
produced was indicative of the approximate amount of vanillin present.

94 OSWALD SCHRBINER AND HOWARD S. REED.

Case, owing to the action of the plant roots and the inorganic salts during
the growth of the first crop.

Tablb IV. — Relative Growth op the First and Second Sets of Plants in Solu-
tions Containing ioo Parts per Million Vanillin with and
WITHOUT Certain Inorganic Salts.

(p.p.m. « parts per million). Relative Relative

trans- green weight
No. Solutions. piration. oftops.

1 First crop in vanillin solution 45 100

2 Second crop in vanillin solution 93 103

3 First crop in vanillin solution + NaNO, 100 p.p.m 114 99

4 Second crop in vanillin solution + NaNO, 100 p.p.m. . . . 190 135

5 First crop in vanillin solution + CaCO, 2000 p.p.m 141 1 1 1

6 Second crop in vanillin solution -I- CaCO, 2000 p.p.m 166 100

7 Controls in piu-e distilled water 100 100

Another experiment upon the action of these fertilizers was performed,
using arbutin as the toxic substance and growing two sets of plants in
the solution. Two cultures, each containing 10 wheat plants, were
made for each concentration of solution employed. Arbutin was shown
to be decidedly toxic to wheat plants, killing at a concentration of
500 p.p.m. and injuring at 25 p.p.m. As before, sodium nitrate and
calcium carbonate were added to the solutions. The set of solutions
receiving sodium nitrate was accidentally lost before it was chemically
examined for arbutin and, hence, does not appear in the records of the
second crop. The growth of the plants in these solutions is shown in
Table V where the figures represent the relative transpiration. The
first crop grew eleven days, the second crop ten days.

Table V. — Relative Growth of Wheat Plants in Arbutin Solutions with and
WITHOUT Certain Inorganic Salts. Growth Measured by Transpiration.

(p.p.m. = parts per million.)

„ - ^, Relative transpiration.

No. Solutions. First crop. Second crop.

I Controls in distilled water 100

Arbutin, 1000 p.p.m 23

3 500

4 200

5 100

6 •• 50

7 " 1000

8 " 500

9 " 200
10 " 100

II " 50

+ calcium carbonate 2000 p.p.m. .
+ •* " 2000 " ..

+ " " 2000 " ..

12 Distilled water + " " 2000 "

13 Arbutin, 1000 p.p.m. + sodium nitrate 100 p.p.m

14 •• 500 " 4- " " TOO

15 " 200 " + " " 100

16 " 100 " + " " 100

17 " 50 " + " " 100

18 Distilled water + " " 100

100

125

148

154

109

147

146

SODIUM NITRATE AND CALCIUM CARBONATE.

It will be seen from these figures that the general order of results was
the same here as in the preceding experiment where a second set of plants
was grown in toxic solutions. In making the comparison it is of course
necessary to use the growth in '* replanted" distilled water as the basis
of the comparison and not fresh distilled water, since all solutions used
for second crops contained the waste products of the first crop. The
second crop was in all cases better than the first. Where the calcium
carbonate had been added to the lower concentrations of arbutin, the
second crop was remarkably good. These results bear out the results
of chemical tests to determine the presence of arbutin.

It was found that Pauly's* diazobenzene-sulphanilic acid reagent could
be used as an indicator for arbutin. While it is only approximately
quantitative it gave good indications. Pure standard solutions of arbutin
are colored bright crimson by the addition of a few drops of diazobenzene-
sulphanilic add reagent. Weak solutions of arbutin, especially after
the growth of plants, are strongly tinged with yellow. The results of the
chemical tests made after the growth of the first and second crops are
given in Table VI.

Tablb VI. — Arbutin Rbmaining in Solution aptsr Growth op First and Sbcond Crop

OF Wheat.

(p.p.m. » parts per million.)

ReaulU of tests to
indicate arbutin.
After first After second
Solutions. crop. crop.

Originally containing arbutin looo p.p.m abtmdant moderate

500 " abtmdant moderate

200 " moderate weak

100 " weak trace

MB •

50 none none

1000 " + CaCO,, 2000 p.p.m. abtmdant weak

500 " + " 2000 ** moderate weak

200 " 4- " 2000 " weak trace

100 " + " 2000 " trace none

50 " + " 2000 " none none

No.

I
2

5
6

7
8

9
10

An inspection of these results shows that the calcium carbonate had the
same action upon arbutin as the inorganic salts used in the previous
experiment had upon vanillin. There had been a disappearance of
arbutin from the solutions in which plants grew, but much more had
disappeared from solutions containing the inorganic salt.

In regard to the question as to how the toxic substance was caused to
disappear, three possibilities seem to present themselves: (i) The plants
themselves absorbed part of the toxic substance and oxidized some of
it; (2) the inorganic salts had a direct action upon the toxic substance;
(3) the plants and inorganic salts working together had a direct or in-
»Z. physiol. Chem., 42, 508 {1904); Ibid., 44^ I59 (1905)-

96 OSWALD SCHRBINBR AND HOWARD S. RBED.

direct action upon the toxic substance. The first possibility was tested
by the experiment ahready described in which a second set of plants
was grown in the toxic solutions. It was there shown that the plants
did have an ameliorating action, although they were only able to over-
come the toxic substances in the lowest concentrations.

In connection with the second possibility, inz., that the fertilizer sub-
stances had a direct action upon the toxic substance, the following ex-
periment will be of interest. A solution containing loo p.p.m. of vanillin
was prepared; a portion of it received sodium nitrate equivalent to loo
p.p.m. of NO3, another portion received calcium carbonate at the rate of
2000 p.p.m., and still another portion received both sodium nitrate and
calcium carbonate. These solutions were then allowed to stand ten
days. At the end of that time another set of solutions, exactly similar
to the first, was prepared. They were compared by growing plants in
them and using the growth of the plant as an indicator of the ameliorating
action of the salts added. Four cultures comprising forty plants were
used for testing each solution. It was thought that any action the in-
organic salts might have had, would be shown by the growth of the plants
in the solutions.

The results of the experiment are presented in Table VII.

Tablb VII. — ^Effbct of Tbn-day Action of Sodidm Nitratb and Calcium Carbon-

ATiS ON THS Toxicity of Vanillin Solutions. Growth Mbasurbd by

Rblativs Transpiration, (p.p.m. »> parts per million.)

Solutions Solutions

prepared prepared ten

at time of days before
No. Solutions. planting. planting.

1 Controls in distilled water 100 100

2 Vanillin^ 100 p.p.m 63 56

3 " 100 " + NaNO,, 100 p.p.m 127 100

4 " 100 " + CaCO,, 2000 " 125 166

5 " 100 " + NaNO,, 100 " + CaCO,,

2000 p.p.m 225 215

From these figures it will be seen that the action of sodium nitrate
was not the same as that of calcium carbonate. The latter produced
beneficial effects which were appreciably increased when it stood in con-
tact with the toxic solution for ten days. Sodium nitrate, on the con-
trary, did not show any increase in its ameliorating powers upon stand-
ing, in fact the solutions appeared to become somewhat poorer in both
the cases in which sodium nitrate was allowed to stand. The results
seem to suggest that calcium carbonate has the power to act independently
in ameliorating the toxicity of vanillin, and the lack of ameliorating
action of the sodium nitrate may be due to the formation of intermediate
compounds having as great or greater toxicity than the original. These
points seem worthy of further study.

SODIUM NITRATQ AND CAU^IUM CARBONATB. 97

Regarding the third possibility it might be said that the experimental
data thus far obtained, go to show that the physiological activities of the
plants were able to ameliorate the toxic conditions and also that the
action of certain inorganic compotmds in the solution is able to bring
about some improvement due probably to direct action on the toxic
substances, but far greater improvement is effected by the combined
action of plants and inorganic salts. In other words, the plants and in-
organic compounds working together are able to accomplish more in the
way of destroying a toxic substance than either can do working alone.
That such a substance as vanillin is actually destroyed, does not admit
of doubt, if any conclusion is to be drawn from the experiments previously
recorded. Whether the destruction of vanillin took place in the solu-
tion or within the cells of the wheat plants was not ascertained, but it
probably took place in the solution. This may be safely inferred from
the beha^or of the plants. Vanillin itself has a very inhibitive action
upon the growth of the roots of wheat seedlings in water cultures. All
plants g^own in vanillin solutions showed more harmful effects in the
growth of roots than in the growth of tops. When, however, fertilizer
ingredients were added in sufficient amounts to ameliorate distinctly the
growth of plants, the roots made as good or better relative growth than
the tops. (Fig. 2.)

Summary.

The activities of the plant roots are able to decrease the toxicity of
organic compounds to a certain extent, provided the original concentra-
tion of the solution is below that able to cause death of the plants. It is
probable that the oxidizing power of the root plays a greater or less part
in the process of amelioration. The first set of plants may have absorbed
directly some of the toxic material from the solutions, but the greatly
diminished toxicity of the solutions as well as the formation of dyestuffs
indicated that other changes had taken place.

The addition of certain inorganic salts to solutions of toxic organic
compounds was distinctly beneficial to plant growth. That the in-
organic salts and the physiological activities of the plant working to-
gether, had accomplished the destruction of toxic substances was shown
by both plant growth and chemical tests.

BURBAU OF SOXUI,

UmrBD States Dbpa&tmbitt op Ao&xcni.TURB,

WAtBIMOTOIf, D. C.

98 C. A. CRAMPTON AND L. M. TOLMAN.

[Contribution prom ths Laboratory op thb Bursau op Internal RBVBNue.

PUBLISHBD BY PBRMISSION OP THB COMMISSIONBR OP INTERNAL RBVBNUB.]

A STUDY OF THE CHANGES TAKING PLACE IN WHISKEY

STORED IN WOOD.

BT C. a. CRAMPTOZr AND L. M. TOLICAN.

Received August 24, 1907.

This investigation was planned and begun by C. A. Crampton, F. D.
Simons and A. B. Adams in the laboratory of the Bureau of Internal
Revenue, in 1898, in order to obtain more definite information concern-
ing the changes taking place in whiskey when stored in wood, in connec-
tion with internal revenue laws relating to the bonding of distilled spirits
and the sale of the same under government stamp. The analytical work,
was completed and manuscript prepared under the direction of L. M.
Tolman, with the co-operatian of L. M. I^w, A. L. Sullivan, E. H. Good-
now and L. B. Forst.

Some work had previously been done in the laboratory by E. Rich-
ards upon the maximum quantity of solid matter that could be extracted
from oak shavings by proof spirits.*

The experiment was conducted with the co-operation of a large num-
ber of whiskey distillers, who furnished the material, which was placed
in bonded warehouses under the supervision of United States storekeep-
ers.

Thirty-one barrels of new spirits were set aside at that time in as many
different warehouses and from as many different distilleries, and a quart
sample from each barrel was taken for analysis and sent to the labora-
tory. These packages were carefully sealed by the officer in charge, to
prevent any possible accident to the contents. The gauge of each pack-
age was taken, and the condition of the barrel and the warehouse noted.

Each year for eight years during the bonded period, the seals on the
packages were broken and a quart was taken for analysis. These sam-
ples were all set aside in glass containers and a complete chemical ex-
amination made in 1906, after all of the samples from the various pack-
ages had been received, except that the determinations of alcohol, solids,
color and color soluble in ether were made each year as the samples were
received.

When the chemical work was begun (1906), we had a series of nine sam-
ples from each of the thirty-one packages, all of which are exactly the
same age, the difference between them being the length of time each had
been kept in the barrel. Sample No. i, or the new spirit, had been kept
eight years in glass. Sample No. 2 had been stored one year in wood
and seven years in glass. Sample No. 3 had been stored two years in
wood and six years in glass, and so on, up to the last of the series, which

^ Annual Report, Commissioner of Internal Revenue, for 1889, page 60.

STUDY Olf WHISKEY STORED IN WOOD. 99

had been stored eight years in wood. In the discussion of these sam-
ples they will be designated by the length of time they were stored in
wood.

The first sample, which was taken from the barrel when it was placed
in the warehouse and which represents the fresh distillate, will be spoken
of as "new spirit/' although, as has been said before, it is of exactly the
same age as all of the other samples.

The objection to the plan of keeping the samples until all were col-
lected before making the chemical analysis is that we must not take into
account the changes taking place after the sample was placed in the bot-
tle; but it seems evident from the chemical analysis of the samples
that this factor, the change taking place in the glass, may be neglected,
as the results obtained on the new spirits, which had been kept for eight
years in glass, showed that practically no change had taken place.

The advantage of this mode of procedure can be readily seen when we
consider to what extent the methods of analysis of spirts have changed
in the last few years; the results would have been of little comparative
value if they had been made by different methods and different analysts.
Further, by making all the determinations at one time on the nine sam-
pks which make up a series, we are able to detect slight changes that
might not have been noticed if the samples had been analyzed eight years
apart by different methods and by different analysts. This is especially
true in the determination of fusel oil, in which case a uniform method of
analysis has made the results very satisfactory, showing the gradual
increase taking place from year to year. Even if the method employed
should later be shown to be faulty, the chief value of the results, which
is that they are strictly comparable, will not be lost. In the fusel oil
work a special effort was made to keep the conditions of analysis abso-
lutely tmiform for each period, so as to eliminate the effects of tempera-
ture, reagents, etc. All of the nine samples making up the series were
started at the same time and carried through to completion by the same
analyst. The results obtained by this plan are very remarkable, and
would have been entirely impossible with the var}dng conditions to which
they would have been subjected if they had been analyzed year by year.
The same plan was followed with all of the other determinations, thus
makmg it possible to show the very small differences which occurred.

Methods of Analysis.

The methods of analysis used were those of the Association of Official
Agdcultural Chemists, with the exception of the determination of fusel
oil, which was made by the modified Allen- Marquardt method. It was
found, however, after a large amount of experimental work which will
be published later, that it was necessary with this method to change the

lOO C. A. CRAMPTON AND h- M. TOLMAN.

oxidizing solution used in order to obtain satisfactory results. The fol-
lowing solution was finally adopted : 5 grams of potassium bichromate,
5 cc. of concentrated sulphuric acid, the whole made up to 50 cc* with
water.

It was also found that it was necessary to take smaller amounts of the
whiskey for this determination on account of the high content of the
fusel oil found in some cases, so that in all these determinations 50 cc
of whiskey were used instead of 100 cc, as given in the method.

The determination of the amount of color in the whiskies was made
in a half -inch cell by a comparison with the standard glasses of the Lovi-
bond colorimeter, using those of the brewer's scale, and the results are
all reported in degrees of this scale referred to the half -inch cell.

The determination of amotmt of color removed by ether was made by
the method of Crampton and Simons,* and results reported as per cent,
of color removed.

The determination of the amount of color insoluble in water was made
by the method of Walker and Schreiber,^ and results reported in per cent,
of color insoluble in water.

The following modification of the paraldehyde test was employed:
To 5 cc of the whiskey in a test-tube add 10 cc of paraldehyde, and
shake vigorously ; then add absolute alcohol, a few drops at a time, shaking
after each addition tmtil the mixture becomes clear, and allow to stand
for about ten minutes. A marked turbidity is shown in samples which
contain caramel coloring-matter. This turbidity is best observed by
holding the test-tube before and somewhat below a source of light. None
of the samples of this series showed any turbidity by this test, while
whiskey containing a very small amotmt of caramel will give a marked
turbidity, showing this to be a reliable positive test for caramel.

The samples were also tested by the Marsh method for caramel color,
which is as follows: To 5 cc of the whiskey add 10 cc of the amyl alco-
hol reagent, as prepared below. Shake vigorously for a few minutes,
and allow to settle. With pure whiskey, the lower layer will be per-
fectly colorless, while if caramel coloring is present, the lower layer will
be colored, depending on the amount of caramel present. The reagent
is prepared as follows: To 100 cc of amyl alcohol add 3 cc. of sirupy
phosphoric acid and 3 cc of water. Shake to form an emulsion before
using. The results obtained on the pure whiskies with this method
were very satisfactory, the lower layer being water-white, while with
whiskey colored with caramel, the color is concentrated in the lower layer,
so that very slight additions of caramel to whiskey can be detected.

* This Journal, 23, 810-813 (1900).

* Proc. Assoc. Official Agr. Chemists, Bulletin 99, U. S. Dept. Agr., Bureau of
Chemistry, p. 61.

STUDY OP WHISKEY STORED IN WOOD. lOI

The amotint of color in the two layers also gives a very satisfactory
indication of the amount of color due to the whiskey, and the amount
of color which has been added.

Statement of Results.

The results of analysis are reported, first, in grams per loo liters of
loo proof alcohol, and second, in grams per loo liters calculated to the
original volume of the whiskey. This second statement of results was
made to show how much the increase taking place in the various con-
stituents as the spirits aged was due to the actual increase of these sub-
stances in the barrel, and how much was due to the large decrease in vol-
ume which takes place at the same time.

For example, the solid matter in solution in whiskey increases each
year to a very marked degree, but when we consider the fact that the
volume of the spirit in the barrel has diminished about half during the
eig^t-year period of storage, it is seen that a very large amount of the in-
crease in solids is due to this loss in volume, so that the portion of the
table where the results are calculated back to the original volume shows
the actual increase of each substance in the barrel, while the first por-
tion of the table shows the changes as they would appear in the whiskey
when diluted to loo proof.

Unfortunately, the data showing the change in volume taking place
each year was not available in most cases, only the volume of the spirits
as it was stored and the volume left at the end of the seventh and eighth
years having been recorded, but from the results found on the few sam-
ples for which the data showing the change in volume from year to year
were obtained, we feel justified in plotting the curve of the loss in volume
from the three points which were determined, and so estimating the yearly
changes. All of the results, however, which are based on these calcula-
ted yearly changes have been starred in the table.

The loss of the spirits in volume is not due strictly to evaporation (since
it must be remembered that the barrels in which these spirits are stored
are made as tight as possible to prevent any leakage), but to the passage
of the spirits through the pores of the wood. The barrel acts in many
ways like the porous membrane of an osmotic cell and has a very decided
selective action on the materials passing through it, as is shown by the
large percentage increase by reason of concentration of the ethyl alcohol
taking place during storage.

The results indicate that water passes through the wood with much
greater rapidity than alcohol under the usual conditions in this coun-
try.

The results would also indicate, as shown by chart No. III., that the
higher alcohols are completely held back. The same selective action

I02 C. A. CRAMPTON AND L. M. TOLMAN.

is shown with the acids, esters, aldehydes and furfural. Apparently,
^ these substances are left in the barrel by the selective action of the wooden

j membrane. It is evident that the chen^ical changes taking place in the

spirit in the formation of acids, esters, etc., are dependent on the storage
of the spirit in the porous receptacle, as none of these changes occur
when spirit is placed in glass, tin, or even barrels, the inside of which
has been covered with paraffin or glue. It is also evident that the con-
dition of the outside of the barrel as to temperature, moisture, etc., will
have a decided effect on the rate of osmosis.

Explanatioii of Terms Used.

The definitions of the terms "Rye" and "Bourbon whiskey" are those
given in the circular issued by the Committee on Food Standards of the
Association of Official Agricultural Chemists.

" Rye whiskey is whiskey in the manufacture of which rye is the prin-
cipal cereal used, and Bourbon whiskey is whiskey in which Indian com
is the principal cereal used."

By sweet mash whiskey is meant whiskey in which yeast is used in
the fermentation of the mash, and by sour mash whiskey is meant whis-
key produced from a mash, the fermentation of which was started by
the use of spent beer or slop and barm from tubs previously set and fer-
mented.

The sweet mash fermentation requires much less time for completion,
and produces as a rule, a higher percentage of alcohol.

By the term "charred package" is meant a barrel, the staves of which
have been charred on the inside more or less deeply by the action of fire.
The charring of barrels in which whiskey is to be stored is an almost uni-
versal practice in this country.

In order to compare the effects of storage in charred and unchanged
packages, two samples of whiskey, Nos. 2625 and 2637, were placed in
uncharred packages.

"Proof" is the term used to denote the alcoholic strength of a liquor,
and 100 proof is equivalent to 50 per cent by volume of alcohol.

Description of Samples.

Following are the descriptions of the various samples, giving the kind
of whiskey, conditions under which it was stored, composition of the
mash used, type of still, with amount and kind of rectification, and yield
per bushel of grain, taken from the survey of the distillery, as made by
the Bureau of Internal Revenue, together with a discussion of the effect
of these varying conditions as shown by analytical results :

Whiskey Aged in Charred Packages.

No. 2598 (See Table I): Sweet mash, rye whiskey, entered into warehouse Jan-
uary 6, 1898, in a new, charred, oak barrel; warehouse, dry and above ground ; average

STUDY OF WHISKEY STORED IN WOOD. I03

temperature, 80° to 85° F. ComposLtion of the mash: malt, 1,064 pounds; and rye,
8,456 pounds. Distilled from copper still, capacity 1,125 gallons; redistilled in a second
still of the same capacity. The spirit is run directly from the still to the cistern-
room without any form of rectification. Yield per bushel of grain, 4 gallons of proof
spirits. The results show the effect of aging in a heated warehouse, there being a large
loss in volume, increase in proof and a great amount of acids and esters formed. The
product shows a very high flavor.

No. 2599 (See Table I): Sweet mash, rye whiskey, entered into bonded warehouse
January i, 1898, in a new, charred, oak barrel; warehouse, dry and above
ground, package was on the fourth floor; average temperature, 85^ F. Composition of
the mash: malt, 3,696 pounds; rye, 15,008 pounds. Distilled in three-chambered still,
with a doubler, all of copper. The spirit is taken directly from the still to the cistern-
room without rectification. Yield, 4 gallons of proof spirits to the bushel of grain.
This package shows a very great loss of volume on storage, 58 per cent, in six years,
and 61 per cent, in seven years, when the package was taken from storage and sold.
The proof also increased to an extraordinary degree, changing from 102 at the begin-
ning to 141 at the end of seven years. These changes are greatly hastened by the
high temperature of the warehouse, 85° F. the year round. The matured product
shows a very fine flavor and taste.

No. 2600 (see Table I): Sweet mash, rye whiskey, entered into warehouse Jan-
uary I, 1898, in a new, charred, white oak barrel, free of sap; warehouse, dry and above
ground; average temperature in winter is 77®, in summer 85® F. Composition of
the mash: 997 pounds of malt; 14,041 potmds of rye. Distilled in a large wooden still
of 10,000 gallons capacity, with a copper doubler of 2,000 gallons capacity. The
spirit goes directly from the doubler to the cistern-room, without any form of recti-
fication. Yield per bushel of grain, 4 gallons of proof spirit. This represents the
old style whiskey distilled in the large wooden still.

No. 2601 (see Table II): Sweet mash, rye whiskey, entered into bonded ware-
house January 3, 1898, in a new, charred, white oak barrel; warehouse, dry and above
ground, with an average temperature in the winter of 80°, in the summer of 90°.
Composition of the mash: malt, 2,844 potmds; rye, 14,032 pounds; molasses, 30 gal-
Ions. Distilled from a large three-chambered copper beer still, heated by direct steam;
capacity, 5,412 gsdlons, with a small copper still heated with a steam coil, having a
capacity of 1,200 gallons, which is used as a doubler. The spirit is taken directly
from the still to the cistern-room without rectification. Yield, 4 gallons of proof
spirit per bushel of grain. This package shows a large loss in volume and a high
temperature in warehouse, 80° in winter and 90° in the summer. This is also the
only sample in which molasses or any other form of sugar has been added to the mash .

No. 2602 (see Table II): Sweet mash, rye whiskey, entered in warehouse January
I, 1898, in a new, charred, white oak barrel; warehouse, brick, steam-heated, dry and
above ground; average temperattve during the first year 85°; afterwards, average
temperature 70° Composition of the mash: 3,625 pounds of barley malt; 24,375
pounds Michigan rye. Distilled in a three-chambered copper still, capacity 7,534
gallons, and a copper doubler, capacity 1,000 gallons. The spirits are rim from doubler
to condenser and then direct to cistern-room, with no other form of rectification.
Yield, 4 gallons of proof spirits per bushel of grain.

No. 2603 (see Table II): Sweet mash, rye whiskey, entered into warehouse Jan--
uary i, 1898, in a new, charred, oak barrel; warehouse, brick, dry, and above ground,
with an average temperature of 82°. Composition of the mash: malt, 2,520 pjounds;:
12,600 pounds rye. Distilled in a large wooden still, capacity 3,554 gallons^, bi^te<i

I04 C A. CRAMPTON AND L. M. TOUIAN.

by live steam, with a copper doubler, 66i gaOons capacity. High wines are diluted
to 80® proof, and redistilled in a copper still of 1,050 gallons capacity, heated with a
steam coil. Spirit from high wine still, goes directly to the dstem-room without re-
ceiving any treatment. Yield per bushel of grain, 4.30. gallons of proof spirit.

No. 2604 (see Table III): Sweet mash, rye whiskey, entered into bonded ware-
house January i, 1898, in a new, fully charred, white oak barrel; warehouse, dry and
above ground; average temperature, 80**. Composition of the mash: malt, 6,552
pounds; rye, 22,848 pounds. Distilled in a three-chambered copper beer still, which
is charged at intervals. The low wines are redistilled in a copper doubling still, capacity
1,389 gallons. The spirit is run directly from the still to the dstem-room without
any form of rectification. Yield, 4 gallons of proof spirit to the bushel of grain.

No. 2605 (see Table III): Sweet mash, rye whiskey, entered into bonded ware-
house January i, 1898, in a new, charred, white oak barrel; warehouse, brick, dry and
above ground; average temperature, 70®. Composition of the mash: malt, 2,072
pounds; rye, 12,040 pounds. Distilled in two wooden stills, one of a capacity of 3,827
gallons, the other of 2,625 gallons. Yield, 4 gallons of proof spirits to one bushel of
grain. This is an old style "pot still" whiskey, distilled in two large wooden stills,
and shows by the amount of fusel oil present that the spirits are not rectified to so great
an extent as occurs with the other styles of stills in use.

No. 2606 (see Table III): Sweet mash, rye whiskey, entered into warehouse
January i, 1898, in a charred, white oak barrel ; warehouse, brick, dry and above ground :
average temperature, 75^. Composition of the mash: malt, 1,680 pounds; rye,
9,520 pounds. Distilled in a large wooden still, capacity 9,695 gallons, with a doubler,
capacity 6,060 gallons. The spirit is taken directly from the doubler to the cistern-
room without rectification. Yield per bushel of grain, 4 gallons of proof spirit.

No. 2607 (see Table IV): Sweet mash, rye whiskey, entered into warehouse
January 3, 1898, in a new, charred, oak barrel; warehouse, brick, dry and above ground;
average temperature, 75^. Composition of the mash: malt, 1,260 poimds; rye, 7,140
pounds. Distilled from two wooden pot stills, one with a capacity of 2,158 gallons
and one with a capacity of 1,754 gallons, and a copper doubler. The spirit runs di-
rectly from the still to the dstem-room, no leach tubes or rectifiers being used in the
process. The yield is 4 gallons to the bushel of grain.

No. 2610 (see Table IV): Sweet mash, rye whiskey, entered into warehouse
January i, 1898, in a new, charred, white oak barrel; warehouse, brick, dry and above
ground, temperature varying between 60° and 75°. Composition of the mash: malt,
6,720 pounds; rye, 38,080 pounds. Distilled from a copper still, capacity 1,900 gal-
lons; low wine still, capacity 1,500 gallons, with a doubler. The spirit from the stills
is run directly to the cistern-room without treatment. Yield per bushel of grain, 4
gallons of proof spirit.

No. 261 1 (see Table IV): Sweet mash whiskey in which com has been used in
excess, entered into bonded warehouse January i, 1898, in a new, charred package,
third floor of warehouse; average temperature, winter 40°, summer 70^. Composi-
tion of the mash: malt, 5,376 poimds; rye, 31,360 pounds: com, 41,664 pounds. Two
large wooden stills were used. The spirit from the still is doubled in copper doub-
lers, but does not run through charcoal leach tubes going directly from still to dstem-
room. Yidd, 4 gallons to bushd.

No. 2612 (see Table V): Sweet mash, Bourbon whiskey, entered into bonded
warehouse January i, 1898, in a new, heavily charred, oak barrd; warehouse, dry
and above grotmd; average temperature in the winter is 45°, and in the summer
^^, Composition of the mash: malt, 8,960 pounds; rye, I1792 pounds; com, 78,848

STUDY O^ WHISKEY STORED IN WOOD. I05

pounds. Distilled in a oontinuons, copper still with a oolunm and doublet, receiv-
ing no rectification after leaving the still. Yield, 4.5 gallons of proof spirits per bushel
of grain. This sample is evidently very highly rectified in the process of distillation.
As shown by the analysis, it contains only a trace of fusel oil and suffered very small
I0S8 of volume in aging, perhaps due to being stored in an unheated warehouse. The
sample was practically cologne spirits aged in a charred package. The taste of the
aged whiskey was very little different from that of new spirits, although the odor
was much like that of whiskies stored in wood. Even the eight-year old sample had
the taste of spirits, differing in a marked degree in taste from the products distilled
in wooden stills in which there was little rectification.

No. 2613 (see Table V): Sour mash, rye whiskey, entered into bonded warehouse
January i, 1898, in a new, charred, oak barrel; warehouse, dry and above ground; aver-
age temperattu-e, 70^. Composition of the mash: malt, 3,576 pounds; rye, 29,344
pounds. Distilled in a three-chambered still with a doubler, capacity of the still
6,665 gallons, capacity of the doubler 1,770 gallons. The spirit from the doubler
is passed through a charcoal filter before being taken to the dstem-room. Yield,
4.5 gallons to one bushel of grain. This was the only sample of rye whiskey made
by the sour mash process, and the only one which was rectified by passing through
charcoal filters.

No. 2614 (see Table V): Sweet mash, rye whiskey, entered into bonded ware-
house January 15, 1898, in a new, charred, white oak barrel; warehouse, new, dry
and above ground; average temperature, 62^. Composition of the mash: rye,
5*950 pounds; malt, 1,050 pounds. Distilled in a large wooden still, capacity 7,766
gallons, with a copper doubling still of 1,065 gallons capacity. The spirits are run
direct from the still to the cistern-room without rectification. Yield, 4 gallons of
proof spirit to one bushel of grain. This sample shows to a remarkable degree the
fact that spirit kept in bottles does not undergo any change. The new spirit having
been kept over eight years in bottles was alkaline to phenolphthalein, due doubtless
to the water used in diluting the spirit to proof, as compared with the same spirit
which had been eight years in wood, which contained 91.9 grams of add per 100 liters.
This shows that the chemical action taking place is largely brought about by the wood
and char in the barrd, which act as catalyzing agents.

No. 2623 (see Table VI): Sweet mash, rye whiskey, entered into warehouse
January 28, 1898, in a new, charred, white oak barrd ; warehouse, brick, dry and above
ground ; average temperature in winter 65°, in summer 70^. Composition of the mash:
rye and barley malt. Distilled in a large three-chambered wooden still, capacity
I5fi77 gallons, with a copper doubler of 1,116 gallons capadty; redistilled, or run
through a copper still before going to the dstem-room. Yidd per bushel of grain,
4.5 gallons of proof spirit. Odor and flavor very good in matured goods; body, very
light.

No. 2608 (see Table VII): Sour mash, Bourbon whiskey, entered into bonded
warehouse January i, 1898, in a new, charred, white oak barrd; warehouse, dry
and above ground; average temperature, 70^. Composition of the mash: malt,
2,340 pounds; rye, 1,400 pounds; com, 16,500 pounds. Distilled in two copper stills,
capacity 2,303 gallons and 2,315 gallons, with twodoublers of 1,089 gallons capadty
eadL Yidd, 4 gallons to the bushd of grain.

No. 2609 (see Table VII): Sour mash, Bourbon whiskey, entered into ware-
house January i, 1898, in a new, charred, white oak barrd; warehouse, new, brick,
dry and above ground ; average temperature in winter 36^, in summer 80 ^ Com-
position of the mash: malt, 2,800 pounds; rve, 4,144 pounds; com, 21,056 pQvnds,

I06 C. A. CRAMPTON AND L. M. TOLMAN.

Distilled in a pot still of 2,000 gallons capacity, with a doubler of 1,000 gallons capac-
ity, both of copper. The spirit is taken directly from the doubler to the cistem*rooni
without any form of rectification. Yield, 4.5 gallons of proof spirit to the bushel
of grain. This sample shows a comparatively small loss of volume during the eight
years' storage, due to the fact that the warehouse is imheated.

No. 2635 (see Table VII): Sour mash, Bourbon whiskey, entered into bonded
warehouse February 7, 1898, in a well-seasoned, charred, white oak barrel, stored on
the third floor of a dry, brick warehouse ; average temperature, 70® to 80**. The mash
is distilled from double pot stills, each of 2,000 gallons capacity, and with a doubler
of 900 gallons capacity. The spirit is taken directly from the stills to the cistern-
room without passing through any form of rectification. Yield per bushel of grain,
3.75 gallons of proof spirit. Composition of the mash: rye, 2,688 pounds; malt,
4,144 pounds; com, 30,800 pounds. The low fusel oil indicates that in the doubling
of the whiskey a great deal of rectification has taken place. Distilled with wood fire
under doubler.

No. 2636 (see Table VIII): Sweet mash, Bourbon whiskey, entered into bonded
warehouse February 11, 1898, in a new, charred, oak barrel; warehouse, dry and above
ground; the water used for reducing proof is distilled; average temperature in the
summer 85^, in the winter 70^. Composition of the mash: malt, 1,288 pounds; rye,
1,400 pounds; corn, 9,744 pounds. Distilled in a chambered still, capacity 4,500 gal-
lons. The spirit nms directly to the cistern-room, no leach tubes or rectifiers being
used. Yield, 4.5 gallons to the bushel of grain.

No. 2637 (s^ Table VIII): Sweet mash, Bourbon whiskey, entered into bonded
warehouse February 11, 1898, in anew, charred, white oak barrel; warehouse, brick,
dry and above ground; average temperattire in the summer 80®, in the winter 48°.
Composition of the mash: malt, 6,496 pounds; rye, 8,288 poimds; com, 27,776 pounds.
Distilled in a pot still of 3,525 gallons capacity, with a doubler of 1,395 gallons capac-
ity. The spirit nms directly from the still to the cistern-room without rectification.
Yield, 4.5 gallons of proof spirit to the bushel of grain.

No. 2639 (see Table VIII): Sour mash, Bourbon whiskey, entered into bonded
warehouse February 8, 1898, in anew, charred, white oak barrel; warehouse, brick,
dry and above ground; average temperature in summer 60^-65®, in winter 55**-
60^. Composition of the mash: malt, 4,704 potmds; rye, 9,408 pounds; com, 32,704
pounds.. Distilled with a pot still, capacity 4,526 gallons, and a doubler, capacity
1,600 gallons. Yield, 4 gallons of proof spirit per bushel of grain. This sample shows
the effect of the temperature of the warehouse in loss of spirits in storage, also the
small increase in proof as compared with the samples that show high losses of vol-
ume.

No. 2684 (see Table IX): Sour mash, Bourbon whiskey, entered into bonded
warehouse March 31, 1898, in a new, charred, oak barrel; warehouse, frame, dry and
above ground; average temperature in summer 70**, in winter 40**. Composition
of the mash: malt, 2,800 pounds; rye, 1,680 potmds; com, 15,680 pounds. Distilled
in a large pot still, capacity 8,000 gallons, with a doubler of 1,400 gallons. The spirit
is taken directly from the doubler to the cistern-room without rectification of any
kind. Yield, 4 gallons of proof spirit to one bushel of grain.

No. 2683 (see Table IX): Sour mash, com whiskey, entered into bonded ware-
house April 9, 1898, in a new, charred, oak barrel; warehouse, brick and heated by
steam. Average temperature in summer 75®, in winter 40®. Composition of the
mash: malt, 1,176 pounds; rye, 840 pounds; com, 9,184 pounds. Distilled with a
three-chambered still to 60** pr6of, then doubled in a pot still heated by fire. The

STUDY OF WHISKEY STORED IN WOOD. I07

spirits from the doubler are run through charcoal leach tubes before gomg to the cis-
tem-room. Yield, 4.25 gallons per bushel of grain.

No. 2644 (see Table IX): Sour mash, Bourbon whiskey, entered into bonded
warehouse January 20, 1898, in a new, charred, white oak barrel; warehouse, iron-dad,
dry and above ground, not artificially heated. Composition of the mash: malt, 616
pounds; rye, 924 pounds; com, 7,952 poimds. Distilled in a pot still with a capacity
of 3,374 gallons, and with a doubler, capacity of 1,500 gallons. The spirt is run di-
rectly into the cistern-room, without any form of rectification. Yield, 4.5 gallons
to one bushel of grain. This whiskey sdows a slight loss of volume, as compared with
the other whiskies, and at the same time the changes taking place in the package
have been much less active, as shown by the slight change in proof and relatively
small amoimts of color, solids, acids, and esters. This emphasizes the fact that re-
actions taking place in the aging of the product are due largely to the temperature
and the rate of osmosis, or the passage of the spirit through the wood.

No. 2685 (see Table X): Sour mash, Bourbon whiskey, entered into bonded
* warehouse April 9, 1898, in a new heavily charred white oak barrel; warehouse, dry
and above ground. Average temperature in the summer is 80®, in the winter 45®.
Composition of the mash: malt, 1,960 pounds; rye, 7,840 pounds; com, 5,600 pounds.
Entered as a Bourbon whiskey, but contains rye in excess, as shown by the compo-
sition of the mash. Distilled in a continuous, copper, beer still, the low wines from
which are doubled in a copper doubler and are not rectified or refined in any way
after leaving the doubler, and before reaching the cistern-room. Yield, 3.5 gallons of
proof spirits per bushel of grain.

No. 2686 (see Table X) : Sour mash, Bourbon whiskey, entered in bonded ware-
house April 12, 1898, in a new, heavily charred, oak barrel; warehouse, dry and heated
by steam. Average temperature in summer 70**, in winter 65**. Composition of
the mash: malt, 6,272 potmds; rye, 5,432 pounds; com, 36,512 pounds. Distilled
in a beer still 23 feet high, with a doubler, the column being about the same height
as the beer still. Capacity of the still 3,480 gallons; capacity of the doublers 1,440
gallons. The spirit is taken directly from the doubler to the cistern-room without
any form of rectification.

No. 2689 (see Table X): Sour mash, com whiskey, entered into bonded ware-
bouse April 4, 1898, in a new, charred, oak barrel; warehouse, dry and above ground;
samples on the top shelf; average temperattu'e in summer 72^, in winter 42^. Com-
position of the mash: malt, 336 pounds; rye, 224 pounds; com, 2,968 pounds.* Dis-
tilled in a copper still with a column. The low wines are redistilled in a doubler which
is practically a large pot-still. The beer still has a capacity of 1,200 gallons and the
doubler of 800 gallons. The spirit from the doubler is leached through tall, narrow
vats, into which powdered charcoal has been packed. The whiskey slowly percolates
through this charcoal, taking several days to reach the cistern-room. The charcoal
used in this leaching is made by burning sugar-maple wood in the open air. The
yield per bushel of grain is 3.75 gallons of proof spirits. The very low content of
fusel oil shows that considerable rectification has taken place. The high ash of the
product indicates that in leaching through the charcoal large quantities of mineral
matter are dissolved. The high amount of solids is due largely to the ash. The
flavor is good and also the aroma.

No. 2689 A (see Table XI): Sour mash, com whiskey, entered into bonded
warehouse April 9, 1898, in a new, charred, oak barrel. Composition of the mash :
malt, 672 pounds; rye, 728 pounds; com, 5,824 pounds. Distilled in a chambered
still with a large doubler, 4,000 gallons capacity. The spirits from the doubler is run
through leach tubes, filled with charcoal. Yield, 4.5 gallons per bushel of grain.

Following are the tables of analyses I to XII:

IIO

C. A. CRAMPTON AND L. M. ITOLMAN.

d
11
§1

8
8

•n

•%g9% qsJVK

')N) 9p^qdpiw«d

a) 9iqni08ai jojoo
Xq paAoniM ioioo

'loqo9|v

iXnXB S« IIO (Mlljl

'ivaiyjod

*3})9dB SB 8a9)S3

.^:

s 3 ;i a 9 3 it

a 3 ^ 3 3 :s 3

o • o

OMQOOsOOOOO
O 00 •O O O »0 O f<5 eo

• •■■■>•••

M N ^lOiOM row (S

* * » « * »
•O r^ fO rCO r>»vO rf«0

» * « *
« t>»00 Cn "O

O !>• t^yO

X
CO

•loqooiB u

lAoiB BB (to (MUa ^

CO
IBjnjjnd &

'3U93B S

SB BdpAqapiv ^
'oildOB SB saa)8H Oi

'oi)ddB SB sppY

CivOOOOvi-iMi^MO

« » » » « ♦
«O00 -^00 OS ►^ •-< i-i ON

00 -^ o 00 ^ fo ^«o «o

« *< >0 ^ O^ ^00 ^ •-•

r«io^C4 ovw Ohi>»i>»

O O r*»oOoO NO ^

00 Ci <^x
W fO »OsO

00 Ci ^« Q O r^ r- ON

~ 00 "O t^OO 00

i-i M w c» rOfOf*^ro^

« tr> ON "^ to »o lOvo vo

ION onO n N lOlOcO

00 vo <^vp ^ r^ N 00 to
M CO »OvO »0 >0 r>.vO t^

X
(A

H
CO

H
(A

N NO »0 OnvO N On ^ r*
r^ t^ l>»00 OnvO

*^ *^ '^oo QNvp vg O f*^

•spHOS

'joojdooi

O) pd1BlllOI«3 J0[00

oqooooqoo

W lOl^OO ON»^ M fO^

O NOO fONOO fOrffO
O *^ "^nO nO 00 On On On

qqqooqqqo

• ^

. bp

a;^ 3

. ^3

::z:

:a:

: bo

. V a 3 3

3 3 3 3 3 3 3

4> 3 3 3 3 3

• o •

• 0

. M

• >o

::g

• r* «

O 0

0 0

• "^ 0 0 »o fooo NO

• M ^ ^ to fO «0 fO

• ^00

• • •

ON« 0

• • «

■ « •

0 0

•
•
*

■

•
•
•

»o

• • ■

. CO •

• ^3

•O 0

0 O 0

0 0

0 0 0 <O00 *- -♦
5f^^r0<0^<t

Oob

«OnOnO fON lOfON
00 OnI^I^OOOOOOO

O O O coco

• fO

•
•o

CO ^ ^-^ CO CO

# ♦ * «

OnnO ih t>.« 00

•

* *

. ON^

0 0 Onoo qc

4t 4t #

«0 «0 On cow

. I-I 00
• CO CO

*^*
N -^ r>i

NO On lO On i^
to lOvO lOvO

* * «

sO t^«0 ^ 1^

• O 0

•^W O 00 •- ON
lO "O lO lO lO ^

0 »0 PJ fO lO CO

. -^-^fN CO-^iO-^O

• On ^ O 0 •'i -^vO 00
NiClMrtNNNN

• CO • l>.00 •- N OO i-i

\0 r^ r^ ON lO N »OQC 10

00 00 t>«X CO On Onoo 00

iTi^ # * • «

OstioOO «OcOfO-0

CO • CO ^ m lo ^ lo

. Tf lO !>.« VO «0

• N Ci

lO o »0

N N N CO

M M M M
CO ON Tf O

• CO "-1 0 NO

. CO ^lOO

io«o >o lO
r>.oo 00 o^

CO
(A

* ■ •

•1 N li*

■

•

pi N PI

00 0 ONt*»^cor-*'0«

^lO ^lO lO

»o "Otfi

* *

•M ro M

*
o^x «

^NOOO

i-i N O

OO
l>»\0

^.^;t

^■•t^t * it

NO O NO 00 00

f^-C »

cO«0 00

io«o

0 0 0

■ ■ •

•

0 0c

\0 CO lO
00 O'OO

»4

irj ^i^r^

N- « ft f,

M M M M

O* N On CO

r>.

•

On 0 M

•-•N^NNNPIfOeO

00 "^ "^ m -^so ro ir. Qv

^ ON ^00 « -^^t^rs

"* l!r t^ ^ **^ «^>5 oo

M N r^ t>.oo 00 O ONn

N OnnO vo 00 *^ p« i^ CO

•* r-»o0 "H 00 M On co«0
CO ^«o «0 r^ t>. i>» r^

qqqqqoqoq

OWN OnnO ^nO CO N
d cOrfiOOO N ^VO

MMMMMP4P4M

O N CO Tf o '^ N Onoo

O 00 O « CO ^ t^OO O

M M »-l I-I M M N

qqqqqqri-qq

Sconq on«ooo ^ r- n
000»^»^«Nco

NO ^nO 00 ONvO « OC
On lO M N N>

»o r>.oo 00 ON

COOniOmN P<>00^
O»0 0

qqooooooo

t^vO CO ^ N O 00 i^iC

O »0 r>» Oi 0» p« lOgo
•^•^•-iiNiNpiwn

O On CO N O ON «0 'to

O 1^ c< -^ lO 1^00 0 «

M 1^ 114 M M N (1

ooooooooo

OOOi^^^ciM

IWJBq

r^ CO

00 CO

vO »0
P* «

'pooM af anifx

^' 8 0) 3

P«

■

3 3 3 3 3

^ c8 rt

^ « «3
0) >^ ?^

00 00
vO N

« N

(1 fi

r*ifl

N Pi

u u

3 3

4^3

3 3 3

M N CO ^ lOvO |>.00

9J >^>*
J5 « p« co^iosO r>i00 ^ ^ N ro ^ lO^o t^ao

STUDY OP WHISKEY STORED IN WOOD.

Ill

8
t

«
E

3
0

• >

• 4

-0

9
U

'1S91 q«i«K

•)f39 spi^qapiVJBd

01 diqniocnf iO|oo
Xq pdAomdJ <iO|03

'(oqooi*

'ivjnjjnd

sv tapXqdpiV

'*■• *** JTji 9 a si 9

:z;

«
• • O • O -00

9 3 3 9 8 9 9

JU9999999

:a:

999999

tfl

000000000

• •••■•••■

0>O00^Ot^«N'*5

O • • O^ t*5 O vO 00 t^
t>> • • vO 1^ 1^ lO^ sO

•

• 0

. 0 0

•
•

:<?

tfi\0

0 0 Q 0 0

?$?!J5

•

• •

- 00 \0

• • •

9 9

. 0 •

0 •

. 00 •

\0 i>.

Q ^ 9^ O^
00 i>.vO ^

»0 On

vO>2

9i)»3« «i uaiiB

-OTIMV 99 «ppy

'Ioqoo|«
\£wm SB i|o iMiid

a
b

M
fa

e
s

'iwi^jnd

fB B9pAqapiY

« * « « «

JL Jl JL ^ M

• "T^oo r-.vO 0 • 0*

vO >0 vO ^ vO ^
1 1« trj^^to 0 0

•

vO NO vO NO 10 10 >0
• • 0 »'i f^ 0 0 0

NO 00 00 00 !«• r^r^

10 to "«)■ to t^ r^ t^ t^vo

o o *^ <^

i^C^^OOOOOO
(^ f^** O 00 ^ fO »OvO

* * * » » *

0^*0 *OvO t^O « vO

ooO«oO»OON«roON
0NM>^0000mOOv

^ M M M MM

y O O fO lO ^ O w «o

« eO"T^t^oooo
O ^00 «0 -^ ONOO f*i

(O ^ r»00 00 00 00 00 1^

* « « * * «

-fSMIOOvO^*^

»0 OS

W l>»00 00 r*- !>• !>. lOvO

# vb # # # #

VO 00 »0 ^ ^ OS «ONO o
CJ fi O OS !>• !>• I>k00 OS

00 00 (O C4 OsOO 00 00 r^
M f*5^iO»0»0»0»0«0

MQO OnOvo r^Mso M
M M 10 10 to tOsO >OvO

• • Os»0 t>. 0
■ f^ O^ ^00

N M g

M M S
M M (^

• ^for^oo »o

:°° w

• W Ci « W f*^

(/I
X

•

• <*5-*^ . -^

• 0 « fO • fO

. M M M • M

• 1? H

• 0 CI « N 00

• 0 Qi

O i^'OO O^oOvOOOS O«0»0c0r>. «OvO

vo w »ooo os ON o» Os r^ '
toioioio>o>oto>o '

00NO'*000«O 'MON
M vO 00 U^ fOOO • OS fO ^

NONOoooooor>> -OsM

'9pa3B'BBU9)BH

'31)»B IB BppY

t^ • 00 00 r^ M M

M . \o i>»oo ON ON

•■Plios

jooid 001
o) pa|B|n9iB0 joi<o

•joaij

ON • lo r^oo 00 NO o t^

«0 • to O «0 ^00 ^no
• t»00 00 00 00 0« On

oqooqoooo

10 M Os ^00 M On O O
1^ ON « «0 lO ^«o so

MMNWMWCHpi

Ooo r*fO-ON« ^t»M-
O «OsO 00 00 OS d 6 d

MMMMMMriM

qoooooooo

8tONO ON M w lOsO d
O00l^*-4I^MC(

lO «O\0

CO iOnO vO

0 M

»o «

N N

est

OS l^sO

ON CS| 00 vO NO 00

'^^oo

OnO

M
M

M M

0000

«o

N 0

»o

0 t>.

M \0 COOO

r* t^ao 00

«o

fo t>. fo 1^ »o 1^ 0

«o

»0 M !>.
lOvO NO

M 00 ^nO
r>. t*.oo 00

rO
OS

•

0 0

a •

■

•

0 0

■ •

a « ■

"*0 0

• • •

0 ThvO

• ■ •

1^ t^ t>.

^I0i0i0«0i0i0i0

0 0* fO W 00

l^fOO.

!*• »OvO

0 «o»o

M M M

S8>

M M

ON
fS|

M 00 0
M M M

OSM

CO
10 so

00 CO
^COO

OS
u

t^ ON OS ON Os ONOO 0

•M

^ ^

«00

M «

CO t>.00

fO ■'t w

Ci w

CO CO NO 00

c>< «

CS|

Csi « W

^00 «

r^ M

CO r^ r*vo

<s» ^

CO N

^ « 0 CO

sO 1^ 1^00 00 Os

^»o M

M 0

u^nO >0 m

10 r* Os 0

1^ CI 00

'^ 10 10 t>> r^ 1^00 00

0 0

• a

0 0

• a

•

0 0 0

• a a

SO N « ^NO M fO «0 ^
CNW iO»O00 OnOnc*

^4 ^4 M »i« ^4 »« Ci

OnOnO Onm •- co^N
OsOOO w N I0»0«0»0

OOOOOOOOi>»

piw^r>.MN^O'*
0000^»^»^c<w

CO « O "^nO r^ ^ «ooo
tONsOOO ONM ro»Or>.

O r^

MOOsOOO COM ^OsiO
tOOO »^ CO ^sO sO r*00

t>.io000«O00«

8M fO »O00 N sO O ^
OOOOixMMM

'lUJBq M

N On

^82

vO M
coo

COM V r^

«o

so CO • »0

. . 'O .
Ci

SO '<^

o
S5

o
2;

^38999999 ^'SS993999 __

55 M « cO^iOsO t>.00 J5 M « rO Tf lOvO 1^00 5? m « co ^ lOsO t^OO

■pooiiiiiamix ^* I 89 9 9 9 9 9 "^ ^* S i» » 3 9 9 9 ^* 5 S« 9 9 9 9 9

112

C. A. CRAMPTON AND L. M. TOLMAN.

« 1
8

*)89) quvpi

*)S9) »piCll9pi«J«d

«t « «t « « a

« 9 « if 9 9

« « « « 9 «| 3

155

8*5)

»8

•c

a
8

o
C

«
g

'JK91B41

n| 9]qniosa| jlo\oo
Xq psAomaa jopo

'loqooiB
iXinv s« {}o lasQii

•

• o

• 0

. 0 0

•
•

:>2

0 ■-•

0 0

0 O 0

0 o^ «o « ►* « «

«* fO I*- -* ij- '^ ij-

•

•N «0 CO

• ■ ■

. o

o o

«*5 !>. 10>0 00 ON »0 fO ■*

isinjjna

§s

4JMMMMMMMM

^<OeocOfO« «

*9IY93B

ev sdpAqapiY

'Of^MV SB SJd^Sa

OO^OOO lOM oqo O^m
MMroOOOONO^O

M M M M

^ t^OO >0 ^ W t^ it" On O

CO to CO to ^ "^ "^

_ *
QMCOiOCHCiroCICt

'dpaOB SB Bppv

*loqo9iB
I^OIB SB ifo idsn^

•IBjnjjnj

SB sspifqdpiy

'3|P9B SB SJ9;8H
*9l)93B SB Sppy

-spiios

'JOOJd 001
on p9)B|n91B9 JOIOO

'jooja

COM "*ior«.o^O "^^JS
"«*■'+ -^ ^ ■* »o lo lo S

OOOOO OslT) ti ^^o

cor«ior^o^Mvo t^ONrtS

M MM-*

*-'''-''''!

oor*ioOO»-"WON^^
M«oo««c<c<«coP

M M M « M «

r^ ^00 ^ « 00 O NO M "
On O O M »0*0 CO r»oo

O loiooo O O CO On r«»

CO COOO M NO ON CO ON M

<^ ^ lo lO iono no r^

000000000

■^locoo O «nO «^r«.
ON CO too 00 On O M

M M M M M M M

OOO Onm -^iOcOm CO

O ON ON C« N ^ lONO NO

M M M M M M

oqooooqq<<f

8c« cocoior^po On O
OOOOOOOm

• ^ • CO •

• cO • CO •

• • • •

0 • OnnO
lO • lO >o

OOO 0 0 0

O 0 0 0

O cOmoooOnOnOnOnO

M TfCOCOCOCOCOCO

o1>%^^

«^ Ooo NO

0 0 « « M

"8??"8

^ » » »

* I^COt^
. 00 M On On
. CONO "OiO

. lO© 0

• lO lO lO

.^oooo

^ On w r«.00

. © © ©

2 4r^d d

^ MM

•000

• M M M

* « ♦ *

M -^CO M 00

0^% ^C*

vo »o r-NO 00 ON ON r^ On
>0 "«*■"*'*'*'* '^ -*

^* * * *

0 M o 00 00

ON"* IT) CO

NO 00 COOO 00 r^ M t^ o

«0 -^ Tf -^ 'i- lO -^ lO

•O ON© ^«

*iO© CO

M qq lo M o C4 lOvo Q\
O 0 CO ■* Tf CO CO ionO

. ^0 O 0
• cor^r^r*

• «^00 00

• 0 0 0 0

■ 0 © ©

g M 0 MOO

• MOO ■*

ij »oo6 c« c«

^ MM

• CO CO »0

. M M M

C10000 lor^o Ono m

nOno « N r^»OTf^«
ioioionOnO no r*

Ooo 0 cor*TfM MOO

NO OnOO »0 t"^ On r* »0 CO
CO <^ lO to ionO no «^

OOOOO

• ■ • • •

•

• ©

. o ©

•

■

•NO

■■S

. « M

l^co© ©

•

CO ONM lO^cOiO©
M CO^COCOCOCO*

•

>*nO CO

• • ■

• •

^On-**-

■ ■ ■

■H-NC Tf ci 1^ r* ON M CO

8 ■*♦ * «^0 ONOO
M <^NO ^cor^NNO *

o«««c««cocici

«- <*NO* « «

*no m ON*t«>ci ro m

^lONOOO C4 C4 c« M o

» ♦ « ♦ * « _

Q ^OniOm OnCIOO *

jC ONOO OniO CO M -^eo
« '♦^ *nO no no "O io

* *^» ^t*-.*^-..
.NO -tM^O »O«i*-*00

:^NOciOcoc«Mioio
a * »ono no no no »o io

r^ij-^MOoo M j>»o

coOco«oc*0©©M
*00 00 00 On O M M ft

On On M cOnO CO
0 CO ONOO ONiONO M 0

CO On On © »ONO On lO "i*-

* iOnO m iOnO nO no r>

M M M M IN M

ooMr^i-iior«©oOM

\0 M *oO M CO IO * On
CO to iOOO 00 00 00 t^

, 00 © CO ONOO "t^ CO On

^j^OO iOmOO m iOiOm
a IOnO t<« r«00 00 00 On

©©©©©©©00

M coco»ocoO cor*M
M onci ^lor^ON© N

M M M M M C< n

•pxisq
at saonsS 9a{AL

*pooii nf 9m\x

NO
M

M . © IOOO IO IOOO

*ON .iOMvOMr>.*

r^co '©ooNOiocow

rh * -^rococococo

• 00

00 IO

OMJ>»MCOONMOr*

nO©©©©©©©no

§2§n? = SS

R : : : : : rg'R

•

NO © r^

'<4* cow

•

• • ••••••

CO ^ r* On M IOOO ON M
MMMMC«CIC4CO

©©©t^©©oqt«

« « CO IOOO COOO M *

©©©©©MMMCt

Cioo

M Oi

« if « S 9 3

^ M M to* IONO r>»oo

0»

^ M « to* IONO t>.oo

9 « 9 9 a

55 « « CO

3 if 9 9 9

IONO r«oo

STUDY OF WHISKEY STORED IN WOOD.

113

'%S9% qsJVK

.*^!l 9 « 9 « 9 I

8
t

-)ta) ap^q»pi«jBd

* S^9 9 «l a 9 is «

U9)BM

• 0 "O *00

• fO • ^ ... |>» to

• « . to • • • ^>o

Aq p»AOinaj| aoioo

0 ON«-l 000000

OCHooOoo^OQOr^
11 CO -* ro CO ^ •<*• -*

[ *ioq(K>i«
liCm* SB 110 \9ma

0 0^00 Mvboo io\oo (O

•

0 r«*»ooo w gvit'^c*
00 r>. i>. !>• r*vO r^ r- r^

-{minpna

♦ » ♦ ♦ * ♦
• vbo i>.cot^^c« 0

** «

.MMMMMMMM

§•6

• 0 >0 ^ CO CO •O
0 • ^»0<0 »0 0 M »0

r©

SB s^pAqapiY

vo • r>- 1^ r^ «^oo 00 00

rami

^« « # # «
"^•or^cooM^-^oovo

OJ5
8

*3f)93B SB SJ9183

• ••••••■•

■*\0 00 0 0 O^ Cv 0 1^
Ci eoco « « CO «

« * « « « «
so \0 0 »00 W vO !>• CO

[ '3!)99B SB SppV

CO 10 d •4' ^ »oo6 1^ M

M M M M M M Ci

■

-(oqcoxB

VO^OC* •VOOOMMM

•

iXmB SB i]o \98ns

(>»oo 00 • 00 00 O^ 0 0

M M

• 0 0 w M lOvO Ooo

'IBJitjjnj

.M«C<C«WCiCO«

*3JU3B

SB s^pXqapiv

O^00 0 iOO MVO tOsO

to t^OO 00 0^ On 0 M M

M l-« M

COOO « CO COO woo On

'OpaOB BB SJ99SH

rhNO •-• ^NO r^ On •-• 0
CO CO CO CO CO '^ ^

»ooo 0 ■* CO 0 NO r^ t^

'OpaoB SB sppv

coiOMvo r«.ON-*"*M

M M M M CI CI CO

000000000

•spiios

NO r^ CO « r^oo no ^ Q
M 0 CO "O «ono «^ on 0

MMMMMMMd
• M

'joojd 001
o) pa^BinofBd ioioo

. 1000 NO r»oo ^M 00

• r*ON«Mwcicoco

0 onono 0 0 0 0 on

•JOOJJ

« c0^t^« tM^WnO
0000>^i^*-*C^c*<

•WJJBq

On OnnO

^ NO CO

-^ CI W

•

• •••••••

* MriT ••••••

-pooM ni »inix

> <U S)4 3 9 « « if

^ M « CO ^ »oo t^OO

114

C. A. CRAMPTON AND U M. TOLMAN.

•\W9% qSJVIV

■)t») 9pXqapimi«a

8'S

I
8

•c

'J9q)d
Aq p9Aom9J JoioD

'loqoaiB

*l«jnjand

99 sdpAqaptv

'3|)99« 8V BJ9)Sa

&
6

m
u

*9D39« ra sppv

'loqooiB ^-^
\£m9 ffB |(o iMnd ^^

•B t9pAq»piv

9 «t ;i 3 3 a 9

• o • o • o • o o

• 6 • Q • »o • !>• Ci

• to • vO ■ >0 • O ^0

O^OOmOOOOO
O «*500 « « O W CO O^

-. ft. t^ft. <L "It «

O 00 COOO O O (OOO M

• ••••••••

vO O 5^ ^ lOvO 00 Ni M
iH«000«00ON

SP.

«i 4 9 ;s

•.is

•55

M M M M M M

ob«

0 0

0 0 «

M M M

M M

S-^trS)

•^%8,

COCV

M M 0
M M M

O Ov

* « «

•o r* «

« * *
M lOOO

0 •*

ro « O

vO VO NO

On t^
to to

too r* <o cooo

M r^

0 « 1-4 T|-r»vo »o
CO -^ lO to lO lO lO

•

0 COM

to ooo

• • ■

0 ■♦

• •

• 0 -0 -0 • 0 o

. r^ . «o . C4 . « VO
to • Tf -to • NO ^

O r^oo O 0 0 O 0 O

0 tooo 0 ^ "^ 0 to to
M to -^ tO tO -^ -^ -^

_ » « * » * * ^ ^
0 « '^'* to too OnO

to l>» ON r» On to OnoO to

\o tototor^tototO"^

• O NO # # # #

• 00 00 tovo i>>^ NO VO

• OOmmmihmih

•« ♦ ♦ * » *

• NO to « '^OO On tooo

• « NO ON ON On On OnQO

• * $> 5) t) %-loNO 0

• 0 ^ « to "^to H-00

• to to ■* <* -^ ^ rh "*

'. * VO ^% ^^ »o '^^

• w ^00 w ON 0 M to
^■^^^■^toto to

000000t>.0 0MtOM

c* ^r^oo to ^ w M r»

00 00
M M to M C4

M Wt O l^tOONl^tOON

NO NO l> t».00 t». r^OO On

Q 00

ONQO r« On M M d to

r^QtowOtor^to

«t>.OM««CIN

M NO O 00 On r«00 NO

M r*i>0 tot^QvO
CO CO ^ to to tONO ^0

tooo 00 NO NO O to M

o
n

§
PQ

fOl>TfONQ tOON-^

^Thto -^tONONO r^

S?=

. bo . bp .bo

•

• o

• 0 •

•
•

• to

• to •

CI M

to NO

to to i>. 0 0 0

ONM -^oooooo
•^■^toeo"*'*

Oob

* «

t^«o

• *

• • •

• •

VO ON to to M OnnO t>»00

00 r^oo 00 00 r- t^oo r*

* » « » ♦ ^

M ^ ^00 'O to O

^ M to 0 "^ ^P O
to toNO so r»<0 »^

•

• •

o o

odd

* « «

* w t^OO

to n 00 M

Vss

c^ r»oo

r» t^oo 00

* * * « _

M >0 00 « to On

ONO

M r>. OnnO 0 00
-it "* to to <* to

c«

•00

ON lO to On

r^ to-*
to-^

• M M VO NO O

. to ^ ^ -^lO

•

■

•

0 O

• ■

0 0 0

• « ■

•

^ to "^nO tooo On to ^

OOOO OnOnOnOnOnm m

VO •nOnOOOOO OnOnm

d '^ 666666*^

O ^00 00 tooo too O

« tOt>.00 ON On O P» ti

M M M

to ^00 0» "H t^ r^ OnnO

ONOt^WNO«OOP«0^
M ^lO^-^totoiO

OC<tO .NOMMtOW

r>.^io -Ot^^tO"^

to ^ 'NO tONO NO l^

OOOOOOOOO NOOOOOOOOO

^ »H tONO to ^ « to M
w On « ^ tONO 00 On M

M M M M l-« M M

'JOOld OOI

o) pdiBinoiBO joioo

to On OnnO to O 00 r^
t^OO O •-• CI ^ CO ^

•H'^'^O'^OOtO^

r»ONM w fOto^^

i>»vo NO ^ « tooo 00 to

M 00 O to Ci -^ "^ tONO

OnOn 'NOOOOOOtO

tONO • 00 M d M ci

'JOOId

OOOOOOOOO OOOOOOOOO Oc^OOOOtoOO

S? S? If V? Cr'Ofi dwto c>iwfOcoTt-r>.ONd« SJCSUZf VPS? ^00
OOOOOOmmhi OOOOOOOmm OOOOOOOmi-i

IWJBq
DftaoiiBS aufAL

00
O

to ON

M 00

to W

3 3 ;t 9 s a

^ « Ci to '^tONO «^00

^ON
M 00

2L0
to to

:s ;s 9 3 a 3

J5 w « rO'^toNO r^oo

■

'^J-

00 ON
00 00

m On
to ti

c3 c3

^ M M co^toNO r«.oo

0
u

STUDY OF WHISKEY STORED IN WOOD. II5

*M91 qU«N * ^^ 33999a ! ^3 999999 *4|999999tt

• 60 'to * bo

119) spXqdpivivj •^9999999 •a|9999999 •O9999999

Z> -2 -2

3
«•
0

••0
• 0

•

• 0 •

0-00

•

• cO •
> 10 •

tooo »o 0

0000

i>.oo r^oo ro »o »o ix.

w «*5 «*5 CO ^ ^ ^ -^

•

• • •

• ■ • ■

m aiqnioBOf 10103
Xq paAomdj -10103

'loqoaiv _

linn ■■ no lasn a «0>^»00o0 -fOwr^ oovp«^N -mnvO vO»0»OmQi-i iooo \Q

mmmmm mm mmmmm mmm mmmmmmmmm

♦ * « _ * * * * ^ ^

» » -^oo* tttno -,» g^oooo* tj Q Q -, *^» <L <t -t ♦

go »Ooo Mr>.poo H"/. 0'^0^«0f*50 SJ'^r»OoO«»00

•

. 0 •

0 •

•
•

• M •

to •
10 .

10 >o

lOO 0

0 0

Onoo 0 fO « m n
to ^ to **5 fO fJ

•

• • • •

• ■

•

• 0 • 0 •

0 0

•

• to • 10 *

>?s?

M 0 0 0 0

0 0

« 0 t^ 0 0
•0 ^to-^^

•

^1,1,^1,^

0 0

• •

•OMOOModo £66060060 a

n t9pifq9piv

M (>4 ro M u
■ O to « M c

000000000000 t>»t^r^

9l'|»3B O'^OtOClMC^r^CI 0^tO ^VO O • tOoO »0 « O^ O "^v© O vO tO O

>All9DtV O -^ ^ O^ »O00 to OnoO

'snao* SB Si»)«S . o^ to -^ CO r^oo mm'i-L ^^6i^6N<^c>4rito •o>6 n v6 -^ to>6 -4-
s S n^'^^^ioio^Mdcitotoco^^^^ .•-'NtOTi-^'^Tf^

^O«Ot^OcO«00 O^00 ,^ ^ O **^* M00r>'«>0^< NOniOmcO-^OxW

'3pMBs« tppv -, VO ^ CO CO 4* 1006 c^r^ Mt^M^-ir^dNOM^S

loqoaiB ^

h 6 6 6 6

0 0 0 0

♦ « *
* 10 ^^

OMO'^vO 0

* _

• 10 0

• COOO »0

0 r^ t^oo 00

• OnOO 00

N On •-• tO^

r^vO On r>«

^ TfO <^ On
«« CO COCO

"^ N « to

« * * «
c« 0 tooo M

* *

00 r>. 0 On

tOCI "^"^

r» ON ON«

to -^ N vO 0

• M w 0

• • • •

M VO t^vo 00 r>. r^vo >o
t^>0 •^ONMOootoO non>-" tovO r^NO ONr>. 00ONONOON«r»o«

Svo ^ to CO -^ IOOO ONi>» Mt^M^-^r»ONONM^ r«.wi>.M^«co»o-^
CO ^>0 to 10 10 10 to h to Ci -^ "^ Tf -^ ^»o ^ "^ ^ »o 10 »o 10 to 10

H 00 tovo too 'Ci^ONW co-^Nvoo -mwo 5 •-•r»t^«^iOMO\t>.

lOiOiOh to Ci "^ ij- Tf Tf ^ 10 ^

-_ -_ _ ci^ONWto-^NvOO-«wo5'-« __

M 10 ^ ^vO • !>■ O to ^ « tooo Q Q^ • r» M M (>5 "^92 **> *^* vO 00 00 PI
wtocowc* .wcocoCQ «>r* r»oo o*-MCi« 'MOwtOMcowio^

^ _>^MMMM MMM ^MMMMM CICINsT'^'^^^'^'^'^'^^^

3" &-ooooooot^o«g ON o ootog^ SS'Ss'i^ ^ *

p 0 'funpn^ S -ooiHMMWNOtogg t^oo ononononO« * o^^*^ *^* *^^ °°

J^^l tf) !nMMMMMJJ^.bnnnnnnMMSflMMMMMMMM

e
s

_ Om>^»-IMMMM 2*3660000'-'>^ 2 PMMMMMM

'Dndov ^ ^vo o to »o t>.w o r* ^ onoo c««on -vor^O^ 0»i> ^00 m \o »o co m

■^ ^ "^ ^ t>.00 OOOnOnOnOnOO 55 O «>00 OOn -tHWCOfi ONO«>t>.OONONONON

A MMQ M.MMMg M

S^O**«»*fOCOCi«io9 O00v0v0r»»0«-«^g^00 OnvO vO •-< « ^oo M

■^IIW •• «**I»H a On "^ t^ ON »o r>. ^vo 4- 5 '*n6 CO COOO r* r^ tovp o tor^^cicocoo Onoo

Q « "^ ^ 10 »OvO vOvO O WWCO'^^iO tovo ^ PQ •-• W tO to to tovO to lO

oq «

ON COOO ^00 OOOmcO m^CO^mOOMOn «^»o <*>00 ih \o r* ^ O

"apaav »• SPPV lo iono ONwtocor^ON MoocoN^Mt^cioo r^^^oo^-^ON-^^

CO ^ too vo r« t^ o CO w >o tovo NO «^ r^ -^ »o too vo vo r* r^

000000000 000000000 oqooooooo

"■PH**® ^ r>.00 CI N N vO ^ to OnvO vOtOOr^OtOto O ^NO ^ 0> ^ O vO vO

On M ^100000 O « •-' "^00 t>.0 f* ^»0 •"• »^ to to tooo O "^ "-i

MMMMMCiCl MMMM««N« MMMMM«MP«

M 00 ■^ M ON ^ "^.^

•jOasd OOI O « r* On On to M |>\0 O*^'* 'COtOt^^O OvOOOC«m^I>»cOOn

01 p»)«|1K)|«3 J0103 o t^-OO OnmcO-^^iO 6a0ON*co^ lOvO 00 O 00 m W CO -^ "^VO to

|i4MMMM .MMMMM MMMMMMM

O O to O O «OvO 00 O^or^OOOOOO r»W0000O00to

TOOiA "HNwcotoix.ONNc^ MM«co tooo "-"co^ Q«5jc«»co4-«or»to

0000000»-«M OOOOOO'-''-'^ OOOOOOOO'-'

OOtO CI vOtO to tOM

'piivq 00 c^oo O vO>-« ^ ^^

m moiivai aaiM io:i::::m6n r^::::::oor^ v6--:i^:«6

vO ■* <*5W ^ ^ « « ON "^ ^^

^ . . . CO '^

NO : NO :":':; ^

« . . . . . \ '. : : y* ......... <^

• •••■•■••■ • ••••••••••

O o o

Z : : : 2 • : • ^

U3 . • J3 * . ^ . • •

•poo* «! »<««<X ^p^»»,,,, '^^P^sjsaa.i ^$«»

9 9 9 9 9

V >>>* Z >*»>N 4» >^5>N

55 « « cO^tOvO «^00 2 « M to ^ tovo r>-« 2 m w ro^iovo r^oo

Ii6

C A. CRAMPTON AND U M. TOLMAN.

•

r 8^« a « « tt « It

•;»; 9piCq»pi«j«d

* 8^« » « » «: 9 9
•2

or aiqniotni joxo3

• 0 0 0 • 0 • O 0

• t^ O 00 • VO • CJOO

• i/> lO to • lO • O 1/3

o
S

« tt « » 9 « «

l^n 9 9 «| 9 « «

• 55

•j»q;»
▲q paAomdJ io|03

'loqooi*
|XmB s« |io |9in^

'ivjnjjna

s« 09pAq9piY

iXuiB 8« iio \a^A >r

0
>AI

0<0tf)000000
O O (OOoovo M o^r*

O «0 -^OO 00 CO M t^ ^

r^eiQ cioo ^«vo ^

t^OO GO 00 00 00 00 (>» fo

•

0 0 0

• 0

• o

•
•

0 coeo

• vOvO

M lOO

0 o

0 O

M « O M « fO fO

•

• • • • •

• •

• 0

• O • 0

. 00

\ lO 'lO

• « m

Oso^ O 0 0 O 0 0

O «*i

^0 woo

O t^oo <*5 c^ r^

% 0«

mmmO^^^OO

00 O^OO M t^fOM <Ori

MMMMMMMMH

*9I)99« BB tppy

• lo M r>i<^vovO fO Q

• 0 t^vo o^ ON r«» r*oo

• ••••••••

OOOOOOOO

^^M t>»f0»O»OM lO

_^» • « « « *

eooo »OsO W »OvO ** ^
C( VO O (^ O^OO «^ «^vO •

^ ♦ * ♦ <L ft * ^ ^ S
>0 00 00 kOO^O M «1-^ ^

fOtOM eo(0^4-ior« ^

^O^mOOxO^MOOm O

^ to t^oo e^ »OpO VO vO ^
i^oo ooooOOOOO ^

4j vO lO to r<«vO tOvO O
CI»OC1fO«N«M«

»o M CO r^ o»ao >0

l>. fO t^ M ^ 0» M

lO t^oO QsOO 00 O^

S6----M-M

(i M M^ ^T^^o o

0 lOiOi-i O^^'^O »o

« «>• 1^00 r«.«^t^t^vo

^lO ^^M « iO« lO

>0»OtO«MWN<iO«

oooooo5>©v^»ot^^

^ 2. i*fi. !2 2 !i !!: ^

0*0 «o« eo«o«o« On

OOOOOOOO

* ♦ *Jk <L • * ^
00 M coioo ts.o 0

30C<COO>OOOC1mO
O t>> foQO On On On 0^0^

^ On « ** to On ONOO r*

00
ON<

00 O
OnC*

* joojd 001

O) p9)BII10lB9 JOIOO

NOO 00 O M O

.ooo<-**->m6m^

CO lO On « CO t>iNO t^>0
CicoONMC<«r«c«c«cO

ON m O r«oo r<« ^ M r<« o

M t^ »0 w ^ »onO i^ t^ '^
w M eo«OcOcO«OeOfO

'^ ro •!«• ^ ^ «^ O O r^

«OnO '^OO O « «0 On eo
« CO CO -^ •It' ^ ^ lO

oqqoooqoo

OniOiOOnm io« t^C«
nOOOOnOCIcO^iO

M M M M M

c00nQ000»O«O

ONONoq^-^coO"*
O "^ "O r>. t^oo ON o d

lO ^ to io*o «>»nO (»

§6

r« r^ CO M so r«oo 0 c

CO ^-NO NO QnnO (^flO V)

lO to to^ ^ ^ ^ ^ c

0% O t^ • <> « o^«

M O r<«oo > ^OnO N
M lO to to • toio«o^

00 to CI O tOONONO ^

so c« to ^ ^ «000 NO cv

M Ci « COCOCOCO^*

0 0 0 0 Q
OnOO m m to <o onoo n5

M r^oO On On ON On ON On

NO to O CO cOOO O 00 to ^
PQ

toto r^oo ON M lo CI «

« « "^ ^ ^ to too NO

(O COOO O OnnO O m \0

NO CI t^co^-'i'toci «
M ^ ^ to to tOvO NO t^

OOOOOOOOO

tOOO O M O ON N4 M 00

00 M d CO -^to t>-00

00 t^ O
O On On CO to *i« ^00 **

O OOO OnOn« tOcOfO

• 00 r*_

•00 (>«Cl

• tor* Ci

O Q OOOO

M O M M <0

0 0 «

ON 0

i^tor^oo

M t>.^00 0

0 r«*t^oo

M M M M n
M M M H M

c« ONOO OOO r^ONt^n

COtOONCOONONCOl^N

to lo to t«* r^ t^co 00 00

ONOO to 00

• to COOO 0

M lONOtO

• ^ COOO «*>
. so t>iC«O0

ONO 0 o

O O 0 00

cOnO ^ on CO CO to f*)0
« t<«ONM fOtotot^r"

M M M M M M

C4 CI O
O ^lOONtOOO 0 ««

O ^ r>-oo ONM CI M (1

totoqqooNoot^ oooooonooo ooo t^oo o o « «

'xoojj cpciMQ««»^c«ci cocifO^to r^oo « ^ m ci co co ^sd 00 0» ?
* ** OOOOOOOOO OOOOOOOmm OOOO O O OOO

toco

'lAUBq
uf snoiiBS aufAl t^ I ^ I I : I co ci

4 '^ coco ^

^ 00

NO I I ! ^0

« ! ! ! ! «

ON to

00 00

I0«
I0«

oONO r<«
! ! CI !

cort

• • • •

^ . • •

3 :iiS

•pooii ni 9niix fe88ti««««itt

S >N>^

#i^ M CI (O^toto r^oo

2 M c« CO ^ toNO r^oo Si^ M d «o ^ lOto t*^

STUDY OP WHISKSY STORBD IN WOOD.

8
8

fli

6
lo

a
u

It ■

M
U

■

E
&

m^iiuvR

*)I9) ap^qdpivivj

-jaqia
Xq p»AOwa4 -10103

1oqo3i«
iXm SB ||o lasn^

*IBaii|jna

SB s^pi^q^piv

"spdaB SB sja)sa

•apwBSBsppY'":

'loqoofB W
lioiB s« ]|o lasn^ ^

IBiiyjiid B

SB Sdp^qapiY
*9p93« SB sjaisa pq

*9p»3BBBSppy

•sp!io6

:s!:

« tt tt « «i ;i tt

r^B B « « B tt B

• o • o • o • o o

• to • \o ■ ^ ■ ^ ^

o cht>.o o q o o o

6 t^ fi 6 f^f* r^oo ^

O »« ceo S> % 00 >0 CI
ODMOOOO M ^** tn6\

.4>«litt«BBB

• 55

. V« B B B B B B

•

• O

• 0

•

0 0

•
•

• so

■■S

•
•

•

tor* 00

o o

O O

coo « t** O

ro «

Ooo

• •

« * • «

Ooo VO M

« ■ ■ •

■ • •

q q

o o

90»Om fOt^cOONOO

fOCiMCO^M 00^O00O^O^ 0^>0 l^

* « «

M M 0\ M

0 M

M 0 M t^

•>o Ooo

* #

^r^ ON CO

• (<< M M

« w 0 0

M M M M

• 00 ^ (Ooo

•sa^$$a$5

• • ■ •

* * «

t^ « loeo Ci

* « * *

• C0»0& vO

. 0 M 0 »^

♦ *

. M M M M

0 O^O^^ «

M M 0 M

io»o«o o

•1 o^^r*

^r^i^oo ONONoooo 1^

OOO esq o q q q

OOm»oo\coc«c«oo

O O O to ^ 0^^0 CO M

f* w O^OO VO »o to ^ '^
^^fOcOfO«0«OcocO

8R8 8lla2-8

g ON o^ Onoo r^ t^ r* r^

W^ 9 • • • • • •

♦-00000000

* *^ QnO Oncom OnO

CO 4^ 5f lo »0 ^ lO «

O ON W ^vo CO lO M CO O

• •••••••• ^fi

!>• r*oo «o ^ M CO « o

ON CO C<l lO ^ ^ J>»vO 00 wT

•OOMCO^CIVOQO to
• Omi^mmmmJ^

NOVO ^

j>»i^ cococi cOfOt^ I,

Ci M (0^tOiOiO<0

lor^m (>«ioo O ^

^ lO to lO^ VO NO NO W
M CO cOvO O O Ov M

'JOOJd 001

01 pa)Bind|B3 J0|03

700JJ

'laiuBq

Bl SnonBJf 9D|A1

. 00
VO

CO^MOO « OnOnO
CO ^ to 10 t^vO v5 r^

qqqqqoooo

00 vQ On to CO O CO t>.NO
M ON M ^\0 O •^ CO CO
•-• M M W N CI «

O NO ON CO M O Ci OnnO

'^^^r^t^ONtocoM

M ^ 1^ ^VO vO NO VO NO

« « « « « *

, t^ cOvO OnnO 00 tovO

S cOno on i>* r* 0 0 0
13 Ci 10 to tOVO NO NO NO

CI to M 00 M <^ CO ^00

10 c^ r^ M r« (OvO CO w
00 r«»Ov« co^« ^^

• «^i>.000Q«^0

• « M w '^^^coto

COON

ONCivo Ci M ioonOnO

COOO 00 O M M M M M

OnnO «i^ C« vO t« M O •-•

CON© 4-4-dr^(>»Qci O
M ^ 1^ lOoO 00 00 ON On pq

. r«*^« O OnO -^^
!h ^MOO cOoO M \p M

a CI \o NO r^ r<«oo 00 On

qqqqqqqoq
00 w i>« 10 10 to cooo CI

M M ^r>>oo On CI ^10

MMMMMMdd

O O CO CO On CO r*vo 00

« « «
• «t CI 00 00 00 r^ CI

• 10 to to iO to tONO

* * #
• COt^NO •i*'^«^ On

• 0 0 M CO CO to «o

N4 frH ^4 ^4 ^i^ ^4 P'^

• ^-^oo -^ CI too

M • CO06 COOO O ^nO
• CI CI coco ^ ^^

00000 totOMVOvONO

• ■••••■ •

« CI CI CO CO to ^no r^
^ -^ "^J" •* ^ -* 4 '^ "^

^ ^ Ovoo _ r^ Q
O O ONONO O'O

OtHMOO^^O^
O •ONONONOr*t>»t>»

O c^ O CI ^nooooo on O Onm CI cOtONO t-» r»
qtoqiotootooo toOtotoOOOt^r»
OOOOOOOmm OOoOO>-i>^mc^

to M

CI 00

CI
CI

too

ON to

CO CI

•

•
•

• ■ •

to 10 NO

« • • •

r^r^r^oo

■

CO On CO

•
•

M
M

to «>• !>•

M M M

CI
CI

•

^WNOvO

•
•

tow Ooo CO
CI co-^-^j-to

M
NO

CO 'I- to ON

ON IN <^^

** W « CI

ONNO

0 CI
coco

On Ovoo

r^io ^ON

1^0

'^J-NO

M M

t^ OnO

M M «

»i4

M
0

CO CO CO

000

M M M

■

•
•

CO
VO

•

•
•
•

•

•
■
•

•

'pooii ni 9«u,

* b a
^ M CI co^iono f^OO

NO
CI

00 NO

•??

to
to

•s

s ;■ 9 a « 9

00
VO
CI

•

B a B a a 9

41 >%>*

S24 M c« co^iovo r-00 55 « « co^tovo r*oo

4) >\>^

Ii6

C A. CRAMPTON AND h. M. TOtMAN.

a
t

m
a

•c

o
C

«;

•^•1 qu«K

*;») 9pXq9p|mx«d

•J919M.

u\ 9iqn\OBn} 10103
Aq paAomaj 10103

'loq<»i«
liCinv IB i;o f9«n^

*I«injjinH

SB BdpAqapiy

-9n»0B 0B BJd^BH

«««««««

« « « V 9 « «

!^« « « 9 «t 9 a

•

000

. 0 •

0 0

•
•

• 10 •

rooo
\0 10

lOlOO

00000

0 roooovo M o^r^

•

fO^OO 00 «*3 M

• •

« « 9 9 9 « 9

0 0 0

o o

O «OeO

too 0

0 0

« 0 « « to «*5

•

00 ^-^

• • ■

•

• ■

•

t^«OciOO^«vo^
r«oo 00 00 00 00 00 t<« r«

• t/) M r>i^ ^ ^ CO

• QNrOQsO^^^O

• 0 r«»NO o^ o^ t^ t^

00000000

^^M r>.fo»o»OM 10

^» ♦ * * ♦ ♦
COOO lOvO M >OVO M

mmm0m^*^0O

rj>9 9 « 9 9 9 C

.V9 9 9 999«

* O * O • O '00

• NO • ^ • O • ^V)

• ^ • 1/) • ^ • ^ IC

oso^ 000000
d <o^d woo d it

^ * it "fL*^* ^ -,

O t>.00 eO cor>.0 0 M
00 o^oo M r«>cot« t<)n

MMMMMMMMM

V vO 10 10 f^sb *0^0 O
fliOMfOMWWww

QOmmmmmmm

« « M ^ ^ W^SO O

O lOiOi-iON^^O »o

N «>• r^oo r* r>. r* r«*vo

* * *^*
r* cor* M

*919»9B SB ippy

'loqooiB *^

liCoiB SB IJO IMtlH >r

•IBinjjnd g

*3})MB
>AI

« vO O «^ O^oo «^ r*vo .\0»oeo«i^*^i^»0« Ji

OOOOOO^O^^iOfH^^

vd d 't>d «o CO r* ^ d^ m

^» * « « ft* ^ ^

vO 00 00 »OOnO •^ ^>0

cOiOM roco^^ior* ^
« cocococ^cococo C

OMH O O^O^t«00 M Q \0^ COM cOtO(Or« 0^^

^ CO r^oo coiooovo^o ^

t^oo 000000000 •>r

M M M M M M

'dddddddo

S> * %>to% t*« 0
00cirOO>O00C«M(i

0 t^ t^OO O^O^O^OkQ^

« • • « « « ^

r<« f« CO M Co rooo 0 «

CO ^^ NO Qv>o 1^00 ic
m 10 lOvo >o ^ ^ ^ ^

O&^r* .%* O^rt

c< 0 t<*oo • ^0^0 N

M 10 1010 • IOIO««

00 lOM 0 10 O^ 0^0 ^

^ ON « "^ CO ON 0^ 00 «^ ^»
WMCiClc0^rh^iO><

<» W O C4 O

I-I M 6 M ^

00 O
ONM

^ OnO O
lO^lO to

O O O M

5«

r«vO

On «

'jooid 001

0% pd}BinO(B3 JOfOO

•JOOid

*^JBq

CO to On CI CO «^nO r««NO

CICOONMCIMMCirO

On « O «^oo r^ ^ w r* O

M f* «o M ^ tONO r^ r^ '^
M M cococococococo

^co-^^^r^O O r*
cO^ ^00 O « lO On ro

800000000
Cxioioonih ioc« r«»«

>O00 ONO C* CO^IO

M M M M M

^cOOnQOOO»Oio
OnOvO O ^^coO ^

O ^ »0 «>• t^OO On d d

M M

lOiOOOOONOOr*

COC^wO'^^^'^CICi

000000000

SOmmmmmmm^

__ 00 0 0 Q S5
OnoO m m 10 CO OnoO O

M r>«O0 On On ON ON On ON

NO M SO ^ ^ tOOO NO 0^
M Ci w f^c^coco^^

MMMMMMMMH

• 00 t^
•00tN>OOOO000

NO «0 O CO COOO O 00 10 «5

• • • • W

«ONO r^oo ON M lo c» c«
w « rh ^ ^ «o «ono no

CO coco O ON^ O •-• vO

no « r^co^^coci «
M ^ ^ 10 «o »ONO NO r>-

ooqoqoooq

•000 O C< d On M M 00
00 M M CO ^tO t^OO

00 r>> O
O O^ On CO »0 M ^00 w

O OOO OnOnw cOtOcO

• • • •

0 0 «

r* »or*<»

w r^^>oo 0

0 ts.r^QO

M M M M n
M M M M M

C« ONOO 0 QO r^ On r« A

•oiooN*ooNON«or*t*
10 10 10 r« c^ r- 00 GO CO

ONOO NO 00

• lOCOGO 0

M ionO no

ONO 0000000

cOnO ^ On CO CO »0 iO<0
M r^ONM co»o»or*i^

M M M M M M

« c« 0
0 ^»oon«ooo 0 fOrt

q 0 q o o qNo o o

CO « CO 'i'lo r»

000000

CO « co^ioi>qp c« "^
0*4*^

O lOt^OO On iH fl d fi

M M M M

0 0 0 r^QO o 0 NO eo
M w CO CO ^nO qp QnO^

00000 00 O 0

10 CO

On»o

0000

00
10

ion

10 V)

CO «

COCO

'pooii ni 9iaix

o
:z;

00 NO

9 9 9 9 9 9

;z:

to CI
«o«

^ ^ ^ ^88»«9999 ^88««9«9«

^ P4 t% co^iONO r^oo 2 M d co^iONO r^oo ^ m «« co^iono r»«o

STUDY OP WHISKSY STORBD IN WOOD.

ftO

3
U

? I
h ft

B
g

l»)q^viv

1S99 9(iuCq9p[W«d

*J9)BM

0! 9iqnioeo| 10(03
Xq paAonxdJ JOioo

'loqcoiv

*lBJiijind

99 8»pi^q9piv

'3|YM« SB U»)9H

*3i|93« wm tppY

'Ioqo3i«
(Xniv *B fio |9tnji

iviyjna

SB BdpAqapiY

'SpaOB BB S J91SI8

'9P99* SB SppV

•spnos

'jooid 001

0) p»)Bin9lB3 10I03

'i0O4<I

lauBq
BlSOOIlBJtf 9niAL

«i (t It ;i «i « «

• o • o • o • o o

.^-««tt«iBBe

• o • o • o • o o

9 B B :i V tt It

o o

• •

• 0 • 10 10

0 OM^O 00000

0 «

^^«0«0«0C0^

0 «

• •

0^ QOvO d

M • r<« • Oh • eo N

tOt^QO 00000

OOMOOOOCi^MlOO^
0\«0m «<5« M COM M

O M '^^co^rh'**-^
^» » ♦ « * ♦

OOiOMroc>>cOONO 0
00 \0 CO O^ O^ 0^^0 r>«\0

• ^ fj • o^ • o^ w

• •MM* W

000 ChO 0000
O O «»oo>«o« woo

"^"^MWM«««W

# # # «

O O O »0 "* O^vO to «

w M o\oo ^ "<? '^ 3t !jr

^ ^ (O to CO «o to

• • 0 Q « c« i>.\0 O

• • 00 O 1^ M O^ M M

O M M M O 1^ M

♦ »

• ^ t^« « # «

•>o Ooo 11 ^r^ONco

• «

M M
M

M M

M 0 0
M M M

:S>

V*>8,»-

•S5

&«s.$$a$!f

* ■ • •

* *

• •

»OfO«

• • «

» ♦ « « « *

• to lOqp vo rh«ovO O
•0•10MM0O^0

• ••■••■■a

• mmmmmmO^

♦ <t ^L* * *

M vo ^vO to 10 10 O
O O^OnOvn m 9\^t^

■ •••■••••

^r^r-»oo On 0^00 00 r*

»o

« «

On OnoO t>«

COM O

00000000

■ ?1Q^ ONtO« OnO
O Ov N ^vO CO lO W CO

• ••••■•••

r* r«*oo «o ^ M CO « O

On to CI 10 '^ -^ r«*vO 00

•00 W CO "^ WVO 00

_♦ » * » « *

M t>«lO^CI « "^^M

^'t^r^r^ONiotOM
M ^ 10 ^V3 nO ^O NO VO

, r» cono On^O 00 "OVO

:^ (OnO On t>» M O O d
08 M 10 10 iOnO no no no

« up M 00 M ^(O«t00

10 C4 r>« M r^ (ONO to M
00 «^onc^ to^w ^-^

w«MC|«C«W««

•r^r^OOOOr^O
• M M w ^^^co»o

• O *1 M M M M M
NO NO

r^M cococi cotor^

M M CO^lOlOtOiO 2

M M M M M M n

«or>.cir>.iooo^p

^ 10 10 10 NO NO NO NO ^

•-• CO cOnO O O On m

* # #

• ^C4 00 00 00 r* c*

• 10 10 to to 10 tONO

• (Ot^NO ^^r*. 0>

• 0 0 •-• to CO 10 10

^4 N4 ^4 |b4 ^4 ^4 ^4

• ^rhoo ^r» too

• tooo cooo 0 ^>o

• W « COCO'^'^'^

00000 lOlOMVONONO

« « « CO to 10 ^NO «^

go onoo r>» Q

2 ONO O OnOnO OnO

Yz. •■••••••

^OmmOOiO>i

to On
OnMnO Ci M iOOnOnO

tooo 00 O 11 *^ >i W M
OnnO ^ONOMCiOM C

tONO 4"^d t^r^d « o

M ^ 1^ lOoO 00 00 ON On pp

. r^-^ w O OnO -^-^

tO^WOO f* OnOO
CO ^ lO lO 1>.nO no !*•

000000000
00 NO ON »o to o to r-NO

M On M ti^^o O « to to

M M M « Ci « C*

^C1
O NO ON CO M O Ci OnnO

O r^O Ci ^NOoood ON
O »00»oioo«ooo

5 5^^'8?8'S2

^ w 00 tooo M >o M

Ci NO NO «^ 1^00 00 On

oooqoqooo
00 Ci 1^ 10 10 10 tooo Ci

•^ 1 •iM>.00 On« it-iO

MMMMMCiCICi

t*» to
O O CO to On to t^NO 00

• On On ON 0 t^ l^ t>»

'♦

• »oiONO r^r^r^oo

• Ci MOO COON tOM

• M Ci to t«* !>. M M

. M M M M M Ci Ci

• 10 M ^ M NO NO M

• »o « 0 00 too ^

• Ci to ^ ^ iONO NO

ooqqoqqqq

M CO ^ 10 On O t^ OnnO
NO Onm ^ t}-00 00 O Ci
MMCiCiCiCiNCOCO

lOQ

Ci ON

• 10 M

• Ci 00

tOCI

•

•
M
M

•

to IONO

M M M

■
M M

•

0 iO»O0 0

r* r^

co»o

0 0

M M

^sO 0

►4 M M
M M M

•

•
•

•

• •

•
•

100

On 10

•

• •

•

00 NO

Ci Ci

On Onoo !>• «0 t}- On

M r* O Ci "^nO t^ On O

M M M M M M Ci
000000000

MMMMCitOtOtO

00000000

tooo . • M • Tf

NO O • • On • to

*pooA ai doifx

to CO

Ci
O

« It It 9 « « M

^ M (i CO ^100 t^OO

3 It 9 It II It

^ IX « to ^ IONO t».oo

On
00

^g^BltaitltJt
^ M Ci CO <4' IONO l>»00

ii8

C. A. CRAMPTON AND L. M. TOLMAN.

a
s

•|ga; qu«K

:S5

;> « 3 3 9 !>

i
1

•a

bo
• o • o • o • o o

ni »iqnio«ai aoioo ! ? ! ? ! ^ : ^S

« •oq o o o o o o

Xq pdAoniM ao|OD S m ro 55- « ^^ po Jo

OO ^MvOO N 0 CO
-loqoaiB 44o6oQd4<oriw

iXmV tB no IMBd ^'^cOcOTMO^f'^f'^}-

* «

lO O^vO O vO 00 O O

©©©•-"•-••-••-iMW

•3>n99« 8 S Jo S S> W & R5

8B nap^qapiv ^c^dsdsc^ci^ododoo

S ««N«rO«0<H

•»n»»«»«ppv 0< 4o^«^«oQ 6 « «o^

p « eo ^ ^»o 10*0 lo

(2 00»oOOON«fOw

iXniB BB i|o i»Bna >: -^i- '* S^^xS io r^oo S

»0 O S « »0 CO »o & S.

BB BdpAqapTV S v0^6>-<i->6»^>-<^

ft «>»fO^N«r^^ciy^

'w%99m BB Bja^BH r^ 4 4 i^oo r^ 6^^ q

« w « « to ^ ^

r«VO vO O CO ^o o >o^
•DflMBBB Bppy ^OModtodN-^dd

CO ^ <^ IO »o>o 1^ JiC

qoooqqooo

'•Plios Q 1^ vd 1^ ^ cooo »o o
00 CO IO r*oo O •-• CO c<j

'joojd cox M>qMioioci<>ao«

oi pa^BinoiBO jo|oo Nr^d^MW««o4l^

O O O O O O r^>o O

•'^^ SS588SSJS
•i^iJBq o ^ : : : o • • N o

« _

d ::::::' i *

jzj iF • •

•pooM u) 9ai\x ^^^""""^^

2 M d CO ^ *nyO r*co

STUDY OF WHISKBY STORED IN WOOD.

119

o
u

a
3

o
o

>< i

•IB9J qsjviv

')89) 9piCq9pi«i«d

ni aiqiiioBO) Joioo
iCq pdAomaj Joioo

'loqo3|«
(iCuv SB iio iMnj

'lunjjnd

SB MpAqapiY

'3|)90B SB saa^sH

'loqooiB oi
iXmB SB i|0 lasnjl •

*9I)93B HB SppV

. bo hfi bo

O'«h>o000000
O CO r» O On ^ fOvO NO

O O

N fO

O •O'^

t^ »O00

OfOfO»ooOOOC

O 00 O^ N OnQO
« N N

O O 00 « f*ioo

f5 CO f)

•-VO O

ON ON O 00 r-*

^ « o> »o N o
00 00 r>.oo 00 00

r-» r-» 1^

8 On On "* N O 00 00 O
rOWvO lO^Pi coro

* * *
.» t»l^O0

vO On ON On On

O M « Q
0000

« N «

N « « « W N

0 «>• «

• -^OO •^ NO «

cOnO 00

• 00 00 OnqO 00

0* #

• • •

» « «

• C< NO iH M 00

OOOOQmmmm

00 «ovb

00 M 10 M

»OvO o
« ^On

NO N O

O «^nO t^ CO
»0 »0 lO »o »o

"^ 10 »OvO NO NO r-» r*NO

_» »^» * * *

COO CO t>.oO ^ M NN ro

•^nO no O CO »0 on O •«1-
roco^»o»o^^io»r

O ON« « * « *

M 00 N ON© O NO w

»-NO»O^QQ«rO«
rh »OnO no nO no >0

fOqO -OnOwOi-o

•HOOM -JOOnOO^O

t^iOON •wwcOO^•^

lOiOO OnOnOnw 4

0 r>. CO ^

• « r^ 0 '^

« »o 0 ^

• NO 00 On M

« VO « N

• • • •

!-• 0 M « 0

SB sdpAqapiY

'3n»dB SB sj9)sK pq

•dl)93B SB SppV

'spiios

« « w

CO N « w fO CO

888

• ONoO NO r* 0

CO r^ ON

• ON 0 M M M
M M M M

0 ^
w r* 0

• N r* t^ 0 »o

so CO"^

• On 0 M so 00

• «o r^ r^ »>« r>»

0 0

N « «

<^ t>. 10 M 0 On

•N I>.OnC4 0 cOOniOO
■^nO t^ !>. t^OO 0»

000

• • •

000000

mmmiOmOmOcO
NnOOO OnO « fO^»0

rO ^NO NO r* Q 00 00 M
OOOOOO OnOnO OnOnO

ONOO Q O O O O 00 O
iOOnO *•* M cO'^I-cOtJ-

dOMW^M^N^ni

O Q O Q O fO
On On CO O l^vO co "^ r^

CO »OvO r-» r>. t^ On O On

O r* O « O « Tj-oo »^

^»O0 !>• ^nO CO r-»NO
cO CO »0 «OnO iOnO nO I^

OnioiOm C* moo •-• «0

M iocOOno r^NvO fO
« N ^ »0 10 iOnO no t^

ONO 0 0 0 0 0 0 0

•-• •*00 Q N ^ !>. ^ M

•t lOr^o^ONM M fO'«t'

'joaid 001
01 paiBinoiBo J0103

•jooia

'laiisq

'pooM u| ani)x

•o O ^'^ O O »o »o
O t>»ior^o O "or*!^

O « ^ Tf tONO NO NO NO

oqqoooooo

•^OOO^NcOcOcO

000000000

_ O O «*^ O O 0

O »0 ON M CO l>.00 N CO
O « CO "^ ^ 10 »OvO NO
OnOOOOOOOOn

M M « CO^tOt^ONO

oooooooo*-*

CO 10

o «o

10 o^

NO 00

CO O NO

CO CO ,4 -^

« ■

W ON

CO «

08 I * tf*

3 3 S 3 3

^ § §;: 3 3 3 3 a
S >^>^ ^ >N>«

J? « « CO-^iOnO l>'00 ^ IN w cO^iOvO 1^00

I20

C. A. CRAMPTON AND U M. TOI^MAN.

Whiskey Aged in IJncharred Packages.

No. 2627 (see Table XII): Sour mash, Bourbon whiskey, entered into bonded
warehouse February 5, 1898, in a new, uncharred, white oak, seasoned barrel (4 years
old); warehouse, dry and above ground; average temperature in the winter 42^, in
the summer 80°. Composition of the mash: malt, 6,048 pounds; rye, 4,704 pounds;
com, 45,024 pounds. The spirit was distilled in a copper still of 5,182 gallons capac-
ity, connected with a second still of 2,133 gallons capacity, and with a doubler of a
capacity of 950 gallons. The spirit is taken directly from the doubler to the cistern-
room without rectification or filtration. Yield per bushel of grain, 4 gallons. The
flavor is rank, even in eight-year old goods, and very different from that of all sam-
ples aged in charred wood.

No. 2625 (see Table XII): Sour mash, Bourbon whiskey, entered into bonded
warehouse January 31, 1898, in a new, uncharred, white oak barrel; low-boxed ware-
house, about three feet from the ground; average temperature in winter 48^, in the
summer 80^. Composition of the mash: malt, 952 pounds; rye, 448 pounds; com,
8,400 pounds. Distilled from a column still, capacity 150 gallons per hour. No
leach tubes used. Yield, 4 gallons of proof spirit per bushel of grain. Flavor not so
rank as No. 2627. Odor, not like any American whiskey.

Table XIII. — ^Avbrags, Maxima and Minima Data on All ths Samplss op

Whiskey.

Grams per 100 liters, 100 proof spirit.
Proof. Color. Solids.

Age.
New

lyr.

2 yrs.

3yrs.

4 yrs.

5 yrs.

6 yrs.

7 yrs.

8 yrs.

Data for
31 samples.

Average

Maximum

Minimum

Average

Maximum

Minimum

Average

Maximum

Minimum

Average

Maximum

Minimum

Average

Maximtun

Minimum

Average

Maximum

Minimum

Average

Maximum

Minimum

Average

Maximum

Minimum

Average

Maximum

Minimum

101.9
104. o
100. o

102.0
104.0
100. o

103.6
109.0
100. o

105.2

112. o

100. o

107.6
118. o
100. o

109.8
125.0

lOI.O
112. 8

132.0

102.0

II5-3
141 .0

102.0
117. o

134.0

102.0

8.2

138

4.6

10. 1
16.7

5.7

II. 7

18.3

7.0

12.4
18.9

7-4
14. 1
19.2

8.4

150
21.2

9-3

159
22.7

10. 1

16.3
24.2
10.5

20.1

161. 0
50

109.4

193.0

540

135.0

214.0

78.0

160. 1
245 o

90.0

167.9

249.0
92.0

189.0
280.0
1 14.0

203.5

287.0
132.0

220.9
309 o

1340

231.6
326.0
141 .0

Acids.

Esters.

Alde-
hydes.

Fusel
Furfural, oil.

6.4

29.1

1.2

16.3

53-2

1.3

3.9
150

trace

0.9
2.0

trace

95.2

171. 3
42.0

43-6

60.5

5.8

32.6

64.8

6.8

6.7

15.5

1.5

1.7

7-9
0.2

no. 7

194.0

42.8

48.6
63.0
II. 0

46.6

75-1
II. 2

9-3
18.7

5.9

1.8

9.1
0.4

114. 0

214.0

42.8

58.5
81.8

16.4

54.8

839
12. 1

11.5
22.1

5-9

2.1

9.5
0.6

121. 2
202.0

43-5

62.2
83.8

17.3

61. 1
89.1
13.8

12.4

22.2

6.4

2.3
9.6

0.7

125.8

237.1

43-5

64.6
92.6
19.0

65.0

105.5
17.3

13.1

23.1

6.6

2.5
9.6

0.8

126.8

254.2

45.1

69.7
96.8

24.3

695
109.0

17.9

13 I

23-7

7-5

2.6

9.5
0.7

139.9

245 -3
44.6

74-3
100. 0

247

730

114.9

21.3

13.8
26.7

7.5

2.5

8.5
0.8

141. 0

264.5
46.6

79-4
112. 0

76.6
126.6

14.3
28.8

2.7
10. 0

148.8
280.3

31.7

22.1

7.9

0.8

47.6

STUDY OF WHISKEY STORED IN WOOD. 121

Discussion of Table XIII.

This table shows the average, maxima, and minima data* for each
year on all of the thirty-one samples, with the exception that the fusel
oil results on Sample No. 2612, which was practically cologne spirits
and the results obtained on the color and solids of samples Nos. 2625
and 2627, which were aged in uncharred packages, were omitted.

The maximum and minimum figures alone have little value except to
show the range obtained, because they do not establish any relationship
between the various substances, as is done by the average figures.

One would not be justified in using these maximum and minimum
limits in judging the purity of the whiskey, because the results show
that in a properly matured spirit the maximum in color and solids, for
instance, does not occur with a minimum of acids and esters, so that if
a sample of commercial whiskey should show a maximum color, it should
not show the minimum adds and esters, and if such a condition were
found, it would indicate that the product was a compounded article.

In the discussion of the analysis of a whiskey, all of the determina-
tions and their relations must be considered before making a decision
as to its purity. The simple fact that it falls between the limits shown
by the maxima and minima figures is alone no definite indication as to
its genuineness.

An interesting study might be made of the conditions which brought
about these extreme results, which would be of practical value to the in-
dustry, but such discussion is beyond the scope of this paper. The facts
atone will be presented, so that any one who is interested may interpret
them.

It is evident from this work that there are two sources of furfural in
whiskey. This is shown by the fact that some of the new distillates
contain mere traces, while the mature spirits from the same source have
amaderable amounts, indicating that it must have been derived from
the barrel; again, other samples of the fresh distillate contain considera-
ble amounts, indicating that it must have come over in distillation and
been derived from the grain of the mash. All of the samples, however,
when calculated to the original (see Table 15) showed a slight increase
of furfural on aging, which was undoubtedly derived from the charred
wood.

Discussion of Table XIV.

This table gives the average, maximum and minimum of the deter-
minations made on the rye whiskies.

The maxima, except on the color and solids, are of little value in judg-
ing the purity of other whiskies. These maxima for the various years
give us limits for color and solids which will rarely be exceeded, and as
shown by the average will be, as a rule, much less. The close relation-

122

C. A. CRAMPTON AND L. M. TOLMAN.

ship which should obtain between solids and color is shown by this chart.
The sample showing the maximum color in each year also contains the
maximum solids, except in the sixth year, where there is another sam-
ple showing slightly higher solids.

Tablk XIV. — ^AvBRAGB, Maxima and Minima Data on Ryb Whiskibs.

Calculated to loo proof.

Color. Solids. Acids. Bsters.

Age. Data.

New Average. . .
Maximum .

Minimum . .

1 yr. Average. . .

Maximum .

Minimum . .

2 yrs. Average. . .

Maximum .

Original
proof.

Alde-
hydes. Furfural.

Pnael
Oil.

3 yrs.

Minimum. .

Average. . .
Maximum .

Minimum. .

4 yrs.

Average. .
Maximum

Minimtun.

5 yrs. Average. .
Maximum

Minimum.

6 yrs. Average. .

Maximum

Minimum .

7 yrs. Average .

Maximum

Minimum .

8 yrs. Average. . .
Maximtun .

Minimum. .

IOI.2
I02.0

lOO.O

102.5
104.0

lOI.O

104.9
109.0

100. o

107.7

112. 0

104.0

III. 2

118. 0

105.0

II3-8
125.0

108.0

118. o
132.0

IIO.O
121. 4

141. o

III.O

123.8

132.0

112. 0

0.0
0.0

0.0

30 o

119. 7
171. o

93.0

92.0

1447

199.0

121. 0
94.0

171. 4
224.0

145.0

119. o

185.0
238.0
156.0

153 o

206.5
251 o
170.0

168.0

223.1
284.0

193.0

176.0

242.2
306.0

195.0
194.0

256.0

339 o
214.0

200.0

4-4
72.0

12.0

46.6
60.5

33- 1
5.8

519
75.6

44-3

II. o

62.7
81.8

52.3
16.4

659
83.8

58.6
173
67.6
92.6

59-4
19.6

72.4
95-8
67.1

243
76.7
100. o
60.9
24.7

82.9
112. o

73-7
317

16.3
21.8

4-3

5.4
150

7.0
15 5

2.8

10.5

18.7

5-4

12.5
20.8

6.5

13 9
22.1

6.4

150
22.4

6.6

•14.6
22.3

7-3

15-5
25.2

7-5

16.0
26.5

7-9

i.o
19

0.7 trace

1.8
3-3

0.4

2.2
5.7

0.7

1-5

112. 7

6.1

202.0

0.7

$79 0
)6o.o

2.8

125. 1

6.7

203,5

0.7

$83.8
)67.8

3-2

128. 1

7-7

254-2

(86.8

1-4

h8-5

3-3

145-5

8.3

245 -3

0.7

)99-2
)8o.o

3-2

145-2

8.5

264.5

0.8

)99 2
)86.2

3-4

154-2

9.2

280.3

0.8

J 109.0
^107.1

The minima, however, are of considerable interest, not so much for
color and solids as for the acids and esters. Taking the adds for instance,
the minimum for each year is given by one package. No. 2623, which is
abnormal in many ways, and omitting it from consideration the minimum

STUDY 01^ WHISKEY STORED IN WOOD.

123

of adds for the first year would be 33.1, only slightly below the average;
for the second year it would be 44.3, and so on.

In order to show this and to eliminate the abnormal, the next to the
lowest figures for the color, solids, adds, esters, and fusel oils are also
included in the table.

Table XlVa. — ^Av^ragb, Maxima and MimifA Data for Boxtrbon Whiskies.

Calculated to 100 proof.

Proof. Color. Solids. Acids. Esters.
lOI.I 0.0 26.5 10. 0 18.4

104.0 0.0 z6l.O 29.x 53.2

Age. Data.

New Average. . .
Maximum .

Alde-
hydes.

3.2
7-9

Furfural.
0.7

Fusel
oil.

2.0

Minimum.. 100. o

lyr.

Average .
Maximum

101.8
103.0

Minimum... 100. o

2 yis. Average. . . . 102 . 2
Maximum . . 104.0

Minimum... 100. o

3 yrs. Average 103.0

Maximum.. 106.0

Minimum . . . 100 .0

4 yrs. Average 104 . 3

Maximum.. 108.0

Minimum... 100. o

5 yrs. Average 106. i

Maximum.. 113. o

Minimum... loi.o

6yrs. Average. . . . 107 .9
Maximum.. 116.0

Minimum. . . 102 .0

7 yrs. Average.... 109.6
Maximum . . 120.0

Minimum. . . 103 .0

8 yrs. Average zii.i

Maximum.. 124.0

Minimum... 102.0

0.0

4.0 12.0

99.6

193.0

61.0

540

126.8

214.0

81.0

78.0

149-3
245.0

95 o

90.0

151 9
249.0

lOI.O

92.0

1733
280.0

123.0

114. o

185. 1

287.0
132.0
127.0

200.9
309 o
140.0

1340

210.3
326.0
1520
141 .0

41. 1

55-3
24.7

7.2

45.6
61.7

25-5
233

54-3
64.8

38.4
32.1

58.4

73 o
40.4

40.4

56.3
78.9

48.2
42-7
67.1
81.0
53-6
450

71.9
86.4
60.6
49.0

76.4
91.4
64.1

53-7

13-0

i.o trace

5-8
8.6

8.4
12.0

10.5
22.1

5.9
II. o

22.2

6.9

II. 4
23.1

7.1

II. 9
233

7-7

12.4
26.7

7.7

12.9
28.8

8.7

1.6
7-9

2 . 7 trace

1.6
91

5.9 04

1.7
9-5

0.6

1-9
9.6

0.8

1.9
9.6

0.8

1.8
9 5

0.9

1-9
8.3

0.9

2.1
10. o

1.0

The point is especially noticable in the esters, for, omitting the minima,
the lowest ester found on a four-year old whiskey was 57.7 grams, which
is only slightly below the average, and would indicate that one might

124

C. A. CRAMPTON AND L. M. TOLMAN.

expect in a four-year old rye whiskey 50 to 60 grams of esters to 100
liters.

Table XIV shows the average, maxima and minima data for the Bour-
bon whiskies.

The same may be said of this table as of the previous one in regard to
the value of the maxima and minima figures.

All of these results indicate that we should expect lower figures with
Bourbon whiskies than with rye.

The high solids shown in the maximum for the new spirit was due
largely to ash, this being sample 2689, and the high ash changes, in this
particular sample, the ratio between the color and solids.

Table XV. — Avbrags Data for All Samplbs and por Ryb and Bourbons Sbpa-

RATSLY.

Caloilated to original volume.

Age.

New

xyr.

2 yrs.

3yrs.

4 yrs.

5yrs.

6 yrs.

7yrs.

8 yrs.

Original volume. Color.

Whole

Rye

Bourbon

Whole 7.3

Rye 8.4

Bourbon 6.4

Whole 8.6

Rye 10.6

Bourbon 6.7

Whole 10.2

Rye II. 5

Bourbon 8.3

Whole 10.2

Rye 1 1. 6

Bourbon 8.9

Whole 1 1. 1

Rye 12.2

Bourbon 10. o

Whole II. I

Rye i2.3

Bourbon 10. i

Whole 1 1. 1

Rye 12.0

Bourbon 10.2

Whole 10.5

Rye II. I

Bourbon 10. o

Solids.

Adds.

Bsters.

Aldehydes. Furfural.

, Fusel oil

20.0

6.4

15-0

4 03 0

•71

96.8

136

4-7

137

4.91 0

•97

83.2

26.0

7.7

17.2

3.26 0

•44

108.6

IOI.5

37.8

29.9

7.08 I

•5

106.2

114. 6

41.8

35.3

8.71 I

•7

106.8

90.1

34-4

24.9

5-55 I

•3

105.8

124.2

46.1

42.9

8.34 I

■ /

108. 1

1336

49.8

49-3

9.02 I

109.7

114. 8

42.7

37-3

7.78 I

•4

107 3

140.2

5I-I

48.4

9-47 I

106.3

150.4

54-4

54.3

9.80 2

104.4

130.7

47.8

42.5

9.15 I

•5

107.3

140.4

51-6

50.9

10.2 I.

104.3

153 I

54-2

57-2

II. 2 2

102.0

127.7

48.9

45.0

9-3 I

106.3

149.2

52.2

5I-I

10.2 I.

100.4

158.8

54.8

57-5

II. 3 2.

100. 1

140.2

49.8

45.0

9.2 I.

100.7

151. 4

53.2

50.7

10.2 I.

104.7

161. 0

54.8

55.5

11.3 2.

105.9

142.5

51.8

45.2

9.1 I.

103.8

154.0

52.2

5I.I

9.8 I.

99-9

161. 3

51.9

56.6

10.6 2.

98.8

147. 1

52.4

46.4

9.0 1.

101.6

155-2

53.1

50.9

9.6 I.

98.0

163.8

52.6

56.7

10.6 2.

99 0

147.7

53.6

45.9

8.8 I.

97- 1

This table gives the average of all the whiskies and the average of
the rye and Bourbon separately, calculated back to the original volume,
in order to eliminate the eflFects due to the concentration taking place
while the spirit was stored. The more important points brought out
by this table are brought out in the various charts. These figures show

STUDY O^ WHISKBY STORBD IN WOOD. 1 25

more nearly the actual changes taking place, and prove that after the
third year there is very little change in the amounts of the various sub-
stances present. As has been said before, the changes taking place in
the whiskey after the third or fourth year are almost entirely due to the
concentration which occurs.

These figures also show that the difference between rye and Bourbon
whiskies holds good even when the results are corrected for loss of vol-
ume, proving that with the rye whiskies there has been a greater activity
in the aging processes.

The fact that after the third year, the acids, esters, and aldehydes do
not show any appreciable increase when corrected for loss in volume, does
not necessarily mean that there is no formation of these substances in
the barrel, as there may be some loss of them through the wooden walls
of the barrel, with the alcohol and water which is constantly passing off;
but the fact that they remain so constant in amotmt would indicate
that these substances are left behind, in the same manner as the fusel
oil and solids, by what might be called the selective action of the wooden
membrane constituting the walls of the porous cell in which these prod-
ucts are stored.

Chart I shows the average of the changes taking place in the proof
and the volume.

The proof increases, starting from approximately 100 proof, so that
all that is shown on the curve are the degrees above 100 proof.

The changes in volume are plotted from the per cent, of loss in volume.
The changes in proof and volume are important in their bearing on the
changes that take place in the other substances.

The wide difference in rate of increase of proof shown by the rye
and Bourbon whiskies is typical of the differences between them, and ap-
parently has a simple explanation. In the case of the rye whiskies nearly
all were aged in heated warehouses where the changes taking place are
aided by the higher temperature. There is also a large loss of volume
in the rye whiskies, as is shown by the lines. In short, there is much
greater chemical activity, and so, greater changes. The amounts of
adds and esters formed apparently have a direct relationship to the
amount of the whiskey which passes through the wood of the barrel.

The Bourbon whiskies, on the other hand, are as a rule stored in un-
heated warehouses, so that there is less loss of volume in the spirit, and
less increase in proof, and smaller amounts of acids and esters are formed.
As a matter of fact, a much lighter bodied whiskey is formed, and from
these results it would seem to be due to a greater extent to the method
of storage than to the inherent differences in the whiskies themselves.

Chart II shows changes in solids and coloring matter, and the same
differences between the rye and Bourbon whiskies as were noted on Chart

126

C. A. CRAMPTON AND L. M. TOLMAN.

§ 1 ^ § 5 ^

«^\

•

:Sf\

^Ui

\
\

kj .

«l \

^ .

\
\

«o

\
\

5 .

1 1
1 1
> 1

■^

•

1 1

<a ^ ^ Nl ^

\
\

q) ^ ^ ^ <^

•

\\
\\
\\

€4

1 \

V M -0 ^ S

\ k'
i \l\

\
\

\ Ml\

«0

-J

•o a to 5 v> Q

N «M < ^ "^

STUDY Olf WHISKEY STORED IN WOOD.

127

g|!3 S !o $ ^o ^

1 1

1
\
\

^iB>

«h

/'■

7
!l

\
1 1
1 1

\ \

\ 1

■

\ 1

•

1 1

•
t

\\ A

\ 1

J-.

\ 1

1 1

•

\
\

\ \

»0

\
\

\ \

\ i

\
\

1
1

«t

■ ■ 'i "i ;■.!■ 1 III

^^^

1
-I

1 i 1 1 1 \ 1

' • ' ' ■ i\i
* * * * * 1 t . ^ \

}

•0

i u

* >r^

■^

\ \|\\

\ 1\

\ ^

\ 1

#•

i t %****•*«* V *\| \\

\ '^v

•

, \^

.M\

*
1

^ 1 1

t
•

•
■

T-

•9

Q ^ 9

\ \\

% \

1 :

^ ^ ^

^\^

\ \

E 1

^
^

J9 ^ «» Is

$ § $ .^^

.\.!..''\

» 1

X
^

''^^

^ ^»fc!^

s § «» j^N

•v;

^^;

iii

-U

-4-

- n

?s.."T«

CM <

5 -0

s^/^j./y 00/ & 3 c/ S'U/a

I20

C. A. CRAMPTON AND h. M. TOI^MAN.

Whiskey Aged in Uncharred Packages.

No. 2627 (see Table XII): Sour mash, Bourbon whiskey, entered into bonded
warehouse February 5, 1898, in a new, uncharred, white oak, seasoned barrel (4 years
old); warehouse, dry and above ground; average temperature in the winter 42^, in
the summer 80^. Composition of the mash: malt, 6,048 pounds; rye, 4,704 pounds;
com, 45,024 pounds. The spirit was distilled in a copper still of 5,182 gallons capac-
ity, connected with a second still of 2,133 gallons capacity, and with a doubler of a
capacity of 950 gallons. The spirit is taken directly from the doubler to the cistem-
room without rectification or filtration. Yield per bushel of grain, 4 gallons. The
flavor is rank, even in eight-year old goods, and very different from that of all sam-
ples aged in charred wood.

No. 2625 (see Table XII): Sour mash, Bourbon whiskey, entered into bonded
warehouse January 31, 1898, in a new, uncharred, white oak barrel; low-boxed ware-
house, about three feet from the ground; average temperature in winter 48°, in the
summer 80^. Composition of the mash: malt, 952 pounds; rye, 448 pounds; com,
8,400 poimds. Distilled from a column still, capacity 150 gallons per hour. No
leach tubes used. Yield, 4 gallons of proof spirit per bushel of grain. Flavor not so
rank as No. 2627. Odor, not like any American whiskey.

Tablb XIII. — ^AvSRAGB, Maxima and Minima Data on All thb Samplbs op

WmSKBY.

Grams per 100 liters, 100 proof spirit.

Aide- Pusel

Color. Solids. Acids. Bsters. hydes. Furfural, oil.

.... 20.1 6.4 16.3
.... 161 .0 29.1 53.2
5.0 1.2 1.3

Age.
New

Data for
31 samples.

Average

Maximum

Minimum

I yr.

2 yrs.

3yrs.

4 yrs.

5 yrs.

6 yrs.

7 yrs.

8 yrs.

Average

Maximum

Minimtun

Average

Maximum

Minimum

Average

Maximum

Minimum

Average

Maximum

Minimum

Average

Maximum

Minimum

Average

Maximum

Minimum

Average

Maximum

Minimum

Average

Maximum

Minimum

Proof.
101.9
104.0
100. o

102.0
104.0
100. o

103.6
109.0
100. o

105.2
112. o
100. o

107.6
118. o
100.0

109.8
125.0

lOI.O
112. 8

132.0
102.0

"5-3
141 .0

102.0

117. o
1340
102.0

8.2

13-8

4.6

10. 1
16.7

5-7
II. 7
18.3

7.0

12.4
18.9

7-4
14.1
19.2

8.4

150
21.2

9.3

159
22.7

10. 1

16.3
24.2
10.5

109.4

193.0

54 o

135.0

214.0

78.0

160. 1

245 o

90.0

167.9

249.0

92.0

189.0
280.0
114.0

203.5
287.0
132 o

220.9
309 o

1340

231.6
326.0
141.0

Acids.

6.4
29.1

1.2

43-6

60.5

5.8

48.6
63.0
II. o

58.5
81.8

16.4
62.2

83.8

173

64.6
92.6
19.0

69.7
96.8

243

74.3

100. o

24.7

79-4
112. o

317

32.6

64.8

6.8

46.6

75- 1
11.2

54.8

83.9
12. 1

61. 1
89.1
138

65.0

105.5
17.3

69.5
109.0

17.9

730

114. 9

21.3

76.6

126.6

22.1

3-9
150

trace

6.7

15.5

1.5

9.3
18.7

5-9

5
12
22.

23
6.

5
I

4
I
I
6

13 I

23-7
7-5

13.8
26.7

7-5

14.3
28.8

7.9

0.9

2.0

trace

1-7

7-9
0.2

1.8

9-1
0.4

2.1

9-5
0.6

2-3
9.6

0.7

2.5
9.6

0.8

2.6

9.5
0.7

2.5

8.5
0.8

2.7
10. o

0.8

95.2
171-3

42.0

no. 7

194.0

42.8

114. o

214.0

42.8

121. 2
202.0

43.5
125.8

237.1

43-5

126.8

2542

45- 1

139.9

245.3
44.6

141. o

264.5
46.6

148.8

280.3

47.6

STUDY OF WHISKBY STORED IN WOOD. 121

Discussion of Table Xm.

This table shows the average, maxima, and minima data' for each
year on all of the thirty-one samples, with the exception that the fusel
oil results on Sample No. 2612, which was practically cologne spirits
and the results obtained on the color and solids of samples Nos. 2625
and 2627, which were aged in uncharred packages, were omitted.

An interesting study might be made of the conditions which brought
about these extreme results, which would be of practical value to the in-
dustry, but such discussion is beyond the scope of this paper. The facts
alone will be presented, so that any one who is interested may interpret
them.

It is evident from this work that there are two sources of furfural in
whiskey. This is shown by the fact that some of the new distillates
contain mere traces, while the mature spirits from the same source have
considerable amounts, indicating that it must have been derived from
the barrel; again, other samples of the fresh distillate contain considera-
ble amounts, indicating that it must have come over in distillation and
been derived from the grain of the mash. All of the samples, however,
when calculated to the original (see Table 15) showed a slight increase
of furfural on aging, which was undoubtedly derived from the charred
wood.

Discussion of Table XIV.

This table gives the average, maximum and minimum of the deter-
minations made on the rye whiskies.

The maxima, except on the color and solids, are of Uttle value in judg-
ing the purity of other whiskies. These maxima for the various years
give us limits for color and solids which will rarely be exceeded, and as
shown by the average will be, as a rule, much less. The close relation-

122

C. A. CRAMPTON AND L. M. TOLMAN.

TaBLB XIV. — ^AVBRAGB, MAXIMA AND MINIMA DaTA ON RyB WhISIUBS.

Calculated to loo proof.

Color. Solids. Acids. Esters.
IOI.2 CO 13.3 4.4 16.3 5.4 i.o
102.0 0.0 30.0 72.0 21.8 15.0 1.9

Age.
New

Data.
Average. .
Maximum

Original
proof.

Aide-

hydes. Furfural.

Pnsel
Oil.

I yr.

2 yrs.

3yrs.

Minimum . .

Average. . .
Maximum .

Minimum . .

Average. . .
Maximum .

Minimum . .

Average. . .
Maximum .

Minimum . .

4 yrs.

Average. .
Maximum

Minimtmi.

5 yrs.

Average. .
Maximum

Minimum .

6 yrs. Average..
Maximum

7 yrs.

Minimum .

Average .
Maximum

Minimum .

8 yrs. Average. . .
Maximum .

Minimum . .

100. o

102.5
104.0

lOI.O

104.9
109.0

100. o

107
112

7
o

104.0

III
118

.2
.0

105.0

II3-8
125.0

108.0

118.0
132.0

IIO.O

121. 4
141 .0

III.O

123.8
132.0

112. 0

0.0

8.8
13-8
2
6

1.6

6.7
8.8

8.6

3
8.

I ,

I .
I,

5
9

4
I

9
6

1.8

7.0
1.2

3.7
2-4
8.0
2.7

4
3
8

24
3
3

6
I

6
,2

119. 7

171. o

93 o

92.0

144-7
199.0

121. o
94.0

171-4
224.0

145.0

119. o

185.0
238.0
156.0
153 o

206.5
251.0

170.0
168.0

223.1

284.0

193.0

176.0

242.2
306.0

195.0

194.0
256.0

339 o
214.0

200.0

12.0

46.6
60.5

33-1
5.8

51-9
75.6

44-3
II. o

62.7
81.8

52.3
16.4

65.9
83.8
58.6

17-3
67.6
92.6

59 4
19.6

72.4
95-8
67.1

243
76.7
100. o
60.9
24.7

82.9
112. o

73-7
31-7

4-3

0.7 trace

7.0

15.5

2.8

10.5
18.7

5-4

12.5
20.8

6.5

13-9
22.1

6.4

15-0
22.4

6.6

14.6
22.3

7-3

15-5
252

7-5

16
26

7-9

1.8
3 3

0.4

2.2

5-7

0.7

1-5
6.1

0.7

2.8
6.7

0.7

3-2

7-7

1-4

3-3
8.3

0.7

8.5

0.8

3-4
9.2

0.8

154-2

280.3

I109.0

)io7.i

The minima, however, are of considerable interest, not so much for
color and solids as for the acids and esters. Taking the acids for instance,
the minimum for each year is given by one package, No. 2623, which is
abnormal in many ways, and omitting it from consideration the minimum

STUDY Of WHISKEY STORED IN WOOD.

123

of adds for the first year would be 33.1, only slightly below the average;
for the second year it would be 44.3, and so on.

In order to show this and to eliminate the abnormal, the next to the
lowest figures for the color, solids, acids, esters, and fusel oils are also
included in the table.

Table XlVa. — Avsragb, Maxima and Minima Data for Bourbon Whiskibs.

Calculated to 100 proof.

Age. Data.

Proof.

Color.

Solids.

Acids.

Bsters.

Alde-
hydes.

Furfural.

Fusel
oil.

New Average

lOI.I

0.0

26.5

10. 0

X8.4

3-2

0.7

100.9

Maximum .

104.0

0.0

161. 0

29.1

53.2

7-9

2.0

171-3

MitiiTnttm . .

100. 0

0.0

4.0

12.0

13-0

I.O

trace

( 7^ 3
\ 42.0

I yr. Average . . .

IOI.8

7.1

99.6

41. 1

28.6

5.8

1.6

no. I

Maximum . .

103.0

10.9

193.0

55.3

55-9

8.6

7.9

173.4

Mifiitniim

100. 0

I 5-4
I 4.6

61.0
540

24.7
7-2

1T,2\

10. 4j

2.7

trace

\5S.0

2 yrs. Average. . . .

102.2

8.6

126.8

45-6

40.0

8.4

1.6

108.9

Biaximum . .

104.0

II. 8

214.0

61.7

59-8

12.0

9-1

197.1

Minimum...

100. 0

I 6.9

i 5.7

81.0
78.0

235
23.3

24 A
11.2)

5-9

0.4

\86.2
U2.8

3 yrs. Average

103.0

10. 0

149 -3

54-3

48.1

10.5

1.7

112. 4

Maximtun . .

106.0

138

245.0

64.8

730

22.1

9.5

221.8

Minimum.. .

100. 0

I 8-9
I 7.0

930
90.0

38.4
32.1

27 2\
12.15

5.9

0.6

\88.o
/43-5

4 yis. Average. . . .

I04-3

10.8

151 9

58.4

53-5

II. 0

1.9

123.9

MftTimntn

108.0

148

249.0

73 0

80.6

22.2

9.6

237.1

Minimum...

100. 0

( 8.6
i 7.4

lOI.O

92.0

40.4
40.4

28,2\

13.85

6.9

0.8

\950
)43-5

5 yn. Average

106. 1

12.3

173-3

56.3

55-9

II. 4

1-9

125.3

Maximum . .

II30

16.7

280.0

78.9

87.2

23.x

9.6

243-4

Minimum. . .

lOI.O

J11.8
I 8.4

123.0
114. 0

48.2
42.7

27^7\
I7-3J

7.x

0.8

S 9^0
\ 45-1

6713. Average. . . .

107.9

13 I

185. 1

67.1

64.0

II. 9

1.8

135.3

Maximum . .

1x6. 0

17.5

287.0

81.0

83.9

23.3

9 5

240.0

Minimum...

102.0

^12.0
} 9.8

132.0
127.0

53-6
45.0

3^ A
17.95

7-7

0.9

\ 98.1
I 44-6

7 yrs. Average. . . .

109.6

13.9

200.9

71.9

63.3

12.4

1-9

137.2

Maximum . .

120.0

19.4

309.0

86.4

90.0

26.7

8.3

243-4

Minimum. . .

103.0

^11.8

140.0
1340

60.6
49.0

37 A

21.35

7-7

0.9

\ 98.2
\ 46.6

8 yrs. Average

III. I

14.2

210.3

76.4

65.6

12.9

2.1

143.5

Maximum . .

124.0

20.9

326.0

91.4

93-6

28.8

10. 0

241.8

Minimum . . .

102.0

^2-3

1320
141. 0

64.1
53-7

37. 7(
22.15

8.7

1.0

\llO.O

\ 47.6

The point is e

specially

noticable in the esters, for, omitting the minima,

1 a fmir-vpflr nIH -whislfpv -wa^c C7.7 jrranis. which

is only slightly below the average, and would indicate that one might

124

C. A. CRAMPTON AND h. M. TOLMAN.

expect in a four-year old rye whiskey 50 to 60 grams of esters to 100
liters.

Table XIV shows the average, maxima and minima data for the Bour-
bon whiskies.

The same may be said of this table as of the previous one in regard to
the value of the maxima and minima figures.

All of these results indicate that we should expect lower figures with
Bourbon whiskies than with rye.

The high solids shown in the maximum for the new spirit was due
largely to ash, this being sample 2689, and the high ash changes, in this
particular sample, the ratio between the color and solids.

Table XV. — ^Avsragb Data for All Samflbs and por Rys and Bourbons Sbpa-

RATSLY.

Calculated to original volume.

Age.

New

I yr.

2 yrs.

3yrs.

4yr8.

syrs-

6 yrs.

7 yrs.

8 yrs.

Original volume. Color.

Whole

Rye

Bourbon

Whole 7.3

Rye 8.4

Bourbon 6.4

Whole 8.6

Rye 10.6

Bourbon 6.7

Whole 10.2

Rye II. 5

Botu-bon 8.3

Whole 10.2

Rye II. 6

Bourbon 8.9

Whole 1 1. 1

Rye 12.2

Bourbon 10. o

Whole II. I

Rye i2.3

Bourbon 10. i

Whole II. I

Rye 12.0

Bourbon 10.2

Whole 10.5

Rye II. I

Bourbon 10.0

Solids.

Adds.

Esters.

Aldehydes. Furfural

. Fusel c

20.0

6.4

15-0

4.03 0

•71

96.8

136

4.7

137

4.91 0

•97

83.2

26.0

7-7

17.2

326 0

•44

108.6

IOI.5

37-8

29.9

7.08 I

•5

106.2

114. 6

41.8

35-3

8.71 I

106.8

90.1

34-4

24.9

5.55 I

105.8

124.2

46.1

42.9

8.34 I

• /

108. 1

133.6

49.8

49-3

9.02 I

■9

109.7

114. 8

42.7

37.3

7.78 I

107 -3

140.2

51 I

48.4

9.47 I

106.3

150.4

54-4

54-3

9.80 2

104.4

130.7

47.8

42.5

9.15 I

•5

107.3

140.4

51.6

50.9

10.2 I.

104.3

153 I

54-2

57.2

II. 2 2

102.0

127.7

48.9

45.0

9-3 I

106.3

149.2

52.2

51.1

10.2 I

100.4

158.8

54.8

57-5

II. 3 2.

100. 1

140.2

49.8

45 0

9.2 I.

100.7

151. 4

53-2

50.7

10.2 I.

104.7

161. 0

54.8

55-5

11.3 2.

105.9

142-5

51.8

45.2

9.1 I.

103.8

154.0

52.2

511

9.8 I.

99.9

161. 3

51.9

56.6

10.6 2.

98.8

147. 1

52.4

46.4

9.0 I.

101.6

155.2

531

50.9

9.6 I.

98.0

163.8

52.6

56.7

10.6 2.

99.0

147 -7

53-6

45.9

8.8 I.

97.x

This table gives the average of all the whiskies and the average of
the rye and Bourbon separately, calculated back to the original volume,
in order to eliminate the effects due to the concentration taking place
while the spirit was stored. The more important points brought out
by this table are brought out in the various charts. These figures show

STUDY OI^ WHISKBY STORED IN WOOD. 1 25

These figures also show that the difference between rye and Bourbon
whiskies holds good even when the results are corrected for loss of vol-
ume, pro\dng that with the rye whiskies there has been a greater activity
in the aging processes.

The fact that after the third year, the acids, esters, and aldehydes do
not show any appreciable increase when corrected for loss in volume, does
not necessarily mean that there is no formation of these substances in
the barrel, as there may be some loss of them through the wooden walls
of the barrel, with the alcohol and water which is constantly passing off;
but the fact that they remain so constant in amoimt would indicate
that these substances are left behind, in the same manner as the fusel
oil and solids, by what might be called the selective action of the wooden
membrane constituting the walls of the porous cell in which these prod-
ucts are stored.

Chart 1 shows the average of the changes taking place in the proof
and the volume.

The proof increases, starting from approximately 100 proof, so that
all that is shown on the curve are the degrees above 100 proof.

The changes in volume are plotted from the per cent, of loss in volume.
The changes in proof and volume are important in their bearing on the
changes that take place in the other substances.

Chart II shows changes in solids and coloring matter, and the same
differences between the rye and Bourbon whiskies as were noted on Chart

126

C. A. CRAMPTON AND L. M. TOLMAN.

vl ■

§ 1 S § S o

>l ■

T
1

\
\
\
\

kj .

\
\

1
I

^ ..

ji>

k\ i

«t

L«

«9

\
\

\ \

«o

S .

1 i
1 1

i 1

<d ^ ^ ^ ^

\
\

^ ^ ^ ^ <^
5 '§ 5 .

L—

^
w

\
\
\

i\
\\
\\
\\

•

5! '^

s M -0 ^ 'o

Wll

y 1

\
\

§.

— M

._^

— u

-J

13 ^ SJ 5 •« *

StUDY Ol^ WHISKEY STORED IN WOOD.

127

"■"

^S S !o 5 ^ 0

U-

"■"^

""""*

^■■■i

"TT"

1 1

•^ b\

\
\
\

t 1
1 1

\
\

\
I

^1 —

I' (

■■' '

^~'

"^1

^^^

"**^

' ^^

^ '

' .

\l 1

»:

\ V

\' '

\ \

1 \

•
1

1 \

1
1

1 1

±_i_.

•^%

i I

1 I

\
\

i' \

•

\
\

;

• —

\ \

»s^

i i

— r

ft:

\
\
\

1
1

«t

.^b

V 1 . .'.■ 1 ii \

.MM

•

-1

i * ! \ !
Ill 1 r,i

* * ^ • ' 1 * • ^ \

•0

^ \ tk

' U

Y^rn

(

\ \|\>

\ ^,

^ 1

••

, % % » ft * » • * 5 * V *M \

^ ^ *{v

•

\ I

V M\

J w

\N^;

^ L i\

0 ^ V

\ NN

% \ K\

1 :

^ ^ ^

^ V"^

\ v\\

B 1

& .w «» 1^

:
1

1 ■

~ 2

$ S ^ ?>

?i-

j>^^\

j\\

■ •
• 1

w ■ N^^M

^■1

«» § «» $5!

$«««(%«% «ft«t

©i «$;>: «0>tf rctf 0; §• ^ 5; 5 5

^^.

^
«*

*•

1 •

■ •

ri 1 1 1 1 1 1 1

=1-

fdOSi

& ^ ^
J> ^ §

«vi «S| <

5 *o

S' tf 3 ± /7 OOi & 3

^cy

r •

/•

128 C. A. CRAMPTON AND U M. TOLMAN.

I are seen, namely, a marked increase during the first year, and a regular
increase thereafter.

A comparison of the solids as calculated to loo proof, and as calculated
to the original volume, shows that the increase in the former is almost
entirely due after the third year to the concentration taking place, the
actual amount of solid matter in solution in any single barrel remain-
ing practically the same after the third year. The same is true of the
coloring matter and apparently little material is extracted from the wood
after that time.

The remarkable similarity of the solids and color curves shows that
there is a very close relationship between them. In fact, the relation-
ship is so close that from the determination of the solids one could very
closely calculate what the color should be, or from the color calculate
the solids.

This relationship is of great value in the detection of the artificial col-
oring of whiskey, and in judging whether the solids are normal.

Chart III shows first the changes in fusel oil, calculated to loo proof,
and to the original volume. A gradual increase of fusel oil is shown as
the whiskey ages, but it will be seen from the curve plotted from the cal-
ctdations to the original volume that this change is due to concentra-
tion. The actual amount of fusel oil in a barrel of whiskey remains the
same during the period of storage, but the whiskey itself shows a per-
centage increase.

These results, however, do not prove whether the fusel oils undergo
change in aging, or whether any of the fragrant esters present are due
to the fusel oils, because the method of analysis employed, first saponi-
fies all of the esters and determines only the higher alcohols. The ques-
tion as to the change taking place in the fusel oils must be answered by
fractionating large quantities of a new and of a mature spirit and study-
ing the composition of the higher boiling-point portions.

This chart also shows the changes occurring in the acids and esters,
and further emphasizes the fact of the differences between rye and Bour-
bon whiskies. In every way the r}^e whiskies are the stronger, if one may
use the term; that is, they contain more solids, color, acids, esters,
etc., but all of these differences can be traced to the method of aging.

A very striking point which is also brought out by this chart is the
change in relationship between the acids and esters in young spirits and
the constant relation in matured spirits.

The average of the acids in the new spirit is 6 grams per loo liters,
and the esters average i6. In the spirit one year old this relation has
changed, the acid being 44 grams against 33 grams for the esters. The
second year the acids are 49, and the esters 47. The third year the acids
are 59, and the esters 55. The fourth year the acids are 62 and the es-

STUDY OP WHISEBY STORSD IN WOOD.

ters 6i, and they remainTpractically the same during the next four years.
Tlus shows that these two substances gradually approach an equihbrium,
wtncb they reach about the fourth year, and which does not change after-
wards.

This point will be more^fully discussed under chart V, showing the re-
lationship of the adds and esters for a number of years.

igo C. A. CRAMrtON^ANb L.'m. TOUIAtJ.

N0.^.lt£LATIONSmP BEIfELOPEOON STORAtEM » YEAR OLB SOOOS.

Chart IV, plotted from the results obtained on the four-year sam-
ples, shows the relationship of the various products present to each other
and to the change in volume and the increase in proof.

Perhaps the most striking point brought out is the close relationship
between the acids and the esters, the two lines following the same course
across the sheet in the most remarkable way. But when we consider

STUDY OF WHISKEY STORED IN WOOD. 13I

the dependence of one on the other this is not so strange. It is apparent
from this, as well as from Chart I, giving the averages, and from Chart
V, on which are plotted the acids and esters, that these two substances
reach an equilibrium, at the end of three or four years, as previously
stated, which relationship does not change during longer storage. The
meaning of this point will be discussed later.

This chart also gives the maxima and minima found for these sam-
ples, and in many cases shows how they may be explained by the loss
in volume and increase in proof.

It also calls our attention to the fact that whereas when a whiskey
has been matured in wood, and has a low acid content, we should expect
to find a small amotmt of esters, it is shown in the new spirit and that
when a year old this relationship does not hold, the two substances not
having yet reached an equilibriiun. This point may be of value in de-
tecting young whiskies.

The color, solids, and concentration lines show a very close relation-
ship, as might be expected, as they are dependent one on the other.

A study of all the lines reveals a marked relation among them. A
high color, high solids, and high concentration are generally accompanied
by high acids and esters, and low color and solids go with low adds and
esters, which is a fact that will be of value in judging the purity of whis-
kies found on the market.

In fact, a study of the relationships fotmd in these whiskies will be
of great value in the determination of the adulteration of commercial
whiskies. The mnge of variation that may take place in whiskies stored
under different conditions is also shown by this chart.

Chart V shows the changes taking place in the acids and esters, using
the results obtained on the new spirits, and those one, two and eight
years old.

The great irregularity in the amounts of these substances in the new
spirit, and the seeming lack of any relation, are at once noted ; the acids,
however, in every case, are lower than the esters, but at the end of the
first year this has changed to a great degree, only seven samples show-
ing higher esters than adds. The average of the esters, as shown in Chart
III, is 9 grams per 100 liters below that of the adds.

At the end of the second year the esters are on the average 7 grams
bek)w the adds, and by the end of the fourth year the acids and esters
average the same, retaining this relation up to the eighth year, when the
experiment ceased.

This shows that in the aging process the adds are formed more rapidly
at first than the esters, but later the esters form more rapidly, so that by
the end of the fourth year they are present in about the same amounts,
and remain the same during storage. The fact that this equilibrium is

C. A. CRAHPTON AND L. H. TOLUAN.

reached at the end of three or four years indicates that there is some very
close relation between the equilibrium of the acids and esters and the
ripening of the whiskey. This is, of course, only one of the factors in the
aging, but it seems probable that, when this condition of equiUbrium is
reached, the whiskey is matured in so far as the acids and esters contribute
to that condition.

The chart also calls attention to some very abnormal samples, for in-
stance, Nos. 2623 and 2689, both of which were aged in very cool ware-
houses, which fact probably explains in a degree the cause of their ab-
normalities.

STUDY OF WHISKEY STORED IN WOOD. 1 33

Effect of Aging in Charred and XTncharred Packages.

In order to determine the effects of the material extracted from the
barrel in the process of aging on the flavor of the whiskey, especially the
difference between charred and uncharred packages, two samples of
sour mash Bourbon whiskey, which had been produced by the same dis-
tiller from very similar mashes, as shown by the table below, and dis-
tilled in the same general type of still, were taken for comparison.

Whisky in Whisky in

uncharred charred

Composition package. packase.

of mash. Pounds. Percent. Pounds. Percent.

Malt 6048 10.9 4144 II 'O

Rye 4704 8 . 5 2688 7.4

Com 45024 80.6 30800 81.6

The samples had both been stored the same length of time in a ware-
house and practically the only difference in the two products was in the
kind of packages in which they were stored. The flavor, however, of
the two whiskies was entirely different: the whiskey stored in the un-
charred package lacked almost entirely the flavor so characteristic of
American whiskies, having more of the flavor of Scotch or Irish whiskey,
but without, of course, the smoky flavor of the Scotch, and was consid-
ered a very good-flavored whiskey by several persons who had a taste
for such whiskies ; the whiskey aged in the charred wood had the strong
aromatic flavor which is so characteristic. In fact, the products are en-
tirely different types of whiskey, showing that the method of aging is
the important factor.

A study of the analytical data obtained, given in the following table,
shows that there is a remarkable similaritv in the amount of esters and
adds and fusel oils in the two samples at the same age.

Color Grams per xoo liters.

insolu. * >

ble in Aide- Pur- Fusel

Color H«0. Solids. Acids. Bsters. hydes. fural. oil.

in Per Per Per Per Per Per Per

Serial number. Proof. H''. cent. cent. cent. cent. cent. cent. cent.

3627, imcfaarred, 4th year 104.0 4.5 96.0 58.8 66.0 8.0 1.2 loi.o

2635, charred, 4th year. . 105.0 9.0 1550 63.6 48.4 10. o 0.8 99.0

3627, uncharred, 8th year no. 9 7.0 23 160.0 81.6 84.5 zo.8 1.5 112. o

3635, charred, 8th year. . izo.o 13.5 61 181. o 81.6 65.6 13.2 1.2 125.0

The main difference between the two samples is in the amount of color
and solids, and the composition of these solids and color, as shown by
their solubilitv in water. The solids and color in the whiskey from the
uncharred package are very much more soluble in water, being less of
the resinous nature shown in the solids from the charred package, and in
this respect it is very much like the Scotch whiskies which are aged in
uncharred wood.

The resinous or oily nature of the solids in the whiskey from the charred

t34 C A. CRAMPTON AND U M. TOLMAN.

package is also shown in the whiskey itself, the foam being much more
oily and remaining much longer than the foam on the whiskey from the
uncharred package.

The actual amount of solids in the eight-year old goods is not so very
different in the two samples, but the amount of color in the uncharred
is only 7.0, as against 13.5 in the charred, and only 23 per cent, of the
color in the uncharred is insoluble in water, while 61 per cent, of the color
in the charred is insoluble, showing a wide difference in composition.
As stated before, the whiskies are remarkably similar as far as the adds,
esters, aldehydes, furfural and fusel oil are concerned, and the chief differ-
ence is in extractive matter.

The esters and adds in both have reached their equilibrium, and the
maturity of the two samples, as far as these constituents are concerned,
has been attained, but the marked difference in the flavor is due, with-
out question, to the matter extracted from the barrels.

In order to determine the flavor due to the extractive matters, 100
cc. of each sample were taken and evaporated to 5 cc., so that as much as
possible of the esters and aldehydes and other volatile substances would
be driven off.

To this residue were added a few cubic centimeters of water. The
odor of the residue from the charred package was very strong, having
the peculiar aroma of American whiskey. The taste of this residue was
also very strong, being both resinous and astringent, and there can be no
doubt that these materials, which are left in the residue, make up, to a
great extent, the peculiar aroma of American whiskey, as distinguished
from Irish or Scotch whiskey.

The residue from the uncharred package has a somewhat similar odor,
and a slightly sour, astringent taste, entirely different from that of the
charred package. This great difference in the flavor and odor of the
residues from charred and uncharred packages is to a great degree the
difference in the flavor and odor of the matured whiskey.

Whether all of these solids come from the package has not been proved,
as part of these solids may come from the oxidation of some of the vola-
tile oils which are distilled over, rendering them non-volatile, but a sam-
ple containing only traces of fusel oil, which had been produced with a
rectifying still by which practically all of these oils are removed, and
which had been aged in charred wood, was examined, and the solids pre-
pared as above, had practically the same flavor. This at least indicates
that the peculiar flavor which characterizes American whiskies, as com-
pared with Scotch and Irish whiskies, is largely due to the kind of pack-
age in which they are aged.

It cannot be considered, however, that the adds, esters, and other
ingredients do not enter into the flavor of the matured product, but the

STUDY OF WHISKEY STORED IN WOOD. 135

peculiar flavor of American whiskies wlfich distinguishes them from other
whiskies is undoubtedly derived from the extractive matter from the
charred wood in which they are stored while aging.

This fact is brought out by sample 2612, which was alcohol aged in
wood, and lacked the taste of the whiskies produced in other type of
stills, in which little rectification takes place.

Flavor.

It must be remembered that the flavor of these whiskies was deter-
mined at the same time, when all were of the same age, about nine years —
the new spirit having been this length of time in bottle — and when we
consider that it still had the odor of new whiskey, and showed no develop-
ment of adds and esters, but was still raw spirit, it seems proved that
none of the maturing processes take place in glass, and that whiskey after
it is bottled will not improve.

In fact, none of the changes that take place in the wood occur in the
bottled sample. Take, for instance, the year-old samples which have
a very considerable amount of color and solids, and yet do not, in the
eight years in which they are in glass, change as to solubility, like the
solids and color of whiskey kept in wood. In fact in no way was any
change indicated in the samples in glass.

No effort was made to judge when a certain whiskey had matured.
The only comparison made was of the new spirit, the spirit four years
in wood, and that eight years in wood, and then only in a general way
to see whether the four- and eight- year old goods could be considered
good whiskies, and whether the new spirit could in any way be said to
have improved. In no case could the new spirit be said to be anything
but raw spirits with the new whiskey flavor, not to be compared with
the spirit one-year old.

The judging of the flavor and taste of the whiskies was done by a special
gauger of the Bureau of Internal Revenue, who had a long experience
in deciding between straight whiskies and compound whiskies, and has
shown his ability to detect the diflFerence. The samples were submitted
without name for an unprejudiced decision. The results will not be
given in detail, as all of the matured samples would doubtless be con-
sidered good whiskey, and a judgment as to quality is very difficult and
unsatisfactory, and unnecessary for this paper.

Special attention was given to a few samples which were peculiar:
No. 2612, which was practically cologne spirits aged in wood, No. 2623,
which was abnormal in content of acids and esters, No. 2689, which was
abnormal in content of esters, and Nos. 2625 and 2627, which were stored
in uncharred packages.

The judgment of No. 2612 was that it had the peculiar aroma of Amer-

136 ALVIN S. WHEEI<ER.

ican whiskey, but had the taste of spirits, and little difference could be
noted between the two-year old sample and the eight-year old.

The judgment of No. 2623 was that it was a good flavored whiskey,
but with very light body.

No. 2689 was pronounced a good flavored whiskey.

Nos. 2625 and 2627, the samples aged in uncharred packages, were both
called immature whiskey, even when the eight-year old sample was tested.
But this judgment was based on a comparison with those aged in charred
packages.

On that basis the eight-year old sample was thought to be about i to
2 years old. This shows what a marked effect the storage in charred
wood has upon the flavor of the whiskey.

Conclusions.

1. There are important relationships among the acids, esters, color*
and solids in a properly aged whiskey, which will differentiate it from
artificial mixtures and from young spirit.

2. All of the constituents are undergoing changes as the aging pro-
cess proceeds, and it is evident that the matured whiskey is the result
of these combined changes.

3. The amount of higher alcohols increases in the matured whiskey
only in proportion to the concentration.

4. Acids and esters reach an equilibrium, which is maintained after
about three or four years.

5. The characteristic aroma of American whiskey is derived almost
entirely from the charred package in which it is aged.

6. The rye whiskies show a higher content of solids, acids, esters, etc.,
than do the Bourbon whiskies, but this is explained by the fact that heated
warehouses are almost universally used for the maturing of rye whiskies,
and unheated warehouses for the maturing of Bourbon whiskies.

7. The improvement in flavor of whiskies in charred packages after
the fourth year is due largely to concentration.

8. The oily appearance of a matured whiskey is due to material ex-
tracted from the charred package, as this appearance is almost lacking
in whiskies aged in uncharred wood.

9. The "body" of a whiskey, so-called, is due largely to the solids ex-
tracted from the wood.

[Contribution from thb Chemical Laboratory op the Univbrsity op North

Carolina.]

THE COITOENSATION OF CHLORAL WITH PRIMARY AROMATIC

AMINES, n.

By Alvin S. Whbrlbr.
Received October ai, 1907.

A number of condensation products of chloral with primary aromatic

PRIMARY AROMATIC AMINES. 137

amines have already been described. The first mention of such a re-
action is probably that of Maumen6,* who hoped to obtain indigotin by
the action of chloral (2 mols.) upon aniline (3 mols.). His product was a
brownish black uncrystallizable substance containing no chlorine. Schiff
and Amato' first described a condensation product of chloral (i mol.) and
aniline (2 moK) with the formula CCl8CH(NHCeH^)2. In the same year
Walla ch* described this comix)und. Later/ he gave a full description of
the products obtained from aniline, /)-toluidine, and a sample of xylidine
boiling at 212-216°. Eibner* studied the condensation of chloral with
/>-nitraniline, w-chloraniline, />-chloraniline, and 1,2,4-dichloraniline and
showed that 1,2,4,6-trichloraniline and 2,6-dichlor-4-nitraniline do not
react. Wheeler and Weller^ prepared the o- and w-nitraniline com-
pounds and Wheeler and Daniels^ showed that only addition products
could be obtained with the naphthylamines. Niementowski and
Orzechowski® found that one molecule of chloral condensed with one
molecule of anthranilic acid but later* obtained the expected diphenamine
compotmd. Finally Riigheimer*® describes the compounds with 0- and
/>-phenylenediamine and 1,2,4- and 1,3,4-toluylenediamine. He also
states that only addition products arc* obtained with the naphthylamines.

The chloral diphenamine compounds vary considerably in stability.
Most of them may be kept for years. They possess great crystallizing
power. Their behavior toward alkalies is variable. The aniline deriva-
tive is decomposed by alcoholic potash into aniline, chloroform and
phenyl cyanide according to Wallach. The />-nitraniline derivative, on
the other hand, is converted into a hydroxy compound, one chlorine
atom being replaced by a hydroxyl group according to Wheeler and
Glenn." They are not stable in the presence of strong mineral adds,
which split them so as to re-form the amine. Eibner has shown that
boiling acetic anhydride and benzoyl chloride give the acetyl or benzoyl
derivative of the original amine. Finally, the writer has found that all of
them react with great readiness with bromine in the cold. There is a sub-
stitution of one hydrogen atom in those which have been analyzed. This
substitution probably occurs in the methylene group of the chloral residue.

* Bcr., 3, 246 (1870).

* Gazz. chim. ital., i, 376 (187 1).

* Ber., 4y 668.

* Ann., 173, 274.
» Ibid., 302, 335.

* This Journal, 24, 1063.

* J. Elisha Mitchell Sci. Soc, 22, 90 (1906).
•Ber., 28, 2812.

* J^Wd., 35, 3898.
»• Ibid,, 39, 1653.

" J. Elisha Mitchell Sci. Soc., 19, 63 (1903).

138 ALVIN S. WHEELER.

Chloral and p-Bromaniline. Trichlorethylidenedp-p-bramphenamine,
CCljCHCNHBiCeHJ,. With c. w. miller. — ^Ten grams of /^-bromaniline
were dissolved in 50 cc. benzene and 8 grams chloral (4.2 grams required
by theory) in 10 cc. benzene were added. The mixture was concen-
trated one-half on the water bath and cooled. A white flocculent pre-
cipitate came down, giving a melting-point of 135®. On further evapora-
tion a second and much larger crop was obtained, showing a melting-
point of 119®. By several recrystallizations from benzene the melting-
point was raised to 140°. The yield was very high. Anal3rsis: calcu-
lated for CMHuNaClgBr,, C 35.45, H 2.34, N 5.93, CI 4- Br 56.24; foimd,
C 35.03, H 2.46, N 6.38, CI -I- Br 55.58.

Trichlorethylidenedi-/>-bromphenamine consists of fine colorless needles,
melting at 140° and decomposing at 205°. It is extremely soluble in
alcohol, acetone, glacial acetic add and hot benzene. It is sparingly
soluble in cold benzene and insoluble in ligroin. It is readily purified by
using a mixture of benzene and ligroin. It is not decomposed by boiling
water but is split by boiling concentrated hydrochloric add with the
regeneration of />-bromaniline. A bromine derivative is easity obtained
by adding bromine to a gladal acetic add solution. The product, con-
sisting of colorless plates, melts at 203° after several recr)rstallizations
from glacial acetic acid. Determinations of carbon, hydrogen and nitro-
gen give very satisfactory figures for a monobrom compound. Chlorine
gives a similar reaction. The product, crystallizing in long colorless
needles, melts at 93° after recrystallization from glacial acetic add. A
study of the constitution of these halogen derivatives is imder way.

Chloral and o-Anisidine. Trichlorethylidenedi-o-methoxyphenamine^
CCl3CH(NHOCH3CeHJj. With w. s. dickson.— Two molecules (12.3
grams) of o-anisidine were dissolved in 50 cc. benzene and one molecule
(7-3 grams) of chloral was added. After warming a short time on the
steam bath a small quantity of colorless needles deposited. These de-
composed at about 215° and weighed 0.95 gram. On concentration of the
filtrate in a desiccator a mass of fern-like crystals was obtained mixed with
a thick liquid. After filtering, the crystals were pressed on a porous tile.
The product was white, melted at 112-114® and weighed 9.7 grams. On
recrj'stallizing from benzene the melting-point was raised to 121®. The
thick liquid finally solidified, considerably increasing the yield. Analysis:
calculated for CwHiyO^NaCl,, CI 28.35, N 7.47; found, CI 28.35, N 7.3a

Trichlorethylidenedi-o-methoxyphenamine crystallizes from ligroin or
benzene in magnificent rhombohedra, one-half inch or more long, with a
slight yellow color. It is easily soluble in cold benzene and carbon tetra-
chloride and in hot glacial acetic acid. It is slightly soluble in cold ligroin
and fairly soluble in hot ligroin. It crystallizes from alcohol in long
slender prisms. One hundred cc. of boiling alcohol will dissolve ap-

PRIMARY AROMATIC AMINES. 139

proTcimately 7 grams and at 25^ about 2.5 grams. It is insoluble in and
unchanged by boiling water. When boiled in concentrated hydrochloric
add the odor of chloral could be detected in the vapors. A bromine
derivative is readily obtained by adding bromine to a concentrated glacial
acetic acid solution. The product crystallizes in clusters of needles which
decompose at about 230°. This compound is being further investigated.

Chloral and p-Anisidine. Trichloreihylidenedi-p-meihoxyphenamifief
CCl,CH(NH0CH,CflHj2. — ^To a solution of 12.3 grams of />-anisidine in
20 cc of benzene (a nearly saturated solution) is added 7.3 grams of chloral.
The solution turns to a dark red color at once, much heat is developed
and a deposition of 0.22 gram of small colorless crystals occurs. These
decompose at about 215^ as in the case with o-anisidine. After filtering,
the reaction mixture is boiled fifteen minutes and then allowed to stand
several hours. An abundant crystalline precipitate formed. After
filtering and pressing on a clay plate, the product melted at 115^ and
weighed 10.5 grams. A further yield was obtained from the mother-
liqtior. Purification by means of the mixed solvent, benzene and ligroin,
raised the * melting-point to 118-120^. Analysis: calculated for
C„H„0,N,Cl3, CI 28.35; found, 28.41.

The para compound crystallizes from ligroin in brilliant scales, show-
ing a strong pink color in the mass. It melts at 118-120° and decom-
poses at 158°. It is fairly soluble in cold benzene, alcohol and ether.
It is readily soluble in glacial acetic add, hot benzene and hot alcohol.
The alcoholic solution emits a most disagreeable odor and on spontaneous
evaporation to dryness a jet black crystalline mass remains. On treat-
ment with bromine in glacial, acetic acid solution a crystalline product is
obtamed which blackens at about 198*^. This compound is being studied
further.

Chloral and Anihrantlic Acid. — The product obtained in this case de-
pends upon the proportions used. One molecule of chloral will con-
dense with one or two molecules of anthranilic add with the elimination
of one molecule of water. The two products have been described by
Niementowski but his method yields a mixture and since we wish to pre-
pare the compounds in order to study their bromine derivatives we have
improved upon his method.^

Trichlorethylidene-o-aminobemoic A cid (Chloral-anUiraniltc A cid) ,
CCliCH:NCeH,COOH. With w. S. dickson.— Five grams of anthranilic
add were dissolved in 40 cc. of boiling benzene (a saturated solution)
and 5.5 grams chloral in 10 cc. benzene were added. The wdghts are in
the proportion of one molecule to one molecule. The mixture was boiled

' Since writing the above, larger quantities of the mono-compound have been pre-
pared in benzene solution and a small amount of the di-compound has been isolated
from the product.

140 ALVIN S. WHBELBR.

under a reflux condenser for three hours, filtered from a small precipitate
and cooled. A crystalline deposit, weighing 5 grams and melting at
148-151°, separated. The crystals were large elongated tables occurring
in clusters. From the filtrate was obtained 3 grams of material, melting
at 145-150°. Several recrj'stallizations from benzene raised the melt-
ing-point to 152°. Niementowski and Orzechowski^ prepared this com-
pound without the use of any solvent. They used an excess of chloral
and got several by-products. We have tried their method but have
employed theoretical proportions. Even so we get the same by-products.
The mortar was placed in a block of ice and the previously cooled sub-
stances rapidly stirred together. The mixture liquefied and then rapidly
became very hard. This product decomposed at about 127°, after two
hours on ice at 124° and after three hours more at room temperature at
118°. It was then rubbed up with a little water and filtered. The de-
composition point rose to 135°. Now, taking advantage of the marked
difference in solubility in benzene of the mono- and di-compounds (not
observed by Niementowski) the crystalline mass, weighing 8.2 grams, was
extracted with 45 cc. boiling benzene. From the extract there separated
a mass of colorless needles, weighing 3.7 grams and melting at 149-152®,
consisting therefore of the nearly pure mono-compound. On evaporating
the filtrate a residue was obtained, weighing 1.3 grams and melting at
160°, a fair quality of the di-compound. A second extraction was made
with 35 cc. of boiling benzene. On cooling, this yielded a product weigh-
ing 0.8 gram and melting at 162° and a residue melting at 157** after
evaporation. There still remained an insoluble residue, dark purple in
color. These results are in marked contrfist to those obtained by our
method of boiling in benzene, for we get practically only the mono-com-
pound and consequently a much larger yield. We further identified
the mono-compound by a chlorine determination. Calculated for
C,HANC1„ CI 39.92; found, CI 39-43.

On treating a glacial acetic acid solution of this compound with bromine
a bromo derivative is obtained in large quantity. On cooling a hot
glacial acetic acid solution, it deposits in clusters of fern-like cr^'^stals which
decompose at 237''. This compound is under investigation.

TrichlorethylidenedirO-aminobenzoic Acid (Chloral Di-anthranilic Acid),
CClsCH(NHCeH,COOH)a.— Five grams (2 mols.) of anthranilic acid in
40 cc. boiling benzene were treated with 2.9 grams (i mol.) of chloral in
10 cc. benzene and boiled under a reflux condenser for three hours. Dur-
ing the boiling there separated 3.25 grams of the di-compound, melting
at 164-165®. On evaporation to dryness the residue was found to weigh
4.0 grams and to melt at 157*^. The pure compound melts at 165^. The

^ Ber., 28, 2812.

PRIMARY AROMATIC AMINES. I4I

method of Niementowski^ was tried and although found to be better
than for the preparation of the mono-compound, it gave a smaller jdeld
and a larger amotmt of unknown colored by-products. Analysis: calcu-
hted for C^H^jO^NjClj, N 6.96, CI 26.11; found, N 6.76, CI 26.10.

The di-compound consists of a crystalline powder and may be purified
by precipitating its ether solution with ligroin. Upon boiling eight hours
with acetic anhydride and cooling, a crystalline substance deposits,
melting at 183° and crystallizing from benzene in needles. This corre-
sponds to acetyl-o-aminobenzoic add. On treating a glacial acetic acid
solution with bromine, there is almost instantly obtained a heavy pre-
cipitate which after recrystallization from glacial acetic acid melts with
decomposition at 236°. This behavior is surprisingly like that of the
biomo derivative of the mono-compound.

Chloral and o-Toluidine, Trichlorethylidenedi-o-tolaminey CCI3CH
{NHC^4CHj)2. With strowd Jordan. — Chloral and o-toluidine were
brought together directly in the proportion of one molecule to two mole-
cules. No advantage was found in using benzene as a solvent. To
28 grams o-toluidine 19.3 grams chloral were added; the mixture turned
dark red and the temperature rose to 80°. After standing for some time,
often over night, a quite hard crystalline cake formed. This was dis-
solved up in ether or successively extracted with benzene. In either
case, a small residue weighing 0.7 gram remained. This was pale greenish
in color and melted at 213°. The main product of the reaction was re-
crystallized from ether until the melting-point reached 80°. The yield
was 70 per cent, of the theoretical. Analysis : calculated for Cj^H^NjClj, CI
30.95; found, 30.77, 30.40, 30.96. The Stepanow method was employed
in the secend and third analyses and found to be extremely convenient.
With somo of our compounds we have found it impracticable on account
of the deep color of the solution. We foimd it advisable to follow the
suggestion of Rosanoff and HilP and filter off the silver chloride before
titrating.

Trichlorethylidenedi-o-tolamine crystallizes in very long silky needles.
It is not very stable in solution or when exposed to the light. It is de-
composed by water into chloral and o-toluidine. It is soluble in cold
alcohol, ether, acetone, chloroform, carbon tetrachloride and glacial
acetic acid, in hot ligroin and hot benzene. The pure substance melts at
80® and will melt repeatedly at that temperature. A bromine derivative
is readily obtained in glacial acetic add solution. This consists of color-
less plates which melt with decomposition in the neighborhood of 268®.

Physiological Action. — ^The tolamine was found to have a physiological
action by an accidental observation. Mr. Jordan unintentionally got a

' Ber.. 35» 3898.

' This Journal, a9,*269.

142

NOTES.

small amotuit in his mouth and in about an hour a feeling of numbness
spread over him to such a degree that pinching the flesh produced little
sensation. Mr. Jordan was frightened by his condition but in an hour
and a half he returned to a nearly normal state. The physiological
action of this and other diphenamine compotmds of chloral will be studied
with care, some preliminary experiments on rabbits by Dr. MacNider. of
this University, having confirmed the observation of such an action.

Crapbl Hill, N. C,
October i6, 1907.

KOTES.

The boiling point of isohviane, which is given in the literature, is based
on a determination by Butlerow.^ He found that the gas begins to con-
dense to a liquid at — 17°.

Some years ago Mabery^ isolated from petroleum a hydrocarbon which
boiled at o® and which he considered to be isobutane on the basis of the
chloride obtained from it, which boiled at 68-69° and which he considered
to be isobutyl chloride. Since 2-chlor-2-methyl-propane boils at 67.3-
67.8°* and has a specific gravity closely approaching that of Mabery's
product, it seems probable that the chloride which he obtained was in
reality a derivative of normal butane, and not of isobutane. This view
is further supported by the work of Pelouze and Cahours,* who foimd
that a chloride boiling at 65-70° is obtained by the action of chlorine on
normal butane.

It seemed of interest to prepare isobutane again, and make a new
determination of the boiling-point. This was done by Mr. E. F. Phillips
under my direction in the laboratory of the Rose Pol)rtechnic Institute
several years ago, and the results were reported to Professor Mabery, think-
ing that he would, at some time, publish something further upon the
subject. As he has not done this and informs me that he does not ex-
pect to take up the subject again, it seems proper to give the results of our
experiments.

The isobutane was prepared by the reduction of isobutyl iodide with
zinc and dilute alcohol. The gas was purified and dried by passing it
through bulbs containing alcohol and also bulbs containing concentrated
sulphuric acid. It was condensed to a liquid by a freezing mixture and
the temperature at which the vapor of this liquid exerted a pressure of
760 mm. was determined. This temperature was found to be — 11.5°.
A considerable part of the liquid was allowed to evaporate and the deter-

^ Ann., 144, 13.

^ Am. Chem. J., 19, 247.

■ Norris and Green, /Wrf., 36, 308.

* Jsb., 1863, 524. . , V *

RBVIBW. 143

minatioii repeated, proving that the liquid was practically homogeneous.
A determination of the density of the gas also gave results agreeing satis-
factorily with the theory.

There seems, therefore, to be little question that the butane obtained
by Professor Mabery was in reality normal butane, and that the boiling-
point of isobutane is — 11. 5**. W. A. NovES.

The following note has been received from Prof. Mabery to whom this
note was submitted in manuscript:

'*! have no reason to doubt the accuracy of Professor Noyes' observa-
tion on the boiling-point of isobutane. It is not incompatible with our
results on the butane in petroleum. I have intended to refer to this
subject more fully in a later paper, a resum^ of the composition of
American petroleum." C. F. MabeRV-

Univbrstty of Illinois,
Urbana, III.

The Stereochemistry of Indigo. — ^The last paragraph of this paper (This
Journal, December number, 1907, p. 1743,) in which the structures of the two
diacetyl indigo whites are discussed on the spatial hypothesis, assmnes that
both compounds possess the ketone structure. If both rearrange to the
tautomeric enol forms, it should be pointed out that there would still be
two stereoisomers (cis and trans) which it would not be possible to resolve
intp optically active isomers. K. George Fai^k and J. M. Nelson.

REVIEW.

RESEARCHES ON THE DENSITY OF GASES

CARRIED ON DURING I904, I905 AND I906 IN THE PHYSICAL CHEMISTRY

LABORATORY OF THE UNIVERSITY OF GENEVA.^

By Philippe A. Guyb.
Received October i, 1907.

The present article contains a risvimi of the results obtained during
the course of three years* work on the exact density of gases. The work
has been carried on in collaboration with Messrs. Jaquerod, Pintza, Davila,
Oazarian and Baume, and until now has been the subject of only isolated
publication (Jaquerod and Pintza, Compt.rend., 139, 129 (1904), (SO,
and 0,); Guye and Pintza, Ibid., 139, 679 (1904); 141, 51 (1905), (N3O,
CO, and NHg); Guye and Davila, Ibid,, 141, 826 (1905), (NO); Guye and
Gazarian, Ibtd., 143, 1233 (1906), (HCl); Baume, unpublished (1907),
(SO,)). These have contributed to the problem of the physicochemical
determination of exact molecular weights, with a view to checking up the

* From Archives des Sciences Physiques et NcUwreUes, 34, 32-62. Translated by
Hden Isham.

144 REVIEW.

atomic weights. This problem has become one of great importance, and
has been taken up by Leduc (Leduc : Recherches sur les gaz, Paris, Gauthier-
Villars (in part), Ann. Chim, Phys. (1897)), Lord Rayleigh (Rayleigh, Proc,
Roy. Soc, 43, 353 (1888), (H3andOa);50,449 (1892), (HjandO,); 53» i34
(1893), (0„ Hj, Nj atm. air); 55, 340 (1894), (N,); S9i 198 (1896), (Ar,
He); 62, 204 (1897), (CO, CO2, NjO); 74, 181 (1904), (N,0); Phil. Trans.
Roy. Soc. 204 A, 351 (1905); Rayleigh and Ramsay, Phil. Trans. Roy.
Soc. 186 A, 187 (1895)), Morley, (Z. physik. Chem., 20, i, (Oj);
22, 2, (Hj), (1896)), Gray, (/. Chem. Soc, 87, 1601 (1905), (N,;
O2, NO); Proc. Chem. Soc. (1907), (HCl)), as well as Perman and Davies
(Perman and Davies,* Proc. Roy. Soc. 78 A, 28 (1906)). Details of the
methods, and the method of calculating the results, will be given under
the head of generalities ; next a review of the results obtained at Geneva
compared with those of other experimenters, and finally a summary of
those values which appear to the author to be most reliable.

I. Generalities.

Methods. — ^Two general methods have been used in density determina-
tions, that of the balloon and that of the volumeter.

The balloon method, the details for which have been worked out by
Regnault (weighing the balloon empty and filled with gas, against a
counterpoise of the same volume, reducing to weight in vacuOy etc.) has
been used in recent work, taking into account the correction for the con-
traction of the empty balloon, and the correction for the compressibility
of the gas (deviation from the law of Mariotte between the existing pres-
sure and that of 760 mm. Hg). The recent work is also characterized
by a gradual reduction of the capacity of the balloon. Morley worked
with balloons of from 8 to 21 liters capacity. Rayleigh used one of about
1.8 liters capacity, Leduc one of about 2.3 liters. The work in this lab-
oratory has led to a still further reduction of the capacity of the balloon,
and two balloons, destined to act as checks upon each other, one 0.8
liter, the other about 0.4 Hter in capacity, filled at 0° and under the same
pressure conditions, have been used.^

Perman and Davies have used a balloon of 0.5 liter capacity, while
Gray has reached the lowest unit, and in the determination of nitric
oxide used one of about 0.267 liter. Contrary to what one might
expect, a priori, the determinations with the small balloons are at
least as concordant, among themselves, as those with balloons of
larger volume. The experiments with the two-liter balloons be-
low, show, on the whole, a better agreement than those with
balloons varying from 8 to 21 liters. This may be due either to the fact
that the corrections for the original weights (for contraction, etc.) are
smaller with the small balloons, or that the chances of accidental error,
especially those due to the condition of the surface of the glass, are dimin-
ished.

The only objection which can be raised to the use of the small balloons
is the possibility of the condensation of gas on the inner wall. If there
were such condensation, the density of gases measured in small balloons
should be greater than in the larger balloons, and this error should be
especially great in the case of the hygroscopic gases, sulphur dioxide,
ammonia, hydrochloric add gas, etc.

^ Experiments using 3till smaller balloons are being carried on at present.

REVIEW. 145

The recent determinations show that while the capacity of the bal-
kxms varies to a great extent, the results obtained are very concord-
ant, especially in the case of ammonia and sulphur dioxide.

By the volumeter method (capacity = about 1.8 liters) Perman and
Davies found 0.77085 gram for the weight of a liter of ammonia gas under
standard conditions, while by the balloon method they obtained 0.77086
gram for the same gas, using a balloon of 0.5 Uter capacity. Guye and
Pintza, using a volumeter of 3.5 liters capacity, found 0.7708. Thus there is
no appreciable difference between the values obtained by the large and
small balloons. Jaquerod and Pintza have obtained the value 2 .9266 grams
for the weight of i liter of sulphur dioxide, using a volumeter of 3.5 liters
capacity; Leduc, using a 2.3 liters balloon, found 2.9266; and Baume
bas recently repeated this determination, using two baUoons of 0.53 liter
and 0.32 liter capacity, respectively, on gas prepared under the same
conditions as that used by Jaquerod and Pintza, and has found 2.9266
grams.

The only necessary precaution consists in rinsing the balloon, not only
with air dried over phosphorus pentoxide, but also several times with
the well-dried gas, the density of which is to be determined, taking care
that between each rinsing operation the vacuum in the balloon is as per-
fect as possible, and that air does not enter the balloon between each
determination.

It might, nevertheless, be assumed that there is a slight condensation
of the gas on the inner wall of the glass, similar to the moisture adhering
to glass. The author's experience (in collaboration with Gazarian) on
the density of hydrochloric add gas, which is extremely hygroscopic,
does not justify this assumption. In fact the balloons were at first
rinsed only two or three times with dry hydrochloric acid gas. Under
these conditions, with a balloon of 0.385 liter capacity, the values for
the density showed a regular decrease, and did not become constant
nntil after the fourth determination. Using a larger balloon (0.818
Kter capacity) this point was not reached after the seventh filling. These
facts led to the conclusion that after twelve successive rinsings^ with the
dry gas, each rinsing followed by a complete evacuation of the balloon,
the condensation of the gas on the interior wall of the balloon is not suf-
ficiently great to affect the results to an appreciable extent, provided
air has not entered the balloon during the process of rinsing. (If Bun-
sen's observations are correct, the moisture given off by the glass at high
temperatures is not due to condensation on the surface, but to water
incorporated chemically or physically in the mass of the glass.) ^

* In the case of non-hygroscopic gases 5 to 6 successive rinsings, each followed
by complete evacuation, are sufficient.

' If we admit that the condensation of the gas on the interior surface of the bal-
loon is not generally appreciable, still we do not daim that it is absolutely nil. In
order to determine the condensation most accurately, two or more balloons of very
different capacities should be filled at the same time, under the same conditions of
pressure and with gas from the same source. The determinations of the densities
of nitric oxide and sulphur dioxide, using two balloons filled simultaneously, furnish
some preliminary information on this subject. The following averages have been

146 REVIBW.

The volumeter method, used as an exact method for the first time by
Morley in the determination of the density of hydrogen, has been used
in two ways. According to the first method the gas is evolved from an
apparatus which may be weighed, and constructed in such a fashion
that the gas alone, in,a state of perfect purity and dryness, escapes. The
gas then passes through a tight-fitting joint into a system of one or more
balloons of known capacity, maintained at o*', and which have been evacua-
ted by a mercury pump. The pressure is measured either with a special
manometer, or simply by connecting the apparatus with the tube of a
barometer. Morley employed the latter method in his fourth series
of measurements of hydrogen, and it has also been employed at Geneva.
Knowing then the temperature and the pressure, the capacity of the bal-
loons and the loss of weight of the generating apparatus, the density
may be calculated.

According to the second method the volumeter, previously evacuated,
is filled with the pure gas. The process of purifjdng may be more com-
plete, as there is no Umit to the weight of the apparatus used for that
purpose. The volumeter being filled with the pure gas is held at o®
and the exact pressure (about 760 mm.) is read. Then the gas is ab-
sorbed in a suitable apparatus, previously evacuated and weighed, and
connected to the volumeter by a tight-fitting joint.

The densities of hydrogen (Morley), oxygen and sulphur dioxide
(Jaquerod and Pintza) have been determined by the first method, by
the second those of nitrous oxide, carbon dioxide and ammonia (Guye
and Pintza), and, more recently, ammonia (Perman and Davies).

The volumeter method used does not take into accotmt the correction
for the contraction of the balloons, but only that necessary for the reduc-
tion of the weights to weights in vacuo (for the evolution or absorption
apparatus is weighed with a counterpoise of the same glass and the same
volume) and finally the pressure coeflfident of the gas in order to calcu-
late the volume at a pressure of 760 mm.

Whichever method is employed (balloon or volumeter), the correc-
tion for the pressure coefficient of the gas is negligible, since the pressure
differs from 760 mm. by only a few milUmeters.

found for the weight of a liter of these gases, determined simultaneously with two
balloons.

Wbight of a Litbr of no (G. and D.).

No. of obser- Capacity of Weight of

vation. balloon. i liter.

I«iter. Gram.

7 0.8 Z.34OX

7 04 1.3403

5 0.5 2.92666

4 0.3 2.92659

It may be claimed that this condensation was not on the surface of the glass*
but in the special grease used to insure a tight cock. It may be well to cite here the
recent work of Swinton {Chem. News, 95, 1349 (1907)), who has shown that the gases,
hydrogen and helium, have no chemical action on the surface of glass, and that the
surface condensation is of a purely mechanical nature. The recent observations of
Travers {Proc. Roy. Soc. 78 A, 9 (1906)) on the condensation of gases by solid bodies
should not be overlooked, if the density of a gas is to be determined with an accuracy
exceeding i part in 10,000.

tmviitfw. 147

Choice of a Unit. — ^A common unit is indispensable. Leduc has re-
duced bis vahies eitber to tbe density witb regard to air, or to the weight
of a Hter of gas at Paris, or both. Morley expresses bis results in weight
per liter under standard conditions, that is, at o^ and 760 mm. pres-
sure, at sea level and latitude 45°. Rayleigh calculates his results either
to density compared with air or compared with oxygen (taken equal
to 32) or as tbe weight of i liter at London or at Paris. At Geneva the
weight of tbe liter under standard conditions has been directly deter-
mined. Gray, as well as Perman and Davies, has likewise used this unit,
which is tbe unit for all results which are given in the following
pages.

D. Bertbelot (CompL rend., 144, 269 (1907)) has recently calculated
the greater number of measurements, comparing densities to that of
oxygen. Instead of taking one value for oxygen, he adopts, in each case,
the value obtained by each investigator for the density of this gas. This
system presents several disadvantages: First, in the case of those deter-
minations made by investigators who have not determined the value
for oxygen, the choice of such a value becomes somewhat arbitrary.
Second, it might happen that the value for oxygen, even when deter-
mined by the investigator, was the most inaccurate of all his determina-
tions; thus the accuracy of his other values would be diminished. This
is the case, for example, at Geneva, where the value for oxygen has been
determined with less precision than for the other gases.

These disadvantages disappear when all results are calculated to the
weight of one liter of the gas under standard conditions. All the inves-
tigators, with the exception of Leduc, have recorded the exact volume
of the balloons employed. Leduc's values may be calculated, since he
has determined the weight of a liter of air at Paris.

There is, finally, another reason for expressing the density by the
weight per normal liter, and that is the fact that the volume of the bal-
loon can be calibrated more accurately than the weight, of a gas for
comparison, can be determined in the same balloon. Leduc reports that
after an interval of two years the capacity of a balloon originally found to
be 2.27636 liters, was 2.27630 liters, or tiie variation was i : 38,000. At
Geneva an accuracy of i part in 30,000 to 35,000, between successive
calibrations, has been easily realized. It is necessary to determine the
capacity of the balloon for water at 0°, and under these conditions no
correction for the expansion of the glass is required.^

Choke of Final Values. — ^The following are some of the rules which
have guided us in making a choice between the various values obtained
for the density of each gas.

First, the nature of tie method employed for the production and puri-
fication of the gas has been taken into consideration. This is, to our
mind, the most important consideration in the density determinations.
Those results showing concordance in the values obtained for the gas
ptepared in two or three different ways have, therefore, been given most
weight.

> Difficulty was experienced in weighing the balloon filled with water at o^, as
tile water expands when the balloon is placed on the balance. This has been over-
come by attaching a bulb tube to the tubule of the balloon, by means of a ground
joint or a rubber tubing, the whole apparatus having a counterpoise of the same glass
and the same volume.

148 REVIEW.

A second consideration, equally important, is the concordance obtained
by different investigators for the same gas. When the average values
of two observers agree, while that of a third experimenter differs by a
considerable amount, they have not been given equal weight, but the
third value has been given less weight than the first two. Frequently,
an experimenter has made a more or less accurate determination of the
limit of error of his work. We have applied the correction in such a way
as to bring the value nearer the mean of the other two observers. Simi-
larly, when an error was discovered in the course of subsequent work,
we have not taken into account that value known to be erroneous.

As will appear in the following discussion, while the preceding rules
are suggested, an accurate choice is difficult to make. We would add
that in abandoning several of Leduc's values, giving preference to others,
generally more recent, in their stead, there has been no intention of di-
minishing the importance of Leduc's work. If the modem determina-
tions have benefited by a more perfect technique, still we must
recognize that to Leduc belongs the credit of having been the first to
give a complete solution, theoretical and experimental, of the problem
of the rigorous determination of the molecular weights of gases as a func-
tion of their densities.

II. Discussion of Results.

Oxygen, — The system of atomic and molecular weights being referred
to oxygen (0= 16), the exact determination of the density of this gas is*
of considerable importance.

Morley (Z. physik, Chem,, 20, i (1896)) has made 41 determinations
of this value, of which the averages of three series are :

No. of deter- Weight of a stand- Capacity of

Series. minations. ard liter. baUoon.^

Grams loiters.

I. 9 1.42879 21.6 and 8.8

11. (a) 6) „„ . tt ,

).v > 1.42887 20.06 20.56

III. (a) 7 ) „ o , ti

(J) jqJ- 1. 42917 8.83,16.52 15.38

In the first series the temperatures varied between 15° and 20°, meas-
ured with mercury and air thermometers, the apparatus being connected
with a manometer open to the air. In the second, the balloons were
submerged in pounded ice, and the pressure measured by a differential
manometer. In the third, the balloons were submerged in ice, the pres-
sure being read on a barometer, the cistern of which communicated
with the interior of the balloon.

The oxygen was prepared from potassium chlorate and by electrolysis
of a solution of caustic potash for Series 111(6). The extreme variations
between the individual observations were as follows:

—-'- — for Series I, for Series 11(a), — ^— for Series 11(6), — ' —

10,000 10,000 10,000 10,000

for Series III (a), — - — for Series 111(6). Referred to the averages, these

10,000

extreme variations would be reduced to about one-half.
^ Capacities are given in round numbers.

R8VIBW. 149

Morley gives double weight to the average of Series III, and thus ar-
rives at the value, i Hter oxygen = 1.42900 it 0.000034.

Rayleigh (Proc. Roy. Soc, 53, 144 (1893)) used oxygen prepared by
three different methods: (a) by heating a mixture of potassium chlorate
and sodium chlorate; (b) by heating potassium permanganate; (c) by
electrol3^s of water, leading the gas thence through a column of copper
oxide heated to redness. The balloon was filled at 0° under a pressure
of about I atm. at London, its capacity being 1836.52 cc. The aver-
ages of the results (weight of oxygen contained in the calibrated balloon)
for each series are :

No. of expcri- Weight of

Method. ments. oxygen. Extreme error.

Grams.

Chlorates (a) 5 2 . 6269 2 . 3/10,000

(6) 5 2 . 6269 2 . 3/10,000

Permangaiiate 3 2 .6271 i . 1/10,000

Electrolysis (a) i 2 .6271

(6) 2 2.6272

Average, 2 . 62704
Correction for contraction, 0.00056

Corrected weight, 2 . 62760

I Hter of oxygen at Paris = 1.42952.

Dividing the value by 1.00033, the gravity factor, i normal Uter of
oxygen = 1.42905 grams.

Leduc {Recherches sur les gaz, Paris, 1898), using the balloon method,
has reported the density relative to that of air, at Paris. He prepared
the oxygen (a) by electrolysis of an aqueous solution of caustic potash
or sulphuric acid, (6) by decomposition of potassium permanganate,
(c) by electroljrsis of dilute sulphuric add, the gas then passing through
a colunm of hot copper oxide.

By the method (a) the values, referred to air, varied between 1.10501
and 1. 105 16, average «= 1.1051, extreme variation 1.4/10,000. By the
method (6),* 1. 10527; by the method {c) (3 determinations), 1.10521.

Rejecting the results for (a) Leduc adopts 1. 10523 for the density of
oxygen referred to air at Paris. From this the weight of the normal liter is

-^— ^ '- — ~— = 1 .42876 grams (taking into account the weight of a

liter of air at Paris, determined by the same investigator to be 1.29316,
which he claims is accurate to 1/20,000); or in round numbers, allowing
for the accuracy of 1/20,000 claimed by Leduc, i Hter of oxygen = 1.4288
grams.

Jaquerod and Pintza (Compt. rend., 139, 129 (1904)), working in this
laboratory, have made five determinations of the density of oxygen,
obtained by heating potassium permanganate, by filling a volumeter
of about 3.5 liters capacity, at o*' and about i atmosphere pressure.
These determinations, showing an extreme difference of 6/10,000, give
for the weight of the standard Hter

I liter oxygen = 1.4292.

* Lcdtic does not mention the number of determinations. It would appear that
only one was made.

t5o REVtl^W.

These determinations were made by the volumeter method. The
authors consider their results to be a little too high, and for their further
calculations have adopted the value 1.4290.^

Gray (J. Chem. Soc, 87, 1607 (1905)), in the course of his work on the
density of nitric oxide, has made six determinations of the density of oxy-
gen, which show remarkable agreement among themselves. The gas
was prepared by heating recrystallized potassium permanganate. The
results give for the weight of a balloon of 267.43 cc. capacity, filled at
o*^ and one atm. pressure at Bonn, the value 0.38228, the extreme varia-
tion being only 1.6/10,000. A correction brings the capacity of the bal-
loon to 267.388 cc. so that the weight of a normal Hter, taking into account
the factor for the gravity at Bonn, is

r* r 0.38228 - _

I hter of oxygen = — - — tttttz =* 1.42896 grams,

■^^ 0.267388 X 1.000505. -^ ^ »

a value differing from that of Morley's by only 1/36,000.

Risumi, — Morley's determinations, being the most numerous, and
carried on under the most varied conditions, would appear to deserve
the greatest weight, even though the individual variations are relatively
large. Rayleigh's value is 1/20,000 greater than Morley's, while Gray's
is 1/36,000 less. On the other hand, the less accurate work of Leduc and
Jaquerod and Pintza differ from this by 1/7000, one greater and one
less. Whether the mean of all five determinations, or of only the first
three, which merit the greater confidence because of the greater number
of single determinations and the close agreement between the single de-
terminations, be taken, the value is the same — 1.42900. The variations
-f 0.00005 (Rayleigh) and — 0.00004 (Gray), from the mean of Morley's
values, are of the same order as the probable error calculated for the lat-
ter (-f 0.000034).

The value, i liter of oxygen = 1.42900, is therefore adopted, and is
certainly correct to i/ioooo and probably to 1/20000.

Nitric Oxide, — Leduc (Siances soc. franf. physique, 1893, p. 214) de-
termined the density of the gas in 1893, and reported it to be 1.0388
referred to air, from which the weight of a normal Hter can be calculated
to be 1.3429. Leduc suppressed tiis value in his memoirs published in
1898; therefore, it can hardly be taken into account. It is evidently too
high. During the investigation carried on with M. Davila we obtained
very nearly the same value when using a gas which had not been puri-
fied by Hquefaction and fractional distillation, so that it would appear
that such a high value is due to the presence of traces of nitrous oxide
in the gas.

Gray was the first to make an accurate determination of the density
of this gas, using the balloon method. The gas was prepared by the
action of acetic acid on sodium nitrite and potassium ferrocyanide,
washed with caustic potash, dried over phosphorus pentoxide, then
liquefied and fractionated at low temperatures. Six determinations,

^ D. Bexthelot has recently reported all the density determinations made in this
laboratory, referred to this unit 1.4292 {Compt. rend., 144, 260). In view of the few
determinations made, and the reservations made by the authors, this has diminished
the accuracy of our other results. In all our other work we have adopted the value
1.4290. See /. chim, phys., 4, 333.

REVIEW. 151

showing an extreme difiference of 3/10,000, gave as the weight of the gas
contained in the balloon, 0.35851. Six determinations of oxygen (ex-
treme variation 1.3/10,000) gave 0.38228. Adopting Rayleigh's value
for oxygen (1.42905) the value for nitric oxide is calculated.

rx f •* • -J 142905 X 0.35851

I hter of nitnc oxide = r. — tt^- = i-3402 grams.

0.38228 ^^ ^

On the other hand, he calculates, from the capacity of the balloon,* i
Bter of nitric oxide = 1.34011 and i liter of oxygen = 1.42896. Gray
adopts the value 1.3402 grams. This practically conforms with that which
would result if the value O2 = 1.42900 were used.

. 1.4290 X 0.35851

^^*"^N° ^:^28 - = ^•34015.

Guye and Davila have carried out three series of determinations on
gas prepared from three different sources, (a) by the action of ferrous
sulphate on nitric acid; (6) by decomposition by means of mercury
of a solution of sodium nitrite acidified with sulphuric acid; (c) by
the action of sulphuric add on a concentrated solution of sodium nitrite.
The method of purification was in each case the same as that employed
by Gray, with the exception that the washing with caustic potash was
omitted, as it changes a part of the gas to nitrous oxide.

Weighings were made in two balloons of about 0.8 and 0.4 liter capacity,
respectively.

Rxtreme error.

No. of cxperi- Weight of *— ' »

Method. menls. i liter. Large balloon. Small balloon.

Mercury 6 i . 3403 3/10,000 6 . 7/10,000

Ferrous sulphate 6 i . 3402 3/10,000 4 . 5/10,000

Sodium nitrite. 2 i . 3401 3/10,000

The mean of the 14 experiments, together with the mean for each
method, gives i liter NO = 1.3402, a value identical with Gray's and
which we hereby adopt.

Carbon Dioxide. — Rayleigh reports the density of this gas to be 1.52909
as referred to air. He gives no details as to the method, the number of
determinations, or their probable error. The gas was prepared, (a) by
the action of hydrochloric add on marble, (6) by the same acid on sodium
carbonate.

Leduc obtained the mean value of 1.52874 for three very concordant
determinations on a gas prepared by the action of hydrochloric acid on
marble. He adopts the value 1.5288, in view of the probable error due
to traces of air. He also observes that the work of Regnault points to
the value 1.5290.

The value for a normal liter can then be deduced as follows :

I Uter of CO2 = 1.29284 X 1.52909 = 1.9769 (Rayleigh).

I liter of CO2 = 1.29273 X 1.5288 = 1.9763 (Leduc).

Guye and Pintza have determined the weight of a normal liter directly,
with a gas generated by heating sodium bicarbonate, and have found
1.9768, the mean of three experiments having an extreme difference of
0.8/10,000. The value finally accepted,

I liter of COj = 19768,
* By private commtinication.

152 RBvmw.

agrees to about 1/20,000 with Rayleigh's value; it gives a certain weight
to the lower result of Leduc and is justified by a consideration of the proba-
ble error. In the course of the work at Geneva, the difficulty experienced
in obtaining carbon dioxide entirely free from air has been constantly
kept in mind, and the gas which was used in these determinations was
entirely soluble in cautic potash.

Nitrous Oxide, — Leduc, Rayleigh, and Guye and Pintza have deter-
mined the density of nitrous oxide.

Leduc used the commercial* liquefied gas, rectified by distillation. He
obtained the densities, referred to air, of 1.5304, 1.5298, 1.5301. Aver-
age, 1. 5301; extreme error, 3.9/10,000. Rayleigh made his first series
of determinations in 1897 with a gas obtained by the decomposition of
ammonium nitrate, and purified by dissolving in water — heating to ex-
pel the gas, then drying. Five determinations, with a probable error
of 1.7/10,000, gave the mean value 1.52951 referred to air. In 1904 Ray-
leigh (Rayleigh, Proc. Roy, Soc, 74A, 181) repeated his work, using the
commercial liquefied gas, purified in the same way as above, and obtained
practically the same result. He then purified the commercial gas by
fractional distillation at the temperature of liquid air, till the density
remained constant, and obtained, as the mean of three very concordant
observations, with an extreme dLfference of 0.8/10,000, the value 1.5297.

From this last value of Rayleigh's and from Leduc's, the following
values for the weight of the normal liter may be calculated:

I liter of NjO = 1.9780 (Leduc),

I liter of N2O = 1.9777 (Rayleigh).

Guye and Pintza have determined the weight of the normal liter di-
rectly, using the volumeter method. The gas was prepared by the double
decomposition of hydroxylamine sulphate and sodium nitrite. The
average of three experiments, having an extreme variation of 2.8/10,000,
is

I liter of NjO = 1.9774 (Guye and Pintza).

The impurities which might be found in the commercial nitrous oxide
are nitrogen peroxide, air or nitrogen. The fractionation at low tem-
peratures might not bring about a complete elimination of these' impuri-
ties,' or it might be that the decomposition of hydroxylamine sulphate
did not furnish a gas entirely free from air, even though the evacuation
of the apparatus was repeated several times. Under these conditions,
and considering the good agreement of the separate determinations
made by Rayleigh in 1904, we would adopt for the weight of the normal
liter,

I liter NjO = 1.9777 grams,

which is the mean of the values obtained by Rayleigh, Leduc, Guye and
Pintza.

Hydrochloric Acid Gas, — Leduc has determined the density of hydro-
chloric acid gas, produced by the action of sulphuric acid on sodium
chloride, and dried over phosphorus pentoxide, to be 1.2692 as compared
with air. There is no indication of the number of determinations or

* Obtained by heating ammonium nitrate.

^ The washing with alkalies, as practiced commercially, is not sufficient to remove
the last traces of NO^

MviEW. 1 53

their accuracy, although he considers the last decimal doubtful. A
nonnal liter would be 1.29273 X 1.2692 = 1.6407 grams.

Guye and Gazarian have repeated this determination with a gas pre-
pared in the same way, but further purified by liquefaction at the tem-
perature of liquid air and fractionation at a low temperature.^ A pre-
liminary series of four determinations gave a mean of 1.6398, with an
extreme variation of 9/10,000.

Gray, who has undertaken a revision of the density of hydrochloric
acid gas, reports a series of six determinations, with a probable error
of 3.7/10,000, of which the mean is, one normal liter = 1.6397. This
differs from the preceding by 1/16,000. The mean of these two concord-
ant values would be 1.63975 or, suppressing the last decimal, 1.6398.
This value is given only provisionally, until further determinations shall
•be announced.

Ammonia Gas, — Leduc used ammonia gas obtained from a commer-
cial ammoniacal solution called **pure,'' which he dried over molten
caustic potash. Without reporting the number or the accuracy of his
determinations, he gives the value 0.5971 as compared with air, or the
weight of one normal Uter,

1.29273 X 0.5971 = 0.7719 gram.

Guye and Pintza have determined the weight of a standard liter, using
commercial liquefied ammonia. Assuming that Leduc's high value was
due to the presence of organic bases, these experimenters previously
purified the gas by leading it over red-hot quicklime, to transform the
nitrogen of the organic bases into ammonia. The gas was then collected
as ammonium chloride, in this form was recrystalUzed, and was finally
set free by warming the salt with lime, and dried over long columns of
recently molten caustic potash. It was entirely soluble in sulphuric
add. Five determinations (using a volumeter of 3.5 liters capacity)
with an extreme error of 3.9/10,000 gave a mean value 0.77079, or in
round numbers, i normal liter = 0.7708. Perman and Davies (Proc.
Roy. Sac.y 78A, 34 (1906)) have repeated this determination, both by the
volumeter method, using a balloon of 1.7783 Uters capacity, and by the
balloon, method, using a balloon of 0.50476 liter capacity. With the gas
from a commercial ammoniacal solution, repeating Leduc's procedure,
they have arrived at a value 0.7717, almost identical with Leduc's, thus
proving the necessity for removing the organic bases. They then puri-
fied the gas by three different methods ; (a) by the method of Guye and
Pintza, (6) by decomposing with caustic potash, ammonium oxalate which
liad been ten times recrystalUzed, (c) by reduction of sodium
nitrite, in a caustic soda solution, by aluminum. They obtained the
following results:

NO. of experi- Wcigfht of
Method. ments. i liter. Bxtreme error.

Volumeter, prepared by (a), (6), (c). . 7 0.77085 5.2/10,000

Balloon, prepared by (a) 4 0.77086 0.8/10,000

* These fractionations were for the purpose of eliminating the volatile phosphorus
compound, which, according to Richards and Wells (This Joiu-nal, 27, 459) is formed
when hydrochloric acid passes over phosphorus pentoxide, and which would increase
the density of the gas. Our work has also shown that the desiccation of the inside
walls of the balloon is accomplished only after repeated rinsings with the hydrochloric
acid gas. For these reasons I^educ's value must certainly be a little too high.

154 RBVIEW.

The extreme error in the first series is undoubtedly due to the fact of
the ammonia gas having been prepared by several methods. The authors
advise the value 0.77085.

It appears from these experiments that Leduc's value cannot be con-
sidered in deciding upon a final value, as it is unquestionably affected
by a constant error due to the presence of organic bases. The mean
of the other three values is i liter of ammonia = 0.770837, which can be
taken in round numbers to be

I liter of ammonia = 0.7708,

differing from the above by 1/19,000. It takes into account the fact
that any source of error (presence of air or of organic bases, traces of
moisture or condensation on the surface of the balloon)^ would tend
toward raising the final value rather than lowering it. The mean of the
determinations by Perman and Davies, using the volumeter method,'
which they consider most accurate, and gas prepared by method (a), is
0.77080 gram.

Sulphur Dioxide. — Leduc reports 2.2639, referred to air, as the mean
of several determinations (he does not state the number) having an ex-
treme variation of 1.3/10,000. The gas was prepared by the action of
mercury on pure sulphuric acid.

Jaquerod and Pintza, using the volumeter method (capacity of bal-
loon = 3.5 liters) and a gas prepared by repeated fractional distillations
of the liquefied sulphur dioxide of commerce, have obtained as the weight
of one normal liter of sulphur dioxide 2.9266 grams, which represents the
mean of seven determinations having an extreme diifference of 1.7/10,000.

This result agrees exactly with Leduc's,

I liter of sulphur dioxide = 1.29273 X 2.2639 = 2.9266.

Baume, using two balloons of 0.3 and 0.5 liter capacity, respectively,
has recently repeated this determination in the Geneva laboratory and
obtained the same result. The gas was purified by the same method
as that employed by Jaquerod and Pintza. The extreme variation was
3.8/10,000 for the large and 12.7/ 10,000 for the small balloon. The mean
of the two series is

I normal liter = 2.9266.

This, therefore, is the accepted value.

Conclusions.

The various results have been tabulated as follows: In column I are
the values obtained by Leduc ; in column II those obtained by Rayleigh ;
in column III those of various other investigators (Morley, Ramsay,
Gray, Perman and Davies) ; in column IV those obtained in the Geneva
laboratory; in column V the finally accepted value ; in column VI the same
value referred to the density of oxygen as one. This table has been com-

* Of the three methods, (a), (6), (c), the first would undoubtedly give the purest
gas. The recrystallization of ammonium oxalate does not assiU'e a complete elimination
of organic bases, which is likewise true of the reduction of sodium nitrite by alumi-
num, as the latter may contain small amounts of carbon capable of being transformed
into organic bases. In the course of my work with Pintza we proved conclusively
that the ammonia gas obtained by the decomposition of magnesium nitride contained
traces of organic bases. Its density was 1/1935 higher than that of the gas prepared
by method a,

REVIEW.

155

pleted by the addition of the better recent determinations of other gases
(/. chim. phys., 5, 203 (1907)):

Wbigrt op Normal LrrSR (in grams).

Gas.

H,
N.

CO,

N,0

]>duc
(1.4288)

0.08982
1.2503
I. 2501

(1-3429)

Rayleigh.
1.42905

(0.08998)
1 . 2507
1.2504

1.9763
1.9780

HCl (1.6407)
NH| (0.7719)

1.7809
1.9769
1.9777

others.

f 1.42900 M.

1 1.42896 Gr
0.089873 M
1.2507 Gr

1.3402 Gr
1.7808 R

Iraboratory.
Guye.

Accepted
value.

Referred

to density

o< Of.

(l.4292)J.P. 1.42900 1. 00000

1 . 3402 G.D.

1.6397 Gr

1.9768 G.P.
1.9774 G.P.
1.6398 G.G.

SO, 2.9266
Air 1 . 2927

0.77085 P.D. 0.7708 G.P.
i 2.9266 J.P.
\ 2.9266 B.

z . 2928

0.08987
1.2507
1.2504
1.3402

I . 9768

1.9777
1.6398

0.7708
2.9266
I . 2928

0.062890

0.87523

0.87502

0.93786

1.2463

I • 3833
1.3840
I . 1475
0.53940

2.0480
0.90469

* Dapier, /. chim. phys., 5, 203 and Arch. sci. phys. nat. [4], 24, 34.

The conclusions to be drawn from the work carried on at Geneva dur-
ing the last three years, compared with that accomplished in other labora-
tories, are as follows :

1. The method gives results generally agreeing to at least i/ 10,000,
when one takes the mean of a half dozen determinations, the extreme
difference between any two of which is not more than 3/10,000 or 5/10,000.

2. This agreement may be obtained even with gases as difficultly dried
and purified as ammonia, hydrochloric acid, sulphur dioxide or nitric
oxide. For these it is only necessary that the balloon be well dried
with dry air, and then rinsed several times in succession with the dry
gas which is to be studied. Under these conditions there is no appre-
dabk difference between the mean obtained for a balloon of 3.5 Uters
capacity and one of 0.5 liter. The surface action is th«i negligible in
work of an accuracy of the order of 1/10,000.

3. The purification of the gas is especially important. The method
based on the liquefaction of the gas, followed by distillation at low tem-
perature, is recommended.

4. It is preferable to calculate directly the weight of a liter of the gas
under standard conditions, rather than to report the density as compared
with that of another gas (oxygen or air). By calibrating the balloon
at o^, its volume may be determined to 1/30,000 — an accuracy which
could be approached, in the density determination, only by a very con-
siderable number of determinations.

5. The most probable values for the density of the gases, determined
with an accuracy of the order of 1/10,000, are collected in columns V
and VI of the last table.

I^ABpRATORT OP PHYSICAL CHEBCISTRY,
UMITBS.8ITY OF GENEVA,
SwrrZBSXAND.

156 NEW BOORS.

NEW BOOKS.

Annual Reports on the Progress op Chemistry for 1906. Issued by the Chemical
Society. Vol. III. London: Gumey and Jackson. 1907. 387 pp. Price,
$2 net.

The arrangement of the material in this, the third, volume of these
epitomes of the progress of a year along each of a number of lines of chem-
ical science is the same as that previously employed, and the reports arc
written by the same authors as in 1905, with the exception of that upon
General and Physical Chemistry, which is prepared by A. Findlay instead
of James Walker.

General and Physical Chemistry are treated apart from Inorganic
Chemistry; Organic Chemistry is subdivided into aliphatic, homocydic,
and heterocyclic divisions, and Stereochemistry is separately treated.
Anal)rtical, Physiological, and Mineralogical Chemistry, and Radio-
activity, each has a separate reviewer, while Agricultural Chemistry and
Vegetable Physiology are combined in one report.

As in the earlier volumes of this series, an effort has been made to render
the reviews readable and more attractive than a mere compilation of
data. The results are excellent, and seem to have been attained without
sacrifice of accuracy. It is almost inevitable that such a presentation
should involve the frequent expression of the personal convictions of the
authors, and it is not surprising that these should, at times, become a bit
obtrusive. As between this evil and a prosy cataloguing of articles, the
reader will quickly accept the style adopted in this volume as the less
objectionable, particularly since the authors may claim to speak with
authority in their respective fields.

These reports include only notices of papers which represent an ad-
vance in our knowledge of chemical science. The selections made by the
respective authors appear to be wisely chosen, and the statements, neces-
sarily very concise, are adequate to lead the reader to institute further
search among the original papers, when his interest is aroused. The
volume as a whole constitutes a valuable aid to the busy worker. It
would seem to the reviewer that, in spite of the references to industrial
chemical progress which are included in the Reports as now subdivided,
they would gain in value to the technical chemist, if a review devoted to
Industrial Chemistry were added. This field is, of course, exceedingly
broad, but with the same judicious selection of topics which is shown in
the present reviews, the important advances of the year might well be
brought within proper compass. The fact that this field is covered in
another journal in English does not seem to entirely excuse its omission
from these volumes. H. P. Tai^bot.

Organic Chemistry, Including Certain Portions of Physicai, Chemistry, for
Medical, Pharmacexttical and Biological Students (wtth Practical Ex-

NEW BOOKS. 157

BRCisBs). By H. D. Haskins, A.B., M.D., Instructor in Organic and Bio-Chem-
istry, Medical Department, Western Reserve University; Professor of Chemistry,
Qeveland School of Pharmacy, and J. J. R. MacLbod, M.D. (Aberdeen), D. P.
H. (Cambridge), Professor of Physiology, Western Reserve University. New
York: John Wiley and Sons. London: Chapman and Hall, Limited, 1907. Small
8vo, xii + 367 pp. Price, $2.00.

This text-book aims to give not only the facts of organic chemistry
(with instructions for laboratory exercises) but also the most important
facts of physical chemistry which have an essential bearing on medical
science. The introductory chapters on physical chemistry are well
written and on the whole correct (see, however, an error in the last para-
graph on p. 43 in regard to the osmotic pressure of a solution), and they
aroused the expectation of finding an equally well written short text on
organic chemistry. As in many recent similar texts, there is much that
is thoughtful and praiseworthy in the discussion of structures and the
development of the experimental evidence in favor of given structures
and m the ordinary treatment of organic chemistry; but one is disap-
pointed in finding in a book that aims to include in its work a few of the
fundamental concepts of physical chemistry so little of its spirit realized
in the treatment of organic reactions; for instance, the old superficial
parallel in the equations of ester and salt formation is emphasized (p. 93)
and the esters are still called "ethereal salts" '* comparable with the
salts of inorganic chemistry,'* (p. 125), although they have none of the
properties of salts. It is true that the diflferences in behavior and for-
mation of esters and salts are merUioned, but the authors do not seem to
have had the courage to break with the old, wrong conception and thus
there is left a confusing impression on the reader. Aside from a brief dis-
cussion of the reversibihty of esterification and saponification, no use what-
ever is made of the facts of reversibility and equilibrium in this book,
although they are of especial importance for physiologists and are essential
in a book that claims to use the modem concepts of physical chemistry.
In part, even the ordinary treatment of purely organic topics is decidedly
faulty; the emphasis laid on the proportion of hydrogen and oxygen in
the carbohydrates (p. 209) is misleading and gives a wrong conception of
the essential nature of carbohydrates. In no place, for instance, is any
mention found of the true sugar group — CH(OH) — CO — and from mis-
leading remarks about the reducing power of ketones (pp. 138 and 216)
it appears the authors are not clear in their own minds about the reducing
power of the " sugar group." On page 2 14 we are told that there are eleven
steteoisomeric acids having the structure H00C.(CH0H)4C00H! The
most glaring fault of the book is in the language used in giving the in-
structions for laboratory work (see pp. 85, 86, etc.) ; they are not written in
English but in the laboratory jargon which is the pitfall for most fresh

158 NEW BOOKS.

young doctors of philosophy in writing their dissertations and which
makes abominable reading! So while the book was undertaken in a
praiseworthy attempt to bring the subject of organic chemistry into
closer relationship to the modem conceptions of physical chemistry, it is
hardly a success in this respect and is only of average value and in part
badly written as an ordinary text-book of organic chemistry.

UNIVERSITY OF CHICAGO, J UlrlUS STIEGLITZ.

November 16, 1907.

A TbxT-Book op Organic Chbmistry. By A. F. Hollbman, Ph.D., F. R. A.,
Amsterdam, Professor Ordinarius in the University of Amsterdam. Translated
from the Third Dutch Edition by A. J. Walkbr, Ph.D., assisted by Owbn
E. MoTT, Ph.D., with the cooperation of the author. Second English Edition,
Rewritten. New York: John Wiley & Sons. London: Chapman and Hall,
Limited. 1907. 8vo. xv + 589 pp. Price, $2.50.

According to the author's preface to this second English edition of his
text-book on organic chemistry, the chief changes made are in the chapters
on the constitution of benzene and on pyrrole. In presenting the vexed
question of the constitution of benzene the plan is adopted of giving all
three of the most prominent fonnulae, Kekul6's, von Baeyer's and Thiele's,
with an explanation and a very brief criticism of each.

There is legitimate ground for a wide divergence of opinion in regard
to the best order of arrangement for the presentation of the facts of or-
ganic chemistry. It seems to the writer, however, that with the facts
of isomerism and the theories of structure, stereoisomerism and tautom-
erism.* the fundamental point which must be most clearly understood
by the student for a working acquaintance with organic chemistry is thf
difference in behavior between saturated and unsaturated compounds.
This difference is brought out most effectively from the experimental
and theoretical side, by a study of the properties of the unsaturated hy-
drocarbons. The point of view acquired there is most useful, in fact,
essential, in the study of the reactions of the aldehydes and ketones as
unsaturated compounds, in which absorption reactions play an extremely
important r61e, in which the smaller degree of stability of the addition
products involves no difficulty in presentation after a thorough discussion
of the olefines and acetylenes, and. in connection with which relative
instability, a further fundamentally important point of view for organic
as for inorganic chemistry may so easily be developed ; namely, the con-
ception of organic reactions as reversible ones, which should be treated

^ This question is treated on p. 305, and the discussion is restricted to the i : 3
diketones. It is a question affecting very many important classes of organic com-
pomids (add amides, nitroparaffins, mono-aldehydes and ketones and their hydra-
zones and oximes, phenols, etc.), and it seems to the writer that in a book of this class,
it ought to be taken up as a part of the question of isomerism at as early a point as
possible.

NBW BOOKS. 159

on the basis of the eqtdlibrium laws. A modem comprehension of the
behavior of organic acids and their derivatives, the esters, amides, nitriles,
etc., it seems to the writer again, is altogether impossible without a pre-
liminary knowledge of the behavior of aldehydes, ketones and olefines,
unsaturated bodies whose absorption products are so far more stable
than the absorption products of the derivatives of the organic acids,
which nevertheless must play an important rdle in the proper theoretical
treatment of their reactions. Yet we find in HoUeman's text the order
of treatment exactly reversed, the acids with their most complex be-
havior first, the olefines with their well-defined simple properties treated
only after such complex unsaturated bodies as the nitriles, isonitriles,
acids and their derivatives, aldehydes and ketones. There is room, it is
believed, for a text-book on organic chemistry in which the reactions are
treated on the basis of our equilibrium laws in a very simple and ele-
mentary but eflScient way with the aid, not of hypotheses, but of well-
known simple facts. Julius Stibglitz.

Tbb Univbrsitt op Chicago,
Nov. 16, 1907.

Poisons, Thbik Effects and Dbtkction. By A. Wyntbr Blyth and M. W. Blyth.
London: Charles Griffin & Co., 4th Ed. 1906. Svo. xxxii + 772 pp. Van
Nostrand Company. Price» $7.50 net. ^

The announcement by the publishers of a forthcoming new edition of
this standard work aroused great interest among anal3rsts and toxicolo-
gists, and they awaited the appearance of the work with the curiosity
natural to the interval of ten years between editions. The fact that
Blyth's poisons is the only comprehensive work of its kind in the English
language should make a new edition doubly valuable.

The fourth edition carries an additional name upon its title-page — ^that
of Meredith Wynter Blyth, Public Analyst for the Boroughs of Brighton
and Eastbourne. This would lead one to conclude that the presence of
poisonous substances in food products and the relations of such materials
to the public health would receive more attention than was accorded
them in the third edition. In this, however, we suffer a very great dis-
appomtment, for this phase of the field of the toxicologist and investigator
is practically ignored, there being essentially no change in the subject-
matter treated, the additions and alterations being mainly in the ar-
rangement and elaboration of contents of the old edition. However,
these changes in the manner of presentation and the addition of newer
and better methods of chemical analysis are sufficiently numerous and
extensive to justify the claim of the publishers that the fourth edition is
"thoroughly revised, enlarged, and rewritten."

The work is divided into nine "parts" as follows: I. Introductory, the
old Poison-Lore, the Growth and Development of the Modem Methods

l6o NEW BOOKS.

of Chemically Detecting Poisons; Bibliography. II. Definition of Poison;
Classification of Poisons, Statistics, Connection between Toxic Action and
Chemical Composition; Life Tests; General Methods in Searching for
Poison; the Spectroscope as an Aid; Examination of Blood Stains. III.
Poisonous Gases. IV. Acids and Alkalies. V. Substances Capable of
being Separated by Distillation. VI. Alkaloids and Poisonous Vegetable
Principles. VII. Poisons derived from Living or Dead Animal Sub-
stances. VIII. Oxalic Acid Group. IX. Inorganic poisons.

Appendix — ^Treatment of Cases of Poisoning. Domestic Ready Remedies
for Poisons.

Parts I and II are thus devoted to the discussion of what may be called
General Toxicology, the remaining parts to Special Toxicology.

Part I has been much improved, both by the addition of new material
and the suppression of doubtful facts; thus revised, the chapters are more
readable. Following Part I is to be fotmd a bibliography of the chief
works on toxicology, which unfortimately has neither been revised nor
brought down to date, only a single addition — ^Vibert's Precis de Toxi-
cologic— has been made to the works listed ten years ago.

Part II has had much new matter added to it. The arrangement of
the classification of poisons has been changed by a more logical com-
bination of doubtful substances imder a single group — ^Vegetable Principles
Not Readily Admitting of Classification — and in this group have been
placed Tutin, Illicium Religiosum, Picric Acid and Picrates, Ictrogen,
Lathyrus Sativus and Arum Poison, toxic substances not treated in the
third edition. A further addition is fotmd under poisons derived
from animal substances in a sub-group — Mammalian Poison; Epinephrine.
The statistics — Deaths from Poisons in England and Wales — are now for
the period 1 893-- 1903, the older statistics being suppressed.

The discussion of the relation between chemical composition, chemical
properties and toxic effect has been extended and is an excellent smnmar\'
of the facts now known, save for the omission of all mention of the con-
tributions of physical chemistry to our knowledge of how toxic sub-
stances act upon living cells. There are also many other portions of the
work where the introduction of physico-chemical methods and theories
is greatly to be desired.

The chapter on Blood and Blood Stains has been entirely rewritten
and Formanek's excellent charts of the absorption spectra of blood pig-
ments have been reproduced, so that the analyst is now given reliable
data and guidance to enable him to undertake an examination of sus-
pected material with a fair prospect of success.

As might be expected, the authors have made the greatest changes in
the specialized portions of their book. Improved methods of separation,
identification and determination are to be met with in the case of almost

RECENT PUBLICATIONS. l6l

every poisonous substance discussed. The chapters devoted to the
vegetable alkaloids are completely rewritten and greatly elaborated by the
introduction of the latest knowledge relating to the chemistry of these
substances. An exceedingly valuable feature is the introduction of
structural formulas and an indication of the relationships of allied com-
pounds. In the third edition the chapters devoted to the compounds of
carbon were, on the whole, the most unsatisfactory portions of the book,
but the authors have succeeded in the fourth edition in well rounding out
their work. The analytical methods are now well chosen, both on accotmt
of their convenience and reliability.

Considering the number of very rare and utterly unimportant poisons
of organic origin treated it is to be regretted that many common poisons
have been omitted; most prominent among these may be mentioned
formaldehyde, methyl alcohol, and acetanilide and other dangerous heart
depressants.

The general make-up and typography of the work is excellent, there
being remarkably few typographical errors; the only serious one noted
by the reviewer is the formula — NiC04 — ascribed to nickel carbonyl.

Providing the analyst does not regard the work as authoritative on
chemical properties and industrial processes, he will find it a safe guide
for the detection of toxic substances and a source of valuable information
relative to physiological effects. Blyth's poisons should be in the library
of every analytical chemist. E. M. Chamot.

RECENT PUBLICATIONS.

BiLTz, H. AND W. Ubbttngsbbisfisi^E aus dbr unorganischbn Kxpbri-
MBNTALCHBMiB. Leipzig: 1907. M. 7.

Brbubr, C. Kitte und Klbbstopps. Geschichtliche und technische Aus-
fiihnmgen. Hannover: 1907. 268 ss. M. 3,40.

BROWNI.BB, R. B., PuuuBR, R. W. AND Othbrs. Fkst PRiNapuss OF Chbm-
ISTRY. Boston: AHyn & Bacon. 1907. 419 p. $1.25.

CoHBN» Juuus B. Organic Chbmistry for Advancbd Studbnts. London,
Eng : Edward Arnold. 1907. 632 p. 21s.

Duncan, Rob. Kbnnbdy. Chbmistry op Commbrcb. New York: Harper &
Bros. 1907. $2.00.

JONBs, Harry Clary, and Othbrs. Conductivity and Viscosity in Mixbd
SoLVBNTs. Washington, D. C: Carnegie Institution of Washington. 1907. 235 p.
$2.00.

JoNBs, Harry Ci^ary. Thb Klbmbnts of Physicai* Chbmistry. 3rd ed., revised
and enlarged. New York: The Macmillan Co. 1907. 650 p. 8 vo. $4.00.

HanausBk, T. F. Thb Microscopy of Tbchnical Products. Revised by the
anthor and translated by Andrew L. Winton and Kate O. Barber. New York: John
Wiley & Sons. 1907. 8 vo. 471 p. $5.00.

Hazbn, AuuBn. Clban Watbr and How to Gbt It. New York: John Wiley
&S00S. 1907. 12 mo. 178 p. $1.50.

Knauthb, K.: Das Susswassbr. Chemische, biologische and bakteriologische

l62 RBCHNT PUBLICATIONS.

Untersuchungsmethoden, unter besonderer Berilcksichtigung der Biologie und der
fischereiwirtschaftlichen Praxis. Newdamm: 1907. M. 18.

LiBBiG, J. V. UND GOssEPELD, E. L. F.i Briefwechsel, 1862-66. 22 Bride
Liebigs zugleich Beitrag zur Geschichte der Indtistrie k^stlicher Dimger in Deutsch-
land. Mit Anmerkungen und Erld.utermigen herausgegeben von O. E. Gtisselfeld.
Leipzig: 1907. 72 ss. M. 3.

LOfpo-Cramer. Photographische Probleme. (Photochemie VerSlnderung der
Silberhalogenide ; ilber den sog. Chemischen Schleier ; {iber die photohaloide Leas ; etc)
Halle: 1907. 220 ss. M. 7,50.

tiERNST, W. Experimental and Theoreticai^ Applications op I'her-
MODYNAMics TO Chemistry. London: 1907. 134 p. $1.30.

NoYES, A. A. The Electrical Conductivity op Aqueous Solutions: a
report presented by Arthur A. Noyes upon a series of experimental investigations
executed by A. A. Noyes, W. D. Coolidge, A. C. Mdcher, H. C. Cooper, Yogoro
Kato, R. B. Sosman, G. W. Eastman, C. W. Kanolt, and W. Bottger. Washington,
D. C: Carnegie Institution of Washington: 1907. 352 p. $2.50.

Ppister; J. Das FArben des Holzes durch Impragnierung. Wien: 1907.
M. 2.

Post, G. Chebhsch-technische Analyse. Handbuch der analytischen Unter-
suchtmgen zur Beaufsichtigtmg chemischer Betriebe, ftir Handel und Unterricht.
3., vermehrte Auflage, herausgegeben von B. Neumann. (2 Bande in 8 Heften)
Band I. Heft 3, und Band II. Heft 2. Braimschweig: 1907 M. 17.

Band I. Heft 3: Eisen, Metalle (ausser Eisen) und Metalloide, von A. Ledebur
und B. Neumann, ss. 209-685. M. 7. Baud II. Heft 2: Rubenzucker, Starke,
Bier, Wein, Spiritus, Essig, Etc., von R. Fr&hlung, E. Parow, H. Hanow, u. a. ss.
209-658. M. 10.

Reich, A. Reinigung und Beseitigung stAdtischer und gbwerblichbr Ab-
wAssBR. Hannover: 1907. 139 ss. M. 2, 20.

Sammlung chemischer und chbmisch-technischer VoRmAGE. Heraus-
gegeben von F. B. Ahrens. Band 12. Heft 1-3: Kauffmann, H., Die Auxochrome.
Stuttgart: 1907. ss. 1-112. M. 3, 60.

Semmlbr, F. W. Die Atherischen Oele nach ihren chemischen Bb-
STANDTEiLEN unter Beriicksichtigung der geschichtlichen Entwicklung. Leipzig:
1907. Lieferung 15 u. 16: ss. 257-490 (V. Band IV: Benzol derivate und hetero-
cyclische Verbindungen). Subscriptionspreis. M. 9, 50.

Das jetzt vollstandige Werk, 4 Bande, 1905-1907. 876, 620, 836, und 498 ss.
M. I, 32.

Taschenbuch pCr die anorganisch-chemische Grossindustrib. Heraus-
gegeben von Dr. G. Lunge und Dr. E. Berl. Vierte umgebeitete Auflage des taschen-
buches ftir die Soda-, Pottasche-, und Ammoniak-fabrikation. Berlin: 1907. 288
p. M. 7.

Vanino, L. Das Natriumsuperoxyd. Wien: 1907. M. 2.

Wauklyn, J. A. Water ■ Analysis. Practical treatise on the examination
of Potable Water, iithed. London: 1907. 266 p. $1.30.

Weston, Frank E. A Scheme por the Detection op the More Common
Classes op Carbon Compounds. London, Eng.: Longmans, Green & Co. 1907.
8vo. 95 p. 2S. 6d.

WoKER, G. Probleme der Katalytischen Forschung. Leipzig: 1907.
M. I, 20.

Vol. XXX. PsBRUARY, 1908. No. 2.

THE JOURNAL

OF THE

American Chemical Society

AMERICAN CHEMICAL SOCIETIES.'

By Marston Taylor Booert.

There are many ways in which a man may endeavor to serve his age
and generation by devotion to the science of chemistry, and surely, not
the least important of these is the labor which has for its object the stimula-
tion and development of our great national chemical society, an organiza-
tion which has been one of the most potent factors in the growth and
progress of the science.

Among the prime requisites for the success of any organization are:
(i) commtmity of interest, and (2) an enthusiastic and intelligent co-
operation on the part of the membership. That we all have the same
interests at heart — the advancement of chemistry, and the development
of our Society for this purpose — ^is true beyond any shadow of doubt. The
needed cooperation on the part of the membership has been manifest in a
high degree. Those who were fortunate enough to be present at our
recent meeting in Toronto will, I know, bear testimony to the fine enthu-
I aasm and esprit de corps evident. That is the spirit that should permeate
the entire organization and penetrate to all comers of our coimtry wherever
our members are located, for it is certain to bring a rich harvest of growth
and power to the Society.

In the rapid progress of our Society during the past few years, it has
not been an easy matter to keep the members properly informed as to its
exact condition and the lines of contemplated development, and yet
without such information intelligent cooperation is difficult.

For any adequate comprehension of the requirements of an organization
of the size and character of the American Chemical Society, not only is a
searching scrutiny of its own history for the last thirty-three years neces-

^ Preadential address delivered at the Chicago Meeting of the American Chemical
Society, January i» 1908.

164 MARSTON TAYI.OR BOGERt.

sary, but also a careful consideration of past and present conditions of
chemical organization in this country. On such knowledge as the founda-
tion, it should be possible to base a fairly trustworthy judgment as to the
most promising lines of development.

I have, therefore, made as thorough an examination as my time would
permit of the histor>'^ of chemical organizations in this coimtry, partly
that we might benefit by the experience of others, and partly to gather
facts concerning other existing chemical organizations in the United
States, in the earnest hope that a way may later open for the union of all
American chemical organizations in one great society or federation.

It seemed to me that the subject was one of suflSdent interest and im-
portance to constitute an acceptable presidential address, for the next
few years will be critical ones for us, and the more familiar our members
are with such matters the more intelligently will they be able to assist our
Coimcil. Such a topic enjoys the obvious merit of appealing to every
member present, an advantage which, I fear, would scarcely pertain to a
text selected from my own chosen field of synthetic organic chemistry.

In the early history of our country, when it was striving to win for
itself a place among the nations, there was little opportunity for the
pursuit of science for science's sake. The facilities for acquiring chemical
knowledge were necessarily very meagre, what little training there was,
being given mainly in medical schools. Chairs of Natural History and
Physical Science were generally held by physicians, for it was felt that
their education more nearly fitted them for such work. Thus chemistry
in the early days was largely a side issue, being taught often with materia
medico. It is particularly interesting to me as a Columbia man to find
that chemistry was first recognized as a branch of the curriculum of
medical study at King's College (now Columbia) in 1767. In 1774, the
Right Rev. James Madison was Professor of Chemistry and Natural
Philosophy at Williams and Mary. The first separate Chair of Chemistry
at a non- medical institution was established October i, 1795, at the
College of New Jersey (now Princeton). It was filled by Dr. John Mac-
Lean, from whom the elder Silliman received his early chemical instruc-
tion.

Although the latter half of the i8th Century was the period of Franklin,
Rumford and Priestley, when we recall that this was also the period of
our war for independence, it is not surprising that progress in science was
relatively inconsiderable.

It was natural in these early years of science in America that what
scientific organizations there were should be general in their character,
as scarcely any individual science, with the possible exception of medicine,
was strong enough to stand alone. It is appropriate, therefore, that a

AMERICAN CimmCAt SOCIETmS. 165

few words be said about these general scientific societies before passing
on to those strictly chemical.

The first of these general societies was the American Philosophical
Society, founded by Benjamin Franklin, at Philadelphia, in the year
1743, and recognized by Provincial Charter in 1769. Franklin was its
president from the date of its charter until his death in 1790. The first
volume of its Transactions appeared in 1769 and contained a paper by Dr.
DeNormandie on "An Analysis of the Chalybeate Waters of Bristol in
Pennsylvania," which appears to be the first chemical analysis ever
published in this coimtry; and what are probably the first papers pub-
lished by Priestley on this side of the water, entitled *' Experiments and
Observations Relating to Analysis of Atmospheric Air" and "Generation
of Air from Water." This venerable association is still active and vigorous,
its aim being to cover all branches of so-called natural science.

On May 4, 1780, the American Academy of Arts and Sciences was
incorporated at Boston.

In 1799, the Connecticut Academy of Arts and Sciences was established
at New Haven and incorporated by legislative enactment.

The Literary and Philosophical Society of New York, founded in the
early part of the 19th Century, had but a brief existence, and published
but a single volume of Memoirs (containing one paper on chemistry).

On April 20, 1816, the New York Lyceum of Natural History was in-
corporated. In 1876, it changed its name to the New York Academy of
Sciences.

At about the same time, the Philadelphia Academy of Natural Sciences
was established, the first volume of its journal appearing in 1817.

These early organizations were rapidly augmented by the establish-
ment of various scientific associations and academies in other parts of the
country, and most of our great cities now have one or more of this class.

The first of these general scientific organizations to wield more than a
merely local influence and power was the American Association for the
Advancement of Science. The original progenitor of this association
was the American Association of Geologists, foimded in 1840 by those
engaged in the geological surveys of various states. It later became the
Association of American Geologists and Naturalists, which body, at its
meeting in Boston, 1847, resolved to enlarge its sphere of action to in-
clude physics, chemistry, astronomy, and the allied physical sciences,
and thus began its enlarged existence in 1848 as the American Association
for the Advancement of Science.

In most of these general organizations, it was customary to subdivide
science more or less closely into sections. At the outset, chemistry was
gnmped frequently merely under the general section of natural sciences,
then with physics or materia medica, and finally it was found necessary

l66 MARSTON TAYLOR BOGBRT.

to establish separate sections for chemistry. In the American Association
for the Advancement of Science a chemical sub-section was established
at the Hartford meeting, August 17, 1874, ^^h S. W. Johnson as Chair-
man, and F. W. Clarke as Secretary. This became a full-fledged section
(Section C) at the Montreal meeting of the Association, 1882, with H. C
Bolton as presiding officer and Alfred Springer as secretary.

CHEMICAL SOCIBTY OI^ PHILADELPHIA.

The first organization devoted specifically to chemistry was the Chemical
Society of Philadelphia, founded in 1792. This was probably the earliest
chemical organization in the world, as it was bom forty-nine years before
the first European chemical society (Chemical Society of London, 1841).
Priestley appears to have been one of its active members. In i8oi-*02
its president was James Woodhouse, who was at the time Professor of
Chemistry in the Medical Department of his alma mater, the University
of Pennsylvania. This Chair had been held by Dr. James Hutchinson,
and on his death, in 1793, ^'^s ofiFered to Dr. Joseph Priestley, who had
but just arrived from England. Priestley, however, preferred the quiet
and retirement of Northumberland, and Dr. Woodhouse was then elected
to the position. The most important communication presented to this
society was that announcing the discovery of the oxyhydrogen blow-
pipe. It was made on December 10, 1801, by Robert Hare, Jr., then
but twenty years old, who subsequently became Professor of Chemistry
in the Medical School of the University of Pennsylvania. Just when this
society ceased to exist is uncertain, but it appears to have been about
1803.

COLUMBIAN CHEMICAL SOCIETY.

It was succeeded by the Columbian Chemical Society of Philadelphia,
fotmded August, 181 1, by **a number of persons desirous of cultivating
chemical science and promoting the state of philosophical inquiry." Its
first president was James Cutbush and its membership list included
sixty-nine honorary and thirteen junior members. Vol. I of its Memoirs
appeared in 18 13. As *' patron'* of this society appears the name of Hon.
Thomas Jefferson, Esq., who had been president of the American Philo-
sophical Society for many years, relinquishing it finally to become the
President of the nation. Seventeen months before the founding of this
society, he had retired from the presidency of the United States and was
living at his country seat in Monticello. James Cutbush was at the time
Professor of Natural Philosophy, Chemistry and Mineralogy at St. John's
College. From June 1820 until his death, December 15, 1823, he taught
at the U. S. Military Academy, West Point.

The constitution of this society provided iii addition to other officers
an "orator," who should deliver "an oration on some chemical subject

AMERICAN CHEMICAI. SOCIETIES. 1 67

within two months after the commencement of the medical lectures in the
University of Pennsylvania, in each year.'* As the Memoirs of the
society contain no "oration," it has been suggested that the incumbent's
efforts were not satisfactory. The roll was called at the opening and
dosing of every meeting and all absentees were fined twelve and a half
cents each. Any member elected to office and declining to serve was
fined one dollar. Once in each month the society appointed some mem-
ber to read an original chemical essay, "for neglect of which, the member
so appointed shall be fined one dollar." Candidates for admission were
required to "read an original essay on some chemical subject."

Among its members were, Dr. Benjamin Smith Barton, of the Uni-
versity of Peimsylvania, sometimes called "the father of American natural
history;" Dr. Archibald Bruce, of Columbia College, founder of the
American Mineralogical Journal', Dr. John Griscom, "the acknowledged
head of all teachers of chemistry in New York City" for more than thirty
years; Robert Hare, of the University of Pennsylvania; Dr. David Hosack,
of Columbia College, foimder of the first public botanical garden in the
United States (1801); James Madison, President of the United States;
Dr. John Maclean, first Professor of Chemistry at the College of New
Jersey; Hon. Samuel L. Mitchell, Professor of Chemistry and Natural
History at Columbia College and a United States Senator; Dr. Benjarnin
Rush, of the University of Pennsylvania, who was regarded by Benjamin
SilHman as " imdoubtedly the first Professor of Chemistry in America,"
his appointment dating August i, 1769; Benjamin Silliman, Professor of
Chemistry at Yale; and many noted foreign chemists. This was the
first chemical society of a truly national character. Unfortunately, it
survived but a few years.

DEtAWARH CHEMICAL AND GEOLOGICAL SOCIETY.

The Delaware Chemical and Geological Society was organized at Delhi*
Delaware County, N. Y., September 6, 1821. It is stated that it was
composed of "between forty and fifty well-informed and respectable
inhabitants" of Delaware County. Its first quarterly meeting was held
at Edgerton's Hotel in Delhi Village, with Charles A. Foote as president.
It, too, was short-lived.

So far as the writer is aware, there were no important chemical societies
in existence, certainly none of national influence, between this time and
the founding of the Manufacturing Chemists' Association of the United
States (1872) and of the American Chemical Society (1874). Original
commimications in chemistry were presented before the various philosoph-
ical societies, academies and institutes, and published in their Transac-
tions, in the Journal of the Franklin Institute and, chiefly, in Sillman's
Journal, With the appearance of the American Association for the Ad-

1 68 MARSTON TAYI.OR BOGERT.

vancenient of Science upon the scene (1848), the chemists rallied more
and more strongly to its support imtil it had enrolled on its membership
list a large number of the leading chemists of the country and its chemical
section (Section C) was unquestionably the most powerful organization
of chemists then in America.

MANUFACTURING CHEMISTS* ASSOCIATION OF THE UNITED STATES.

On May 29, 1872, the Manufacturing Chemists' Association of the United
States was organized at the Astor House, New York City, Mr. Thomas
S. Harrison, of Philadelphia, presiding. Manufacturers of chemicals
whose annual product is at least $50,000 in value are eligible for member-
ship. Its objects are to protect its members against unwise legislation
and unjust freight discrimination, and to promote and aid any matter of
general or special interest in the chemical industries. Its membership
at present comprises forty-one representative corporations. Annual
meetings are held in cities selected by vote. This brings me up to the
foimding of our own

AMERICAN CHEMICAL SOCIETY.

In the American Chemist for April, 1874, there was published a letter
from Dr. H. C. Bolton, of Columbia College, entitled: "Centennial of
Chemistry, 1774-1874," in which he referred to the many notable chemical
discoveries of the year 1774, as the fruit of the labors of such men as
Scheele, Lavoisier, Priestley, Cadet, Bergmann, and others. As the dis-
covery of oxygen by Dr. Joseph Priestley on August i, 1774, resulted in
the overthrow of the Phlogistonists and the establishment of chemistry
on its present basis, the writer points out that the year 1774 may well be
regarded as the birth year of modem chemistry, and suggests that it
"would be an agreeable event if American chemists should meet on the
first day of August 1874, at some pleasant watering-place, to discuss
chemical questions, especially the wonderfully rapid progress of chemical
science in the past one hundred years." This suggestion met with the
hearty approval of the editors of the Am£ric,an Chemist and they re-
quested all chemists interested in the matter to send in their views at
once. Among other letters, one was received from Miss Rachel L. Bodley,
Professor of Chemistry at the Woman's Medical College, of Pennsylvania,
which contained the following: "I made a pilgrimage last August to the
grave of Priestley in Northumberland, Pennsylvania, and was deeply
impressed by the locality, its associations, and its charming surroundings.
My proposition is, therefore, that the centennial gathering be arotmd
this grave, and that the meetings, other than the open air one on the
cemetery hilltop, be in the quaint little church built by Priestley, where
might be exhibited the apparatus devised by the great scientist and used
in his memorable experiments."

AMKRICAN CHEMICAL SOCIETIES. 1 69

At a meeting of the Chemical Section of the New York .Lyceum of
Natural History, May 11, 1874, on motion of Dr. H. C. Bolton, the fol-
lowing resolutions were adopted: "Whereas, the discovery of oxygen
by Priestley on August i, 1774, was a momentous and significant event in
the history of chemistry, being the immediate forerunner of Lavoisier's gen-
enJizations on which are based the principles of modem chemical science ;
and, whereas, a public recognition of the one hundredth anniversary of
this brilliant discovery is both proper and eminently desirable; and
whereas, a social reunion of American chemists for the mutual exchange
of ideas and observations would promote good fellowship in the brother-
hood of chemists; therefore, resolved, that a committee of five be ap-
pointed by the Chair, whose duty it shall be to correspond with the
chemists of the coimtry with a view to securing the observance of a
centennial anniversary of chemistry during the year 1874." The com-
mittee appointed consisted of Messrs. Bolton, Chandler, Wurtz, Leeds
and Seeley. The suggestion of Miss Bodley that the meeting be held at
Northumberland met with general approval, and the above committee
having made all necessary preparations, issued the call, with the result
that the chemists of the country assembled at Northumberland on July
31, 1874, where they were received and most hospitably entertained by
the direct descendants of Priestley. Altogether about eighty chemists
attended the celebration. The first session was held in the Public School
building and Professor C. F. Chandler was chosen president. Various
historical addresses were presented and congratulatory cablegrams ex-
changed with the chemists of Birmingham who were to imveil a statue
to Priestley the following day. Some of Priestley's original letters were
read, and a memorial address by Henry Coppee, LL.D., president of
Lehigh University, was delivered at the grave.

At the afternoon session on July 31st, Professor Persifor Frazer "pro-
posed the formation of a chemical society which should date its origin
from this centennial celebration," since America had "not a single society
to represent the chemical thought of the country.*' This was opposed
by Professor J. Lawrence Smith, chiefly on the ground that there were
"already two great organizations, the American Scientific Association
{sic) and the American Academy of Sciences, which undertook to em-
brace in their proceedings everything connected with chemical research."
Others expressing themselves as of similar opinion, it was finally re-
solved "that a committee of five be appointed from this meeting to co-
operate with the American Association for the Advancement of Science
at their next meeting, to the end of establishing a chemical section on a
firmer basis." The committee appointed consisted of Messrs. Bolton,.
Silliman, Smith, Horsford and Himt.

The attitude of Dr. J. Lawrence Smith towards the question will be

^o MAKSTON Taylor bogeht.

etter understood when it is recalled that he was president of the Ameri-
an Assoctation for the Advancement of Science two years before (1S72).
.t the Portland meeting of this Assodation the following year (1873), a
iparate heading in the phy^cs section was given for chemistry and more
apers (six) were presented in that science than there had been in both
hysics and chemistry (five) the year before. At the close of the meet-
ig, an informal gathering of chemists was held (August 26), with Professor
. A. Lattiniore as chairman, at which resolutions were adopted asking
>r the organization of a. chemical sub-section of the Association. These
^solutions were favorably received, and the chemists of the Association,
^operating with the coniniittee appointed at the Northumberland cele-
ration, organized a chemical subsection at the Hartford meeting, August
7, 1874. As already stated, in the year 1882 this became a section, with
>r. Bolton as its first chairman.

The formation of a chemical sub-section by the American Association
)r the Advancement of Science did not, however, fully satisfy many of
lose who were present at Northumberland. These chemists felt that
'hat was needed was an independent American chemical society, which
lould unite in one active, aggressive organization the chemists of the
suntry, and that only through such an organization could the progress of
hemistry be properly stimulated and hastened. This need was felt most
eenly in New York City, and led finally to a meeting on January 22,
876, at the home of Dr. C. F. Chandler, at which a committee was ap-
ointed, consisting of Messrs. Chandler, Habirshaw, Endemann, Alsberg,
[orton, Walz, Hoffmann and Casamajor, to attend to the preliminaries
f organization. A circular was prepared and mailed to about one hun-
red chemists residing in the vicinity of New York City, suggesting the
irmaUon of a chemical society in New York. The replies were so
umerous (forty) and encouraging that it was decided to attempt the
>rmation of a national instead of a purely local society, and a circular
:tter was sent out to chemists in all parts of the country. Sixty chemists
utside of New York City signified their desire to join. The first meeting
)r organization was held April 6, 1876, at the New York College of
harmacy, with Dr. Chandler as Chairman, Isidor Walz as Secretary, and
lirty-five chemists in attendance. Drs, Bolton and Egleston both
lought the time inopportune for such a movement, as both the New
'ork Academy of Sciences and the American Association for the Ad-
ancement of Science had chemical sections. Nevertheless organization
'as proceeded with, a constitution and by-laws adopted, and at an ad-
)umed meeting, April 20th, Dr. John W. Draper was elected president.
he first regular meeting after organization was held May 4, 1876, with
ice- President Chandler in the Chair, thirty members and fifteen victors
eing present. The first paper read was "On the Determination of the

AMERICAN CHBMICAI^ SOClETmS. 1 7 1

Rektive Effectiveness of Disinfectants/' by Dr. H. Endemann. Arrange-
ments were made with the editors of the American Chemist to publish the
Proceedings of the Society and to supply the members with reprints.

On June i6, 1876, the Society gave a dinner at the Union League Club
of Philadelphia, to the foreign chemists officially connected with the
Centennial Exposition, which was attended by about seventy .chemists.

On November 16, 1876, in Chickering Hall, New York City, Dr. Draper
delivered his presidential address, on ** Science in America," before a large
and distinguished audience. At the close of the year, the membership
of the Society was about 230. In November, 1876, the Society was in-
corporated in New York State. The first report of the librarian (Casa-
major), published at the close of 1878, showed that the library then con-
tained 344 volumes. On March 6, 1879, the publication of the Journal
of ike American Chemical Society began.

By the close of the year 1880, the Society was $900 in debt, with nominal
assets in the shape of uncollected dues amounting to about $1,000. The
publication of the Journal was therefore temporarily suspended and the
debt liquidated by personal subscription. During this year there was
frequently no quorum (fifteen) at the meetings. The Journal for 1881
(VoL III) covered only 189 pages and contained twenty-seven original
articles, eighteen of which were contributed by three authors, Leeds,
Stebbins and Casamajor. In 1882, the lack of original material for the
Journal appears to have been felt even more than the lack of funds. In
1884, it was found necessary to reduce the quorum from fifteen to ten.
The Journal for 1888 (Vol. X) contained but ten original papers, and the
following year the Society appeared almost moribund. There were but
few papers submitted for the Journal and not enough money to publish
even these. Personal subscriptions were again necessary to meet ex-
penses. It was apparent to all that a radical change in policy was neces-
sary or the organization would inevitably succumb. The fact that the
Society was chartered in New York and all its meetings held in New York
City, rendered the Society essentially a local one, in spite of the fact that
the president and other officers were frequently selected from non-resi-
dents. The direct outcome of this condition of affairs was a steadily
increasing discontent on the part of chemists in other parts of the country,
resulting in a withdrawal of their interest and support, and the New York
members, upon whom the burden was falling with ever-increasing severity,
were rapidly becoming discouraged by this lack of support. As a natural
corollary, there had begun a movement for the establishment of another
chemical organization which should be more truly national.

At the meeting of Section C in Cleveland, in 1888, a committee was
appointed on the formation of a national chemical organization. At the
Toronto meeting, the next year (1889), this committee, after conferences

'• > '• , .>

t. : ''. 1 . ' I

172

MARSTON TAYLOR BOGERT.

with committees appointed by the American Chemical Society, Associa-
tion of Official Agricultural Chemists, Washington Chemical Society, and
the Chemical Section of the Franklin Institute, recommended the es-
tablishment of a national organization. As the result of this report, the
constitution of the American Chemical Society was revised so that an
advisory council, local sections, and migratory meetings were authorized.
The headquarters, however, remained in New York City. The first general
meeting outside of New York was held at Newport, R. I., August 6-7, 1890,
and was a great success. It was followed by the organization of the
Rhode Island Section, the first of the local sections, which was duly
chartered the following year. At the second general meeting, held in
Philadelphia, December 30-31, 1890, a conference occurred of committees
representing various chemical organizations to bring about consolidation,
the basis of which was to be the union of all as the ''American Chemical
Society," the present New York organization to become a local section.
At the third general meeting, at Washington, August 1 7-18, 1891 , as the up-
shot of a similar conference of committees, it was resolved to imite on the
above basis. The organizations represented at this conference were:
The American Chemical Society (290 members). Section C of A. A. A. S.
(200 members), Assoc. Official Agr. Chemists (75 members). Chemical
Section of the Brooklyn Institute (75 members). Chemical Society of
Washington (70 members). Chemical Section of the Franklin Institute
(70 members). Chemical Society of the University of Michigan (60 mem-
bers), Louisiana Sugar Chemists' Association (52 members), Cincinnati
Chemical Society (29 members), and the Manufacturing Chemists' Asso-
ciation of the United States. A complete reorganization of the American
Chemical Society followed, and a further revision of its constitution, and
on April 29, 1892, the New York organization became the New York
Section of the American Chemical Society. Just one month before this
date, the "Chemical Society of Cincinnati and vicinity," organized Decem-
ber 19, 1890, was chartered as the Cincinnati Section. The Chemical
Society of Washington, an organization of chemists working for the
United States Government, and founded in 1884, became a local section
the following year, retaining at the same time its original title. In 1893,
Dr. Hart's Journal of Analytical and Applied Chemistry was consolidsited
with the Journal of the American Chemical Society and Dr. Hart appointed
editor. The condition of our Journal at the time was not very encourag-
ing, for when he took charge there were but two papers ready for pub-
lication, with five numbers of the Journal in arrears.

The seventh General Meeting occurred in this city (Chicago), August
21-26, 1893, in connection with the World's Chemical Congress of the
Columbian Exposition. At the Buffalo meeting, August 21-22, 1896,
the present plan of holding joint meetings with Section C of the American

AMERICAN CHEMICAL SOCIETIES. 1 73

Assoc, for the Advancement of Science was agreed upon. By its terms, the
first two days are devoted to the American Chemical Society, with the
exception of sufficient time on the first morning for Section C to organize
and in the afternoon for the address of their vice-president, the rest of the
meeting being then given up to Section C.

At the close of the year, 1897, the Society arranged with Dr. A. A.
Noyes to take over the publication of the Review of American Chemical
Research, and Vol. Ill of this Review was published with the 1897 Journal.
The publication of abstracts and patents had been undertaken by the
Society years before, beginning with Vol. I, No. 4, but it finally ceased
with the January number of Vol. XVI (1893).

In 1902 associate membership was abolished. This attempted classifica-
tion of our members was partly responsible for the separation from us of
the industrial chemists and electrochemists.

The first recognition of the growing strength and importance of the
various branches of chemistry was made in 1904 when the general meet-
ings were for the first time held in sections.

This year, as you know, we have taken a great stride in advance by the
successful publication of Chemical Abstracts, superseding the Review of
American Chemical Research. In spite of the necessary increase in dues
from $5.00 to $8.00, we are gaining new members more rapidly than ever.
Within the past year three new local sections have been chartered, Syra-
cuse, St. Louis and Wisconsin, and others are in process of formation.

It was hoped at the time the American Chemical Society was reorganized
that all existing chemical societies would come into the fold and that the
American Chemical Society would be the one organization to include all
the chemists of the country. Unfortunately, this has not yet been realized.
Not only did some of the chemical societies then existing fail to come in,
but other separate and independent organizations have since arisen. A
brief consideration of these is necessary to any proper understanding of
existing conditions. I cannot, of course, take up those of purely local
character, although some of them (for example, those at Detroit, Cleve-
land and elsewhere) are strong in numbers and influence. We hope that they
will all ultimately form new local sections of our Society or unite with
existing local sections. Nor is there time to take up the consideration of
chemical sections of academies of science or of scientific institutes, al-
though some of them exert a powerful influence locally (as the chemical
section of the Franklin Institute, for example).

It is more important to consider those chemical societies which possess
a more national character, pointing out the reasons for their establishment,
the particular field covered, and their present strength in numbers and
influence. The most important of the chemical organizations established

174 MARSTON TAYLOR BOGBRT.

once the founding of the American Cbemical Society, taking them up in
chronological order, are as follows:

iS8o, Assoc, of Agricultural Chemists (later the Assoc, of Official Agri-
cultural Chemists).

1894, New York Section, Society of Chemical Industry.

1898, New Sngland Association of Chemistry Teachers.

1900, New York Section, Verein Deutscher Chemiker.

1902, American Slectrochemical Sodety,

1904, Western Assoc, of Technical Chemists and Metallurgists.

1906, Society of Biological Chemists.

ASSOCIATION OP OFFICIAL AGRICULTURAL CHEMISTS.

The condition of agricultural chemical work in the United States in
[880 has been described by Dr. Wiley as "chaotic." There was no unity
>f purpose, action or methods, no standard of comparison or reference.
The great differences in analytical methods led to constant wrangling and
itigation between buyer and seller.

It was to put an end to this state of affairs that the Hon. J. T. Hender-
lon. Commissioner of Agriculture for Georgia, at the suggestion of Mr.
H. J. Redding, issued a call for a convention of agricultural commissioners
ind chemists for the purpose of securing uniformity in analytical methods.
This convention was held in Washington, July 28, 1880, with Judge
Henderson as Chairman and A. R. Ledoux as Secretary, the chief topic
ronsidered being fertilizer analysis. It was further decided to form a
livision of the sub-section of chemistry in the American Association for
he Advancement of Science and to hold the next meeting with this
Association. This joint meeting occurred in Boston, August 27, 1880,
ind a committee of three was there appointed to secure the formation of a
)ermanent chemical section in the American Association with agricultural
chemistry as a sub-section. The third meeting of the Association, held in
he room assigned to the chemical sub-section of the American Associa-
ion, Cincinnati, August 18, 1881, transacted its business separately, but
ead its papers before the American Association.

After the Cincinnati meeting, the interest of agricultural chemists in
x>llaboration seemed to die out, due apparently to the difficulty of har-
noniztng the conflicting interests of the trades chemists and official
chemists. After the lapse of three years. Judge Henderson again called
I meeting, which was held at Atlanta, Ga., May 15, 1884, and once more
iiscussed the unification of analytical methods. At the next meeting,
leld in Philadelphia, September 8-9, 1884, in conjunction with the Ameri-
an Association, it was decided to form a separate and independent
trganization instead of a sub-section of the American Association, and on
September 9th, the Association of Official Agricultural Chemists came
nto existence.

AMBRICAN CHEMICAI, SOCIETIES. 175

At the eleventh annual meetmg, Washington, August 23, 1894, the
Society of Leather Chemists was absorbed and a reporter on tannin ap-
pointed.

According to the constitution of this Association, its objects are (i)
to secure uniformity and accuracy in the methods, results, and modes of
statements of anal3rsis of fertilizers, soils, cattle foods, dairy products,
and other materials connected with agricultural industry; and (2) to
afford opporttmity for discussion of matters of interest to agricultural
chemists. Its voting members must be analytical chemists cotmected
with the U. S. Dept. Agriculture, or with any state or national agricul-
tural experiment station or agricultural college, or with any state or
national institution or body charged with official control of materials
named in (i) above. Other anal3rtical chemists are permitted to attend
meetings and participate in the discussions but not to vote or offer motions.
Referees are appointed to prepare and distribute samples and standard
reagents and to tabulate and present results before the Association.

The adoption and publication of these *' official methods" has brought
order out of chaos. The results of the work of the Association are pub-
lished by the Division of Chemistry, U. S. Dept. Agriculture.

There are no dues, as publication and postage are paid for by the De-
partment of Agriculture, while samples are sent out by C. O. D. express.

Its membership is assumed to include all official chemists — State, Munici-
pal and Federal.

At the recent Jamestown meeting of the Association, 90 members were
present. The meetings are usually held in Washington, where the at-
tendance is often double this number.

AMERICAN SECTIONS, SOCIETY OF CHEMICAI. INDUSTRY.

The New York Section of this Society was founded May 2, 1894. Since
then, New England and Canadian Sections have also been established.
The present membership of the New^ York and New England Sections
combined is about 1500. It is the chief organization of industrial chemists
in the country, and its journal is the best known pubHcation of its kind.
Among the reasons for. its establishment, as stated by one of its founders,
are (i) that the meetings of the New York Section of the American Chemi-
cal Society were at the time devoid of interest to chemical manufacturers;
(2) that the American Chemical Society itself did nothing to promote
chemical industry; (3) that it discriminated against industrial articles
submitted for publication; and (4) that manufacturers were admitted
only as Associates.^

^ It need hardly be said that these statements were based on an almost complete
misapprehension of the attitude of the American Chemical Society toward Industrial
Chemistry.

176

MARSTON TAYLOR BOGERT.

'k^n

NEW ENGLAND ASSOCIATION OF CHEMISTRY TEACHERS.

Of quite a different type is the New England Association of Chemistry
Teachers. Founded February 19, 1898, "to promote efficiency in the
teaching of chemistry/' this Association now has a membership of nearly
200. Any person interested in the teaching of chemistry is eligible to
membership. The number of Active members is limited to 75, but no
limit is set for the number of Associate or Honorary members. Three
meetings a year are held, most of them occurring in the vicinity of Boston.
The Anntial Report of these meetings makes a volume of about 1 10 pages.
Active members pay $2.00 a year, Associates $1.00. There are no local
sections. The character of the permanent committees gives a good idea
of the scope of the work. They are as follows: College requirements,
current events and publications, high school course of study, industrial
chemistry, laboratory construction, new apparatus, instruction, and
social committee.

NEW YORK SECTION, VEREIN DEUTSCHER CHEMIKER.

Established in December, 1900, this New York Section has at present
125 members. Its meetings are generally held at the Chemists' Club in
New York City, at the close of those of the New York Section of the
Society of Chemical Industry, and are purely social in character. The
official publication of the Verein is the Zeitschrift fiir angewandte Chemie.

AMERICAN ELECTROCHEMICAL SOCIETY.

The original call for the organization of this society was dated Phila-
delphia, October 19, 1901, and was sent to about thirty people. It was
signed by Messrs. Hering, Reed, Richards, Roeber, Sadtler and Wahl,
and began thus: *'The rapidly growing importance of the subject of
electrochemistry, and the want of suitable occasions in this country for
the discussion of papers and questions pertaining thereto by those espe-
cially interested, have suggested the advisability of founding a national
Electrochemical Society, similar in its organization to the American
Chemical Society and the American Institute of Electrical Engineers."

The preliminary meeting was held at the Engineers' Club, Philadelphia,
and it was decided to organize, provided seventy-five members were
assured. A second circular letter was therefore sent out to engineers,
chemists and metallurgists, including all members of the American Chemi-
cal Society and of the American Institute of Electrical Engineers. This
was dated New York, November 25, 1901, and contained the following:
"The products of electrochen.ical industries in this country at the present
time amount to about one hundred million dollars per year. The grow-
ing importance of these industries and the fact that scientists and en-
gineers interested in electrochemistry are now distributed among at
least half a dozen different societies, and therefore have no common

ABISRICAN CHSBUCAI/ SOCIBTIBS. 1 77

medium of communication, suggested the formation of an American
Electrochemical Society, on the same general plan as the American Chemi-
cal Society and the American Institute of Electrical Engineers

The bringing together in this way of those engaged in the scientific study
of electrochemistry and the practical engineers and pioneers of the in-
dustry, will be of inestimable value to both."

It is appropriate to add that one of the most prominent of the charter
members states that one of the reasons for forming a separate society
was that electrical engineers interested in chemistry were, at that time,
admitted by the American Chemical Society only as associates.

The meeting for organization took place at the Manufacturers' Club,
Philadelphia, April 3, 1902, fifty-two members being present. The total
number of charter members was 336.

Two volumes of Transactions are published annually, and three local
sections have been formed: New York, Philadelphia and Madison (Wis.).
The total membership at present is about 700.

WESTERN ASSOCIATION OF TECHNICAL CHEMISTS AND METALLURGISTS.

In 1904 this Association appeared on the scene. It was established
because, in the opinion of its organizers, there was then no society in the
country that met the needs of the men engaged in the extraction of metals
and rare earths from their ores. It is incorporated under the laws of
Cobrado, with headquarters at Denver.

Its first general annual meeting was held at Denver, Colo., the second at
Salt I/ake City, and the third is now being held at Deadwood, S. D. Local
sections have been established at Denver, Salt Lake City and Butte, and
others are in process of formation. The total membership at present is
about 250. The objects of the Association are the general advancement
of technical chemistry, the improving and promoting uniformity in
methods of metallurgical analysis and assaying, and the encouragement
of research in the metallurgy of precious and rare metals. Any one
interested in these objects is eligible for membership. The official organ
of the Association is the Western Chemist and Metallurgist.

SOCIETY OF BIOLOGICAL CHEMISTS.

The latest separate chemical society is that of the biological chemists.

In 1899 a number of New York physiological chemists established the
Society of Physiological Chemists. This is a purely local organization,
meeting about once a month during the winter for the presentation and
discussion of recent important work in this field. It is somewhat of the
nature of a seminar in physiological chemistry, abstracts and reviews
being submitted rather than original contributions.

The growth of this Society of Physiological Chemists, the development
of the Biochemical Section of the American Chemical Society (established

MARSTON TAYI,OR BOGERT.

), the increasing number of chemical papere on the programmes of the
rican Physiological Society and the great success of the recently
)Iished Journal of BiologUai Chemistry, "were among the influences
stimulated thoughts of a national organization of biochemical work-

: the meeting for organization, New York City, December 26, 1906,
Iwenty-nine biological chemists present were addressed by Professor

as follows: "We have become convinced that there is need in this
try for an organization which shall further the interests and foster
growth of biological chemistry. Biological chemists at present are
ited with widely different societies and come little in contact with the
: body of men who are interested in biochemical work. Whether we
lemists have as our field of work the physiological chemistry of our
cal schools, or deal with the chemical problems of botany, zoolog>~,
ology, pharmacology, or medicine, we all have one common nieet-
[rotmd, and that is, chemistry as applied to animal or vegetable struc-
'; living or dead. As distinguished from the work of pure chemists,
oic or inorganic, our efforts are directed towards throwing light on
life processes and functions of living structures, with the help of

lical and physico-chemical methods Organization develops

iination of effort, encourages research, it furnishes the mechanism

competent criticism and helpful discussion and makes it

;nt to faculties of science and medicine, and to scientific and medical
ties, that a great and growing department of research demands its
ig place in the general scheme of higher education I believe

we can have a society on broader lines than is possible to a mere
9n. We wish to draw into our society biological chemists in all de-
nents of biology, including those organic and physical chemists who

a Uvely interest in our subject."

ghty-one chemists enrolled as charter members. Their contributions

lublished mainly in the Journal of Biological Chemistry, which I under-

1, is not an official organ, but the private property of Messrs. Abel

Herter.

the limited time available, I have endeavored to give you some idea
e more prominent American chemical societies of the past as well as
le present, the reasons for their establishment, the fields covered,
data indicating their present size and influence.

'en this hasty and imperfect survey unavoidably forces upon us the
lusion that, in the judgment of many of the chemists in the countrj-,
American Chemical Society has not adequately met the needs in all
ches of the work, otherwise these separate societies would not have
n. Such a condition of affairs must, of course, cease, or total dis-
ration will only be a question of time.

AMERICAN CHEMICAL SOCIBTIES. 1 79

I am happy to say that the Society is fully alive to the situation and
that a larger future seems opening up before us. The first step in this
direction has been taken — ^the publication of Chemical Abstracts. It
might justly be called a stride rather than a step, for, in my opinion, no
single act on the part of our Society could have done more to unite the
chemists of the country than the publication of these abstracts, covering
as they do every branch of the subject.

Any national organization of chemists, if it would be successful, must
take into account the following factors:
L Specialization.
II. Publication.

III. Geographical location.

I. Specialization. — ^The history of chemical organization shows the
usual evolution from the general to the special. First came the societies
for science in general, then those devoted to chemistry alone, and now
those for special branches, which, in their turn, may tmdergo a still further
division in the years to come.

Every chemist has some one branch of the subject in which he is particu-
hrly interested. But little progress would be recorded were it not so.
It is as fitting as it is inevitable, that those following the same specialty
should wish to meet together periodically to discuss matters of mutual
interest. It is equally certain that as they grow in numbers they will
seek organization, and demand a medium of publication which shall give
suitable recognition of the importance of their special branch of chemistry.
The society which fails to take cognizance of the growing strength of
specialization and to lay its course accordingly, fails to grasp its oppor-
tunities and slowly but surely will be crowded to the wall. We should
not labor under the delusion that if our society fails to recognize this
tendency to specialize, specialization will therefore cease.

The vital point is that the American Chemical Society must show that
it can adapt itself to this condition, and by suitable changes in its plan of
organization do far more for the fostering and stimulating of special
branches of chemistry than could be accomplished by separate and in-
dependent societies. That this can and will be demonstrated, I con-
fidently believe, and I am sure that ever}'^ well-considered move in this
direction will meet with the hearty approval of all.

The practice of holding our semi-annual meetings in sections is a move
along this line, and I believe that a further advance would be the or-
ganization of these sections on a- somewhat different basis. As a first
step, they might elect their own ofiicers to serve for a year, let us say,
who shall do what they can to make the semi-annual meetings as success-
ful as possible, by providing attractive programmes and a large attend-
ance, inducing other organizations with similar interests to meet with

l8o HARSTON TAYLOR BOGBRT.

them (as the Society of Biological Chemists is now meeting with our
Biological Section), pointing out to the Council in what ways the Society
may be made more valuable to their particular group of members, and in
general, doing whatever they can to stir up increased interest. These
general sections should be given as large a measure of self-government as
possible. Their presiding officers should be ex-officio members of the
Council, and should be associated with the president and secretary in the
preparation of the programmes for the general meetings. This is essen-
tially the policy recommended in 1903 by the committee of which Dr.
A. A. Noyes was chairman, and it is not unlikely that such a policy,
particularly when accompanied by a suitable development of our pub-
lications, would lead some of the existing specialized societies to unite
with us.

To those of our members who fear that such a move may introduce a
disintegrating element, I would cite the recent action of the Verein
Deutscher Chemiker.

This great German society, at its recent general meeting in Danrig,
May 23-25, 1907, changed its constitution and by-laws so as to provide
for the estabUshment of separate sections or groups (Fachgnippe) for
those working in the same field.

So far, sections have been organized for Paper Chemistry, Fennenta-
tion, and Technological Chemical Instruction, and the following additional
anes are contemplated in the immediate future: Heavy Inorganic Chemi-
::als, Legal Protection of the Trade, Color Chemistry, Medico-pharmaceuti-
:al Chemistry, Analytical, Organic, Photo- and Electrochemistry, and
Chemical Apparatus and Machinery. Calls have already been issued for
the organization of several of these.

At the preUminary meeting for the organization of the Fermentation
Section, in Berlin, the nth of last November, Geheimrath Delbriick, in
the course of his opening remarks, said: "that in the profession there is
great need for the founding of such a section, is shown by the large at-
tendance at this meeting and the numerous apphcations for member-
ship." He stated further that 66 chemists had already signified their
desire to join and that "there would thus be added to the Verein a whole
^oup of new members." Some of those present even thought it wise
to prepare for the formation of sub-sections within the sections.

To any one reading in the Zeiischrift fur angewandte Chemie the re-
ports of their meetings, it will appear that a wave of enthusiasm is sweep-
ing all along the line as the direct result of this plan of establishing sec-
tions. Not only have the members no doubts whatever as to the bene-
Scial effects of such a sub-di\ision of the Verein, but they appear con-
Sdent that it means a brighter future and a wider sphere of usefulness
for their organization.

AMERICAN CHEMICAI/ SOCIETIES. l8l

11. Publication. — ^The publications of the Society must truly represent
the varied interests of the membership. To those of our number who are
situated at a considerable distance from any large city and have no library
accessible and no local section, the character of our publications is of
preeminent importance. The development of specialization already
alluded to must be provided for either (i) by the establishment of separate
journals by our Society to more fully cover these special fields, or (2) by
the development of appropriate divisions within our present Journal.

Under the former plan, some such division as the following might be
suggested, although probably no two men would divide up the field in
exactly the same way :

(i) Our present Jowrnal, to contain articles of general interest, reviews,
pnKeedings, and the like, as well as articles in those branches for which
no special journal is provided.

(2) Chemical Abstracts. A publication needed by every chemist, no
matter what his specialty, and a most valuable bond, therefore, in hold-
ing all chemists together.

(3) A Journal of Industrial and Engineering Chemistry, to include the
best features of the existing journals of industrial and engineering chemistry
and metallurgy.

(4) A Journal of Biochemistry. If consolidation with the Society of
Biological Chemists could be brought about, the preient Journal of Biolog-
icd Chemistry might become the official organ.

(5) A Journcd of Inorganic and Physical Chemistry, to include the
present Journal of Physical Chemistry, or to be a continuation thereof.
There is very good reason for believing that this can be brought about
when the time seems opportune.

(6) A Journal of Organic Chemistry. If a way could be found acceptable
to Dr. Remsen, by which the present American Chemical Journal would
take this place, the problem would be a much simpler one.

Of these various journals, all the members would receive the first two,
while the others would be furnished at a price slightly in excess of the
actual cost — a price, by the way, far below that for which separate so-
cieties could afford to supply them. Or perhaps, some different arrange-
ment would be deemed preferable.

The other plan contemplates segregation rather than separation, and
has in mind the development of our present Journal until it somewhat
resembles a union under one cover of all the various journals mentioned
above, with the possible exception of Chemical Abstracts.

This latter plan has in its favor evident advantages. It would cost
far less for publication, and such an enlarged Journal could probably be
sent to all the members without any additional charge whatever. Un-
certainties as to the proper place for an article to appear would arise only

I83 C. JAMBS.

as between subdivisions of one and the same journal, and not as between
different journals. Repetition could be avoided by proper cross-ref-
erences. There would never be any very great danger of a lack of suffi-
cient material for such a journal. Its large circulation would also make it
a most desirable medium for advertising, and the increased income from
this source would still further reduce the cost of publication.

Perhaps the chief objection to such a plan lies in the rapidly increasing
amount of material submitted to our various chemical journals. The
bulk of this material if not already too great to be handled properly by
any one journal, would certainly become so in the course of the next few
years.

It may be that a combination of these two plans would appear best,
arranging for the issuing of those sepaTBte journals which seem roost
urgently needed at the present time, if there are any such, and meanwhik
developing the journal by a segregation of its contents, perhaps with
suitable divisional headings, so that when the moment is most opportune
these sub-di\'isions may start an independent career as special journals.

Already an able and energetic committee is at work on the question (rf
the advisability of our publishing a Journal of Industrial and Engineering
Chemisiry, and we hope to have a report from them at this meeting.' .

III. Location. — As far as practicable, opportunities should be pro-
vided for our members in all parts of the country to hold periodic meet-
ings. To insure this we have our Local Sections and migratory General
Meetings. So far as the latter are concerned, however, the country is so
large and our membership so widely scattered, that only a small propor-
tion find it pos^ble to attend. It might be wise, therefore, to hold
one general meeting during the Christmas hoUdays, and during the
summer have several separate gatherings in different parts of the
country — say, one West of the Rockies and two or more in the Hast.

To return to my original statement, what the American Chemical
Society needs is the enthusiastic and intelUgent co&peration of its mem-
bere. I am sure that the enthusiasm will be forthcoming, and I trust
that the data presented may be of some service in helping you to decide
intelligently as to the best plans for the development of our Society.

THE BROMATES OF THE RARE EARTHS.

Part I. A New Method for the Separatioii of the Yttrium Earths.

Bt C. Jun.

Received December «, I90T.

During recent years chemists investigating the rare earths have dt-
' The report will be found in Proceedings of this number.

THE BROMATBS OF THE RARE EARTHS. 183

rected thdr efforts more towards the cerium group than towards the
yttrium earths, with the result that the elements cerium, lanthanum,
praseodymium, neodymium, samarium, europium and gadolinium are
fairly well known. On the other hand, the yttrium earths, with the
exception of yttrium, )rtterbium and scandium are still but poorly de-
fined. Many investigators have been of the opinion that **new" erbium
is complex, while almost nothing is known about thulium, holmium,
dysprosium and terbium. In addition to the above, victorium, annotmced
by Crookes^ is believed by Urbain and some others to be a mixture, al-
though its discoverer still claims that it is a separate and new, individual
element.

In the case of the cerium group, several excellent methods of fractional
crystallization have been developed. On the other hand, no rapid and
simple method has been applied to the yttrium earths with the possible
exception of the method of Urbain, using the ethyl sulphates, and some
methods emplo3dng compotmds of a costly nature, such as the acetyl-
acetonates and the metanitrobenzenesulphonates. As one must deal
with many kilograms of material, expensive compounds are out of the
question tmless the separation should prove almost quantitative, which
is never the case.

Urbain's ethyl sulphate method* gives good results. He sajrs that
yttrium, neoerbium and )rtterbium accumulate in the most soluble por-
tions, with no trace of the lanthanum group and no earths of the samar-
ium and gadolinium groups. It is very difficult to obtain erbium from
fractions rich in holmium by its use and for separating earths of the same
group, the method of fusing the nitrates is still the best. Another diffi-
culty is due to the fact that the ethyl sulphates are inclined to hydrolyze
unless special precautions are taken.

The fractionation of the simple nitrates from concentrated nitric add
proposed by Demarcay* is very tedious and the use of a solvent of such
a character causes great inconvenience in the laboratory.

The separation obtained by taking advantage of the difference of solu-
bility of the oxalates in a saturated solution of ammonium oxalate, de-
vebped by Carl Auer von Welsbach,* gives interesting results. It has
drawbacks, however, as it involves the use of two temperatures and re-
quires a large number of operations.

Among the remaining methods only a few need be mentioned : Fusion
of the nitrates and separation of the most easily decomposed portions
by their insolubility in water, separates ytterbium and scandium at one
end, and yttrium and terbium at the other. To obtain ytterbium free

'Compt. rend., 126, 835; 127, 107; 132, 136.
*7W., 123, 728; 130, 1020.
*Monatsh., 37, 935.

[S4 C. JAHBS.

from erbium and thulium, requires about seventy operations. It is
ilso extremely difficult to get yttrium pure by this method, since terbium
:lings.

Fractional precipitation of the concentrated neutral nitrate solution
by means of magnesium oxide,' tending to throw down the less baac
■lements first, gives similar results.

The preparation of pure yttria by Muthman and Bohm's chromate
method* is comparatively simple, but unfortunately it is of no value for
separating the other members.

The fractionation obtained by boiling a solution of the oxalates in
immonium carbonate* works well for obtaining erbium free from hoi-
nium and dysprosium, on the one side, and thulium, ytterbium and
icandium on the other.

Fractional precipitation to be of value must be very rapid. Fractional
:rystallization is to be preferred, for it is much easier to carry out a lai^
lumber of operations. As a rule, in fractional precipitation methods,
especially where dilute solutions are employed, a good deal, if not most,
)f the material is washed away. It is highly desirable, therefore, that a
nethod consisting of fractional crystallization of some type of isomorphous
x>mpounds, with greatly varjang solubilites, should be found. With
:his object in view, the author has examined the sulphites, xanthates,
iuccinates, double carbonates with sodium glycollates, methyl sulphates,
lormat propyl sulphates, camphorates, iodates, thiocyanates, mono-
^hloracetates, monobromsucdnates, oleates, bromates, etc, besides
learly every compound proposed in Uterature for the purpose of fractiona-
,ion. The bromates are the best suited for the purpose of all those ex-
tmined up to the present time.

The bromates are easily prepared by using barium bromate, which,
n turn, is formed by mixing boiling solutions containing the required
imounts of barium chloride and potassium bromate. Because bariuni
)rcmate is not very soluble even in boiling water, it is highly important
Jiat the precipitate should be finely divided, otherwise the double decoln-
wsttton between the rare earth sulphate and barium bromate will take
nnsiderable time. The formation of large crystals is prevented by rap-
dly cooling the mixed boiling solutions. As potassium bromate can be
)repared cheaply, the rare earth bromates are not costly to obtain.

The rare earth material, generally in the form of the oxalates, is mixed
nto a paste with sulphuric acid and the temperature raised until the
umes of sulphuric acid cease to be evolved. The residue is then finely
TOwdered, dissolved in ice-cold water, and the resulting solution poured

' Muthman and Rolig, Ber., 31, 1718.

' Ber., 33, 49; Chem. News, 81, (69,

■ This Journal, 19, 495; Chem. News, 95> 181.

THB BROBiATES OI^ THE RARE EARTHS. 185

over an excess of barium bromate. This operation is best carried out
in a large evaporating dish placed on the water-bath, care being taken
to keep the mass well stirred.

After a time the precipitate is allowed to settle and some of the clear
Hquid taken up by means of a pipette and added to a warm solution of
barium bromate; if no precipitate is obtained the liquid is filtered ofif.
Sometimes, however, a precipitate is formed which consists of barium
biomate and, therefore, it is best to dilute with water and boil. If the
precipitate persists, either more stirring or more barium bromate is re-
quired.

When the double decomposition is complete a little bromine is often
liberated, but there is not sufficient to cause any inconvenience in the
laboratory. This is evidently due to the fact that a small amount of
bromic add is formed by the action of a trace of free sulphuric acid ac-
compan3dng the rare earth sulphates. The latter should, therefore,
be well ignited.

The filtered liquid is evaporated until a drop, removed on the end of
a glass rod, nearly solidifies when stirred on a watch glass. Under these
conditions just about half of the substance in solution crystallizes out
on cooling. After a little experience there is absolutely no diflSculty
in judging the most convenient concentration. If the fractionation is
carried out in porcelain dishes a little water should be sprayed on the
surface so as to prevent the top from solidif)dng to a crystalline mass.

Casseroles are by far the best utensils to use for this work, especially
if small amounts of substances are being separated, as they can be cov-
ered by large watch glasses. This prevents rapid crystallization and
the tendency of the material to creep up the sides of the vessel. Also
there is no need to spray any water on the surface after evaporating or
dissolving. Very often, but usually when working with small quanti-
ties, the liquid refuses to crystallize or else the crystals separate out as
a fine feathery mass, so that it is quite impossible to pour off the mother
Hquor. If it does not crystallize, the best procedure is to add a trace of
the solid, when the whole immediately solidifies, forming the feathery
type of cr3rstals as mentioned above. The mass is then carefully heated
so as to dissolve all but a very little, which will start the crystallization
as the liquid cools. An even better plan is to commence the operation
by the addition of a crystal while the liquid is still quite hot.

When working on the large scale, very fine hexagonal prisms are often
obtained, some being more than two inches in length and over a quarter
of an inch in thickness.

Even after six series of crystallizations a considerable change is easily
apparent, although this is shown more by the spectroscope than by the
color. The most soluble fraction is very pink and it gives an inteo^

S6 C. JAMBS.

rbium spectrum. The thulium red is also very strong and the band
1 the blue has made its appearance while the bands of holmium and
ysprosium have nearly disappeared. The least soluble fraction is still
inkish and gives intense dysprosium and holmium bands and a weak
rbium spectrum. The color of the oxide of this fraction is orange,
lowing that terbium accumulates at this end. After the process has
een continued until some twenty fractions have been obtained, the
last soluble portion forms brilliant colorless crystals, which dissolve
1 water with a tinge of greenish yellow color, seen only in concentrated
ilution. The absorption spectmm of this fraction shows very faint
imarium and holmium bands, while those of dysprosium are much
ironger. Practically the whole of this fraction consists of an earth ^ving

colorless salt, yttrium. And as the oxide is of a brown ochre color it
lows that terbium collects in the least soluble portion. As one goes
own the series, the fractions become yellower and the oxides paler. In
le fractions that show the strongest yellow, the dysprosium and holmium
ands are very intense, and the oxide becomes yellowish. Farther along
le series, the lines of erbium make their appearance and even while the
rbium absorption is still weak the hquid assumes a pink color. This
icreases until it reaches a rosy pink, at which stage the spectroscope
lows only erbium bands, and the oxide is of a pure rose tint. Further on
till the thulium band in the red shows itself, while the solutions become
aler and give a very stong thulium spectrum, the erbium bands becom-
ig decidedly weaker. The most soluble fraction is reached when the
)lution is nearly colorless, the erbium spectrum is faint, while that of
lulium, although fainter, is still intense. The oxide of this last fraction
i white and dense and consists largely of ytterbium.

The above series show that the rare earth bromates arrange them-
;lves in the following order of solubility:

Samarium (Europium?, Gadohnium?), Terbium, Yttrium, Dyspro-
um, Holmium, £rbium. Thulium and Ytterbium — which is dmilar to
le solubihties of the oxalates in ammonium oxalate, but different from
le ethyl sulphates; since according to Urbain, yttrium, erbium and ytter-
ium ethyl sulphates are found in the most soluble portion.

A fair conclusion can be drawn that the use of the ethyl sulphate method
ould prove valuable in conjunction with the bromate, especially for
le separation of yttrium from dysprosium and holmium and perhaps
)r the separation of thulium from ytterbium.

Finally, the author would like to point out some of the rapid separa-
ons obtained by this method, the most remarkable being the sepa-
ition of thulium from erbium. And since ytterbium is still more solu-
le than thulium, its removal is even easier. For example, erbium
laterial, supposed to be quite free from thulium, was converted into

REVISION OF THE ATOMIC WEIGHT OF I.EAD. 1 87

the broraate and crystallized four times, when the most soluble portion
gave the thulium spectrum.

Dysprosium and hohnium also separate from erbium with compara-
tive ease, and as yttrium places itself between terbium and dysprosium,
the latter element can be obtained terbium-free. The division between
dysprosium and holmium is not so marked.

The absorption spectra and the colors of the various fractions show
curious changes which are not altogether understood and may be simply
due to the comparatively small amount of material under examination.

More material (15 kilos) is at present being fractionated and in the near
future the less basic portion* of earths derived from about 100 kilos of
euxenite and 100 kilos of yttrotitanite, etc., will also be included. The
bromate method will be further investigated and also applied to the
cerium group and the results published at an early date.

Nbw Hampsbxrb Collbob,
Dnrium, N. H.. November 18, 1907.

[Contributions From the Chbmical Laboratory op Harvard College.]

A REVISION OF THE ATOMIC WEIGHT OF LEAD.

Preliminary Paper — ^The Analysis of Lead Chloride.

Grboort Paul Baxtbr and John Hunt Wilson.
Received December 2, 1907.

Although lead is one of the most common elements, its atomic weight
has received comparatively little attention, the value at present accepted
being based almost wholly upon the work of Stas.^ Of the earlier de-
terminations of this constant those of Dobereiner* and Longchamps*
can hardly be considered as possessing other than historic interest. The
first results which can lay claim to accuracy are those of Berzelius,*
who obtained values ranging from 206. 7 to 207.3 by reduction of litharge
in a current of hydrogen. Berzelius also synthesized the sulphate from
metallic lead with the result 207.0.* Shortly after, Turner" criticized

^ Earlier work on the atomic weight of lead has been carefully summarized by
Qarke. Smithsonian Miscellaneous Collections, Constants of Nature, "A Recalcula-
tion of the Atomic Weights," 1897.

In recalculating the data of earlier determinations the following atomic weights
have been used in this paper:

0=16.000; Ag=io7.88; 0=35.46; N=i4.oi; 8=32.07
Richards and Wells, Pub. Car. Inst., No. 28 (1905); Richards and Forbes, Ibid., No.
69, p. 47 (1907); Richards and Jones, Ibid,, No. 69, p. 69; Report of International
Comniitte on Atomic Weights, This Journal, 29, no (1907).

*Schweig. J., 17, 241 (1816).

• Ann. chim. phys., 34, 105 (1827).

• Pogg. Ann., 19, 314 (1830).

• Lehrbuch, 5th ed., 3, 1187 (1845).

• Phii Trans., 527 (1833).

i88

GREGORY PAUL BAXTER AND JOHN HUNT WIl^ON.

■ rHi^

'-^•'. 11:

'fS:

•* -^

( , ■?<

■ :V ■■

' i >

■■: X ;-i

- •• *■'* '

..- 5 ■ " .

-. v^

M C

X ^ * -

?-r'

■ ^ ■■ :>

- V

^•:ir

L- ' t

^;■a

P'^rr

the first method employed by Berzelius and attributed the irregularity
of his results to the action of lead oxide on the siliceous matter of the
tube at the temperature employed in the reduction. By the conver-
sion of both the metal and the oxide into sulphate Turner, in a pains-
taking research, deduced the values 207.0 and 207.6 respectively, and
by converting the nitrate into sulphate, 204.2. Marignac* converted
metallic lead into the chloride by heating in a stream of chlorine and
obtained the result 207.42. Both Marignac* and Dumas* analyzed
lead chloride. Marignac, who dried the salt at 200**, by titration against
silver found the atomic weight of lead to be 206.81, and from the ratio
of lead chloride to silver chloride, 206.85. Dumas subsequently showed
that lead chloride, even when dried at 250°, retains moisture and is
somewhat basic, and in one analysis in which corrections are applied
for these errors, found a somewhat higher value, 207.07, as was to be
expected. Chloride analyses by early investigators are, however, to
be universally distrusted, owing to neglect of the very considerable solu-
bility of silver chloride, thus producing too low results.

Stas's* work upon the syntheses of lead nitrate and sulphate from the
metal is undoubtedly the most accurate contribution upon the subject,
although a careful consideration of his work discloses minor defects,
many of which he recognized himself. The metallic lead used in the
syntheses was finally fused under potassium cyanide. Whether or not
this treatment introduced impurities into the metal is uncertain. Stas
himself suspected the presence of alkalies in the metal. Since the nitrate
could not be dried above 150° without decomposition, it undoubtedly
contained moisture, and Stas calls attention to this point. The sul-
phate was made by treatment of lead nitrate, resulting from the nitrate
syntheses, with sulphuric acid. The sulphate was dried finally at dull
redness, and was probably free, or nearly free, from moisture, although
it may have contained traces of lead oxide resulting from occluded ni-
trate, as well as sulphuric add. Most of these probable errors tend to
lower the observed atomic weight, so that Stas*s value from the series
of nitrate syntheses, 206.81, and that from the sulphate series, 206.92,
are to be regarded as minimum values. The reader of Stas's own ac-
count of his work upon lead cannot fail to be impressed with the fact
that he was somewhat dissatisfied with the outcome of his research.
Mention should also be made of the work of Anderson and Svanberg*
on the conversion of lead nitrate into oxide, although the method was

* Ann., 59, 289 (1846).

• J. pr. Chem., 74, 218 (1858).
•Ann., 113, 35 (i860).

* CEuvres Completes, i, 383.

• Ann. chim. phys. [3], 9, 254 (1843).

RB VISION OF THE ATOMIC WEIGHT OF LEAD. 1 89

primarily employed in an endeavor to fix the atomic weight of nitrogen.
Their results yield the value 207.37.

The discrepancies between the results of these various experiments
only serve to emphasize the need of a redetermination of the value in
question, and it was with this object in view that the work embodied
in this paper was undertaken.

The search for a suitable method for determining the atomic weight
of lead failed to reveal any more promising line of attack than those
already employed for the purpose. With an element of so high an atomic
weight as lead, in any method involving the change of one of its com-
pounds into another^ errors which may be insignificant with elements
of small atomic weight are magnified in the calculations to undesirable
proportions. Furthermore, during the following investigation, reduc-
tion of the chloride and oxide in hydrogen was investigated far enough
to show that complete reduction of either compound was extremely
difficult, if not impossible, without loss of material from tlie containing
vessel by sublimation, aside from the fact that all available material
for containing vessels is acted upon by either the fused salt or the reduced
metal. The elimination of moisture from lead nitrate or lead sulphate
without decomposition of the salts seemed likely to prove a stumbling
block in the use of these substances. Finally, in spite of the slight solu-
bility of lead chloride, the determination of the chlorine in this salt by
precipitation with silver nitrate was chosen as presenting fewest diffi-
culties. In the first place, the determination of a halogen can be effected
with great accuracy. In the second place, the elimination of moisture
from lead chloride is an easy matter, since the salt may be fused in a
platinum vessel in a current of hydrochloric acid gas without attacking
the platinum in the least and without the production of basic salts. In
the third place, silver chloride, which has been precipitated from a dilute
solution of lead chloride by means of silver nitrate, does not contain
an amount of occluded lead salt large enough to be detected.

Purification of Materials.

Water. — All of the water used in either the purification or the analyses
was distilled twice, once from an alkaline permanganate solution and
once from very dilute sulphuric acid. Block-tin condensers were used
in both distillations, and rubber and cork connections were avoided.
Generally, receivers of Jena glass were employed, but in certain cases
the water was collected in platinum or quartz vessels.

Hydrochloric acid. — Commercial C. P. hydrochloric acid was diluted
with an equal volume of water and distilled with a quartz condenser,
only the middle fraction being collected.

Nitric acid. — ^Nitric acid was distilled with a platinum condenser,

90 GREGORY PAUL BAXTER AND JOHN HUNT WIL30N.

intil free from chlorine. Two distillations were invariably i

o accomplish this end, if the first third of each distillate was rejected.

SUver. — Pure silver was obtained by methods already many times
mployed in this laboratory. Silver nitrate was dissolved in a larfe
olume of water and the silver was precipitated as chloride with an ex-
ess of hydrochloric acid. The precipitate was thoroughly washed
,nd reduced with alkaline invert sugar. The reduced silver, after being
k^shed, was dried and fused on charcoal in the flame of a clean blast
imp. After the buttons had been cleaned by scrubbing with sand
,nd etching with nitric add, they were dissolved in pure dilute nitric
.cid and the silver was precipitated as metal with ammonium formate.'
'his silver was washed and fused in the flame of a blast lamp on a crucible
>f the purest lime. The buttons were cleaned as before, and then elec-
rolyzed.* Finally the electrolytic crystals were fused in a boat of the
lurest Ume in a porcelain tube in a current of pure electroljrtic hydrogen.*
'he bars of silver were cut in pieces with a fine steel saw, etched with
lilute nitric acid until free from iron, washed, dried, and heated in a
■acuuni to 400". The silver was kept in a desiccator containing solid
■otassium hydroxide.

Lead chloride. — Three samples of lead chloride from two entirely dif-
erent sources were employed. Sample A was prepared from metalUc
ead. Conmiercial lead was dissolved in dilute nitric acid, and the so-
ution, after filtration, was precipitated with a slight excess of sulphuric
icid. The lead sulphate was thoroughly washed, suspended in water,
ind hydrogen sulphide was passed in until the sulphate was almost com-
)letely converted into sulphide. Next, the sulphide was washed with
vater, dissolved in hot, dilute nitric acid, and the solution was freed
rom sulphur and unchanged sulphate by filtration. The lead nitrate
hus obtained was crystallized twice, dissolved in water, and predpi-
ated in glass vessels with a slight excess of hydrochloric add. The
hloride was washed several times with cold water and then crystal-
ized from aqueous solution eight times, the last five crystallizations being
■arried out wholly in platinum, with centrifugal drainage after each crys-
allization. In crystallizing the lead chloride the whole sample was
lot dissolved at one time, but the same- mother liquor was used for dis-
olving several portions of the original salt. Needless to say, the chlo-
ide was not exposed to contact with the products of combustion of
Uuminating gas, lest lead sulphate be formed.

Sample B was prepared from commercial lead nitrate. This salt was
lissolved and crystallized from dilute nitric add once in glass and six

■ Richaids and Wells, Pub. Car. last.. No. 26, 19 (1905).
* Abrahan, J. Chem. Soc. Proc., 1893, p. 660.

■ Baxter, Proc. Am. Ai»d., 39, 349 (1903).

REVISION OI^ THE ATOMIC WEIGHT OI^ I^BAD. I9I

times in platinum vessels, with centrifugal drainage. Hydrochloric
add was then distilled into a large quartz dish, and the solution of the
nitrate was slowly added with constant stirring with a quartz rod. The
chloride was freed from aqua regia as far as possible by washing with
cold water, and was once crystallized from aqueous solution in quartz
dishes to remove last traces of aqua regia. Finally the salt was crystal-
fized three times in platinum.

It could reasonably be expected that both of these samples were of a
high degree of purity; nevertheless, upon heating the salt in an atmos-
phere of hydrochloric add, the salt itself turned somewhat dark, and
upon solution of the fused salt in water a slight dark residue remained.
Although in a few preUminary experiments attempts were made to de-
termine this residue by filtration and ignition, it was subsequently found
that even a small filter paper adsorbs appreciable amounts of lead com-
pomids from a solution of the chloride, which cannot be removed by
washing with water. From three to thirteen hundredths, of a milligram
of residue were obtained in several blank experiments, by ignition of
filters through which half per cent, solutions of lead chloride had been
passed, with subsequent, very thorough washing. In order to avoid
the uncertainty of this correction, further attempts were made to ob-
tain a sample of the salt which would give a perfectly clear solution in
water after fusion, and thus render filtration unnecessary. With this
end in view a considerable quantity of Sample A was fused in a large
platinum boat in a current of hydrochloric add. The fused salt was
powdered in an agate mortar, dissolved in water in a platinum vessel,
and the solution was freed from the residue by filtration through a tiny
filter in a platinum ftmnel into a platiunm dish, where it was allowed
to crystallize. This sample was then twice recrystallized with centrif-
ugal drainage. Notwithstanding the drastic treatment to which it
had been subjected, when a portion of this material was fused in hy-
drochloric add, the same darkening as before was observed, and the
same residue was obtained. The suspidon that the difficulty was due
to dissolving of the filter paper by the solution of the salt^ led to
a second more successful attempt by crystallization from hydrochloric
add solution in platinum vessels. In this way it was found possible
to prepare salt which showed no tendency to darken upon heating, and
which, after fusion, left absolutely no residue upon solution in water.
Portions of Samples A and B were thus recrystallized three times more.
Smce these two spedmens of material gave identical results, for two
final experiments portions from each of these samples were mixed and
then subjected to three additional crystallizations. This last sample
was designated Sample C.

* Mr. P. B. Goode in this laboratory has recently found a similar difficulty with
the dibrides of the alkaline earths.

GREGORY PAUL BAXTER AND JOHN HUNT WILSON.

Method of Analysis,
be lead chloride contained in a weighed platinum boat was first fused

current of pure, dry hydrochloric acid gas. This gas was generated
dropping concentrated sulphuric acid into concentrated hydrochloric
1, and after being washed with a saturated solution of hydrochloric
, was passed through five towers filled with beads moistened with
tily boiled, concentrated sulphuric acid, to dry the gas. It has al-
ly been shown that phosphorus pentoxide may not be used for this
30se.' After the salt had cooled, the hydrochloric acid was displaced
dry nitrogen, and this in turn by dry air. Nitrogen was prepared
^ssing air charged with ammonia over red-hot rolls of copper gau/e,
excess of ammonia being removed by means of dilute sulphuric acid.

gas was passed over beads moistened with a dilute silver nitrate
tion and over solid caustic potash to remove sulphur compounds

carbon dioxide respectively, and was finally dried by concentrated
huric acid and phosphorus pentoxide. The air was purified and
d in a similar fashion. The apparatus for generating the hydro-
iric acid and for purifying the hydrochloric acid and nitrogen was
ttructed wholly of glass with ground-glass joints. The platinum
t containing the fused chloride was next transferred to a weighing
,le without exposure to moist air, by means of the bottling apparatus,
:h has frequently served for a similar purpose in many atomic weight
^stigations in this laboratory.' After standing some time in a desic-
ir in the balance room, the weighing bottle was weighed. In nost
he analyses the lead chloride was dissolved from the boat by pro-
;ed contact with boiling water in a Jena glass Rask. In the last two
lyses, in order to show that no error was introduced through solu-
y of the glass, the solution was prepared in a large platinum retort,

was not transferred to the precipitating flask until cold,
ery nearly the necessary an ount of pure silver was then weighed
and dissolved in a redistilled nitric acid diluted with an equal volun;e
rater in a flask provided with a column of bulbs to prevent loss by
tering. After the silver was all dissolved, an equal volume of water

added, and the nitrous fumes were expelled Iiy gentle heating. The
tion was then further diluted until not stronger than one per cent.,

added slowly, with constant agitation, to the solution of lead chlo-

contained in the precipitating flask. The precipitation and band-

of the silver chloride were conducted in a room lighted with ruby
t. The flask was shaken for some tin-e and allowed to stand for a

days, with occasional agitation, until the supernatant liquid had
)me clear. Thirty cubic centimeter portions of the solution were
' Baxter and Hines, This Juurnal, i8, 779 {1906).
* Richards and Parker, Proc. Am. Acad. Ans and Science, 33, 59 (1S96).

REVISION OF THE ATOMIC WEIGHT OF LEAD. I93

then removed and tested with htindredth normal silver nitrate and so-
dium chloride, in a nephelometer,^ for excess of either chloride or silver,
and, if necessary, standard silver nitrate or sodium chloride was added,
and the process of shaking and testing repeated until the amounts of
sBver and chloride were equivalent. The test solutions were always
returned to the flask, since they contained appreciable amounts of silver
chloride, and the weight of silver chloride subsequently obtained was
corrected for the quantity thus introduced. Furthermore, if an excess
of silver was found, a negative correction of an equivalent quantity of
silver chloride was necessary.

After the exact end point had been obtained, about two tenths of a
gram of silver nitrate in excess was added in order to precipitate the
dissolved silver chloride, and the flask was thoroughly shaken, and al-
lowed to stand again until the solution was perfectly clear. The silver
chloride was washed, first several times with a very dilute silver nitrate
solution containing four htmdredths of a gram per liter, and then eight
times with pure water. It was next transferred to a Gooch crucible
and dried for several hours in an electric oven, the temperature being
gradually raised to 180°, and was cooled in a desiccator and weighed.
In every case the moisture retained by the precipitate was determined
by fusion in a small porcelain crucible. The silver chloride, dissolved
in the filtrate and washings, was determined by comparison with stand-
ard solutions in the nephelometer in the usual manner. Care was taken
to treat both tubes in exactly the same manner, and final readings were
taken only when the ratio had become constant. Before proceeding
to the nephelometer tests, however, the filtrate and washings were passed
through a very small filter in order to collect a small quantity of asbestos
shreds mechanically detached from the Gooch crucible. The filter was
ignited and weighed, the ash being treated with a drop of nitric and
hydrochloric adds in order to convert any reduced silver into chloride.
In order to find out whether lead or silver nitrates were appreciably
adsorbed by the filter paper, a solution containing lead nitrate, silver
nitrate, and nitric acid of the concentration of these filtrates, was passed
through several small filter papers, which were then very carefully washed.
In four cases, after incineration of the papers, there was found, — 0.00001,
+0.00002, -f 0.00003, -I-0.00001 gram of residue, exclusive of ash. This
correction is so small that it is neglected in the calculations. In all the
analyses the platinum boat behaved admirably, the loss in weight never
amounting to more than a few hundredths of a milligram.

The balance used was a short arm Troemner, easily sensitive to a
fiftieth of a milligram. The gold-plated brass weights were carefully
standardized to htmdredths of a milligram. All the weighings were
* RichaTds and Wells, Am. Chcm. J., 31, 235 (1904) ; 35, 510 (1906).

GBEGORY PAUL BAXTER AND JOHN HUNT WILSON.

ly substitution with tare vessels as nearly like those to be weighed
ible.

um corrections: The values of the density of lead chloride as
ly various observers range from 5.78 to 5.805,' the mean of the
ccurate determinations being 5.80. This gives rise to a vacuum
on of +0.000062 for each apparent gram of lead chloride, the
of the weights being assumed to be 8.3. The other vacuum
ons applied were silver chloride +0.000071, and silver — 0.000031.
nalyses which were carried to a successful completion are recorded
ables.

PelBht of
PbCI, U

Thb Atomic Wbioht op Lead.
Series I. PbO,: aAg.

Jeighl

:a<UciJ

4.67691

63061

— 0

00074

3,63987

3.6770s

85375

00000

3.8537s

4,14110

31.188

000 30

3.31408

4.56988

M673

00000

3.54673

5.13387

97596

— 0

00038

3.97568

3.85844

99456

00000

3.99456

4.67344

63638

00000

3,63638

3 «03i7

40837

00000

3.40837

4.39613

33437

00030

3 33407

307.179

As^l07.SBo.
C1-3S-4SJ,

307.079

307.073
307.085
307.101
307.086
307. 0S9
307.088

307.103
307. 0S8

Wclgbt WeJKht
PbCIfln AECIJn

Series II. PbCl,;3Aga.

Weigbt

lbt fro

weigl
Coriected of h
weight Ag=

[hi

™fb.
A«=

■«bcit«, wilen, of AgCl.
jCIi, Gramt, Grama. Gram Gram. GratD. Gram a, t(= J5.<73, CI^m <;t

A 4.67691 4,82148 0,00100 o OO031 0,00304 4.82173 307 188 307.088
A 4,14110 4.2684S 0.00030 o 00008 0,00180 4,37016 307,193 307,093
B S-I3387 S-38116 0.00054 o 00013 0,00197 5.28272 307. iSi 307.081
B 385844 3-97759 0.00035 0.00033 0,00193 3. 97949 J07, 136 307.036
C 3.10317 3 19751 0.00045 0.00014 0.00189 3-19909 307.361 307.161
C 4.39613 4.43730 0.00030 0,00004 0,00368 4.43983 307,304 307.104

307.193 307.093

rejecting tbe least satisfactory analyBcB, 13 and 14 107.191 307,091

>t Series I and II 307.190 307.090

Jose agreement of the averages of the two series is strong evi-
hat no constant error, such as occlusion, affects the results,
more, in all, 19.55663 grams of silver produced 25.98401 giams
■ chloride, whence the ratio of silver to silver chloride is 132.865,
in close agreement with the result 132.867 obtained by Richards
idolt-Bdmstein-Meyerboffer, Tabellen.

CATALYTIC DECOMPOSITION O^ HYDROGBN PHROXIDB. 1 95

and Wdls.* The dififerent samples, A, B, and C, all give essentially iden-
tical results.

It appears, then, that if the atomic weight of silver is taken as 107.93
(0=i6.ooo), the atomic weight of lead is 207.19, nearly three tenths
of a unit higher than the value now in use. If the atomic weight of
silver is 107.88, a value probably nearer the truth than 107.93, lead
becomes 207.09, a number still much higher than that depending upon
Stas's syntheses, as is to be expected.

We are greatly indebted to the Carnegie Institution of Washington
for assistance in pursuing this investigation, also to Dr. Wolcott Gibbs
and to the C)rrus M. Warren Fund for Research in Harvard University
for many indispensable platinum vessels.

Cambsidob. Mass.,
October 18, 1907.

CATALYTIC DECOMPOSITION OF HYDROGEN PEROXIDE UNDER

HIGH PRESSURES OF OXYGEN.'

By B. B. Spbar.
Received October la, 1907.

Introduction.

The mechanism of catalysis by means of the metals in their different
forms and especially their decomposing effect on hydrogen peroxide has
been the subject of a great deal of very thorough experimental investiga-
tion. Since Bredig published his work on the preparation of colloidal
solutions by electrical' means, the study of catalysis has had an additional
impetus, and our knowledge of the subject has been greatly increased
by the investigations of this author and his co-workers, Miiller v. Bemeck,*
K. Ikeda,' W. Reinders,' Fortner,^ Teletow' and v. Antropoff.*

It has been conjectured by several authors that the dissolved of chemi-
colly bound oxygen in the metal phase plays a necessary part in the
catalytic decomposition of hydrogen peroxide. The experiments of
Haber and S. Grinberg,*® Euler,*^ and of Engler and Wdhler" indicate

'Local.

* Diasertation, Heidelberg (1907).
'Z. angew. Chem., 1898, p. 951.

* Z. physik. Chem., 31, 258.

* /W., 37f 2.

* /Wd., 37» 323-

' B«., 37» 798.

* Z. Elektrochemie, xa, 581.
•Ibid.

*• Z. anorg. Chem., 18, 37.

" Oefers af. K Vetenskaps Fdrhandl., 1900, p. 267.

*' Z. anorg. Chem., a^, 1.

196

9* B* wPJSAR*

that the presence of oxygen is necessary before many of these catalytic
processes take place.

Bredig and Miiller v. Bemeck^ attempted to investigate the influence
of the dissolved or chemically bound oxygen on the catalytic decom-
position of hydrogen peroxide by subjecting their colloidal solutions to
reduced pressure before using them as catalyzers. They were not able,
however, to discover any marked difference in either the rate of reaction
or in the reaction constant as a result of this treatment. Bredig/ there-
fore, concluded that the decrease in catalytic effect observed by Geraez,
after he had boiled his colloidal solutions, or had heated his platinum
black, was caused by some change that the metcU had undergone, rather
than by the loss of oxygen. This assumption of Bredig's agrees very
well with the experimental facts, at least in the case of colloidal solutions,
because of the coagulating effect of the boiling. Again, if the loss of
oxygen were the only cause of the observed retardation in the rate, the
colloid must very quickly regain its original activity because oxygen is a
product of the decomposition of hydrogen peroxide. It is doubtful if this
is the case.* It might be pointed out, however, that if a compound of plati-
num and oxygen is an essential factor in the catalysis it docs not necessarily
follow that this compound will be destroyed by pumping out the dissolved
oxygen unless it has an appreciable dissociation pressure and a rapid rate
of decomposition. Bredig's experiments are not conclusive, they indicate
that the dissolved oxygen does not play a vital part in the reaction. This
latter statement is in accordance with the results given in this article.

Liebermann and Genersich* endeavored to investigate the r61e of the
oxygen in the decomposition of peroxide by boiling their colloidal solu-
tions or by bubbling gas through them in order to free them from oxygen.
They then treated different portions of the catalyzer thus prepared with
oxygen, hydrogen and nitrogen gas respectively and compared the effective-
ness of each portion with that of the original colloidal solution. Un-
fortimately we are not able to conclude very much from their experi-
ments because they worked with much too concentrated solutions (about
3 per cent, hydrogen peroxide). Their reaction mixture must have
been saturated with oxygen as a product of the catalysis itself before
they could make a single titration to learn the progress of the reaction.
Their gasometric method also does not give us the necessary information,
for the difference in height of the two columns of mercury is not an accurate
measure of the amount of decomposition during the first few moments
of the reaction because of supersaturation. But it is precisely this
amount that we wish to know.

* Loc. cit., p. 336.

' Liebermann and Genersich (Arbeiten aus den hy. Institut der Universit&t Buda-
pest; Archiv. ges. Physiol., 104, 139) state that their platinum solutions regained their
original activity in a few instances.

CAXAI^YTIC DECOMPOSITION OF HYDROGEN PEROXIDE. 1 97

The object of the work described here was to ascertain experimentally
the effect of increasing the concentration of oxygen on the catal>i:ic de-
composition of hydrogen peroxide. This was accomplished by means
of the apparatus described on page 198.

Experimental.

Preparation of the Solutions. — ^The colloidal solutions used in this work
ireie prepared by Bredig's well-known electrical method. The pure
metals were obtained from Heraeus, Hanau. The coUoidation of platinum
tookplaceinneutral,thatof all the other metals mentioned in dilute M/iooo
alkaline solutions as free as possible from carbonate. The concentrations of
the platinum, palladium, iridium and gold solutions were determined by
precipitating the metals with hydrogen sulphide in dilute sulphuric acid
solution, heating the sulphides to constant weight and weighing as pure
metal The colloidal silver solutions were analyzed by dissolving the metal
in concentrated nitric acid, evaporating almost to dryness, diluting and
titrating with a potassium sulphocyanate solution according to the
method of Volhard. The hydrogen peroxide solutions were prepared
by diluting a 30 per cent, preparation from Merck with pure water.

The water used in the preparation of the solutions was purified by
Hulett's method.^ The conductivity of the water thus prepared was
not determined because these measurements are not suflSciently delicate
to be regarded as a safe criterion for the absence of many of the so-called
poisons for colloidal platinum; furthermore, many of these poisons are
Don-electrol3rtes.

Cleansing of the Glass Apparatus. — Experience teaches that reproduc-
abk results cannot be obtained with metal catalysis unless the glass
with which the solutions come in contact has been cleaned with extreme
care. New vessels should be boiled for some time with concentrated
hydrochloric add and then with pure water. Vessels not in use should
be carefully cleaned and placed in pure water. Immediately before they
are required they should be well steamed out, using pure water to generate
the steam.

In spite of the most careful manipulation, however, one often obtains
irregular results that cannot be repeated. This is especially the case
where the experimental 'difficulties preclude the possibility of preventing
occasional poisoning effects. If many irregular results were obtained
with a particular set of solutions they were thrown out, the glassware
thoroughly cleaned and new solutions made up.

Description of the Bomb. — ^The experiments under pressure were carried
out in a bomb made of "Rotguss" (see Fig. i , a, p. 198). The ring c was for
screwing the lid b on to body of the bomb a. The manipulation is much
' Z. phyak. Chem., az, 297.

mote difficult when the ring c and the lid b consist of one piece as in the
case of bomb I> Fig. 3, p.
204. The compressed gas
was conducted into the
bomb through the metal
tube e. The detachable
tube / was connected with
the glass tube g by a piece
of rubber tubing p. The
glass tube g dipped into the
reaction mixture contained
in the glass vessel d and
served as a siphon throu^
which small portions of the
liquid in d could be taken
out at will and analyzed.
The short metal tubes t and
q were soldered on to the
lid. The bomb was made
gas-tight by means of a steel
ling, m, and two rings of soft
lead, /, which were each
about 2 mm. thick. This
arrangement is not entirely
satisfactory and difficulty
was experienced in keeping
the bomb from leaking at
high pressures. The original
plan of the author seems
much preferable, but owing
to a misunderstanding it
was not carried out by the
firm that made the bomb
(see Fig. 1,6). Here there is
only one lead ring employed
and no plain surface between
the body of the bomb and
the hd. The under side cA
the thread ic on the bomb
and the upper side of the
thread on the ring c should
be right-angled and not bev-
Fig. i(a). eled as it is ic the figure.

CATALYTIC DECOMPOSITION OF HYDROGEN PEROXIDE. 199

The bomb was placed in a thermostat and partially filled with water
in order that the reaction mixture might rapidly come to and remain

at the temperature of the thermostat. The tube e was connected with a
manometer by means of a bent metal tube 2, Fig. 3, p. 204 and the manom-
eter was in turn coimected with an ordinary oxygen bomb. By means
d this arrangement a pressure of i-ioo atmospheres of oxygen gas could
be obtained at will in the reaction-bomb. It was found possible to open
the valve i so slowly that the liquid ran out of / drop by drop or in a
gentk stream. At very high pressures (100-250 atmospheres) the liquid
came out as a spray, owing to the expansion of the dissolved gas.

Blank experiments without a catalyzer proved that a small portion of
the hydrogen peroxide was decomposed on being forced out through the
metallic tube /. This difficulty was overcome by coating the inside of
the tube with paraffin which was accomplished by filling the tube with a
saturated alcoholic solution of paraffin and then sucking air through.
It was found necessary to renew the coating every two or three days.
These precautions prevented the decomposition of the hydrogen peroxide
even when the solution was allowed to stand some time in the tube.

Analytical Method. — The progress of the reaction was followed by the
method so often employed by Bredig and his co-workers, viz., by titrating
the still undecomposed hydrogen peroxide in a known volume with a
dilute solution of potas^um permanganate (0.24 gram per liter) after
the addition of a few cubic centimeters of dilute sulphuric acid. Occa-

sionally it was found necessary to add a drop of manganous sulphate to
start the reaction.

Operation under Pressure (i-ioo Atmospheres). — The colloidal solu-
tions were diluted to the desired concentration and kept at 35° for at
least twelve hours before the experiment because very irregular results
are obtained, if freshly diluted solutions are employed for catalytic pur-
poses. The hydrogen peroxide solutions were also warmed to 25° be-
fore the mixing took place.

The desired amounts of the different solutions were now placed in the
reaction vessel and thoroughly mixed by stirring with a glass rod. Two
cc, were immediately pipetted out and titrated. In many cases, how-
ever, the first titration was made after the pressure had been applied.
The lid was then placed on the bomb and quickly screwed down. The
bomb was connected with the source of pressure as quickly as pos^bk
and the pressure slowly applied. In about ten minutes after the first
titration the valve i was slowly opened and 5 to 10 cc. of the liquid were
allowed to run out before the second sample for analyds was collected
in order that no liquid that had remained in the tube / might be titrated
and possibly cause an error in the result. The sample thus obtained
was violently shaken to free the liquid from ^s before the usual 2 cc
were measured for titration.

The compressed oxygen used was that generally employed for analytical
purposes. In order to ascertain if it contained anj-thing poisonous to
colloidal platinum, the gas was bubbled through a hydrogen peroxide
solution for one-half hour. No difference in the rate could be detected
resulting from this treatment. One dare not bubble gas through the
colloidal solutions. ' In later experiments the possibility of poisoning
by the compressed oxygen was reduced very greatly by the use of the
glass vessel described on p. 202,

In order to compare the results obtained under pressure with those
under ordinary circumstances, experiments with the saiBe solutions in
the same vessel on the same day were carried out in the bomb at atmos-
pheric pressure. The solutions were mixed as already described and
the first titration made as usual. The glass vessel containing the reaction
mixture was then placed in the bomb and the lid screwed on. In about
ten minutes a pressure of 1-3 atmospheres was applied and a sample
run out as before. The pressure in the bomb was then relieved.

Catalysis with Colloidal Platinum. — In the course of this investigation
irregular results were occasionally obtained and the reactions were con-
sidered poisoned or accelerated if they could not be repeated under the
same experimental conditions. More than 200 measurements were
made with colloidal platinum, a few of which are given below.
> Liebermann and Geneincb, Loi:. cU.

CATALYTIC DECOMPOSITION OF HYDROGEN PEROXIDE.

20 1

In this article:
i « time in minutes;

Titer = number of cubic centimeters of the potassium permanganate
solution used to titrate the undecomposed hydrogen peroxide.
P = pressure in atmospheres;

Cone. = concentration; /

Atm. = atmospheres;
Exp. = experiment;
k = reaction constant calculated from the integrated formula,

A — Xi

where A is the titer at the beginning of the reaction, x^ and Xj the amount
of hydrogen peroxide decomposed at the time t^, /j expressed in cubic
centimeters of potassium permanganate solution. In every case, k
bas been calculated from pairs of consecutive measurements. The con-
centration of the metal in the reaction mixture is expressed in gram atoms,
that of the hydrogen peroxide in gram mols. per liter.

TABLE I.

Expt. 63(a).
P = 70.

Ooacentrations. /. 0.4343 k.

Expt 63(3).
P= I.

/. 0.4343 >fc. Concentrations. /.
30.5 0.0028 29 0.0023

Bxpt. 131.

P = 70.
0.4343 *.

Bxpt. 133.
P=I.

/. 0.4343 k.

Pt-~—
145,000

"*"• 30

62.5 0.0029 ^-5 0.0024 Ft —

76.5 0.0029 76 0.0024 H,Os

100,000
I

10 0.0176 10

0.0169
0.0187

0.0150
0.0186

35 0.0174

Bxpt. 64.
P = 70.

Bxpt. 65.
P= I.

DapHcates of

63 (a) & (6) 39 0.0024 29 0.0021 Pt<

77 0.0027 67 0.0025

Bxpt. 135.
P = 70.

Bxpt. 136.
P = i.

2,000,000

109.5 0.0027 III. 5 0.0027 H,0,—— -

Bxpt. 80. Bxpt 81. Duplicate of

P = 70. P = I.

236

382

0.00028
0.00023
0.00020

0.00047
0.00028

Bxpt. 137.
P= I.

Pt *—

150,000

^" 30

29 0.0040 29 0.0037

74.5 0.0059 63 0.0047

126.5 0.0042 106.5 0.0052
167 0.0051

126

118
359

0.00022

O.OOOII

Bxpt. 119.
p = 70.

Pt-

^A-- 34-5 0.0067

Bxpt. 120.
P=- 1.

18 0.0057
40 0.0061

56.5 0.0070 63 0.0065
78.5 0.0069 90 0.0063

E. B. SPEAK.

[though a gas does not dissolve very rapidly in a motiooless liquid
nay safely assume that the concentration of the oxygen in the mixture
very much greater towards the end of the reaction in the experiments
:r high pressure than in those at a pressure of one atmosphere, yet
constants in the two cases are almost identical. Quantitative ex-
tents with regard to the amount of dissolved oxygen will be given

de objection might be raised to the foregoing results that the coni-
sed oxygen contained a poison or an accelerator for the reaction
se influence exactly counterbalanced the effect of the increased con-
ration of the oxygen. This posdbility was very much reduced by
use of the glass vessel described below. By this arrangement tlie
ice of contact between the gas and the reaction mixture was reduced
ery small dimensions, so that not very much of the outer gas conld
into the reaction mixture during the experiment, while the chemically
oxygen that had been generated by the catalysis could get out of
liquid only under high pressure. By beginning with the correspond-
concentration of hydrogen peroxide in the reaction mixture it was
possible to obtain any desired concentration
of oxygen, while the outer compressed fas
served to keep the generated oxygen from
escaping from the liquid.

a (Pig. 2) is a glass vessel (ca. 3 cm. deep)

that exactly fitted into the vessel in which

the reaction was carried out. The dphon

tube g (Fig. i) fitted just as neatly into

tube h (Fig. 2). Vessel a was placed on the

surface of the reaction mixture and was free

to sink as the liquid was siphoned out for

'" *■ analysis. If the glass was free from grease

liquid rose between the sides of the inner and outer vessels to the

;r edge of vessel a. A series of results obtained by this method is

n in Table 2.

TABLE 1.

ga 0.0094 90
' Preparation several months old.

60. '

43«». CODMD

ntlODB

^m...>.

Eipt. 19J.

.0071 Pt— -

XI,OO0

0.0055

O.O03S

.0072 HA-

■i

0.0060

0.0037

.ooSi
.0076

60
137

0.0064
0.0061

O.cxHi

Bxpt. 354.
P=- 1.

Bxpt. 255.
P-70.

10 0.0059

0.0054

30 0.0074

0.0060

60 0.0074

0.0073

120 0.0078

177

0.0089

CATALYTIC DSCOMPOSITION OF HYDROGEN PBROXIDE. 203

Bzpt. 356. Bxpt. 357.

DopUcates of P = i. P = 70. Duplicates of

254*255 II 0.0050 10 0.0055 252 & 253

30.5 0.0065 30 0.0066

60 0.0070 60 0.0080

117 0.0064 139 0.0094

188 0.0053 169 0.0122

If all the oxygen generated in Expts. 255 and 257 remained in solu-
tion the pressure toward the end of the reaction must have been 120
atmospheres. We may, therefore, safely assume that the partial pressure
of the free oxygen in solution was 30 to 40 times as great as in the corre-
sponding experiments, 254 and 256, yet we see that the rate was practically
the same in both cases.

The influence of still greater oxygen concentrations (100 to 200 atmos-
pheres) was investigated by the aid of the apparatus shown in Fig. 3.
As the author could not obtain the use of a pump suitable for the pur-
pose, the high pressure was obtained by allowing liquid oxygen to evapo-
rate, a is the reaction bomb already described. It was placed as usual
in a thermostat and was coxmected with a manometer for high pressures
(i-iooo atmospheres). The manometer was in turn connected with the
bomb b. If now a calculated amount of liquid oxygen was put into
bomb b and allowed to evaporate, the corresponding pressure was ob-
tained in the reaction bomb a.

It was impossible to pour liquid oxygen into the bomb without cooling
it before hand if the liquid came into contact with the metal, because the
heat capacity of the latter is so great that the liquid evaporated in a
very few minutes. Cooling the bomb before introducing the oxygen
was too expensive and impracticable. This difficulty was overcome by
pasting strips of asbestos paper on the inner walls of the bomb and blow-
ing a glass bulb inside it. The narrow space between the glass and the
metal was filled with zinc oxide. With this arrangement the liquid
oxygen (ca. 400 cc.) evaporated completely within thirty minutes, conse-
quently, the high pressure was obtained quickly enough for the purposes
of the experiment.

Details of the Experiments at High Pressure. — The experiment was
begun in the usual manner and a pressure of 70-80 atmospheres was
obtained in the reaction bomb from the usual source. The liquid oxygen
was next poured into the bomb 6, and the whole apparatus connected
as in the figure. As soon as the pressure in the bomb, b, had risen to
70-80 atmospheres, bomb a was connected with the source of pressure by
openmg valve d (Fig. 3). The pressure rose to its maximum in about
thirty minutes but occasionally fell somewhat towards the end of the ex-
periment because the apparatus was not always gas-tight.

An analysis of the gas obtained from the liquid oxygen gave a content

204

E. B. SPEAR.

o^ 93-5 per cent, of oxygen.* The remainder consisted principally of
nitrogen. The results are given in Table 3.

Fig. 3.

In Experiments 259 and 267 an open glass vessel was used in bomb
a, while in Experiment 268 the vessel described on p. 202 was used. We
see quite cleariy from these results that a high concentration of oxygen
has no perceptible effect on the catalysis.

Expt. 259.

p = 160 »-^ 140.

Concentrations. /. 0.4^3 Jk.

^ I

Pt — 10.5 0.0042

200,000 ^ ^

32 0.0050

63.5 0.0049
107 0.0052 120.5 0.0047
193 0.0050 197.5 0.0049
256 0.0052 ...

TABLE 3.
Hxpt. 263.
P = 70-
/. (*.4343 ft. Concentrations.

0 0.0026 Pt =

^ 350,000

30 0.0047 H,0,=«|

60 0.0050

Bzpt. a66.
P= 1.

0.4343*.
0.0020

30.5 0.0026

67 0.0029
151. 5 0.0025
346.5 0.0021

Bxpt. 367.
P = aoo »-► 100.

/. 0.4343 *.
12.5 0.0020

0.0028
0.0030

32
62

174
269

0.0026
0.0023
0.0019

Pt-

350,000

HA-*

Bxpt. a68.
P = aoo»-»-i6o.

Bzpt. 269.
P= I.

0.0025

26.5 0.0024
60 0.0030

120 o 0022 120

231 0.0019

30
60

0.0033

0.0036
0.0032
0.0022

The effect of increasing the concentration of the oxygen was tried
where the k decreased from the beginning of the reaction. In such cases
it is generally assumed that some one of the solutions contains a poison
for the catalysis. Experiments 270 and 271 were carried out with the
same solutions 8 days later than 260-3.

TABLE 4.

Bxpt. 260.
P = I.

Bxpt 261. Bxpt. 262.
P = 70. P = I.

Bxpt. 263.
P = 70.

Concentrations. /.

0.4343*-

/. 0.4343 il. Concentrations. /. 0.4343 A.

i' 0.4343 k.

^ I

0.0075

10 0.0069 Duplicates 10 0.0047

9 0.0026

x^^ 10

200,000

HA-j;, 30

0.0068

29 0.0060 of 260& 261 32.5 0.0037

30 0.0047

0.0055

60 0.0057 121. 5 0.0036

60 0.0050

120

0.0050

117. 5 0.0052 207.5 0.0024

120.5 0.0047
197.5 0.0049

CATALYTIC DI^COMPOSITION OF HYDROGEN PEROXIDE. 205

Bxpt. 370.
I ^=-^-

Erpt 271.
P = 70.

Pt-— i— 10
200,000

0.0043

0.0044

H|0.-i 31

0.0042

0.0040

0.0043

0.0042

137

0.0035

126

0.0038

215

0.0032

• •

Many authors have thought that the platinum becomes more active
through the presence of the hydrogen peroxide or of the oxygen. A
glance at the tables given above will reveal the fact that the k often
increases in value until the very end,* while in many experiments (usually
with different colloidal solutions) k goes through a maximum.^ A large
number of experiments not given here were carried out and in every
case no relation was found between the amount of dissolved oxygen and
the rate of reaction, or the course of k.

Experiments with Colloidal Gold under Pressure. — ^The gold catalysis
of hydrogen peroxide at ordinary pressure has been studied by Bredig
and Reinders-' Experiments under high pressures of oxygen gave similar
results to those obtained with platinum.

The reaction mixture was made alkaline because the rate in neutral
or dilute acid solutions is very slow. The alkaline solution was made
according to Paul's method by dissolving freshly cut sodium in water
that had been boiled to free it from carbon dioxide.

TABLE 5.

Exp. 155.
P = i.
Coocentrationfl. /. 0.4343 Jk. '•

Exp. 157.
P = 8o
0 4343*-

Concentrations
Ai, '

Kxp. z68.
P=x.
0.4343 *•

Bxp. 169.
P=8o.

0-4343 *.

M
NaOH— — 40 0.0025 40

0.0027
0.0027

Am ^ —■ ■■•

200,000

NaOH-^

20
40

0.0045
0.0045

0.0040

•

0.0051

H|Q,-yj 70 0.0033 70

0.0026

«'«'-i

0.0079

0.0056

100.5 0.0046 95 0.0044
96.7=tA/a ioo«tA/2

0.0045

52.3-tA/2

■ •

• •

55-6 = tA/a

Au- ' Expt. 154. Expt. 160.
400,000 P - I. P - 80.

Au- '

400,000

Bxpt. 162.
P= I.

Bxpt. 163.
P= 1.

NaOH« 19.5 0.0020 20

0.001

, NaOH-^^^

0.0021

0.0019

H/),-- ~ 70 0.0027 50

0.0024 H,0,= --

O.OOII

•

Duplicates
of 162 & 3
under pres.

40
60

O.OOII

Bxpt. 164.
P = 6o.

0.0018
0.0012
0.0016

]

60
81

Bxpt. 163.
P-80.

0.0019
0.0015
0.0013

* See Experiments 65, 80, 81, 121, 250, 253, 254, :

* Sec Experiments 232, 237, ^66, 267, 268, 269.
'Loccit.

* Partly coagulated.

80
255,

0.0013 . .

257, 244, 245.

• • •

Experimeiits with colloidal gold more dilute with respect to alkali
ban — : gave very irregular results, very probably because of the presence
)f a varying amount of carbon dioxide. It is very difficult under com-
>licated experimental conditions to prevent errors from this source.

As Bredig and K. Ikeda have pointed out, the metal catalysis of hydro-
gen peroxide in alkaline solutions does not follow the simple law for first-
uder reactions. Nevertheless, the values of k given in the table above
lave been calculated according to this law because they serve as a method
if comparison if the same intervals of time are considered. Wherever
t was practicable the time tj^j^ necessary to decompose one-half the hydro-
;en peroxide has been calculated by interpolation.

Experiments with CoUoidal Palladium under Pressure.— Tht decom-
wsition of hydrogen peroxide by colloidal palladium under ordinary
)ressure has been studied by Bredig and Fortner. The reaction under
ligh pressures of oxygen has been followed by the author in precisely
he same manner as in the case of gold. The results are very regular
ind show that the palladium solutions used were very much more active
han those of the other metals having the same concentration.
TABLE 6.

Hipl. 175. Bxpt. ij«. Expl. 177 topi. ITS

ooceDtTalioni. (, atMl^- '■ 0-4M3*. ConcentrWiom. I. 04J43*. /. (MMjt-

'd — -----^ 30 0.0073 20 0.0069 Duplicates of 30 0.0067 20 o.ooSi
!/>,— -~ 40 0.008843 0.0093 I75&t76 40 0.0104 40 0.0107

laOP— — 60 0.0093 60 0.0093 60 o.ojio 60 a.0134

38-tA/i 38.7-tA;2 36-i~tA;. 34-4-tA|i

Eipl. 179. Expt. iSo.

p = I. P - Bo.

hiidicates ao 0.0070 ao 0.0073
f 175&6 40 0.0095 43-5 0.009a
60 0,0100
37-4-tA). 38,i-tA(j

Investigation with Colloidal Iridium. — The catalytic properties of colloidal
ridiutii have not yet been carefully worked out.' It has been the object
if the following experiments to go into the question only in so far as it
ras necessary in order to ascertain the effect of high concentrations of
ixygen on the catalytic decomposition of hydrogen peroxide by colloidal
ridium.

The electrical colloidation of iridium is much more difficult than that
if the other metals used In these experiments. Iridium is brittle and
he wires often fuse together because of the large amount of current
equired (14-16 amperes, 70 volts).

* Experiments on this problem are being earned out by Dr. Brossa, in Hdddbefg.

CAXAI^YTIC DECOMPOSITION Olf HYDROGEN PEROXIDE. ^OJ

No experiments were carried out with iridium in add or neutral solu-
tion. Those in dilute alkaline solution show the remarkable fact that the
reaction very nearly follows the law for first-order reactions. No very
great weight should be laid on this result until the subject has been
thoroughly investigated, because all the following experiments with
colloidal iridium were carried out with a single preparation.

TABLE 7.

Hxpt. 182. Ezpt. 183. Ezpt. 184. Bxpt. 185.

P = I. P = 70. P = 70. P = I.

Concentrations. /. 0.4343/1. /. 0.4343 /fc. Concentrations /. 0.4343 ik. /. 0.4343 /k.

20 0.0054 20 0.0048 Duplicates 21 0.0062 21 0.0049

I »

60fOOO
Na0H»— 40 40 0.0052 of 183 & 182 60 0.0049 40 00052

HjO,«— 60 0.0060 60 0.0058 60 0.0051

56.3-tA/2 60— t A/2

n..«i'.^*». Expt. 186. Expt. 187. Expt. 189. Expt. 190.

Duplicates p = 70. P = X. P = I. P = I.

ofi83&i82 20.5 0.0043 20.5 0.0049 Ir=r— — ■ 20 0.0018 20 0.0018

42 0.0050 40.5 0.0045 NaOH = 49.5 0.0019 51 0.0019

60 0.0062 60 0.0050 H,0,«— 93.5 0.0020 104.5 0.0022

6l=*tA/2 6o=»tA/2 147 -O 0.0022

Ir-r- ^ Expt. 191. Expt. 192. It =2 Expt. 193. Expt. 194.

60.000 p = i: p*= if 60,000 p = I. p = u

I 2

NaOH— -- 20 0.0023 20 0.0024 NaOH— — 20 0.0021 20 0.0021

Od • Oo

Bfit^— 40 0.0025 40 0.0026 H,0,—— 50 0.002Z 50 0.0021

85 0.0025 80 0.0026 80.5 0.0020 80 0.0021

169 0.0028 166 0.0028

Experiments with Colloidal Silver under Pressure, — Macintosh'
measured the rate of decomposition of hydrogen peroxide by colloidal
silver and found that the constant calculated from the equation for
first-order reactions decreased, whereas it increases in the case of
gold, platinum, etc. He explains this phenomenon by assuming that
colbidal silver is slowly dissolved by hydrogen peroxide. The author's
experiments verify Macintosh's results. It was found necessary to work
in dilute alkaline solutions because the silver dissolved very rapidly even
in neutral solutions of hydrogen peroxide. For instance, the decom-

. . M I

position of a — solution of hydrogen peroxide by a gram atom

solution of colloidal silver became immeasurably slow after 20 minutes.

' Professor Bredig has very kindly informed the author that subsequent investiga-
tion in his laboratory has shown that the method of analysis employed here for iridium
gives faulty results. These figures do not, therefore represent the true concentrations.

' Partly coagulated.

" J. Phys. Chem., 6, 15.

B. B. SPBAR.

TABLE 8.

Bipt. 131.

ipl, JJ». Ei

"iJl*-

0.4J43*. Coocentnitloni. (.

o.lMl *-

0.0189

0.0181 NaOH-^ 15.5

O.OI3S

.5 0.0165

0.0151 HV3,— "^ 36.5

0.0094

O.Ot36

103

0.0119 755

0.0088

Eip«.«S.

Bipl. 316.

36.5 o.moi 36 0.0118

76.5 0.0076 76 0.0097

Theoretical Consideratioiis.
make the reasonably safe assumption that the reaction takes
a heterogeneous system let us consider under what conditions
gen concentrations would appreciably affect the catalysis.
ie the reaction under pressure goes completely as far as we can
e by ordinary analytical means, any chemical equihbrium be-
I'drogen peroxide, oxygen and water may be left out of COD-
1. It has been suggested by several authors that the oxygen
L intermediate compound with the metal or, perhaps, a solid

Neither of these hypotheses preclude the existence of a layer
ised oxygen on the surface of the solid phase.
■e, then, the first step of the reaction is represented by the equa-

I. >O+ttM=M„0y,

,0j. is either a chemical compound or a solid solution and that
is followed by a second according to the equation

II. M„O^+j'H,0,=«M+>O,+j'H,0.
and Brunner' have shown that the measured rate will be affected
te of the chemii-al reaction in heterogeneous systems only when
c is slow in comparison to the rate of diffusion of the substance
ag decomposition. In the light of these considerations, pressure
ve an effect on the measured rate of reaction when:
e activity of the platinum- oxygen phase in the second step of
ion changes with increasing concentration of oxygen,
e rate with which the chemical compound or sohd solution M,0^
st step of the reaction is formed, varies with the concentration
ly^k. Chem., 51, 95 and 494.

SBPAitATioN OP moK FkoM tm>njM. 209

of the oxygen. As the experimental results given in this article prove
definitely that pressure has no measurable effect on the rate, we are forced
to conclude either, that the two above cases do not exist, or that the
chemical reactions are rapid in comparison to the rate of diffusion of the
hydrogen peroxide. The latter hypothesis agrees very well with the
experimental results obtained by J. Teletow.*

Pressiu'e would also have no appreciable effect on the reaction if an
active platinum compound were formed directly from the union of plati-
mum and hydrogen peroxide and not from platinum and oxygen, even
if the rate of union were not rapid.

Summary.

The chief results of this article are: A method has been worked out
whereby rates of reaction may be measured under high gas pressure.

It has been experimentally determined that the catal3rtic decomposi-
tion of hydrogen peroxide by colloidal solutions of platinum, palladium,
iridium, gold and silver is unappredably affected by increasing the pres-
sure of oxygen gas above the reaction mixture from i to 200 atmospheres.

This investigation was carried out in the chemical laboratory of the
University of Heidelberg during the years 1905-6, under the direction of
Prof. Bredig to whom my sincere thanks are due for friendly and valu-
abk advice.

Massachusbtts iNSTrruTB OF Tbchivoloot,
October 9, 1907.

A METHOD FOR THE SEPARATION OF IROIT FROM INDIUM.

By p. C Mathbsb.
Received December 16, 1907.

One of the most difficult steps in the purification of indium is its sepa-
ration from iron. Winkler' obtained a separation by the fractional
precipitation of the stdphide, the indium sulphide being less soluble than
the iron sulphide. A more satisfactory method was devised by Bayer,'
who treated a solution of the mixed chlorides with sodium sulphite.
Basic indium sulphite is precipitated from this solution upon boiling.
The precipitate, after filtration, was dissolved in a solution of sulphurous
add and basic indium sulphite was again precipitated by boiUng. This
solution and repredpitation was repeated several times to completely
purify the indium. The Bayer method was tested in this laboratory,
but gave unsatisfactory results. Weselsky* treated the chlorides of
indium and iron with sulphur dioxide or sodium thiosulphate and then

^ Dissert., Heidelberg, 1906.

» J. pr. Chem., 94, i (1865).

» Bayer, Licb. Ann., 158, 372 (1871).

* J. pr. Chem., 94, 443 (1865).

0 p. C. UATHERS.

ith barium carbonate, which precipitated indium hydroxide, together
ith traces of iron and zinc. Meyer' treated the material that contained
dium and iron with sodium carbonate until neutral and then witb a
ilution of potassium cjranide until the precipitate that first formed
as redissolved. The solution thus formed' was diluted with lo volumes
water and was boiled. This decomposed the potasaum indium cyan-
e and indium hydroxide was precipitated while the iron remained in solu-
}n as potassium fern- or ferrocyanide. This method was found, upon
ial, to give fair results. The indium, however, is not completely pre-
pitated by boiling, considerable amounts of it being found in the fil-
ate. The precipitate is very gelatinous, difficult to wash, and often
isses through the filter paper before the washing is completed. For
lese reasons the method is not satisfactory. Some quantitative results
itained in this laboratory with the Weselsky method follow;

dium Qilde predplUtfd
h ammoalum hydrazide.

0.0320 gram O.0O33 gram or 6.8%

0.0545 gram 0.0528 gram or 96.8% 0.0014 gram or 3.5%

Dennis and Geer* have proposed the remoi^l of the iron by extracting
.e ferric sulphocyanate with ether. This method has been applied to
le removal of iron from nickel, cobalt, copper, aluminum, etc. Indium
Iphocyan&te, however, is quite soluble in ether and for this reason
dium is always present with the iron in the ether extract. The labor
volved in recovering the indium thus dissolved constitutes a serious
'awback in this method, which is otherwise very satisfactory and yields

1 indium that is free from iron.

Several other methods have been suggested for the separation of iron
id indium, but those cited above seem to have been the ones most gen-
ally used and to have been the most satisfactory.

The method here proposed for the removal of iron from indium is the
edpitation of iron from nitroso-/3-naphthol. This reagent* quanti-
tively precipitates cobalt, copper, and iron, but does not precipitate
uminum, lead, zinc, or nickel. Experiment showed that indium also
is not precipitated, and this led to the development of the following
ethod of separation:

A solution of the indium chloride or sulphate containing a small quan-
:y of iron was evaporated to a volume of 20 to 25 cc., was neutralized
ith ammonium hydroxide and then an equal volume of 50 per cent.
etic acid was added. The iron in these solutions was precipitated
ith nitroso-^-naphthol dissolved in 50 per cent, acetic add. The solu-

' Meyer, Lieb. Ann., 150, 137 (1869).

* This Journal, 36, 437 (1904).

'Ilinsld and Knoire, Bei., 18, 173S (1885); Knorre, Ber., 20, 383 (1SS7).

SOME NEW COMPOUNDS OF INDIUM. 211

tion may be either hot or cold when precipitated, but it should stand
several hours before filtering, and should be cold when filtered. The
residue is washed with 50 per cent, acetic add and finally with water.
Before using this method the bulk of the iron must be removed by some
other method. Since the precipitate of iron with nitroso-/?-naphthol
is very bulky, 0.05 gram of iron is about the maximum quantity that
can easily be handled on a 12 cm. filter paper. The indium precipitated
by electrolysis from a solution containing indium sulphate and quite
large amoimts of ferric sulphate and strongly acidulated with sulphuric
add, contain^ only small amountsi of iron. This metal can then easily
be freed from the iron that it still contains by predpitating the iron with
iutroso-/9-naphthol. Colorimetric analysis of the indium, after the pre-
dpitation of the iron by this method, showed that the content of iron was
very low. This small amount of iron^ could easily have been introduced
into the solution during evaporation by the dust from the air. Quan-
titative determinations of the iron remaining in the indium gave the
following results:

Indium
oxide taken.
Grmm.

Ferric
oxide added.
Gram.

Nitrosc^

naphthol used.

Gram.

Ferric oxide still
present in the indium oxide.
Gram.

0.3148

0.0328

5.0

Less than 0.00005

0.1738

0.0262

4.0

" " 0.00005

0.2184

0.0197

" " 0.00005

0.3061

0.0262

2.5

" " O.OOOI

0.2142

0.0262

2.5

Very faint red

Some indium remains in the residue with the iron and a second pre-
cipitation will not remove all of it. The indium in these residues can
easily be detected with the spectroscope. The total quantity of indium
that is lost in these iron residues is, however, quite small, as is shown
by the following data:

Ferric oxide obtained from ludium oxide in ferric oxide in
Indium Ferric First Second First Second

oxide taken. oxide added, precipitation, precipitation, precipitation, precipitation.
Gram. Gram. Gram. Gram. Gram. Gram.

0.3061 0.0262 0.0286 0.0273 0.0024 O.OOII

0.1520 0.0262 0.0283 0.0273 0.002X O.OOII

0.2142 0.0262 0.0281 0.0266 0.0019 0.0004

CORMKX UNIVBItSITy,

1907.

SOME NEW COMPOUNDS OF INDIUM.

By p. C. BCathbrs and C. G. Schlitbdbrbbro.
Received December 16. 1907.

This paper describes the preparation and properties of four new com-
pounds of indium: the perchlorate, the iodate, the selenate, and the
caesium-selenium alum.

^ Stokes and Cain, This Journal, 29, 409 (1907).

^. C. MATHBKS AND C. G. SCHLUBDBKBBRG.

Hum Percklorate, In(C10,)r8H,O. — Metallic indium was dissolved
ute perchloric acid. This solution was evaporated upon a hot plate
concentration such that when cooled in a mixture of ice and salt,

crystals were formed. The beaker was then placed in a vacuum
ator containing concentrated sulphuric add where the crystalliza-

continued in a satisfactory manner. The crystals were rapidly
I with a small amount of water, dried on filter paper, and then placed
; vacuum desiccator for a short time.

; metallic indium that was used had been deposited electrolytically
;hen fused in a charcoal cnidble in a current of hydrogen. This
I was necessary to remove salts that were occluded from the elec-
.e. The perchloric add that was used was purchased from Schuck-
. It contained a small quantity of sulphuric acid, which was re-
d by the addition of the proper amount of barium hydroxide.

0 different methods were used to determine the indium in the indium
lorate. The shorter procedure consists in heating the sample to
iperature of about 200°, at which decomposition of the perchlorate

place, then evaporating with nitric add, and finally igniting the
e to the oxide. In the other method the indium is predpitated
: hydroxide with ammonium hydroxide and is weighed as the oxide,
econd method is preferable, since it permits determinations of both
idium and the chlorine (ClO^) in the same sample. To determine
hlorine, the filtrate from the indium hydroxide was treated with
cess of sodium carbonate and a few drops of potassium permangan-

It was then evaporated to dryness on a water-bath and heated to
Fusion, this treatment decomposing the perchlorate and yielding a
de. The mass was dissolved in dilute nitric add and the solution
reed from the insoluble manganese dioxide by filtration. The chlor-

1 the filtrate was predpitated and weighed as silver chloride.

Theory lor
lD(C10iVBBtO. Found.

Indium oxide >4-93 34.82 ; 24.73 '• >4-83'

Chlorine 19.08 1905 ; 18.67 l

empts to make direct determinations of the water were unsuccess-

L account of decomposition and the formation of a basic salt with
of chlorine. Samples of the crystals that had been dried for several
in the vacuum desiccator gave higher values for the indium oxide,

md 25.7. These values agree with the theory for In(C10,),.7H,0.

8 evidently due to the loss of water of crystallization, since several

;nt samples that were dried for only a few minutes gave results

howed 8 molecules of water.

ium perchlorate is a colorless, crystalhne, deliquescent compound

>etemuned by Method i.

SOME NEW COMPOUNDS Ol? INDIUM. 213

that is soluble in water and in absolute alcohol, but much less soluble
in ether. It easily forms saturated solutions in water. Basic salts are
precipitated from a neutral water solution when the temperature is
raised to 40°. When heated in the open air, the crystals fuse at about
80®, but do not form a clear liquid. When the heating is carried to a
higher temperature, but not to a red heat, the substance decomposes
with a liberation of chlorine. The residue is insoluble in water, but is
easily soluble in dilute nitric acid, and this solution gives a strong test
for chlorine.

Indium lodaie, In(I08)3. — ^Anhydrous indium trichloride was prepared
and sealed off in a glass tube by a method previously described.^ Potas-
sium iodate was recrystallized twice from water. This sample was anal-
yzed by treating it with dilute sulphuric acid and potassium iodide and
titrating the iodine that was liberated. Taken, 0.0750 g. KlOgi found,
0.0749 g-

Equivalent quantities of the indium trichloride and of potassium
iodate were dissolved separately and then mixed. A precipitate formed
immediately but it showed no crystalline structure when examined with
a microscope. The entire solution was evaporated to dryness upon a
water-bath and the residue was collected upon a Gooch crucible and
was thoroughly washed with cold water. ^ The crucible and contents
were dried over sulphuric acid in a vacuum desiccator. Analyses of
several samples prepared in this way did not g^ve constant results. The
substance was then dissolved in boiling nitric acid (i: 10 by volume).
On evaporation of this solution, the indium iodate separated in imper-
fect, broken crystals whose crystal system could not be determined.
These crystals were washed with water and dried in a vacuum desiccator.
The results of the analyses were as follows:

Theory for Iti(IOt)t. Found.

Per cent. Per cent.

Indium oxide 21.71 21.76 and 21.93

Iodine 59-53 59-19 and 59-30

Water at 125** 0.08

Loss at 160°. . . . (decomposition, brown color) 0.40

The indium was precipitated as the hydroxide with ammonium hy-
droxide and was weighed as the oxide. The iodine was calculated from
the titration with sodium thiosulphate of the iodine that was liberated
when a sample was treated with potassium iodide and dilute sulphuric
add.

Indium iodate is a white, crystalline, anhydrous compound. It is
soluble in 1500 parts of water at 20° and in 150 parts of i : 5 nitric acid

* Mathers, This Journal, 29, 485 (1907).

* This is the method emplojred in his research upon iodates by A. Ditte, Ann.
chim. phys. (6), 21, 145 (1890).

I'. C. UATBBRS AND C. G. SCHLUBDBttBBRG.

All attempts to grow large crystab from these solutions were
:ssful. It is soluble in dilute sulphuric add and in hydrochloric
It the latter causes a decomposition with the liberation of chtor-
he crystals decrepitate when heated with a free flame, become
in color and give off iodine vapor. If touched with a red-hot
ey explode and form a cloud of iodine vapor.
KM Selenate, Inj{SeOJ,.ioHjO. — Indium selenate was prepared
)lvihg indium hydroxide in selenic acid, which had been made ac-

to the following method :

dust known to t>e rich in selenium was heated in a porcelain dish

strong solution of potassium cyanide for several hours. On
;, a clear, amber-colored solution of potassium selenocyanate,
£), was obtained. A current of filtered air was passed throu^
ution for several hours in order to separate any tellurium that
iiave been dissolved with the selenium. A small quantity of a
Mored precipitate that separated was filtered off. The filtrate
sled to about zero degrees and cold, so per cent, hydrochloric
iffident to entirely predpitate the selenium, was added, care be-
en to add the hydrochloric add slowly to avoid an appreciable
lemperature. The selenium is thrown down as a bright red, amor-
predpitate, which was filtered off and thoroughly washed. This
lissolved in a ten per cent, solution of potassium cyanide, treated
r, predpitated in the cold as before, filtered off, washed very thor-

and dried.

selenium was then dissolved in concentrated nitric add and the
1 evaporated to dryness on the water-bath, redissolved in very
hydrochloric add and again evaporated to dryness. The sele-
ioxide thus obtained was dissolved in a little water and added to
; water, through which a stream of chlorine was allowed to bub-
us oxidizing the selenium dioxide to selenic add. After being
eated for several hours with chlorine the hquid was freed as far
ible from excess of that gas by passing through it a current of

solution of selenic add was neutralized with copper carbonate
shly predpitated copper hydroxide. The blue solution of copper
i thus obtained was filtered, concentrated on the water-hath

aside to crystallize. Blue tricUnic crystals of copper selenate
ed. These were recrystallized from water solution and again
;d in water and the solution freed from copper by electrolysing
arinum dish at low current density.' Electrolysis was continued
o test for copper could be obtained with dther ammonium hy-
:, potassium sulphocyanate or potas^um ferrocyanide in small

Metzner, Compt. rend., 137, 54 (i8g8).

A NBW FORM OF COU)RIMBTER. 215

portions of the solution. The solution of pure selenic add thus obtained
was treated with indium hydroxide. Indium selenate crystallized out
in white, easily soluble, hygroscopic crystals. Analysis gave the follow-
ing results:

Theory for Iiit(Se04)s.ioHtO. Pound.

Per cent. Per cent.

Indium oxide 33-o8 32.9 — 33.0

Selenium 28.30 28.3 — 28.2

Indium Caesium Selenate, CsIn(SeOj3.i2HjO. — Caesium-indium-selen-
ate, the alum, was made by crystallizing a solution of caesium selenate
(prepared from caesium hydroxide and selenic add) and indium selenate.

The alum crystallizes in beautiful colorless octahedra, belonging to the
tetragonal system, which are soluble in water, efflorescent in the air
and are of the following composition :

Theory for CsIn(Se04)t.i2HsO. Pound.

Per cent. Per cent.

Selenimn 21.18 21.04

Indium 15.3 15.5

Caesium 17.7 17.8

In the above analyses the selenium was determined by repeated pre-
cipitation with sulphur dioxide gas from a hot dilute solution of the salt.
The predpitate was collected on a Gooch filter, and the filtrate was
boiled down and again treated with sulphur dioxide. It]|was fotmd neces-
sary to repeat this operation several times in order to remove all of the
selenium from the solution. The Gooch filter was dried at 102-5*^
and the selenium was weighed.

The indium was precipitated by ammonium hydroxide as the hydrox-
ide from the hydrochloric add solution of the salt, collected on an ash-
less filter, heated in a porcelain crudble and weighed as lUjOg.

For the determination of the caesium, the selenium was first removed
and the caesitun was then predpitated from the chloride solution by chlor-
platinic add and wdghed as caesium chlorplatinate.

CoKMBu. UNiVBKsrnr,
December, 1907.

A NEW FORM OF COLORIMETER.

By Gborob Steiobr.
Received November 6, 1907.

The many uses to which colorimeters can be put are too well known
to need mention here.

Instruments using the prindple upon which this one is based — ^the
ratio of the thickness of the liquid through, and not the actual dilution
to equal concentrations — are not applicable to all colorimetric deter-
minations. It will be found, in comparing such a solution as is used
in the colorimetric determination of manganese and some other substances,

■»• I*

-• ■^. .. ■

• •• # ' ■

;,•:■.•..■.. t

1- ■ • i

• r. V ' - •

. -rf - - -

•'■ •'.'. i'

•1 .

•4

•1

1 * , . ' • '

II'

» • ■•,

• I.

2l6

GEORGE STEIGER.

that there is a change not only of the intensity of the color, but also of
the color itself, making it impossible to find a point at which two solu-
tions of different concentrations will have the same depth of tint. In
some other cases, as for instance the yellow color of the higher titanium
salts, this principle gives perfect satisfaction.

The instrument to be described consists of two wooden boxes, the
interior portions of which are finished in dead black. In Fig. i, AA is

Fig. 1.

a piece of finely groimd glass, and this should be illuminated with the full
light of the sky. B is a mirror mounted to swing so that light may be
thrown perpendicularly through the hole C.

The second portion of the apparatus consists of a box, as shown in Fig.
2, made with two parallel grooves in the bottom, in which the two glass
cells, CC, can be moved back and forth, and the hole E, which admits
light reflected by the mirror B of Fig. i. These cells are about 15 cm.
long, 2.5 cm. wide,^ and 5 cm. deep. On the bottom of each cell and near
the outside edge is engraved a scale, a convenient unit for which is the
millimeter. FF are glass tubes with mirrors, GG, attached to the lower
ends at an angle of 45*^. These tubes may be lifted up when it is desired
to remove the cells, they may also be removed entirely from the clips
RR for cleaning purposes, but they should be pushed down when in use
so that the lower edges of the mirrors touch the bottoms of the cells.
When ready for use this box is placed in the space marked DD, Fig. i.

A NEW FORM OP CXJLORIMBTER. 21?

Care should be taken to place the mirrors at an exact angle of 45°.

Under these conditions, in each cell, all light coming through the bot-
tom of the ceil and reflected through the end K, will go through the same

thickness of liquid, and if the mirror were a reflecting surface coming
in direct contact with the liquid this distance would be represented by
the line 0 P, Fig. 3. There is a small error here, due to the converging
of the rays to the eye; this is so slight, however, as not to cause any per-
ceptible uneven illumination.

The mirrors being made of ordinary looking glass, the reflecting sur-
face will be the upper side G H G, Fig. 3, and the light must go through
the glass of the mirror before striking the reflecting surface, and the same
on leaving. The distance which the Ught travels through the glass of

GEORGE STEIGER.

e mirror will be represented by twice the length of the hypotenuse of
isosceles- right- angle-triangle, the equal sides of which are each equal
the thickness of the glass, and must be deducted from the length 0 P.

A ray of hght entering the glass at the point marked N will travel to
and then be reflected to (■ From / to w it will go through the colori-
:tric solution, and this distance is therefore the length to be measured,
point, H, is marked on the mirror near the outer edge, so that it may

seen in the same line of vision as the scale on the bottom of the cell,
d perpendicularly above (. In looking through the end K, this mark
11 be recorded at the point i' directly below it on the scale, and CP
ing the same at Iw, the distance desired can be read off. The posi-
m of the point H is determined by measuring off, on the back of the
irror, a distance from the lower edge equivalent to three times the
ickness of the glass. It may be convenient, if thin looking glass has
en used, to have this point farther up on the mirror (H'), in order that
may be seen more plainly, but if so moved an addition must be made

the observed reading, equivalent to one of the sides adjacent to the
[ht angle of an isosceles- right-angle-triangle, the hypotenuse of which
equal to the distance this point has been removed from H. It is con-
nient in making the graduation on the cell to allow for this correction,
le reading can then be made directly.
Glass cells to answer the purpose may be had of any of the large sup-

SMALL AMOUNTS OF FLUORINE. 219

ply houses, but not graduated, the graduation must be done in the lab-
oratory. The supports F F can be made of rather heavy walled glass
tubing, about i cm. outside diameter.

The mirrors are made of a good grade of looking glass, the lower and
top edges blackened, and cemented to the ground ends of the glass tubes
with Canada balsam, after which the backs are coated with paraffin.
Parafl&n answers well as a coating for a large number of colorimetric
solutions. In case a liquid is to be used which attacks paraffin, a sub-
stitute must be employed which is unaffected by the liquid in question.
It will be found necessary to replace the mirrors from time to time, as
it is not possible to so protect the silvered surfaces as to prevent the grad-
ual eating in from the edges by the various solutions used.

The comparison is made by pouring a solution of known strength into
one of the cells. The unknown solution made up to a definite volume,
is put into the other. The left-hand cell is then placed at a convenient
point, which should be determined by the depth of color of the solution
it contains. The right-hand cell is then moved back and forth till, on
lookmg in the end M of the apparatus. Fig. i, the two mirrors appear
to be of the same shade.

The strengths of the two colorimetric solutions being inversely pro-
portional to the thickness of the liquids looked through, by substituting
in the following equation the amount of the material to be determined
may be found.

Let R equal the reading of the cell containing the known solution with
a concentration C, and r the reading of the cell containing the unknown
sohition, which has a concentration c, then

Cbsmxcal Labosatort,

U. ^. Gbolooical Survbt,

October 31, 1907.

THE ESTIMATION OF SMALL AMOUNTS OF FLUORINE.

By George Stbiobr.
Received November 6, 1907.

The estimation of fluorine in such substances as rock mixtures, which
require a carbonate fusion to bring the fluorine into a soluble form, is
not only a long and difiicult operation, but also the final results are far
from being satisfactory.

When as much as one tenth of one per cent, is present, a negative
result will often be obtained by the Berzelius method, the one usually
employed.

The method to be described is based on the well known fact, that the
presence of fluorine has a powerful bleaching effect on the yellow color.

230 GBORGB STEIGBR.

which is produced by the oxidizing of a titanium solution with hydro-
gen peroxide. A solution of definite volume is made containing tbe
fluorine to be estimated, also having a known amount of titanium pres-
ent ; this is compared in a colorimeter with a second solution containing
an equivalent amoimt of titanium per cc, and the bleaching effect re-
corded. From the extent of this bleaching, the percentage of fluorine
can be calculated.

Although the results obtained are not so accurate as those given by
many methods for the estimation of other elements, yet consdering
tbe difficulty of the fluorine determination, and the time and labor re-
quired by methods now in use, the present one may well be employed,
where small quantities of fluorine are to be determined. The operations
require not only less skill to carry out, but are fewer in number and take
much less time.

Traces of fluorine amounting to several hundredths of one per cent
are easily detected, and an approximation to the quantity can^be made.
In amounts up to a few tenths of a per cent, the method seems to be
more reliable, and if not more than two per cent, is present, the results
compare favorably in accuracy with the standard methods. It is hardly
to be expected, however, to find a colorimetric method using only a
few milligrams of the material to be determined, that will compare in
accuracy with the gravimetric methods u^ng much larger quantities,
where considerable percentages of fluorine are concerned.

Description of the Method. — In rock mixtures containing only a few
tenths of a per cent, of fluorine, two grams of the finely ground rock
powder are fused with four or five times its weight of a mixture of sodium
and potassium carbonates. It may be necessary in the case of a sub-
silidc rock, to add silica, 50 per cent, of which should be present.

From the aqueous leach of the fusion, all of the alumina, and most
of the silica, are separated by adding ammonium carbonate, 'heating
on the water bath for fifteen or twenty minutes, allowing to cool an hour .
or more and filtering. The filtmte is evaporated to small bulk (25 or
30 cc.) and filtered a second time to insure a perfectly clear solution,
a condition absolutely necessary for a satisfactory comparison of tbe
color. After this treatment the solution should be entirely free from
alumina, and contain no more than 25 milUgrams of silica. This amount
of alica is not sufficient to interfere with the reaction.

The solution is now put into a 100 cc. measuring flask, sulphuric acid
added to almost neutral reaction with care not to add an excess, and well
shaken to fre^ from the excess of carbon dioxide, and then fully acidi-
fied. Care should be taken not to have an excess of acid present be-
fore shaking, for the reason that the escaping gas will carry off some
fluorine; even under the above conditions a slight loss occurs. If a con-

SMALL AMOUNTS OF FLUORINE. 321

ddeiable amount of fluorine is present, an aliquot part of the solution
should be used containing not more than 2 or 3 milligrams. 20 cc of
standard titanium sulphate solution' (i cc. of which contains 0.0001

\ii

READING
' Directions foe the preparation of the titanium solution will be found i
second part of this paper.

222

GEORGE STBIGBR.

■ .a » • •

i t^i

gm. TiOa) are now added, together with 2 or 3 cc. of hydrogen peroxide,
and the flask filled to the mark with water. The solution is now ready
to be compared with the standard. The latter is prepared by using
20 cc. of^the standard titanium solution, 2 or 3 cc. of HjOj and bringing
the volume up to 100 cc. with water.

These two solutions containing the same amount of titanium per cc
should be of the same depth of color, but the one having the fluorine
present will be found to be of a lighter shade, owing to the bleaching
effect which that element has on a solution of titanium oxidized by HjOy
The extent of this bleaching is not directly proportional to the amount
of fluorine present, but by reference to the curve below, the quantity
corresponding to a given bleaching can be found.

The two solutions are now compared in a colorimeter and their ratio
recorded. Suppose a ratio for example, of 100 to 85, that is, the fluorine
present has caused a bleaching of the solution equivalent to 15 per cent,
of the titanium present; then by finding the point on the abscissa marked
85, with the help of the curve the amount of fluorine (0.00055) can be
directly read off on the ordinate.

It is necessary to employ a colorimeter^ whose error is not more than
two or three per cent.

Experimental Work«

Directions for the preparation of the solutions used in the experimental
part of this work:

Titanium Solution. — A quantity of potassium-titaniimi-fluoride was
dissolved in water, a large excess of sulphuric add added and the
solution evaporated till fumes of sulphur trioxide came off, then cooled,
more water added and the operation repeated. This was done several
times to insure the entire removal of fluorine; the solution was then
largely diluted with water, and its contents of titanium determined by
precipitating a portion of it with ammonia, and weighing the titanium
directly as its oxide. The solution was next so diluted as to contain
0.000 1 g. TiO, per cc., at the same time adding enough sulphuric add
to make a 3 per cent, solution.

Fluorine Solution. — 1.236 gm. of potassium-zirconium-fluoride were
dissolved in one liter of water. One cc. of this solution contained 0.0005
gm. of F.

Silica Solution. — Five grams of silica and 10 gm. of sodium carbonate
were fused together, dissolved in water, filtered, and made up to 500
cc. This solution contained 0.0 1 gm. SiO, per cc.

Aluminum Solution. — Common alum was dissolved in water sufficient
to make a solution containing 0.0 1 gm. Al^O, per cc.

' A description of the colorimeter used in this work will be found in the preced-
ing article.

SBiALL AMOUNTS 01^ FLUORINB. 223

Phosphorus Solution. — ^A solution of microcosmic salt was made, one
cc. of which contained 0.005 &^ ^fiy

Iron Solution. — ^A solution of ferric sulphate containing 0.005 g™- F^
in each cc.

In the method described, after fusing the rock with carbonates and
leaching with water, there will be in solution besides the fluorine, sili-
cates and aluminates of sodium and potassium, and the excess of sodium
and potassium carbonates, all in large amotmts, a small qtiantity of
iron, and all of the phosphorus, chlorine, vanaditun, and some other
elements usually fotmd in very small quantities. The effect of silicon,
aluminum, sodium^ and iron salts, and the phosphorus, has been deter-
mined; the possible effect of traces of such salts as vanadates, tungstates,
and chlorides, has not.

In all the following experiments 20 cc. of the titanium solution and
1} cc. of the fluorine solution were used, oxidized with 2 or 3 cc of hy-
drogen peroxide and made up to 100 cc. A large number of compar-
isons were made of this solution with one of the same composition, ex-
cepting that it contained no fluorine. The a.verage ratio fotmd was
100 to 82.5, that is, I J cc. of fluorine solution (0.00075 g- F) should
cause a bleaching effect equivalent to 17.5 per cent, of the titanium pres-
ent, when the solution contains no interfering substances.

The following results were obtained, when the several salts were in-
troduced into solutions containing the above quantities of titatiium
and fluorine.

The figures given tmder the head of ** reading" denote the ratio of the
cotor of the solution containing the fluorine, etc., to that of the stand-
ard titanium solution, the latter being taken as 100.

Effect of Sodium Salts. — ^The sodium carbonate was dissolved in water,
the solution acidified with sulphuric add and well shaken to free it from
dissolved gas, the titanium, fluorine, and hydrogen proxide added, and
made up to 100 cc. with water.

Na^„ gms 8. 8. 8. 8. 8. 8.

Reading 85.3 85.7 83.7 86.4 85.2 84.4

Na/^Oa, gms 8. 8. 8. 8. 8. 8.

Reading 82.9 83.5 83.3 83.4 83.1 86.3

Effect of Silica. —

Silica solution, cc. added i. 2. 5. 5. 5.

Reading 84.4 83.3 83.8 85.7 84.2

Silica solution, cc added 5 . 10. 30. 50.

Reading 84.2 84.9 85.9 88.8

Effect of Alwminum. —

Alnm solution, cc. added i . 3. 5 . 5 • 20.

Reading 92. 96.3 97.3 96.6 99.3

224

GEORGE STEIGER.

i'''H

• =-:.iik

EjjPec/ of Iron. —

Iron solution, cc added 5 . 5 . ....

Reading 90. 89 . 9 .... .... ....

Effect of Phosphorus. —

PjOj solution, cc. added 5 . 25 .

Reading 84.5 80.3

The following experiments were made having phosphorus present
but no fluorine.

P,05 solution, cc. added 5 . 25 . ....

Reading 98.4 94-3 ••••

This bleaching effect on the color of an oxidized titanium solution
by phosphorus has been observed before.

These results show that sodium salts in large amounts have a slight
effect in making the observed reading higher, but not enough to seriously
alter the results. Silica in amounts up to o.i g. has but little effect.
It is easy, however, by one precipitation with ammonium carbonate
to leave not more than 30 milligrams of silica in solution. Aluminum,
even in small quantities, has a very marked effect on the bleaching,
but this base is entirely separated by the ammonium carbonate treat-
ment. Phosphoric acid has the same effect as fluorine in bleaching the
color, but as this is present in much smaller quantities than those used
in the above experiments, its effect can be neglected. Iron prevents
the bleaching, but as the quantity coming out in the leach water is hardly
more than a trace, and even this small amount is probably separated
by the treatment with ammonium carbonate it can also be neglected.

Various mixtures were made roughly representing commonly occurring
rock mixtures, containing accurately known amounts of fluorine. Re-
sults of the fluorine determinations in these mixtures follow:

SiOs.

0.60

* 0.70

II
<i
II
II
II
II

A1«0«.

0.20
II

0.15
II

FetOs. MKCOs.CACOs.NAHNH4PO4.TiOs. PcalculAted. P found.

0.0XX4 0.0102 Gm.

0.05
II

0.03
II

0.05
II

O.IO

0.05

II
II
II
II
II
II

O.OI

0.008

O.OI

0.005

II
II
II
II
II
II

0.0052

0.00284

0.0005

0.00526

0.00253
0.0x228
0.00516
0.00536

0.00525

0.0027

0.0003

0.0056

0.0040

0.0020

0.0089

0.00486

0.00530 "

To give an idea of the amount of silica in solution after the treatment
with ammonium carbonate, several determinations were made in the
above solutions after the colors had been compared. The results follow:

^ In this determination the fusion of the mixture was made as usual, and the
fluorine added to the leach water before the treatment with ammonium carbonate.

VOLUMETRIC DETERMINATION OF ZINC. 225

StO, 0.0244 0.0740* 0.0268 0,0555* 0-0159 0.0298 0.0174 0.0113 0.0263 Gm.
The following determinations were made of fluorine in natural rocks
and compared with the gravimetric results:

Gravimetric determinatioii o. 15 3 .01 3 .01 per cent.

Colorimetric determination 0.21 2 . 58 3 . 20 per cent.

A number of determinations of fluorine were made in rocks contain-
ing quantities var3dng from 10 to 20 per cent., but the results were not
satisfactory, being several per cent, out of the way.

Chkmicai. Labokatort,

U. & Gbolooicajl Sukvby,

October 31. 1907.

VOLTTMETRIC METHOD FOR THE DETERMINATION OF ZINC«

By Wm. Hbkbbrt Kbbn.
Reodved Noyembcr 11, 1907.

Several schemes have been advanced for the estimation of zinc by titra-
tion with potassium ferrocyanide, but so far no one of them has fotmd uni-
versal application. The methods which have fotmd favor are either so
complicated or so difficult in manipulation that a large personal error is al-
ways introduced and it seldom happens that very close checks are ob-
tained by different operators, even when the same procedure is followed.
Nearly a year ago I became interested in the volumetric determination
of this metal with the idea of substituting it for the rather tedious grav-
imetric method which has always been in use in our laboratory.

At first, and for quite a long time in fact, I was not very successful and
encountered numerous difficulties. The samples which I used in this
preliminary work were spelters, the zinc content of which had been very
carefully determined by difference. Finally, after a trial of all the meth-
ods which seemed reliable, with varying successes, I concluded that
changes might be made to good advantage in nearly all of them, so with
what experience I had already gained, I attempted to work up a scheme
which would embody the good points of all of these methods and as far
as possible none of the bad ones. The method which I am about to de-
scribe, therefore, is not new, but is rather a re-modeling of the older ones.
The scheme is one which is identical in certain parts for almost every
condition, but there are some slight variations which are necessary for
different products and, if accuracy is desired, they should be observed.
Accordingly, I will describe the method, applying it to typical cases.

Preparation of the Ferrocyanide SoluUon, — Dissolve crystals of c. p.
potassium ferrocyanide in water in the proportion of 22 grams to the
Hter. If the solution is not clear, it should be filtered before it is diluted
to the desired volume. It is a good plan to make up several liters at a

' In these two cases the silica was filtered immediately after being precipitated
with ammonium carbonate, which accomits for the large quantity found.

226 W. HERBERT EEEN.

time, as tbe solution remains constant almost indefinitely, and if many
determinations are necessary, a bottle of fifteen, or even twenty, liters
of solution will not last very long.

Preparation of the Standard Zinc Solution. — This solution should be
made up with the greatest care, as all subsequent work depends on its
accuracy. In order to make up two liters, weigh very carefully and trans-
fer to an 800 cc. beaker 10 grams of c p. zinc or an equivalent weight
of the oxide (13.4465 grams)— ^if the latter is used, it must be freshly
ignited, cooled in a desiccator — and dissolved in 50 cc. of hydrochloiic
add diluted with water. Heat until solution is complete and then dilute
to about 300 cc and add a considerable excess of bromine water. Heat
until the bromine is entirely expelled. Wash the cover glass and the
sides of the beaker and add an excess of ammonia. Put in a warm place
on the hot plate where it will remain just below the boiling-point, or
boil very gently for about fifteen minutes. By this time the small amount
of iron, which is nearly alwajrs present, should be completely precipita-
ted. Filter carefully into a graduated liter flask, washing the beaker
and precipitate once with water containing a little ammonia and then
several times with hot water. Replace the flask under the funnel by a
small beaker (about 350 cc. capacity) and dissolve the iron precipitate
in hot dilute hydrochloric acid. Re-precipitate the iron with ammonia
and filter again, adding the filtrate to the main solution. Bum the filter
paper containing the iron precipitate and ignite. Weigh as Fe,0, and
calculate to Fe, deducting this weight from the original 10 grams of zinc
Sometimes small amounts of silica and dirt are also present in the so-
called c. p. zinc, and the weight of this should also be deducted. These
are small corrections and make very little difference as a rule, but it is al-
ways well to make them, as it gives the operator more confidence in his
results and leaves no loop-hole for inaccuracy in later work. Now
make the solution barely acid, using a small piece of litmus paper as an
indicator and add 30 cc. of hydrochloric acid in excess for each liter of
solution. Add also 10 grams of ammonium chloride for each liter. Di-
lute to the mark with water, after cooling, and pour carefully into a
dry two-liter bottle. Do not wash out the flask, but after it has drained
as much as it will, fill it to the mark with distilled water and add this
to the rest of the solution. Shake until thoroughly mixed and the solu-
tion is ready for use. Its value will be 0.0050 gram of zinc per cc of
solution — less a small correction due to the impurities deducted from
the original weight.

Standardizalion, — By means of standard pipettes, measure out three
portions of the zinc solution of 20 cc. each, three portions of 50 cc, three
of 70 cc, and three of 100 cc, into 400 cc. beakers of the wide shallow
type, which will greatly lessen the amount of stirring necessary in order

VOI/UMBTRIC DETERMINATION OP ZINC. 227

to obtain a homogeneous solution after each addition of the ferrocyanide.
Add about i cc of hydrochloric acid to each of the smaller portions.
Dilute each to 150 cc. and all are ready for the titration. Except for
very accurate work it is not necessary to take more than three or four
portions altogether (one for each amoimt) to determine the average
value of the ferrocyanide solution, but it has been found^ that a slightly
di£ferent factor should be used for different amounts of zinc, hence the
necessity in accurate work for establishing these factors for amounts of
solution likely to be used most frequently. Fill a clean burette with
ferrocyanide, heat one of the portions of zinc chloride nearly to boiling
and titrate as follows: Pour off about 20 cc. of the hot zinc solution
mto a small beaker and set it aside. Titrate the remainder by running
in ferrocyanide, a few cc. at a time, tmtil a drop tested on a porcekin
tile with a drop of a five per cent, solution of uranium acetate shows
a decidedly brown color. Now add all but about 2 cc. of the reserved por-
tion, and having tested to be sure that the endpoint is not now over-
stepped, add the ferrocyanide in portions of half a cc. at a time imtil
the end point is again passed. Finally add the last of the reserved por-
tion, washing it all out with distilled water and carefully washing down
the sides of the titration beaker, and finish the titration by adding two
drops at a time, testing after each addition. The endpoint at this time
will be sharper if instead of one drop two or three drops are added to the
drop of uranium acetate on the tile. The amoimt of zinc lost at this
stage of the titration will be so small as to be insignificant. When the
final distinctly brown tinge is obtained, wait a minute and observe if
one or more of the preceding tests do not also develop a color. Note
the number of spots by which the endpoint has been overstepped, and
since each test means an addition of two drops, or o.i cc. of solution,
make the necessary correction when reading the burette. A blank should
also be made and the final reading and all other readings corrected by
this amount. The blank should be made with a solution containing 10
cc. of hydrochloric add, neutralized with ammonia, made add again,
and after 3 cc. of add have been added in excess, diluted to 150 cc. and
heated nearly to boiling. This will usually require about 0.2 cc. to give
a dedded test. Rim all the titrations in the same way and calculate
the values for one cc. of the solution for the different portions, taking the
average of the three results in each case if they are close enough (it is
easy to obtain checks within o.i cc), thus obtaining the different factors
for the varying amotmts of zinc — or what is equivalent to the same thing,
for varying amotmts of ferrocyanide. In actual work later, the factor
should be used for the amount of solution which corresponds most nearly
to that required for the titration. Ordinarily, however, the one average

^ This Jottmal, ag, 205.

W. HERBERT EEEN.

)r for three or four different amounts of zinc will be accurate enou^

,11 determinations.

le last solution which I standardized had the following factors:

SUndiTd

noololion. Gnmi of linc. FeirocyBnide Hluilon, Ptcton.

lo 0,09995 194 0.005153

, 50 0.34988 48.1 0,005195

70 0.34983 67.! 0,005306

100 0.49975 95.5 0.005333

reater accuracy could be obtained with a slightly weaker solution,

15 or 18 grains to the liter, but for general application I find a solu-
of the above strength the most convenient.

ttermination 0/ Zinc in a Spelter. — Method of sampling. — There are
ral methods in common use, but probably the most accurate one
I break off small pieces from slabs in different parts of the pile until
igh has been obtained to make a fair average — say five pounds for

car load, if the shipment is homogeneous. The sample thus ob-
;d should then be heated in an ordinary clay crucible until just melted,
1 it should be poured into a wooden box, previously rubbed with
Ic — and shaken violently. A box suitable for this must be made

tight joints and no cracks and should be supplied with a cover. This
:ment should result in a finely granulated sample. Screen out the
ser pieces by passing through a 20-mesh and reduce the bulk of the
stuff by coning and quartering until a small working sample is oh-
m1. This not only will be found very easy to wei^, but will in ad-
m almost perfectly represent the whole consignment.
nalysis. — Weigh five grams of the sample of spelter, place in a 600
>eaker, cover with a watch-glass and dissolve in 50 cc of hydrochloric

{1.20) diluted with water. Heat until completely dissolved. The
Lition of hydrogen, occasioned by the solution of the zinc, predpi-
i such metals as copper, lead and tin; but the heating should be corn-
ed until these metals go into solution again, as they occlude small
unts of zinc. When solution is complete, neutralize with ammo-
and readdify with hydrochloric add (1.20), adding 5 cc. in excess,
;e to about 300 cc. and pass a current of hydrogen sulphide gas through
warm solution for about twenty minutes. If the predpitated sul-
es do not settle readily, allow the solution to stand for an hour or
t a temperature just below the boiling-point. This will not be neces-
, unless tin is present. Filter off the sulphides, allowing the filtrate
an into an 800 cc. beaker and wash the filter thoroughly with hot
:r. Heat the filtrate to gentle boiling and aftej the free hydnagen
hide has t>een expelled add a large excess of bromine water and con-
£ the boiling until the bromine has been expelled. Now add a large
ss of ammonia and boil gently for about fifteen minutes. If the

VOhVUEtlStlC DEtEftMlNATlON OF ZINC. 229

amount of iron is considerable, it will not be necessary to boil so long, as
it is only in the case of very small amounts that the precipitation is likely
to be incomplete, imless the boiling is continued. Filter into a graduated
Hter flask, washing first with dilute ammonia — both the beaker and the
filter — and then with hot water. Put about lo cc. of hydrochloric add
(1.20) in the beaker, dilute with hot water and pour it over the precipitate
on the paper, allowing it to run through into a small beaker of about 250
cc capacity. Re-predpitate with ammonia and filter into the main
solution in the flask, washing as before. If the amoimt of iron is very
large, a third predpitation will be necessary, but in the ordinary spelter
this is not the case. Put a small piece of litmus paper in the flask and
add hydrochloric add (1.20) until the solution is just add, and then add
30 cc in excess. Cool to room temperature and then make up to the
mark with water. After mixing the solution thoroughly by several
decantations back and forth from the flask to a beaker, measure out
several portions of 100 cc. each (equivalent to 1/2 gram of spelter) by
means of a pipette. I find it convenient to use a 400 cc. beaker for the
titration. Usually only one portion is necessary, but it is well to have
a second one in reserve. Dilute to 150 cc, heat nearly to boiling (about
85°) and titrate with ferrocyanide, as described in the procedure for
standardization. Calculate the percentage of zinc by multipl)dng the
standard value of one cc. by the number of cubic centimeters required
(kss the blank) and divide by 0.5, the weight of spelter taken as a sam-
ple.

// mat^anese is present in the spelter — a rare occurrence — ^the above
result will be too high. If time is an object, the quickest way to over-
come this difficulty is to make a separate determination of the manga-
nese, although it may be removed before titration in a way I shall describe
later. To determine the manganese, the best and most rapid way is
to use the Bismuthate Method as applied to steels: Dissolve i gram
of the spelter in a 150 cc. Erlenmeyer flask in 50 cc. of nitric add (i to 3)
and boil oflf the nitrous fumes. Cool in ice-water and add about a gram
of sodium bismuthate. If manganese is present, the pink color of per-
manganic add will develop at once. If such is the case, filter through
an asbestos filter into a 3*0 cc. Erlenmeyer, washing with 100 cc of ice-
cold 3 per cent, nitric add. Add a measured amount of ferrous sulphate
solution (12 grams to the liter) until the solution is perfectly colorless;
titrate the excess with a standard permanganate solution (i gram to the
liter). Subtract the volume of permanganate required from the amoimt
of the latter equivalent to the volume of ferrous sulphate used and mul-
tiply by the manganese value of the permanganate solution. The factor
is about 0.00035 for a solution of this strength. This will give the per-
centage of manganese in the spelter. The chemical equivalent of this amoimt

) W. HEKBBRT KEEK.

zinc should be deducted from the percentage of zinc previously ob-
ned.

If cadmium is present, it will also remain in solution with the zinc and
U aSect the titration in a manner dmilai to the manganese. It is
mlly present in such very small amounts that it is not readily predpi-
,ed-by hyrdogen sulphide except from a nearly neutral solution, in
lich part of the zinc is also precipitated. It is therefore necessary to
ike a separate determination of the cadmium and correct the apparent
lount of zinc, as was done in the case of manganese. Another altema-
e, which has given me doubtful success in the case of such small amounts,
to boil the solution for a few minutes with a piece of aluminum fni
■t previous to titration. The aluminum thus introduced into the
ution does not affect the result. Usually, however, unless specially
i:ed for, cadmium is always counted as zinc.

To Determine Zinc in Ores. — Decompose one gram of the sample in a
iker with hydrochloric acid, or if necessary with aqua regia. Evapo-
e to dryness, and if nitric add has been added, re-dissolve in hydro-
oric acid and evaporate again. Dissolve in 15 cc. of hydrochloric
d (r.2o),dilute with water, filter and wash. Bum off the filter (unless
rontains lead) in a platinum crucible and treat with hydrofluoric and
phuric adds — exactly as in a silica determination. Dissolve the residue
hydrochloric add and add to the solution. Usually this treatment of the
idue is not necessary, as it rarely contains more than a few hundredths
a per cent, of zinc. If lead sulphate is present in this residue, use a
■fomted crudble with an asbestos felt for a filter. Remove the Si-
te after washing the residue thoroughly and dissolve out the lead
phate with a strongly ammoniacal solution of ammonium dtrate.
ish with hot water and then transfer the remaining reddue, together
;h the felt, to an ordinary platinum crudble and treat wih hydrofluoric
d as before, finally adding the hydrochloric add solution of the re-
ining residue to the filtrate. (Advantage may be taken of the above
ps to determine the lead in the same portion— the only changes neces-
y being to evaporate with sulphuric add instead of hydrochloric and
weigh the perforated crudble before and after treatment with the
monium dtrate, calling the difference lea# sulphate.) Neutralize

filtrate with ammonia and again acidify, adding alnnit 3 cc. of add
excess for every 200 cc. of solution. Pass hydrogen sulphide through

solution to remove the copper group and, having filtered out the pre-
itate thus obtained, heat until the free hydrogen sulphide has boiled

and then add an excess of bromine water and evaporate to about

cc. Predpitate the iron with ammonia, udng a large excess and
shing with water containing ammonia. If the predpitate amounts
more than a few tenths of a per cent, it should be dissolved and re-

VOLUMETRIC DETERMINATION OF ZINC. 23 1

precipitated and in case of large amounts of iron a third precipitation is
sometimes necessary. Acidify the filtrate with hydrochloric acid, add-
ing 3 cc. in excess. The solution is now ready for titration, unless man-
ganese is present, if its volume does not exceed 150 to 200 cc. If it is
of larger hulk than this, it should be evaporated until of approximately
this volume, before titration. In the case of very large amoimts of iron
the sulphide of zinc should be separated from it just as described below
when nickel is present, or, if preferred, the ether separation may be used.

When Manganese is Present in the Ore. — If the iron is small in amount,
add to the solution after evaporation, but before removing the iron,
five grams or more of potassium or sodium bromide. Make strongly
ammoniacal and stir for an hour by means of some such device as that
used for the solution of steels in the determination of carbon. If no ap-
pliance of this sort can be had, the solution should be allowed to stand
over night at room temperature. The manganese and iron are com-
pletely precipitated and should be filtered off and washed with water
containing ammonia. The filtrate should then be acidified as above
and titrated. If the iron is present in large enough quantities to require
re-predpitation, it should be removed before treatment with the bro-
mide; but if both the manganese and iron are precipitated together, as
above, and it is desired to re-precipitate them, they should be dissolved
in a mixture of hydrochloric and sulphurous adds, the solution boiled
to expel the sulphurous fumes and then oxidized with bromine water.
The bromine should be boiled off as before and the predpitation then re-
peated with the alkali bromide and ammonia. The amount of bromide
may be varied without any effect on the titration. If it is desired to de-
termine the manganese, it may be dissolved in sulphurous add alone,
and after solution is complete, nitric add added, the nitrous fumes boiled
out and the determination completed by the Bismuthate Method.

Of course the manganese may be readily avoided by predpitating
the zinc as sulphide from a slightly add solution and this is to be advised
in the case of large amounts of manganese — also, if any nickel is in evi-
dence. The presence of nickel will be made known by a blue color of
the ammoniacal filtrate from the iron. If it is necessary to predpitate
the sulphide of zinc to avoid these metals, the best plan is to evaporate
the hydrochloric add as first obtained to a small bulk in order to expel
most of the free add, dilute with a little water and neutralize with sodium
carbonate solution, using methyl orange as an indicator. The neutral
point should be carefully obtained and then ten drops of hydrochloric
add added. Wash down the sides of the beaker, dilute to 200 cc. and
pass hydrogen sulphide through the cold solution for an hour. Let
the solution stand for several hours, or, if convenient, over night before
filtering. Zinc sulphide precipitated in this way should filter out like

232

W. HERBERT KEEN.

so much sand. Put the paper containing the precipitate into a beaker
and digest until dissolved in lo cc. of hydrochloric add diluted with water.
Pass hydrogen sulphide for a few minutes to be sure that all the copper
is precipitated and then filter and wash thoroughly with hot water.
Evaporate the filtrate after adding a slight excess of bromine water to
oxidize any iron that may have been occluded by the zinc, and when it
is reduced to a small volume add an excess of ammonia and boil. Filter,
if necessary, and neutralize the filtrate with hydrochloric add, adding
3 cc. in excess. Dilute to 150 cc. and titrate. This method is excellent
if the conditions are just right, but my idea has been to avoid a predpi-
tation of the zinc, if possible, and thus get rid of what may be a trouble-
some filtration. Accordingly, I have endeavored, except in cases where
nickel is present, to predpitate from the zinc all metals interfering with
the titration.

Brasses. — Remove the tin, copper and lead in the usual way— by
solution in nitric add, filtration of the metastannic add, and deposition
of the two latter metals by electrolysis. If it is certain that the copper
is entirely rfemoved, evaporate to dryness the nitric add solution, which
now contains only the zinc and perhaps a small amotmt of iron. Dis-
solve the residue in 10 cc of strong hydrochloric add and, after dilution,
predpitate the iron with ammonia, being careful to boil long enough
to insure a complete precipitation, as iron comes down very slowly when
present in such small quantities. Filter and wash with dilute ammonia
water and then with hot water. Neutralize the filtrate with hydrochloric
add, add an excess of 3 cc., dilute to 150 cc., and titrate with the stand-
ard ferrocyanide solution. In the case of a manganese bronze, the only
modification would be the treatment with alkali bromide, as described
above imder ores.

It is sometimes easier, in the case of brasses, to make the filtrate from
the electrolysis ammoniacal and add colorless ammonium sulphide.
Digest at a temperature near the boiling-point for about an hour, allow
to settle and filter, keeping as much of the predpitate in the beaker as
possible. When the liquid has run through the fuimel, replace the fil-
trate beaker by the one containing the predpitate and pour dilute hy-
drochloric add over the filter, allowing it to run through on to the main
bulk of the precipitate. Wash the filter once or twice and heat the solu-
tion until the zinc is dissolved. Pass a current of hydrogen sulphide
for a few minutes and filter out any small traces of copper that may be
predpitated as sulphide. Oxidize the filtrate with bromine water and
evaporate to a small volume, finally precipitating the small amotmt of iron
and finishing as before.

Aluminum Alloys Containing Zinc. — Aliuninum does not affect the
ferrocyanide and so the natural idea would be to dissolve in hydrochloric

DETERMINATION OF BENZENE IN ILLUMINATING GAS. 233

add and proceed directly with the titration as soon as solution was com-
plete. But iron is almost always present in such alloys to a considera-
bk amount, and hence the necessity for a more roundabout method.
Dissolve one gram in either nitric or hydrochloric add. Add a few crys-
tals of dtric add, make ammoniacal, and heat to boiling. Remove from
the source of heat and add a little colorless ammonium sulphide. Al-
low to settle and then filter, being careful to wash out the last traces of
dtric add with a wash water containing a little ammonium sulphide.
The predpitate will consist of sulphide of zinc, iron and copper metals.
Treat it exactly as described for bmsses under similar conditions. If
desired, the metal may be dissolved in hydrochloric add and the zinc
predpitated as sulphide from a slightly add solution just as when it is
required to separate it from nickel.

In general, any one working intelligently, realizing the final conditions
and knowing something of the material in hand, will find no difficulty
in following out the plan of this method.

Labokatort of Booth, Gaxxbtt & Blair,
Fhikuldpliia.

THE DETERMmATION OF BEHZENE IN ILLUMmATING GAS.

Bt L. M. Dennis and Bllbn S. McCarthy.
Received November 25. 1907.

L The Absorption of Benzene by Ammoniacal Nickel Nitrate.

In 1903 Dennis and O'Neill described* an absorption method for the
determination of benzene in illuminating gas. The absorbent there
recommended was an ammoniacal solution of nickel nitrate, the use
of such a solution for the determination of benzene having been suggested
by the statement of Hofmann and Kiispert* that when illuminating
gas acts upon a mixture of nickel hydroxide and ammonia, there is formed
a compound of nickel cyanide with ammonia and benzene, Ni(CN)2.
NHrCeH..

In practice, this method for the determination of benzene in some local-
ities, has given most excellent results, while in other quarters it has
been far from satisfactory. Morton recently demonstrated* that when
mixtures of benzene and air are analyzed with the use of the reagent,
the results are scarcely better than might be obtained with water alone,
and that, moreover, the efl&dency of the absorbent steadily decreases
as the amotmt of benzene that it has taken up increases. These state-
ments of Morton have been substantiated in this laboratory and the

* This Journal, 25, 503.

* Z. anorg. Chem., 15, 204 (1897.)

* This Journal, 38, 1728 (1906),

; , 'I *

■,.^'r^

234

L. M. DENNIS AND ELLEN S. MCCARTHY.

only explanation of the excellence of the results in the analyses of mix-
tures of air and benzene as published by Dennis and O'Neill would seem
to lie in the fact that but few absorptions of benzene were made and that,
consequently, the reagent had not taken up sufl5cient benzene to render
the loss in its absorbing power noticeable.

Confronting the facts adduced by Morton, however, were arrayed
statements from several University and technical laboratories testify-
ing to the satisfactory character of the method and the imiformity of
results obtained by it in analyses of illuminating gas. In seeking for
explanation of these variations in the reports of different chemists it
occiured to the authors of the present paper that those samples of illumi-
nating gas on which the method gave good results might have contained
cyanogen compounds necessary to the formation of the compoimd that
Hofmann and Kiispert described, while the poor results on other sam-
ples of illuminating gas might be caused by the absence of cyanogen
compotmds from the gas. To test the validity of this assumption, the
ammoniacal nickel nitrate solution recommended by Dennis and O'Neill
was used in a series of analyses of a gas mixture containing benzene,
hydrocyanic add and air.

The hydrocyanic add gas for these analyses was prepared by filling
a Hempel nitrometer with mercury, then introducing from five to ten
cubic centimeters of a concentrated solution of potassium cyanide and
adding gradually through the side arm about the same volume of hydro-
chloric acid (one part of add to one part of water). The gas set free
by this reaction was completely absorbable by caustic potash. It prob-
ably contained small amounts of cyanogen and carbon dioxide but it
will be referred to below as "hydrocyanic add." A mixture of this
gas with benzene and air was obtained by measuring off a volume of
air, then passing this into a Hempel simple gas pipette containing liquid
benzene, drawing the gas back into the burette and noting the increase
in volume, and finally drawing into the burette the hydrocyanic add
gas that had been evolved in the nitrometer, and again measuring the
volume. The measurements were made in a jacketed Hempel gas burette
with mercury as the confining liquid. Several mixtures prepared in
this manner were analyzed by the Dennis and O'Neill method. The
same sample of ammoniacal nickel nitrate was employed for all of the
absorptions listed in Tables I, II and III.

Table I gives the results obtained with the mixture of air, benzene
and hydrocyanic add. The completeness of the absorption may be
seen by comparing the volume of air taken with the volume of gas re-
maining after absorption. It will be noted that the agreement is prac-
tically exact in every case except the 4th.

DBTBRIONATION OF BENZENE IN ILI<UMINATINO GAS. 335

Tabls I.

Volume retnainliig
Hydrocyanic after abftoii'tion
Nnmberof Air Benzene acid gaa by ammoniacal

experiment, taken. taken. taken. nickel nitrate.

cc. GC. cc. cc.

1 56.2 i.o 3.2 56.2

2 56.4 0.6 2.4 56.5

3 56.5 1.9 2.4 56.5

4 56.8 3.4 5.5 57.1

5 57.7 1.5 1-2 57.7

6 58.1 1.4 1.5 58.1

7 58.2 2.1 6.1 58.3

8 591 0.8 1.3 59.1

While the above table demonstrates that, in the presence of cyanogen
compounds, ammoniacal nickel nitrate quantitatively absorbs benzene
vapor, the accurate results obtained by some analysts with the reagent
could even yet not be explained, if these cyanogen compounds in a gas
mixture ate completely removed by potassium hydroxide, the reagent
that precedes the ammoniacal nickel nitrate in the analysis of illumina-
ting gas. To ascertain whether any of the cyanogen compounds are
left in the gas mixture after it has been passed into the potassium hy-
droxide pipette, the analyses given in Table II were made. In each
analysis the gas mixture was allowed to remain in the potassium hydrox-
ide pipette for such a length of time as had been found to suffice for the
absorption of all of the hydrocyanic add when that was mixed with
air alone. The results show that when hydrocyanic acid, benzene and
air are present together, potassium hydroxide does not remove all of
the hydrocyanic add, the benzene appearing to exert ^ a deterrent effect
upon the absorption. It will be noted that the ammoniacal nickel ni-
trate quantitatively removed the benzene and the residual hydrocyanic
add in nearly every case.

Table II.

Volume

absorbed Volume re-
Volume by ammo- mainingafler
Hydrocyanic absorbed niacal absorption
Air Bensene add gaa by potassium nickel by ammoniacal
Nnmberof taken. taken. taken. hydroxide. nitrate, nickel nitrate,
experiment. cc. cc. cc. cc. cc. cc.

9 59.4 1.4 7.2 6.9 1.7 59.4

10 59.9 0.5 8.6 8.4 0.5 60.1

II 60.2 I.I 3.1 3.0 I.I 60.3

12 56.1 0.9 5.5 4.8 1.6 56.1

13 56.4 0.9 1.7 1.6 1.0 56.4

14 56.7 1-3 2.0 1.8 1.5 56.7

15 57-0 0.7 14.3 13. 1 1.8 57.1

16 57.8 1.7 1.9 1.9 1.7 57.8

If, from the results in the above table, we are justified in assuming
that after treatment with caustic potash some cyanogen compounds
still remain in illuminating gas, it at once becomes evident why, the

i! It; !■■■!:'

236 L. M. DENNIS AND ELLEN S. MCCARTHY,

Dennis and O'Neill method gave satisfactory results with such samples
of illuminating gas as contained cyanogen compotmds.

After the first three determinations in Table II had been made, it
was thought possible that the ammoniacal nickel nitrate had then taken
up sufficient hydrocyanic add to enable it to absorb benzene vapor with-
out the addition of further hydrocyanic add gas to the mixture. This
assumption seems also to be borne out by the fact that in Determina-
tions II and 13 the absorption of benzene was complete even when only
i/io of a cubic centimeter of '^hydrocyanic acid" gas was present with
the benzene vapor. The results of these analyses are given in Table
III.

Tabls III.

Volume absorbed by
Number of ammoniacal nickel

experiment. Air taken. Benzene taken. nitrate.

cc. cc. cc.

17 60.5 1.5 1.5

18 60.5 1.9 1.8

19 60.9 0.9 0.7

20 60.8 1.4 1.4

21 61.2 0.7 0.7

The results, Nos. 12 to 21 inclusive, demonstrate that after the am-
moniacal nickel nitrate has taken up some cyanogen it is able quanti-
tatively to absorb fairly large amounts of benzene vapor, and that if
cyanogen compounds are present in illuminating gas, the benzene con-
tent may accurately be determined by the method of Dennis and O'Ndll.
If, however, the reagent contains no cjranogen compounds and if none
are present in the ^as that is being analyzed the method will not give
accurate results.

n« The Absorption of Benzene by Ammoniacal Nickel Cyanide.

The foregoing experiments having demonstrated that a solution of
ammoniacal nickel nitrate cannot be relied upon for the absorption of
benzene from all samples of illuminating gas, the necessity arose for
such modification of the method as would render it imiformly accurate.

It is apparent that to insure complete absorption of the benzene from
all gas mixtures through the formation of the ammonia benzene nickel
cyanide, cyanogen must be present in every case. This led to the use
of an ammoniacal solution of nickel cyanide^ as an absorbent in place
of the ammoniacal solution of nickel nitrate.

Preparation of the Ammoniacal Solution of Nickel Cyanide. — ^To 50
grams of nickel sulphate (NiS04.7HjO), dissolved in 75 cc. of water, are
added 25 grams of potassium c)ranide dissolved in 40 cc. of water. After
the addition of 125 cc. of ammonium hydroxide (Sp. Gr. 0.91) the mixture
is shaken until the nickel cyanide has completely dissolved and is then

^ See also Hofmann and Amoldi, Ber., 391 339 (1906).

DETERMINATION OF BENZENE IN ILLUMINATING GAS. 237

allowed to stand at a temperature of o*^ for twenty minutes. The clear
liquid is decanted from the crystals of potassium sulphate that have been
precipitated, and is treated with a solution prepared by dissolving 18
grams of crystallized citric add in 10 cc. of water. After the mixture
has stood again at 0° for ten minutes, the greenish blue supernatant
solution is decanted and is introduced into a gas pipette. Two drops
of liquid benzene are now added to the reagent through the large tube
of the pipette and the pipette is shaken until the benzene has combined
with the reagent. This is effected in two or three minutes. This addi-
tion of benzene to the reagent is made because it was found that at times
a freshly prepared solution of the ammoniacal nickel cyanide did not
qxiantitatively remove benzene vapor until it had been used for four
or five determinations and had absorbed some of the substance. Hof-
mann and Amoldi used acetic acid in the preparation of the absorbent.
This we have replaced by citric acid because of the appreciable vapor
tension of acetic acid at ordinary temperatures.

Apparatus Employed. — ^The reagent was placed in a Hempel gas pipette
of the form used for the absorption of heavy hydrocarbons by means
of fuming sulphuric acid, the upper bulb that is fiilled with broken glass
being, however, 4.6 cm. in diameter which is somewhat larger than in
the usual Hempel pipette. The measurements of the gas volumes were
made in a Hempel gas burette provided with a water jacket and if changes
of temperatures occurred, corrections were made for them. Numerous
e3q)eriments showed that satisfactory results may be obtained with
either water or mercury as the confining liqtiid in the burette.

Analytical Procedure. — After the sample of gas has been measured in
the burette, the burette is connected by means of the usual capillary
tube with a pipette containing the ammoniacal nickel cyanide solution,
and the gas mixture is repeatedly passed over into the pipette and drawn
back into the burette for a period of about two minutes. The pipette
is then disconnected and the gas is passed into a "fuming sulphuric acid ''
Hempel gas pipette containing a five per cent, solution of sulphuric acid,
this being done to remove the ammonia that enters the gas mixture from
the reagent. The ammonia is not easily absorbed by the dilute sul-
phuric add and the gas mixture must, consequently, be passed into
and withdrawn from the pipette repeatedly for about two minutes to
effect its complete removal.

Experimental Results, — ^The experiments that were made to ascertain
the applicability of the reagent to the absorption of benzene and the
accuracy of the method may be classified under the following heads:

(i) Determination of Benzene in . Known Mixtures of Benzene and
Air;

23,0 L. U. DBNNIS AND ELLEN S, MCCARTHY.

(a) Study of the Behavior of an Ammoniacal Solution of Nickel Cy-
anide toward Known Mixtures of Ethylene and Air;

(3) The Determination of Benzsne in Coal Gas;

(4) The Determination of the "Analytical Absorbing Power" of the
Ammoniacal Solution of Nickel Cyanide;

(5) The Determination of Benzene in Coal Gas that Had Been Freed
from the Benzene that it Originally Contained and Had Been Mixed vitb
Known Volumes of Benzene Vapor; and

(6) The Determination of the Vapor Tension (with Reference to Ben-
zene) of the Partially Exhausted Reagent.

(1) The Determination of Benzene in Known Mixtures of Beraene and
Air. — Air was drawn into the gas burette and measured. It was next
passed into a gas pipette containing mercury and a few cubic centimeters
of liquid benzene, and was then drawn back into the burette and the
increase in volume was noted. The percentage of benzene was then de-
termined by absorption with the ammoniacal solution of nickel cyanide
with the subsequent removal of the ammonia by dilute sulphuric add
in the manner described above. A large number of these analyses was
made. The few results in Table IV illustrate the average efficiency
of the method.

Tabub IV.

(2) A Study of the Behavior of an Ammoniacal Solution of NicM Cy-
anide towards Known Mixtures of Ethylene and Air. — Ethylene was pre-
pared by reducing an alcoholic solution of pure ethylene bromide with
a dnc-copper couple. The ethylene employed in the first experiments
was made in a flask and the resulting gas was found to contain about
90 per cent, of ethylene. In later experiments the preparation was
carried on in a Hempel nitrometer, and the resulting ethylene was then
found to be completely absorbable by ao per cent, fuming sulphuric
add. Mixtures of ethylene and air were prepared and passed over into
the ammoniacal nickel cyanide. In no case was any absorption of ethyl-
ene noted after the reagent had been shaken once with the gas mixture
and thus saturated with it. The data of two experiments are given as
illustrative of the results here obtained: Per cent, of ethylene taken
4.3, found 0.0; again, ethylene taken 10.4, found 0.0.

DSTBRMINATION Ot BBKZBNB IN XLLmONATlNG GAS. 239

(3) The Determination of^Bemene in Coal Gas. — ^A large number of
determiaations of the benzene naturally occurring in coal gas was made
by the above method and the results are given in Table V. It is per-
haps unnecessary to state that here, as in all other absorption methods
in gas analysis, freshly prepared reagents should first be shaken with
the sample of the gas to be analyzed in order to saturate the reagent
with the slightly soluble constituents of the gas mixture. Moreover,
the same sample of reagent should not be used for the analysis of gas
mixtures of markedly different compositions.

Tabls V.

• Tim^ of contact Bensene

Nnmber of with reagent. found

ezpciimcnU Date. Minntes. Percent

33 Jan. 7 4 0'8

34 Jan. 7 2 0.8

35 Jan. 7 a 0.8

4 more 0.8

36 Feb. 13 2 0.7

37 Feb. 13 2 0.7

38 Feb. 13 4 0.7

39 Feb. 13 2 0.7

40 Feb. 15 6 0.6

41 Feb. 15 4 0.6

42 Feb. 15 2 0.6

43 Feb. 18 2 0.8

44 Feb. 18 2 0.8

2 more 0.8

45 Feb. 18 6 0.8

46 Feb. 19 2 0.5

47 Feb. 19 2 0.5

2 more 0.5

48 Jimc 3 2 X.2

49 Jmic 3 2 1.2

50 Jmic 3 2 1.3

51 Jime 3 2 1.2

52 Jmic 3 4 J -2

53 J«ne 3 4 1.2

54 Jmie 3 2 1.2

2 more X . 2

The results tabulated above demonstrate that an ammoniacal solu-
tion of nickel cyanide completely removes the benzene from coal gas
in two minutes. It should be understood that the gas mixture is repeat-
edly passed into and drawn out of the gas pipette during this time.

(4) The Determination of the *' Analytical Absorbing Power** of the Am-
moniacal Solution of Nickel Cyanide. — ^The term "analytical absorb-
ing power" was introduced by Hempel and means one-fourth of the
total volume of the gas that one cubic centimeter of the reagent is able

yo h. M. DENNIS AND ELLEN S. MCCARTHY.

» absorb.' Theoretically, one cubic centimeter of the ammoniacal
tlution of nickel cyanide will absorb about 20 cc. of benzene vapor
leasured iinder ordinary conditions of temperature and pressure. This
>rTesponds to an "analytical absorbing power" of 5 cc of benzene vapor
id inasmuch as the gas pipette holds 200 cc. of the reagent, this volume
lould suffice for the absorption of 1000 cc. of benzene vapor. The
impound that is formed, Ni{CN),.NH3.C,H„ exhibits, according to
le experiments of Hofmann, no measurable vapor tension.

To ascertain whether this theoretical "analytical absorbing power"
as in accord with actual results, the reagent that had been used in
xperiments 48 to 53 inclusive, was shaken with 2 cc. of liquid benzene
univalent to about 500 cc. of benzene vapor) until this was completely
isorbed. This reagent was then employed in Experiment 54 and in Ex-
iriments 68 to 72 inclusive. The completeness of the absorption ob-
lined in each case shows conclusively that the limit of the absorbing
jwer of the reagent had not yet been reached even when one filling
' the pipette had taken up over 500 cc. of the benzene vapor. Furtier
camination of the absorbing power demonstrated that even after a
ipetteful of the reagent had absorbed 800 cc. of the benzene vapor quan-
tative results could still be obtained. Since, however, the compound
' benzene with the reagent separates as a precipitate, it is advisable,
I order to avoid the clogging of the pipette, to filter the reagent after it has
iken up fairly large amounts (from 200 to 400 cc.) of benzene vapor.

{5) Ths Determination of Benzene in Known Mixtures of Coat Gas and
ernene. — These determinations were carried out to ascertain whether
snzene can be determined accurately in the presence of the other gases
lat are usually found in coal gas. The sample of coal gas was first
eed from benzene by treatment with the ammoniacal nickel cyanide
ad then known amounts of benzene were added to it in the maimer
tK>ve described.

In Experiments 64 to 67 inclusive, one sample of coal gas was em-
loyed and to this successive portions of benzene vapor were added,
nd were then removed by absorption with the reagent. The fact that
le volume of the gas residue remained constant throughout these analy-
a demonstrates that the constituents of ordinary coal gas Other than
enzene {with the exception of those gases absorbed by a concentrated
jlution of potassium hydroxide) are not absorbed by the reagent This
anclusion is corroborated by the fact that in Experiments 57, 58, and
1, a prolongation of the contact between the gas and reagent did not
ause further appreciable diminution in the volume of the gas residue.

. ' See Hempel, Methods of Gas Analysts, Translated from 3rd German Editian
y Dennis, page 145.

DBTBRKINATION OI^ BBNZBKB IN nXUMINATING GAS. 24 1

Tablb VI.

Time

of contact Bencene Bensene

Namber of with reagent, taken. found.

ezperimenL Minutes. Percent, Percent.

55 2 I.I I.I

56 2 2.8 2.8

57 4 0.8 0.7

58 8 0.8 0.8

59 2 4.3 4.3

60 2 2.7 2.7

61 12 2.7 2.8

62 2 0.9 0.9

63 2 0.7 0.7

64 2 0.9 0.9

65 2 I.I I.I

66 2 1.3 1.4

67 2 1.3 1.3

68 2 1.4 1.4

69 2 1.2 1.2

70 2 i.o j.o

71 2 2.6 2.7

72 2 3.6 3.6

(6) The Determination of the Vapor Tension (with Reference to Ben-
zene) of the Partially Exhausted Reagent. — ^Although Hoftnann states that
the ammonia benzene nickel C3ranide exhibits no appreciable vapor ten-
don, it was still deemed desirable to ascertain whether the reagent, after
somewhat extended use, would show a measurable benzene vapor ten-
sion. To this end a measured sample of air was brought into contact
with a pipetteful of the reagent that had previously absorbed about
800 cc. of benzene vapor. The sample of air was then drawn back into
the burette, was passed into a pipette containing dilute sulphuric acid,
and was then drawn back into the burette and measured. The results
were as follows:

Vdtune air taken (cc.) 60.7 61.4 74.8 89.2 83.1 73.9

Volmne after contact with'reagent (cc.) 60.7 61 .3 74.8 89.3 83. i 73.9

From these experiments it appears that the benzene vapor tension
of the partially exhausted reagent is so slight as to be negligible.

Dennis and O'Neill* determined experimentally that benzene is not
taken up by the concentrated solution of potassium hydroxide that is
employed in the Hempel method for the determination of carbon diox-
ide, and that consequently absorption of carbon dioxide may properly
precede the determination of benzene. This point has been examined
further by the authors of the present paper who also have found that
potassium hydroxide solution does not absorb benzene from mixtures
of benzene vapor and air. The order of procedure then in the analysis

* Z^. ci«„ p. 507.

(.3 L. U. DENNIS AND ELLBN S. HCCASTHY.

: coal gas should be that recommended by Dennis and O'Neill,' the
isoTption of benzene being effected by tfae ammoniacal solution of
ickel cyanide in place of the solution of ammoniacal nickel nitrate
hich they there suggested.
I. The Uorton Method for the Determination of Benzene in niuminst*

ing Gas.
D. A. Morton has recently described* a simple method for the deter-
ination of benzene by absorption with concentrated sulphuric add
Ip. Gr, 1.84) and before leaving this work it was thought advisable
I examine Morton's method with a view to ascertaining the accuracy

the results that can be obtained through its use. It is self-evident
lat for the determination of benzene in coal gas, there must be empk>yed
1 absorbent that does not remove ethylene. Morton states that the
feet due to the absorption of ethylene by concentrated sulphuric add

in^gnificant. This statement is somewhat surprising, for inasmuch
1 both benzene and ethylene are quantitatively absorbed by fuming
ilphuric add it would seem reasonable to suppose that concentrated
ilphuric acid would also take up ethylene, and that, although it might
)Sorb this gas in smaller amounts than did the fuming sulphuric acid,
le percentages of ethylene removed by it would yet be appreciable.
Lirther, it was deemed possible that the apparently correct results in
le sepamtion of benzene and ethylene obtained by Morton might be
cribed to an acddental bftlandng of errors in the removal of the two
ises in question. In looking up the literature of the subject it was
und that Frankland and Thome,* in working upon "The Luminosity

Benzol" found "that ordinary sulphuric acid absorbs benzol vapor
,ther rapidly, and may be used for its eudiometric determination."
hey were, however, separating benzen^ from hydrogen, carbon monox-
e, and methane, and no ethylene was present in the gas mixture that
as employed. Butlerow and Gorjainow* found that concentrated
ilphuric add will completely absorb ethylene at a temperature of from
io" to 170". Berthelot' states that when ethylene is shaken with
incentrated sulphuric add, 100 grams of the add absorb 6.7 liters
20 volumes) of ethylene, and further that the ethylene that escapes
)sorption upon passage through a Liebig bulb containing fuming sul-
luric add is afterward absorbed by shaking it with ordinary sulphuric
ad. In a paper by Faraday' it is stated that when ethylene was al-
^Loc.cil.p. 511.
■ This Journal, 38, 173* C'9o6).

• J. Chem. Soc., 33, 89 (1878).
' Ber, 6, 196 (1873).

• Ann. chim. phys. [3], 43, 385 (1855),

• PhiL Trans., 1825. p. 448- . t. . . ■ . ■ ..

DSTERMINATION OF BENZENE IN ILLUMINATING GAS. 243

lowed to stand for eighteen days in contact with concentrated sulphuric
add one volume of the add absorbed 84.7 volumes of the ethylene.

Experimental.

The Absorption of Ethylene by Sulphuric [Acid [(Sp. Gr.^1.84). —
Ethylene was prepared in the manner above described. Admixture
of air and ethylene containing 22.5 per cent, of ethylene was made and
was shaken for one minute with about 150 cc. of concentrated sulphuric
add contained in a Hempel simple gas pipette. It was then drawn
back into the burette and the diminution in volume noted. It was
passed back again into the pipette and shaken for another minute and
measured. This was repeated several times and the results are given
in Table VII. It will be noted that 18 cc. of the 22.5 of ethylene present
m the gas mixttire was absorbed in thirty minutes, the rate of absorption
diminishing as the percentage of ethylene decreased.

Table VII.

Length Total

Ethylene of each length of

pretent shaking with coutactwith Total volume

in 100 cc. of Ethylene concentrated concentrated ethylene

Nnmherof gas mixture. absorbed. HtS04. H1SO4. absorbed,

experiment. cc. cc. Minutes. Minutes. cc.

73 22.5 2.3 I I 2.3

74 20.2 1.9 I 2 4.2

75 18.3 1-3 I 3 5.5

76 17.0 1.3 I 4 6.8

77 15.7 1-2 I 5 8.0

78 14.5 1.2 I 6 9.2

79 13-3 o.B 1 7 10. o

80 12.5 0.8 I 8 10.8

81 II. 7 i.o I 9 II. 8

82 10.7 1.0 I 10 12.8

83 9.7 0.8 I II 13.6

84 8.9 4.4 19 30 i8*o

The Action of Concentrated Sulphuric Acid upon a Mixture of Air,
Ethylene and Benzene. — ^A mixture of air, ethylene and benzene con-
taining 1.5 per cent, benzene and 10 per cent, ethylene was prepared
and was shaken with concentrated sulphuric acid in the manner that
Morton describes, except that in Analyses 89 and 91, the shaking was
continued for two minutes instead of only one minute. The results
are given in Table VIII. It will be seen that of a total of 11.5 cc. of
hydrocarbons in 100 cc. of the gas mixture, 8.2 cc. were absorbed in nine
minutes. There was but 1.5 per cent, of benzene present in the original
mixture and yet 4.3 per cent, of the hydrocarbons was absorbed in the
first minute of shaking with the sulphuric acid, demonstrating that at
least 2.8 per cent, of ethylene was removed by the reagent. If the ana-
lyst is to regard the absorption by sulphuric add as indicative of the
per cent, of benzene present, the actual error in this case would be very

244

L. H. DENNIS AND BLLBN S. MCCARTHY.

large. By further shaking of this mixture with the concentrated sul-
phuric add from 0,3 per cent, to 0.7 per cent, more of the hydrocarbons
was absorbed per minute.

Tabi* VIII.

Lcaglli of Total length

rateid fajdrocaibiMu

cc. EU mixture. ab«OTb«d.

H,S04.

BtSO,.

9»

The Action of ConcerOraUd Sulphuric Acid upon Mixtures of Air -wiA
Varying Amounts of Ethylene and Benzene. — In this series of experimeots
the gas mixture was first shaken for one minute with the concentrated
sulphuric add and was then passed into a pipette containing ammoniacal
nickel cyanide solution to ascertain whether all of the benzene originally
present had been absorbed by the sulphuric add. An examination of
the results in Table IX shows that the gas absorbed by the concentrated
sulphuric add is not benzene alone, for upon comparing the figures in
the third and fourth columns it will be seen that in Experiments 96 to
Tabls IX.

Bthvlene

HTdrocarboni

■Wirbed In

one minute tiy

■nlpburie icld.

■burbed by mni-
monUcal nickel

cyanide noIntlaB.

• DETBRIdlNATION 01^ BBNZENB IN ILLUMINATING GAS. 245

III, with the exception of the looth and 102nd, the volume of gas ab-
sorbed is greater than the volume of benzene taken, in many cases con-
siderably greater. Moreover, the absorption of the benzene by the
concentrated sulphuric add is frequently incomplete, as is demonstrated
by the results in the last column. In some cases the measurements
showed that the volume of gas absorbed by the sulphuric acid is greater
than the volume of benzene added, and yet even in these cases not all
of the benzene that had been added was removed by the sulphuric acid.
(See in particular, Experiments 9^, 104, and 105). The results in Table
IX would further demonstrate that with so large and so irregular varia-
tions a correction factor for the absorption of ethylene as suggested
by Morton would be of no value.

The Determination of Benzene in Coal Gas by means of Concentrated
Sulphuric Acid. — ^A series of determinations (Table X) of benzene in
coal gas was made by the Morton method. At the time when these
analyses were made the coal gas showed by the ammoniacal nickel cyan-
ide method an average content of less than one per cent, of benzene.
The results by the Morton method are in every case much too high and
furnish further proof that concentrated sulphuric acid removes other
constituents than benzene.

Tabls X.

Length of con tact Hydro-

with concentrated carbons "

Namber of sulphuric acid. absorbed.

experiment. Date. Minutes. Per cent.

112 Feb. 14 I 2.0

113 Feb. 14 I 1.9

114 Feb. 14 2 2.2

115 Feb. 14 4 2.8

116 Feb. 15 I 1.7

117 Feb. 15 I 1.9

118 Feb. 15 I 1.8

119 Feb. 15 2 2.1

120 Feb. 15 4 2.3

The Effect in the Morton Method of Varying the Speed of Shaking. —
Morton directs that the gas be passed into the pipette containing the
concentrated sulphuric add and be shaken vigorously (twice per second)
with this reagent for one minute. The experiments in Table XI were
made to ascertain whether the amotmt of absorption by the sulphuric
add would vary if the speed at which the pipette is shaken is varied.
With acceleration of the shaking slightly higher results were obtained,
but they can not be taken as conclusive upon the point at issue, because
the concentrated sulphuric acid when shaken foams to so great an ex-
tent as to render it difficult to obtain correct measurements of the resid-
ual gas volume. Moreover, minute gas bubbles become suspended |

34-6 L. H. DENNIS AND BLLBN S. UCCAStHY.

through the sulphuric add and do cot all of them rise to the surface of
the reagent even after that has stood undisturbed for two minutes.
Tablb XI.

LcDSIIi orcoDUct Number HTdroattooi
Number of with retKCnt. ofiluke* abiorbed.

To avoid the difficulty that results from the foaming of the reagent,
due to shaking, the concentrated sulphiuic add was placed in a ptpette
of the type used for fuming sulphuric add.' In the first detenninaUon
the sample of coal gas was measured in the burette and was then passed
into a pipette and dmwn back and measured. In the second determina-
tion a fresh sample of the gas was taken and was passed into and drawn
out from the pipette twice. In the next determination another sample
was run into the pipette three times, and so on. The coal gas used in
Experiments 127 to 134 inclusive, was found to contain by the ammonia-
cal nickel cyanide method 0.8 per cent, benzene. The sample of gas
on the following day, February 19th, Experiments 135 to 141 includve,
contained 0.5 per cent, benzene.

Tablb XII.

Number ot titde*

^1 mliture Hydrctarbom

Number of paned Iota ubaorbed.

uperimcat. Date. pipette. Percent.

la? Feb, 18 I I.I

laS Feb. 18 a I. a

139 Feb, iS 3 1.4

130 Feb. 18 3 i.o

131 Feb, 18 3 1 .6

131 Feb, 18 4 i.s

133 Feb, iS 6 1.6

134 Feb. 18 10 a.o

13s Feb. 19 1 o.a

136 Feb. 19 a 0.4

137 Feb. 19 3 0,7

138 Feb, 19 4 i.o

139 Feb. 19 5 1.7

140 Feb. 19 8 a.o

141 Feb. 19 II a. I

The results given in this table show wide variations in the apparent

amount of benzene present and demonstrate that the method will not
give accurate results even when the foaming is avoided by the use of
the fuming sulphuric add pipette.

' See Hempel's Gas Analysis, page 339. ...._-

BWECT OF COAL GAS ON WROUGHT IRON PIPE. 347

Sttmmary,

An ammoniacal solution of nickel cyanide prepared according to the di-
rections given in this article will quantitatively absorb benzene from mix-
tures containing benzene and air, and from ordinary coal gas. It will not
absorb measurable quantities of ethylene or of the other constituents of
ordinary coal gas, with the exception of those absorbable in potassium hy-
droxide solution.

The method proposed by Morton for absorbing benzene by means
of sulphuric add (Sp. Gr. 1.84) does not give constant results even when
the conditions are the same, and yields widely var3dng results when the
conditions are changed. Moreover, it absorbs both ethylene and ben-
zene, and from mixtures containing both of these gases it does not quan-
titatively remove the benzene but does remove an indeterminate amount
of ethylene. Moreover, in the manipulation suggested by Morton, the
reagent foams to such an extent as to make the accurate reading of the
gas volumes well nigh impossible.

CoaitBz.L Univbkutt,
November, 1907.

THE EFFECT OF COAL GAS OH THE CORROSION OF WROUGHT

IRON PIPE, BXTRIED IN THE EARTH,^

BT WM. t,. DUDUIT.

Received November 14, 1907.

This investigation was undertaken in connection with a study of the
conditions causing the corrosion of pipe laid under the streets in the
City of Nashville.

Five samples of earth were collected as representative of the various
types in which the pipes are laid.

Sample No. i was taken from a street at the depth of the gas pipes.
It was an old filL Sample No. 2 was taken from about 4 feet from the
surface on private property in the heart of the city, 50 feet from the side-
walk, and was a yellow clay. Sample No. 3 was taken from between the
car tracks just under the street metal. It was a loamy clay. Sample
No. 4 was a clay taken from a vacant lot in the northern part of the city.
Sample No. 5 was taken from under the car tracks in the southern part
of the city; and ¥ras a mixture of clay and loam.

The anal3rses of the samples gave the following results in percentages:

Sample. No. i. No. a. No. 3. No. 4. No. s*

Moistitre (aampk air-diied). . . 15.37 15.82 14.41 i3-95 14-98

Chlorine 0.04304 0.01501 0.0018 0.0006 0.0024

Nitrogen, as nitrates 0.01501 0.22523 0.006 0.003 0.0015

Nitrogen, as nitrites 0.00008 0.000002 trace none trace

Nitrogen, as ammonia 0.000306 0.000165 o.ooooix 0.000039 0.000042

* Bead before tb^ J«^f ir York Section of the American Chemical Society.

a^ WM. L. DUDI.BV.

Stmple. Ho. t. No. i. No. 3. No. 4- No. 5.

Albuminotd ammonia o.ooizS 0.000053 0.00003 0.000031 o.ooooii

AtkaU equivalent (Na^O^.. . 0.0383 0.0478 0.0334 0.0057 0.031s
Sulphuric anhydride (SO^.... 0.0134 0,019 0.0448 0.0073 0.0434

Humus 3.04 3.40 3.05 1.4 3.7

Wrought iron pipes, one inch inside diameter and ten and a half inches
long, were carefully cleaned and freed from all oxide by immersing in a
warm ammoniacal solution of ammonium citrate for twenty minutes
(or longer if necessary), brushing with a stiff brush, rinsing with distilled
water, and drying rapidly.

The ammonium citrate solution is made by dissolving 40 grams of
citric add in water and neutralizing with the requisite amount of anuno-
nium hydroxide of sp. gr. 0.896, required by the equation:
C;,H,0,.H,0 + 3NH,0H - Etc.

The solution is then diluted to 460 cc. and contains approximately
10 per cent, of ammonium citrate. The citrate solution acts more rapidly
on the iron oxide when warmed and has very little effect on metallic iron.
A clean and bright piece of iron, having a surface of about 15 square
inches, was allowed to remain in the solution for 24 hours, and no effect
on the surface was observed nor was any appreciable loss of weight noted.

The pipes were weighed and plugged at one end with corks driven
in about one-eighth of an inch beyond the edge, and this space was filled
with melted paraffin to keep the moisture in the earth from getting inside.

The samples of earth were each put in a wooden box of one cubic foot
capacity, painted inside with asphalt paint to prevent the soluble con-
stituents of the wood from permeating the earth. The pipes were placed
in an upright position in the boxes (with the closed ends down) and the
samples of earth were carefully tamped around them to within one inch
of the top.

The earth in each box was to have been sprinkled daily with 50 cc
of water, so as to maintain natural underground conditions as far as
possible, but the person in charge of this series of experiments neglected
this precaution during a large part of the time. However, as each re-
ceived the same treatment, the results are comparable, although they
are all too low, except in the case of sample No. 5, which will be referred
to later.

At the end of twelve months the pipes were removed, cleaned with
the citrate solution, as previously described, and weighed.

The results are as follows:

Original weight ttnal weljcht V>h of

Sample of pipe. of pipe. wdgln.

oT earth. On ma. Grami. Grama.

No. I 61449 603.93 *o-56

" 2 652.45 638.91 13.53

" 3 665.76 663.30 3.56

" 4 680.81 678.30 3.51

" 5 618.90 606.50 13.40

EFFBCT 01? COAL GAS ON WROUGHT IRON PIPE. 249

The results, except that for sample No. 5, show that the amount of
corrosion is determined practically by the chlorine content in the earth.
There is some error in the result obtained from sample No. 5 which I
am not able to discover, but that the result is entirely abnormal will be
shown clearly later on.

The series of experiments designed to show the effects of coal gas on
the corrosion of the pipe was conducted under the specified conditions
in every detail. Into each box was placed an L, made of one-half inch
gas pipe, one limb passing vertically down one comer of the box and the
other along the bottom to the diagonal comer. The limb along the bot-
tom was perforated on the under side with numerous one-eighth inch
holes uniformly distributed and the end was plugged. The gas was ad-
mitted to this pipe through a standard meter reading to one-htmdredth
of a cubic foot.

One-half of a cubic foot of coal gas was admitted into each box daily,
except Stmdays, and at the same time the earth was moistened by sprink-
ling with 50 cc. of water. The boxes were covered with canvas so as to
prevent the moisture from evaporating too rapidly.

In the samples of earth Nos. 2, 3, 4 and 5, two pipes were placed and
the amount of corrosion of each pair shows very good agreement. The
results, after twelve months, are as follows:

Original weight Pinal weight I^osa of Average loss

Sample. of pipe. of pipe. weight. of weight,

ofearth. Grams. Grams. Grams. Grams.

No. I 661.53 652.36 9.17 9.17

< 658.30 651.77 6.53? ^
) 654.07 647.92 6.I5J

M J 663.60 662.45 1.25

•* ^668.71 666.88 1.83' ^^

u . J 665.62 664.37 1.25?

^ J. 671.45 670.36 1.09*

" 5 i!f3-«' tV'^ '■^\ «.96

'^ {663.14 661.02 2.I2J

The following table shows the comparative amoimt of corrosion of
the pipe in the earth alone and in the earth saturated with coal gas:

Loss of weight I/Oss of weight in

Sample in earth alone. earth saturated with gas.

ofearth. Grams. Grams.

No. I 20. 56 9. 17

" 2 13.53 6.34

3 2.56 1.54

4 2.51 1. 17

" 5 (12.40) 1.96

The effect of the gas in retarding corrosion of wrought iron pipe is very
marked, especially when we consider that the results showing the corro-
sion in the earth ^lone are all low (except No. 5, which is manifestly

250 WM. L. DUDLEY.

wrong) because of the failure to maintain the proper amount of moisture
in the earths throughout the whole period of the test

The accompanying chart shows the curves, indicating the relation-
ship between the amount of corrosion and the chlorine content; abo the
error in No. 5. In the series of experiments with the coal gas, Sample
No. 5 shows the normal amomit of corrosion. In the series with the
earth alone, No. 5 should have shown about 3 instead of 13.40 grams
of corrosion. My opinion is that No. 5 was inadvertently subjected
to electrolysis, as the appearance of the pipe indicated it and electrolytic
experiments were being carried on at the same time.

No. 2 shows a rather high corrodve power in both series of experiments,
which is doubtless due to the abnormally high content of nitrogen as
nitrates in this sample of earth. It is said that the location from which
Sample No. 2 was taken was formerly the dte of a livery stable, which
would account for the high percentage of nitrates.

CHEMICAL EXAMINATION OI^ MICROMERIA CHAMISSONIS. 25 1

[Contribution prom thb Wsu^comb Chbmical Rbssarch Laboratories, London.]
CHEMICAL EXAMINATION OF MICROMERIA CHAMISSONIS.

(Yerba Buena.y

Bt pRBDBiiicK B. Power aicd Arthur H. Salway.
Received Noyember ao, 1907.

The labiate plant Micromerta Chamissonis (Benth.) Greene (syn*
M. Dougldssi Benth.), commonly known as ** Yerba Buena," is a peren"
nial, trailing or creeping, sweet-scented herb, which is indigenous to the
Pacific coast of the United States. The generic name is stated to be
derived from the Greek mikroSy small, and meros, part, on account of
the small size of the flowers (compare Jepson's ** Flora of Western
Middle California," p. 463). A description of the anatomical char-
acters of the plant, with illustrations, has been given by J. Moeller.*

The above-mentioned species of Micromerta is used to some extent
medicinally, but, so far as known to us, it has never been the subject of
chemical investigation, and we have therefore availed ourselves of the
opportunity of making a complete examination of its constituents.

Experimental,

The material employed in this investigation consisted of a bale of the
entire, air-dried herb, which was kindly placed at our disposal by Messrs*
Burroughs, Wellcome & Co., of London. The genuineness of the ma-
terial was assured by the fact that it had been specially collected for
them in California under the supervision of a competent botanist, Mr.
P. E. F. Perr^dfes, B.Sc., F.L.S.

Distillation of the Plant with Steam. Characters of the Essential Oil.

A quantity (56V4 pounds = 25^/, kg.) of the herb was submitted to steam
distillation by Messrs. Stafford Allen & Sons, of London, and our thanks
are due to them for having kindly conducted this operation for us. The
amount of essential oil obtained was 42.3 grams, corresponding to 0.16
per cent, of the weight of air-dried plant. Towards the end of the dis-
tillation a small quantity (8.5 grams) of a semi-solid substance separa-
ted in the condenser and receiver. This was separately collected and
examined.

The essential oil had a pale yellowish brown color and an agreeably
aromatic, somewhat mint-like odor. Its density was 0.9244 at 20®,
and its optical rotation — 22^48' in a i dcm. tube. It was not com-
pktely soluble in ten times its volume of 70 per cent, alcohol, and gave
no coloration with ferric chloride.

The above-mentioned, semi-solid substance was spread on a porous
plate, when the oily constituent was absorbed, and a white, crystalline

* Read before the New York Section of the American Chemical Society, Nov. 8,
1907.

* Am. J. Pfaarm., 1883, 54, 461, from Pharm. Centralhalle, 1883, No. 29.

252

FREDERICK B. POWER AND ARTHUR H. SALWAY.

*;f '

: I : ■

solid remained. This was crystallized from acetic acid, from which it
separated in pearly leaflets, melting at 61.5°. It was analyzed with
the following result:

0.0914 gave 0.2506 COj and 0.1073 H,0. C = 74.8; H = 13.0.

CuHjjO, requires 0 = 75.0; H=»i2.5 per cent.

This solid substance was thus identified as palmitic add.

Examination of the Alcoholic Extract of the Plant.

For the purpose of a complete examination of the constituents of the
plant, 36 poimds (16*/, kg.) of air-dried material were extracted by con-
tinuous percolation with hot alcohol. After the removal of the greater
portion of the alcohol, a dark green, thick extract was obtained, which
weighed 4170 grams. Of this extract, 3.5 kg. were mixed with water,
and the mixture distilled with steam until all the volatile substances
present had been removed.

Volatile Constituents of the Alcoholic Extract.

The distillate obtained, as above described, had a distinctly acid re-
action and contained a quantity of essential oil floating on the surface.
It was extracted twice with ether, after which the ethereal liquid was
washed with a little water, dried with anhydrous sodium sulphate, and
the ether removed. About 20 grams of a pale yellowish brown essential
oil, possessing an aromatic, mint-like odor, were thus obtained. When
distilled under a pressure of 25 mm. it passed over between 80® and ito*',
but the greater portion boiled between 120® and 140°. The oil had a
density of 0.9450 at 20®, and an optical rotation of — 26° 44' in a i dcm.
tube. It was readily soluble in 70 per cent, alcohol, thus diflfering from
the essential oil distilled directly from the dried plant.

The adds contained in the aqueous distillate from which the essential
oil had been removed were neutralized with baryta, and the liqtiid con-
centrated. About 3 grams of a barium salt were thus obtained in the
form of a thick syrup, which afforded reactions indicating the presence
in the distillate of formic, acetic, and butyric adds, the latter bdng
relatively small in amount.

Non-volatile Constituents of the Alcoholic Extract.

After the removal of the essential oil and volatile adds by distillation
with steam, as above described, there remained in the distilling vessel
a reddish-brown, aqueous liquid (A) together with a large quantity of
a dark green, soft resin (B). The resin was separated by filtration from
the liquid, while still hot, and washed well with hot water, the washings
bdng added to the filtrate.

Examination of the Aqtieous Liquid (A).
The aqueous liquid, which was of a reddish brown color, and pos-

CHEMICAL BXABflNATlON OF MICROMERIA CHAMISSONIS. 253

sessed a slightly bitter taste, was concentrated to a convenient bulk
by heating under diminished pressure in a bath of hot water.

Isolation of a New Phenolic Substance^ Xanthomicrol, C^H^fiJipH)^*

The concentrated aqueous liquid was extracted repeatedly with ether,
and the several portions finally mixed. This ethereal liquid was washed
with a little water, and then extracted with small successive portions
of a saturated solution of sodium carbonate. The first few extractions
with alkali afforded opaque liquids containing small amounts of a dark
green, semi-solid sodium compound of a resinous nature, but neither
from this nor from the products which separated on acidifying these
liquids could anything crystalline be obtained. The alkaline liquids
obtained by the subsequent extractions were, on the other hand, of an
Qrange-yellow color, and, when acidified, yielded orange-yellow precipi-
tates, which were collected, washed with water, and crystallized from
alcohol A substance was thus obtained in the form of fine, lemon-
yelbw, silky needles, melting at 225°, which was analyzed with the fol-
lowing result :

0.1114 gave 0.2535 CO2 and 0.0453 H3O. C=62.i; 11=4.5,

After another crystalUzation the melting point of the substance re-
mained unchanged, and it was again analyzed:

0.1099 gave 0.2522 CO, and 0.0450 H,0. C = 62.6; H==4.5.

C„HuOj requires C = 62.5; H=4.2 per cent.

The only known compound of the empirical formula CuHuO,, which
has properties similar to those of the above substance, is the so-called
*'datiscetin,*' a hydrolytic product of datiscin (compare Beilstein's Hand-
Imch der org. Chemie, Bd. III., p. 580). Datiscetin, however, is stated
to melt at 237®, and its solution in sulphuric acid, which has a yellow
coter, shows a blue fluorescence. Although the above-described sub-
stance (m. p. 225°) also dissolves in concentrated sulphuric acid with a
yellow color, its solution shows no fluorescence. It thus appears not
to be identical with any substance hitherto recorded, and it is therefore
proposed to designate it xanthomicrol, with reference to its yellow color,
its phenoUc properties, and the generic name of the plant from which
it has been obtained.^

Xanthomicrol is readily soluble in alcohol, ethyl acetate, and acetone,
and moderately soluble in ether, but only sparingly so in chloroform
and benzene. It is practically insoluble in cold, but slightly soluble in
hot water. It dissolves readily in cold aqueous alkalis, forming solu-

' According to the recent investigations of Korczynski and Marchlewski (Chem,
Ceniraib., 1906, II., 1265 and 1907, II., 700) the substcmce designated as datiscetin,
when purified, possesses the empirical formula C,5H,oOe, melts at 268-269^, and
conUdns four hydroxyl groups. It would thus be essentially different in its 00m-
position and character from xanthomicrol.

14 FREDEtaCK B. POWER AND ARTHUR H. SALWAY.

3ns which are yellow when dilute and orange-red when concentrated.

is, however, not acid towards litmus, and the sodium compound, pio-
iced by shaking an ethereal solution of the substance with the tbeo-
tical quantity of sodium ethoxide, is readily decomposed on exposure
I the atmospheric moisture and carbon dioxide. The property which
is substance possesses of combining with alkaUs is therefore to be at-
ibuted to the presence of one or more phenolic groups.
Di-acetylxanikomicrol, CuH,bO,{CO.CH^,. — A small quantity of xantho-
icrol was boiled for a short time with acetic anhydride, the greater part

the latter removed by distillation, and the residue poured into water,
lie product of the reaction, which soon solidified, was colIected„wasbed,
id dried. It was then crystalUzed from ethyl acetate, from which it
parated in pale yellow needles, melting at ii6.°

The number of acetyl groups in this compoimd was determined by
rdrolymtg weighed portions of the substance with a solution of sodium
Mroxide, subsequently acidifying with sulphuric add, and distilling
itil add ceased to pass over.

(I) 0.1763 gave an amount of acetic add equivalent to 9.4 cc N/io
aOH.

(II) 0.17S9 gave an amount of acetic add equivalent to 9.7 cc N/io
aOH.

(I) CH,.C0 = 2J.9; (II) CH,.CO-23.3.
C«H,(,0,(CO.CH,), requires CH,.C0 = a3.i per cent.
Di-acetylxantkomicrol is extremely soluble in most of the usual organic
Ivents, but insoluble in water. It is immediately hydrolyzed by aUta-
., with the reproduction of xanthomicrol, CuH„0„ melting at 325°.
The ethereal liquid from which the above- described new phenolic sub-
ince had been obtained, and from which nothing further was extracted
1 shaking with a solution of sodium carbonate, was washed with a little
iter, dried with caldum chloride, and the ether removed, when only a
ly small amount of an uncrystallizable resin remained.
The aqueous hquid {A) which bad been extracted with ether, as above
scribed, was found to give a deep yellow predpitate with basic lead
etate. A slight excess of the latter was therefore added, the predpi-
te collected, washed, suspended in water, and decomposed with hydro-
n sulphide. On filtering the mixture an orange-red hquid was obtained,
lich gave a greenish black color with ferric chloride, thus indicating
e presence of tannin, but nothing of a crystalline nature could be
)lated from it.

The filtrate from the basic lead acetate predpitate was treated with
drogen sulphide for the removal of the lead, and filtered. A portion
this liquid, when treated with phenylbydrazine acetate, readily
:lded a crystalline osazone melting at 217°, and it therefore evidently

CHBMICAI^ EXAMINATION OP MICROMERIA CHAMISSONIS. 255

contained a considerable quantity of glucose. The remainder of the
filtrate was concentrated to a small bulk under diminished pressure,
when a thick, dark-colored syrup was obtained, from which nothing of
a crystalline nature separated, even after standing for several months.

Examination of the Resins (B).

The resin which had been separated from the aqueous liquid, in the
manner previously described, was dried at loo^, and then weighed 480
grams, corresponding to 3.5 per cent, of the weight of air-dried plant.
This resin was intimately mixed with purified sawdust, the mixture thor-
oughly dried, and then extracted in a Soxhlet apparatus with various
solvents, when the following amounts of extract, dried at 100^, were ob-
tained:

I. Petroleum (b. p. 40-60^) extracted 177. 12 grams — 36.9 per cent.

Ether

178.08 «

- 37- 1

TTT.

Chloroform

24.00 *

- 5.0

IV.

Bthyl acetate

10.56 «

— 2.2

Alcohol

48.00 «

— 10. 0

Total 437 -76 grams ■- 91 . 2 per cent.

I. Petroleum Extract of the Resins.

This was a soft solid, of a dark greenish color. It was boiled for two
hoars in alcoholic solution with 40 grams of potassium hydroxide, when
a small quantity of ammonia was evolved. After this treatment the
greater portion of the alcohol was removed, water added, and the alka-
line liquid repeatedly extracted with ether. The ethereal extract, which
was of an orange-red' color, was washed with a little water, dried with
calcium chloride, and the ether removed, when an orange-colored, semi-
solid residue was obtained. This was distilled under diminished pressure,
(25 mm.), in order to remove a small quantity of resinous matter, when
it passed over between 240® and 350®. The distilled product was
tiighly fluorescent, both in the fused state and in solution, and, on cool-
mg, it solidified to a wax-like solid. On fractionally crystallizing the
latter from ethyl acetate, it was found that the greater portion consisted
of a compound which separated in small, glistening, pearly leaflets, and,
after a few crystallizations, was obtained in a pure state. It then melted
at 66-7^, and was analyzed :

ai420 gave 0.4434 CO, and 0.1822 £[,0. C =-85.2 ; H = 14.3.

CjjH^ requires 0=85.3; H=»i4.7 per cent.

This crystalline substance was thus identified as hentriacontane.

The mother liquors from the crystallization of the hentriacontane
yielded a further small quantity of a hydrocarbon, which, after repeated
fractionation, melted at 71-4®, and probably consisted of pentatria-
oontane. The more readily soluble fraction that was finally deposited

256 FREDERICK B. POWER AND ARTHUR H. SALWAV.

condsted of a mixture of hentriacontane with a substance which ciys-
tallized in large, thin plates. The latter substance was mechamcally
separated and recrystallized from a mixture of ethyl acetate and alco-
hol, when it was obtained in the form of pearly leaflets, melting at 135".
This compound, when dissolved in chloroform, gave, with a little acetic
anhydride and a drop of sulphuric acid, the color reaction characteris-
tic of the phytosterols,

0.1590 of the air-dried substance, heated to 105°, lost 0.0074 ^fi-
H,0=.4-7-

0.14S0 of anhydrous substance gave 0.4536 CO, and 0.1633 H,0.
C = 83.6; H = i2.2.

C„H„0, H,0 requires H,0=4.5 per cent.

CjjH^O requires 0 = 83.9; H = ii.9 per cent.

The ultimate mother-liquors, from which the above-mentioned hydro-
carbons and phytosterol had been isolated, were collected and the solvent
removed. A considerable quantity of a thick yellow oil was thus obtained,
which was distilled under diminished pressure, and the product collected
in two fractions. The fraction of higher boiling-point sohdified on cool-
ing, and a further quantity of phytosterol was isolated from it, whereas
that of lower boiling-point consisted of a thick, gummy substance, having
an unpleasant odor.

The alkaline, aqueous liquid, from which the at>ove- mentioned neutral
substances had been extracted by ether, contained an insoluble potas-
sium salt, which was collected on a filter and separately examined. The
salt was boiled with alcoholic sulphuric add, in which it was soluble,
and the liberated acid precipitated by the addition of water. The acid,
after being collected and washed, was crystallized from alcohol, but, as
it separated in a somewhat indefinite form, it was distilled under dimin-
ished pressure and then crystalUzed from ethyl acetate. It was thus
obtained in the form of small, glistening plates, which, after a few re-
crystallizaUons, melted constantly at 80-3°.

0.0714 gave 0.2034 CO, and 0.0830 H,0. C = 77.7; 11 = 12.9.

CyH„0, requires 0=77-6; H = 12.9 per cent.

This substance was evidently behenic add. The mother-Uquors from
the crystallization of the latter gave deposits melting below 70", but
these were too small in amount to permit of further purification.

The alkaline liquid, from which the above-described potassium salt
had been removed, was acidified with sulphuric add and distilled with
steam. The distillate contained a very small amount of volatile add
which, after conversion into a barium salt, afforded reactions indicating
it to consist chiefly of butyric add, with traces of formic and acetic adds.
The contents of the distilling flask were subsequently extracted with
ether, the ethereal solution being washed, dried, and the ether removed,

CHBMICAL EXAMINATION OF MICROMBRIA CHAMISSONIS. 257

when a quantity (88 grams) of a dark green, soft solid was obtained.
This evidently contained a considerable amount of/ resinous matter,
which was separated by washing with light petroleum, in which the resin
was insoluble. The petroleum washings, after the removal of the solvent,
yielded a quantity (46 grams) of a green, wax-like solid, which was dis-
tilled under diminished pressure. The larger portion (37 grams) passed
over between 230° and 250® at 20 mm., while about -3 grams were col-
lected above 250®. Both fractions, on cooling, solidified to 'a wax-like
mass. The larger portion "was fractionally crystallized from alcohol,
when from the least soluble fractions a substance was finally isolated
which melted at 71-3**, and this melting-point was not altered by further
crystallization.

0.0780 gave 0.2193 CO, and 0.0905 HaO. 0 = 76.7; H = i2.9.

CjoH^oO, requires 0 = 76.9; H = 12.8 per cent.

This substance was thus identified as arachidic add.

The mother-liquors, which still contained a large proportion of the
fatty adds present in the mixture, gave deposits from which, by long-
continued fractionation, a substance was obtained which melted quite
constantly at 55-7°, and this was analyzed.

a 1522 gave 0.4182 CO, and 0.1740 HjO. 0 = 74.9; H = i2.7.

CmHjjO, requires € = 75.0; H = 12.5 per cent.

It is evident that this substance was nearly pure palmitic add.

From the portion of mixed fatty adds distilling above 250^/20 mm.
a very small amount of a substance was isolated which melted at 77-
&^. This apparently consisted of slightly impure behenic add, which,
as already noted, had been isolated from the insoluble potassium salt
formed in the hydrolysis of the petroleum extract. The mother-liquors
from the above-described compounds did not decolorize bromine, and
no unsaturated adds were therefore present in the original mixture.

II. Ether Extract of the Resins.

This was obtained in the form of a light green powder and amounted
to 178 grams. It was digested with a large volume of ether, which dis-
solved about one-half of the total amount of extract. The sparingly
soluble portion was separated by filtration from the ethereal solution,
washed with a little ether, and the two portions independently investi-
gated.

Isdaiian of a New Crystalline Alcohol, Micromerol, C^H^fi^-0Ht2Hfi.

The portion of the ether extract of the resins which remained undis-
solved by the above treatment was dissolved in alcohol, in which it was
only moderately soluble, and the solution heated with animal charcoal
under a reflux condenser for several hours. By this means the dark green
coter of the liquid was completely removed. After filtering the solu-

1 FREDERICK B. POWER AND ARTHUR H. 5AI,WAY,

1 it begas to deposit a quantity of a crystalline substance in the form
fine, colorless needles. The crystals were always accompanied by a
itively small amount of amorphous, fluffy matter, but it was found
t this could be removed by filtration through fine muslin, which re-
led the crystalline substance while the amoiphous matter was carried
ough the filter with the mother-liquor. By repeating this operation
crystalline substance was obtained in a pure condition, and furthec
intities of it were deposited on concentrating the mother-liquors.
; compound, when rapidly heated, melts at 277°. An analyds of it
■e the following results:

1.9268 of air-dried substance, when heated for 2 hours at 105**, lost
567 H,0. H,0-7-a-

I) 0.1291 of anhydrous substance gave 0.3652 CO, and 0.1301 H,0.

II) 0.1338 of anhydrous substance gave 0.3934 CO, and 0.1276 H,0.
I) C-77-i; H = io.3. (11) C-77.3; H = io.2.

^HuO(.3H,0 requires H,0 -6.6 per cent.
I^HjiO, requires C = 77.3 ; H - 10. i per cent.

Ls no substance of the empirical formula CnH„0„ having the proper-
of this compound, has hitherto been recorded, it is proposed to desig-
e it micromerol, with reference to the generic name of the plant from
eh it has been obtained.

licromerol is only moderately soluble in cold, but readily in hot alco-
, and is therefore easily crystallized from this solvent. It is also mod-
tely soluble in ethyl acetate and in acetic acid, but only sparingly so
icetone, chloroform, and benzene, while in water and in light pet|p-
n it is insoluble. It is likewise insoluble in aqueous alkalis, but if
ttle micromerol be dissolved in ether, and the ethereal solution shaken
li a solution of sodium hydroxide, a white, insoluble sodium com-
nd is at once precipitated. The latter compound, however, is very
table, being decomposed through the influence of atmospheric carbon
dde with the liberation of the original substance. On the other hand,
lOUgh the micromerol is insoluble in aqueous alkahs, it immediately
ses into solution if a few drops of alcohol are added, presumably as
impound soluble in alcohol, for the addition of a drop of hydrochloric
1 reprecipitates the original substance, whereas the addition of water
s not throw it out of solution.

licromerol is optically active, and its specific rotatory power was de-
nined with the following result :

.1609 gram of the anhydrous substance, in 50 cc. of absolute alcohol,
e ap -|-o''22' in a 2 dcm. tube, whence [ajj, +57.0°.
licromerol is an extraordinarily stable substance. When fused with
issium hydroxide at a temperature of 270-80" it may be recovered
the most part unchanged. A strong solution of chromic anhydride

CHEMICAI, EXAMINATION OP MICROMERIA CHAMISSONIS. 359

in acetic add does not attack it in the- cold, and only slowly on heating.
A strongly alkaline solution of permanganate very gradually acts upon
it in the cold. It is not altered by bromine in chloroform solution, even
after heating for several hours, and, therefore, evidently contains no
unsaturated Unkings. The presence of one hydroxyl group in micro-
xneiol was proved by the formation of both a monoacetyl and a mono-
methyl derivative.

Acetylmicramerol, C„Hji04(C0.CHg). — ^This was prepared by heating
the micromerol for a short time with an excess of acetic anhydride,
then distilling off the greater portion of the latter, and pouring the prod-
uct into water, when it solidified immediately. Some difficulty was
experienced in crystallizing the product, for it was found that alcohol,
which appeared to be best adapted for this purpose, hydrolyzed it
quite readily, and unless the crystallization was conducted very rapidly
an impure acetyl derivative was obtained. In most of the other usual
organic solvents it is extremely soluble. It was ultimately ascertained
that the substance could be crystallized from light petroleum contain-
ing a small proportion of ethyl acetate, whereby all danger of hydrolysis
was averted. From the latter mixture of solvents it separated in fine,
colorless needles, melting at i88^.

ao823 gave a2272 CO, and 0.0726 H,0. € = 75.3; H = 9.8.

C^HjiO^CCO.CHJ requires C= 75.8; H=9.7 per cent.

CaH^04(C0.CHg), requires C « 74.5 ; H = 9.4 per cent.

For further confirmation respecting the number of acetyl groups in
this compound, a weighed quantity of it was hydrolyzed by boiling with
aqueous alcohol, and the acetic distillate collected in a standard solution
of alkali, which was re-titrated.

a326i gave an amount of acetic add requiring 6.1 cc. N/io NaOH
for neutralization. CH3.CO » 8.0.

a326i gave, on hydrolysis, 0.3002 of C^tfi4' CH,.CO = 7.7.

CaHjP^CCO.CH^ requires CH,.CO = 7.8 per cent.

The optical rotatory power of this acetyl derivative was determined
with the following result :

ao878 gram, dissolved in 25 cc of chloroform, gave a^ +0® 20' in a
2 dcm. tube, whence [a]^ +47.1°.

Mdhylmicromerol, C3,H5i08.0CH„H20. — ^This was prepared by heat-
ing a solution of micromerol in absolute alcohol with an excess of sodium
ethoxide and methyl iodide on the water-bath under a reflux condenser.
After several hours the reaction was complete, when the greater por-
tion of the alcohol was removed by distillation, and the product poured
into water. The precipitated solid was then dissolved in ether, and the
ethereal solution washed a few times with strong aqueous alkali, whereby
any unchanged micromerol was converted into an insoluble sodium

260

^RBDBRICK B. PQWBR AND ARTHUK H. SALWAY.

:fl:V

M =

compound, which could be removed easily by filtration. The ethereal
solution was then washed with a little water, dried over calcium chlor-
ide, and the ether removed, when the methyl derivative was obtained
in a practically pure state.

Methylmicromerol readily crystallizes from alcohol in clusters of thin
needles, which contain water of crystallization. The substance, when
air-dried, melts at 116-7°, ^^^ when rendered anhydrous it melts at
167 ^

0.3897 of air-dried substance, on heating at 110° for 2 hours, lost 0.0155
HjO. HaO=4.o

(I) 0.1918, dried at 100®, gave 0.5424 CO, and 0.1864 H3O.

(II) 0.1802, dried at 100°, gave 0.5102 CO, and 0.1656 H,0.
(I) C=»77.i; H = io.8. (II) 0=77.2; H = io.2.
C88H6i08.0CH3,H30 requires 1130=3.3 per cent.
C33H51O3.OCH3 requires C = 77.6 ; H = 10.3 per cent.
C8sH5o03(OCH8)3 requires C = 77.8; H = io.4 P^r cent.

As the combustion of this substance affords no conclusive evidence
respecting the number of methoxyl groups that have entered the mole-
cule, this was established by a separate determination of the methoxyl,
for which Perkin's modification of the Zeisel method was employed.^

0.2 1 75 gave 0.0993 Agl. CH3O = 6.0.

C33H51O3.OCH3 requires CH3O = 5.9 per cent.

It may be noted in this connection that the methoxyl group was at-
tacked only with great dilB&culty by the hydriodic acid, it having been found
necessary to heat the mixture for 3 hours at 140° before the reaction was
complete. It also follows from the above result that micromerol, C83H5,04,
contains no methoxyl group.

Methylmicromerol is readily soluble in the ordinary organic solvents,
but insoluble in water.

It was finally deemed desirable to determine the molecular weight of
micromerol or one of its simple derivatives, and for this purpose methyl-
micromerol appeared to be well adapted on account of its ready solubility
in cold benzene, whereas micromerol itself is only sparingly soluble in
the latter.

0.3339 of anhydrous substance in 26.5 benzene gave Jt — 0.121°.
M.W. = 5io.

C33H5i03-OCH3 requires M.W. =526.

Isolation of a New Crystalline Alcohol, Micromeritol, C^HJD^{0H)^,2Hfi'

The more readily soluble portion of the ether extract of the resins,

which, as described under Section II, had been separated by digesting

the total amount of this extract with a limited quantity of ether, was

examined as follows : The ethereal liquid, which was of a dark green color,

* J. Chem. Soc., 83, 1367 (1903).

CHEMICAL EXAMINATION O^ MICROMERIA CHAMISSONIS. 26l

was first extracted with relatively small quantities of a solution of sodium
hydroxide. The first extraction removed a small amount of resinous,
uncrystallizable matter, whereas the next few extracts were of an orange-
yellow color, and, when acidified, yielded a brown precipitate. The
latter was collected, washed, and crystallized from alcohol, from which
it separated in yellow, silky needles, melting at 225°. This substance,
of which about i gram was obtained, was found to be identical with the
new phenolic compound, xanthomicrol, CuHuOe, which had been isolated
from the aqueous liquid (A), as previously described. After all of this
yellow substance had been removed by extraction with a solution of
sodium hydroxide, further treatment of the ethereal liquid with the
same alkali resulted in the formation of a large quantity of an insoluble,
dark green sodium compound. The latter was removed by filtration,
and the treatment with alkali continued until no more solid substance
separated. The ethereal liquid then contained only a small amount of
a thick, pitch-like resin, from which nothing crystalline could be ob-
tained.

The above-mentioned solid sodium compound, which amounted to
nearly 50 grams, was digested with dilute sulphuric add, when a sub-
stance was liberated which was collected on a filter at the pump and well
washed with water. The solution of this substance in hot alcohol was
treated with animal charcoal for the purpose of removing some green
coloring matter when, after filtration, a faintly yellow liquid was ob-
tained, which was set aside to crystallize. The first deposits from this
liquid were not of a distinctly crystalline character, and melted somewhat
indefinitely at about 250®. They were found, however, to contain appre-
ciable quantities of the previously described micromerol, C8sH6j04, since
by treatment with an amount of ether which was insufl&dent to dissolve
the whole, and recrystallizing the undissolved portion from alcohol, fine
colorless needles were obtained which melted at 277®, and were identical
with the compound previously isolated.

The alcoholic liquid from which the above-mentioned first deposits
had been removed, together with the alcoholic solution of that portion
of these deposits which was more readily soluble in ether, was concen-
trated. A gelatinous mass was first obtained, from which a further
small quantity of micromerol was isolated. The mother-liquors from
this gelatinous substance were further concentrated, when, after long
standing, feathery needles separated which were different in character
from the previously isolated micromerol. Some difficulty was experi-
enced in isolating this compound in a pure state on account of its being
accompanied by a little amorphous, fluffy matter. It was found, how-
ever, that by allowing the substance to crystallize slowly from a dilute
solution it could be obtained in well-formed crystals, and by means 'Of

i2 niBDBRICE B. POWBK AND ARTHUR H. SALWAY.

fine muslin filter these were easily separated from tfae amorphous un-
iiities, which passed through the filter with the mother-liquors. Tbe
bstance, after washing with a little alcohol and lecrystalli^g from
e same solvent, was finally obtained pure. It then melted at 194-
', when rapidly heated. On analysis it gave the following results:
aai44 of air-dried substance, when heated for 2 hours at no**, lost
5158 H,0. H,0 = 7-4-

(I) 0.1633 o* anhydrous substance gave a4537 CO, and 0.1488 H,0,

(II) 0.1416 of anhydrous substance gave 0.3945 CO, and 0.1275 ^fi-
(I) C-76.2; H = ia2. (II) C = 76.o; H-iao.

C„H<,0„ 2H,0 requires H,0 = 7.i per cent.
C,|h„0^ requires 6 = 76.6; H = 9.8 per cent.

This substance evidently possesses the empirical formula C„H^O„
id as no substance of this formula, mth the same properties, has hither-
been recorded, it must be regarded as a new compound. It is there-
re proposed to deagnate it by tbe distinctive name, micromeritol, with
Terence to tbe botanical name of tbe plant from which it has been
itained.

Micromeritol is moderately soluble in cold, and readily soluble in hot
»bol, from which, on cooling, it crystallizes in clusters of fine, coloi-
is needles. It is also readily soluble in chloroform and ethyl acetate,
it only moderately soluble in benzene. In water and in light petro-
im it is insoluble. It is optically active, and a determination of its
edfic rotatory power gave the following result:

*^<^543 gi^ni of the anhydrous substance in 25 cc of chloroform gave
, +o°i6' in a 2 dcm. tube, whence [a\j, +61.4".

Micromeritol, in its general characters, resembles micromeroL It
for example, insoluble in aqueous alkalis, but readily yields an un-
tble sodium compound on shaking an ethereal solution of the sub-
ince with aqueous sodium hydroxide. Unlike micromerol, from which
differs by the elements C,H„ it contains two hydroxyl groups. The
iter, however, are dissimilar in character, for if the diacetyl derivative
boiled with aqueous alcohol, one of the acetyl groups is eliminated
die the other remains unaffected.

A property of the above-described alcohols, which is also possessed
a greater or less extent by others of a similar character and represented
the same general formula, C,H,,^,^0„ is that of forming colloidal
lutions. If, for example, a httle micromerol, CuHg04, be dissolved
boiUng ethyl alcohol, a httle ammonia added, and tbe liquid heated
til the alcohol and ammonia have been for the most part removed, no
sdpitate is produced on diluting tbe cooled liquid with water. This
ar hquid, on being allowed to stand for about an hour, forms a per-
tly transparent jelly. When, however, an aqueous solution of an

CHEMICAL EXAMINATION OP MICROMERIA CHAMISSONIS. 263

electrolyte, such as ammonium chloride, is added to the liquid before it
has formed a jelly, the micromerol is immediately precipitated in a floccu-
lent form. In the case of micromeritol, C,(|H4e04, a slight precipitation
was produced on the addition of water and the solid separated completely
in a crystalline state on allowing the liquid to stand for some time, where-
as complete precipitation immediately ensued on the addition of a drop
of a solution of ammonium chloride. On testing the monohydric
alcohol lippianol,^ CsgHgeO^, in the same manner, it was found that on
diluting the ammoniacal liquid with water a crystalline precipitate was
immediately produced, and the amount of this was not increased by the
addition of a solution of ammonium chloride. The alcohol, morindanol,'
CjjHcO^, when tested under the same conditions, exhibits a behavior
very similar to that of micromerol, with perhaps still less tendency to
deposit the solid substance except when an electrolyte is added. The
tendency to the formation of these colloidal solutions, therefore, evi-
dently increases in proportion to the molecular weight of the substance.

Di-acetylmtcromerilol, CajH4404(CO.CH3),. — When micromeritol is
subjected for a short time to the action of boiling acetic anhydride,
and the solution allowed to cool, the dirocetyl derivative separates in long,
colorless needles, which melt at 204^. It can also be crystallized from
ethyl acetate or from alcohol, and separates from the latter solvent in
glistening plates.

0.1611 gave 0.4350 CO, and 0.1312 HaO. 0 = 73.6; H=9.o.

C„H4i04(CO.CHj), requires € = 73.6; H=9.o per cent.

The molecular weight of the di-acetyl derivative was determined by
the cryoscopic method with the following result:

0.3308 in 21.2508 benzene gave At — 0.15®. M.W.=509.

CJH^^O^CCO.CH,), requires M.W.=«554.

The somewhat low result of this determination is probably due to the
fadlity with which the di-acetyl derivative loses some of its acetic add.
It, nevertheless, suffices to show that micromeritol possesses the molecu-
lar formula assigned to it.

Mono-acetylmicrameriiolf Cj^jH^jO^CCO.CHg). — If di-acetyl micromer-
itol be boiled with aqueous alcohol for a short time it loses one acetyl
group, and the solution, on cooling, deposits thin, colorless needles of
a mono-acetyl compound, which melts with decomposition at 255^.
a3090 of the di-acetyl derivative gave 0.2846 of the mono-acetyl com-
pound. Loss = 7.9 per cent.

C„H440^(CO.CHjj)2 •^ C^H^O^CCO.CH,) requires a loss of 7.6 per cent

Mono-acetylmicromeritol is readily soluble in chloroform, and moder-
ately soluble in alcohol, ethyl acetate, and benzene.

» Amer. J. Pharxn., 79, 455 (1907).
* J. Chem. Soc., 91, 19 18 (1907).

54 FEBDBRICK B. POWER AND ARTHUR H. SALWAY.

[I, IV and V. Chloroform, Ethyl Acetate and Alcohol Extracts of &e Resins.

Tbe portion of resin extracted by chloroform (III) was relatively
nail in amount, and formed a hard, black, brittle mass. It consisted
! a complex mixture of amorphous substances, and, with the exception
! a very small quantity of micromerol, nothing of a crystalline nature
mid be isolated from it.

The portions of resin extracted by ethyl acetate (IV) and alcohol
/) respectively, hkewise consisted of mixtures of black, amorphous
ibstances. They were dissolved in amyl alcohol and the solutions
ctracted with sodium carbonate, but nothing of a crystalUne character
mid be separated from tbem.

Summaiy.

The results of this investigation have shown that Mtcromena Ckamis-
mis (Benth.) Greene (syn. M. Douglassi, Benth.), commonly known
1 "Verba Buena," contains, in addition to some essential oil, resins,
id other amorphous substances, the following compounds:

1. Xantkomicrol, C„H^,04(OH)„ a new phenoUc substance, which
ystallizes in fine, lemon-yellow needles, melting at 225". It yields a
'■-acetyl derivative, C„H„Og(CO,CH,)„ which crystallizes in pale yellow
iedles, melting at 116°.

The amount of xanthomicrol obtained corresponds to about ao2 per
;nt. of the weight of air-dried plant.

2. Micromerol, C„Hj,0,-OH,2HjO, a new monohydric alcohol,
hich crystallizes in fine, colorless needles, melting at 277°. It is op-
cally active, having [«]„ -f-57°. Its acetyl derivative, Cj,H„04(C0.CHJ,
irms colorless needles, which melt at 188° and have [ajj, -{-47.1°. Its
ethyl derivative, C„H5,0,.0CH„H,0, crystalhzes in thin needles, which
lelt at 116-7° or, when anhydrous, at 167".

The amount of micromerol obtained corresponds to about 0.25 per
:nt. of the weight of air-dried plant.

3. Micromeritol, C„H„0j(0H)„2H,0, a new dihydric alcohol, which
ystaUizes in fine, colorless needles, melting at 294-6°. It is optically
:tive, having [aj^ -t-6i.4°. Its di-acetyl derivative, Q^HJIi^iQ.O.C'S.^i^
id mono-aceiyl derivative, C„Hi,0,(C0.CH5), form colorless needles,
hich melt respectively at 204° and 255°

The amount of micromeritol obtained corresponds to about 0.05 per
int. of the weight of air-dried plant.

4. Hentriacontane, C,iH„ (m. p. 66-7°), about 0.05 per cent, with
iparently a very small amount of pentatriacontane.

5. A Phytosterol, C„H„O.H,0 (m. p. 135°), in small amount.

6. Glycerides of palmitic, arachidic, and behenic acids.

7. Formic, acetic, and butyric adds in a free state.

8. Glucose (phenylglucosazone, m. p. 217°), a considerable amount.

MARRUBIIN. 265

The total amount of crude resin corresponded to 3.5 per cent, of the
weight of the plant, and from it most of the above-described crystalline
substances were isolated.

The amount of essential oil obtained by the direct distillation of the
air-dried plant corresponds to 0.16 per cent, of the weight of the latter.
This oil had a pale yellowish brown color, an agreeably aromatic, mint-
Kke odor, and possessed the following constants: (i=» 0.9244 at 20°;
Wd —22*^48' in a I dcm. tube.

Among the above-mentioned substances the two crystalline alcohols,
micromerol and micromeritol, are of special interest. Their empirical
composition is represented by the same general formula C^^^.ifi^^
and, as they contain no unsaturated linkings, they are evidently cyclic
compounds in which a benzene nucleus is doubtless present. As micro-
merol contains but one hydroxyl group, whereas micromeritol contains
two such groups, they are not simple homologues, but a similarity in
their general characters renders it probable that some fundamental rela-
tionship exists between them. It is of further interest to note that a
crystalline, monohydric alcohol, C^Hg^O^ (m. p. 300-8°; [aj^ + 64.9®),
designated as lippianol, which was recently isolated from a South African
plant, Lippia scaberrima, Sonder (Nat. Ord. Verhenaceae), possesses the
same general formula as those above mentioned, namely, C^H2„_i404,
and has similar properties (compare Power and Tutin, Archiv der Pharm,
24s, 344 (1907), and Amer, /. Pharm., 79, 449 (1907)). Another com-
pound of this class, having the formula CjgH^O^ (m. p. 278°; [a]D -1-65.9°),
and designated morindanol, has likewise been isolated in these labora-
tories from a West African plant, Morinda longiflora, G. Don (Nat. Ord.
Rvbiaceae). Cf. /. Chem. Soc, 91, 1918 (1907).

Our thanks are due to Mr. H. H. Dale, Director of the Wellcome Physio-
logical Research Laboratories, for having conducted a test with micro-
merol. One g^m of the substance was administered to a small dog, but
no symptoms of any kind were manifested, and it therefore appears to
be devoid of any pronounced physiological activity.

MARRUBim.'

By H. M. Gordxn.
Received November aa, 1907.

The bitter principle, marrubiin, was discovered in horehound (Marru-
bium vulgar e, Linn6) by Mein and, without indicating the method by which
it was obtained from the plant, sent to Harms for investigation.' Later

* The Wm. S. Merrell Chem. Co. deserves my thanks for not only supplying the
horehound but for preparing for me an extract in accord with my directions.
*Arch. Phann. (2), 83, 144; (2), 116, 141.

366 H. U. GORDIN.

the bitter principle was examined by Kromayer,' Hertel,* MorrisoD*
and Matusow.' The different chemists employed different methods
for the isolation of marrubiin, and their results differ so much from each
other that it seemed advisable to undertake a thorough examination
of this constituent of a popular plant. As mamibiin is only sparingly
soluble id water, the yield obtained by this menstruum is extremely
small. This accounts for the very small yield obtained by Harms (2
grams from 2$ lbs.). On the other hand, owing to the great expense
involved, ether, while a fairly good solvent for the bitter principle, is
not suitable for extraction on a targe scale. It seemed, therefore, that
Matusow's method in which acetone is used as a menstruum was tlie
most convenient for the purpose. Matusow's method conasts in extract-
ing the horehoimd with acetone, distilling off the solvent and treatiuf
the readue with hot benzene. According to Matusow, the benzene solu-
tion on cooling deposits the mamibiin in crystalline condition. I haw
faithfully followed Matusow's directions and obtained the crystals, but
a careful examination of these showed them to be not mamibiin but
potassium nitrate. The identity of the salt was established by a large
number of the reactions of both the metallic and the nitrate ions, as also
by a crystallographic comparison with pure potassium nitrate carried
out by Prof. Kraus, of the Univerdty of Michigaiu The appearance
of saltpeter in considerable quantities in the acetone extract is possibly
due to the presence of water in ordinary acetone. The small crystals
of the salt cannot be noticed in the very dark readue left after distilling
off the acetone; when the residue is heated with benzene the potas-
sium nitrate, which is extremely easily soluble in hot water, goes into
solution in the small amount of water present and crystallizes out upon
cooling. Having failed to prepare marrubiin by Matusow's method,
I worked out the following method. The coarsely ground horehound
is extracted by percolation with cold acetone till the latter comes out
nearly tasteless, the solvent distilled off and the residue digested repeat-
edly with conaderable quantities of petroleum ether. After pouring
off the petroleum ether the readue is digested several times with wann
water to remove soluble inorganic salts and then dissolved in just enougb
hot alcohol to bring all into solution. On cooling, the whole liquid solidi-
fies to a mass of crystals. After washing the crystals with cold alcohol,
they are repeatedly reciystaUized from hot alcohol, using animal char-
coal freely, till their color is snow-white and their melting point is con-
stant. The yield of pure marrubiin by this method was 0.25 per cent;
but as it had to be recrystallized nearly a dozen times and marrulain
' Ardi Pharm., 108, 258.

* Am. J. Phann., iSgo, 373.
■ ;Wrf„ 1890, 337.

* Ibid., i8g7, 301.

MARRUBIIN.

267

is quite soluble in cold alcohol, most of it remained in the mother liquors,
from which I hope to obtain additional quantities. The melting point
of marrubiin is 154.5-155.5°; under a pressure of 15 mm. it boils at 297-
99°. It crystallizes from hot alcohol in two different forms. On quickly
cooling the alcoholic solution the crystals are small and fine ; by slow crys-
tallization large heavy crystals can be obtained. If the hot saturated
solution is free from any solid marrubiin, the solution can be kept for
several hours in a supersaturated condition, and when crystallization
begins, the crystals are flat and over two centimeters long. As shown
by the melting point and analysis, the two modifications are chemically
identical. A complete separation of the two modifications could not
be effected. Following is an accoimt of crystallographic measurements
carried out by Prof. Kraus:

"Marrubiin crystallizes in two distinct modifications from a solution in alcohol.
The crystals of the first modification are well developed and allowed accurate gonio-

metric measurements to be made. The
second modification does not show as well
developed faces, so that for the present only
approximate measurements for several angles
can be given.

"Modification One. — These crystals are
readily obtained by slow evaporation of an
alcoholic solution. The crystals are usually
quite small, about 2 mm., perfectly clear, and
transparent. The following forms were ob-
served: a (100), tti (iio),c (001), d (loi),
and 9 (01 1). Of these forms the ortho pina-
coid a (100) and the prism m (i 10) are usually
quite large, giving rise to a prismatic devel-
opment. The basal pinacoid c (001) may,
however, predominate, causing the tabular
habit to result. The domes d (loi) and q
(001) are generally quite small but give ex-
cellent images. Fig. i shows the usual combination and development. In all,
eight crystals were measured with the following results:
"Crystal System — ^Monoclinic.

"Axial Ratio— a : b : c=i.555i: i: 0.885, ^=61** 17'.

Observed. Calculated.

a:m= (100): (110) = 53*^ 45'

a:c == (100): (001) = 61® 17'

c:q = (001): (on) = 370 49'

a: d = (100): (loi) = 39 *> 51.3' 39« 55.7'

c: f» = (001): (no) = 73<» 32.5' 73° 29.5'

'Afodificaiion Two. — ^The second modification also belongs to the monoclinic system.
The crystals are flat tabular in development and show a prism angle of 94° 36'. The
angle between the ortho- and basal pinacoids, 0(100) and c (001), is about 71** 3.5'.
More accurate measurements will be made later. The above values are, however,
sufficient to show a decided difference in the crystallization of the two modifications.

Edward H. Kkaus."

Fig. 1.

z6S H. H. GORDIN.

The formula of mamibiin is C„H,jO,. Analysis gave C, 73.00, 73.14,
72.90, 72.97; H, 8.28, 8.57, 8.32, 8.43; calculated for CjiHjjO,; C, 73.21;
H, 8.20. A molecular weight estimation by the cryoscopic method
with phenol as solvent gave M = 327. Calculated for CjiHjjOj; 344.22.
Mamibiin is isomeric with the menthyl ester of acetoxy a-phenylacrylic
icid and with the menthyl ester of benzoylacetic add, which were pre-
pared by Lapworth and Hann. ' The first of these esters melts at 51-52°,
the second is a liquid. Marnibiin is dextrorotatory. [«]d = 45-68°
[c=4,794, solvent, acetone from bisulphite and redistilled by myself).
The polariscope was a Josef and Jan Frif apparatus. Marrubiin is solu-
ble in about 60 parts of alcohol at 20° and 20,835 parts of water at 21.5°.
Lt is very easily soluble in acetone, chloroform, hot alcohol, warm phenol,
pyridine and warm glacial acetic acid ; diificultly soluble in ether or ben-
Eene. It does not reduce Fehling's solution or ammoniacal silver nitrate
either before or after warming with dilute mineral acids. It does not
:ontain CH,0 groups (hy Zeisel's method) and does not decolorize bro-
mine in glacial acetic acid solution. It has a very bitter taste and a neu-
tral reaction. It does not react with acetic anhydride, benzoyl chloride,
Hydroxylamine or phenyl hydrazine, showing absence of hydroxyl or
:arbonyl groups. In all cases unchanged marrubiin was recovered as
ihown by crystalline appearance, melting point and bitter taste. Cold
iqueous or alcoholic potassium hydroxide has no effect upon it. Ven-
lilute alcoholic potassium hydroxide attacks it slowly on prolonged
soiling. When boiled for a short time with ten per cent, alcoholic potas-
tium hydroxide it takes up a molecule of water and is quantitatively
:onverted into the potassium salt of a new acid which I have named
narrubic acid, CjoH^Oj.COjH. The acid is made as follows: Ten
jrams mamibiin are boiled for about half an hour with 150 cc. of alcohol
:ontaining 10 to 12 per cent, of potassium hydroxide, the Uquid is then
liluted with twice its volume of water and concentrated till the odor
)f alcohol disappears, adding water from time to time. The solution
S set aside in a cool place for 24 hours, and the very small amount of
ilimy matter which separates out removed by filtration. The filtrate
s acidified with hydrochloric acid and the extremely bulky white pre-
lipitate which separates out, collected on a filter, thoroughly washed
vith water and recrystaUized from hot alcohol to which hot water is
^dually added. Yield over 98 per cent, of the theoretical. Thus obtained
narrubic acid forms snow-white, long silky needles. The acid is ex-
:remely bulky and looks like i]uimne sulphate or caffeine, but is even
luffier than these. According to Prof. Kraus the acid crystallizes in
xtreniely fine, long, prismatic crystals. The prism faces show, ap-
»arently, parallel extinction. The acid melts at 173-4°. ^ts formula
' J. Chein. Soc„ 81, 1497 and 1507,

MARRUBIIN. 269

is C^H^Oj-COaH. Analysis gave: C, 69.02, 69.08; H, 8.76, 8.67. Cal-
culated for C20H3JO3.CO2H : C, 69.57; H, 8.35. Marrubic acid is very
easily soluble in alcohol, warm phenol and pyridine, difficultly soluble
in ether, glacial acetic acid and benzene, almost insoluble in water. It
has an add reaction in alcoholic solution. It is isomeric with antiari-
genin. The latter is not an add and becomes intensely yellow at 170°
melting at 180®. Marrubic add is very easily soluble in ammonia water,
but upon concentration of the solution all the ammonia evaporates, leav-
ing the add unchanged. It is easily soluble in alkalies and alkali car-
bonates and forms salts with metals, none of which, however, could be
obtained in crystalline condition. The salts all seem to be very easily
sohible in water and alcohol with the exception of the copper salt, which
is amorphous and difficultly soluble in water. Titration with standard
alkali showed the add to be monobasic. 0.2200 gram of the add required
6.05 cc. N/io KOH (phenolphthalein as indicator), and 0.3963 gram
required 10.8 cc. N/io KOH. Calculated for C20H29O8.CO3H, 6.1 and
10.9 cc respectively.

Barium marrubate, (C2oH2903.C02)2Ba, was prepared by digesting
marrubic add with an excess of barium carbonate in dilute alcohol, filter-
ing, concentrating to a small bulk, again filtering (to remove traces of
barium carbonate) and evaporating to dryness. The salt is amorphous
and very easily soluble in water or alcohol but insoluble in ether. It
could not be obtained in crystalline form. For analysis it was con-
verted into barium carbonate. 0.2818 gram gave 0.0645 g. BaCOj, and
0-5605 g. gave 0.1293 g. BaCOg. Calculated fo¥ (C2oH2Q03.C02)2Ba,
^5-97 P^r cent. Ba. Found: 15.93 ^tnd 16.06 per cent. Ba. At water
bath temperature marrubic add quickly reduces ammoniacal silver nitrate
or Fehling's solution. Neither the alcohol solution of the free add nor
the aqueous solution of the barium salt are colored by ferric chloride.
Marrubic add is dextrorotatory, [a]^"^ = 7-86. (0 = 2.5456; solvent, ace-
tone).

When heated to 190-200° under a pressure of 15 mm. the acid boils
up suddenly, loses one molecule water and changes back to marrubiin.
This was shown by treating the melt with dilute sodium carbonate solu-
tion to remove traces of unchanged add and recrystallizing the insoluble
part from alcohol. The crystals were insoluble in alkali carbonates and
melted at 154-5°. The same reconversion of marrubic add into
marrubiin can be effected by warming the add with ten times its amount
of acetic anhydride and a trace of zinc chloride for a few minutes to about
50**. On largely diluting the liquid, which assumes a dark yellow color,
the color disappears and an oily liquid separates out, which on stand-
ing under water becomes solid. The solidified predpitate was digested
with sodium carbonate and then recrystallized from alcohol. It was

H. M. GORDIK.

. alkali carbonates and melted at 154-5°. '^^^ liquid ob-
digestion mth sodium carbonate gave no precipitate upon
excess of add ; hence the reconversion of marrubic acid into
)y this method is quantitative. The acid is also reconverted
3iin by boiling for an hour with alcoholic hydrochloric add
er cent, add and 50 cc. alcohol). On throwing the solution
water and recrystallizing the predpitate from alcohol, it was
e mamibiin (insoluble in alkali carbonates and melted at
?art of the acid is resinified in this reaction. On digest-
on of marrubic add in pyridine with benzoyl chloride (oiw
ch) for 24 hours and then adding 750 cc. water an oily liquid
Lit which becomes solid upon long standing. The mass was
th dilute sodium carbonate solution and then reciystallized
[>1. It was found again to be marrubiin. It constituted
;r cent, of the amount of add taken. The part soluble in
bonate does not seem to be unchanged marrubic add. It
tated from the alkaline carbonate solution by addition of
; add and recrystalhzed from alcohol to which hot water
lly added. On standing fine needles separated out. They
ween 164-70° and represented irost probably a benzoyl
if marrubic add, but the amount was too small for an analy-

rfufra/e.— Marrubic add was digested with a solution of potas-
xide on the water bath, leaving the add in slight excess, the
ed to remove undissolved add and evaporated to drjTiess.
s of the potassium salt thus obtained were mixed with 4 cc.
and zo cc. acetone and the mixture boiled for two and a half
r refiux condenser. The liquid was then evaporated to dry-
;sidue treated with dilute sodium carbonate solution to re-
ble traces of unchanged add and recrystallized by dissolv-
alcobol and adding water to turbidity. As the sodium car-
tion did not give any predpitate with adds, the esterifica-
iTUbic add by this method must be quantitative. From a
alcohol and water the ester crystallizes in very pretty glitter-
from a mixture of ether and petroleum ether large hea^7
1 be obtained, which consist of layers of leafiets. Unlike
nd marrubic acid, the ethyl ester is tasteless. It is extremely
le in ether and pyridine, a little less soluble in alcohol, chloro-
zene and still less in petroleum ether. It melts at 87°, When
jout 100° under a pressure of 28 mm. it boils, gives up a mole-
9hol and is reconverted into marrubiin. This was shown
illizing the reaction product from alcohol and comparinR
rrubiin. For analysis an ethoxyl estimation was made by

DBCOliPOSinON CURVES OP SOMB NITROCBLLULOSBS. 371

Zeisel's method. Found: 10.78 and 11.50 per cent, of CjHgO. Calcu-
kted: 11.54 P^^ cent, of CjH^O. The ester fomis an acetyl derivative,
but this and several other derivatives of niarrubiin will be reported upon
in my next paper. The results of the investigation so far show that
marrubiin is a lactone behaving like a r-lactone in that it easily takes
up a molecule of water and changes to a hydroxy acid. The formulae
of marrubiin, marrubic acid and ethyl marrubate can therefore be written
as follows:

C»H„0,— CO, C,^„0,(OH).COOH, CjoH„0,(OH)COO.C,H5.

L ^ ^

The investigation is continued.

NORTHWSSTBRN UnXVBRSITY SCHOOL OP PHARMACY,

Cbicaoo.

DECOMPOSITION CURVES OF SOME NITROCELLULOSES OF

AMERICAN MANUFACTURE.'

By OSWIlf W. WiLLCOX.

Received November 11, 1907.

In the course of a study of the test proposed by ObermuUer* for the
stability of nitrocellulose, it was observed that the rate of decomposi-
tion of an ordinary collodion cotton during a given period of time de-
pended on whether or not the gaseous products of decomposition had
been allowed to remain in contact with the sample during previous pe-
riods. As the ObermuUer apparatus is perhaps the most accurate and
convenient means yet devised for following quantitatively the progress
of the decomposition of a nitrocellulose at temperatures below its ignition
point, some of the results obtained are deemed of interest.

ObermuUer's test is essentially as follows: A weighed quantity of
the nitrocellulose to be tested is placed in a glass tube, which is then
evacuated by means of a good air pump. When the air has been re-
moved as much as possible, the tube is plunged into a bath previously
brought to a standard temperature, which is maintained constant through-
out the duration of the test. The nitrocellulose in the tube immediately
begins to decompose and to give off gaseous products; the tube being
in connection with a mercury manometer, the rate at which the
products of decomposition are evolved is measured by the increase of
pressure shown by the manometer. This rate will naturally be greater
for nitrocellulose of poor stability and less for nitrocellulose of good
stability. Working at a standard temperature of 140° C, and with a
tube which, with its connections, had a volume of 37 cc, Obermuller

* Published by permission of the Honorable the Secretary of War.
^MiUheilung aus dem Berliner BezirksvWein des Vcreins deutscher Chemiker,
October 11, 1904.

-■'.■' •■ * -i I

.' v.

« ■ . «

' ■ r.

•' ^ •

':'' • :.v ^

I .

•I

''-'^•■-A\C''A

, k

. .' ».

• ■••■!

.. y

4
I

• ■• ■ I

272

OSWIN W. WILLCOX.

found that one gram of a stable nitrocellulose does not cause in one hour
a greater increase of pressure than 100 mm. of mercury.

The apparatus employed is diagrammatically shown in Figs, i and 2,
which are copied from Obermuller's paper. The decomposition tube

/^=^

\_y

Figs. I and a.

Z (Fig. i) is of glass and of about 12 cc. capacity; the top of Z is ground
to fit the larger end of T, by means of which it is connected to the manom-
eter M. Behind the manometer is a scale S. By means of the side
tube R the decomposition tube can be put in communication with the
air pump. V is a glass bottle, on either side of which is a glass slop-
cock (H and H'). The lower part of the manometer is connected with
the mercury reservoir G. The bath, L, which may consist of a saturated
solution of calcium chloride, or, better, of oil, such as is used in the cylin-
ders of locomotives, serves to heat the tube Z. The sheet-iron case
K, which is provided with a window of heavy glass, serves to protect
the operator from possible explosions. In Fig. 2 is shown an enlarged
sketch of the tube Z and its connections. G is a glass rod sealed by
its upper end to T; the object of this rod is to hold the nitrocellulose
in place in the lower end of Z. X is a mark i cm. under the cap.

The method of procedure adopted in this study is as follows: Ex-
actly 2 grams of the nitrocellulose previously dried by heating for i hour
at 100° were placed in the decomposition tube Z; any nitrocellulose

DECOMPOSITION CUHVES O^ SOME NlTROCELLXJIX>SES. 273

adhering to the walls of the tube were swept down by means of a small
wad of pure dry absorbent cotton, which was allowed to remain in the
tube during the heating. The tube was then fitted in place and evac-
uated as completely as possible, usually down to an internal pressure
of less than 5 mm. of mercury. The bath having been brought to 140°
the positions of the mercury in both limbs of the manometer were noted,
all stopcocks were closed and the tube lowered into the bath up to the
mark X, the time being noted at the same instant. Exactly 15 minutes
from the time of immersing the tube, the surface of the mercury in the
left limb of the manometer was brought back to its original position
and the rise of the mercury in the right limb was noted; and this was
repeated at intervals of 15 minutes until four such readings had
been taken. The accumulated gases were now pumped out and
a vacuum maintained in the tube for 15 minutes. At exactly 75
minutes from the time of beginning all stopcocks were again closed
and the gaseous products of decomposition allowed to accumulate
for a period of 15 minutes. The pressure developed during these
15 minutes was read, the tube again kept vacuous for 15 min-
utes, and so on alternately for 5 hours from the time of beginning.
For convenience in discussion, the whole tiire of heating may be regarded
as divided into periods of 15 minutes each. During the first four periods
the gases were allowed to accumulate in the tube and to exert pressure;
after the first hour the gases were being continually withdrawn during
the odd periods, whereas they were allowed to accumulate during the
even periods. The procedure followed during the first hour will be re-
ferred to as "test with increasing initial pressure," and that followed
after the first hour as "test with constant initial pressure." All read-
ings of pressure are given as millimeters of mercury per gram of sub-
stance. The data obtained are tabulated in Table I.

Table I.

PreMure in nim. of mercury per period of 15 minutes.

Na of Test with increasing
umple. initial pressure. Test with coastant initial pressure.

ist. and. 3rd. 4th. 6ih. 8th. loth. lath. 14th. i6th. i8th. 3oth.

2920 20.3 23.9 28.7 32.9 27.8 28.6 30.6 31.0 33.5 33.0 33.5 3A.O

3831 18.7 23.4 27.7 33.0 27.7 28.7 30.6 31.3 33.7 32.7 34.5 34.5

3383 18.6 22.2 27.8 32.0 27.6 28.2 30.3 30.6 32.2 32.5 33.0 33.5

3863 21.0 29.5 35.0 42.7 38.0 39.7 41.8 43.3 45.7 45.7 470 47.5

3607 21.9 24.5 28.7 34.2 31.5 33.5 34.1 34.5 36.7 36.2 39.0 38.0

3608 19.2 20.9 24.6 29.1 25.5 27.0 28.1 29.1 29.7 31 7 31.5 32.5

3609 18.1 22.2 25.8 25.8 29 4 26 I 28.5 28.3 29.5 29 7 32.0 32.0

The samples of nitrocellulose here reported on represent the product

of the principal factories of the United States. They are **decanitro-

celluloses** of about 12.60 per cent, nitration and 99 per cent, solubility
in ether alcohol.

, • '- "M ■ .J' » - ••* "■

.'• .' . ~ ■• • •-
r' '■■ * . ' ** '4

I .. ■*.

■r .,■

• m

•••■• "=}

274

OSWIN W. WILLCOX.

' f

• V . • • • .4

. • - - •-<

" • :

• r.'

'* ^ '

» ... ' * * -

'' ' 'K. • J

• *■ • ";

1 ^

* - ■

!•'

-.1
I

«
•-1

The figures at the top of the columns in Table I are the numbers of
the periods, of 15 minutes each, during which the gases were allowed
to accumulate (it will be remembered that a vacuum was maintained
in the tube during the odd periods). The results are also plotted in
the accompanying curves. An examination of the figures given in Table

Pig. 3-

DBCOMPOSITION CURVES Of SOME NITROCELLULOSE S.

Pig. 4.
1 will show that the course of the decomposition in "test with increasing
initial pressure" and in "test with constant initial pressure" follows
different laws. In the case of sample No. 3831, for example, the quarter
hourly increase of pressure in the first period is 18.7 mm., in the second
23.4 mm., in the third 27.7 mm., and in the fourth 33.0 mm., an aver-
age constant difference of about 4.5 mm. By the aid of this law of in-
crease it may be calculated that during the sixth period the increase
of pressure would be 42 mm., if the gases were allowed to accumulate
in the tube without interruption. The calculated rate for the sixth
period was found by experiment to be the actual rate under the condi-
tion named. If, however, a vacuum be maintained in the tube during
the fifth period, then the increase of pressure during the sixth period is
not 43 mm,, but only 27.7 mm., a difference of 14.3 mm. The differ-
ence in the rates of increase of pressure under the two conditions meas-
ures the catalytic effect of the presence of the gaseous products of de-
compodtion on the speed of the reactions of decomposition. It may
be noted that while removal of the gases lowers the "decomposition pres-
sure" of the heated nitrocellulose it is not restored to the original value;
the initial decomposition pressure gradually rises with the time of heating,'
' At tbe s&me time the heated nitrocellulose is being transformed into a modi-
fication soluble in absolute alcoliol. A separate report, with quantitative data, will
be nude on this subject in tbe future.

.-.v.-..--"v
*1 • ..-."

•• 1*

X..

■

-8 . J

' ■■ ■

>* I

m f.

— ■ V I

'••■ :. .M

. ^
1
I

•:; • 'M

I *
1

••i--

■

•I

• I-

. I

276

OSWIN W. WUrtCOX.

The results reported in Table I were obtained by working at a con-
stant volume of 30 cc. It is obvious that as the volume of the tube
and its accessories determines the concentration of the gaseous pnxlucts
of decomposition, the rate of increase of pressure will vary with this vol-
ume. To deduce the volume relations of the test a number of exper-

■M-

-♦-►

■tT+

"-'*"*■■■■■■■■■■■■■■■■■•■■■■■■■■■■■■■■«■■■■• ■•■■■•••

•■■■■■■■■■■■■•■■■■■■■■■■■■■■■■■■■■•■■■•I

■■■■•■■■•■■■■■■■■■■•«■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■•■■■■•"■»■>•■■*!
■■■■■■■■■■«■••■■■■■■«■■■■■■■■■■■■«■■■■■■■■■■■■■■■■■■!■•■■•■■■■■■■*"!

:■■■■•■■■■■■■■■■■■■■•■■•■■•■■

■•■■■■•■^■■•■■a ■■■■■■■■■■■■■■«■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■•■■■■•«■■■■■■■■■«■■■■■■"■■■"■ *■■■*"*■'
■■■■■■■■■ -«.ap .'/ / ■{■■^■■■■•■■■■■■■■■■■■•■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■"■> ■■■■■•
■■■■■■■a <^» ' ' J ■■■■■■■■«■■■■■■■■■■ ■••■■■■■■■■■■■■■■■■■■■■•■•■■■■■■ ■■■■■■■■^ rv^.^aaaaaaaaaaaB ■■■■••■••
aaaaaaaa ^aaaaapaapaaaaaaaaBaaaaaaaaaBaBaaBaaBaaaaaaaaaaaaaaaaaaaraBaaaaaf < a -• laaaaasaaaaaBa ■■■■■■ >«•
■niaama.^. ■■■BaifBiiaaanaaaBaaaaaaaaaaaaaaaaBaaBaaaBBaaaaBaaaaaaBaBaaaaaaavaaBaBaiaaBaaaBaaaaaaaiiii
BaaaaBM^* BBaiiaiaaaaBaaBaBBaBaaaBaaaaaBaBBaaBaaBaaBaBaBaBaaBKBaBaaaBaaaaa'.aaaaaBBBBBaaaBaBBa ■•> ■ ■ ■■•
Baaaaaaa ■^aaaaBBaaBBBaaBaBaaaaBBaBBaaaaaBBBaaaaBBaBaaBBBBBaaaaaBBBaBBBBBB^.aaaaaaBaBBaaBBaBaaaaaaiaa ti

BBBBBBBaaaaBBBBBaBaBBBBBBaBaaBBBaBaaBaaB^BBBaaBaBaaaaaaaaBaBaa

BMBBBBV^BBaa BBBBBa'' •"' ■^■aaaaaaaaiaai

laaaBaaa.BBaaaBaBaaBaaaaaaaaaBBBaBaBaaBBBBBaBBBPBaBBaaBBBBBBBBBBBBBar^BaBBaaaBBaaa-aaaaBaaaaaaaaaa.o

'i'.'iaaaaBBBBaaaBBBaaBBaaBBBaaaai

raaaa a ■ a

aaao. BaaaaBaBaBBBBBaaaBaaBaaBaBBaBBBaBaBBaaaaaaaBaaaaaBBBaBa'.aaaBaBBBaBaa'-^aaaaa ■■aaaaaaaa •■■■• ■
,aaaa. aaaaBiaBaaaaaiaBaaBaBaaaBBBBaaaaaaBBaBaaBBBaBBBBBBBBBa;.BBBaBaBBBaBa ..BBaaaaa aaaaaaaaaa laaa • ■

aaiaaaaa v^aaaaBaaiaaaaaaaaaBaaaaaaBaaBaaaaaaaaaaaaaaaaaaaaaaaar^aaaaaaBaaaa'.aBaaBa' >.*<r^aaaaaaiaa«ij
■ aaaaaaa.'H aaaaiaiiaaaBBaBBaBaaBaBaBBBaaaaaaaaBBaaaaaBaBBaaBBarjBBBaBBKBaBr^BBBaBBa '*< 2.*!"!""*""*!!

|M:::iii::|:i::ii

aBaaBaBBBBBBBaaBBBBaBBr^aaaaaaBBBaa'', aBaBBB'>'*<r«.aaaaaaiaa«ij
■aaaaBBaaaaaBaBBaaBBarjBBBaBBKBaBP^BBBaaaB ' *< ^ «aaaa aa aaa* n

•aaB8BaaaaB«aaaBaaaB'.aBB8BaaaaBB'.«aBBBaaaB~a aaaaaaaa ■■■«aaait
BaaaaBaBaBBaBBBaBBa'.aaaaaaBBBa'^aaaBaaBa^^aaaaaaaaaa aaaiaam

mffm immtfflfttfgmff^ffl

■ ■ I

aaBaaaaa«.v BBaaBaaaaaaBaBBaaBaaaaaaBaBaBaaaar aaaaaaaa- «aaaaaaaa'.«BBaBBa'^BBaaaBBaBaaaBaaaaaaaaa a
■aSaaBaa^-«BaaaBaaeaaaaaaBBaaaaaBBaaBBaaaflBraBBaaaaa'^aaaBaaBB''^aa8Be»-4BBBBBBBBaaBBBBBBBaB8«Baaa
aaaaiaaa.'^aaaaa8aaiaaaaaaBaBaBBaaaaBaaBBa'.aaBBBaar4aBBaBaar..aBBBBP'..BaBBaBBBBaBaBBaaBaaBaaBaaat

-,aaBaaBBr:.aaiBBr'^#aBBBBaaBBBBBBaBBBBaBaaaaaa

aaBaaaaa«»a< ". .bbbi

Hiiiii

BaaBaaaaaa-.
aaaaBaaar ^BBav'^aaaa'

~ -aaaaaaBBBaaasB

■ ••:.aBaaB' ^aaBaaaaaaBaBBBaaBBBBaa as aaaaaaaa ■ a •«
dBBaaaaaaaaaBaaBBaaaaaBaaaaBBBaaaBaaii*''**

BBaaaaaBaBaa

aaBaa■■t■naaaaaaa^aaar^aBa'->aa■:;daaaaaaaBaaBBBBa8BBaaBBaBaBBBBB■BBaBaaaBaBaaaaBBBBB■■■t>*•t

laaaaBaaaBaaBaaa t« « ••!
BBBBaBaaBBaBaBaaaBBBaaBaBv

r;^«aBBaaaaBBaaBBBiaii»aaBaBBBBaaBBBBBBaaaaBaaBBBBBB«B ■aaaB88«8aaaa»«aa«>i*a>

B8aaBBaaB%aaa«aaaaaaaBBBBBBaaBBBBBBaBBBaaaBaaaBaaaaaaaBaBaBB«aBBaBBaBB8aBBKasaaaaaaaaaBaaBaBaai«*>

aflaaaaBBBAaaaElf vaMiaa , .iBBBaaaa' /Aaaaaaai 'i , JBaaaaaa^ ..aaaaBaay. ' iaaBaar > .. jbbbbbb' ' x. BBaaaBB»*r

ii»^paBBa«BM(«Hi'« j—Ml*^ ' '.BaBaaaar* •<aaB»^»i'<^ ■.aaaaaBBi.* 'jaaaaaaM '.aaaaaB..- "'■■■■■■l^'.^"*;* ■"•!•"
iaaiiBa»aaaBlilB«aa#aBaaBaaaaaaaaaaBaBaaaaa«aBBaaaBBBaaaaaaaBBBaBBaaaBB8aBaBaBBBaBBaBaBBBB8Ba8BaBa

w.*» \.rf X -BB ■«'

ilii Si

" 4n93

a aai

aaBBBaaaaaai'i'^ tt ft-'M* ti*/ tn. 'S'.m if
BiiaBBaaBaaiaaaaaaaaaiaBaaaaMaaaaaaBBi
BiiaBfaaaa '/>r .,v,* yfvfj»i: ^aa v r. $0 ^
aaaaaaaaafjiiii/ jb* "»»'/** '.' ^mrr '*.ii-'* •
aiaiaaaaaaa— aaaaaaiaaaaBBaaaaawaaaBi

4-L

ffi

'i.ff • . ' m' i9',rr "tv •*•*' -r'.a '. vj j'/ /t mf. jm'. iaaaaaaaaaBBBaBaa ■ aaai
kaaaaaBBaaaaaaaaBBaaBBaaflaaBaiaaaaBBaaaBBBaaBaaaaaaaBBaaira aa Bj
ir-.f.a^ *//• t'/' / r J'/ - 0 v 't ' ■ *u't /rrfttm [»^v- aaaBiaBaaai

tjtswm, 'St'. J : •itwi.wt'M t O// tf.'.rti/, ' %r 9d' »'•' . //-aaaaaaaa a ■•«
laaaalaaaaaaaaaaaaaaaanaatBataaaaaBaaBaaaaaaaaaaaaaaaBaRB a aaai

Fig. 5.

iments were made on sample No. 3831. The total volume of the appa-
ratus as far as the glass cock H' was measured and found to be 60 cc.
On leaving H open during a determination, the gases would exert a pres-
sure corresponding to this volume. By adding 10 cc. of merciuy to the
glass bottle V, the volume could be reduced to 50 cc., another 10 cc. of
mercury would reduce it to 40 cc, and so forth. The pressures obtained
by working at the volumes 30 cc, 40 cc, 50 cc, and 60 cc. are given in

Tablb II.
Obbrmullbr Tbst at Dippsrbnt Volumbs.

Pressure in mm. of mercury per period of 15 minutes.
Volume Test with increasing
in cc. initial pressure. Test with constant initial pressure.

30
40

50
60

ist.
18.7
12.5
10. 0

8 o

initial pressure.
7nd.

23.4
17.0

12.5
9-5

3rd.
27-7

19-5
150
12.0

4th.

330
22.0

16.0
13.5

6th. 8th loth. 1 3th. 14th. i6th.

27.7 28.7 30.6 31.3 33.7 32.7

21 O 21. o 23.0 25.0 24.5 24.5

15.5 17.0 17.5 17.5 18.5 19.5

13.5 146 15.5 15.0 15.5 15.5

x8th.

aoCh

34.5

26.5

25.5

20.5

21 0

16.5

16 0

' «

DECOMPOSITION CURVES OP SOME NITROCEU.ULOSES. 377

Table II, and are plotted in Figs. 5 and 6, As was to have been ex-
pected, the pressure increases as the volume decreases. During a given
period of heating the evolution of gaseous products of decomposition
in "test with constant initial pressure," is constant. The equation

PV = K
where P is the pressure developed during a period, V the volume, and
K a constant, holds for all volumes between 30 cc. and 60 cc. The av-

Plg. 6.

erage values of K are plotted in the curve shown in Fig. 7. This curve
represents as neariy as may be, the course of the decomposition of a
nitrocellulose when heated in a vacuum. In "test with increasing initial
pressure" as the volume decreases the pressure increases according to
a law which can be stated as follows: If the original volume be 60 cc.,
and if this volume be successively decreased by a constant quantity,
then the pressure developed in one hour at any other volume can be
found from the equation

P» = P„ + "^"--'^c,

where P is the pressure expressed in millimeters of mercury, ti the num-
ber of times the volume has been decreased by a constant quantity,
"and C a constant. For stable American collodion cotton P„ is about

DBCOMPOSrriON CXTRVES OP SOME NITROCBLLULOSBS.

279

43 mm. and, wider the conditions above described, C is about 9. The
above equation has been proven for volumes between 30 cc. and 60 cc.

The values of P^, when plotted, give the rectangular hyperbola shown
m Fig. 8.

A noteworthy feature of the curves showing the rate of increase of
pressure developed by a nitrocellulose in **test with constant initial pres-
sure" is that they are not straight lines nor smooth curves, but are made
up of oblique and horizontal lines. This is discernible in the curves
shown in Fig. 4 and is well marked in Fig. 7, wherein are plotted the
average values of K, calculated from the data given in Table II. The
step-like character of these curves reminds one of the curve represent-
ing the progress of the dehydration of a hydrated salt like CuSO^-sH^O,
which can be transformed into lower hydrates, each having its own vapor
pressure. The resemblance is sufficient to suggest the explanation that
a given sample of nitrocellulose when heated at 140° has a character-
istic decomposition pressure which is represented by the first approx-
imately horizontal portion of the curve (A B, Fig. 7). As the heating
progresses a new phase or component appears and adds its decomposi-
tion pressure or accelerative action to the previous one, giving the new
horizontal section C D. The rate of increase of pressure represented
by this part of the curve remains fairly constant until a new phase or
component appears, whence results the horizontal section EF. It ap-
pears to be characteristic of nitrocellulose of good stabiUty that the
curves obtained in "test with constant initial pre^ure** have this step-
like character plainly developed. In Fig. 9 the transition between two

UfcotffXfsmm cmyi ofM^/ii^ /ypjsu.

'4C

60 /MO /OQ

Tffie /jf/nf/rtmest

Fig. 9.

J9O0

X40

adjacent horizontal sections is well brought out. The curve shown in
this figure was taken from the record of a different sample. No. 3222;
in this case a fresh portion was taken for each pressure determination,

• ■ ■ ' - J

•«.-■*.' * ■ •

« •

•• •.

•» ■

• r
•• •

• r

. - t . ■ ,, ..." «
• ■ • • • 4

• ■

■

V .«

••••■.•*" 1

••

• * .

• ■ «

• ■ •

r*
i

»<■

280

OSWIN W. WILLCOX.

and the tube was kept evacuated until just before the period for which
it was desired to measure the decomposition pressure of the material
According to Saposhnikov* the relation of velocity of decomposition
of a nitrocellulose to temperature for temperatures between 125° and
140°, and between 145° and 155° can be expressed by the equations

(dv\ , , (dv\

\~f.] max. = 24.6 + 0.201/ and I ■\- J max. = 136.5 + 0.985/, respect-
ively; the ratio of reaction velocity to temperature within each of these
ranges of temperature can therefore be represented by a straight line.
Solving the two equations simultaneously, 143® is indicated as the tem-
perature at which these two lines will intersect. At this temperature

/JV

iOO

ca/fi^r 6HomMf £rrrcr or

/iATf Of P£C0rifOS/r/(¥f //f

/Jtf

/tW

/40

Me M4

/¥S

/4a

/so

Fig. 10.

a sharp change in the ratio takes place in either direction. Before I
became aware of Saposhnikov's work I had thoroughly studied the tem-
perature relations of the ObermuUer test, employing a temperature
interval of 2** instead of 5°, as employed by Saposhnikov. My results
between 138** and 148° are as follows (volume, 50 cc):

Temperature: 138° 140° 142° 144° 146*" 148''
Pressure (mm.) 38 53 77 103 130 183

From these data are obtained the curve shown in Fig. 10. Instead

* Russ. Phys. Chem. Soc, 38, 11 86 (1906); reviewed by H. M. Gordin in Chemical
Abstracts, May 20, 1907.

CASEIN IN cow's MILK. 28 1

of one sharp change in the rate of decomposition at 143® there are here
shown two such breaks, one at 140** and one at 146°. As the Obermuller
apparatus uniformly gives concordant results within ±2 mm. the posi-
tion of the curve is sensibly free from experimental error.

CHKinrAi> Laboratory,
Samsv Hook Provino Ground.

[Contribution from thb Agricultural Chemical Laboratory of the Wiscon-
sin Experiment Station.]

VAIUATIONS IN THE AHOXTirT OF CASEIN IN COW'S MILK.

B. B. Hart.
Received November 11, 1907.

There is a general belief among dairymen and some dairy chemists
that casein and fat are present in milk in very constant relative propor-
tions; that given the percentage of fat in milk, the percentage of casein
can be directly calculated by rule. This rule was formulated by Van
Slyke* and is based on averages of numerous analyses made at the New
York Agricultural Experiment Station. The rule is to be applied espe-
cially to milks ranging from three to four and one-half per cent, of fat and
is stated as follows: to find the per cent, of casein in milk when the
per cent, of fat is known, subtract 3 from the per cent, of fat in milk,
multiply the result by 0.4 and add the result to 2.1. The limitations
placed on the rule as applicable to milks containing but from three to
four and one-half per cent, of fat led the writer to inquire how applica-
ble it might be to milks. of higher fat content. Hill,* as early as 1890,
showed that in individual cows the proportion between fat and casein
is widely different. He, however, obscured this important fact by con-
clusions based on averages of many milk analyses. His conclusion was
that normal milks, whether rich or poor, have on an average, one-fourth
as much casein as total soUds, though he further says that single sam-
ples may depart widely from this standard.

Shuttleworth,* from work on individual cows, showed that a consid-
erable variation in the proportion of casein to fat existed among differ-
ent animals, and that a ratio established for one period of lactation in
any single animal may not be the same as the ratio found at some other
period for the same animal.

A priori there seems to be no good reason why we should expect a defi-
nite quantitative relation between these two constituents of milk. They
are entirely unlike in chemical constitution and their elaboration has
.been along different lines of synthesis. If we could suppose that they

' Modem Methods of Testing Milk," p. 192.
* Fourth Annual Rept. Vt. Agr. Exp. Sta.
*Rept. of Ontario Exp. Farm, 1895.

B. B. HAKT.

resulted from the splitting of a. single chemical entity, then there
i be reason for a definite relation between the amounts of these
ubstances in this secretion. But the facts regarding the production
Ik constituents do not appear to support any such hypothesis. The
variable milk constituent we have Is the fat, which may rise and
xim day to day in no incondderable arrount, dependent on feed and
nvironment to which the animal is subjected. During such fluc-
>ns of the fat' content, the casein may remain constant. Instances
n record where the fat content of the milk has dropped from 3.25
;nt. to 2.20 per cent., while the casein content of those milks remained
d8 and 2.18 j)er cent., respectively. This at least would indicate
,he precursor of these two important milk constituents was not a singfe
ical entity, which seems a necessary assumption if the two substances

0 remain in constant relative proportion. On the other hand, it
ites rather a differentiated process, with the formation of fat and

1 as distinctive and dependent upon inherent cell characteristics,
ain, the relation of fat to casein in the cow's milk established for
leriod may not be found to be the same at some later period of lac-
1. In fact, it appears to be the normal tendency for the nitrogen
ounds of milk to increase relatively to the fat with the advance of
ion.

rther, while there is no doubt that the rule above formulated is
accurate when applied to the mixed milk of herds made up of grade
lis, it appears entirely possible that it might not be applicable to
1 milk produced by animals of high fat-producing capacity.
The very fact that the efforts of progressive dairymen is to displace
Dw fat-producing animals with animals producing higher fat yields,
introduce a tendency to move away from the application of the

)m the standpoint of the breeder of dairy cows and the cheese in-
y, it would appear extremely important to know whether or not
lis producing milks of five to six per cent, fat content, are produc-
definitely related percentage of casein, and with animals producing
to five per cent, of fat, whether that same relation holds true. If
;s not, then it would appear that here is important work for breed-
f dairy cows in the selection and production of animals producing
more fitted for the butter or the cheese industry, as the case may be.
idies of the University Herd. — No attempt was made to follow the
il through long periods of lactation. The only attempt made here
o learn whether the variations of fat to casein in different animals
if any significance and what might be expected any ti^ie an analy^
nade.
Technical BuU, No. i, N. Y. Agr. Exp, Station.

CASEIN IN cow's MILK.

283

Casein determinations were made by the Official Agricultural Chem*
ists' method. The only variation was the use of the factor 6.38 instead of
6.25. The samples were from a mixture of night's and morning's milld
Fats were run by the Babcock test from composite samples selected
over a period of one week in the usual way. The collection of the sam-
ples for casein determination was made in the middle of the week dur-
ing which the fat sample was being taken. The analyses cover a period
from July 26th to August 7th. The data relating to the animals are
arranged in the table according to breed. The table contains, besides
the percentages of fat and casein, a column showing the amount of casein
calculated by Van Slyke's rule from the fat content. Besides these data,
two separate columns show the relations of fat to casein and casein to fat*

Analyses op Mtlk op Univbrsity Herd.

Breed.

No.

Per cent,
of casein
found.

Per cent.

orca<<ein

calculated.

Percent,
of fat.

Relatic

offal

•nd caa«

m Relation
t ofcaacin
>in. and fat.

Jersey

. . . I

2.45

2.60

4- 27

1.74:

1 0.57:1

311

2.83

4.83

X.55:

I 0.64: I

3.31

3.28

5.95

1.79:

I 0.55:1

3.65

324

5.85

1.60:

I 0.62:1

300

2.81

4.79

1.59:

I O.62M

2.92

3- 30

6.02

2.06:

I 0.48:1

Guernsey

... 7

3.50

3.38

6.21

1.77:

I 0.56:1

2.77

3- 30

6.01

2.16:

I 0.46: I

309

3.02

5.31

1. 71:

I 0.58:1

3.12

3.08

5.46

I.75--

I 0.57:1

2.91

2.85

4.89

1.67:

I 0.59:1

2.60

2.94

5. II

1.96:

I 0.50:1

2.47

305

5-37

2.17:

I 0.46:1

2.98

3.31

6.04

2.03:

I 0.49:1

Holstein

...15

2.10

2.17

2.96

1. 41:

I 0.70:1

2.13

2.17

3.19

1.49:

I 0.66: I

2.50

2.27

3.44

1.57:

I 0.72: I

2.16

2.32

3- 56

1.55:

I 0.60: I

1.88

2.17

3.18

1.52:

I 0.65:1

2.15

2. II

3.03

1.40:

1 0.70:1

Brown Swiss. . . .

. .. 21

2.66

2.61

4.29

1. 61:

I 0.61:1

2.70

2.49

3-99

1 . 51 ••

I 0.66: I

Ayrshire

. . . 23

2.56

2.34

3.61

1. 41:

I 0.70:1

•

2.14

2.87

4.93

1.57:

I 0.63: I

2.47

2.28

3.47

1.40:

I 0.71:1

2.61

2.32

3-57

1.36:

I 0.73:1

The table shows that in a large number of instances the application
of the rule gives data agreeing closely with actual determination. There
are, however, as Dr. Van Slyke has already emphasized, high fat milks
where the agreement is not very close, and actual determinations would
alone disclose their true casein content. No. 13, with a fat content in

284 B. B. HART.

the milk practically identical with that of No, 10, nevertheless shows a
casein content 0.65 per cent, lower than the latter.

The table further shows that there is considerable variation not only
among animals of different breeds, but between animals of the same breed.
Percentage variation of casein ranges from 1.88 for Holsteins to 3.65 tot
Jerseys, Among Holsteins themselves the range of percentage is from
1,88 to 2,50, while among Jerseys it is from 2.45.10 3.65. Reduced to a
ratio of pounds of fat to pounds of casein, we have among different breeds,
for instance, at the time these analyses were made. No. 6, a Jersey, show-
ing 2.06 pounds of fat for every pound of casein, while No, 17, a Hol-
stein, shows 1.37 pounds of fat for one pound of casein. No. 8, a Guern-
sey, shows 2.16 pounds of fat for every pound of casein, while No. 15,
an Ayrshire, shows 1.4 pounds of fat for one pound of casein. These
ire the extreme cases among the number of animals investigated.

Among breeds themselves, we have No. 6, a Jersey, with a ratio of 2.06
punds of fat to i of casein, while No. 3 of the same breed shows a ratio
Df 1.79 pounds of fat to every pound of casein. Stated in another way,
No. 6 shows 0.48 pound of casein for i of fat, while No. 3 shows 0.55
pound. The data on the milks of these two cows clearly shows that
relative to their fat, No. 3 is the greater casein producer. The yield of
:heese from the milk of No. 3 must necessarily be larger, under uniform
conditions of manufacture, than from that of the other animal. Again,
No. 7 showed a relation of 1.77 pounds of fat to i of casein, while No. 8
showed the relation of 2.16 to 1. No. 7 shows the relation of casein to
[at as 0.56 to 1, while No. 8's relation is 0.46 to i. The milks from these
inimals were at about the same period of lactation.

A further consideration of the table reveals the fact that among breeds
:he Holsteins, Brown Swiss and Ayrshire uniformly show a higher rela-
;ive proportion of casein to fat than do the Jerseys and Guernseys. It
ilso shows that certain individuals among the two latter breeds may
ihow as high a relation of casein to fat as certain individuals among the
)ther breeds.

What these animals will do for a whole year is not known, but enough
lata is at hand to emphasize the fact that individual differences in casein-
iroducing power do occur among animals of different breeds, and surely
nay occur among animals of the same breed, and that the casein-pro-
iucing power does not necessarily bear any close relation to the fat-pro-
lueing power. That a higher fat holding milk means an increased
casein holding milk is not here denied, but that the increase is in a fixed
»roportionate ratio, the data do not support. It emphasizes, it seems,
he fact, that the casein -producing function is in part, if not largely,
ndividualistic, and capable of being used in producing dair\' t^'pes of
Liiimals, either for an industry in which fat plays the most important

NOTES. 285

r6fc, or for a cheese industry, where both fat and casein are primarily con-
cerned.

Summary.

1. The relation of casein to fat in cow's milk is a variable one.

2. One of the prime factors controlling its relation is individuality.

3. The relation of casein to fat varies among animals of different breeds
and among animals of the same breed.

4. Direct determination of both fat and casein seems necessary in de-
termining the value of the milk of any single cow for cheese production.

NOTES.

The Use of the Centrifuge, — ^Attention has recently been called to the
advantage of the laboratory use of centrifugal action for separating
crystals from their mother-liquor — a process which has long been of
great service in technical operations on a large scale. ^ The object of
this note is to point out certain important precautions necessary in the
use of this highly serviceable apparatus. The word of caution seems
to be especially demanded because new apparatus is being put upon the
market by several firms, and the novice may be unfamiliar with the
intensity of the centrifugal effect, and the consequent danger inherent
in improperly constructed machinery.

It is well known that the forces acting to drain out the liquid in a cen-

4x*nV
trifuge are times as great as they would be in a gravity- vat with

a perforated bottom, if n == the number of revolutions per second, r
the radius, and g = 980.6. Thus if n = 20 (i. €., 1200 revolutions per
minute) and the radius of the centrifuge is 10 centimeters, the drying
is nearly 160 times as great as that effected by gravity — a very great
advantage. It must not be forgotten, however, that the strain upon
the apparatus increases in the same proportion, being quadrupled for
each doubling of the speed. Therefore with high speeds great strength
is necessary. Even great steel fly-wheels sometimes burst under their
strain. For this reason, centrifugal apparatus constructed of fragile
material should never be nm rapidly, and even with the simplest and
strongest apparatus, the machine should always be surrounded by a very
strong casing or box of wood or metal, so that no harm would result if any-
thing should break. For the same reason rapidly revolving centrifugal
apparatus should never be constructed of glass, unless the glass is enclosed
in metal in such a way that the fragments will not fly if broken. Glass
apparatus is frequently not well annealed, and is liable to break under
the heavy strain.

* Richards, This Journal, 37, 104 (1905); Ber., 40, 2771 (1907). Kdthner, Chem.
Ztg., 1907 (No. 73).

At Harvard the funnel-centrifuge, alluded to by KSthner, is. made of
itinum, but porcelain funnels and receivers may be used without dan-
r if the rate of revolution is not too great. I have never dared to
e glass funnels in this apparatus. The porcelain basket-centrifuges,
which seveml forms are on the market, will stand considerable strain;
e speed at which they may safely run varies with the form and stout-
ss, and should be carefully determined for different loads and indicated
■ the manufacturer. The porcelain receiver surrounding them is never
-ong enough to hold the fragments if the basket should break; tbei?-
re this whole apparatus also should be surrounded by a strong guard-
X. Both porcelain and glass should be supported on some Idnd of mb-
r cushion, so as to distribute the strain as evenly as pos^bl&
It is not out of place to call attention to another somewhat less se-
lus but nevertheless important precaution, namely, the equal distti-
tion of the load. This is essential if the apparatus is to run smoothly
d the strain is to be evenly distributed. In the case of the funnel-
itrifuge the adjustment is very readily accomplished by hanging the
posite funnels upon the two arms of a common balance, and filling
em to equilibrium with similar crystals about equally moist. In the
se of the basket-centrifuge, the distribution must be made with a
itula, before the machine is started.

If these simple and obvious precautions are taken, the centrifuge
11 be found, as has been said before, a very \^luable aid in the puli-
ation of small quantities of substance in the laboratory of the inves-
;ator. In the course of twenty years no serious accident has resulted
>m its use at Harvard, and much has been gained. As has been said
fore, the gain to be expected varies greatly in different cases; it is
;atest in the case of very soluble substances which do not carry im-
rities in isomorphous solid solution with them into the solid state,
d least in the case of slightly soluble substances, which contain isomor-
ous contamination.

To summarize the contents of this brief note — the value of centrifu-
I action in purifying substances has been once more emphaazed,
t the importance of equal distribution of the load, the danger of using
Lss or other very fragile material in the centrifuge, and the necesaty
caution in regulating the speed and in always guarding the operator
a stout casing around the machine, is pointed out.

T. W. RiCHAROS.

Apparatus for the Centrifugal Drainage of Small Qwtnlities of Crys-
s. — The high efficiency and importance of centrifugal drainage in
: removal of mother liquor in purification by crystallization has re-

NOTES. 287

cently been strongly emphasized by T. W. Richards,' who describes vari-
ous convenient forms of apparatus for the purpose. The use of these
devices in this laboratory has led to the construction of a new very con-
venient modification of centrifuge which not only makes possible the
complete removal of mother hquor from small quantities of substance
without undue loss of material and in a cleanly fashion, but also provides
for the preservation of the mother liquor and rinsings from the crystals
in an equally satisfactory manner. This latter point is frequently of
importance to any chemist who is purifying small quantities of precious
material, especially if the substance is very soluble.

The following diagram explains the construction of the device, which
has already been briefly described :*

A is a cup, preferably of aluminum on account of its lightness and
deanUness, with trunnions by means of which it may be hung upon
one end upon of a metallic arm at-
tached to the head of a centrifugal
machine. A similar cup, suitably
wd^ted, serves as a convenient
counterpoise. These cups are of the
form commonly employed to contain
the flasks used in the determination of
fat in milk.* A hard rubber sleeve B
fits loosely upon the top of the alumi-
num cup, the inside of the sleeve
being turned to the proper size and
angle to hold a platinum Gooch cruci-
ble C, of any desired aze, which
serves as the basket for the crys-
tals. Sleeves of metal, platinum plated, might be used, but are ob-
jectionable on account of their weight. With a low-speed centrifuge,
a porcelain Gooch crucible would probably be safe. For obvious rea-
sons the top of the crucible should extend a few millimeters above the
sleeve. The mother Uquor drains into a platinum crucible D. Where
platinum is unnecessary for the sake of purity of the mother liquor, the
CTudbk D may be replaced by a suitable stout glass vessel. Thin beak-
ers, however, are likely to be fractured by the weight of the liquid, if
the speed of revolution is high. Although the surface of the aluminum
cup may easily be kept bright and shows no tendency to abrade and
thus contaminate the mother liquor, all possible danger from this source
may be avoided by lining the inside of the cup with a cylinder of platinum

' Tbis Journal, 37. 3io (1905).

' Baxter and Coffin, Ibid., aS, 158a (1906).

■ Ahnniunm cops of this sort are made by the Inteniatioiial Instrument Co.,

turned over the upper edge of the cup. The latter precaution is
rourse more necessary when acid vapors are emitted by crystals ot
ther liquor. The Gooch crucible, or the whole top of the cup, may
rovered with a circular piece of platinum foil, the edges of which have
n turned down to hold it in place. A hole in the bottom of the alumi-
n cup faciUtates the removal of the vessel containing the motber
lor,

t is, of course, possible to make the apparatus more elaborate, for
;ance, by providing the aluminum cup with an especial porcelain
platinum lining. The system as described above, however, has the
at advantage that it may be constructed with materials available
most laboratories. G. P. Baxter.

Cakbiudok. Mau..
NovemlKT M. 1907.

REVIEW.

RECEHT PROGRESS IK PHYSICAL CHEMISTRY.

P. G. CoTTmiu..
RccelTciJ Norember ij, 1907.
Tie issuance by our Society this year of "Chemical Abstracts" has
lewhat changed the requirements for the annual .reviews of special
ics. Heretofore their chief aim has been to present a brief synopsis
the more important foreign literature, to supplement the "Review oi
lerican Chemical Research" included in the monthly numbers of the
imal. Since, as far as collection of data is concerned, all of this field is
V covered by the "Abstracts," the author has, in what follows, con-
id himself to the discussion of a few selected topics which have either
racted special attention of late or seem to open up or emphasize new or
viously neglected fields of inquiry. In order to present some of these
:heir true perspective it has seemed necessary to trace the same line of
ught back among the earlier workers, and in this respect there has been
attempt to limit the present article strictly to what has appeared dur-

the past twelvemonth. As heretofore, the subject of radioactivity
. been left for treatment in a separate article.

'erhaps the most significant trend of recent work is to be found in the
icentmtion of eSort toward narrowing the gap between molecular and
lar phenomena. This field in which, among other important matters,

whole subject of colloids ultimately belongs, stands to-day in very
ch the same relation to physical chemistry and ordinary mechanics
t physical chemistry stood to physics and chemistry twenty or thirty
irs ago. The classification of natural phenomena into sharply defined
ijects may in most cases be interpreted simply as an admission that we

omitting a region between, in which, as we enter it from either side,

methods of treatment gradually fail us. Thus it was that the trouble-
ne and outgrown distinction between chemical and physical combina-
ti was swept away by Gibbs, and under the broader conception ot
base" and "component" the two fields merged in one as far as hetero-

REVIEW. 289

geneous equilibrium was concerned; but here Gibbs' definite contribution
ended. The very conception of a ** phase" in whatever words it is de-
fined rests eventually upon the distinction between molecular and molar
magnitudes. To emphasize this point of view we may frankly define a
phase in the sense in which Gibbs uses the term, as a portion of matter
homogeneous down to molecular magnitudes.

This gives us a perfectly sharp classification of systems until we reach
the colloids. Then classification by its means ceases to be independent
of theoretical assumptions, and depends upon where we arbitrarily draw
the line between molecular and molar magnitudes in a series which passes
through all possible values intermediate between the typical unquestioned
extremes. It is this characteristic of a boundary or frontier region which
has lent the subject of colloids much of its fascination for the investiga-
tor. Many, apparently overlooking the necessity that colloid phenomena
must eventually furnish the connecting link between molar and molec-
ular types, have striven to build up a new and entirely independent science
on this foundation. This tendency has undoubtedly been greatly stimu-
lated by the importance of colloids in biological problems. Where we
may hope for the greatest permanent advances is, however, in the ex-
tension, with the necessary modification, of the well established notions
from over both its border line^ into the new field until they meetaiid
merge in one more generaUzed set covering the whole extent. When,
for instance, we are able to trace the phenomena of osmotic pressure
up through the solutions of substances of large molecular weight such as
the starches and simpler proteins on to the inorganic colloids and finally
connect it with its equivalent in the unquestioned suspensions, we Will
have done much to realize Newton's dream of laws broad enough to em-
brace both chemistry and astronomy, stellar and atomic mechanics.
Several recent papers in this direction deserve particular notice. Einstein
(A. Einstein, Ann. Physik [4], 17, 549-60 (1906); 19, 289-306 (1906);
22, 569-72 (1907)); with the aid of the molecular kinetic theory has
developed the laws for **the osmotic pressure of suspended particles,'*
also the ** viscosity of a liquid holding in suspension a multitude of rigid
spheres." The law of this generaUzed osmotic pressure, as we might
style it, proves to be identical with that we have long been familar with in
purely molecular phenomena, while the change in the apparent coefficient
of viscosity of a liquid produced by the introduction of suspended particles
is shown to be equal to the ratio of their total volume to that of the liquid,
at least for small values of this ratio where the suspended particles are
large compared with the molecules of the liquid.

From these two relations he then proceeds to calculate the rate of
diffusion, in terms of the size of the particles and by comparing the re-
sults thus obtained with the values of molecular dimensions derived
from other sources and the known diffusion coefficients of true solutions,
shows the new formulae to agree with our present conceptions in those
regions where the two overlap. A few of the most important formulae
derived by Einstein, and the values of the symbols used are as follows:

. , ^ RT n , , ^ RT I

(0 P = -^ XT (3) D =

V N '"'^ N 6^Kr

(a)|'=x+* (4)X '<^^t'^\^,

P, V, R and T have the usual signification as in the gas laws.

N — number of molecules in a gram molecule.

n = number of suspended particles in volume V.

r « radius of suspended particles.

D — diffusion constant of suspended particles.

K — specific viscosity of pure liquid,

K' — spediic viscosity of liquid with suspended particles.

^ ■= combined volume of suspended parades in unit volume of the
liquid.

w — ratio of circumference to diameter of circle.

A <- mean displacement of a particle in a particular direction duiing
time t,

c — molecular concentration of dissolved substance.

Perhaps it is only fair to point out that the weakest point in these papers
as a whole lies in the author's tacit assumption in the original development
that the suspended particles are large as compared with the structure
of the liquid and then in subsequently applying the formulae to cases of
typical solutions such as sugar in water where the question might be
raised whether the sugar molecule in solution were large enough in com-
parison to the water molecule to justify this procedure. The author's
detennination of 'p by equation (2) for a i per cent, sugar solution leads to a
vahie four times that of the solid sugar, which he explains by each sugar
molecule carrying an envelope of over 30 water molecules with it. These
possible objections however lose significance entirely when we tum to
suspended particles of relatively large dimensions, which are, after all,
of chief interest to us in the present connection. Einstein suggests ui
passing that this opens the way to settling the disputed question as to
the ori^ of the Brownian movements. What makes this work of particu-
lar ^gmficance at present is the surprising manner in which its predictions
have been met by entirely independent experimental evidence from
another quarter.

It is some years ^ce Siedentopf and Zsigmondy brought out their
"ultramiscroscope " which is simply an application to the study of ex-
cessively minute objects under the miscroscope, of a principle which has
long been familiar to us in viewing excessively distant objects in the
telescope; viz., that an object which subtends thousands of times too
small a visual angle to have its outlines seen, becomes visible as a point of
light when strongly illuminated and viewed against a dark background,
provided that the distance between such points is still perceptible in the
ordinary sense. Thus by powerfully illuminating a submiscroscopic
suspension of particles by a beam of light at tight angles to the line of
»ght, we are able to perceive the position of each particle, although its
shape is not merely far beyond our power of vision, but also beyond the
theoretical limits of the miscroscope, which are set by the wave length
of light itself. The relative sizes of the particles may be estimated roughly
from their apparent brightness. The limits of ordinary miscroscopic
vision may fairly be set at about 0.00025 mm. (*. e., 250 ^/a) while the
ultramiscroscope extends the limit of perception down to particles about
6 or 7 ^/i in diameter. The closeness with which this approaches ordinary

RBVIEW. 291

molecular dimensions may be judged by comparison with the values
derived from the molecular kinetic theory. Thus Loby de Bruyn sets
the diameter of molecules of soluble starch at 5 /^ft, Jager those of chloro-
form at 0.8 /Aft and O. E. Meyer the water molecule at o.i ftjut. A most
excellent and conservative view of the progress made in this subject up
to about two years ago is to be found in Zsigmondy's own book. (Zur
Hrkenntnis der Kolloide; R. Zsigmondy, Jena (1905), G. Fischer 186 pp.).
Much valuable work has since appeared along the same lines, but we
can consider here only one of the most recent developments which bears
directly upon the theoretical matters pointed out above.

Svedberg (T. Svedberg, Z. Elektrochem., 12, 853-60, 909-10 (1906)) in
working out the technique of the above method has devised a very in-
genious scheme for observing and measuring the amplitude and period
of vibration of colloid particles, and has applied this to the study of
colloidal solutions of a number of metals including sodium, potassium
and calcium in various organic solvents, the method of preparation of the
alkali metal colloids being also due to Svedberg himself (T. Svedberg,
Ber., 38, 3616-20 (1905); 39, 1705-14 (1906)). The mode of opera-
ting consists essentially in allowing the colloid solution to flow through
the field of the ultramiscroscope at a known speed. The motion of each
individual particle due to the flow, compounded with that component
of its vibratory motion at right angles to both this and the optical axis
of the miscroscope produces a path clpsely resembling a sine curve which
is traced so rapidly by the luminous point that it appears as a continuous
line of light which allows of its dimensions being estimated with fair
accuracy on an eye-piece micrometer. Svedberg finds that for particles
of equal size and character in different liquids, the product of ampUtude
into the coefficient of viscosity of the Hquid is a constant as required by
Einstein's formula. The numerical value of the constants as derived
from the observations is about six times that computed from Einstein's
formula, but remembering that the theoretical formula contains two very
uncertain quantities — diameter of the particle and number of molecules
in a gram molecule — the agreement is astonishingly close, more especially
as Svedberg's experimental data were published without knowledge of
Einstein's work, only his last paper above referred to taking cognizance
of the latter. What lends this kind of work special interest at present is
the growing deadlock between what we may call the extreme thermo-
dynamic school including such men as Mach and Ostwald on the one
side and the extreme mechanists such as J. J. Thomson, Rutherford and
Arrhenius on the other. The question is as to which represents the
more fundamental conception of natural laws. Each form of treatment
is applicable to the whole range of molar phenomena. It is only in case
we come to deal experimentally with individual molecules that a choice
between them becomes anything more than a matter of philosophic
taste. If we do come to deal with individual molecules in the sense of
the kinetic theory, Maxwell's Demon will have been realized and the
limits of applicability of the second law of thermodynamics exceeded,
and the old quest for perpetual motion of the second type once more
reinstated among serious scientific pursuits. On the other hand, if ther-
modynamics, as far as the second law is concerned, succeeds throughout
in its application to indefinitely small portions of matter, then the molec-

12 Review.

ar theory, in its present form at least, must be given up as an objectiw
mception of nature.

In this connection a remark of Ostwald's (Z. physik. Ckem., 57, 383
907)) in reviewing Zsigmondy's book is significant as showing the
feet that such recent work is having on even the most extreme repre
ntatives of the non- mechanistic school. He says "Von den vielen
igentiimlichkeiten sei in erster Linie die merkwiirdige Eigenbewegung
;r submikroskopischen Teilchen erwahnt, die an die Brownsche
ewegung erinnert, aber von ihr verschieden ist. Der Berichterstattci
luss bekennen, dass er noch nicht absehen kann, wie diese ausser Zveifel
ehenden Tatsachen sich ungezwungen mit dem zweiten Hauptsatze
erden vereinigen lassen. Hier scheinen Maxwell's Damonen, die man
n molekularen Gebiete als ungefahrlich ansehen durfte, im Endlichen,
Sichtbaren ein freies Feld fiir ihre experimentelle Widerlegung des
veiten Hauptsatzes zu baben." Svedberg has more recently (T. Sved
;rg, Z. physik. Ckem., 5g, 451-8 (1907)) contributed another article
[ggesting certain hypothetical mechanisms violating the second law.

It is probable that the wonderful success which mechanistic methods
ive met with in the field of radioactivity has had much to do with
rengthening their position of late among chemists. To follow out this
ae would carry us too far afield, but it may be pointed out in passing
lat owing to the extreme delicacy of the methods of measurement the
idioactive phenomena offer a promising field for the study of indi\'idual,
:omic, and molecular phenomena, as illustrated for example in the
Hnthariscope and yet more quantitatively even in a recent paper by
.ohlrausch (F, K. W. Kohlrausch, StUsb. Akad. Wiss. Wten., 115 [za].
?3-82 (1906); Physik.-chem. Cenlr., 4, 219) upon variations in the rate of
idioactive change from the mean value.

Turning now from the progress made along molecular kinetic hnes to the
ibject of thermodynamics proper, we meet a set of no less important
ivances from the chemist's standpoint. These run back genetically
ir several years and include contributions from a number of well-known
orkers, chief among whom may be mentioned, somewhat in chronological
rder, the names of: Lewis (Proc. Am. Acad. Arts and Sci., 35,
-38 (1899}; 36, 145-68 (1900); also Z. physik. Chem., 32, 364-400;
5, 343-68), Richards (Proc. Am. Acad. Arts and Sci., 37, 1-17 (1901);
J, 399-411 (1902); 38, 293-317 (1902)), Tammann (Krystallisiereii und
chmelzen { 1 903) espec., p. 42), van't Hoff (Boltzmatin Testschrif t, p. 233-^ i
1904)), Trevor {_/. Physic. Chem.. 9, 269-310 (1905)), Bell (/. Physic,
hem., 9, 381-91 (1905)), Haber (Thermodynamik technischer Gasreak-
onen {Miinich, 1905)), the whole as viewed at present seeming to have
ilminated in a most remarkable paper by Nemst {Nachr. Ges. Wiss.,
otiingen (1906), [1], 1-40), pubhshed early last year and since
rought more directly before American chemists by the Silliman lectures
'Experimental and Theoretical Applications of Thermodynamics to
hemistry," Chas. Scribner's Sons, New York, 1907, 123 pp.) pven by
rofessor Nemst in New Haven, and of which it formed the basis.

It represents what promises to be the most important contribution of
lermodynamics to chemistry since the work of Gibbs and Helmholtz,
ad forms a fine example of the magnificent results which the pure ther-
lodynamical method is capable of accomplishing within its most fruitful
eld of application. The presentation in the original is already so com-

kEVlEW. ^9i

pact and derives its individual steps from such diverse sources that it
is impossible to give comprehensively even an outline of it here. (See
Abstract by Abegg, Z. Elektrochem,, 12, 738-43 (1906)). In as far as the
outcome of such a thorough-going and detailed contribution as this can
be reduced to epigrammatic statement we might perhaps say that it rep-
resents the amplification and correction of the old *'law of maximum
work" so as to cover clianges in heat capacity and concentration in the
reacting system, it being recognized that the disregard of these factors
was what most seriously limited the applicability of the law as first ad-
vanced. Nemst points out that in its original form the law was true for
all systems at absolute zero and very near the truth at higher tempera-
tures for many changes involving only pure solids and liquids or at best
concentrated solutions. He then introduces as his fundamental hy-
pothesis the proposition that not only are the free and total energy of any
system equal at absolute zero but their first derivatives with respect to
temperature are also equal at this point. This necessitates specific heat
being a strictly additive property at absolute zero and consequently the
variation in heat of reaction with temperature depends solely on the tem-
perature coefficient of specific heat. " Starting now with the well-known
van't Hoff equation for equilibrium

assuming in the neighborhood of absolute zero a difiference of 3.5 cal.
between the molecular heat of the vapor at constant pressure and that
of its corresponding solid or liquid, he deduces the working equation

log K = - ~^-- + Svi.75 logT -I- -^^^ T + 2vC
4-571 A 4.571

0^= heat of reaction for T = o.

V = number of molecules of each species counted positive on one
side of equation and negative on the other.
? = temperature coefficient of atomic heats.
C = specific "chemical constant" for each molecular species.

The additive nature of these "chemical constants" is perhaps the most
significant result of Nemst's development. This permits us to predict
the true chemical equilibrium in any system from the heat of reaction
and certain purely physical constants of the individual substances origi-
nally brought together, and since the purely physical phenomena of
vaporization is merely a special case of equilibrium, the numerical value
of C for each substance may be derived from a study of the temperature-
vapor pressure curve for that pure substance. By a skilful use of a
modified law of corresponding states, Nemst has even gone further and
made it possible to calculate this constant C if we have an approximate
value for the critical pressure of the substance and in addition inow for
some one temperature: (i) the vapor pressure, (2), the heat of vaporization,
(3) the temperature coefficient of the latter. The value of these * 'chemical
constants" seems to vary between relatively narrow limits. Of the
twenty- three examples given, the lowest is hydrogen, 2.2 and the highest
alcohol, 4.1. They rise fairly regularly with the boiling-point in the case
of simple substances but are abnormally high in the case of polymerizing
substances. For the application of the equation to actual cases, including

294 REVIEW.

a discusdon of permisable simplifying assumptions, the original must be
consulted. (See also K. G. Falk, This Journal, 39, 683-87 (1907)).
Among the interesting consequences of the theory may be mentioned the
deduction of a more exact form for Trouton's rule including a logarith-
metric terra, and the treatment of the question of the stability of chemical
compounds from a perfectly general standpoint. This latter aspect has
tieen developed at some length by Brill (Z. physik. Chem., S7, 711-38
(•907)) both for homogeneous and heterogeneous systems. For the case
of a gaseous compound, each of whose molecules dissociates forming two
new molecules, as, for example NjO, into aNO,, he deduces for a pressure
of one atmosphere, the following table of approximate temperature (T)
on the absolute scale, at which the dissociation will be one-half complete
for reactions with a given heat of reaction (Q) at constant pressure and ordi-
nary temperature , irrespective of the specific character of the substances in-
volved.

Q. T (absolute).

10,000 cal. ago"

15.000 405

20fioo 525

30,0(x> 780

50,000 1220

100,000 2350

200,000 4500

Leaving the subject of energetics and passing on to the properties of
matter, we must place in the forefront Landolt's paper (H, Landolt, Silzb.
Akad. Wiss. Bert., 8, 262-98 (1906); Z. pkysik. Chem., 55, 589-621
(1906)) on the change in total weight during chemical reaction. This
paper represents the result of years of patient work (H. Landolt, Z.
pkysik. Chem,, 12, 1-34 (1893)) in what to many must have appeared a
hopeless, not to say thankless, cause. Step by step the accuracy of
the experiments were increased until the probable error of a complete
experiment had tieen forced down to so small a fraction of the minute
changes which Landolt was here in search of, that he could be podtive
of the existence of new phenomena which all our work since Lavoiaer
had demonstrated to be undetectable. To summarize the actual re-
sults it may be said that out of the 14 reactions of various types studied,
only two gave systematically a change in weight decidedly larger than
the errors of observation. These were the reductions of silver sulphate
or nitrate by ferrous sulphate and the reduction of iodic acid with hy-
driodic acid. The experiment involved a mass of reacting material of
about 250-350 grams, the total experimental error determined from
blank control experiments not exceeding 0.03 mg. Every one of nine
separate experiments on each of the above reactions gave a loss in weight
which ranged in the first reaction from 0.068 nig. to 0.119 mg. and in the
second reaction from 0.047 to 0.177 nig. Out of the 75 experiments
performed in all, 61 showed loss in weight and the greatest gain in weight
in any experiment was 0.019 mg. or well within experimental error. From
the data thus far obtained it does not seem possible to connect the
type of reaction with the sign or magnitude of the changes in weight.
In the paper the possibility is suggested that the loss in weight merely
represents the escape of electrons. Landolt's transfer from the University

REVIEW. 295

to the Reicbsanstalt temporarily interrupted this work but it is now
reported that he has it actively under way again.

The relation of chemical constitution to crystal structures, although
almost of necessity a close one, has been one of the least productive fields
of inquiry either for the experimenter or the theorist. We have, to be
sure, made some progress in the Umited field where we have had the
rotation of the plane of polarized light to help us, but this merely scratches
the surface of the general problem, and leaves the most fundamental
geometrical aspects of the question as obscure as ever. It is with all the
more satisfaction, therefore, that we welcome the recent attempt of Pope
and Barlow (/. Chem, Soc, 89, 1675-1744 (1906) ; C. A., 1907, 1809; see also
Review in Am. Chem, J., 37, 638-54) to indicate a line of attack for this
problem which really reaches the basic geometric principles in a perfectly
concrete manner. The greatest stumbling block at the very outset has
been the erratic way in which series of related compounds are distributed
among the various crystallographic systems or classes of symmetry.
Tutton's classic work (see /. Chem. Soc^ for past ten years) on isomorphic
salt series has given us a fairly clear idea of the relations existing within
these latter series. The present paper on the other hand derives much
of its strength from its method being essentially independent of the
distinction between even the crystal systems or classes of symmetry, the
system of reference adopted being a modification of the topic axis idea
in which the sum of the valencies of all atoms in the compound takes the
place of the specific volume heretofore used in defining these axes. The
physical hypotheses upon which the whole development is based are:
(i) that crystal structure is determined by the arrangement of the atoms,
or better their spheres of influence, in such wise as to present the closest
possible packing of the assemblage and (2) that the volumes of the spheres
of influence of the different atoms in a compotmd are directly proportional
to the valencies of the atoms. Starting with these simple arbitrary
hypotheses it is surprising what a vast mass of heretofore apparently
unrelated chemical and crystallographic facts the authors are able to
coordinate and explain. As many of them are cited in the abstract
above referred to they need not be repeated here. It is interesting and
suggestive to compare this theory of valence, derived as it is primarily
from the crystallographic standpoint, with that of Abegg and Bodlander
which had its origin in electrochemical phenomena and the closely related
theory of Werner (see last year's review This Journal, 27, 908, also A.
Werner, Ber., 40, 15-69 (1907)) which was purely chemical in origin.
Perhaps the most fundamental difference between the present theory
and its predecessors is its freedom from any assumption of intrisic polarity
within the atoms themselves, the definite assignment of which forms so
important a part in Abegg's scheme. Pope's conception of valence
leads here to the admission that the valencies of different atoms need not
necessarily bear exactly the simple ratio to one another that we are accus-
tomed to assign. There is, to be sure, nothing in pure chemistry to con-
tradict this, but such a condition of affairs would be hard to reconcile
with Faraday's laws of electrochemical equivalents.

Traube, also (J. Traube, Ber.^ 40, 137-39 and 723-33 (1907)), has re-
cently attempted to refer valence to atomic volume, basing his deduc-
tion, however, on the refractive index of isotropic substances with the
aid of the electron theory and a number of rather specialized assumptions

296 KEVIEW,

as to the division of the atomic volume among the essential parts ol
what may be termed the atom as a whole. Still another radically different
view of valence based on an assumed eccentricity of the atomic nucleus
in its ether envelope is presented by Ensrud (G. Ensrud, Z. physik. Citem..
58, 257-87 (1907)) and supported by evidence drawn from the specific
heat of gases. Thus far our theories of valence have for the most part
dealt only with the building up of atoms into molecules but the time
seems fast approaching when they must also be made to connect with the
theories concerning the structure of the atom itself which the study of
radioactivity and related topics are fast forcing upon us. On the whole,
the subject of valence seems to present just at present an exceptionally
promising field for the chemist who can without prejudice freely conelate
the partial truth evidently contained in each of the above rather specialized
treatments of the general subject. We may, in passing, refer to a his-
torical sketch of the conception of valence recently published by Hen
{Chem.-Ztg., 30, 1273-5 and 1284-6 (1906)).

The subject of Hquid crystals still continues to attract much interest,
but thus far the work has been chiefly from the preparative side. Vor-
lander (D. Vorlander, Z. physik. Ckem., 57, 357-64 (1907)) has described a
number of very interesting and beautiful examples, chiefly azo compounds,
which show not only one but two liquid crystal modifications and some
which remain in the crystalline state over more than a hundred degrees of
temperature. Under certain conditions he states that it is possible to
obtain liquid crystals with well defined angles and straight edges. Leh-
mann (O. Lehniann, Z. physik. Ckem., 56, 750-66 (1906), and Jaeger
(F, M. Jaeger, fffc. Irav. chim., 25, 334-51 {1906}, and Versl.Akad.lVet. Am-
sterdam, 15, 345-8; 389-401 ; 401-10 {1906)) have described a long series of
esters of cholesterol and phytosteroI,all of which show anisotropic liquid
forms. According to Lehmann, every one of the cholesterol esters may be
had in two anisotropic liquid forms although in most cases, one of these is
labile while Jaeger claims a total of five possible, distinct, hquid phases,
for the cinnamic acid ester. Double bonds and the presence of certain
groups, such as the azo and azoxy, seem to favor greatly the appearance of
these very soft crystals, which indeed is what they really are. What may
seriously be called quantitative physical measurements on these crystals,
have as yet scarcely been undertaken, but with the rapidly increasing
list of substances now at our command to draw from, the opportunities
here presented for fundamental work, the direct comparison of crystal
building forces with surface tension and ordinary mechanical stresses,
cannot long remain unappreciated by investigators. Merely by way of
suggestive example may here be cited the work of Lehmann (O, Lehmann,
Ann. Pkysik, 21, 381-3 (1906)) on the relative orientation of the two
forms of crystal as they pass over into each other at the transition point;
also the fact that fluid crystals on glass surfaces seem always to orient
themselves with their optic axes as nearly perpendicular to these surfaces
as possible. Silver iodide seems to be practically the only typical elec-
trolyte which has been shown to possess an appreciably fluid anisotropic
form. Kohlrausch (W. Kohlrausch, Wied. Ann., 17, 642 (1882)) IcOTg
ago showed that the electrolytic conductivity of this substance under-
goes a tremendous change in passing from the hard to the soft crystalline
state at 145°, but that the change from the soft or liquid crystal state to

REVIEW. 297

that of an isotropic liquid occasions scarcely a perceptible break in the
temperature-conductivity curve. This, in the light of our present con-
ception of liquid crystals, brings us face to face with the question of the
degree of electrolytic dissociation in crystals of strong electrolytes.
Lorenz (R. Lorenz, "Elektrolyze Geschmolzener Salze," Bd. Ill 297-311;
Manogr. uber angew, Elektrochem,, Bd. 22, Halle (1906)) has come out
flatly in favor of the assumption of a high degree of electrolytic dissocia-
tion in solid crystalline salts, attributing their low conductivity in the
solid state almost exclusively to the high internal friction. Much work
has been done on metallic conductivity in different directions through
the crystal, but a corresponding study in a thorough and systematic
manner for electrol)rtic conductivity is still lacking, and although pre-
senting many difficulties it promises very interesting results (Tegetmeier,
Wied. Ann., 41, 18 (1890)).

The study of electrolytic dissociation in non-aqueous solutions has
received its most important contributions of late from Walden in a con-
tinuation of the work already reported in last year's review (This
Journal, 28, 905 (1906)). Nine articles in this series have now appeared,
viz: I, General Introduction (P. Walden, Z. physik. Chem,, 46, 103-88
(1903)); 11, Electrical Conductivity (Ibtd,$4j 129-230 (1905)); III, Vis-
cosity and its Relation to Conductivity (/6id., 55, 207-49 (1906)); IV, Mo-
lecular Weight by Boiling-Point Method (Ibid., 55, 281-302 (1906)); V,
Sohibility (Ibid., 55, 683-720 (1906)) ; VI, Heat of Solution (Ibid., 58, 479-
511 (1907)); VII, Heat of Dissociation and Heat of Solution (/6irf., 59,
192-21 1 (1907)); VIII, Relations of Refractive Index and Electrolytic
Dissociation (Ibid., 59, 385-415 (1907)) ; IX, Electrostriction (Ibid., 60, 87-
100 (1907)). One of the most salient features of the work is the sharpness
with which Walden has been able to separate the effects of changes in the
degree of ionization from those due to the friction of the moving ions. The
key to Walden's success is to be found in his systematic plan of campaign.
He first worked over in a qualitative way a very wide range both of solvents
and solutes and from these selected a working list of about fifty typical sol-
vents and some four or five solutes. The strictly quantitative work was then
chiefly confined to these substances and their mixtures. The solutes
were for the most part iodides of organic ammonias, the most complete
series of measurements being carried out upon solutions of tetraethyl
ammonium iodide. The first important outcome of the work, as already
noted in last year's review, was the giving of a quantitative form to the
Nemst-Thomson relation between dissociating power and the dielectric
constant of a solvent, Walden finding that if the same electrolyte were
dissolved in two different solvents to such concentrations that its degree
of dissociation was the same in each, then the volumes of the two solu-
tions, for equal quantity of the solute, were inversely proportional
to the cubes (in last year's review this was incorrectly stated as the
first powers) of the dielectric constants of the solvents. He further
established the fact that at complete dissociation and high dilution the
mokcular conductivity of the same electrolyte in different solutions is
very nearly inversely proportional to the coefl&cient of viscosity of the
pure solvent. Thirdly the heat of dissociation for a given electrolyte
was found to be independent of the solvent. Many more interesting
details are brought out in the papers, but the above are perhaps the most

itriking features on account of their ampUdty and
Tork is being continued and it will certainly take its pi
;lasdcs in this branch of inquiry. Bauer (E- Bauer,
[2, 725-6 (1906)) points out that Walden's results suppi
vhich both he and Malmstrom (R. Malmstrom, Z. Eleki
tog {1905); E. Bauer, Z. Eleklrochem., 11, 936-38 (
aously made upon theoretical grounds, that the ioni
)f a binary electrolyte in two immiscible solvents after
lartition equilibrium between the two solutions should
)f the cube of the dielectric constants of the solvents,
ng also requires that the solution pressure of a metal in
x; proportional to the cube of the dielectric constant
ind consequently, the order of elements in the absolut
ihould be independent of the solvent. Van Laar (J.
>Aj'jtfc.CA«m., 58, 567-74 (1907); 59, 212-17 {1907)) she
'esults corroborate and explain the relation between solu
;iation empirically established by Abegg and Bodlam
iie equation of Luther connecting distribution and disso
The lugh migration velocity of ions common to solute
sxplained by Danneel (see This Journal, 28, 904) 1
itudied by Hantzsch and Caldwell (A. Hantzsch and Ket
Z. physik. Ckem., 58, 575-84 (1907)) using formic and
jyridine as solvents. In conformity with the theory adva
,hey find pyridinium salts in pyridine, formates in formic
n acetic add all show abnormally high migration velo
Another long and praiseworthy campaign which is
brth positive results of great importance is the direct
>smotic pressure undertaken by Morse and Frazer with
itudents (H. N. Morse, J. C. W. Frazer and students, A
io-6: 28, 1-23; 29, 173-4; 32i 93-119; 34, 1-99; 36.
(24-60, 425-67, 558-95; 38, 175-226). After sevi
ievoted to perfecting the Pfeffer cell and auxiliary app;
lique has been reduced to a certainty and a wonderi
iccuracy secured. In the past, the lack of this very tecl
is, perhaps too willingly, to rely entirely upon the in
neasurement by freezing- and boiling-point displacement
md electrochemical measurements, and we have come
act that these, though perhaps more convenient, are
neasures of the quantities sought. For instance in e
[ mg. molecule of solute per liter produces only 0.001
:he freezing-point but would still represent on an oil co
m osmotic pressure of 10 cm. We must look to direct
neasurements to determine the very high molecular W(
This must eventually form another important link in the
.ypical solutions and suspensions referred to in an earlii
Lfticle. In this same connection, reference should here
vork of Berkeley and Hartley (Earl of Berkeley and
^roc. Roy . Soc, London, A, 78, 68; also Pfei/. Tf-afW.,2o6A,
vho have succeeded in carrying the measurements for cai
ip to pressures of 135 atmospheres. Morse and Fra;
lublished results obtained by their methods on a wid

REVIEW. 299

solutions and still more data of a similar character are promised in the
near future. One of the most interesting results of the measurements
thus far published, at least from a theoretical standpoint, is the fact
that it is apparently the volume of the pure solvent rather than that of
the finished solution which must be multiplied by the osmotic pressure
to give a constant; or as they express it, we should use the * 'weight normal"
rather than the "volume normal." The molecular osmotic pressures
which they have measured at 20° are some 2 to 4 per cent, greater than the
corresponding molecular gas pressures, while at o*^ they are from 6 to 1 1 per
cent, higher. The authors carefully refrain, however, from drawing
general conclusions until they shall have extended their measurements
to a wider range of temperatures.

An attempt has been made by van Laar (J. J. van Laar, Proc. Acad.
WeL Amsterdam, 21, 53-63 (1906); Physik,-chem. Centr., 4, 11) to es-
tablish the same ** weight normal" relation from a thermodynamic stand-
point, the theoretical values thus obtained agreeing wefi with Morse
and Frazer's measurements. Another relation pointed out by van Laar
is that we are led to expect that the divergence from the PV = RT law
will be in opposite senses in the gas and osmotic pressures for most sys-
tems. Caldwell (R. J. Caldwell, Proc, Roy. Soc, A 78, 272-95 (1906))
in studying the dBfect of salts and non-electrolytes upon the inversion of
cane sugar by adds also points out how much simpler relations are ob-
tained by referring concentrations to equal quantities of solvent. This
paper, together with one by Armstrong (H. E. Armstrong, Proc, Roy.
Soc.j A 78, 264-71 (1906)) dealing further with the same data, aims to
establish the association of solvent and solute in contra-distinction to
dissociation of the solute as a general explanation for the peculiarities of
electrolytes. Both papers contain much very suggestive matter, how-
ever little one may agree with their main contention. The electrolytic
dissociation theory in fact has been steadily absorbing of late many of
the original concepts of its old opponent, the so-called hydrate theory.
A good illustration of this is to be found in H. C. Jones's work on hydrates
in solution which has recently been collected and published in book form
by the Carnegie Institution of Washington. Another paper touching
the same subject by Bousfield (W. R. Bousfield, Proc. Roy. Soc, A 77,
377 (1906) also Phil. Trans., 206A, 101-59) presents experimental
data in support of his formula (W. R. Bousfield, Z. physik. Chem., 53,
257-313 (1905)) for the diameter of the hydrated ion at different dilu-
tions of the solution and shows how this may be directly connected with
such properties as migration velocity, viscosity, degree of dissociation,
density and refractive index.

The subject of photochemistry has come to stand further and further
apart from the other divisions of physical chemistry chiefly because, in
the latter, it has been the study of reversible processes which have served
to bring out the logical connection between apparently unrelated phenom-
ena, while in photochemistry imtil very recently such processes were
almost unknown. In other words in the great majority of laboratory
reactions brought about by the action of Ught, its function may be com-
pared to that of a catalyzer, in that it removes passive resistances from
reactions already potentially possible rather than making permanent
contribution itself to the free energy of the system. When we stop to

ze that the photochemistry of chlorophyll ac
nic world absolutely depends, represents on thi
rent type of process, in which radiant energy is

chemical energy and stored as such, the uni
tie subject are apparent. The first noteworthy

by Luther, Wiegert and Wilderman was cited
IS Journal, 28, 909 {1906)). Since then a gr
iect has manifested itself. Luther and Wiegert
Ic on anthracene and dianthracene (R. Luthei
nk. Chent., 53, 384-427 (1905)) and Wilderman (
. Soc. London, 206^, 335~40i; also Z. pkysik
-5 (1907)) has also published a second very exte
ts on the action of light on a series of galvanic cell
f, Ann. Physik, 13, 464-76 (1904)) pointed ou
formation of ozone by various fonns of electri'

in the main a photochemical process due to
;t light thus produced. This has since been
jl and quite recently Regener (E. Regener, An
1906)) and R.USS (P. Russ, Z. Etekirochem., 12
m tlwit ultraviolet light of wave length 200-3CK

oxygen, while light of still shorter wave len(
rse reaction. The simplicity of the substance h
I this unique relation to the different portions

reaction an exceptional importance. In this
d in passing, that Ladenburg and Lehmann (
aoaxm, Ann. Physik, ai, 305-18 (1906)) ha
lin tran^tory bands in the absorption sped
:h they attribute to a still more unstable alio
jen. Among the most suggestive of recent ci
iFCtical standpoint are two papers by Trautz
!., 4, 160-72, 351-59 (1906); also Z. Elektroc
hich he points out the close relation between re
inescence and those whose velocity is affected 1
MDpIete thermodynamic equilibrium must be ra

"black body." If the system has not reache
1, Trautz believes it highly probable that t
tion will not be that of the ideal black bod;
1 the black body spectrum for the temperatu
es sufBdently great for us to detect, we have e
tion or a case of chemical luminescence depe:
fhich the equilibrium between mdiation and al
3 further pointed out under what restrictions
pplied to photochemical reactions and what tyj
' be expected, experimental evidence for som
ented. The small temperature coefficient of tl
tochemical reaction has often been pointed out, an
;e it probable that this quantity is larger the furthe
spectrum the re^on of photochemical sensitiv
tal confirmation of the view, new data on the
>gallol solution are presented. In the dark this i
inescence, and it is powerfully accelerated by re(

all, by violet. Its temperature coefficient is al

REVIEW. 301

croft (W. D. Bancroft, /. Physic, Chem., 10, 721-28 (1906)) has lately
attacked the general problem of photochemistry from still another aspect,
pointing out how by treating each active color of light as a new variable
comparable with temperature and pressure, the phase rule may be gen-
eralized to include photochemical systems even though these merely
represent stationary states and not true thermodynamic equilibria. In
ilhistration and confirmation of this treatment one of his students (G. A.
Rankin, /. Physic, Chem,, 11, 1-8 (1907)) has studied the equilibrium
of rhombic and amorphous sulphur in carbon bisulphide under the in-
fluence of light of varying intensity. Another reversible photochemical
reaction has lately been reported by Dewar and Jones (J. Dewar and O.
H. Jones, Proc, Roy, Soc. Lond,, 79A, 66-80 (1907)) in the case of iron
tetra- and pentacarbonyls in presence of carbon monoxide. It appears,
on the whole, to resemble closely the case of anthracene and dianthracene
studied by Luther and Wiegert. In closing, we must not omit mention
of the extremely interesting work of Usher and Priestley (F. L. Usher
and J. H. Priestley, Proc, Roy, Soc, Lond,, 77-B, 369 (1906), and 78B, 369
(1906)) on the primary reactions of the chlorophyll assimilation. They
have apparently been able through very ingenious technique to experi-
mentally realize outside of the plant the much looked -for lower aldehyde
stage of the process. The importance of this question and the experi-
mental difficulties are both so great that independent confirmation and
extension of the experiments are much to be desired. The purely cata-
lytic action of light has also been studied by itself of late by Wiegert (F.
Wiegert, Ann. Physik, 24, 55-67 and 243-66 (1907)), both from the theo-
retical and from the practical side. The reactions between chlorine, car-
bon monoxide and phosgene fall in this category. On the whole, it
seems not too much to hope that we are rapidly approaching an epoch in
photochemistry similar to that which electrochemistry passed through
some twenty or thirty years ago when it was first really made an integral
part of general physical chemistry. May it not be that in the process of
this development, photochemistry, like electrochemistry, will furnish us
with a new wealth of experimental and theoretical methods for attacking
the old problems of general chemistry?

When we come to consider the results of the mere application of ph)rsico-
chemical principles and methods to other branches of chemistry we meet
a mass of material which, to do justice to, would carry us far beyond the
scope of the present review, but it may still not be out of place to note a
few suggestive cases, although such a selection must of necessity be
rather an arbitrary one. Toward the problem of protein synthesis Taylor
(A. E. Taylor, /. Biol, Chem,, 3, 87-94 (1907)) and Robertson (T. B.
Robertson, Ibid., 3, 95-9 (1907)) have each contributed an exam-
ple of the reversal of the process of protein digestion, the former
working with protamine sulphate and trypsin from the Uver of the clam,
and the latter with certain stages in the digestion of casein by pepsin.
It is instructive to note that both authors were led to their respective
methods by purely physicochemical lines of reasoning. The first paper
also contains a good set of references to earlier work on reversions of
physiological interest. Still another instance of the reduction of a bio-
chemical process to physicochemical control is offered by the work of
Buchner, Meisenheimer and Shade (E. Buchner, J. Meisenheimer and H.

302 REVIEW.

Shade, Ber., 39, 4217-31 (1906); also H. Shade, Z. ]
1-46, 60, 110 (1907)) who have studied the ferment;
colloidal platinum solutions and shown that the progre
is entirely comparable to that of the ordinary yeast fen
the dynamical treatment of the process they are furt
out the probable type of the intennediate stages of t
the work on establishment of a convenient absolute sc
alkalinity by the use of indicators standardized by
ments, which was begun by Friedenthal (Friedenthal, Z
114-9 (1904)) and Salm {E. Salra, Ibid., 10, 34
99-101 (1906)) has been added a valuable set of d;
Salm, Z. physik. Chetn., 57, 471-501 (1906)) covering
indicators. A much needed thorough systematization
of the constitution of alloys which was commenced 1
Tammann, Z, anorg. Ckem., 37, 303-13. The subsi
this series have also appeared from time to time in t
five years ago, now comprises over fifty papers by hims
students, covering a large number of the important bi
similar systematic work is just begiiming to be und
in ternary systems (R, Sohmen and A. v. Vegesack, Z.
257-83; 60, 507-9; also Janecke, Ibid., 59, 697-71
(1907)). Of course the systems having iron as one coi
some time past attracted much attention, but even he
workers have, for the most part, centered their activi
of special problems. A movement toward a broader
whole subject seems, however, to have set in and is bein
supported by several of the leading institutions and so
Closely related to these metallurgical problems are the ge
of rock formation, and here may be noted the work ol
Laboratory of the Carnegie Institution of Washingtoi
results have already appeared in This Journal (A, L
Shepherd, This Journal, 28, 1089-111-] (1906)). In this
ena of supersaturation, superfusion and vitrification gr
the problems. These latter phenomena in themselves
receive more direct investigation. Young and Burke {{
W. E. Burke, This Journal, 28, 315-47 ; 39, 2
have reviewed the older work and pointed out som
Miers and Isaac (H. A. Miers and Miss F. Isaac, Proc.
A 79, 322-50 (1907}) working with mixtures of salol
determined the condition for various labile and metast
tween a pair of components, both of which show sti
They also discuss briefly the bearing of the work on
geological problems. In this connection for sake of comp
may be made of the older work of Guertler on fusion:
(W. Guertler, Z. anorg. Ckem., 40, 225-53 (1904)), the
of devitrification (W. Guertler, Ibid., 40, 208-79 C1904))
points of the alkaline earth borates (W. Guertler, Ibid.,
that of Zschimmer's paper on the properties of various gl
raer, Z. Eleklrockem., 11, 629-38 (1905)) and Doelter'sUl
genesis" (Die Wissensch^t ^mmlung Hft. 13, Braun
(1906)).

NEW BOOKS. 303

CORRECTION.

The Relative Solubility of the Silver Halides and Silver Sidpho*
cyanate. — In Table I of this paper (Tms Journai,, 30, p. 72) the ratio of
the sohibilit}' of silver chloride to silver sulphocyanate appears in an
inverted form. The mean ratio calculated from the table should read

SAgCNS

SAga

0.0748. ArTHUK H. HiLh.

NEW BOOKS.

A Text-Book of Electro-chemistry. By Max Le Blanc. Translated from the
Fourth Enlarged ^German Edition by Wnjjs R. WmTNBv and John W. Brown.
New York: The Macmillan Company. Prioe, ^2.60 net.

Since the first edition of Le Blanc's treatise appeared in 1895 it has
been accepted the world over aS the standard text-book of electrochem-
istry. In the succeeding editions the author has aimed to keep pace
with the rapidly growing science. It is doubtful whether the resulting
growth of the volume to more than double the original size has increased
the value of the book as a text-book, but it has provided a remarkably
bandy and comprehensive compendium of electrochemical knowledge.

The style remains the style of a text-book, and the subject will seem
easier to the reader than it is in reality. Weak points in the theory
are slurred over. The recent critique of Jahn, the question as to the
correctness of the ionization values calculated from the conductivity,
tbe enormous deviation of strong electrolytes from the mass law, re-
ceive but scant attention. It has been the misfortune of the ionic theory
that its advocates have seldom been satisfied with pointing out its un-
qtiestionabk triumphs, but have claimed for it a perfection which it
has not yet attained.

The author has succeeded to an extraordinary degree in bringing his
work up to date. Important investigations which appeared even up
to within a few months of the date of publication of the book are men-
tioned and frequently their results are incorporated in the text

Occasional misstatements occur, as the one on page 181 that "the
'relations' between the solution pressures of various metals are inde-
pendent of the nature of the solvent, and, moreover, always possess
the same value. " As a rule, however, the statements are accurate and
reliable. This, unfortunately, is not true of the last chapter, in which
the so-called decomposition potentials are discussed. The greater part
of this chapter is devoted to an attempted explanation of phenomena
which have been shown to be as purely subjective as the N-rays of Blond-
lot

The English edition is rather more than a translation. Bxplanatory
and supplementary paragraphs have been added, many new illustra-

304 NEW BOOKS.

lijns have been introduced, and the translators have used tbrougfa
out a new system of notation, devised more methodically than any doh
in use. They have done an important service in thus calling atteatioi
to the need of a rational notation in electrochemistry, whether or not
the special system which they propose be ultimately adopted.

Gilbert N. Lewis.

J. G Genfla's Leiubncli d«r Fulwiifabritaition, revised and enlarged. By Dr. Buki
ROCK, Erster Band, Die Erdfarben. Braunschweig: Priedricb Viewig nnd Sola
1906. Pric«, 6 marks.

The preparation of this book was undertaken by Dr. Buntrock at tb
solicitation of the publishers, Friedr. Viewig und Sohn, and the manuscrip
of this first part, covering the mineral colors, was delivered to them
according to the preface, by the author at the close of 1904, and accordin;
to the inscription of the publisher was published September i, 190*
In view of the very rapid progress being made in the industries, mud
of the apparatus and some of the methods as well as the use of some c
the products described have become obsolete, and the book will there
fore be found by no means up to date by those engaged in the industrie;
and hardly a safe guide to those seeking direction in them.

This first volume is divided into five parts treating principally of tb
mechanical treatment of crude mineral colors to prepare them for use i
the arts. Chapter V treats of the chemical changes which the earth
colors undergo under various conditions. The other chapters trea
respectively of the mechanical means for preparation of these colors, a
follows: Elutriation and elutriating apparatus, drying and drying aj
paratus, milling and mills, bolting and mixing, and the apparatus tteede
therefor. The apparatus described is such as is usually employed in tl
industries for the operations mentioned, and it must be said that tb
forms of construction described and recommended are rather crude a
indicated in the illustrations presented. Many of the illustrations (
apparatus are to be found in better shape in the catalogues of the mam
facturers named in the book, yet for those de^ring knowledge regardio
the sources from which the apparatus and machines can be obtainet
the book is by no means a bad guide. Yet this much may be said al:
of the advertising pages of the current technical chemical journals, wbi
the information offered regarding the mineral colors may be found quit
as fully developed and in as reliable a form in most of the encyclopedia
and works already at hand in our libraries, public and private.

The part now published closes with extensive advertisements of makei
of the machinery described in the text, preceded by an index of thes
advertisements arranged alphabetically. The t>ook is by no means u
to the high standard of the usual publications of Viewig und Sohn.

Wm. McMuktsib.

NEW BOOKS. 305

Der HthrttngBmittelchemiker als Sachventttndiger, Anleitung zur Begutacfaung
der Nahningsmittel, Genuszxnittel trnd Gebrauchsgegenstfinde nach den gesetz-
lichen Bestimmttngen mit praktischen Beispielen von Propbssor Dr. C. A. Nbu-
PSU>, Oberinspektor der Kgl. Untersuchtmgsanstalt ffir Nahnmgs- und Gentisz-
mittel zu Munchen. Berlin: Verlag von Julius Springer. 1907. xix+477 pp.
Price, unbound, Mark 10; bound, Mark 11.50.

The primary object of this book is to guide the food chemist in inter-
preting his analyses and in making suitable recommendations to the court.
It is also designed to aid physicians, court officers and other officials
who may be called upon to deal with food problems in a legal way. De-
scriptions of methods are not included.

The book contains a vast store of information touching the source
and preparation of foods — ^their composition, adulteration, imitation
and deterioration — and also contains numerous examples illustrating
the practical application of this knowledge.

The General Part of the work consists of an introduction with remarks
on the purposes of food control, methods of analysis, and standards,
chapters on official reports and decisions, and one chapter on the general
composition of foods.

The Special Part is divided into twenty-one chapters on the different
classes of vegetable and animal foods and a final chapter on toys, cook-
ing utensils, paints, cosmetics and petroletun products.

As a rule each food or class of foods is considered under the following
heads: i. Defiinition, Origin, Methods of Preparation and Preser-
vation, Composition, and Standards, including the United States Stand-
ards. 2. Forms of Adulteration. 3. Imitations. 4. Forms of Dam-
age. 5. Characters Injurious to Health due to Damage, Adulteration,
etc 6. Practical Examples with Analytical Data; Interpretation of
Results; Decisions under German Law.

Although the data is in the larger part German, the book will prove
a valuable addition to the library of the American food chemist and
should also be within the reach of others interested in the manufacttire
and inspection of foods. A. L. Winton.

Chtpten on Paper Making. By Ci«ayton Bbadls. VoL I., 1904, Vol. II., 1906.,
H. H. G. Grattan, London; Vol. III., Vol. IV., 1907, Crosby Lockwood & Son,
London. Price, $2.00 per vol.

These four small volumes constitute an important contribution to
the chemical technology of paper-making. As a necessary consequence
of the method by which the subject matter is developed, the treat-
ment is fragmentary and disjointed, but in spite of this regrettable
lack of sequence the student of paper-making will find in these ''chapters"
a great amount of useful information and suggestive discussion which is
not to be found elsewhere.

VoL I comprises ten lectures delivered before the Battersea Polytech-

306 NEW BOOKS.

nic Institute and dealing with various special phases of
of paper-making and the chemical and physical properties <

Vol. II opens with a thoughtful discussion of Techn
as Applied to Paper-making, while the body of the volui
of the carefully considered answers made by the author t<
propounded by the City and Guilds of London Institute
who had taken its Course in Paper Manufacture. As tc
themselves it may be said that very few American papei
tendents could hope to pass them with credit. Mr. Bt
cover in what is usually an eminently practical way ma
which paper-makers generally have only the most empir
and as to which it is only fair to say they have few sour
tion.

Volumes III and IV are the outcome of the publicatioi
Pulp of a series of test questions on paper- making technc
to workers in English mills. These questions developer
ingly frank and intelligent discussion of the subjects pi
in these last two volumes Mr. Beadle has brought tog<
of the best answers to each question and has extended a
whole by critical discussion and comments of his own.

In spite of the serious handicap which the method of
imposed on the author, these volumes will well repay ca
all who have to do vnth the art of paper-making. A

The Prindplei of Copper Smaltliig, By Erward Dyer Pstb
Metallurgy, Harvard University. Hill Publishing Co., New 1
Price, $5.oo.
Whoever will carefully read through every page of th
pages will undoubtedly have become acquainted with
upon which the modem practice of copper smelting is bas
hardly, however, agree with the statement that it repre;
down of principles; rather would he be hkely to conside
planation of these principles could have been made con:
concise. This extended style is no doubt admirable in a
but makes continuous reading rather wearisome. It is
a most instructive book. The facts are clearly stated, a
gone into thoroughly, examples and problems being cor
to elucidate the various points, and the influence of busi:
tions is kept constantly in view. The work is divided int
ters, embracing: Methods and Collectors; First Principle
Principles of Roasting; Chemistry of Smelting; Practict
Blast Furnace Smelting; Keverberatory Smelting; Py
Practical Study of Slags; Matte; Production of Metallic
Matte; Refining of Copper; Principles of Furnace Buil

RBCBNT PUBLICATIONS. 307

tions of Thermochemistry; Miscellaneous and Commercial. The chapter
on blast furnace smelting contains a very interesting description of the
development of the gigantic furnaces constructed by E. P. Matthewson
at Washoe, Anaconda, Montana, smelting 3,000 tons of charge per 24
hours. The chapters dealing with pyrite smelting are largely written
and are reviewed by R. Sticht, of the Mt. Lyell mine in Tasmania.
Throughout the parts dealing with Bessemerizing methods, it is disap-
pomting to find that no credit is given to the pioneer work of the recently
deceased John HoUoway, of London (in whose laboratory the reviewer
found his first' occupation), who clearly foresaw, and endeavored to put
in practice the principles of these methods, and from the adoption of
which others have received the rewards. The chapter on Thermochem-
istry is written by Prof. Joseph W. Richards, of Lehigh University. In
the portion dealing with the refining of copper, but passing mention
is made of the electrolytic methods, as they are of very special nature,
and the art has a Hterature, though a meagre one, of its own. While
the book is primarily intended for students, and for those who have
not an exact knowledge of chemistry, it should be found useful to all
who have interests in the mining, refining, or chemistry of copper.

T. Lynton Briggs.

RECENT PUBLICATIONS.

Bausr, Hugo. History op Chsihstry. New York, Longmans Green & Co.

1907. $1.

BoTTLBR, M. Harze u. Harzindustrib Hanover: 1907. 291 ss. M. 4,40.

DucHBMiN, R. La D6natu ration db l'Alcool en France et dans les prin-
dpaux pays d'Europe. Paris: 1907. 8vo. 280 p. M. 6,80.

Dunbar. Lbitpadbn pOr dis Abwassbrrbinigungsfragb. Miinchen: 1907.
386 ss. mit 150 Fig. gr. 8. M. 9.

Fbrchland, p. Dm Englischbn bi^bktrochbmischsn Patbntb. Ausziige
ans den Patentscbriften mit Sadti- tmd Namenregister. Band I: Elektrolyse. Halle:
"W. gr. 8. M. 9.

Fischer, E. Untbrsuchunqbn in dbr Puringruppb (i 882-1 906). Berlin:
"907. gr. 8. 608 ss. M. 15.

Gill, A. H. Enginb-room Chemistry. New York: Hill Publishing Co. 1907.

198 pp. $1.

GoDCHOT, M. Contribution A l'^tudb dbs Hydrurbs d' Anthracenes et
DB LBDRS D^Rivils. Paris: 1907. 8vo. 71 p. M. 4,80.

Gottschau*, M. Lbitpadbn dbr Chemie nach dem Arbbitsprinzip. (2 Teile).
Tdl I: Metalloide. Mfinchen: 1907. gr. 8. 112 ss. mitt 77 Fig. M. 1,20.

Hbiocblmayr, F. v. und Brunner, K. Lehrbuoh dbr Chemie und Miner-
ALOGI8. 3. Auflage. Leipzig: 1907. gr. 8. 184 ss. M. 2,60.

JoNBS, H. C. Elements op Physical Chemistry. 3RD Edition revised and
enlarged. New York: Tbe Macmillan Co. 1907. 8vo. $4.

Kahlsnberg, L. Laboratory Exercises in General Chemistry. Madison,
Wis. 1907. i2ino. 204 p. $.80.

3o8 RBCEKT PUBUCATIONS.

SK MlWHWIBTaCHAFT AUP WISSSNSCBAFTUCKU

neubearbeitete Auflage. BeTlin: 1907. p. t.
701 8S. M. 13.

Lassak-Cohh. Arbbitsmstboden fCk obcamiscii-cbbmischb Laboratoium.
h tmigearbeitete u. vermehrte Auflage. Teil II: Spezieller Tdl. a. HUfte. Htin-
buTg: 1907. gr. 8. ss. 865-1496. H. 20. Das jetzt voUstandige Wnt; 1 Ttilc,
I9c»fr-i907- 364 u- '509 S8. mil 170 fig. M. 58.

LovBBDO, J. DE. Conservation pas ls pkoip des DsuutEs AuMENXAiut.
E>aris-. 1907. Svo. 180 p. H. i.ao.

Haulock, L. Stbbsocbshis. Lchre von dn raumlichen Aaordnung dn
A.tome im MolekOl. Leipzig; 1907. gr. 8. 15* s». mil 58 Fig. M. 5.

UarC, R. Die PRVSlKAUSCH-CHBllISCaBN EusBNSCHAnSN DBS MBTALUSCBU

Sbusns. Hamburg; 1907. gr. 8. M. 4.

MBMtLEN, F. Die Kbenssiten, ihkb Zubahmbhsbtxuno ukd Fabbikatiok
tou Standponktb deb PHVSiKAuacHEN Chbioe. Aim dcin FnnifiKchen ftbti-
Ktzt von F. Goldschmidt. Halle: 1907. gr. 8. H. 5.

Ueemet, M. a. Manipulations et sxsecices ds Cbdoe. L« petit labon-
toite; le materiel nmpli££. Paiia; 1907. 730 pg, avec 378 figures. If. 7.

PAcBBux, H., Ln^NON, A., bt Blanc, L. Lbs Pkoddits CHiutQVB. Pirii:
1907. laiuo, 384 p. M. 5,

Pbbt, Bert W. Labobatori; Experimbnts in Cheuistrv. 3rd Ed, leviMd.
\tm Arbor, Mich.: George WahT, 1907. 139 p. $.60,

PlaTTner, C. F. Probibbkunst hit deu LATrohbb. VollxUudige Anla-
tung zu quolitativen und quautitativen Lfltrohnmtersudiungeti. 7 Auflage, btai-
bcitet von F. Kolbeck. Leipiig: 1907, gr. 8. 531 ». mit 7« Kg. M. 11,

PoTONid, H. Die Entstbruno dbr Stbinkohlb v. vBRWANiMnu Biliio(ce>
IiNSCHLiSssLicH DBS Pbtrolbuhs. 4, verbesserte und erwriteite Auflage, Beifin
1907. gr. 8. 47 ss. mit 38 Fig. H. 4.

Pkics, V. Slater, and Twiss, Douglas F. A Course op Practical Oboane
Drbhistky, New York; Longman*, Green & Co. 1907. 339 p. fi.ao.

ScHWALBE, C. G. NbcERE FXrbsthsoribn. Stuttgart: 1907. gr S. 131
K. M. 3,60.

Thaussino, J. E. TsraoBiE und Praxis dxk Halzbbrbitdno uHd Bmru'
UKATioN. 6, unter Hitwerkung von G. Holzner, A. Jflrgenaen, T. Langer und E
9truve neubearbdtete Auflage. 3 Teile Ldpdg; 1907. Lex. 8. mh 3 Toftln
]33 I^guren und Atlas von 34, meist Kolorierten Tefeln, H. 43.

WiNDiscH, W. Anlbituno iur Untsrsucruno des Ualzbs auf Eztraktgb
BALT sown ADP SEINE AusBBUTE IN DER PRAXIS. 4. Deubeaihcitctc Aufltge. Btt
in: 1907 gr. 8. M. 3,50.

Vou XXX.

March, 1908.

No. 3.

THE JOURNAL

OF THE

American Chemical Society

FIFTEEnTH ANinJAL REPORT OF THE COUMITTEE ON ATOMIC
WEIGHTS. DETERMINATIONS PUBLISHED DURING 1907.

Bt p. W. Cla&kb.
Received January 24, 1908.

Although the number of new atomic weight determinations published
during the year 1907 has not been large, the quality of the work done
was remarkably high. Some of the determinations are of fundamental
importance, as, for example, those of hydrogen, nitrogen, sulphur, potas-
sium and lead. Sulphur and lead are especially noteworthy, on accotmt
of their divergence from the older measurements. The data now avail-
able are as follows :

Hydrogen.

Noyes^ has made complete syntheses of water in five series of deter-
minations. The first series was defective, and is therefore not published
by the author. For the other series very complete details are given,
which would take too much space for repetition here. Noyes's corrected
and final data are as follows:

SSCOND SBRISS, HyDROGBN FROM SULFHURIC AciD, WSIGHSD TwiCB.

Weight 0.

Weight HaO.

At.

wt. H.

Weight H.

H:0.

H : H.O.

3- 72565

29.57891

33.30408

1.00765

1.00767

380318

30.18400

33.98748

z. 00800

1.00799

3.75873

29.83358

33.59127

1.00792

1.00795

2.96328

23.51987

26.48379

z. 00792

1.00790

2.II395

18.89214

1.00795

3- 53136

28.02910

31.56024

I.0079Z

1.00792

3.53959

28.09619

31.63554

Mean,

1.00785

1.00786

1.00787

1.00789

* This Journal, 29, 1718.

3IO

Thud Seribs, Hydrogen f

F. W. CLARKE-

A SvLrauuc Acid, Ozidizbd b

weight H.

Weight 0.

Weighi H,0.

H:0.

H:H,0.

3-44379

19- 39757

21. 84043

I .00746

1,00746

3.18739

17-36305

19.55117

1.00784

I.OO7S0

a.75ia9

21.84345

24.59389

1.00764

1.00768

4.00063

35.75073

1.00803

4.04057

33.07689

36,11763

1.00773

1.0077s

Mean

I .00767

1,00774

TH SBKIES,

Hydrogbn and

Oxygen from

SUI-PBURIC

Acid

wt. H

COMBINBD

27916

12734
17556
19346
30746

08455
76537
13449
27384
30863
48543

79354

Weight H]0.
10,36138

36.S9043
37,30787
37,46453
30.61357

41-43905
40,85834

t. 00823
1.00774

t.oo8tS
1.00823

H:H|0.
1,00830

00831
00839
00797

Wdgbl H.

4,61180
4,63358
4 59853
4 55832
4.20399

Weight o.
36,60909
36,69575
36,50484
36-17887
33.37000

H:0.

H : H,0,

1.00779

I ,00798

1,00806

1.00776

1.00780

1 ,00795

1.00790

1,00783

1,00786

Mean, 1.00786 1,00788

The mean of all four series, assigned equal weight, is H = 1.00789.
The mean of the 48 determinations, as a single series, is H = 1.00793
The series, however, are not of equal weight, the second and fourll
being better than the others. According to Noyes, the value H =
1.00787 is the most probable value to be derived from his determinations
Morley's classical syntheses of water, recalculated, give H = 1,00763
Combining this with Noyes's results, the average is 1.C0775, which is per
haps better than the determinations of either chemist taken alone.
Silver-ITitrogen.
Richards and Forbes,^ in a most careful investigation, have effectei
the synthesis of silver nitrate from pure silver, and so measured tin
' This Journal, 19, 808.

REPORT OI^ COMMITTEB ON ATOMIC WBIGHTS. 3X1

ratios between Ag and AgNOj. The vacuum weights are given below,
together with the ratio Ag: AgNO,: : loo: re:

Weight Ag. Weight AgNO«. Ratio.

6.14837 9.68249 157.481

4.60825 7.25706 157.480

4.97925 7.84131 157.480

9.071OI 14.28503 157.480

9- 13702 14.38903 157.481

9.01782 14.20123 157.480

Mean, 157.480

A small correction for a trace of water retained by the nitrate reduces
this figure to 157.479, which is identical with that obtained by combining
all the older data. If now, Ag = 107.930, N = 14.037; but if the recent
value for N, 14.008, is correct, then Ag becomes 107.880.

Potassium.

Richards and Mueller* have redetermined the atomic weight of potas-
sium by analyses of the bromide. First, the ratio between silver bro-
mide and the potassium salt was measured, with the subjoined results:

Weight KBr. Weight AgBr. At. wt K.

2.19027 3 45617 39- "4

4.19705 6.62285 39.113

2.06723 3.26206 39.112

2.58494 4.07889 39- "5

Mean, 39.1135

Second, the ratio between potassium , bromide and metallic silver was
determined by the usual titration method, as follows:

Weight KBr. Weight Ag. At. wt. K.

4.33730 3.93164 39.113

4.18763 3.79587 39-115

4.15849 3.76943 39- "6

367867 3.33450 39.116

3.60484 3.26776 39. "O

4.78120 4.33387 39. "8

5.67997 5.14860 39.116

6.41587 5.81571 39- "5

2.88134 2.61184 37. "3

3.64383 3.30309 39. 1"

3- 12757 2.83504 39.113

Mean, 39. "43

The value for K was computed with Ag = 107.93 ^^^ Br = 79.953»
All weights were reduced to a vacuum standard. The final value, ob-

* This Journal, 29, 639. The paper is preceded by that of Richards and Staehler
on potassium chloride, which was noticed in the report of this committee for 1906.
It originally appeared in the BerichU.

p. W. CLARKE.

tained by combining these results with those of Richards and Staehler,
is K = 39-1139. Clarke, in his "Recalculation of the Atomic Weights,"
1897, from a combination of all the older data, found K = 39.112. Tk
two values are nearly identicat

Sulphur.
The atomic weight of sulphur has been redetermined by Richards and
Joaes,* who employed an entirely new method. Silver sulphate was
transformed into silver chloride by heating in gaseous hydrochloric add.
The weights, reduced to a vacuum, are subjoined. In the third column
the chloride formed from 100 parts of sulphate is given:

Weight &s,SO(. Wtighl AgCI. Ratio.

5-37924
5 08853
5.36381
5.16313
S- 08383
5 1337*
5-16148

s- 19919

5.37436

From this ratio, if Ag =

If Ag = 107.88 and CI = 35.457, then S — 32.069.

Incidentally, the authors discuss the older values for sulphur, and espe-
cially those derived from gaseous densities. The latter they regard as
subject to serious errors.

Lead.

In a preliminary paper, Baxter and Wilson' give the results of theii
analyses of lead chloride. The data, with vacuum weights, arc as fol-
lows:

79»S9
85330

91 -934
9i-93»

67810

91.934

93118

9>-934

74668

- 9' -934

67374

- 9»-933

71946

- 91.931

74490

- 91.939

77993

91.936

940S8

91-934

Mean

91-933

■n*^

and a -

= 35-473

S =

32.113-

eight PbCU

Weight Ag.

Weight AgCl.

Ag. lalio.

AgCl ratio.

4-67691

3.61987

4 -83173

207-179

207. iSE

3-67705

3.85375

307.189

4.14110

3.31408

37016

307-173

207

193

4.56988

3.54673

207.185

5-13387

3.97568

28272

207.201

307

181

3-85844

3.99456

97949

207.186

207

.36

4.67344

3.63638

307.189

3-10317

3-40837

19909

207.188

207

261

4-3013

3-33407

43983

307.303

207

104

' This Journal, 19, 836.

' Proc. Amcr. Acad,, 43, 365.

REPORT OP COMMITTEE ON ATOMIC WEIGHTS. 313

The mean of both series is Pb = 207.190,^ when Ag = 107.93 and CI =*
35.473. This value is much higher than the usually accepted 206.9.

Palladium.

The determinations of Woemle' were based upon analyses of palla-
dosamine chloride, Pd(NH3Cl)2. Two reductions in hydrogen gave the

At. wt. Pd.

106.68
106.70

Five electrolytic determinations yielded the subjoined figures:

Weight chloride. Weight Pd. At. wt. Pd.

1.02683 0.51749 106.71

1.22435 0.61708 106.72

1.46735 0.73944 106.69

0.59796 0.30139 106.73

2.64584 I 33329 106.69

following results:

Weight chloride.

Weight Pd.

2.94682

I . 48493

I. 83140

0.92296

Mean, 106.708

All weights were reduced into vacuum. The antecedent values for H,
N, and CI are not stated.

Nickel and Cobalt.

Barkla and Sadler,* studying the absorbability of secondary Rontgen
radiations from various metals, which is an atomic function, find regu-
larities to which nickel is an exception. If, however, nickel be given a
higher atomic weight than that now assigned to it, the anomalies disap-
pear. Ten measurements of absorbability, compared with the absorba-
bility of rays from other metals, gave, by interpolation, values for Ni
between 61.15 ^^^^l 61.6. These figures can hardly be assigned much
weight in comparison with the excellent and more direct chemical deter-
minations.

Parker and Sexton,* in a brief note, aimounce that 15 electrolytic com-
parisons of cobalt with silver give, in mean, Co = 57.7. The details
of this investigation are yet to appear.

Indium.

Mathers,* in determining the atomic weight of indium, employed two
methods. First, the ratio between InClj and 3AgCl was measured gravi-
metrically, with the following results:

* With Ag = 107.88 this becomes 207.090. See this Journal, 30, 194. — Editor.
' Sitzungsb. phys. med. Soz. Erlangen, 38, 296.

' Phil. Mag. [6], 14^ 408. Prdiminary notice in Nature, Feb. 14, 1907. In Nature
for April 4, Hackett questions the conclusions stated by Barkla.

* Nature, Aug. i, 1907.

* This Jotunal, 29, 485. *

htlnClt.

1156

9S9«>

98175

54540

46361

0860Z

Welsht AgCl

4.11431

64176

85"5

77904

84557

94054

P. W. CLARKE.

At.Wt.lB.

114.80
114. »5

114.9s
114.90
114.86
114.96

Mean, 1 14. S8
Secondly, similar analyses were made of indium tribromide, as follows:

Weight InBr,. Wdgbt AsBr. AI.wt.lB.

3-73494 434550 114.89

7.69880 la. 23341 114.86

6.17450 9-96917 114-89

5.3664a 8.52741 114.85

5.16113 8.20128 II4>8S

4.98336 7.92009 114.81

Mean, 114.86
The calculations were made with Ag = 107.93, CI = 35-4-73, and Br =
9.953. The author favors In — 114.9 as the value to be accepted.

Tellurium.
The research upon tellurium, by Baker and Bennett,* was primarily
o determine the homogeneity of the element. Various fractionating
irocesses were employed, but tellurium of the same atomic weight was
iroduced in every case. Two methods, both new, were adopted for the
,tomic weight determinations. First, TeO, was heated with sulphur in
ubes of glass, the two ends of the tube being packed with pure silver leaf.
Sulphur dioxide was expelled, and from its amotmt, as measured by the
Dss of weight, the percentage of oxygen in the TeO, was computed,
^or this purpose the value S = 32.06 was assumed. The determinations
ly this method fall into three principal series, as follows, representing
lifferences in the source of the initial substance: i. Fractional crystal-
ization of telluric add from barium tellurate. 2. Fractional crystalUza-
ion of telluric acid from oxidation of the element. 3. Tellurium dioxide
trepared from tellurium hydride. Vacuum weights are given through-
lut. The data are as follows:

Shribs 1.

' J. Chem. Soc., gi, 1849.

weight TeOfc

Louao^

I-51509

0.60838

1.09875

0.44074

1-02150

0.40993

0.90835

0.36472

1.00702

0.40451

1.01515

0.40733

KEFOKT 0» COMHITTSS ON ATOMIC WBIGHTS.

315

SBRISS 2.

Prtction.

Weight TeOf.

Loss SOa.

Per cent. O in TeOa.

1.56837

0.62938

20.046

X. 07852

0.43257

20.035

I . 72627

0.69296

20.052

2.09253

0.83927

20.032

0.83335

• 0.33465

20.059

I. 15372

0.46284

20.04X

1.686x8

O.6766X

20.045

0.90835

0.36472

20.053

Now

Sbribs 3.

1. 022 1 7

0.41050

20.064

0.80697

0.32322

20.051

I . 32003

0.52992

20.053

I . 05207

0.42221

20.047

I . 37043

0.54969

20.032

0.95944

0.385II

20.048

Several other exf)eriments, concordant with these, are cited, but with-
out the detailed weighings.

The other method employed consisted in converting tellurium into
the tetrabromide by direct imion with bromine in an atmosphere of nitro-
gen. Here again, several series of determinations are given. For brevity,
the data are combined in one table :

Series.

Weight Te.

Weight TeBr4.

Per cent. Te

0.6x273

2.14933

28.508

0.56866

I . 99354

28.525

0.59884

2.09951

28.523

0.57894

2.03040

28.5x4

0.54743

X. 9x899

28.527

0. 33859

I. 18732

28.517

0.56866

I . 99354

28.526

0.47643

X . 67025

28.525

0.56622

I . 98597

28.511

0.44271

I ■ 55205

28.524

O.41671

I. 46177

28,508

0.506x1

I . 77489

28.515

0.37382

X.3X08X

28.519

0.31895

X. 1x868

28.512

O.4893X

I 71554

28.522

0.47x56

1.65404

28.510

0.40748

X . 42867

28.523

0.62013

2.17449

a8.5i8

0.37382

X. 3x081

28.519

0.50822

X . 78207

28.5x8

O.X2928

0.45354

28.505

0.42926

I . 50540

28.515

0.80348

2.8x7x5

28.5XX

0.95309

3.34193

28.5x2

316 S. W. CLARKE.

From the dioxide determinations the authors compute that Te •<
127.609. From the bromide syntheses, if Br = 79.96, Te ■• 127.601.
They also give, but without details, several determinations based upon
analyses of tellurium tetrachloride. Four fractions of the chloride gave
for Te the values 127.58, 127.60, r27.64, 127.62, The commonly accepted
value, 127.6, thus receives strong confirmation.

On the other hand, quite different results have been announced by
Marckwald.' He prepared pure telluric acid, H^TeO,, which, by heating
imder proper precautions, was reduced to TeO,. His figures, on the bads
of O = 16 and H = 1.008, are subjoined.

Weight acid. Weight TeOr At. wt. Te.

8.6a77 S-9^4 126.93

12.36S0 8-5135 136.84

13.0051 9.0344 136.80

8.6415 3-9947 ia6.«s

8.45SS 5.S696 IS6.80

8.0113 5.5599 126.94.

Mean, excluding No. 4, 13G.86
From the sums of the weights Marckwald, rejecting the fourth deter-
mination, computes Te = 126.85. This falls below the atomic weight
of iodine, and is in harmony with the periodic law. Since, however
it diverges so widely from many, concordant, higher determinations,
it evidently needs corroboration by other experimenters and other methods
Ytterbium and Lutecium.
According to Urbain,* ytterbium is a mixture of two elements, neo^
ytterbium and lutecium. The atomic weight of neo-ytterbium is not
far from 170; that of lutecium is perhaps a little above 174. Detailed
determinations are yet to be made. In a footnote Urbain remarks that
the* atomic weight of thulium, which is given as 171 in the tables, is cer-
tainly below 168.5. .„ ,

Radium.

Madame Curie* has redetermined the atomic weight of radium, with
purer material than that used in her former researches, and in largei
quantities. The well-known chloride method was employed, witb the
subjoined results:

Weight RiClt. Welghl Agd.' At. wt. Ki.

0.4053 , 0.3906 336.35

0.400 0.3879 336.04

0.3933:5 0-3795 336.15

, „ Mean, 336.18

' Ber., 40, 4730.
' Compt. rend., 145, 759.
' Ibid., 14s, 433.

* From the weights of AgC], o.ooooG gram is to be deducted in each case, repre-
senting filter ash.

REPORT OF COMMITTEE ON ATOMIC WEIGHTS, 317

Calculated with Ag = 107.8 and CI = 35.4. With Ag = 107.93 and
O = 35-45. Ra becomes 226.45, or 226.5 ^ round numbers. It is an-
nounced that Thorpe is also engaged upon a redetermination of this con-
stant.

Gaseous Densities.

Gray^ has determined the density of gaseous hydrochloric add, and
assigns to the weight of one liter, under standard conditions and in
latitude 45®, the value 1.6397 grams.

Guye^ has discussed the data relative to nitrogen and some of its gaseous
compounds, and assigns to it the atomic weight 14.010. D. Berthelot,*
from essentially the same data, finds N = 14.005. In an elaborate dis-
cussion of the whole subject, Guye* has finally adopted the following
figures for the weight in grams of one liter of each gas considered:

O, 1 . 42900 CO, 1 . 9768

H, 0.08987 N,0 1-9777

N, 1.2507 HCl 1.6398

CO 1 . 1504 NH, o. 7708

NO 1 . 3402 SO, 2 . 9266 4

Air.... 1.2928

Hiiscellaneous Notes.

Watson,* in a brief note, has considered the simultaneous calculation of
atomic weights from a group of related ratios. Hinrichs' has proposed
a graphic method for achieving the same purpose. He has also published
several papers^ in which he seeks to establish integral values for the atomic
weights of Br, Mn, and CI.

A number of writers* have discussed relations between the atomic
weights. On the question of standards, see Erdmann,* and especially
Swarts.*® The latter proposes two methods by which the atomic weights
may be connected with the gas equation, and with the absolute system
of imits. General papers on the subject of atomic weights are by Brauner**

* Proc. Chem. See, 23, 119.

* Compt. rend., 145, 1164.
' Ibid., 145, 65.

* Arch. Sd. Phys. Nat. [4], 24, 34; Joum. Chim. Phys., 5, 203; Compt. rend.,
144, 976; Chem. News, 96, 175; This Journal, 30, 143. Other papers, mostly contro-
versial, are as follows: Guye, Compt. rend., 144, 1360; Berthelot, Ibid., 144, 76, 269.

* Nature, 77, 7.
•Compt. rend., 145, 715.

' Ibid., 144, 973; 144, 1343; 14s, 58. See also Chem. Zentralb., 1907, 1958.

•See Wilde, Manchester Lit. Phil. Soc.,51 [i]. No. 2. Stromeyer, /fewf., No. 6.
Minet, Compt. rend., 144, 428; Collins, Chem. News, 96, 176; Verschoyle, Ibid.,
96, i97;[^Delatmy, Compt. rend., 145, 1279.

* Chem. Zeitung, 31, 95.

*• Bull. Acad. Roy. Belg., Classe des Sciences, 1907, No. 3, p. 2 1 2.
" Chem. Zeitung, 31, 483.

[8 Arthur A. noybs and yogoro kato.

id Richards.' Richards's paper is in the form of a lecture delivered
ifore the German Chemical Society.

D. S. Gbolooical SnsvBT.

lOmKIBUTEONS FROM THE RSSBARCH LABORATORY OF PHYSICAL CHBUISTRY OF IRE

MAssACHussrre Institutb of Tbchnology, No, 21. )

HE EQUWALEITT CONDUCTANCE OF HYDROGEN-ION DERIVED

FROU TRANSFERENCE EXPERIMENTS WITH NITRIC ACID.

Contents: i. Outline of the investigation. 2. Preparation and standardiiatioo
the solutions. 3. Description of the experiments. 4- The experimental data.
Summary of the transference numbers. 6. Summary and discussion.

I. Outline of the Investigation.
In an article published four years ago by A. A, Noyes and G. V. Sain-
let' there were described some transference determinations tnade with
1 20, 1/60 and 1/80 normal hydrochloric acid at 10°, ao°, and 30°, which,
hen combined with the equivalent conductance of chloride-ion {usmj
le value of Kohlrausch) yielded for hydrogen-ion a much higher equiva-
nt conductance than that which had been derived from the conductivity
\ acids at high dilutions. Thus the value for hydrogen-ion at 18° de-
ved from the transference experiments was 330, while that of Kohl-
lusch derived from conductivity was 318. This serious divergence
ppeared greater than the possible errors in the transference detemiiiia
ons;' and it seemed as if it must be due either (i) to an error in the ex-
■apolated values of the equivalent conductance of acids at zero concen
ration, (2) to the formation of complex ions or some other abnormality
f the hydrochloric acid, or (3) to a marked difference in the relativt
elocities of the hydrogen-ion and the anion, at moderate and at ven
iw concentrations. To test the first of these posdbilities, a study ol
le effect of the impurities in the water upon the conductance of veri
ilute hydrochloric and nitric acids was made in this laboratory by H
[. Goodwin and R. Haskell,' the results of which showed that, afte;
iiminating the effect of impurities as far as possible, a value for the equiva
:nt conductance of hydrogen-ion at extreme dilution (315 at 18") tva
iwer than that previously derived by Kohlrausch (318) was obtained

' Ber., 40, 2767.

* This Journal, 34, 944-968; 35, 165-168 (1903-3); Z. physik. Chem., 43, 49-7.
;903).

' The experimental results of Noyes and Sammet have recently been fully coo
rmed by those of Jatin, Joachim and Wolfi (Z. physik. Chem., 58, 641 (1907)).

' Phys. Rev., 19, 369-396 (1904); Proc. Am, Acad., 40, 399-415 (1904)- ^
iewed in Z, phy^k. Chem., 5a, 630 (1905).

•

In view of these results it did not seem possible that the divergence
could be due to the first-mentioned cause. The present investigation
was therefore undertaken, in order to test the second explanation, or that
being excluded, to estabish the correctness of the third one. It was car-
ried on with the help of a grant from the Carnegie Institution of Wash-
ington, and a description of it substantially identical with that here pre-
sented forms a part of Publication No. 63 of that Institution.

It was thought that independent transference experiments with an-
other acid, if they yielded results concordant with those with hydrochloric
add, would serve both to exclude any specific error that might arise
from complex ion formation or other individual peculiarity of that acid
and to confirm the experimental accuracy of the transference data, and
that they would thus establish the fact that a marked change in the rela-
tive migration velocity of the ions of acids takes place on passing to ver>'
low concentrations. Nitric acid was selected as the second acid, since
it is of quite a different chemical character.^ Another purpose of this
investigation, bearing directly on the third suggestion mentioned above,
was to extend the transference measurements with both acids to a dilu-
tion of about 0.002 normal.

2. Preparation and Standardization of the Solutions.

The chemically pure nitric acid of trade was freed from lower oxides
of nitrogen by diluting it with two-thirds its volume of conductivity
water and drawing a current of purified air through it. It was carefully
tested (using 5-10 cc.) for chloride with silver nitrate, for sulphate by
evaporation with barium chloride, for ammonia with Nessler reagent,
and for nitrite by diluting and adding starch and potassium iodide. These
impurities could not be detected at all, or were present only in entirely
insignificant quantity. Dilute solutions (from 0.06 to 0.0006 normal)
were made up with water having in all cases a specific conductance lying
between 0.9 and 1.2 X io~* reciprocal ohms at 18*^, and were titrated
with the help of phenolphthalein against a o.i normal solution of care-
fully purified barium hydroxide. The strength of the barium hydrox-
ide solution was determined gravimetrically both by precipitating with
sulphuric acid after neutralizing with hydrochloric acid and by evapora-
ting to dryness with pure nitric add and weighing the residue of anhy-
drous barium nitrate after heating to 1 60°- 180°. The two methods gave
for the content of the solution in milli-equivalents per kilogram 1 10.60
and 110.72, respectively; the value adopted was 110.64. Afterwards
two other solutions of barium hydroxide were prepared and titrated
against nitric acid solutions which had been standardized against the

* A single transference experiment has aheady been made with this acid at 25^ at
0.05 normal concentration by Bein (Z. physik. Chem., 27, 44 (1898)).

320 ARTHUR A. NOYES AND YOGORO KATO.

first barium hydroxide solution. Solution No. 2 contained 119.04, and
solution No. 3 contained 58.59 mtlU-equivalents per kilogram of solution
The five solutions of nitric acid varying from about 0.06 to 0,006 nor
mal, which were standardized for use in this work against these bariun
hydroxide solutions, showed as a mean in each case of 5 or 6 closely con
cordant determinations a content in milh-equivalents per kilogram 0:
solution as follows :

No. I. No, I. No. 3. No. 4. No. 5.

Content 59-3Z S7'4> 1S.436 6.S09 6.605

The very dilute solutions {approximately 0.002 normal) of nitric ant
hydrochloric acids employed could hardly be titrated with sufficient ac
curacy by this method. The concentrations both of the original solu
tions and of the portions after electrolysis were therefore determined bi
measuring their conductance by the usual Kohlrausch method in a cyHn
drical cell with horizontal electrodes, and dividing the correspoodinj
s[>ecific conductance by the equivalent conductance of the acid in ques
tion at this concentration and temperature. Goodwin and Haskell
have recently determined the eqmvalent conductances at 18° in aoo:
normal solution to be 371.3 for HNO„ and 375.0 for HCl at 18°, fron
which follows with the help of D^guisne's temperature-coefficients:
383.4 for HNO, and 387.4 for HCl at 20°, which are the values we haw
used in calculating the original concentrations. The actual conductano
measured in the conductivity vessel, the specific conductance, and tin
concentration in milli-equivalents per Uter calculated therefrom were a
follows:

Kitric acid golution. Hydrochloric acid tolulion.

No. 6. No, 7. No. 1. No,).

Actual conductance X lo* 3,142 2,094 i.975 2.136

Specific conductance X 10'. . . 847-3 818.4 78'-3 845.0

Milli-equivalents per liter 2.210 3. 161 3.017 ^ '^

The conductance capacity of the conductivity vessel was 0,3956 for al
the measurements presented in this article.* Hydrochloric acid solu
tion No. I was made by diluting quantitatively by weight (with wate
of conductivity 0.9 X io~°) a 0.13737 normal solution which had beei
standardized by weighing the silver chloride obtainable from it; the con
centration calculated from the dilution was 2.015, in close agreement witi
that derived from the conductivity (2.017). Solution No. 2 was pre
pared from the same stock solution, which was itself made by treatinj
pure salt with pure sulphuric acid, redistilling the strong acid obtained

' Phys. Rev., 19, 381, 383 (1904). These values like all of ours given below wer
not corrected for the conductance of the water.

* Kohlrausch and Holbom, Leitverm6gen der Elektrolyte (1898), p. 199.

' A 0.009,954 normal potassium chloride solution measured in it showed as an aver
age of several detenninations a conductance of 31 1 1.3 X io~* reciprocal ohms.

EQUIVALENT CONDUCTANCE OF HYDROGEN-ION. 32 1

and diluting it; it was proved to be free from non- volatile matter and
from sulphuric add.

3. Description of the Experiments.

The apparatus, consisting of two connecting U-tubes, was almost iden-
tical with that used by Noyes and Sammet, and the procedure followed
in the transference experiments was nearly the same. Referring the
reader therefore to this article* for the main features, we will here de-
scribe only the modifications adopted in our work. In order to avoid
all danger from leakage, the two U-tubes were joined by drawing over
their ends two thicknesses of light black tubing, tightly wiring this on,
and entirely covering the joint with melted paraffin. The anode consisted
of a circular platiniun plate, convex downward, soldered with gold to
a platinum wire. The cathode was a straight platinum wire which dipped
into the solution always less than i cm., so that by having the current
dense the reduction of the nitric acid was as far as possible prevented.
Since the solution weakened around the cathode and concentrated around
the anode, to avoid stirring, the cathode arm was filled with liquid nearly
to the top, while the anode arm was filled only a few centimeters above
the bend and the electrode was placed just below the surface. To keep
the solution at this level the anode arm was fitted with a rubber stopper
carrying a delivery tube which dipped into an outside vessel of water
whose level could be varied.

Given in outline, the method of carrying out the transference experi-
ments consisted in passing a suitable current for three hours and fifteen
minutes (except when otherwise noted in the table) through the stand-
ard nitric or hydrochloric acid solutions in the apparatus just described,
determining the quantity of electricity by means of two silver coulome-
ters placed in series with it, one on either side, dividing the electrolyzed
solution into a cathode, an anode, and three middle portions, and titra-
ting each of these with barium hydroxide (or, in the case of the 0.002 nor-
mal solutions, measuring the conductance at 20°) to determine the con-
centration changes. From the analyses of the cathode and anode por-
tions two separate values of the transference number were obtained,
and by the analysis of the middle portions it was made certain that no
error arose through convection.

The method of procedure at the end of electrolysis was to transfer
by means of a pipette the three middle portions to tared wide-mouth
Erlenmeyer flasks with rubber stoppers. Then the two U-tubes were
separated from each other, stoppered, well cleaned and dried outside,
and weighed. The solutions in them were then, after thorough mixing,
poured as completely as practicable into tared flasks, again weighed,
^ This Joumal, 24, 946 (1902).

3aa ARTHUR A. NOYBS AND YOGORO KATO.

and finally titrated, allowance being made in the calculation for the small
portion that remained in the tubes, which were themselves cleaned, dried,
and weighed empty. In the titration of all the portions, the quantity
of barium hydroxide solution added was determined by again weighing
the flasks containing them after exact neutralization with the base. In
those cases where the titration was replaced by a measurement of the
conductance, each portion was poured in succession into a cylindrical
conductance cell with horizontal platinized electrodes 2.5 cm. apart and
measured as accurately as possible, using three resistances in the rheo
Stat.

The principal error to be feared was that which might arise in Ihu
analysis of the cathode portion through the reduction of some of the
nitric acid by the electrolytic hydrogen. To reduce this to a minimura
the cathode was, as already stated, made as small as possible. Sines
careful analytical tests' showed (except in one experiment, No. z, when
the cathode was known to be badly arranged) no nitrite or ammonia ir
the cathode portion or nitrous vapors in the hydrogen evolved, there is
good reason to believe that the error from this source was not seriou!
in most of the other experiments. The effect of this error, it may hi
noted, would be to cause an apparent increase in the transference numbei
of the anion when calculated from the cathode change.

In case of the 0.002 normal hydrochloric acid solution investigatec
there was the possibility of an opposite error from the liberation of chlo
rine at the anode, which would have resulted in too small a transferena
number as calculated from the anode change. With so very dilute s
solution and the low current density used, there was probably little dan
ger of this; but to detect any such effect, two different forms of anode:
were employed^ — a short platinum wire in Experiments 1-5 (see Tabk
1) and a platinum discin Experiments 6, 7, 9, and 10. As the mean re
suits (167.8 and 168.8) with the two electrodes with such different sur
face areas agreed almost completely, it seems hardly possible that then
was a serious error from this source, especially in the latter experiments.

In order to determine what error, if any, might arise in the very di
' These tests were made by adding to 10 ec. of the cathode portion after it;
neutralization a few drops of pure sulphuric acid and some starch solution containisi
potassdum iodide; by adding to 10 cc. of the neutralized portion a tew drops of Nessfc
reagent; and by conducting the hydrogen evolved at the cathode through a tube con
taining filter paper moistened with a solution of starch and potassium iodide. A]
these tests gave a slight positive indication in the one experiment mentioned above
but in no other case, though they were tried in most of them.

' The cathodes were also varied in form (since the cathode results were consider
ably higher than the anode results), though there seemed to be no possibility of aj
abnormal reaction. A platiniun disc was used in Experiments 1-5, a spiral wire in 6-8
and a short straight wire in 9-10. The form of electrode had no influence, howeva
In Experiment 8 a silver anode was used.

BQUIVAI^NT CONDUCTANCE O^ HYDROGBN-ION. 323

lute solutions from contamination during the experiment, a ** blank"
experiment was made, in which the solution was treated in absolutely
the same way as usual except that no current was passed. The stock
solution of hydrochloric acid used (No. 2) had a conductance of 21372
and the portion withdrawn at the end of the experiment had conduc-
tances as follows: K,* 21336; M^, 21355; M, 21349; M^, 21349; A, 21356.
There was on an average a decrease of o.i per cent. Although this would
cause a not considerable divergence of the cathode and anode transfer-
ence numbers, yet it would not appreciably affect their mean; therefore
no correction was made for it (except that the use of 21360 as the initial
value eliminated it in great measure in the experiments with this solu-
tion).

4. The Experimental Data.

The data of the experiments and the calculated transference values
for the 0.06-0.007 normal nitric add solutions are given in Tables 1-3.
The first column contains the number of the experiment; the second, the
number of the acid solution used ; the third, letters representing the differ-
ent portions submitted to analysis, K signifying the cathode solution,
Mjp the adjoining middle portion, M the next portion, M^ the portion
adjoining the anode, and A the anode portion itself; the fourth, the weight
in grams of the separate portions; the fifth contains the number of grams
of barium hydroxide solution used in neutralizing the portions after the
electrolysis; the sixth, the initial content, expressed in equivalents and
multiplied by lo*, as calculated from the weight of the portion and the
standardization value;' the seventh, the final content calculated from
the barium hydroxide used; the eighth, the; change in content of the sepa-
rate portions; the ninth, the total change in content, which includes
the changes in the portions adjoining the cathode and anode ;' the tenths
the milligrams of silver precipitated in the coulometers; and the eleventh,
the calculated transference numbers for the anion multiplied by 1000.*

' For the meaning of these letters see the next paragraph.

' See Section 3. BaOaH, solution No. i was used in Experiments i to 6 ; solution
No. 2 in Experiments 7 to 26; and solution No. 3 in Experiments 27 to 32.

' Except where the change in the adjoining portion was opposite in sign to that
in the electrode portion.

* The way in which these were calculated may be illustrated with the help of the
data obtained in the first experiment. The cathode portion submitted to analysis
weighed 214.08 grams and was found to require 107.72 grams of the BaOgH, solution
containing 0.11064 milli-equivalent per gram, so that the final content of the portion
was the product of these last two quantities or 11.918 milli-equivalent s. To deter-
mine the original content the weight of the portion is multiplied by the original con-
centration of the solution (0.05922 milli-equivalent per gram), which gives 12.678
milli-equivalents. The decrease in content in the cathode portion is, therefore, 0.760
milli-equivalent. Adding to this the decrease in the adjoining middle portion (0.005)
and dividing by the ntunber of milli-equivaleuts of silver (523.0/107.93) precipitated

ARTHUR A. NOYSS AND YOGORO KATO.

Table i. — ^Transpekbncb Data p

1 '/„ Normal Nrnac A

314

107
39

313

>I3

i86

303

169

734

107

186

18s

108

29S

171

245

116

126

163

•5

137

268

158

304

'47

109

182

155

273

161

381

'37

134

139

90
38

75
131

304

'49

137

136

133

344

143

267

133

izS

140

131

336

138

356

I30

.48

141

132
274

89
66

69
155

13.678

4.4>5

IS. 897
18,012
6.49^

18, 755 + 760

7.451

8,143

17.517
16, 348

6,501
10,828

9.23'

7.970
7.947
8,385

7.975 + 5

7.947 ± o

8,399 +14

'4,565 +1,455 +'.

14,494

15,906

15. 349

13,601

7.365

7.358

8,091

8,085

7,568

7.580

13.590

15,305

14.705

13, 293

8.537

8.521

8,103

8,094

7.630

7,640

in the coulometer, the
tion for the change in
ference is applied later.

5,770 17. '80 +1,.
transference number is found t<
weight of the electrode portion

be 0.1 579- The
by the elcctroly

EQUIVALENT CONDUCTANCE OF HYDROGEN-ION.

325

Tabls I {Continued).

n
o

1 1

&
o

4
295.01
105.16
148.18

147.77
281.13

258.40

135.90
143.28

139.93
253.26

343-52
148.88
162.04

139.38
256.25

275.18

154.33
150.66

132.84
304- 76

0*0

OS
«**

PQ
5
141.78

54.54
76.88

76.75
157.05
124.30

70.51

74.35
72.63

141.29

167.97

77.20

84.05

72.33

143.14

133.84
80.06

78.12

68.94

167.07

16, 939
6,038
8,008

8,485
16,144

I4» 837
7,803
8,227
8,034
I4» 542
19*725
8,548

9,304
8,003

14,713
15,801

8,861

8,650

7,627

17,499

15,687
6,035
8,506
8,492

17,376

13, 753
7,802

8,227

8,036

15,633

18, 584

8,542

9,300

8,003

15,838

14,808

8,858

8,644

7,628

18, 485

^^
c S

O
8

—1,252

— 3

— 2

+ 7
+ 1,232
— 1,084

— I
± o
+ 2
+ 1,091
—1,141

— 6

— 4

± o

+ 1,125

— 993

— • 3

— 6

+ 1

+ 986

V .

<
10

§1

fie
H

—1,255 861.8 157.2

+ 1,239
—1,085

861.4
758.2

155
154

• • •

+ 1,093
—1,147

757.6
783.3

155
158

+ 1,125
— 996

783.0
687.1

155
156

+ 987 686.8 155

2
5

o
5

Tajsub 2. — Transpbrbnce Data for 0.0184

4,879
2,620

3,325
2,767

6,370

5,037

2,799

14.

288.64

142.34
180.71

150.21

321.64

305 . 45
151.88

40.98
22.01

27.93
23.24

53.39
42.31
23-51

5,319
2,623

3,329
2,767

5,927
5,628

2,799

OR V54 Normal Nitric Acid

—440 —443 300.7
~~^ 3 • *
"""" 4
+ 0

+ 443
—591

AT 20®.
159-0

+ 443
—591

300.8
402.0

1590
.158.7

129.16
308.11
334 04

157.46
164.29

133-13
363.95
353.27
161.93

175.48
136.96
299.07
342.27
154.70

20.01
52.70
46.44

24.34
25.42
20.62

61.59
49.64
25.06

27-13
21.23

51.23
46.04

23.89

2,380

5,677
6,155
2,901
3,027

2,453
6,706

6,509
2,984

3,233
2,524
5,511
6,307
2.851

2,382
6,273

5,529
2,897
3,026

2,455
7,332
5,909
2,983
3,230

2,527
6,099

5,481
2,844

+ 2

+ 596
— 626

— 4

— 1

+ 2
+626
— 600

— 1
+ o

+ 3
+ 588

—826

— 7

+ 598
— 630

401.8
420.8

160.6
161.6

+628
—601

421.0
400.5

161. o
162.0

+ 591 400.5
—833 564.6

159.2
159.2

ARTHUK A. NOYSS AND YOGORO KATO.
TablS 3 {Cortttwiaii.

^ i^ I I 8 «a

% £

M 171-63 36.57 3.11

Ma 145.81 33.60 3,61

A 347-OI 60,64 6i3'

18 3 K 280.38 36.50 5, li

Mi 145-40 3J.50 a,6

M I59-6S 34-70 2,9-

M* 135.00 30.97 2,4

A 307-95 54-44 5,6

19 3 K 340.61 45.45 6,2

Ms 137-96 »9-77 3.3

M 159-9" 34-74 3.9

Hi 15238 33.65 3,8.

A 35591 62.33 6,5,

30 3 K 2B7.80 39.08 5,31

Mc 102.36 15.79 1.8:

M 111.64 17.36 3,0,

M* 131.39 30.30 2,4:

A 43333 70.95 7.7'
Tabls 3. — Transfbkbncb Data 9

31 4 E

3.163
3.690

4.345
3.678
2.940
2.496
6.4S0
5.4»0
3.354
3.945
3.815
7.419

+ 815 553
— 870 588

■4-861 +868
— 650 — 656

8,446
: 0.0067

— 4 ..
+ 647 +647 441
OR */„o NoBHAi, Nnxic i

13
to6
J39

53
87
33

27
15

176

27
)59

78
34

7
34

'.053 — 35

1.584 3.4S3 +838 +848 578.

1,609 3,085 —534 —53' 353.

840 833—7

973 963—9

771 778+7

1, 770 3. 385

+5"5 +533 35a.

3.061

'.978

+ 497 +499 337-

870 873+3
1,447 3,860 +4'3 +4»6 38j,

BQUIVAUNT CONDUCTANCE OF HYDROGBN-ION.
Table 3 (.Coniinuedi.

If li II

u ts I"

H < e

Ma 131

»5

■3

—556

—559

377-5

159-7

— 3

— 9

+ 4

+ 557

+ 561

378.0

160.3

-538

-546

353-7

166.7

771

3.288

+ 517

+ 518

353-5

158- 1

444

1. 861

-583

—593

397-3

161. 1

900

890

— 10

983

978

— 5

770

777

+ 7

605

3.187

■1-583

+ 589

397-5

160.0

688

3.145

—543

—550

371 3

159-9

847

840

— 7

978

974

— 4

853

858

+ 5

750

3.394

+ 544

+ 549

371-3

159-6-

693

3.3'9

-364

-368

336.3

168.3

744

740

— 4

058

1,059

+ 1

791

+ 0

724

3.078

+ 354

336.3

161. 8

551

3,140

—411

— 4>7

375-6

•63 3

+400 +400 375.8 156.6
—699 —703 473.0 160.7

871

877

+695 +701 473.1 160.3

Tables 4 aad 5 present the results obtained with the more dilute solu-
tions, where the concentration was determined by conductance meas-
urements. The first four columns are the same as in the preceding tables.

' In this experiment (No. 30) the period during which the soluUcm was electrolyzed
was greater than the usual time (3} hours), namely, 6 hours.

' In theae expenments (Nos. 31 and 33) the solution was electrolyzed 4) hours and
6 bows, respectively.

C^W^^^-'i 'i' Jr* Jf^i

- » r^i

•^v'«r ;..^V^

» ?IT • .< ■ '> '• ■• • ■ ■ 1.71

328

ARTHUR A. NOYES AND YOGORO KATO.

ABU

* 4.-

-Transitsrencb Data for 0.0022 Normal Nitric Acid

20*.

Bxperiment
No,

•

1
1

Weight of por-
tion.

Change in con-
ductance X

Change in con-
tent.

Total change
in content.

Ag in coulom-
eters.

Transference
number X io».

33*

349.79

1,498

—644

—2,336

—2,345

154.5

163.8

97.13

2,133

— 9

■ ■ •

• a

B • •

151.75

2.139

— 3

— 5

• • ■

• •

• ■ •

1x8.86

2,156

+ 14

+ 17

• • •

• •

■ • •

389.53

2,707

+ 565

+ 2,282

+ 2,299

154

■5

160.6

349.70

1,813

—281

—1,019

— 1,021

165. I

"9-33

2,092

— 2

• • •

• •

• • •

« •

2,086

— 8

• • •

■ • •

• •

■ • «

•

"3.57

2,095

+ I

• • •

m m

• • •

359.49

2,358

+ 264

+ 984

+ 985

159.3

359.94

1,562

532

—1,986

—1,988

131

• « •

106.04

2,092

— 2

•
■ • ■

• •

• • •

134.48

2,089

— 5

— 7

• • •

■ •

• • •

113.76

2,104

+ 10

+ 12

• • •

m •

• • «

393.43

2,571

+477

+ 1,946

+ 1,958

131

161. 0

350.57

1,528

—566

—2,058

—2,073

135

■5

165. I

104.87

2,080

— 14

— 15

• • •

• •

• • •

132.59

2,090

— 4

— 5

• • •

■ ■

• ■ •

"5. 59

2,105

+ II

+ 13

• • •

• •

■ • •

386.20

2,603

+ 509

+ 2,039

+ 2,052

135

163.4

■ •

• • •

134

• ■ a

125.23

2,086

— 8

— 10

• • m

• • 1

• • •

134.42

2,095

+ I

• • ■

• • <

• ■ •

139.43

2,104

+ 10

+ 14

• • •

• •

■ ■ •

376.86

2,606

+ 512

+ 2,001

+ 2,015

134

161. 6

Tabve s

. — ^Transpbrbncs Data ]

POK 0.002 ]

[ Normal Hydrochi/)ric Acid at 20^.

384.26

1,250

725

— 2, 862

—2, 883

178.6

174-2

125.64

1,959

— 16

— 21

■ ■ ■

• ■ 1

• • *

135-42

1,971

— 4

— 5

■ • •

■ • «

• ■ •

126.10

2,000

+ 25

+ 32

■ » •

• • 1

• ■ ■

389.93

2,658

+ 683

+ 2,736

+ 2,768

178.

167. 1

313.49

1,360

—615

—1,980

—1,988

123-

173-3

132.60

1,969

— 6

— 8

■ • •

• • 4

* • •

131.29

1,973

— 2

— 3

• • •

■ • ■

« « »

112.29

1,989

+ 14

+ 16

• • •

• • «

■ > •

385.67

2,455

+ 480

+ 1,902

+ 1,918

123.

167.2

372.66

1,466

509

—1,949

—1,955

120.

175.8

121.26

1,970

— 5

— 6

• • •

• • a

• • •

135.37

1,970

— 5

• « •

• « «

• • •

130.33

1,985

+ 10

+ 13

• • •

119.

• « •

In this ez]

seriment (I

^0. 33) the electrol]

iTsis was continued for 2

lJhoi

UTS instead of

for 3i hours as usual.

BQUIVALBNT CONDUCTANCE OP HYDROOBN-ION.
Table s (Crattfuicd).

385. n

1.395

—680

—2

690

lis. 65

1.935

— 40

—

143- 10

1.959

— 16

136.67

1.995

-f 30

430-37

3.577

+ 603

661

388. a8

1.497

-478

—I

906

.22.78

1,966

— 9

—

138.55

1.970

— 5

—

133.73

1.975

+ 0

434.00

3.383

+408

+ 1

777

420.70

1.546

—590

— 2

549

131. 58

3,116

— 20

—

148.97

3,127

— 9

—

.03.76

3.163

+ 37

458.17

2,664

-1-538

485

443.00

1.761

—375

-rl

706

122.55

3.133

— 3

—

143.53

3.136

+ 0

137.24

3,141

+ 5

476.09

3.473

+ 337

+ 1

648

436.04

1,863

—374

— 1

337

107.3a

3,126

— 10

—

142.93

3.139

— 7

—

118.31

2,078

-58

—

458.83

1.819

—317

—I

494

107.75

3.135

— 1

—

160.12

2.133

— 3

—

104.29

3.143

+ 6

478.33

3,430

+ 394

+ 1

444

438.30

1.835

—301

— I

355

114.76

3.131

— 5

—

143-57

3.133

— 4

—

104.46

2.136

+ 0

471.81

2,411

+375

+ 1

333

73»
687

171.0
170.9

173
169

917

115.0

180

777

114.8

.67

574

161.0

173

513

161. 1

168

710

105.8

174

655

105.8

168

338

78.0

77-7

171

495

93.7

174

450

93.8

168

361

85.0

173

333

84.9

169

The fifth contains the actual conductance X lo*; the sixth, the difference
between this value and the initial conductance X 10* as given at the end
of Section 2 ;' the seventh, the corresponding change in content of the
whole portion, expressed in io~* equivalents, obtained by multiplying
this difference by the conductance capacity of the vessel (0.3956), divid-
ing by the equivalent conductance values 382.1 for HNO, and 385.8 for

' In Experiments 5, 7 and 8 the electrolysis was continued for only 2I hours.

'These initial values ore: 3142 for HNO, solution No. 6; 2094 for HNO, sotution
No. 7; 1975 for HCl solution No. i;and 2136 for HO solution No. a.

I30 AR'THUK A. NOYES AND YOGORO KATO.

^Cl,' and multiplying by the volume of the portion (obtained from its
veight by multiplying it by 1.0018) ; and the eighth, the total change of
;ontent or the sum of the changes in the electrode portion and the ad
oining portion. The ninth column contains the milligrams of silver de-
Misited in the coulometers; and the tenth, the transference number for
he anion X 10*.

5. Summary of the Transfereace Numbers.
The following table contains a summary of the transference numljers
lerived from the preceding experiments together with the means de-
lved therefrom. In finding the separate means of the cathode and anode
'alues a few abnormally high or low values (designated by an asterisk)
lave been omitted." To these means in the case of the two most concen-
rated solutions a correction has been applied to remove a small emii
ntroduced by the method used for the calculation of the separate values,'

' These values are those of dL/dc at 0.003 aonnal, where u represents the spedfu
onductaace and c the equivatent conoentration. We derived them through a caiclu
ODsideration of all the results obtained by Goodwin and Haskell with both acids ai
8° between the concentrations ot 0.001 and 0.005 normal. The values were firs
etived at iS" and were found to be 370.0 for HNO, and 373.5 for HCI, and these wen
ben increased with the help of Wguisne's coeificients so as to make them cortesponi
3 ao". It is scarcely possible that the errors in these values exceed 0.3 per cent.

' The high cathode values in Experiments 1, 3 and 4 were probably due to reduc
ion by the electrolytic hydrogen, which was proved to have taken place in Experimoi'

The cathode value in Experiment 31 was omitted since the middle portion shown
large change in content.

' Namely, in calculating the original content the total weight of the electrodi
ortion was amply multiplied by the initial content per gram. That weight had
owever, been increased, over what it would have been originally, at the anode by th
(dgbt of the transferred nitric add and had been decreased by the electrolysis out of i
f the water corresponding to the hydrogen and oxygen evolved ; and at the cathode i
Ad been decreased by the weight of the transferred nitric acid.

■ ' By considering the effect ot this on the restilt, it will readily be seen that when an;
dd of equivalent weight a, transference number n, and original content c in equivalent
ler gram of solution is electrolyied as in this case with the production of hsrdrogen ani
<x^en, and the calculation is made as above (multiplying the total weight of lb
Mjrtion by c) then the anode transference-number should be increased by the fractiona
mount {a« — 9)c/» and the cathode transference number should be increased by th
ractional amount AC, In this case, with the strongest (0.058) normal solurion, di
orrections, apphed (since a — 63, 11 — 0.156, and c — 0,000058) are +0.03 per ceni
n the anode value and +0.36 per cent, on the cathode value. With the o.oiS
loimal solutions the corrections are one-third of these percentages.

The corresponding correction was not applied by Noyes and Sammet to thei
esidts with hydrochloric add. It would have the effect of increaang their final valu
t 0.05 normal (165.69) by just 0.17 per cent, (to 165.96), while at the lower coocer
rations the correcrion would be scarcely appredable.

A more ample way of calculating transference numbers from the experimenli
lata is to refer the initial content to the weight ot water present instead of to that i
he whole solution, and to calculate correspondingly the wdght ot water in the portio

BQOIVALENT CONDUCTANCB OF MYDROGEN-ION.

331

and the results are designated "corrected means." These cathode and
anode means have then been combined in the case of the three stronger
nitric acid solutions under the assumption that each has a weight inversely
proportional to the square of its average deviation (A. D.). Since the
cathode values show in all three cases much greater variations, this pro-
cedure gives to the anode values a much greater weight, which would be
a priori desirable since they are not subject to the possible error arising
from the reduction of the nitric add around the cathode. It is in fact
very probable that both the larger variations and the greater magnitude
of the cathode values are due to this cause. In spite of this source of
enor, it is to be noted that the mean cathode value exceeds the mean
anode value by only 0.9, 0.6, and i.i per cent., respectively, in the case
of the three more concentrated solutions. Taking into account the fact
that almost all other errors affect the two results in opposite directions,
we believe the final A. D. values give a fair measure of the probable pre-
cision of the final results, which is from o.z to 0.3 per cent, for the 0.06
to aoo7 normal nitric acid solutions.

In the case of the 0.002 normal solutions of both adds the divergence

of the cathode and anode mean values is much greater, and it seemed

best to assign an equal weight to each without reference to the value of

Tablb 6. — Summary ov ths Transperbncs Vai,17Ss.

iLgje noriul HNOi at »°. 0.0184 normsl HNOi at xP. ^,J3o6^ norniBl HNOi at »°.

Catbode. Anode.

Cathode. Anode.

•i6r

i,S6

*i6i

ISS

*i6o

I.S6

i.H

"55

156

15(>

157

"55

156

15.';

"57

"55

IS4-

las

<,S8

".W

156

155

15*.

155

157

155

14 158.7 160.6

15 161. 6 161. o

160.5 "59" '6-
159-5 "59-' 27.
160.3 "58-1 38.

160.10 159-30 29'

30..
160.39 159-33 31-'

A. D

58 Final mean .
41 Final A,D. ,

A. D

Final mean .
KnalAD..

I.S8

■,58

160

163

"59

164

"59

119

160

166

■ ■i8

161

160

I.S9

■59

i6«

161

16,1

•1.16

160

161

159

4«

AD.. 0.37

after the electtolysis by subtracting from i1

. total weight the weight of solute found

in it; but even then a correction must be applied to the anode portion for the v
electrolysed out of it. The present ba^9 of all such transference determinations is
coune the assumption that the water itself does not migrate.

ARTHUR A. NOYES AND YOGORO KATO.
TablS 6 (Continatd).
ilHNO,al»°. ooojiNonnalHClit

EipcHment No. Cathode. Anode. Biperiment Do. Calhode. Anode.

33 163.8 160.6 1 174.2 "le? 1

34 '65- 1 "39-3 2 173-3 'ift? I

35 >63.5 161.0 3 175.8

36 "65-1 1633 4 173-0 'legj

37 ■ ■ ■ 161 -6 5 *i8o. I 'iSt.o

164.4 161.3 6 >7i.5 16*5

a. d

A.D 0.4 0.5 8 171.6

Knal mean.. 161.8 9 174.1 i6S,8

Final A. D... 1.2 10 173,0 ,69,3

Mean. 173.5 16S.8'

A,D 0.3 0.1

Pinal mean .. 171. i

FmalA.D... 1.7

its average deviation; for the divergence probably arises in the main from
a slight contamination of these very dilute solutions during the experi
ment, which would affect the cathode and anode values oppositely and
about equally. The final A. D. values, which expressed as percentage
are 0.7 per cent, for the nitric acid and i.o per cent, for the bydrocfabrii
add, are again a fair measure of the maximum error of which there is
any reasonable probability.

6. Summary and Discusuon.
The final results of the transference experiments described in this art!
ck, as well as of those carried out by Noyes and Sammet' with 0^)5-
0.006 normal hydrochloric add at 2o°,* are brought together in Table 7
In this table are also given the values of the equivalent conductana
of bydrogen-ion calculated from each transference number and from thf
most probable values for nitrate-ion and chloride-ion (64,6 and 68.5, re
spectively) at 20° and extreme dilution,' In the last row of the tabli
are given the corresponding values for zero concentration as derived froir
Goodwin and Haskell's conductivity experiments."

' The mean of all the anode values is 168.3 but it seems best to omit the first foui
in which experiments an anode of small surface was used, and which are somcwba
lower perhaps owing to the evolution of a small quantity of chlorine.

' Z. phyak. Chem., 43, 63 (1903) ; This Journal, 34, 958; 35, 167 (1901-3).

' Corrected for the inaccuracy in their calculation as described in a preodini
footnote.

* The value here given tor the CI is that derived by Noyes and Sammet fror
Kohlrausch's conductivity data and the existing transference data for potasaun
chloride. That for the NO, ion we have obtained by subtracting from that for ihe C
the difference tor these two ions at ao" given by Kohlrausch {Stlzungsber. iomgi
prenss. Akad. der H'ifienjcA., 1901, 1031). These values have then simply b«i
multiplied by (i — n)/n.

' These investigators found for A, at 18° 377.0 for HNO, and 380.1 for HCI. Thi
corresponding values at 30" calculated with Wguisne's coefficients are 389.2 am
393.5 respectively. Subttucting from these the values for the NO, and d ions (64.1
and 68.5) one obtains the values for the hydrogen-ion given in the table.

EQUIVALENT CONDUCTANCE OF HYDROGEN-ION. 333

Tabu 7.— Final Values op the Transpbrbncs Numbers and the Equivalent

Conductance of Hydrogen-ion.

Equivalent p4

sr liter.
HCl.

Transference number
X io».

Equivalent coi

of hydrogen

experimei

HNOs.

aductance
•ion from
itB with

HNOs.

HCl.

0.058

0.051

155.7

166.0

350.3

344.2

0.0184

0.017

159 -6

167.5

340.2

340.5

0.0067

0.0056

160.0

167. 1

339.1

341.4

0.0022

0.0021

162.8

171. 1

332.2

331.8

166.0

174-5

324.6

324.0

It will be seen from Table 7 that, except at the highest concentration
(0.055 normal), there is substantial agreement between the values of the
equivalent conductance of hydrogen-ion derived from the independent
transference experiments with the two different adds, and that the (nearly
constant) value for the concentration interval between 0.018 and 0.006
noraial is nearly 5 per cent, larger than that derived from conductivity
measurements at extreme dilution. The reality of this divergence,
first discovered by Noyes and Sammet, confirmed as it is on the conduc-
tivity side by the investigation of Goodwin and Haskell and on the trans-
ference side by the recent determinations of Jahn, Joachim, and Wolff,
and by these new experiments with nitric add, can, we believe, no longer
reasonably be doubted. It must therefore be concluded that the trans-
ference number of the anion of acids, and therefore the ratio of the velocity
of the anions to that of the hydrogen-ion, is several per cent, larger at very
smaU concentration (0.001 normal and less) than at moderate concentra-
tions (0.05 to 0.005 normal). Thus a change in the relative velodties
takes place even after the concentration of the solute has become so
small that as a meditun the solution scarcely differs from the pure solvent.
The fact that higher transference numbers were obtained with the 0.002
nomial solutions than with the more concentrated solutions of both adds
confirms the conclusion drawn from the comparison with the conduc-
tivity data. The values obtained at 0.002 normal show, moreover, that
even at this very low concentration the velodties have not yet become
identical with those at zero concentration.

This change of the transference number may, of course, arise either
from an acceleration of the anion or from a retardation of the hydrogen-
ion at very high dilution, or from both causes combined. The facts that
salts do not as a rule show any change in their transference numbers
after a moderate dilution is reached and that their ionization values cal-
culated from freezing-point lowering and other molecular properties
agree with those corresponding to the conductance ratio (A/A^y make
it probable, however, that it is the fast-moving hydrogen-ion that is

' See A. A. Noyes, Z. physik. Chem., 52^ 634.

334 ARTHUR A. NOYBS AND YOGOHO KATO.

mainly, if not wholly, affected.' It is under this (possibly
assumption, namely, that neutral ions have the same velocity at
and at very low concentrations, that the values, given in Table
equivalent conductance of hydrogen-ion at various concentrat
derived.

The fact that the values of the equivalent conductance of I
ion are nearly constant for the interval of concentration o.
seems to indicate that these are the normal ones, and that the
at lower concentrations arise from some secondary effect of
character, determined perhaps by the smallness of the ipn-c
tion itself.

The results obtained at the highest concentration (0.05 to 0.0
differ in the case of the two adds, which makes it seem probabl
variation in the stronger solution is due to some different cause,
one of a specific chemical nature, from that which gives rise to t
at high dilutions.

As to the bearing of these results on the calculation of ionizati
it may be said that in the case of largely ionized adds at modi
centrations it seems in the light of now existing knowledge m(
priate to divide the observed equivalent conductance of the ac
value obtained by adding to the equivalent conductance of
that of the hydrogen-ion obtained by the transference experime
described at the concentration in question. On the other hai
case of any add solution in which the ion concentration is less t
normal the older value (324 at 20° or 315 at 18") for bydrogei
be preferred.

It is of interest to compare the ionization of hydrochloric :
puted in the manner just stated with that of neutral salts of
ionic type, like potassium and sodium chlorides. At the com
0.05 normal the ionization value derived from Kohlrausch's vi
of the equivalent conductance of the add at 18** is found to
provided the equivalent conductance of hydrogen-ion is taken
derived from the conductivity of the acid at small concentratio
becomes 0.900 when the equivalent conductance of hydrogen ioi
6.2 per cent, larger than this, in accordance with the transferent
At this same concentration the ionization values for potassiun
and sodium chloride, as derived from their equivalent cone
are 0.891 and 0.878. The approximate agreement of these va
the new one for hydrochloric add seems to justify the extension
ionized adds of the prindple that salts of the same ionic type h£
same concentration roughly the same degree of ionization.

BOSIOH. Denmber. 1907
' It is therefore probable that the decrease Id the conductance of s
always observed at very high dilutions is not vholly due to impurities in tl

SAINTS, ACmS AND BASES. 335

[CONTRIBlxnONS FROM THE RSSSARCH LABORATORY OP PHYSICAL ChBBHSTRY OP THB

Massachusbtts Institute op Technology, No. 22.]

THE COHDUCTIVITY AND IONIZATION OF SALTS, ACIDS, AND
BASES IN AQUEOUS SOLUTIONS AT HIGH TEMPERATURES.

a Report by Artbur A. Noyes upon a Series of InTestigations by A. A. Notes, A. C. MblchiiR,

H. C. CoopBS, G. W. Eastman and Yooo&o Kato.

In a previous paper^ from this laboratory by A. A. Noyes and W. D.
Coolidge an apparatus and method were described for the accurate meas-
urement of the electrical conductivity of aqueous solutions at high tem-
peratures. Measurements with solutions of potassium and sodium chlor-
ides were presented and discussed. Further measurements have since
been made with these two salts, and the investigation has been extended
to other di-ionic salts (silver nitrate, magnesium sulphate, sodium ace-
tate, ammonium chloride, and ammonium acetate), to two tri-ionic
salts (barium nitrate and potassium sulphate), to an acid salt (potassium
hydrogen sulphate), to certain acids (hydrochloric, nitric, sulphuric,
phosphoric and acetic adds), and to certain bases (sodium, barium, and
ammonium hydroxides). With most of these substances the measure-
ments have been made at four or more different concentrations varying
between 0.1 and 0.002 normal and at temperatures ranging from 18° to
306^

For the original data and for a detailed description of the experimen-
tal methods and of the calculations, reference should be made to Publi-
cation No. 63 of the Carnegie Institution of Washington, of a part of
which publication this article is a brief summary.' Only the final results
will be communicated here.

Tables i and 2 contain the values of the equivalent conductance of
the various substances expressed in reciprocal ohms. The values of the
concentration given in the second column express the milli-equivalents
of solute per liter of solution at the temperature to which the conductance
value refers. (In the two cases of potassium hydrogen sulphate and
phosphoric add, however, the concentration is expressed in milli-formula-
weights of solute (KHSO^ or H8PO4) per liter of solution, and the values
are correspondingly the molal (or "formal") conductances instead of
the equivalent conductances). In obtaining these values excepting, how-
ever, the cases of the strong adds, the conductance of the water was subtrac-
ted, and those for sodium acetate, ammonium acetate, and ammonium
chloride have been corrected for the effect of the hydrolysis of the salts.
The atomic weights employed were those given by the International Com-
mission for 1905, referred to oxygen as 16.00. The temperature is the

' This Journal, 26, 134-170 (1904).

' Copies of that publication may be obtained at a cost of $2.50 each, by applica-
tion to the authorities of the Carnegie Institution.

ARTHUR A. NOYES.

• Oo<i~' oo-wo- o»30'<a^S- 01 to t~

1 a ■* n ■ o-ng*- ior~o-OV- ...

lOigor-- o<gSeo«- 0teot>..3>o. ...

rajS, : «S£8: J8 !: S S S X : 5 ?. J :

ill sSsIj KS&Sis ■ lill

53SS *SR8.J

^ a«

00000 00000 ooooooc

°"2ji8 °". SJ88 "••asiai

SALTS, ACmS AND BASBS.

I "■" " ^
I

Z . o « «

>c^tc tf>ci«Or^&

B - ■
5.0.

I I a 8 i ° - S S,

° " 2 j« 8 ° " 2 i«|

ARTHUR A. NOYSS.

.ior«08-*vo Q.>©-- ifif^Mn ... ....

g iflpl: lilSI t«i?! ESS BJ55

I -

?«::::; llsis p?ss rJs sill?

§ ii ; ; : ;: SkSkj pisi lit s*s?5

> o^o-o•e ot'O'^'* owomio f^ior- '^"or%inS

g_g o o o o o

8.H

looo ooooe ooo oooooj

" : " t " i 'i
1 : ;

SALT3, ACIDS AND BASBS.

■S - B : O-03 ...

5 r. P. DO • N

IE»-'

O ■ 5 g> t^ *

I .-saasj

Is^l as^JI ?'

P ha o o o o o oooo oooe

i.|| ° s a,« 8 ° " s a » " 2 s

I i

340

true temperature on the hydrogen-gas scale (as derived at the higbei
temperatures from the determinations of Jaquerod and Wassmer' of
the boiling-points of naphthalene and benzophenone). The conduct-
ance values (A^) at zero concentration were mostly obtained with the
help of an empirical function of the form i/A^ — i/A — K^CA)"'', which
corresponds to the equation C(J|, — A) = KiCA)', by plotting the recip-
rocal of the equivalent conductance (i/A) at the various coocentiatiODS
(Q against (CA)"'', varying the value of « till a linear plot was obtained,
and then extrapolating for zero concentration. For the slightly ionized
substances and for some of the others the J, values were derived from
the others by the principle of the additivity of iori conductances; and
in a few cases, the values at certain intermediate temperatures were ob-
tained by graphic interpolation; values derived in either of these ways
are indicated by enclosure within parentheses.

These conductivity results have interest from a theoretical stand-
point mainly in two respects — first, with reference to the equivalent
conductance of the ions or their specific migmtion-velodties; and second,
with reference to the degree of ionization of the various substances.

The directly derived values of A^ for the largely ionized electrolytes
are summarized in the following table. The substances are ananged
primarily according to the ionic type and secondarily in the order in which
the A^ values at i8° increase. In adjoining columns are given also the
mean temperature-coefficient AAJAt for the succesave temperature-inter-
vals and the ratio j1|)(s)/'*(i(kci) of '^^ equivalent conductance of the sub-
stance in question to that of potassium chloride at the same tempera-
ture.

The results given under -^ooj/^^ockci) ■" Table 3 show that the values
of the equivalent conductance for complete ionization in the case of all
the di-ionic substances investigated become more nearly equal as the
temperature rises, the approach to^rard equality being rapid between
18° and 218°, but comparatively slow at the higher temperatures. This
shows, of course, that the specific migration-velocities of the ions are
themselves more nearly equal, the higher the temperature. Complete
equahty has not, however, been reached even at 306", but the divergence
exceeds 5 per cent, only in the cases of hydrochloric add, sodium hydrox-
ide, and sodium acetate, which have ions which at iS° move with excep-
tionally large or small velocities.

The behavior of the tri-ionic salts, potasdum sulphate and barium
nitrate, is especially noteworthy. Their equivalent conductance increases
steadily with rising temperature and attains values which are much
greater than those for any di-ionic uni-univalent salt. Thus, at 306°,
the value for potassium sulphate is about r.5 times as great as that for
' J. chiin. phys., a, 72 (1904).

SAI,TS, ACIDS AND BASBS.

TabLB 3, — EfiDlVAUSNT CONDUCTANCB AT ZSXO CONCBOTXATION,

Sodium acetate.

a -53

■

a85

3.95

0.69

«o

3.40

0.7*

660

3.00

o.So

934

0.8a

130. 1

3.46

414.

3-77

625

3.13

8a5

3.86

1005

4.60

Barium Mr,

te.

II6.9

3.37

0.90

385

3-84

0.93

600

0.96

A„(KC1)'
0.84

387

4-44

309

0.8;
3-44

o.8<
331

0.9:
3 33

o.9<
4.40

0.91
chlorid*.

3-47
3.80
3-43

176

1.05

1.13-8

1.0a

3-93

455

1. 10

4.64

715

I.I.

S.64

065

1.39

460

6.37

1.45

115. 8

367

Ai ■ Ao(KCl)"

0.89
3.06

0.89

3- 6a

0.91
3-39

0.9s
3-94

0.96
4.00

0.95
im hfdroilde.

83s

!.I6

3.58

PhaaphoTlc add.

^18

3.60

377

'a-90" ' 37^'

,... -,.:^_,^i.

4.78

g.<ri

i ,: ■:' .'/ , .1-

S.76,.:.J IK,

730

3-57

1.76

S36

.3-95.

».99 , S6P;

- Al^A 7 ■

930

1-.49

'047
1330

, 3.95

Iv67,. .,>P«3i :
1.49 ,, }3^.

...■:.■■ 143+..

.1,,,,. : .'»-73
, .... .1.-83

342

ARTHUR A. NOYES.

!*■

potassium chloride. This behavior, which at first sigh'
mal, is in reality in conformity with the principle that
ions subjected to the same electric force approach equ
temperature; for, assuming that the resistance of the i
the same for all ions, the velocity of a bivalent ion, ow
electric charge, should become twice as great as that o1
under the same potential-gradient; and correspondingly
conductance of a completely ionized uni-bivalent salt si
times that of a completely ionized uni-univalent salt
markable is, therefore, not the greater values at high t<
the approximate equality at room temperature of the eqi
ances of bivalent and quivalent ions, especially of the
which might be expected to have not far from the same siz
may be due, as has been suggested by Morgan and Kanol
large hydration of the bivalent ions.

With respect to the form of the temperature-conducts
be seen from an examination of the values of JAJdt ths
crease of conductance is in case of all the neutral di-ic
between ioo° and 156° than it is between 18" and 100°,
and 218°,' and, therefore, that the curve is first conve:
and then again convex toward the temperature axis, w:
diate points of inflexion.

In the case of adds and bases, however, and therefo
and hydroxide-ion, the rate of increase of the equival
steadily decreases with rising temperature, so that the
concave toward the temperature axis. With the tri-io
other hand, the rate of increase steadily increases, owinj
crease in the equivalent conductance of the bivalent ion ; tl
fore always convex toward the temperature axis.

It is of interest to note that the fluidity, or the recip
cosity, of water shows nearly the same increase as the co
di-ionic salts, at any rate up to 156°, which is about tb
previous determinations of the viscosity have extendi
for the viscodty (ij) the data of Thorpe and Rodger and
taking the mean values of A^ for the five uni-univaler
in this research, the product i^, has the values 1.19 at i
and i.oi at 156°. When it is considered that the cot
' This Journal, aS, 57* {1906).

' With respect to this last temperature-interval sodium acetate
* See lAndoIt-BSmstein-MeyerhoSer, Physikalisch-chemiscbe '
From tbe data tbeie given the viscosty in dynes per sq. cm. is fou
to be 0.01053 at 18°, 0,00283 at 'o°°. a"<i 0.001785 at 156°. The
for tbe salts referred to are 113 at 18°, 36931 100", and 566 at 156°; tl
were potassium, sodium and aminonium chlorides, sodium acetati

SAI/rS» ACIDS AND BASES.

343

increase five-fold, this variation in the ratio will be seen to be of secondary
significance.

With respect to the variation of the equivalent conductance (A) with
the concentration (Q, it has been found that between the concentra-
tions 0.1 and 0.002 or 0.0005 normal the results of all temperatures with
all the salts, both di-ionic and tri-ionic, and also with hydrochloric acid,
nitric acid, and sodium hydroxide, are expressed by the function C{Aq — A)
« K(CA)'* provided that to the exponent n a value (varying with the
different substances) between 1.40 and 1.55 is assigned. This is clearly
shown by the summary of the n values given in Table 4. These were

Tabub 4.— Valubs o9 rsB Exponent n in the Function C(Ao — A)

8vb8tancc. i89,

Ka 1.42

NaCl I

AgNO, I

NaCjHA 1

HNO,. . . .
NaOH. . . .
Ba(OH),.

Ba(NO0,.
MgSO,...

42
53
45
45
43
50

55
42
50
43

looP.

1.40
1.48
1.52

1.45
1.38
1.45
1.50
1.45

1.42
1.50

I
I
I
I
I
I
I
I
I
I

.50

.42

.40

.45

.50

.45

.42

.50

• m

50
50

1.50

1.47
1.52

iC(CA)«
306P.
1.48
1.46
1.52

42
50

1.42
1.50

derived by a graphical method (which involved no assumption in regard
to the value of -4o), this being regarded as a third constant to be deter-
mined from the data themselves. In general, the value of n could be
found within 0.02 or 0.03 tmits.

It is evident that, if the conductance-ratio A/A^ can be taken as a meas-
ure of the ionization (y), the latter changes with the concentration in the
case of all these substances in accordance with an entirely similar ex-

(CrY
ponential law, namely, in accordance with the ftmction r-/ \ =const.,

in which n has values varying with diflferent substances only between 1.40
and 1.55.

In a previous article* emphasis was laid on the remarkable fact that
at ordinary temperatures the form of the functional relation between
ionization and concentration is the same for salts of different ionic types.
These results show that this is also true at high temperatures, and, more-
over, that even the very large variation of temperature here involved
and the large consequent change in the character of the solvent affect

' Noyes, "The Physical Properties of Aqueous Salt Solutions in Relation to the
lomc Theory," Congress of Arts and Science, St. Louis Exposition, 4, 317 (1904);
Tedmology Quarterly, 17, 300 (1904); Science, 20, 582 (1904); abstract in Z. physik.
Chan., 52, 635.

344 ARTHUR A. NOYBS.

only slightly, if at all, the value of the exponent in this
ralation. Thus an additional confirmation is given tc
conclusion that the form of the concentration-function
of the number of ions into which the salt dissociates,
show almost conclusively that chemical mass-action ha:
influence in determining the equilibrium between the i
ionized part of largely dissociated substances. How co:
tradiction with the mass-action law is, is seen when it is
di-ionic and tri-ionic salts this law requires that the cont
un-ionized substance be proportional to the square and ci
of the concentration of the ions, while the experimental
it is proportional to the 3/2 power of that concentratioi
be the type of salt.

It has been found by trial that the functions J, -
Jo — A = K(CA)*, which contain only two arbitrary const
also satisfactorily express the results with potassium (
chloride, hydrochloric acid, and sodium hydroxide at any
between the concentrations of 0.1 and 0.002 or 0.0005 '
however, the data at still smaller concentrations, as
Kohlrausch and others at 18°, do not conform to the
these functions, they apparently do not give by extrap
value of Jq, and correspondingly the ratio A/A^ derived 1
a true measure of the ionization. It has therefore nc
while to make a study of the apphcability of these
the substances investigated.

The equivalent conductance and ionization of the sligt
stances, acetic acid and ammonium hydroxide, on the othi
with the concentration at all temperatures, even up to
ance with the mass-action law. It is interesting to note
add, an acid of moderate ionization (60 per cent, at 18° .
at 156° at o.oi normal' concentration), has intermediate 1
1.9), which, however, approach more nearly the theore
than the empirical one.

In order to show the relations between degree of ioniza
ter of the substances, and the temperature, the perce:
of all the substances investigated at the different temp
and o.oi normal solution is shown in Table 5. The su
ranged in the order in which the ionization at 18° decrea:
in the caes of sulphuric acid show the percentage of the
which exists in the form of hyrogen-ion, without referen
arises through the primary dissociation into H**^ and HSO
ary one into H+ and SOi"" ; the values are only approxima
an estimate of the relative extent to which these two sta

SALTS, ACIDS AND BASB5.

a s;a:>2 z wz s Bg'

C "■« S ■!? I t I; Ri« 2!.* o Q& S 2

^ P p p g p § Se p2 •" ; P -°

oeooeooooooooooooooooooooooooo j9

-■eb' Ok-.l>- o-~<- ■ 0 ■ ■«■ «■ .^©■^s■ ? g

ciat

rou{

and

risit
prin
het
up

acid
pen

equ
cen
o-ol

SALTS, ACmS AND BASBS. 347

Thus the rate of decrease in ionization is small between i8° and loo"
for either type of salt; but it t>econies greater at the higher temperatures,
especially in the case of the tri-ionic salts ; and for the highest tempera-
ture interval {28i°-3o6*) it is extremely rapid for both types of salt.
The decrease in ionization of hydrochloric add, nitric add up to 156°,
and sodium hydroxide is about the same as that of the di-ionic salts;
thus the average value of ( — io^AtIAI) at 0.08 normal for hydrochloric
and nitric acids is 0.38 between 18° and 100°, 0.63 between 100' and
156°; and for hydrochloric add 0.76 between 156° and 218°. Between
156° and 306° nitric add decreases in ionization much more than the
other substances of the same type.

The phjrsical property of the solvent which is most closely related to
its ionizing power is, as has been shown by Thomson and Nemst, its di-
electric constant. It is, therefore, of some interest to compare its varia-
tion with the temperature with that of the ionization of salts. Unfor-
tunately, the dielectric constant of water has been determined only be-
tween 0° and 76°. Drude' has, however, derived for this interval a quad-
ratic equation, from which a value at 100° may be calculated, probably
without great error. The values of the dielectric constant obtained from
this equation are 81.3 at 18° and 58.1 at 100°, and the ratios of these is
1.40.

The question now arises, what function of the ionization should be com-
pared with this? It seems clear that, from a theoretical standpoint, it is

c.(i-r^ ^

ized salt which prevail in solutions that at the two temperatures ((j and
y have the same concentration of the ions (that is, solutions for which
C,)-, = Cij-i); for in such solutions the electric force between the ions,
and therefore their tendency to unite to form un-ionized molecules, in
so far as this has an electrical origin, must be inversely proportional to
the dielectric constant. The above ratio is evidently equivalent, since

Crf-, = C,r, to the ratio- tt^* where, however, r, and r, refer to

"' i/i {1— rO/r»

the slightly different concentrations C, and C, (C, being equal to C,j;lj-j).
Now for the four uni-univalent salts pven in Table 5 the mean values
of the percentage ionization at 0.08 normal is 84.4 at 18° and 80.9 at 100°,
or hy interpolation, 80.6 at 100° at 0.08 X r.042 normal (that is, at
Cij-,/;-,), whence the value of the ratio just referred to is found to be
1.30. The value of the corresponding ratio for the two tri-ionic salts
at 0.08 normal is in the same way found to be 1.38,' While the former

' Wkd. Ann. Phys., 59, 50 (1896).

' The niran vslue of tbe percentage ionization for these two salts at 0.08 normal
■ic 71.7 at 18° and 65,6 at 100°, or by interpolation 64.8 at loo'ato.oS X 1.09 normal.

48 ARTHUR A. NOYBS.

f these values differs considerably from the ratio (1.40) of the dielectric
onstants, yet all the values lie in the same neighborhood. Indeed, tbe
greement is as dose as could be expected, considering the character of
tie data involved.
Finally, even though it seems theoretically to correspond to a less corn-
arable condition in the solution, yet, in view of the valence principle
edssdisc just below, it is of interest to note the values of the simpler

itio, -pr, :■■ of the concentration of the im-ionized substance at two

:mperatures at the same total concentration, instead of the sameion-
onceotration. At 0.08 the value of this ratio for 100°/ 18" is i-si for
tie four uni-imivalent, and 1.21 for the two uni-bivalent salts, thus con-
iderably less than the ratio of the dielectric constants.

The degree of ionization of the different substances may be next con-
idered in relation to the ionic type to which they belong and to their
tiemical nature. It has already been pointed out that even up to the
ighest temperatures neutral salts of tbe same ionic type have roughly
le same percentage ionization, the differences not exceeding 8 per cent.
I any case investigated. The strong acids, hydrochloric acid and (up
> 156°) nitric acid, and the strong bases, sodium and barium hydroxides,
Iso conform in a general way to this principle, though their ionization
*ms to be several per cent, greater than that of the corresponding
ilts; it is worthy of mention, however, that this greater value may be
ue to an increase in the equivalent conductance of the hydrogen-ion or
ydroxide-ion with the concentration of the solute, as is indicated to be
le case by the transference results which have been obtained with these
ads.'

It is also remarkable that the rough proportionahty which had pie-
iously been shown to exist at ordinary temperatures' between the iin-
inized fraction of a salt at any concentration and the product of the
Eilences of its ions has now been proved to persist up to the highest tem-
sratures, where the degree of ionization has become much less. This

shown by the following summary: Under A are given the mean values
: the percentage of un-ionized salt 100(1 — y), tor the neutral salts of each

B. '

B. A.

'a. "

17 SO

.11

•5

31 35

»5

.19

3»

.14

30 51

30 93

»3

' See This Journal.

• For a discussioii of this principle, see the author's article on "The Phjacal Prop-
ties of Aqueous Salt Solutioos," Loc. cH.

SALTS. ACIDS AND BASBS. 349

type at the concentration 0.04 molal and for the tini-univalent salts at
0.08 molal; and under B are given the ratios of these values to the product
of the valences (v{v^ of the ions.

It will be seen that the principle continues to hold, especially when
the comparison is made at the same equivalent concentration, even when
the ionization has become very small; thus it is only 26 per cent, for the
uni-bivalent salts at 306° and only 7 per cent, for the bi-bivalent salt
(magnesium sulphate at 218^).

The ionization tendencies of phosphoric add, acetic add, and ammo-
nium hydroxide, and the effect of temperature on them are best shown
by the summary of their ionization-constants which is given in Table
7.* The concentration involved in the constant is expressed in equiva-
lents per liter, and the constants themselves have been multiplied by lo*.

TaBLB 7. — ^lONIZATION-CONSTANTS OP PHOSPHORIC AciD, ACETiC Acn>, AND AlCMONmM

HVDROXmB.

Temperature. Phosphoric acid. A4

O ...

18 10400

25 9400

50 7000

75 4800

100 3400
135

128 2230

156 1420
318
306

It is evident from these results that the ionization-constant for ammo-
nium hydroxide increases considerably in passing from 0° to 18®, then
remains nearly constant up to 50°, and finally decreases with increasing
rapidity as higher temperatures are reached, attaining at 306^, a value
which is only about one two- hundredth of that at 18°; and that at all
temperatures the values for acetic acid are not very different from those
for ammonium hydroxide. Phosphoric acid is seen to have a much larger
ionization, which, however, decreases steadily and very rapidly with
rising temperature.

The interpretation of the results obtained with sulphuric acid is com-
plicated by the fact that the ionization doubtless takes place in two
stages; but in the original publication* a method has been described
which can only be referred to here, by which it is possible to determine

^ In the case of phosphoric add the values vary considerably with the concentra-
tion m correspondence with the fact that the exponent in the concent ration -function
was found to be 1.8- 1.9 instead of 2 as required by the mass-action law. The values
here given are those at the concentration 0.05 formula-weights (HgPO^) per liter.

* Publication No. 63 of the Carnegie Institution, p. 271.

ic acid.

Ammonium hydroxide.

• •

139

8.2

17.2

• •

18.0

• •

18. 1

• •

16.4

I.X

13-5

• •

10.4

• •

5.42

• •

6.28

Z.72

1.80

0.139

0.093

,50 ARTHUR A. KOYBS.

he hydrogen-ion concentration within fairly narrow limits from the con-
luctance alone, without knowledge of the extent to which the separate
tages occur. The method is of general application to dibasic adds;
nd, if the ionization-constant for the first hydrogen be known, as is true
rith many of the organic acids, the method could be used for computinf
hat of the second hydrogen from the conductance at high dilutions
irhere the second ionization is appreciable. The ratio of the hydrogen-
3n to the total hydrogen in the case of sulphuric acid is thus found to
ary in 0.08 normal solution from about 66 per cent, at 18° to 48 at 100°
nd 35 at 306".

Similar calculations of the hydrogen-ion concentrations have been
aade in the case of potassium hydrogen sulphate. These show tbat in
.1 molal solution, at 156°, the hydrogen-ion concentration is not mon
han 3 per cent.; and this justifies the conclusion that the secondar)-
inization of sulphuric acid (into hydrogen-ion and sulphate-ion) in its
wn moderately concentrated solutions is also insignificant at this tem-
lerature and higher temperatures. Interpreted with the help of this
onclusion the conductivity data for the acid show that the primary dis-
Dciation (into hydrogen-ion and hydrosulphate-ion) is about the same
s that of hydrochloric add at temperatures between 100° and 306"; and
: is reasonable to suppose that the same is true at lower temperatures
own to 18°.

With the help of this principle the ionization of the hydrosulphate-ion
t 18°, 100°, and 156° in the solutions both of the acid and acid salt
as been computed. This ionization is thus found to be large at 18°,
Tit it decreases very rapidly with the temperature. Thus in a o.i molal
otassium hydrogen sulphate solution equal quantities of sulphate-ion
nd hydrosulphate-ion are present at 18", while at 100° there is only 15
er cent., and at 156° only 4 percent., as much sulphate-ion as hydrosal-
hate-ion in the solution.

Only rough values of the ionization-constant of hydrosulphate-ion
ito hydrogen-ion and sulphate-ion can be given, since they vary very
luch with the concentration; some idea of its magnitude is fumished
y the following values which hold at about o.oi molal (or 0.002 molal
t 156°): 18500 X io~* at 18°, 1220 X 10"' at 100°, and 115 X 10"*
t 156°, whereas the ionization-constant of acetic acid at 18° is 18 X
o"". From the change of the ionization-constant with the temperature,
be heat absorbed (i£) by the reaction HSO^ = H+ -|- SO,= has been
>und to be given by the expression: ^B = 14,170 — 65 T, where TrepiF-
;nts the absolute temperature. From this it follows that the vahe
t 18" is —4750 calories, and at 100" — 10,070 calories, while from Thora-
m's heat-of-neutraUzation measurements and our ionization data at
8" the value — 5020 calories is derived.

SALTS, ACIDS AND BASES. 35 1

Conclusion.

In the preceding paragraphs have been summarized the generalizations
to be drawn from the results of these investigations, in regard to the be-
havior of the varous types of chemical substances in aqueous solutions
through a wide range of temperature. In conclusion, it seems, however,
desirable to draw attention again to a theoretical principle of even more
general import, which has been already presented in a previous article
by the author as a conclusion apparently justified by a study of the then
existing data, for this principle has now received a further confirmation
through the demonstration of the fact that certain purely empirical laws
relating to the ionization of salts in water still continue to be valid, even
when the physical condition of that solvent is greatly altered by a large
change in its temperature. This principle is that the ionization of salts,
strong. adds, and bases is a phenomenon primarily determined not by
specific chemical affinities, but by electrical forces arising from the charges
on the ions; that it is not affected (except in a secondary degree) by chem-
ical mass action, but is regulated by certain general, comparatively
simple laws, fairly well established empirically, but of unknown theoretical
significance; and that, therefore, it is a phenomenon quite distinct in al-
most all its respects from the phenomenon of dissociation ordinarily
exhibited by chemical substances, including that of the ionization of
weak adds and bases.

The most important facts leading to this conclusion are the approxi-
mate identity of the ionization values for salts of the same ionic t)rpe;
the existence of a simple approximate relation between the value of the
mi- ionized fraction and the product of the valences of the ions; the smal
effect of temperature on the ionization of salts and a parallelism between
the magnitude of that effect and the effect upon the dielectric constant
of water; the validity of an exponential relation between ionization and
concentration, which differs from that required by the mass action law, and
which is approximately the same at all temperatures and for different
ionic t)rpes of salts; and the fact that the optical properties and other
similar properties of dissolved salts (when referred to equal molal quan-
tities) is independent of this concentration and therefore of theif ioniza-
tion, so long as the solution is even moderately dilute.

The molecular explanation of these facts and the more general con-
clusions drawn from them would seem to be that primarily the ions are
united somewhat loosely in virtue of their electrical attraction to form
molecules, the constituents of which still retain their electric charges and
therefore, to a great extent, their characteristic power of producing optical
effects and such other effects as are not dependent on their existence as
separate aggregates. Secondly, the ions may imite in a more intimate
way to form ordinary uncharged molecules, whose constituents have com-

35* ARTHUR A. NOYSS.

pleteiy lost their identity and original characteristics. These two kinds
of molecules may be designated electrical molecules and chemical mok
axles, respectively, in correspondence with the character of the forces
which are assumed to give rise to them. Now in the case of salts and
most of the inorganic adds and bases, the tendency to form chemical
molecules is comparatively slight, so that the neutral electrical molecules
greatly predominate. On the other hand, in the case of most of tbe
organic acids, the tendency to form chemical molecules is very much
greater, so that as a rule these predominate. The facts, moreover, indi-
cate that chemical molecules are formed from the ions in accordance with
the principle of mass action,' but that electrical molecules are fonnedin
accordance with an entirely distinct principle, whose theoretical basis is
not understood.

It is to be expected that with neither class of substances will tjie pre-
dominating type of molecule be alone present ; and that minor deviations
from the mass action law in the case of moderately ionized substances,
and from the usual empirical law in the case of largely ionized substances,
may well arise from the presence of a small proportion of molecules of tbe
other type. In tbe former case, we may indeed with some confidence
predict quantitatively that that proportion of electrical molecules will
always be present which corresponds for the type of substance in ques-
tion to the concentration of its ions in the solution.

A fuller experimental investigation of the properties of dissolved salts,
especially of those of polyionic types, and of the phenomena of the solu-
bility effect and the distribution into a gaseous or another liquid phase
of ionizing substances, if combined with a thorough and persistent study
of all the available data, gives promise of suggesting a fuller theoretical
explanarion of this remarkable behavior of largely ionized substances in
aqueous solution. Even* if such a theoretical interpretation should not
be discovered, one may at least hope to determine with greater accuracy
and certainty the laws of the equilibrium between the ions and un-ionized
molecules, and between the two forms of the latter, in case their existence
shall be more fully substantiated. The facts already known make it
almost certain that we have here to deal with a new kind of equilibrium
phenomenon, and not simply with some deviation of a secondary nature,
' The best evidence of this is that furnished by the change of the conductanot o(
slightly ionized electrolytes with the concentration; but distribution experiments also
indicate it. Thus it is probable that as a rule the chemical molecules ajone distribute
into tbe gaseous phase or into organic solvents and that therefore the concentration of
the substance in such phases is a measure of the concentiation of those molecules in
the aqueous solution; and the few experiments thus far published indicate tiiat tbe
latter is at least approximately proportional to the product of the concentrations of tbe
ions. (Compare the experiments on picric add by Rothmund and Dracker, Z. pbysik.
Chem., 46, 836 (1903)).

REFRACTIVE INDICES OF AIXOHOL-WATER MIXTURES. 353

arising, for example, from a somewhat abnormal osmotic pressure, or a
change in the migration velocities of the ions, as has been assumed by
most authors.

In conclusion, I desire to express to the authorities of the Carnegie
Institution my great indebtedness for the assistance rendered me in the
prosecution of these researches; for without such aid little progress could
have been made up to the present time.

Boston, December. 1907.

[Contribution from ths Laboratory op the Mau^inckrodt Chemicai^ Works.]
THE REFRACTIVE INDICES OF ALCOHOL-WATER MIXTURES.'

Bt I^aukcblot W. Andrews.
Received January 15, 1908.

To Leach and Lythgoe belong the credit of having first determined,
by means of the Zeiss immersion refractometer, the refractive powers of
aqueous solutions of methyl and ethyl alcohols and of publishing the re-
sults' in tabular form for the entire range from zero per cent, to one him-
dred per cent, for the temperature of 20°. An earler table' by B. Wag-
ner comprises the range for ethyl alcohol only from zero to three hundred
and thirty, expressed in grams per liter.

The method used by the first-named authors for fixing the concentra-
tion of the solutions of which they observed the refraction, is not men-
tioned in their publication, but the inference is that they deduced the
concentrations from density determinations by means of Hehner's tables.

Since, in case of nearly absolute and of very strong alcohol, the refrac-
tometer and density constants bear such a relation to one another that
the concentration may be much more accurately inferred from the former
than from the latter, it follows, that the refraction constants should be
fixed independently of observations of specific gravity. For this and
other reasons, I decided to prepare absolute alcohol, and from this to
make, by dilution with known weights* of water, the solutions needed
for the ref ractometric work.

Preparation of the Absolute Alcohol.

Three methods came into consideration for the preparation of the abso-
lute alcohol required, viz.y the usual quick-lime process; the method
of Winkler* with metallic calcium, and the Evans and Fetsch* and Konek^

^Read before the American Chemical Society, January 2, 1908.

* This Journal, 27, 964 (1905).

* Dissertation, Jena (1903).

* The weighings were not reduced to vacuum.
•Bcr., 38,3612 (1905).

* This Journal, 26, 1158 (1904).
' Bcr., 39, 2263 (1906).

\*M

xmi

.*<

Sf+

Uil

»!«

laS»j

.lHv>

^Sif>i

• ■■:'«'

l^i^'l:

i J-

* >.

Pfe^i

r.>i--

;«4 .-•^-. '■.

CV-;

mi'

V4-

■fci

?aj

■~ifi.

•»

f'Si!'^:

i-s'i

\i:>^

354

LAUNCELOT W. ANDREWS.

process with magnesium amalgam. It was decided to operate by all
three methods and to compare the refractive indices of the products.

The plan contemplated the continuance or repetition of each process
until no further change occurred in the constants, and was carried out
in that way. Two criteria of complete hydration were made use of,
first, the constancy of the refractive index, which falls rapidly as the
last portions of the water are extracted, second, the constancy of the
critical temperature of solution. The latter method, devised by Cris-
mer,* cannot be top highly recommended. It consists, as is well known,
in cooling a mixture of equal volumes of the alcohol and of kerosene until
the mixture becomes turbid, a sign of the critical temperature having
been reached, and observation of the temperature. It was found ad-
visable to free the kerosene from its more volatile constituents by passmg
a current of steam through it for some time, after which it was dried.
This treatment obviated the appearance of a preliminary haze before
the critical temperature was reached, a phenomenon that slightly dimin-
ished the sharpness of the observations. The following table shows
the relation between the C. T. S. (critical temperature of solution) and
the percentage of alcohol, as found for the kerosene used by me. It
may be seen that if the temperature be read to o.i°, a difference of 0.005
per cent, in the amount of water present may be easily detected, a fact
previously demonstrated by Crismer. With proper precautions, a much
closer temperature determination may be secured, but is hardly neces-
sary.

Tabls I.

Per cent, of

C. T. S.

Diff. per cent

water.

alcohol.

kerosene.

for i« C.

0.0

100. 0

4.0»

0.050

0.2

99.8

8.o«>

0.4

99.6

ii.6«>

0.6

99.4

15.1**

0.8

99.2

i8.4*>

I.O

99.0

21.7®

0.061

0.3

98.8

24.9**

0.4

98.6

28.o<»

0.6

98.4

31.1**

0.8

98.2

34.0^

2.0

98.0

36.8<»

0.075

0.2

97.8

39.4**

0.4

97.6

42.0**

0.6

97.4

44. 4**

0.8

97.2

46.6®

O.IOO

3.0

97.0

48.6^

The raw material employed for the preparation was that commer-
cially known as ** double Cologne spirit." This was digested for three

* "Les temperatures critiques de dissolution," Bruxelles (1904).

RBPRACTIVB INDICBS OP AIXOHOL- WATER MIXTintBS. 355

reeks, being frequently stirred, with good quicklime, and then distilled,
he first and last tenths being rejected. To the intermediate fraction,
rbich contained about 99 per cent, of alcohol, silver nitrate was added'
1 the proportion of about five grams to each liter and the solu-
ion boiled for about eight hours, under reflux condensation and exclu-
ion of moist air. Then, after standing cold for two or three days, it
ras distilled. The distillate contained about 0.3 per cent, of vrater.
lo reaction for aldehydes could be obtained from it. Various portions
f this intennediate product were used for the further dehydration with,
rst, fresh lime from marble, second, turnings of metallic calcium, third,
lagnesitmi amalgam.

It soon appeared that a too prolonged treatment with either of the last-
amed reagents gave an alcohol which, on distillation, was less dry
[udged by the C. T. S.) than that secured by a briefer treatment. The
rue explanation of this deportment is uncertain, although it would be
ot difficult to suggest an hypothetical one. The proper period of diges-
ion is shown by the appearance of a pale yellowish tint throughout the
quid. When this point is distinctly reached, no matter which of the
Tying agents be used, if the alcohol is distilled with the usual precau-
ioDS,' the following phenomena were observed.

The first portion of the distillate contained a little water and had a
'- T. S. (with the standard kerosene referred to above) of about 6°. After
pproximately ten per cent, had distilled off, in each case the C. T. S.
ose to 4° or 4.15° and remained at that point till nearly all had come
ver. In some instances, especially when magnesium amalgam was
resent in the distilling vessel, the C. T. S. fell again toward the close of
be distillation. Hence, the middle portion alone was used for the de-
emiinations. It may be remarked, however, that the first fraction
tenth) was not distingiiishable in density from the middle one. In
everal instances, the intermediate fraction was subjected a second and

third time to the action of the same desiccating agent as before, but
a no case was it possible to reduce the C. T. S. below 4° or to obtain a
iroduct of which the density or refractive index pointed to greater dehy-
iration than that of the 4° materiaL

It was repeatedly observed that the alcohol dehydrated by prolonged

■ Winkler, Ber., 38, 361a (1905).

* The TEtnark of Winkkr {Loc. eit.), that the hygroscopic character of absolute
Icohol "has been exaggerated" should not be interpreted in the sense that any pre-
uifiocis to exclude moist air may safely be neglected. This author must have been
ivored with exceptionaUy dry atmospheric conditions, or the remark quoted would
ever have beeo made. I can fully confirm the observations of Crismer (Lcc. cit., p. 7)
D the rapid absorption of water by dry alcohol on exposure to humid air and have not
>een able under the conditions of my experiments to confirm the divergent observations
if Winkler.

3S6 LAUNCBLOT W. ANDRBWS.

contact with either calcium or with magnesium amalgam possessed a
peculiar foreign odor, which persisted even after dilution with a little
water. The nature of the substance giving this odor is entirely unknowL,
but its presence does not appreciably influence either of the constants
determined. Its amount is, therefore, probably, extremely minute.

The desiccation with calcium or with magnesium amalgam may bt
effected by contact at the ordinary temperature for several days or at
the boiling point of alcohol in half an hour or two hours. The ama^am
was prepared by agitating magnesium powder with its own weight of
mercury under the surface of 98 per cent, alcohol, acidified with a little
hydrochloric acid. When etched in this manner, the mercury soon flows
over the surface of the magnesium particles. The alcoholic add is poured
off and the magma washed with absolute alcohol. Other methods of
preparation were experimented ^th, but this was found most efGcknt
and has the advantage of rendering a greater amount of magneaum
available in proportion to the mercury.

Apparatus.

The refractometer used was a Zeiss instrument of the immersion type.
Its scale was subject to a small correction ( — 0.24°) which was detennined
by a series of observations on pure water at 25°. All the readings pie
sented in this paper have been so corrected,

The thermostat used in the work was kept at 25° + 0.6°. Its tempera
ture never varied more than 0.05° during an observation and very rarelj
by so much. Each sample of alcohol was observed at least five time;
after the temperature of the thermostat was attained, and the mean ol
the concordant readings taken. This mean was corrected for 25° by the
temperature coefficient given in Table II.

Two thermometers were used. One was divided in 1/5° and had divi
sions wide enough to enable 0.01° to be estimated with considerable ac
curacy. This thermometer was made by Geissler Nachfolger in 1880
of unknown glass. The second thermometer was made of Jena noraia
glass, divided into 1/10° and read to 0.01°. In Order to insure takiaj
the readings on a rising meniscus, in case of all definitive observations
the thermometer was removed from the thermostat, waved in the air s
moment to cool it about i ° and then replaced.

Fixed points were determined for both instruments at 0° and at thi
transition point of sodium sulphate (after Richards). The intenne
diate scale was calibrated by two mercury threads and the reading coi
reeled accordingly and reduced to the international or hydrogen scah
by the table of Marck.'

With these corrections applied, the two instruments agreed, on dired
comparison, for temperatures between 20" and 30° within the limits ot
' I.aiidoIt and Bemsteiii Ta.belleD, 44.

RB7RACTIVB INDICES OF ALCOHOL-WATBR HDCTURBS. 357

the observation errors. A third thermometer was used for some of the
density determinations. It was fused into one of the pycnometers. It
was compared with the others by filUng the pycnometer with mercur>'
and immersing the whole in a mercury bath containing the other instru-
ments and placed in the thermostat.

Two pycnometers were employed, one of about 50 cc. of the Sprengel
type, the other of 35 cc, with fused-in thermometer. The errors of a
dendty determination of alcohol amounted to about two units of the
fifth place of decimals. It is hardly possible to secure greater accuracy
on a liquid of this kind with pycnometers so small as this. The deter-
mination of density was not, however, a main object of this investiga-
tion.

To obtain a degree of precision in the measurement of this constant
commensurate with present means of obtaining pure absolute alcohol,
pycnometers or sinkers of not less than 250 cc. should be used.

In all these determinations, except as previously noted, correction to
vacuum was made for weights and substances weighed, and the errors
of the weights were corrected.

Experimental.

1. Absolute alcohol made by the quicklime method g^ve C. T. S. 3.9*.
Density ',C 0.785103, 0.785087, 0.785107. C. T. S. after each density
determination, 4.1°, 4-1°, 4.0°. Refractometer, mean of 7 observations,
85-30°. Another preparation of the same gave; refractometer readings,
mean of 5, 85.33°. C. T. S. 4-i5° Density 'J° 0.785111.

2. Absolute alcohol made by calcium method, digested at about 35°,
gave C. T. S. 4.o5^ Density «o 0.785091, 0.785112, 0.785108. Re-
fractometer, mean of ten readings, 85.29°. A second preparation, made
at the boiling-point of alcohol: C. T. S, 3,9°, Density '^t 0.785105.
A third preparation, made with an alcohol from another source, dried
in the same manner: C. T. S. 4.1°; refractometer, mean of 5, 85.32°.

3. Absolute alcohol made by cold magnedum amalgam process gave:
C T. S, 4.18°. Density 'Jo 0.785102, 0.785110, 0.785095; refractometer,
mean of six, 85,35°,

A second lot of the same, but digested hot: C. T. S. 4.03°. Density 'Jo
0.785091,0.785105,0.785106; refractometer, mean of five observations,
85.30°.

A third preparation, made like the last: C. T. S. 3.95°. Density "j"
0-785085; refractometer, 85.28°.

4. Total mean of all observations on absolute alcohol:

Density, 0-78510 + 0.00001, ^^°. Refractometer (weighted mean)
85-30° + 0.02 Z, 25°. Index of refraction (/<d) agamst air, 1,359408 +

0.00001, 25°.

i LAtlNCIiT.OT \V. ANlJRKWS.

j. I''ro'ii t!n' absolute alcohol, a series of dilutions with distilltd water
re m::u]u by n'oighinjr, at ititcr\-als of about two percent., txccpt in

neigliborliood of absolute iilcohol, where the iiiter\'als were closer,
ch of thest dilute alcohols was oIiser\-C(] with the rcfractoiTictvr, al
st five obs<.'rvatioiis beins taken at eaeh dilution. Subsequently
ther dilutions, about ten in all, weR- made from different prepara-
IS. These obser\'atioiis were reduced b}' a graphic method and in-
K-ndently by a series of <|nadratic equations, which were found to lit

various parts of the curve closely. The plottinjj served as a check
the iiiathcmatical work and to detect ckrical errors in the compuia-
IS. In this manner the scale readings and the refractive india-s givin
the followini' table were separately coni]>ntcd as they api>ear in Table

When completed, no individual point observed was found to differ

omi;ti;r hor Aqi eoib .^LCcmoLS,

.563
564
5.6s
5-65

^">-'

the first colum

REFRACTIVE INDICES OF ALCOHOL- WATER MIXTURES. 359

national (hydrogen) scale. The figures of the last column give approximately only the
differences of the scale readings between 20^ and 25^. They may be used for calcula-
ting the corrections when the temperature differs by one or two degrees from 25 °. The
readings presupposed adjustment to read 13.63° in water at 25°. Practically the use
of the table is restricted to alcohols stronger than 88 per cent, by weight.

from the corresponding value of the table by an amount larger than could
be accounted for by an error in the temperature of 0.03° and in the aver-
aged refractometer reading of 0.04°, and the mean difference was less
than half this amount. It is of course possible, and probable, that there
are constant errors of greater magnitude. In these, as in previous de-
terminations of the same sort, there is no direct proof that the alcohol
may not have contained traces of higher alcohols sufficient to affect
the results appreciably, though this seems imlikely. Further, it is not
known what may be the magnitude of the errors in the Zeiss table of the
relation of the refractometer degrees and the corresponding refractive
indices.

The observations show the existence of a maximum of the refractive
index at 20.7 per cent, of water. This corresponds very accurately,
perhaps by accident, with a hydrate of the composition represented by
the formula 3C2H8O.2HJO. It is very unlikely that the position of this
maximum is incorrectly located by a larger amount than 0.3 per cent.

The view has recently found expression that improvements in the
technique of making absolute alcohol have reduced the densities found
for this substance from the figures obtained by Mendel^eff to those of
recent authors. That this view is incorrect is shown by the fact that
Crismer obtained exactly the same density numbers as Mendel6eff, al-
though he used the critical temperature of solution method for deter-
mining when his alcohol was dry, and this is without doubt the most
searching criterion for the purpose yet employed.

The results of Mendel^ff and of Crismer, D ^J© = 0.78522 are given
in terms of the mercury-glass thermometer, those of Winkler, Konek,
Klason and Norlin, and of the present writer, in terms of the hydrogen
thermometer.

Since we do not know the expansion curve for the glass of the ther-
mometers employed by Mendel^eff or by Crismer, it is impossible to say
exactly what the difference is. We may, however, assume, with proba-
bly a very small error, that the correction to be applied at 25° is 0.11°
or — 0.000095 on the density. Appljdng this to Mendel^eff's figure, we
have for the density reduced to the international standard, 0.78512.
Morky* has recently recalculated Mendel^eff's table in part, reducing the
densities to the hydrogen temperature standard and he finds for abso-
lute alcohol, at ^J^, 0.78763. This corresponds to 0.78508 at 25°.

^ This Journal, a6^ 1 185.

J. LIVINGSTON R. MORGAN AND RESTON STEVENSON.

:hercfore appears that the absolute alcohol of Mendel^eff was just
re and free from water as that obtained by the most modem methods.

Summary.
is demonstrated that the absolute alcohol, prepared by the use of
ed marble and freed from aldehydes, has the same density, the same
live index and the same critical temperature of solution as that
has been dried by the use of magnesium amalgam or of metallic cal-

: observations of Crismer, to the effect that the critical tempera-

if solution of alcohol in kerosene is the best criterion of the dryness

olute alcohol, is fully confirmed.

iolute alcohol was found to have the following constants:

isity 'Jo 0.78510 + 0.00001.

>s immersion refractometer, 85.30" + 0.02 at 25" H.

ex of refraction (ji) against air, 1.35941 + 0.00001 at 25° H.

ractive power -"^ - = 0.45833; — j" = 0.45779-

able is presented of the refractive indices against air and of the re-
meter readint^ of aqueous alcohols for each per cent, of water
3 to 30, accompanied by an approximate table of temperature coeffi-
of refraction through the same range.

; existence is demonstrated of a maximum refractive index <rf
)I5 at 25° for the mixture containing 20.7 per cent, of water and
KY cent, of alcohol, a composition which very closely corresponds
lance or otherwise, with the formula 3C,H,0.2H,0- (calculated,
per cent.).
Loois, Missouri. ^__

jBunoNs FROU THB Havbhbybr LaboraTOribs op Columbia UNiVMisrrv>
No. 149.]

WEIGHT OF A FALLING DROP AHD TBE LA^ OF TATE. THE
DETERMINATION OF THE MOLECULAR WEIGHTS AND
CRITICAL TEMPERATURES OF LIQUIDS BY
THE AID OF DROP WEIGHTS.-

ReceiTed December j6. 1907.

Introduction. Object of the Investigation.
1864, Thomas Tate,' as the result of his experiments with water,
;nced the following laws :

ixtract from the Dissertation of Reston Stevenson. Our thanks ate due to
Higgins for kind asdstance in the latter part of the work.
'hil. Mag., 4th Ser., if, 176 (1S64). AJl other references to drop weight will
id In the bibliography of that subject at the end of this papa.

WEIGHT OF A FALUNG DROP AND THE LAWS OF TATE. 3^1

I. Other things being the same, the weight of a drop of liquid (falling
from a tube) is proportional to the diameter of the tube in which it is
formed.

II. The weight pf the drop is in proportion to the weight which would
be raised in that tube by capillary action.

III. The weight of a drop of liquid, other things being the same, is
diminished by an augmentation of temperature.

Tate's experiments were all made with thin-walled glass tubing, vary-
ing in diameter from o.i to 0.7 of an inch, the orifice in each case being
ground to "a sharp edge, so that the tube at the part in contact with the
liquid might be regarded as indefinitely thin." His weights were cal-
culated from the weight of from five to ten drops of liquid, which formed
at intervals of 40 seconds, and were collected in a weighed beaker.

Tate's Law, as we know it to-day, is supposed to be a summation of
the first two laws of Tate, but it must be said that it attributes to Tate
a meaning that he never indicated, and probably never intended. The
analytical expression of this faulty law is the familiar

where W is the weight of the falling drop, r the radius of the tube on which
it forms, and /• is the surface tension of the liquid. Of course, Tate's
second law shows drop weight to be proportional to surface tension,
for the weight of a liquid rising in a tube by capillary action is propor-
tional to surface tension; and his first law shows drop weight to be pro-
portional to the diameter (or radius) of the tube; but he did not even
imply that drop weight is equal to the product of the circumference of
contact into the surface tension. The real analytical expression of Tate's
first two laws, as he actually announced them, in place of the above,
should be

W = KiT-D,
where K^ is a constant, and D is the diameter of the tube ; or, when the
drops are all formed on the same tube (i. e., where D is constant),

K being a new constant.

The general result of the work of all other investigators since the time of
Tate, on the subject of drop weight, may be summed up best, perhaps,
in the words of Guye and Perrot (1903), viz. :

"The law of the proportionality of the weight of a drop to the diameter
of the tube is no more generally justified than that of the proportionality
of the weight to the surface tension."

"The laws of Tate are not general laws, and, even in the case of static
(slowly forming) drops, represent only a first approximation."

It will be seen from these conclusions that Guye and Perrot repudiate
not only the form of Tate's law as we know it to-day, but also his first

12 J. LIVINGSTON R. MORGAN AND RESTON STEVENSON.

ro laws in the form that he aanouaced them. It must be said, however,
lat no investigator has as yet fairly tested Tate's laws, for no one has as yet
actly reproduced Tate's conditions. Practically all the results thus
r obtained have been for drops forming on capillary, instead of on tbiii-
illed tubes; and the effect produced by the "sharp edge" of the drop-
ng tube, as described by Tate, has never been even approximately ap-
oached, except under such conditions that the results were obscured
! other factors (Ollivier, Antonow).

The object of this investigation, which was started By one of us, ox
■ATS ago, is to test the truth of Tate's law (and especially the second),

he originally stated them, more fairly and with greater accuracy than
is hitherto been done, reproducing his conditions in a way that others
ive failed to do, paying particular attention to the effect of the form of
e tip, and excluding those errors which are so apparent in the work

some of the previous investigators. And it was hoped that even if
ite's laws were found not to hold rigidly, it might still be possible to em-
oy the temperature coefficient of drop weight of any one liqiud, in a
rmula similar to that of Ramsay and Shields,' in place of their tempera-
re coefficient of surface tension, as a means of ascertaining molecular
;ight in the liquid state, and the critical temperature.
It may be said here, to anticipate, that the results of our work have
oven to be even better than we had hoped, for they have shown that
it only molecular weights in the liquid state and critical tempera-
res, can be calculated just as readily and accurately from the
mperature coefficient of drop weight, as from that of surface tenaon;
it also that the relative surface tensions of various liquids can be found
Dm drop weights, and that, thus found, they agree with those
termined by the capillary rise as well as do those by any of the
her methods, and almost as well as those for the same liquid by the
me method,' carried out by different observers. This relation to sur-
ce tension is true for the interpolated values of surface tendon, and
rther work, using the actual, experimental values, will probably only
ow the relation to be even more rig^d than this.

Apparatus and Method.
In order to avoid the complication which might be introduced by the
ccessive formation of several drops, we have measured, throughout
ir work, the volume of a single drop, for that method, under these condi-
>as, is far more accurate and deUcate than any weighing method.
Although, unUke Tate, we have used capillary tips upon which the
op forms, we have so constructed them that we might expect to ob-
in an effect amilar to that obtained by Tate with the "sharp edge"
' Zeit. f. phys. Qiem., ii, 431 (1893).

WEIGHT OF

i FAI,1.

: LA

of his thin-wallcd tube Apparently tht' effect of this
is to di-li-rit the area of tlie lilt)'.- upon which the drop can
prevent the Hi|uid rising upon the outer walls of the tube,
tip we have cmploy.'d in our !i;e;isure'.i!e;ili is sliow\i, in m
i(O'). both the bottom and bevd bei'is lii;:hly polished.

Obscn'ation of a tip of this form shows Lhat it beli:ives e.
a one as described by Tate, and tint the lower ed^'e of the
Tate's ''sharp edge," is the liniit of the area upo:i which tli
■provided, -of course, thai its diaun ter
drop of the liquid with tiie sr.^allest :i
a diot> on this lip dors iiol. uitthr ,;;/.
■£:ills o'i the tube ur (/ mi'^hl on nn ordi
obtained, duiins: the course of onr
(Antonow and OUivier).

"sharp edge"
hang, and to
The fonn of

^■tion,

1 Pi;

1 lhat of
1 droj). The

This effect:

xactly as such
bevel, just as
le drop harigs,
the maxiinum
liijiiid forming
Tl the hcvel or
IS also been
iveEti,i;itors

by the if

foreign substances, wliii

contaminate the liquid.

The coTiiplete appar

lus used ill our preliii

shown m section, in Tig.
I P IS a tnnslucent por-
ctlain scale, 55 centinie-
tirs long divided into
imlh Titters AU is a
capillan burette of such
a bore that i niillin:eter
contiins ^hout 0.0003 ^^•
This tul)e was carefull)' 1
calibrated with nicrcuiy,
and a cun-e prepared, I
from which the volume
between any two scale
readings could be found.
One end of this burette,
A. was connected by rub-
ber tubing to the rubber
compression bulb K, 'i'his
bulb was Ro arranged in
a scRW clamp that the
prissure upon it could be gr.iihi.'.lly it;',-K'.'.se
lute and delicate control over the ■iiii\e;;i;:il ■
larger tubing BC, which is the contiiur.ilio

.■ burette. The
.es through the

J. LIVINGSTON R. MORGAN AND SBSTON STEVENSON.

ber stopper R, and thus supports the dropping cup D. F is a dipper
ch can be raised, lowered, or swung around to any position by means of
rod G. From this the burette can be filled with liquid, and into it the
J from the tip O ultimately falls. The bottom of the cup D is covered
] a thin layer of the liquid, and the tube H, through which G passes, is
fed with filter paper, saturated with the liquid.

he object of this form of apparatus was to prevent evaporation of the
id of the drop, and to enable us to measure drop volumes at any de-
1 temperature, by immersing the entire apparatus to the point m b
iterbath.

efore making a measurement with this apparatus, the cup, dipper,
; and tip are thoroughly cleansed with chromic-sulphuric acid, water,
hoi, and ether, and dried by a current of air. The liquid is then placed
he cup and in the dipper, from which, after the stopper R is fastened
tly, the tube is filled to such an extent that the lower meniscus is just
ut to enter the tip 0 when the other end of the column (in the burette
; ,43) is at zero, or some point just below it. This point (the zero point)
len recorded, and the bulb very gradually compressed until the drop
led at the tip 0 falls off. The reading of the other end of the column,
he instant of fall, then enables one, knowing the zero-point, to find the
ime of the raaximimi drop that can form on the tip; we shall designate
as the pendant drop (P. D.}, . By drawing the liquid, that is left on the
back into the tube again, until the lower meniscus is once more just
at to enter the tip 0, it is possible to find the volume of the drop that
remained clinging to 0; this we shall call the dinging drop (C D.).
ttacting the volume of this from that of the pendant drop, we finally
the volume of the falling drop (F. D.).

xpeiiments with this preliminary apparatus showed the method to be
llent, but made apparent the fact that greater delicacy was desrable.
second, and final form of apparatus, as shown, in section, in Fig. 2,
mpiy a modification of the first. Here the dropping tube is sealed
a glass stopper, and the cup is provided with a wide rim to allow the
of mercury as a seal. An elastic band, passed from the hooks Q ovei
stopper, and between the two tubes, holds stopper and cup together
prevents the passage of either mercury or the water of the bath into
cup. To obtain a more delicate setting, in determining the zero-p(»nt,
1 is possible by observing the passage of the meniscus into the tip,
dropping tube, here, is constricted at S. and the lower meniscus, in all
ings, is held to a mark at that point.

no pieces of apparatus in this form were used, the burette in one case
e 2) holding approximately 0.000,08 cc, and the other (tube 3) 0.000,056

WEIGHT OF A FAI^LING DROP AND THB I.AWS OF TATB.

365

cc per millimeter.^ In order that a scale of the same length as before
might be used, these measuring tubes were bent in the form shown in Fig.
In tube 2 there were three small

bulbs blown in the first length of
the burette, while tube 3 had a single
bulb Vy with an approximate capac-
ity of 0.027 cc The use of a bulb
or bulbs enabled us to get the total
voltune of liquid necessary for a
drop, without an excessive length of
the tubing. For liquids forming
drops of large volume, the zero-
point must be above the bulb or
bulbs; for those giving smaller vol-
umes it must be below the single
bulb, or, in case there are three
bulbs, below one or more of them.

Wilh these pieces of apparaiuSy
only the volumes of the falling drops
were measured, for the results with the
first apparatus showed that, of the
three kinds of drops, they only were
related to surface tension. Our rea-
son for originally determining the
volumes of all three kinds of drops,
when Tate considered only the fall-
ing drop, was the suggestion of
Ostwald^ that the pendant drop from
a capillary tube would probably cor-
respond to falling drop from a thin-
walled tube, such as Tate used. Ex-
periment shows, however, that here,
also, the falling drop is the impor-
tant factor.

To measure the volume of the fall-
ing drop with this piece of appara-
tus, the zero-point is foimd, just as
before, by drawing the liquid back
into the burette, imtil, when the

* These tubes were calibrated with mercury at room temperature, and no cor-
rection in volume was made when they were used at higher temperattures, for the varia-
tions were found to be well within the experimental error.

* Hand- und Hilfsbuch zur ausfUhrung Physiko-chemischer Messtmgen. Leip-
°8» "893, pp. 300-301.

Fig. 2.

J, LIVINGSTON R, MORGAN AND KESTON STEVENSON.

r ii;i-:iiscTis is nt zero or jiist Ixlovv it, the lower iiuiiistiis is

ly at tlie ir,:irk in the euii^-tiictcd portion of Ihe tube S; llitti,
.-fori.-, the liLjiiiid is v.ry crdhiiily fore, d over iiiiiil tli;.' dro;: on
tip O falls off. No iViidiii.i; for the ;K;iuuit drop ih -U. .iittd,
lie liquid is at oiici' dritwii Ijaek lo llit- y ar!:, and. after Lillowinji-ufli-

ti'iw for tiraiiiage-, the |io.-!lioii of the upper ir.eiiiscus is olwrvtd
dilTtTeiice in voIlitiil- of llic l.nrJte tiilie, l;etwctii the zero-poial ;md
ittt-r iH)int, is then the voiii -.e .>f ihe drop thai h is fallen, 'flu.- ;iri.=^
on the nih!)er bidb in all cases must be increand icry gr.ulii.illy al
istaut when the drop is about to fall, for a sudden increase in prr'-sun
at tiuR' tends to increase the- volui;e of Ihe filing dro{),
was necessiiry before usiii,!; these delidte foru.s of apparatus to \i:o\i
usively Ihiit no evaporation talces pl.uv fro:^ t]w drop ;is it is for- i^;
seerlain tliis the tube w.is fillul with liipild..and (lie zero-point Loled

gradually the |>ressure on the bulb was increased until a larjie drov
;h not sufficiently large to Ml was fon.;ed ;tt <). .After .=l;.ii>i;;.g i;
:oHdilion for sewnd iiii:iutes, caa behig tak^ii, ;.s it nnist ahv.iys Ik
the apparatus was not j.irrvd or d lurbed. llie liijuid is dr,:Hii liac

the lower iiK*niscws is a'.;.:in at the . ,irl: at N. Ar.v tli ere;i;e in vo
of the liquid, fro;n that ori'^^in.dly o^;,erved, is tl.e.i (o be attriiiutii:
such tips as we have usid. to i.\;iporalion Iro;ii the drop. Evti: Oi
observations for this piiryw^e, however, showed that there \v..> ii
jration, for we invarial;ly found an jiK.riiiti. in the voiu:i.e of liie li|Ui(
id of a dccreusc] in other words, liqnid was always deposited upo;i tl:
ing drop, no mailer how ofl^n it w,'s for;;iid and drawn back. .\iii
ni!x?r of atte;i;i>ts to a\oid tins tiej.o-ition U]xjn the fonninir dro]
irti;Llly fdling the cup with glass bead^, siind, or liltcr p.'per, r.ioisleii'.

the ii([uid, and alao by the use- of a \ertically pbeed bundle of slio;
:1 tubes, each filled with the licjiud and presenting a n-.t-iiiscus of a|
in.ately the diameter of the drop itself, i! timj JKiiiid thai H could i
cd entirely by d, />eY(V;(i,^' the li.pud (h-jm tlw ,'xpLt tiiii-nl) ^is a joz. »^
\iUs oj the Clip D. Tliis fo;; can be i)i-oduccd \-evy n.-;idily by hwiii
;up in a wnterbalh (after the LippT.ifus lus been set up and fiUw
to 20°. In this way nihnite drops of the liquid are deposited iipc
rails of the cup, and ehanjje the condition within, so that there isth*
er evaporation from the lunging drop, nor deposition upon it, and tl
r nienisciis always returns to the sa::e point, no mr.tter how ofti
Irop may be formed ami drawn bach, Ik'forc each nieasureiri'i
ssured ouimIvis, in this way, lli.'t such a conciition was attaintd.
will be seen that the delicacy of tliis method depends simply upon tl
if the caiiillary tubing uv. das the imn He. Tube ,^ (i mm. - o.ooo.o.
ivas Ihe smalKst tubing available ■i\. tbe tii..e, except, of coursi-, tl

WEIGHT OF A FALLING DROP AND THE LAWS OF TATE. 367

very narrow thermometer tubing, which offered too great a resistance to
the flow of liquids, for our purposes.^

In all cases the apparatus was immersed in a waterbath with transparent
sides, the temperature of which was kept constant to the point within o. i ^.

Results.

In Tables I, II, and III are given our results for drop volumes and drop
weights, and the relation observed between drop weight and surface ten-

Tabl9 I.
Diameter of tip » 0.622 cm. approximately, i mm. on burtte — 0.0003 ec.

Surface Weight of drop ^ ?
tension, Weights of drop in mgs. Surface tension 7
dynes per cm. > > > ■ .

Sobstaace. Temp. 7. Wp.d. Wf.d. Wc.d. Kp.d. Kp.d. Kcd.

Ether 20.0 16.80 34.6 21.4 13.2 2.06 1.27 0.79

Benzene 22.5 29.38 56.0 35.2 20.6 1.91 1.21 0.71

Ethyl iodide 19. i 30.00 53.2 36.1 17.2 1.77 1.20 0.57

Chlorbenzene 20.0 32.10 66.0 41.4 24.6 2.05 1.30 0.77

Goaiacole* 19.6 37.35 78.5 50.0 28.4 2.10 [1.34] 0.76

Benzaldeihyde 15.4 39.19 77.3 49.8 27.6 1.97 1.27 0.71

Aniline 17.5 44.10 81.8 52.9 28.7 1.86 1.21 0.65

QuinoHne 15.4 45.13 86.6 57.0 29.6 1.92 1.26 0.65

Water 20.0 70.60 127.1 89.1 37.5 1.80 1.26 0.53

Average Kp.D. ■■ i.248±o.oi2
Mean error of a single result — ±0.035

Tabls II.

Diameter of tip » 0.62 cm. approximately, i mm. on burette « 0.00,008 cc.

Weight of Surface ten-
Volume of falUngdrop sion, dynes w
falling drop. Specific *""«^»- percm. Kf.d.= ^^^^^"
Substance. Temp. cc. gravity. Wf.d. 7 7

Benzene 30.5 0.03880 0.867 33-64 26.58 1.260

60.7 0.03420 0.833 28.50 22.77 I. 251

Chlorbenzene... 28.5 0.03561 1.098 39-io 31.02 1.261

... 65.0 0.03220 1.058 34*07 26.91 1.266

Aniline 27.8 0.05033 1.013 50.99 40.69 1.250

58.2 0.04675 0.982 4591 3732 1.230

Qoinoline 28.0 0.04912 1.091 53-58 42.30 1.265

65.0 0.04572 1.060 48.47 38.22 1.268

Water 25.5 0.08812 0.9969 87.85 69.70 1.260

56.9 0.08180 0.9848 80.55 64.79 1-244

79.2 0.07742 0.9722 75.20 60.97 . 1.233

Average Kp.D. - i.253±o 004

Mean error of a single restilt -« ±0.013

^ It has since been possible to obtain still smaller tubing, and the work is now
being continued in this laboratory with a burette on which i millimeter corresponds
to abont 0.000,046 cc.

> Commercial and impure; omitted in computing the average.

J. LIVINGSTON R. MORGAN AND RSSTON STBVBNSON.

i'i.{'Sru-%0'°>'

s .
Si"

In^p 'Doiins) 3»iine '

!r»!ss.s.si lis

O. ""Ai ■«SiQ'do.p8nnnijjoma(»Al "

)o»hoo"-0"-

»iiH[Bjjo imniOA '

i'llf If^

9 S 3 3 0

WEIGHT OF A FALLING DROP AND THE LAWS OF TATE. 369

sion, together with the data necessary for the calculations. The surface
tensions given, except those for water, are interpolated from the results —
determined under the same conditions as our drop weights, i. e., against
saturated air — of Renard and Guye,^ those for water being interpolated
from the results of Ramsay and Shields,* against the vapor pressure of the
liquid. Kp u , Kpu , and K^^^ , in Table I, and Kp jj in Tables II and III,
are the factors by which the surface tension in dynes must be multiplied
to give the drop weight, in milligrams, from these tips. Kp ^ is the con-
stant already mentioned in the real analytical expression of Tate's laws,
when the same tip is employed. The tips used in Tables I, II, and III,
although made from the same tubing, have slightly different diameters
exposed, owing to the bevels being cut at slightly different angles. The
diameter of the tubing itself was about 6.5 millimeters.

Table I shows our reason for determining the weight of only the falling
drop with the more delicate form of apparatus.

In order that errors in our interpolations of the values of surface ten-
sion, as well as possible errors in the surface tensions themselves, might
not influence our conclusions as to the accuracy of Tate's laws, in Table
III, where the determinations are the most accurate, we have also secured
a check, without any direct comparison with surface tension, by substi-
tuting our drop weights, of the same liquid at two different temperatures,
for the surface tensions in the well-known law of Ramsay and Shields,
and then comparing the constancy, for the various liquids, of our constant,
^temp. ^^^ that of those of Ramsay and Shields, (k^ g^ s.) ^^^ Renard and
Guye (k^ g^ q). In other words, for 7- in dynes, in the relation

«=A? = 2.i2 ergs,*

we have substituted W^pu. in milligrams, so that if surface tension (as
altered by temperature) and falling drop weight are proportional, for
any one liquid from the same tip, we should find the expression

w;4^.r-w;.4f

y-^~ir == fetcmp.

just as constant as the other for all so-called *' non-associated" liquids.

All our densities are interpolated from results found in the literature,
as were also those of both Ramsay and Shields, and Renard and Guye,
so that uniformity in the compared results is thus secured.

All chemicals, with the exception of guaiacol (Table I), which was im-
pure, were specially purified for the purpose.

* J. chim. phys., 5, 81 (1907).

* Z. physik. Chem., 12, 431 (1893).

' M is here the molecular weight as a liquid, d the density, and t the temperature.

370

J. LIVIKGSTON R. MORGAN AND RESTON STBVENSON.

Our drop volumes throughout are each the average of several det
tninations, the extreme variation in tube 3 (Table III) being a2-a4 ;
cent.

Tabus IV.
Drop weights for various tip diameters. ( — 27°.

Wp.D

Wp.D.

-<.6»mia

D,"6.Ji

D,= T.ii.

Bj-

Dt.

afi.io

34.60

39- '5

5-577

3-563

5-49«

*9.70

40.40

45.10

6.348

6.495

6.190

41 15

55 00

63.40

8.792

8.843

8.76+

Oilorbenzene 29.70

Quinoline. .

In Table IV are given the drop weights issuing from beveled tips
various diameters. These results are not as accurate as some of the oUk
for tube 2 was used as the burette, and the error in measuring the loi
end of the bevel is necessarily large. Under K' are the values of the a
stant of Tate's first law, t. e., weight of falUng drop divided by the dian
ter of the tip.

Tabub V.
/ WrD \
Kp.D. I — 1 tor tips' of various forms.

Temp. B In Pig. 3. Temp. C In Pig. i

Chlorbenzene 64 . o

23-5

Quinoliae. . . 64.0

934

933
965
124

080

Average 0.98340.025 1,135 ±0.017 i-i45±<»-<»*

Table V gives the results obtained by use of tips of various forms, I
of approximately the same diameter (see Fig, 3}. Tip A , here, is rounc
at the end, B has a bevel at an angle of about 30°, not sufficient tohj

A B

Fig. 3.

> AppFRximately of same diameter.

WEIGHT OV A FALLING DROP AND TlIi: LAWS OF TATE. 37 1

the c-lToct of a sharp cilc^v. and (', without Lvvil. has a wry shaq) edge.
All l!:.>e wero measured in tube 2. and consequently the determinations
are not as accurate as those in Trh^Ie III.

Table VI — Critical Tkmperatures.^

From /M\'-'/-

Ar •-; 7 1 1 ' k(r-cl).

From
Su:.j.tar:ce. '^^''^"^ ^ "a * " "~ ^^*^"^T>. ^^"^^ r. & q.^ R. .S: S.-^ Observed.

Benzene 2S6.6 2S5.8-2S9.6 2S8 280.6-2964

Chlorlx'nzcne 354 . i 357 . 2-358 . 4 359 . 7 360 . 0-362 . 2

Pyridine 35^o 34^.7-346.9 342

Aniline 4.^9-4 44^- 1-449 -i 404-9 425-7

Q'.iinolinc 49-3 49 S • 6-496. 9 466. i <r520

And. finally, in Table VI, are th" critical ten';j)eratures of the liquids
in T.'ble III, a.s c.dculated ])y tlie substitution of the drop weight, U'p j^ ,
and /m^,„„ for the surfiice tension r, and k in the Ramsay and Shields rela-

lion.

(-;;)■■""<-).

when- T is the dilTer nee between the critical tetiijierature and that of ob-
ser\'ation, and A/, d and k have tb»e sa.iie n ec.niiijj: as before.

Discussion of Results.

It will Ix- seen, even from Til^le I, where tlie cx]j(^rimental error in drop
wc!i;ht is comi)arativf'ly larf^e, that .conlrr.ry to the conclusion of Guye
and l-i-rrot, the reli lions! ij) Ix-lwee!! drop Wv i.i;ht . jrom a properly con-
strudi'd tip, and sm'face tension i:i ^-^luraUd rir,^ is very n.uch irore than a
first i'ppro::i'rution. e\'. .1 wh».'n tlie l!cj\iids exa.n.in'..d include that Rivin^^
111:- Iii.^liv-t, and that '^iviw^j, ali o^l: the l()V\vst, Mirf.ice tei:sion known,
i. r., wati-r at 70.6 and elhi r at i().S dy;..s p^r cv..Ai'i etcr.

Thf results in Table II i.-iUl- ti^is conclusion cm n n.'on' strikin.s:, for
th^ V show th'it uiuch of tiie vari -lion in I is due to exjjvriniental error.
And. firully. Table JII, wherr tlK^ a.fXT.racy in the drtc rnination of drop
vo1u:t\' i]'\d drop weii^ht was the 'O'-'trst ro -^Ible at the tiijX\ shows the
w.rii^ioi in the constant nl :t"c); ^ ^'p, for !0\.' of the same lifjuids ex-
ar/inid i'l I and II, to be v^ry s;' . 11 !idi.\d. ll^iv, witli five Hquids,^
varyi:'^ iri surface t(«ision fro." 2s. '*S to 52. (-.2 dvn«.s, each bein^ studied

' Here, in all cases, the tenipcraturc coc'l'cicnt (k or ktcmp) used is the one found
for ij'.e specific liquid, and not the averac;e values.

* Cak^ulated extremes from surface tensions.
M",iven by Kanisa\' and v^liiekls, Loc, c I.

' A coord in ;^' to Ivcnard and Gu\e, siirr.u^e teii>ions in salnrnted air and those un-
dfr the w.vor pres-iure rif tlie H<iuid do vn\ diijcr l.y more than 0.5 per cent.

* rnfiTlunciteiy, ether could not Ijc u-cd in either tuhe 2 or tube 3, owin.i; to in-
tvifcrcnce of a bulb; and the vohune of tube 3 \v:is too small to j)ermil water to be
\i^:\ with tile In-veled tip.

373 ]. LIVINGSTON R. MORGAN AND RBSTON STBVBNSON.

at two temperatures, the mean value of K, „ for all cases, from a certain
tip, is 1.226 + o.ooj6, the mean error of a single result being +0.0083
Although in these results the error is small, the discrepancy is still t«
great — granting the accuracy of the drop weights and surface tensions— ti
conclude that the proportionality is rigidly exact; even though the agree
ment is about as good as that observed in results for surface tenaons b;
different methods, and little worse than that shown in the results by an;
one method, by different observers. The error in drop weight canno
in any case exceed 0.4 per cent., taking all things into consideration, ani
is generally much less, consequently the discrepancy is only to be explainei
either by errors in the interpolated surface tensions, or by actual failur
of the law of proportionality to hold closer than this (due possibly to a ver
slight and variable, but unooticeable, rise of the liquids on the walls 0
the tip). When it is remembered, however, that the interpolations of th
values for surface tension were made from smoothed curves, which coul
not always be made to pass through all the few points available, it become
very apparent that in some cases errors in our interpolated surface tension
even as high as one per cent., are quite possible. If this be true, the lai
of the proportionahty between falling drop weight (from a proper tip
and surface tension becomes rigid. To prove this directly and conclusivel
has been impossible, for it could be done only by aid of a more delicat
apparatus, with measurements of drop weights at the exact temperatuie
at which the surface tensions themselves have been determined.' Below
however, it is shown that the interpolated values of surface tension jo
2ny one liquid are burdened with error, so that analogy would force th
concludon that they, also, are at the root of the error when different liquid
are considered.

We would conclude, then, from Tables I, II, and III, and from the bt
havior of tip C in Table V, that Tate's second law — ^the weight of a fallin
drop (from a proper tip) is proportional to the surface tension (agaim
saturated air) of the liquid — is true. Because surface tensions calculate
from drop weights agree, even with those possibly faultily interpolate
from results by capillary rise, as well as those determined by other met!
ids agree with these, when directly determined.

Consideration of the columns it„„p, fen ^ q , and fes as.- "• Table III, sboii
that our constants, though calcu^ted from results at only two tempcrc
tares, arc as constant as those of Renard and Guye, which are in eac
ase the mean of determinations made at several pairs of temperature;
ind are very much more constant than those of Ramsay and Shields,' froi

' This is now being done in this laboratory.

' Although Ramsay and Shields's values were calculated from surface tensions ol
MTved under different conditions, their constants are still to be compared witb tl
rtbers as to constancy.

WEIGHT OF A FALLING DROP AND THE LAWS OF TATB. 373

results at two temperatures. It will also be observed that the variation
of i^iemp. ^rom its mean value is always (when worth considering) in the same
direction as that of Renard and Guye's, for the same liquid.

This certainly proves conclusively thaty with any one liquid^ from any one
tip, drop weight is proportional to the surface tension, as it is altered by changes
in temperature, for, by substitution of drop weight for surface tension in the
Ramsay and Shields expression, leaving out any direct comparison with
the interpolated values of surface tensions, a result is obtained which is as
constant as that found by the use of directly determined — not interpolated —
surface tensions. And this is true when our interpolated values of surface
tension at the two temperatures lead to a discrepancy in the two volumes of
Kyj^, as calculated for that liquid. Although this proof is not direct, as far
as concerns different liquids, it leaves very little possibility of the slight dis-
crepancy in K^^ being due to anything bui the errors in the interpolated sur-
face tensions as we concluded above.
We would conclude from the constancy of ^tcmp.» ^ Table III, then:
That Tate's third law — the weight of a falling drop decreases with in-
creased temperatiue — ^is true. And, further, that the change in drop
weight for a change in temperature can be calculated accurately for non-
associated liquids, by the substitution of the drop weight at one tempera-
ture for the surface tension, and ^temp. ^^^ ^ ^ ^^^ Ramsay and Shields
relation

and solving for the other drop weight.

Or, knowing the drop weights, fetcmp.» ^°^ ^^^ densities, it is possible to
find the molecular weight of the liquid, with an accuracy equal to that
attained when surface tensions are employed directly in the above rela-
tion.

Since the molecular temperature coefficient, fetcmp.» ^ found to be constant,
it is possible, by extrapolation, to find the temperature at which the drop
weight would become zero; i. e., the critical temperature of the liquid, for at
that point the drop would disappear, there being then no distinction be-
tween the gas and the liquid. It is only necessary, for this calcula-
tion, to substitute W^^^ for 7- and ibtcmp. ^^^ ^» ^ ^^^ other form of the Ram-
say and Shields relation, i. e.,

,(^)'/.., (,_«),

and solve for the critical temperature (t plus the temperature at which ^
(or W^^) is determined). (See Table VI.)

It must be remembered here, however, that in all cases in which we
have applied this method, we have done so at a disadvantage, for we have but

374 J- LIVINGSTON R. MORGAN AND RHSTON STEVENSON'.

two j>oints through which to (h;iw lln' curvt-, IHirlhi r than llmt hv Ii.a
worked at low teiii]H.-r,iluris liK-\i'r siUiw Hc/'K iiini conK'qiit'.jtIy [■iv
cxtritpolatf from tluM' two points lhroiis;h a much j;rt.;'tir dist:i!.a- ih.
either Renard and Gii\r, or Rninsay :iTid Shiiids. from Ihi-ir tarstrnuir.K
The first objection holds for ;i!l our liqiiid^^. lhoii;;h k\i< for htr.znii, \i\
the second hardly atTccts bciizi'ni-, for 7-(.3'' is not far from it;. l)oili;^i!-|K)i::
With all high-hoiling li<|iiids, liolli o'ljiclions lioW. and both iiK-rww wi;
the boiling point (and criiic.il IfmiK-raliiK-).

From the fqtml com/a lu-y cj l;,„„^ iiml I.-, lur.i.^rr. ■! /^ cridail ihaljufl:
aocuratt- critical hmj'i r-iliins can be c:r!cii!ali-d jrom drop -.miilit^ m ji
sutjace tensions. (ii;aiv\f \i>liiralid air. pnKiiid in l>«lh cturs fl:,- dilcrmnalMi
jrom which the moUcuiir tf.npnolttrv C'lrjjicii-nls are jimnd. air yi^ni-
aswiny tempera I lira, nnd carried tn ax liieh a fcnipiraliiir.

Table IV, it is thonght, ^hows that froi-: such tips. Ix'twoi-nlhtsi'.ii.n'
eters, tht-rc is a dirocl ])roportioTiality b.'twuu dro]) weight and dia'ul'
of the tip (Talc's finl lau). Al li-„^.t tlur<.> is no dc.-i.kd trend i:i tlic rr
portional factor, for it varits just as -iin- 'ii;,'lit expect it to from the know
and fairly large, cxjiorinifiital iTror, it !i;ust be rL-membcrcd that ti]
larger than the dianu'ttT of the ni;r:i :.\i;'i drop woidd always deliver pi
constant maximum dro|) vv<i|>'ht; wh:!i\ when thu tip becomes sinall. tin
is probably a point beyond whicli the drop will not dc-crease aiiprcrial
in weight for a considcraljle chaii.^'e in ilii:E'eter, for it would then !
difficult to prevent in any way the ri>e of li^.jnid upon w:it!s of the tip.

Table V shows that when rounded, a (i;) b'.h.ivesdifTereiitly from the oi
in Table III; the liijiiid lisi's to v.'.rions Ik l-lits on liie outer walls, aiid l[
diameter of the basis for the drop vari's with tin- nature of the liiu);
This is also true, though to a Us;-erde.L;r,e. wilh the lube that is ir.suf
ciently bc'veled. In neither aise is A',,,, even a]'proNiii rtely con-rrii
Tip C, on the other hand, compares \X'V\' favo!-abIy wilh the other 6.-:v,'i
one, used with Tube 2 (Table II). \\ b.^lcwr tlk-ory m:'y la- adv.iiici
then, as to the lip, it will iie seen that l!ie point to be ifinsi<Kivd i> tl
effect of the tip (Tale's "sharn edge") in delivi.itin? the. purliou a|(
which the drop can hang. cspi.ei:ii;y by preventing the ri^e of li^aiil i:|X
the walls, for that would be vari:)i,'le with different h'piids. and k-ui
variable weights. Undoubtedly it is only tlie fiihire to follow Titf
directions in this respect Ihiit has can-ni! the d. ler(ninaiio;r. of ilr:
weights, since his litre, to negUive his coneluMons.
Summary.

The results of this inve-.tig:il ^m ii'i-\' Ix- sinin.iirized as follows:

I. Anapj)iiratnsisdi.'ser;iul by whieh i! is j o-.ibk' lo iri^kc a verv :'«
rale estimation of the volun e of j siu'^'lr drop of li.juid fill' ■■ fnr.i) ;i n:'-
and consequently of its wei.ghl.

WEIGHT OF A FALLING DROP AND THE LAWS OF TATE. 375

2. With this apparatus was used a capillary tip," beveled at an angle
of 45°, which, contrary to those used by other investigators, had the same
effect as the one originally used by Tate, i, e., it delimits the area of the tip
wetted, by preventing the rise of liquid upon the walls, and thus forces
all liquids to drop from one and the same area.

3. It is shown that whenever this effect is obtained, either by use of a
properly beveled tube, or one ground to a sharp edge, the drop weight
has a different meaning than 'it has when the drop is formed on either a
rounded tip, or on one insufficiently beveled.

4. The falling drop from a capillary tip, and not the pendant drop, is
proportional in weight to that of the falling drop from a thin-walled
tube with a sharp edge.

5. From such tips as we have used, it is concluded that Tate's second
law — ^the weight of a drop, other things being the same, is proportional
to the surface tension (against saturated air) of the liquid — ^is true.

6. It is shown that from such a tip, Tate's third law — the weight of a
drop is decreased by an increase in temperature — ^is true.

7. Falling drop weights for the same liquid at two temperatures, from
such a tip, can be substituted for the surface tensions in the relation of
Ramsay and Shields, and molecular weights in the liquid state calculated
with an accuracy equal to that possible by aid of surface tensions, under
the same, saturated air, conditions. And, by aid of this formula, know-
ing the molecular weight of a non-associated liquid^ the falling drop weight
at one temperature, and the densities, it is possible to calculate the weight
of the drop falling from the same tip at another temperature.

8. Critical temperatures can be calculated by aid of Ramsay and Shields's

equation y(-i ) = k(T — 6), by substituting a drop weight for surface

tension, and the molecular temperature coefficient of drop weight for k, with
the same accuracy attained by the use of surface tensions (against saturated
air), provided the drop weights (from which the coefficient is found) are
determined at as many temperatures, and at as high a temperature as
the surface tensions.

9. For beveled tips, when the diameters lie between 4.68 and 7.12 mm.,
Tate's first law — ^the drop weight of any one liquid is proportional, imder
like conditions, to the diameter of the dropping tube — ^is true.

BIBLIOGRAPHY OF DROP WEIGHT— ALPHABETICALLY ARRANGED.

AutoQow, G. N. J. chim. phys., 5, 372 (1907).

BoDe, J. Geneva Dissertation, 1902.

Doclaux. Ann. chim. phys., 4th ser., 21, 386 (1870).

IHiprt. Ibid,, g, 345 (1866).

Eschbaum, F. Ber. pharm. Ges., Heft 4, 1900.

Gug&elmo, G. Accad. Lincei Atti., 12, 462 (1904); 15, 287 (1906).

Guthrie. Proc. Roy. Soc., 13, 444 (1864).

Gnye and Pcrrot. Arch. sden. phys. et naturelle, 4th ser., 11, 225 (1901); 4th

ser., 15, 312 (1903).

176 L. w. McCay.

lagen. Berl. Alcad., 78, 1845.

lannay, J. B. Proc. Roy. Soc. Edin., 437, 1905.

tohlrausch, F. Ann. phys., ao, 798 (1906); 12, 191 (1907).

^baigue. J. phaim. chim., 7, 87 (1SG8).

.educ and Larcdote. J. phys., i, 364 and 716 {1902).

C. r., 134, 589; 13s, 95 and 732 (1901).
^hnstdn, F. Ann. phys., ao, 237 and 606; 31, 1030 (1906); aa, 737 ('9°7}-

Jathieu, J. phys. [2], 3, 203 (1884).

)llivier. Ann, chitn, phys.. 8th ser., 10, 229 (1907).

Uyldgh. PhiLMag., 5th ser., ao, 321 (1899).

losaet. Bull. soc. chim., 33, 245 {1900).

fate. T. Phil. Mag., a?, 176 {'864).

rraube. J. pr. Chem. fa], 34, 292 and 515 (1886).

Ber.. 19, 874(1886),
^olfcmann. P. Ann. physik. (a), it, ao6. '

Vonhington. Proc. Roy. Soc., 3a, 362 (1881).

PhiLMag., 5th ser., 18, 461 {1884); 19, 46 (1885); ao, ji
(1885),

Labor A TO sv OF Fhtstcai. CaBMiSTHv.

rHE ACnON OF HYDROGEN SDLPHTOE 05 ALKALINE SOLUTIOIfS
OF ZmC SALTS.

The fact that the zinc sulphide, or the zinc hydrosulpbide, precipitatet
rom alkaline solutions of zinc salts by sodium or potassium hydrosulphidt
s soluble in an excess of these reagents, and that the zinc sulphide, or zin
lydrosulphide, precipitated from alkaline solutions of the metal by hydro
^n sulphide dissolves when the ^s is permitted to act on the solution
or some time, appears to have escaped the notice of the analytical chemists.
Vt all events, in no work on analytical chemistry to which I have acces
s this remarkable behavior of zinc sulphide referred to. The solution 0
he zinc sulphide is a colloidal one, for the zinc in it will not pass througi
)archment paper. The zinc sulphide, or zinc hydrosulpbide. acts toward
odium and potassium hydrosulphides in much the same way that zii)'
ixide, or zinc hydroxide, acts towards sodium and potassium hydroxides
The analogy between the two reactions almost compels one to condudi
hat the change takes place in the sense of the equation:
Zn(SH)j + 2RSH = R,ZnS, + 2H,S.

If, however, an alkali sulphozincate is actually formed, it must be ver
mstable, for concentrated solutions of mineral salts, when added to it
olution, precipitate only zinc sulphide, or possibly zinc hydrosulpbide

' I noticed this peculiar behavior ol' zinc sulphide some three or four yean ag
nd supposed that my observation wns a new one. I found, however, that the rea<
ion was first observed by Julius Tbomsen in 1878 (Ber,, 11, 2044) and subsequent!
xamined by A. VUliers (Compt. rend., lao, 97). Lottermoser (Samlung chem. 1
ech. Vortrage, VI) und Winsinger (Butl. de I'Acad. des Sciences de Bnixelles, [2
4, 321) also refer to it.

ZINC SALTS. * 377

and then too the zinc sulphide separates out gradually, and in a slimy
condition, when the solution is allowed to stand.

In many books on qualitative analysis the student is directed to sepa-
rate manganese from zinc by adding to the solution of their chlorides an
excess of sodium or potassium hydroxide. Now. if the alkaline solution
of the zinc, after its separation from the manganous hydroxide, be treated
for 15-20 minutes with a rapid current of hydrogen sulphide the zinc sul-
phide which is first precipitated may dissolve. The smaller the amotmts
of zinc and sodium chloride present in the solution, and the more rapid
the current of gas, the more readily does the zinc sulphide formed pass
into solution. Should a student pass a rapid current of hydrogen sulphide
into such a solution, and then leave the spot and not return tmtil after
the lapse of some 15-20 minutes, the chances are he will find the solution
clear, or nearly so, and report no zinc. I have convinced myself by a num-
ber of experiments that there is, in a case of this sort, considerable danger
of overlooking the zinc. It is a significant fact that Fresenius^ uses hydro-
gen sulphide water in order to test for the zinc in the alkaline solution.
He is also careful to state that an excess of the reagent is to be avoided.
However, in describing the special reactions of zinc' he states that hydro-
gen sulphide precipitates from alkaline solutions all the zinc in the form
of the hydrated sulphide. Nothing is said about an excess of the reagent,
although he does mention the fact that ammonium chloride greatly pro-
motes the separation of the precipitate.

In the following experiments I used a sodium hydroxide solution of zinc
oxide containing 8 grams of the oxide in a liter:

1. Ten cc. of the solution were diluted to 150 cc. and treated with a rapid
current of hydrogen sulphide. The zinc sulphide was precipitated almost
immediately, .but at the end of 15 minutes it had passed into solution.

2. To lo cc. of the solution, made faintly acid with hydrochloric add,
a few drops of a concentrated solution of manganous sulphate were added
and the metals precipitated as sulphides with yellow ammonium sulphide.
After filtering, dissolving the precipitate in a small amount of very dilute
hydrochloric acid, and separating the manganese with excess of sodiiun
hydroxide the alkaline filtrate was diluted to about 150 cc. and treated
with a rapid current of hydrogen sulphide. In 15 minutes the zinc sulphide
had dissolved and the solution was clear.

3. The experiment was repeated with a solution containing the same
amount of zinc oxide along with considerable amounts of manganese,
cobalt and nickel. The alkaline solution of the zinc oxide, however, be-
haved toward the sulphuretted hydrogen gas exactly as it did in i and 2.

' Anldttmg z. qualit. chem. Analyse, 1874, 291.
* Loc. cit., p. 137.

37^ ■ GEORGE M. HOWARD.

On standing, all three solutions became turbid, owing to a gradual sqa-
ration of the zinc sulphide.

FitlHCITOH, N. J.,

THE DETERimiATIOH OF ABTIHOirY USD ASSEHIC HT LEW

ABTmONy ALLOYS.

BvGbokok U, Howard.

Received Januaiy aa« 190^.

The separation and determination of antimony and arsenic in alloy;
with anything hke commercial rapidity, and at the same time with s
fair degree of accuracy, is by no means a simple matter. The appar
ent lack of satisfactory published methods led to the development sev-
eral years ago of the method here described, which, while involving noth-
ing radically new, has proved very serviceable in the author's labora-
tory.

One great advantage of this method is that tin does not interfere and
does not have to be removed, which makes it available for type metal
etc. Iron and copper in small amounts are also without effect, and in
fact there seems to be nothing at all likely to be present which inter-
feres. The method is applicable also to many other cases besides lead-
antimony alloys — as for instance to the mixed sulphides of antimonj
and arsenic obtained in many analyses. Some operators object seriousl)
to any method involving the use of hydrogen sulphide, but if the de-
tails of manipulation here given are followed there will be no incon-
venience whatever from that source.

The procedure is as follows:

The sample in fine filings, 0.5 to 2 grams according to circumstances,
is weighed into a 125 cc. Erlenmeyer flask, 60-70 cc. of strong hydro-
chloric acid added, and two or three drops (not more) of nitric acid (1.4)
The flask is then placed on a hot plate where it will be just short of boil-
ing until solution Is complete. Frequent agitation con^derably hasteo!
the action. It is sometimes necessary to make further additions oi
oitric add, but this should be done carefully and an excess avoided
When the metal is all dissolved {10-20 minutes) the flask should be moved
where it vrill boil vigorously for a few minutes until the color changes
from reddish yellow to colorless— or, if iron or copper is present, to straw
yellow. Now, while still hot, hydrogen sulphide is passed into the so-
lution until it is completely saturated — 15 minutes is usually sufBdent.
If insuffident hydrochloric acid has been used, or the solution has been
boiled so long on the plate that much has been lost, antimony sulphide
frill be predpitated as the solution cools. The hydrogen sulphide treat-
ment is most conveniently handled by fitting the flasks with two-hole

AXTIMONV AND ARSExNlC IN LEAD-ANTIMONY AL,I.OYS. 379

Stoppers and inlet and outlet tubes, and connecting several in series
to tliL- c^cnerator. If the outlet from the last flask is led into a bottle
of ar.-slic soda solution, no gas whatever conies off free. When satu-
rated, tliv- fl isks are transferred to a current of air, still in series and again
absorbing the gas from the last flask in caustic soda, and the air passed
until all the hydrogen sulphide is removed (1/2 hour is sufficient for two
tiasks). Tlie hydrogen sulphide precipitates nothing but arsenic as sul-
pliide, but reduces all salts capable of reduction — antimony, tin, cop-
per, iron, etc. The current of air then reoxidizes all of these except the
antimony, which remains in the antimonious form.

To the now^ cold solution a little tartaric acid is added, and water until
its bulk is about doubled, when it is filtered, best through a double filter,
into a 16 oz. flask. Practically all of the lead chloride must be washed
out of the precipitate with hot water, but it is not necessary to add all
the wasliings to the filtrate, as the antimonious chloride is readily washed
out bv decantation with cold water.

Atil: !h<ui}\- -Thii filtrate is nearly neutralized by -adding powdered
sodium carbonate in small portions, care being used not to reach the
point of precipitation of the lead, or, if this is reached, making it slightly
acid again w^th hydrochloric acid, 'i'he neutralization is then com-
pkt'jd with sodium bicarbonate and a slight excess added (about one
spoonful of powder), 'fhe antimony is then determined by titr^-ting
with standard iodine solution, using fresh starch solution as indicator.
The precipitate of lead carbonate do.s not affect the titration, and with
a little practice the end pouit can be recognized just as easily as in a
ckar solution. Too much starch should be avoided, as it makes the
end point obscure. A convenient strength for the iodine solution is
I cc. =- 0.005 gram antimony, then with 0.5 gram samples i cc. = i
p.r cent, antimony.

If ars.nic is neghgibly low -which, with a little experience, can be
pretty wi-ll ga^g^jd by the appear.. iice of the precipitate — or for any
reason is not to be determiiud, the filtering may be dispensed with.
In tins case the whole sohition, with the arsenious sulphide, if any, and
free sulphur in suspension, is merely transferred to a larger flask, neu-
tralized and titrated. Tlie suspended arsenious sulphide and free sul-
phur are without effect.

Arsitiic. -'fhe bulk of the arsenious sulphide and sulphur is washed
off tile filler back into the same flask in which the precipitation was made,
using not over 20 cc. or so of water. A few drops of sodium hydroxide
are added (5 drops of 20 per cent, solution is ample), the solution boiled
for I few moments and then decanted through the filter into an 8 oz,
Erlenniever flask. Tliis weak soda readily dissolves the arsenious sul-

380 GEORGE M. HOWARD.

phide, while taking up only a small part of the free sulphur. If the amount
of precipitate is at all considerable, it is safer to give a second treatment
with soda solution.

The filter is washed with hot water and discarded. To the filtrat
is added hydrogen peroxide solution, which should be reasonably fresh
or else its strength known. 30 cc, of 3 per cent, solution is suffiden
for arsenic up to several per cent. The hydrogen peroxide oxidize
the arsenic and also all sulphur compounds, giving a colorless solution
This is now boiled down to a small bulk, about 20 cc, the excess of perox
ide being decomposed in the process. When cool, potassium iodid
solution is added in amount equivalent to about 0.1 gram KI, then 21
cc. of strong hydrochloric add. After standing five minutes it is coob
and titrated with standard thiosulphate, adding three drops of starcl
.solution only when the color is almost gone. A convenient strengt
for the thiosulphate solution is i cc. = o.ooi gram arsenic, and it should
of course, be frequently standardized. It is well also to run a blan
titration, uang the same amounts of reagents as in the analysis, to dc
termine whether there is any constant to be deducted from the burett
reading, due to impurities in the reagents.

The difficulty with the end point experienced by many in this titra
tion appears to be due to using too large an excess of potassium iodid
and too much starch. With the proportions given above, the end poii
is exceedingly sharp, and the reaction seems to be just as complete a
when more potassium iodide is used. The heating due to adding tli
strong hydrochloric add is suffident without any digesting, as is somi
times recommended.

After oxidizing and concentrating, the arsenic is in a form to be reat
ily determined in several ways, although the above method is preferre
on account of ease and rapidity. For instance, the arsenic may be pn
cipitated with silver nitrate as silver arsenate, after neutrahzation wit
acetic add, and the combined silver titrated vrith thiocyanate, as i
the well known modified Pearce method. In this case care must I
used to wash the sulphide free from lead chloride, and also to have a
reagents free from chlorine (most of the hydrogen peroxide solutioi
on the market contain hydrochloric acid). Or, the arsenic may be di
termined gravimetrically by predpitation as ammonium magnesiu;
arsenate. In this case it is better to use ammonia instead of sodiui
hydroxide to dissolve the arsenious sulphide.

PRBCIFItATION OP COPPBR PROU NITRIC ACID. 38 1

[CONTRIBimON PROM THB ChSMICAL LABORATORY, OmO STATS UnIVBRSITY.]

THE DJFUJEKCE OF TEMPERATURE ON THE ELECTROLYTIC PRE-
CIPITATION OF COPPER FROM NITRIC ACID.'

By Jambs R. Withrow.
Received January 13, 1908.

While determining copper electrolytically with stationary electrodes,
it was observed upon several occasions that after the copper appeared }

to be almost all precipitated it slowly re-dissolved and disappeared from
view within a very few minutes. This took place without any change
in the current conditions as registered on either the ammeter or volt-
meter. Evidently the conditions for the determination were very close
to the limits of the ability of the current to precipitate the metal. A
slight variation of the conditions, therefore, in some one of the possible
directions during the course of an experiment, was suflScient to permit *
the free nitric add in the electroljrte to overcome the influence of the
current and dissolve what metal had already been deposited.

Increasing temperature of the electrolyte caused by fluctuation
of the gas pressure was blamed for the difficulty. It was soon found
that higher temperatures than customary^ existed in the electrolytes
which caused trouble. The following work was then undertaken to
ascertain just what were the safe limits of temperature for a variety
of concentrations of nitric add, using a fixed amount of current.

Platinum-iridium dishes approximating the customary form were
used as cathodes. They were 9 cm. in diameter and 5 cm. deep and when
they contained 100 cc. the cathode area was about 100 sq. cm. The
anode was a plattnum-iridium wire nearly 1.5 mm. in diameter (about
No. 15, B. and S. gauge). It was bent in the form of a spiral 4 cm. in
diameter, comprising three complete turns. In all, there were 26.5
an. of the wire exposed to the electrolyte. The current was supplied
by storage cells, and the current conditions given in the tables have been
calculated to the **Normal Density" from the instrument readings, be-
cause the "Normal" cathode surface was not used. The American In-
strument Company's Type 2 switchboard ammeter had a capacity of
one ampere and the scale was divided into hundredths. The voltmeter
had a capadty of 50 volts with scale divisions of half volts.

As a result of some preliminary experiments the distance between the
electrodes was maintained at 2 cm. throughout the work. This made
their actual separation at all points approximately uniform. The cur-
rent strength was kept constantly at 0.08 amp. (per 100 sq. cm.). The
total dilution was maintained at 125 cc. When nitric acid was added
to the electrolyte the metallic deposits were in all cases brilliant and com-

^ Read at the Chicago meeting of the American Chemical Society.
• Edgar F. Smith's "Electrochemical Analysis," 3d Ed., 1902, p. 59.

38:^

JAMES R. WITHROW.

i^J

r^-**

'i?ij

pact except at the higher temperatures (above 70°). In these latter
cases the deposits were not smooth and burnished but were made up of
a collection of lustrous individual crystals of copper. When examined
under the microscope these crystals were of octahedral aspect. Tht
deposits from the solutions in which nitric acid was entirely absent,
were of a different character. All deposits were, however, thoroughly
adherent, and no tendency to sponginess was observed in any case. This
good condition of the deposits enabled them to be weighed without the
use of a drying oven. Washing with boiling water was dispensed with
for the same reason.

They were washed thoroughly with distilled water, carrying off the
excess with a siphon in the usual manner vvithout breaking the circuit.
When washed free from all electrolyte the dish was removed and rinsed
with dilute, and then absolute alcohol. Finally absolute ether was used,
and after the adhering ether had been all vaporized and the dish rubbed
on the outside with a piece of chamois skin it was placed in a desiccator
and weighed in from 15 to 30 minutes. The wash water, or at least
the first portion of it, was always preserved and tested in the usual man-
ner. When, however, this solution was neutralized with ammonium
hydroxide and acetic add added followed by potassium ferrocyanide
the presence of a trace of copper was often observed.

Occasionally, during the progress of the work there was a discolora-
tion of the anode. This discoloration was of a more or less milky appear-
ance, but at times it resembled anodic deposits that have been obtained
from the electrolysis of gold solutions.^ At other times it was like bur-
nished deposits of gold itself. This discoloration was probably of the
nature of a deposit for it could be removed by the touch of a finger, or
immersion in nitric acid. No increase in weight of the anode could be
detected, however, even in several determinations run for upwards of
20 hours. When the deposit was removed by nitric acid no change in
the weight of the anode greater than a tenth of a milligram could be de
tected except in a single isolated case of 0.3 mg. No further attention
was therefore given to the matter at this time. The amount of copper
considered as present in the solutions used in the work was the mean
of several determinations which were run until the electrolyte g-ave no
further tests for copper.

It was found that in the complete absence of nitric acid the copper
from a solution of pure copper sulphate (containing 0.25 gram of metal)
could be completely precipitated in 16 hours. The rate of precipitation
of copper was therefore determined from tliis electrolyte, in the total
absence of nitric acid, to point out the value of the addition of this add.

* This Journal, 28, 1353.

1>RSC1PITAT10N 0I^ COPPER l^ROM NITRIC AClD. 383

CnSOf. HNO».
Cu in grama. cc.

Total
dilution,
cc.

Cnrrent.

Temp.
Centigrade.

Time.
Hours.

Cu found.
Gram.

N.D.iooamp.

Volt*.

0.2503 0

125

0.08

3-3-2.6

0.1112

0.2503 0

125

0.08

3.3-2.5

0.2234

0.2503 * 0

125

0.08

3. 3-3-3

0.2482

0.2503 0

125

0.08

3-3-2.8

0.2510

In the absence of nitric add the copper did not appear as rapidly upon
closing the circuit as it seemed to do in the presence of small amounts
of this add. In the above determinations the solutions appeared to take
an an olive-green tint before the copper became visible. This was also
observed when a solution containing only copper nitrate was electro-
lyzed, but not when free nitric acid had been added. Owing no doubt
to the liberation of sulphuric add, the conductivity of the electrolytes
in the above determinations increased stea^ly, requiring continued in-
crease in the resistance of the rheostat to maintain the current strength
at 0.08 ampere per loo sq. cm. The deposit was about as bright at
the end of one hour as it would be from a nitric acid electrol)rte. After
about two hours, however, the copper began to come down on this bright
deposit in a pulverulent form. It would come off as a powder, if touched,
but never of its own accord, even upon thorough washing. It appeared
velvety and had completely lost its metallic luster. In color it approached
the so-called *'cherry-red" of powdered hematite. This pulverulent
copper was in all cases only a superficial coating, as was shown by treat-
ing it with dilute nitric add, when it dissolved with ease, exposing the
bright copper beneath.

The copper from this electroljrte was completely predpitated in 13
hours. The deposit, however, weighed high. This seems to be a com-
mon complaint against the sulphuric add electrolyte, if the determina-
tions are run too long.

The rate of predpitation of the metal was determined from a solution
of copper nitrate to which no free add had been added.

Cu(NO,)t.
Co m grams.

HNOs.
cc.

ToUl

dilution.

Current.

Temp.
Centigrade.

Time.
Hours.

Cu found.
Gram.

N.D.iooamp.

Volts.

0.2505

125

0.08

2.9-2.5

0. 1087

0.2505

125

0.08

2.8-2.6

0.2208

0.2505

125

0.08

2 . 9-2 . 6

0.2450

0.2505

125

0.08

2.8-3.7

0.2504

These deposits were not all like those from pure copper sulphate. They
resembled the deposits obtained in the presence of free nitric add, but
were not so brilliant and had a slight tendency to be pulverulent at the
center. A comparison of these two tables will show that while the copper
may appear first from copper nitrate, yet slightly more is predpitated in a
given time after one hour and beyond, from copper sulphate.

Using the copper sulphate solution the rate of predpitatiori of copper

}84 JAMES R. WITHROW.

vvas determined at 25° in the presence of varying amounts of nitric ac
[sp.gr. 1.42).

3503
*503

2503
2503

2503
2503
2503
2503
2503
1503
2503
2503
2S03
3503
2503
2503
2503

From the results given above it will be seen that the rate of predp
tion in presence of -0.25 cc. of nitric add was slightly lower than w
no free add is present in the copper nitrate electrolyte and much lo
than with pure copper sulphate. This gave a fairly good idea of
retarding effect of the presence of nitric add. The effect, howe
was not great, and the improvement in the character of the depc
more than compensated for it.

In the presence of i cc. of nitric add a quarter gram of copper
predpitated in 12 to 13 hours. Running the detenuination for 15,
23 or 24 hours had no effect on either the appearance or weight of
deposit. The results in presence of this amount of add were ahi
identical with those from the 0.25 cc. nitric add electrolyte.

When two per cent. (2.5 cc.) of nitric add was present the predp
tion of a quarter gram of copper required 13-14 hours. The ret
ing influence of the free nitiic add was now becoming more emph;

PRECIPITATION OP COPPER PROM NITRIC ACID. 385

and when five per cent. (6.25 cc.) of acid was used, the effect was pro-
nounced. In this case the precipitation of the quarter gram of copper
was still incomplete after 17 hours. As this was too long an interval
for the precipitation of such a small amount of this metal, the work with
these conditions was not carried farther. The result with ten per cent,
nitric add (12.5 cc.) came from a faint film (or line) of metal at the up-
per edge of the cathode surface exposed to the electrolyte, and probably
should not be called a deposit from this strength of solution. Red ox-
ides of nitrogen were evolved from the solution containing 50 per cent,
nitric acid and no copper was deposited. This work showed that ni-
tric add retarded the predpitation of copper, but that as it had such
a beneficial effect on the character of the deposits, the presence of
a small amount was a most desirable addition. The following compar-
ison of the amounts of copper precipitated in six hours from the electro-
lytes containing varying amounts of free nitric add, illustrates very
clearly the retarding effect of this add, a fact already appreciated by
those who have used this electrolyte.

Cu deposited
Temperature Cu present. in 6 hours.

Electrolyte. centigrade. Gram. Gram.

CUSO4 -Ho CC HNO, 25 0 . 2503 0 . 2482

Cu(NO,), + o cc. HNOj 25 o. 2505 o. 2450

C11SO4 + 0.25 cc. HNO, 25 0.2503 o. 2445

C11SO4 + 1 .00 cc. HNO, 25 o. 2503 o. 2439

CUSO4 + 2 . 50 cc. HNO, 25 o. 2503 o. 2418

CuSO^ +6.25CC. HNO, 25 0.2503 0.1585

Having found just what the influence of nitric add was on the precip-
itation of copper at ordinary temperature, the minimum amount of this
add was used in a number of experiments at higher temperatures.

Total Current.

CUSO4. HNOs- dil. " . Time. Cu found.

Cain grams. cc. cc. N.D.iooamp. Volts. Temp. Hours. Gram.

0.2503 0.25 125 0.08 2.4-2.3 40° I 0.0972

0.2503 0.25 125 0.08 2.3-2.4 40® 3 0.2261

0.2503 0.25 125 0.08 2.3-2.4 40® 6 0.2493

0.2503 0.25 125 0.08 2.3-2.5 40° 7 0.2500

0.2503 0.25 125 0.08 2.4-2.4 60** I 0.0833

0.2503 0.25 125 0.08 2.3-2.2 60® 3 0.2288

0.2503 0.25 125 0.08 2.2-2.3 60® 6 0.2501

0.2503 0.25 125 0.08 * 2. 1-2. 1 70® I 0.0683

0.2503 0.25 125 0.08 2.1-2,1 70® 3 0.2214

0.2503 0.25 125 0.08 2.1-2.2 70° 5 0.2488

0.2503 0.25 125 0.08 2.2-2.3 70** 6 0.2501

0.2503 0.25 125 0.08 2. 1-2. 1 80® 3 0.1823

0.2503 0.25 125 0.08 1.7 80® 6 0.2491

0.2506 0.25 125 0.08 2.1-2.2 90® 3 0.0821

0.2503 0.25 125 0.08 ' 2.1-2.3 90° 6 0.2460

0.2503 0.25 125 0.08 70® 6 0.0000

0.2506 0.25 125 0.08 2.4-2.2 25-70° 3 0.2258

Cram.

houn.

0.3503

0.^503

0,1503

0.3513

0.3503

386 JAHBS R. WITHROW.

From these results it seemed that even the slight elevation of
perature to 40° decreased the time by a considerable amount. A
ther elevation to 60° still decreased the time but to a much smaUe
tent, while 70° did not seem to improve the results at 60°. In fad
time reducing effect of increased temperature reached its limit at 6a
and from that point any elevation of temperature had a retarding <
on the precipitation. This is well illustrated by a comparison ol
amounts of metal deposited in six hours at the various tempenit

Cu present. Co dcpo>iUd
Klectrolj'te. Tcmpenlure. ' '

CuSO^+o. 35 cc. HNO, JS"

CuSO,+o.35 cc, HNO, 40°

CuSO,+o, 35 cc, HNO, 60°

CuSO,+o,33 cc, HNO, 70°

CuSO,+o.35 cc, HNO, 80"

CuSO,+o.25 cc, HNO,, , 90' 0,3503 0.3460

Upon one occasion where the current strength had fallen to 0.0;
pere and the copper appeared almost all precipitated, the elect
was heated with the result that about half of the dish was clean
its deposit in spite of the fact that the current increased at the
time to 0.15 ampere. That the increased solvent power of nitric
with elevated temperature, greatly overbalances the current's inci
power of precipitation is further shown in the case of the five per
nitric acid electrolyte. At 70" no copper is precipitated, whereas (
gram is precipitated in the same time at 25°.

In the case of the last determination in the table the temperatur
maintained at 25° for two hours and forty-five minutes. It was
raised to 70° during the next ten minutes and held there for the re
ing five minutes of the three-hour period. l*he result indicates tl
celemting influence of increase of temperature, it being much )
than for the same period of time at either 25° or 70°. This infl
of temperature on electrolysis is completely masked at first in all
where the electrolyte is heated to the desired temperature before
ing on the current. From the table it will be seen that the amoi
copper precipitated in one hour is less and less as the temperatu
creases. After this first hour, however, the beneficial efl'ect of inc
temperature is to be seen in every case until the temperature at
the solvent action of the nitric acid begins to persistently dii
the electrolytic deposition.

Concluuoos.

1. Even the smallest amounts of nitric add have a tendency to
the electrolytic precipitation of copper, under the conditions here

2. The presence of nitric acid is, nevertheless, de^rable becai
its beneficial effect on the character of the depomt.

THE GRAVIMETRIC DETERMINATION OF TELLURIUM. 387

3. While in general increased temperature means accelerated precipi-
tation, vet with the low current strength and conditions here used, the
reverse is the case above 70°, no doubt owing to the rapidly increasing
solvent action of the acid .

Columbus,
August 7, 1907.

THE GRAVIMETRIC DETERMINATION OF TELLURIUM.

By Victor Lekhrr and A. W. Hombbrobr.
Received January 17, 1908.

Of all the methods proposed for the precipitation of tellurium perhaps
the one which is most used is a modification of the original method of
Berzelius. He used sulphurous acid as a precipitating agent.

The method of procedure as commonly carried out consists in adding
to the hydrochloric acid solution of tellurium a strong aqueous solution
of sulphur dioxide and allowing this mixture to remain in a warm place
for a few days in order to effect a complete precipitation. It has been
shown by Schroetter,^ Brauner/ Norris and Fay,' Crane,* Frerichs,*
and others that the precipitation by means of sulphur dioxide is far from
satisfactory. Brauner has pointed out that part of the precipitated
tellurium undergoes oxidation in the liquid, becoming converted into
the tetrachloride, in which form it remains in solution. Crane has sug-
gested that the main cause of the incomplete precipitation by means
of sulphur dioxide is the very rapid increase in the ratio of the adds
to the unprecipitated tellurium in solution, two-thirds of this being due
to the hydrochloric acid set free, and one-third to the sulphuric add
formed. He thought if these could be removed the reduction would
be complete. The hydrochloric acid could be eliminated by evaporation,
but the continuous increase in sulphuric add would soon interrupt the
reaction. This might, however, be kept under control by the addition
of sodium or potassium hydroxide.

Whitehead has suggested a remedy in the use of add sodium sulphite.
He advises a moderately concentrated solution of the sulphite and that
the quantity added to the tellurium solution be sufficient only to just
neutralize the acids present and that formed during the reaction. When
the solution is thoroughly agitated and then allowed to stand in a warm
place, the precipitate will form and settle evenly. He states that "while
add sodium sulphite does not completely remove all of the tellurium
from the solution in the cold, that if not used in great excess and the

* Chem. News, 87* 17.

' J. Chem. Soc, 55, 392.

* Am. Chem. J., 20» 278.

* Ib?d.^ 23, 408.

* J. fiir pr. Chem., 66> 26?. •

385 VICTOR LENHEB AND A. W. HOMBERGER.

mixed solutions be raised to the boiling point, toward the end of
action the precipitation will be perfect, and the tellurium will be obta
in a state of aggregation favorable to easy filtration."

Frerichs has worked on the basis that hydriodic acid and sul
dioxide cause immediate and complete separation of tellurium fro
tellurous solution even in thecold.andMcIvor'hasconfinnedthismel

Norris and Fay' have demonstrated that under ordinary woi
conditions precipitated tellurium increases in weight about 0.5 per e
owing to oxidation and that this increase is balanced by the quai
of the elements left behind as tellurium tetrachloride in the stro
acid solution in which the precipitate is formed by sulphur dioxide. '
believe that it is more accurate to weigh tellurium dioxide than to v
tellurium in the elementary state.

Mclvor' and Donath* have studied the precipitation of tellu
by hydrosulphurous acid. This method possesses the disadvan
of a precipitate of tellurium contaminated by more or less sulphur,
method hardly possesses any advantages over the sulphurous add
cipitation.

Stolba* in 1873 and later Kastner* have proposed the predpit<
of tellurium from an alkaline solution by means of grape sugar.

Later, Gutbier' described the precipitation of tellurium by meat
hydrazine as a method for its determination. His method of proce
is to dissolve telluric acid in warm water in a porcelain dish cov
with a glass cover and add by means of a pipette a 10 to 20 per 1
solution of hydrazine hydrate. A dark blue almost black color is n
and after heating a short time, elementary tellurium is predpit
in a flocculent condition, the liquid becoming colorless. He conti
the addition of hydrazine hydrate until the fluid is no longer col
by further addition of the reagent.

Experimental.

Precipitation by Hydrazine.— In our hands the method of Gutbier
fairly good results. The fact should be noted, however, that the add
of an excess of the hydrazine does not at once predpitate all of the
lurium. It is preferable to add a small amount of the predpib
agent from time to time. This necessitates several hours for com]
predpitation. The following results were obtained by Gutbier's me
in hydrochloric acid solution :
' Chem. News, 87, 163.
' Am. Chem. J., ao, 278.
» Chem. News, 871 163.
* Z, angew. Chem., 18901 314.
' Z. anal. Chem., iii 437.
■ Ibid., 13, 142.
' Ber., 34, 2734.

THE GRAVIMETRIC DETERMINATION OF TELLURIUM. 389

TcOi. Te required. Te obtained.

Gram. Gram. Gram.

0.2247 0.1795 0.1790

0.1988 0.1588 0.1577

0.2006 0.1603 0.1596

0.2056 0.1643 0.1637

Precipitation by Sodium Acid Sulphite. — In order to completely pre-
cipitate the tellurium by this reagent from a hydrochloric add solution
of a tellurous compound, the solution must contain excess of the reagent
and must be allowed to stand in a warm place for twenty-four hours.
In the following experiments, the sodium acid sulphite was prepared
freshly for this purpose by passing sulphur dioxide into a solution of
sodium carbonate. It has been our experience that when acid sodium
sulphite which has not been freshly prepared is used for the precipita-
tion of tellurium, the precipitated element frequently contains sulphur.

That it is necessary for the solution to stand a considerable length
of time is apparent from the following experiments, all of which
experiments were made under exactly the same conditions. The solu-
tion of the dioxide in hydrochloric acid was brought to boiling, a sat-
urated solution of add sodium sulphite was added, the solution allowed
to stand the requisite length of time, then brought on a Gooch platinum
filter washed with water until the filtrate no longer showed chlorides,
after which it was washed with 15 cc. of alcohol and dried at 105°.

Te required. Te obtained. Brror.
Gram. Gram. Gram.

Solution allowed to stand two hours o. 1609 o. 1586 — 0.0023

" " " " 0.1609 0.1590 — 0.0019

" " " " 0.1767 0.1744 — 0.0023

Allowed to stand six hours o. 1609 o. 1600 — 0.0009

0.1609 0.1603 — 0.0006

0.1374 0.1366 — 0.0008

0.1527 0.1517 — o.ooio

Allowed to stand twenty-four hours o. 1609 o. 1615 -fo.0006

" " " " 0.1609 o.i6i8 +0.0009

" " " " 0.1726 0.1730 +0.0004

" 0.1286 0.1289 +0.0003

After tellurium, which has been predpitated by means of sodium
sulphite and hydrochloric add, has been washed thoroughly with water,
and alcohol, it oxidizes very slowly when heated as high as 200® as evi-
denced by the following data:

Length of time of beating. Temperature. Te(i). Te(2).

15 minutes 105° 0.1619 0.1620

15 " 105° 0.1619 0.1620

I hour 120-130° 0.1620 0.1622

I " 200** 0.1620 0.1623

Precipitation by Means of Sulphur Dioxide. — By the treatment of tel-
lurium dioxide dissolved in hydrochloric add with a freshly saturated

f( i< ft

U <l (I

If i< tt

}90 VICTOR LENHER AND A. W. HOMBBRGBR.

solution of sulphur dioxide and allowing to stand for 24 hours, the fi
lowing results were obtained :

Tereqalrcd Tt obtained.

.1609

o. 1617

O.1613
0.161S

0.1609

0.1609 0,1613

Unless the acidity in this precipitation is ten per cent., the tellurii
is not likely to be completely precipitated or it will be precipitated
a very fine state of division. The solution should also be hot in on
to secure satisfactory precipitation.

Simultaneous Precipitation by Means of Sulphur Dioxide and Hyi
Une. — By bringing both sulphur dioxide and hydrazine into a tellurii
solution the whole of the elemout is thrown out of the solution alnt
instantaneously. The solution should have an acidity of 5 to 10 |
cent, and it is desirable to have the solution in a high degree of cono
tration. The solution is brought to boiling and 15 cc. of a satuiai
solution of sulphur dioxide is added, then 10 cc. of a 15 per cent, so
tion of hydrazine hydrochloride and again 25 cc. of the sulphur dii
ide solution. The solution is boiled for a few minutes when the elem
tary tellurium will settle in such a way that it can be rapidly wash
The precipitate is then transferred to a platinum Gooch filter and was]
first with hot water until all of the chlorine is removed, and then n
15 cc. of alcohol. The crucible and contents are then dried at ioo-i(
and finally weighed.

The following results were obtained by the process as outlmed abo
using 10 cc. of solutions of hydrazine hydrochloride of different streng
with sulphur dioxide. Tellurium dioxide was used for the analy
Streogihof

le hydro

Tc obUincd.

O.I731
0.3065

1735

ao68

i-o
+0

0004
0003

0.1638

1641

+ 0

0003

o.idoS

1608

0000

O.ZIU

1310

— 0

0001

0 U35

1434

— 0

0001

0.1605

1607

oooj

0. 1071

1070

— 0

0001

9. 1658

1656

— 0

0001

0.1643

1637

— 0

0005

0.1368

1364

— 0

0004

0.1432

1430

— 0

0003

That hydrazine must be there in sufficient quantity is evidenced
the following series of tests in which a 6 per cent, solution was used ali

LOSS OF PHOSPHORIC ACID IN ASHING OF CBREALS. 39 1

with sulphur dioxide and the solution boiled only a few minutes, other
conditions being exactly the same as in the preceding series of experi-
ments.

Error.
Gram.

.0134
.0258

.0073
.0381

.0122

The following two experiments were made with a large excess of sul-
phur dioxide water along with a 6 per cent, solution of hydrazine and
the solution heated six hours.

Te required. Te obtained. Error.

Gram. Gram. Gram.

Te required.
Gram.

Te obtained.
Gram.

0.1508

0.1374

O.1701

0.1443

0.1608

0.1535

O.1521

O.I 140

0.1903

0.1781

0.1680 0.1545 — 0.0135

0.1516 0.1416 — O.OIOO

The method which has been used in this laboratory for a number of
years and which has proven the most satisfactory for the gravimetric
determination of tellurium is as follows: The tellurium either as de-
rivative of the dioxide or as a tellurate, should be present in a solution
which has an acidity of approximately ten per cent, of hydrochloric acid
and it is preferable ,to have the solution sufficiently concentrated, other-
wise the fine state of division of the precipitate will render it unsatis-
factory for washing. The solution is heated to boiling and 15 cc. of a
saturated solution of sulphur dioxide added, then 10 cc. of a 15 per
cent, solution of hydrazine hydrochloride, and again 25 cc. of a satu-
rated solution of sulphur dioxide. The boiling is continued until the
precipitate settles in such a way that it can be easily washed. This
boiling should not take more than five minutes. The precipitated tel-
hirimn after being allowed to settle is washed with hot water on a Gooch
filter until all of the chlorine is removed, after which the water is dis-
placed by alcohol and the crucible and contents dried at 105°.

University of Wisconsin,
Madison, Wis.

LOSS OF PHOSPHORIC ACID m ASHING OF CEREALS.

By Sherman I«eavitt and J. A. LbClbrc
Received December 26, 1907.

Recent work in this laboratory on the detennination of phosphoric
add in the ash of wheat has brought to the attention of the writers the
fact that whereas the temperature below fusion, at which ashing of grain
is carried on, makes very little difference in the percentage of ash, there
is a loss in the corresponding values of phosphorus, varying with the tem-
perature.

2 SHGRHAK LBAVITT AND J. A. LB CLBRC.

The fact that there is danger of losing phosphorus where grains are ;
wed to become fused during ashing, has been known for some yea:
d the methods of the Association of Official Agricultural Chemists f
e determination of phosphoric acid in ash were originated with ti
;a in view. In grains rich in phosphorus, such as cottonseed- me
e methods of the A. O. A. C. prescribe charring and extracting the chan
iss with acetic add or water. Many analysts, however, do not c»nsid
necessary to make a previous extraction on a grain comparati\'eIy !<

phosphorus, such as wheat. Our results show that there is a loss
lOsphoms below the fusing point of the ash, and, as stated above, tl
IS varies with the temperature.

The greater part of the phosphorus in wheat is in a water-soluble foi
x)wn as phytin, a substance of relatively high molecular weight coi
red to the phosphorus molecule. A comparatively large percenta

the phosphorus may be lost in ashing without appreciably changi
e amount of ash. Two grams of a sample of ground wheat were weigh
to a flat platinum dish and ashed in a muffle at a temperature of li
Iness for five hours until the ash was gray or white. The ash was tb
ighed and dissolved in a few cubic centimeters of concentrated nitric a<
d filtered and the determination of phosphoric add made by the vo
ftric method of the Association of Official Agricultural Chemists.
It was found rather difficult to produce a perfectly white ash at the te
rature used in the muffle, so a higher temperature was tried and the o
^ponding phosphoric adds determined. The substance used in evf
X was wheat ground so as to pass through a one millimeter sieve. T
ies of ashings were carried out, one at low redness, in which the samp
re allowed to heat up from the cold muffle, and the other set ashed
listinct redness.

Tablb I.-

-AvBRAGB Results ok Sbvbn Samples Dupucatsd.

A««.(t«.

Aihal Aahal Percent PfOi. Percent. P,Oi, Per
low redness, redneis. low ledneu. rtdocM. 1

;ven samples.

... 2.tM 1.99 0.74 0.40

The results from which Table I was obtained were all determined in dui
:e and averaged, and the final table made from these averages. In e\f
% the difference in the ash between ashing at redness and at low redn
,s less than o.i per cent., whereas the loss of phosphoric add van
tween 39 and 53 per cent.

As is well known, the addition of calcium acetate prevents the volatilii
n of phosphoric add. Two samples in which the percentage of pb
oric add had been previously determined by ashing with an excess
Idum acetate were then ashed in the ordinary way in the muffle at a co
ratively low temperature, one duplicate of each determination bd
iced in the back part of the muffle and the corresponding ones in t
nt of the muffle.

I^OSS OF PHOSPHORIC ACID IN ASHING OF CEREAI^. 393

Tablb II.

Percent. Percent, Percent Percent. Percent.

Number PiO^ by cal- PaOp. Back P9O5. Front PgOs lost. Back P^Ok lost. Front

of sample. cium acetate. ofmuffle. of muffle. of muffle. of muffle.

2154 1.03 o 85 0.97 17.5 5.8

2155 1.05 0.63 0.90 40.0 14.3

We see from Table II that the loss of phosphoric acid is greatest in the
back of the muffle where it is hottest and least in the comparatively cool
part of the muffle. The loss in either case, however, is an appreciable
one and shows the necessity for ashing at a comparatively low tempera-
ture.

In order to make a more thorough study of this loss of phosphorus a
number of samples were ashed in the regular way at redness; a duplicate
set was also ashed by treating with 5 cc. of a solution of calcium acetate
of such a strength to furnish more than enough calcium to fix all of the phos-
phoric add and render it non-volatile. (A solution yielding about 0.03
grams CaO in 5 cc. of the solution is suitable for 2 grams of grains.) The
samples treated with calcium acetate solution were heated on a steam
bath to dryness and burnt over the free flame and finally heated for fifteen
minutes over the blast lamp and weighed for ash, allowing for the correc-
tion of calcium oxide in 5 cc. of the acetate solution. Phosphoric acid
was determined in the ordinary way.

Twenty-three samples of wheat gave the following results:

Tablb III.

Per cent. Per cent. Per cent. Per cent,

ash at ash, ace- Per cent. PsO^ in ash PfOs, ace- Per cent,

redness. tate method. loss, ash. at redness. tate method. PflOs lost.

2.09 2.07 i.o 0.64 0.99 35.0

Eighteen of the above samples gave a diflference of less than o.io per
cent, between the ash at redness and the ash by the acetate method. The
phosphoric acid varied between 10 and 50 per cent. loss. Table III shows
us that there is no appreciable loss of ash but a loss of 35 per cent, of the
total phosphorus determined as phosphoric acid.

Samples when ashed at a very low temperature (the dishes radiate a
faint glow) for five hours in a muffle, gave practically no loss of P2O6. Nine-
teen samples gave the following results:

Tablb IV.

Ash, ordi- Percent, Percent. Percent.

nai7 method, ash. calcium Per cent. PsOs, ordi- PsOq, calcium Per cent,

low redness. acetate method, loss, ash. nary method acetate method. loss, PjOc.

1.95 1-93 10 O 91 0-93 2.0

Thirteen of the above samples varied less than 0.10 per cent, between
ash results by the two methods and all of the phosphoric add results
checked within o.io per cent. Determination of ash and phosphoric acid
ate almost identical.

14 F> J> MOORB AND R. D. QALB.

All of our results show that the temperature of ashing below fuaon i
ft so important a factor where only the percentage of ash is desired, bu
tien determining the phosphorus as phosphoric add in ash the greates
ution must be observed to keep the temperature below the volatiEzatio:
>int of the combined phosphorus.

THE COLORED SALTS OF SCHIFF'S BASES.

A Contribution to Our Knowledge of Color as Related to Chemical

Constitution.

BV F. J, MOOBE iKD R. D, Galb.
Received January 6. 1908.

The Hydrochlorides of Bases Formed by Condensing p-Amino D
methylaniline with Aromatic Aldehydes.
The starting point of the present investigation was a chance observi
an made upon the compound produced by condensing ;f>-aminodiinetby
liline with piperonal. This product is of a light orange color and hi
le formula

,0. ,CH.

CH/ >C,H, — CH = N — CH, — N<

When this substance, either in the dry state or in ethereal or benzei
ilution, is treated with dry hydrochloric acid gas, one molecule of tl
tter is first added to form a salt of a. deep blood-red color. This sa
in further add another molecule of the acid to form a dihydrochlorid
he color of the latter salt, in sharp contrast to that of the former, is
right lemon-yellow.

We found this phenomenon so striking that we determined to prepa

number of compounds of analogous constitution and study the col
■ their salts. The first substances selected were those most strict
[lalogous to the one already mentioned, namely, the condensation pro
:ts of /»-aminodimethylaniline with aromatic aldehydes. These has
imish the subject matter of the present paper. They all show a t
avior toward hydrochloric acid, and (so far as has been tested) al
)ward other acids, entirely analogous to that described in the case
le piperonal compound. A minor exception has to be noted iii t
ise of anisaldehyde. When its condensation product with ^-amin
imethylaniline is treated with hydrochloric acid, the same color chanfi
re observed as in the other cases, but analysis of the products show th

maximum of nearly three molecules of the acid is here absorbed. Tl
ehavior will be discussed more fully in the experimental part.

COLORED SALTS OI^ SCHIFF'S BASQS. 395

Subsequent papers will deal with other compounds similarly consti-
tuted. Thus we have prepared a number of bases of this class by con-
densing aromatic aldehydes with />-aminodiethylaniline. Of these it
may be said that as far as investigated, they show the same reactions
toward adds as the corresponding dimethyl compounds. The salts
are, however, less stable. One of us, in collaboration with Mr. R. G.
Woodbridge, Jr., is also studying the bases formed by condensing p-amino-
diphenylamine with aldehydes. These bases unite with one molecule
of hydrochloric acid to form dark red salts like the monohydrochlorides
already described. These salts, however, do not add a second molecule
of the add.

From the above it is clear that we are dealing with a quite general
law which may be stated thus: Bases of the general formula

.CH3

R_CH=N-<_J>-N<;^^^

add hydrochloric add to form salts of a dark red color (darker than that
of the free base) while most of them also add another molecule of the
add to form salts of a light yellow color (usually lighter than that of
the free base).

Before attempting a theoretical explanation of this uniform behavior,
a word should be said concerning the color of the simpler benzylidene
compounds containing only one atom of nitrogen. As aromatic alde-
hydes condense so readily with primary amines, a voluminous literature
has grown up on the subject. We have not made an exhaustive study
of this, but the general rule seems to be that ndther such bases nor their
salts are very highly colored. This certainly holds true of those which
have come under our observation. Thus the condensation product of
benzaldehyde with aniline is a pale cream color (perhaps due to a trace
of impurity) and the product formed by treating piperonal with aniline
is perfectly colorless. The hydrochloride of the piperonal compound is a
bright lemon-yellow, that of the benzaldehyde compound a much paler
shade, only slightly stronger than that of the free base. For purposes
of comparison we have prepared several other condensation products
of piperonal. All of these, namely the compounds prepared by treating
piperonal with />-toluidine, /)-chloraniline, /)-bromaniline, />-aminoethyl-
benzoate, and w-nitraniline, all showed the same light color of base and
salt

From this we can draw the conclusion that if the group — CH = N —
is to be considered a chromophore at all, it is a weak one ; certainly far
weaker than — N = N — . Of the intensifying effect of add addition
or salt formation, whether we call it "auxochrome" effect or "halo-

^ ' F. J. MOOKB AND R. D. GALE.

hromy," we may say that in this group of compounds it is noticeabl
ut not striking.
Turning back now to the more complicated bases containing two atom
f nitrogen, inspection of their formulae shows the presence of the sut

tituted amino group, — N^ , which acts as an auxochrome in s

lany important dyes. The efifect of this group is seen in the color {
he bases which contain it. These colors range from a light yellow
hrough golden yellow, to a fairly strong orange. None go as far as dee
ed. The color of the salts has already been dwelt upon.
Our interpretation of the color relations so frequently referred to wi
epend upon our idea of exactly what happens when one molecule i
lydrochloric acid is added to these bases. If we take for purposes i
[lustration the simplest case, that of benzylidene /i-ammodimethj
niline,

/CH,
-N< ,

re can imagine hydrochloric acid being added to the compound in sui
lifferent ways as to form products of any one of the following formula
/CH,
(I) C,H,-CHC1- """" ' ^ -■■

(=)

(5) C.H,-CH,-

)f these, (i) and (2) seem hardly worthy of serious consideration. T
iddition products of acids to the simpler bases of this type, such as benz;
dene aniline, CjHj — CH^N — C,Hj, have always been looked upon
immonium salts. They spUt at once under the influence of water ii
he corresponding aldehyde and the salt of the base from which th
rere ori^ally formed, in this case, aniline hydrochloride and benz
lehyde. The salts of the more complicated compounds now tmder d

COLORBD SALTS OP SCHIFF'S BASES. 397

cussion act in a similar way. This would exclude (i). It would not
necessarily exclude (2), according to which the addition products are
to be looked upon as secondary chloramines. They are, however, rather
more stable than chloramines would be expected to be, and they do not
evolve chlorine when they are treated with hydrochloric acid. A more
conclusive argument, however, against (i) and (2), and in favor of the
salt-like character of the compounds is the fact that so far as tested (and
only qualitative tests have thus far been made) sulphuric add gives
addition products entirely similar in color and properties to those formed
by hydrochloric acid.

Formulae (3) and (4) represent more nearly what we should expect
if the addition of hydrochloric add is to be looked upon as simple salt
formation. The formulae differ only in the nitrogen atom, which is sup-
posed to become pentavalent. In dedding between them we have to
remember that the simpler bases like piperonylidene aniline add one mole-
cule of hydrochloric add to form salts which are not dark red but light
yellow. Now these salts must have their hydrochloric add united to
that nitrogen which forms part of the chain connecting the two benzene
rings. It would therefore seem quite probable that these red salts have
their hydrochloric add botmd differently. Furthermore, if we assign
to the red salt the formula (4), a consistent and plausible explanation
lies near at hand for the difference in color between the salts containing
one and two molecules of hydrochloric add respectively. In the first
case, we have three factors codperating to intensify the color, the chromo-

yCH,
phore group, — CH = N — , the auxochrome group, N^ , and, finally,

whatever influence in this direction is to be ascribed to salt formation —
"halochromy."

Now the compound with two molecules of hydrochloric add can hardly
have any other formula than the following:

CeHj— CH«N--< >— N-<

a CI

H Cl H Cl

Here the nitrogen belonging to the chromophore group, — CH = N — ,
has changed its valence, and this might naturally be expected to prove
destructive to the chromophore character of the group. Furthermore,
the light yellow color of these saturated salts is in entire harmony with
the similar color of the simpler bases which, as already pointed out,
must have the hydrochloric acid upon the central nitrogen.

An argument against the explanation just outlined might be found in
the hydrolysis of the salts. It is a well knov\m fact that the salts of the
simpler bases like benzylidene aniline decompose at once when treated

398 P. J. HOORB AND R. D. GALB.

with water, yielding, in this case, aniline hydrochloride and benzaldehydc
The salts of the more complicated bases we are studjdng react ^mikrl)
except that the operation takes place in two steps. If the yellow hydra
chloride of piperonylidene ^-aminodimethyl aniline, for example, b
treated with water, it immediately turns dark red, obviously owing t
the formation of the red monohydrochloride. The color then slowl
fades out as this salt splits into the hydrochloride of /i-aniinodimetby:
aniline and piperonal. Now the weak spot in the hydrochloride of pipei
onylidene aniline is obviously the molecule of add attached to the eei
tral nitrogen. If the red salt has then a constitution analogous to tha
shown in formula (4), it is not, at first sight, quite dear why it shoul
bydrolyze at all. The above argument, however, against (4) does n<
seem as serious as that which can be brought against (3). If we gi\
the latter formula to the red hydrochloride, we can only ascribe the ii
tensity of its color to salt formation — halochromy. We should then a
pect a further intensification of the color when the second molecule 1
hydrochloric add is added. As we have already seen, this is not tt
case.

There remains the quinoid formula, (5). This also offers a con^tei
explanation. We have only to assume that the red hydrochloride e:
ists in the quinoid, the yellow dihydrochloride in the benzoid form,
seems at first a little difficult to account for the hydrolysis into aldehyi
and amine salt on this basis, but the shifting of double bonds involve
is not of a particularly revolutionary character. It would also seem po
sible that the dihydrochlorides might also exist in a quinoid form. Pe
haps, when more compoimds have been studied, such salts will be founi
In the meantime, some work is under way in this laboratory designs
to throw light upon the probabiUty of a quinoid formula for the red hydn
chlorides.

An explanation, differing from any of the above, has occurred to \
ance reading a recent article by Anselmino' which came to our notii
after the experimental work described in this paper was practically con
pleted. Anselmino found that when ^-homosalicylic aldehyde is coi
densed with aniline the product formed exists in two difi'eient forms,
red and a yellow. That this is not simply a case of dimorphism is shov
by the fact that the two forms differ in chemical behavior and give diffe
ent derivatives. The most probable assumption concerning their coi
stitution would seem to be that they are stereoisomeric in the same sen:
as the isomeric oximes of unsymmetrical ketones. If we try to explai
the behavior of the compounds we are studying upon a similar ba^, v
should have to say that the salts vary in color because they are derive
from different stereoisomeric bases, one red the other yellow. That i
' Ber., 40, 3465 {1907) ; also Ibid., 38, 3989 (1905).

COLORED SAINTS OF SCHIFP'S BASES. 399

the case of the bases themselves, the yellow form is alone stable; that
when the bases add hydrochloric acid, the yellow salts first formed are
unstable and go over to form salts of the other — red base. Finally,
when the saturated salts are formed, that a transformation in the oppo-
site sense takes place. This explanation disregards the question as to
which nitrogen holds the hydrochloric add. Now, in this connection,
we have to remember that previous to this recent work of Anselmino,
no well attested cases of stereoisomerism had been observed among
benzylidene compoimds, though they have long been looked for; and,
further, that Anselmino himself observed no such isomerism in the case
of the salts. In fact he points out that his red and yellow bases gave
identical hydrochlorides. This, of course, is no conclusive argument
against a stereochemical explanation of the facts which we have ob-
served. It does, however, furnish a reason for not accepting such an
explanation hastily.

It will be seen in the experimental part of this paper that the salts of
salicylidene />-aminodimethylaniline undergo change in color on stand-
ing, and it has been already pointed out that compotmds of this class con-
taining the diethylamino group show similar behavior. From what has
already been said, it will be seen in what a variety of directions, an explana-
tion of this fact might be sought. Any discussion of these possibilities would,
however, be premature imtil more work upon these ethyl compotmds has
been done. For the present also we wish to reserve any opinion as tcf
the probable formulae of the red hydrochlorides.

Experimental Part.

The condensation product of />-aminodimethylaniline with benzalde-
hyde was first prepared by Calm.^ It is interesting to note his observa-
tion that this base adds two molecules of hydrochloric add to form a
"white" salt. This illustates the light color of the saturated hydrochlo-
ride, and at the same time shows that Calm attached no significance to
the formation of the red intermediate product. He gives a determina-
tion of chlorine in the saturated salt. The other bases studied in this
paper were prepared by Nuth' at the suggestion of Calm. He did not
ptepare salts. We have little to add to his description of the bases,
except for some melting points already corrected by more recent observ-
ers. We prepared the bases by mixing molecular quantities of the amine
with the various aldehydes, sometimes warming a Uttle on the water
bath. The reaction then proceeded at once and gave good yields. The
product was recrystallized once from alcohol. It was usually foimd
that further recrystallization did not raise the melting-point. The bases
range in color from a light yellow to a light orange.

* Ber., 17, 2938 (1884).
" Ibid., 18, 573 (1885).

F. J. HOORB AND R. D. GALB.

studying the salts, we at first allowed hydrochloric add to act upoi
bases in the dry state. When piperonylidene />-aniinodimetbyI
le, for example, is spread upon a piece of porcelain (a crucible cover]
I current of hydrochloric acid passed over it, the mass suddenly turn;
;p red. This color no sooner appears than it begins again to fade
5 to the formation of the light yellow saturated salt. A simik
, is produced when the base is dissolved in dry ether and hydrochlorii
^s introduced. Here the liquid first turns dark red ; then a precipitab
E same color appears, and, finally, the time varying with the amoun
bstance present, the color of the solution grows rapidly lighter ii
;, the precipitate becomes bright yellow, and at the end, the supei
it liquid is colorless, the saturated salt being practically insolubl
her. It will frequently be noticed that at the mouth of the tub
e the gas enters, and where consequently the add is always in ex
a yellow deposit of the dihydrochloride forms at once. This make
)bable that the red salt, however it is prepared, contains a good dea
e yellow enclosed. This shows itself in the analytical data fumisbe
er on. When the salts were prepared for analy^s, the following pre
-e was employed in order to avoid this kind of contamination a
I as possible. First, the free base was dissolved in sodium-driei
and then a solution of hydrochloric add in dry ether was run i
a burette. When preparing the saturated salt, the addition of th
solution was continued as long as any predpitate formed. To pn
an unsaturated salt, a little less than half as much add was adde
id been found necessary in preparing the saturated one. The pn
ites so obtained'were filtered by suction, washed repeatedly with dr
, and finally dried in a vacuum desiccator containing both coneei
d sulphuric acid and sticks of caustic soda. If lumps formed durin
rjing, these were ground up and the desiccation continued. As fs
r observation goes, these salts melt only with decomposition, and tb
erature observed depends largely upon the time of heating. In tt
of the saturated salts, the point of decomposition lies quite do;
o° in almost all cases. The chlorine determinations were in all casf
: by the Carius method. The salts precipitated and dried as abo\
ibed were used for analysis without recrystalUzation. Alcohols woul
been the only practicable solvents for this purpose, but some exper
s made in this direction led us to fear the contaminating effects (
jlytic action. As we have pointed out above, the salts were probabi
or less contaminated with each other, and perhaps, also with the fn
. Under the circumstances, we have not thought it desirable to mak
1^ number of determinations, in the hope that some of them migl
more closelj' with the results of theoretical computation. I
of the wide divergence to be observed in many cases, we think ths

COLORED SALTS OF SCHIFF'S BASES. 401

the numerical results make plain the one point which we wish to empha-
size at this time, namely, that the light colored salts contain about one
molecule of acid more than the dark ones. The results follow:

Benzylidene p-Aminodimethylaniline. — Red hydrochloride: calculated
for C^Hj^NaHCl, CI, 13.60; found, 17.01. Saturated hydrochloride: calcu-
lated for CisHi^jNj. 2HCI, CI, 23.09; foimd, 20.48.

Cinnamylidene p-AminodimethylanUine. — Red hydrochloride: calcu-
lated for CijHigNjHCl, CI, 12.36; found, 15.33. Saturated hydrochloride:
Calculated for Ci7Hi8Nj.2HCl, CI, 21.95; found, 21.34.

Piperonylidene p-AminodimethylanUine. — Red hydrochloride: calcu-
lated for CieHieOjNaHCl, CI, 11.63; found, 15.19. Saturated hydrochlor-
ide: calculated for Ci^HigOjN3.2HCl, CI, 20.78; found, 21.12, 21.05.

Salicylidene p-Aminodimeihylaniline, — Red hydrochloride: calculated
for CijHieONjHCl, CI, 12.81; found, 13.07. Saturated hydrochloride:
calculated for C15H1JON3.2HCI, CI, 22.65; found, 22.47.

The color of the salts seemed less permanent in this case than in any
we had previously studied. Nuth records that the free base turns red
on standing in the air, and we notice that the red hydrochloride has a
tendency to grow lighter in color. What is perhaps more curious is
that the saturated hydrochloride which, when first precipitated, is almost
colorless, grows considerably darker in color. We have since met with
even more marked changes of this kind in the study of the bases formed
by condensing />-aminodiethylaniline with aldehydes. We consider
any theoretical discussion of this phenomenon to be premature until
these diethyl compounds have been more thoroughly studied. The
fact should be borne in mind, however, that, according to Anselmino,
salicylic acid is one of those which forms isomeric aniles, one red, the
other yellow.

Anisylidene p-AminodimethylanUine. — Red hydrochloride: calculated
for CieHi,ON,HCl, CI, 12.20; found, 12.20. C^^^fi^^.2RQ\, CI, 21.68;
found, 18.34. Saturated hydrochloride: calculated for CieH,gON2.3HCl,
CI, 29.27; found, 28.78, 26.16, 26.61.

In view of the fact that in the case of the red hydrochlorides high re-
sults for chlorine are usually obtained, probably owing to the tendency
of these salts to enclose some of the saturated compounds, the numer-
ical results given above leave it a little in doubt whether the red salt,
in this case, contaiBS, when pure, one or two molecules of acid. It is
quite evident, however, that the saturated salt contains nearly three mole-
cules. This seems to be one of those cases referred to by Baeyer^ where
a base shows itself capable of combining with more acid than its formula
accounts for. It might be supposed that in both salts the additional
molecule of hydrochloric acid attached itself to the methoxyl group to

* Bar., 38, 1 157 (1905).

403 p. J. MOORE AND R. D. GALB.

form an oxonium salt. This may be the correct explanation, but it i:
not necessarily so; for we find that the most simply constituted membe:
of the whole group, benzylidene aniline, though it conains but one nitro
gen and no oxygen, yet adds two molecules of hydrochloric add, as showt
by the following analysis:

Calculated for C,jH„NHCl, CI, 16.29; for CaH„N.2HCl, CI, 27.90
found, 27.68, 27.62.

We were much surprised at this result, as it raised the question whetbei
all of the benzylidene compounds containing one nitrogen might not be
have similarly, and whether in those containing two atoms, all of the acdc
might not be held by one nitrogen. The analyses of the salts of the thret
following compounds, however, show that the addition of more molecule:
of acid than there are nitrogen atoms in the base, is the exception; thougt
from what has gone before, it may well be of more frequent occurrenci
than has hitherto been supposed.

Piperonylidene Aniline. — This base was first prepared by Lorenz.
It is absolutely colorless, the first of these compounds which we havt
been able to prepare in that condition. The hydrochloride is bright
yellow, and this makes perhaps the most striking example of simple "halo
chromy" which we have observed in the group. The hydrochloride con
tains but one molecule of add, as is shown by the following analytica
data:

Calculated for C„H„0,NHC1, CI, 13.55; found, 13.76. 13.51-

Piperonylidene p-Tolwdine. — This substance crystallizes from alcoho!
in cream-colored prisms which melt at 98".

Calculated for C„H,,0^, C, 75.26; H, 5.48; N, 5.87. Found, 75.28
5-55. 6.11.

The hydrochloride is light yellow. The percentage of chlorine wa;
determined. Calculated for C„HyO,NHCI, CI, 12.86; found, 13.03
12.98.

Piperonylidene p-ChloranUine, — The base melts at 78°. It has nol
been analyzed. It forms a light yellow hydrochloride. Calculated foi
C„H„0,NCI.HC1, CI, 23.95; found, 23.43.

We wished to learn something of the color of salts analogous to those
formed by condensing aldehydes with ^-aminodimethylaniline, but
which contained an amino group not substituted by alkyl radicles. We
had some hopes that at least a small yield of such 4X)m pounds might be
obtained by condensing one molecule of aldehyde with one of a diamine.
Accordingly we treated piperonal with f-phenylenediamine, and also
with benzidine. In both cases the only products we obtained were
those formed by the condensation of one molecule of diamine with two
molecules of piperonal. These bases are soluble with difBculty in ako-
' Ber., 14, 79» (1881).

COLORED SALTS OP SCHIPP'S BASES. 403

hol, ether, or the aromatic hydrocarbons. They can be crystallized
from nitrobenzene, from which they separate in bronze yellow scales of
a semi-metallic luster. The compound with />-phenylenediamine melts
at 216®. Calculated for CJHifiJ^ii C, 70.94; H, 4.33; N, 7.55. Found,
70.30, 4.52, 7.90.

The compound formed by condensing benzidine with piperonal melts
not quite sharply at 241°. The liquid formed is not transparent, and
it is possible that it is crystalline in character, as liquid crystals are not
infrequently met with among compounds of similar constitution. The
composition of the base was verified by a nitrogen determination. Cal-
culated for C^HjpO^Nj, N. 6.50; found, 6.43.

We add some incomplete data concerning some substances prepared
during the present investigation, but which, as they have no further
theoretical interest for us, will probably not be worked with further.

Piperonylidene p-Aminoethylhenzoate. — ^This substance was prepared
by condensing piperonal with />-aminoethylbenzoate. The base melts
at 109®, is almost colorless, and forms a yellow hydrochloride.

Calculated for C17H15O4N, C, 68.64; H, 5.09; N, 4.72. Found, 68.34,
5.08, 4.84.

Piperonal condenses readily with m-nitraniline to form a base which
melts at 119®. This forms a yellow hydrochloride. We have a nitrogen
determination in the free base. Calculated for Ci4Hii04N3, N, 10.35;
found, 10.56.

Piperonal condenses with />-bromaniline to form a product melting at
109°. This forms a hydrochloride of a light canary-yellow color. We
have analyzed neither the base nor the salt.

Summary.
When /^-aminodimethylaniline is treated with aromatic aldehydes,
condensation products are formed which have the general formula

.CH,

R- ' ^ '

'-=«-''-<ZXcH.

These bases add one molecule of hydrochloric add to form dark red
salts of a much deeper color than the free bases. The addition of more
hydrochloric add produces salts of a light yellow color, lighter than that
of the free base.

Three explanations of this behavior are considered:

(i) That the first molecule of add adds to the auxochrome nitrogen
augmenting the color, while the second adds to the chromophore nitro-
gen, changing its valence and consequently destroying its chromophore
character.

(2) That the monohydrochloride has a quinoid structure, while the
saturated salt and the free base are benzoid.

4 WM. LLOYD EVANS AND B&NJ. T. BROOKS.

(3) That there is a double series of stereoisoineric red and yellow base
d salts, the red form of the monohydrochloride being the stable one, th
llow form being stable in the other cases.

A decision as to which of these explanations is most applicable mm
deferred until more experimental material can be collected.

HASSACBUaSTTB IHBTITDTB OP TBCHHOLOOr.

4 FKOU THE Chshicai. LABORATORY OP THS Ohio Stats Univbrsity
ON THE OXIDATION OF META-NITROBENZOYL CARBINOL.

Bt William Llotd Bv>hs and Bbnjauik T. Brooks.

Received Januarj ij. i^oS.

The work of Nef on the oxidation phenomena exhibited by many serii

organic compoimds has made it possible to follow by experiment tl
act course taken by such reactions. The recent work of Denis' on tl
havior of various aldehydes, ketones and alcohols towards oxidizii
ents is an excellent example of this kind of experimental study. In
evious paper by one of us,* it was shown that reactions of this type i
2 benzoyl carbinol series lend themselves admirably to this kind of tiea
:nt.

Zincke* and his students were the first to show that benzoyl carbine
len acted upon by various oxidizing agents, gives mandeUc, benzo>
rmic and benzoic adds in varying amounts according to the agents use

a further study of this same substance by one of us,* it has been shon
at benzoyl-fonnaldehyde is also one of the products of oxidation of bei
yl'CarbinoL When benzoyl- formaldehyde is acted upon by alkalies
d copper salts,' at 100°, it undergoes a benziUc add rearrangemea
ring mandelic add exclusively. It has been shown by Denis that acet
rmaldehyde undergoes a similar rearrangement with dilute solutioi

spdium hydroxide, and even with water alone at 100° it suffers a pa

,1 transformation.*

It is a well-known fact that many orthodicarbonyl compounds,

-Lo'

' Ann. Chem., 318, 137; 335, 191; 357, =

'Am. Cbem, J., 38, 561.

'Evans, Ibid., 35, 115.

' Ber., 13, 63s; Ann. Chem., ai6, 311.

* Pechmann, Ber., so, 2904; aa, 2556.
' Evans, Am. Chem. J., 35, 134.

• Ibid., 38, 584. j8s.

META-NITROBENZOYL CARBINOlv. 405

IC5
by the addition of one molecule of water, undergo a ben2dlic add rearrange-
ment, giving an a-hydroxy add,

X.Y.C.(OH).COOH.
In some cases, such as diketosucdnic ester, hexaketomethylene iand di-
ketobutyric ester,^ this transformation takes place in the presence of water,
but in most instances the presence of an alkali is necessary. For this re-
markable rearrangement Nef^ has offered the following explanation: The
orthodicarbonyl compotmd undergoes the following dissociation:

X — C = 0 Xv .0—

I -^ \Ci^+-CO,

Y— c = o y/

these products of dissociation then uniting to form an a-lactone,

\c^^^l^c = O,

which then, with a molecule of water, undergoes hydrolysis, forming the
corresponding a-hydroxy acid, as:

Xv /O- Xv /OH

}C^ C — + HOH m^ >C^^- COOH.

y/ 6 y/

The interpretation of Gabriel' concerning the transformation of o-carboxyl-
benzoyl-formaldehyde into phthalidcarbonic add through the interme-
diate formation of o-carboxylmandelic add becomes, therefore, perfectly
evident. Upon this basis, also, the behavior of benzoyl formaldehyde,
C,H5=X, H=y,* towards oxidizing agents was readily explained. With
mercuric oxide and silver oxide the reactions merely progress through
the first stage, the rate of oxidation of the dissociated partides bdng greater
than that of recombination. With copper hydroxide and potassium ferri-
cyanide, in the presence of alkalies, the reaction progressed through the
second stage, mandelic add being the product in each case. The pres-
ence of benzoylformic add in the oxidation with alkaline potassium per-
manganate is due to the subsequent oxidation of the mandelic add formed
in the second stage, as has been shown by pr^dse experimental data.

Further experiments have been imdertaken in this laboratory having
for thdr main purpose: (i) to ascertain whether the same general course
of reaction is followed in the substituted benzoyl carbinols as has been de-
veloped previously, and (2) to discover what possible effect, if any, the
introduction of substituents into the ring might have on these substances
towards the oxidizing agents previously considered. This paper, which
constitutes a first report on this work, deals solely with the oxidation of

* Denis, Am. Chem. J., 38, 590.
' Ann. Chem., 335, 272, 273.

» Ber., 40, 81, 82.

* Loc. cii.

6 WM. LLOYD EVANS AND BBNJ. T. BROO^

nitiobenzoylcarbmol. Briefly, the results are as follows: (i) m-Nitro
ylcarbinol with freshly precipitated mercuric oxide, freshly precipit
ver oxide, and potassium permaDganate alone, or in the presenc
ustic alkalies, gives m-nitrobenzoic acid and carbon dioxide exclusi^
) m-nitFobenzoylcarbinol, with cupric hydroxide and caustic alki
/es only m-nitromandelic add; {3) m-nltrobenzoylcarbinol, with p
im ferricyanide and potassium hydroxide, gives both m-nitrobei
id and m-nitromandelic add. By comparison, it is seen that the d
ces thus far developed lie in -the behavior of m-nitrobenzoyi carl
wards potassium permanganate and alkalies, and potassium ferncya
d alkalies. The general method previously employed for the pre]
<n of benzoyl carbinol was found to be the most convenient for the pi
ion of OT-nitrobenzoyl carbinol, viz., the hydrolysis of the corresji
f acetate which had, in turn, been prepared from the bromide.

Experimental Part.
Preparation of m-Niiromonobromacelophenone. — The following me
: the preparation of m-nitromonobromacetophenone was found t
ich more satisfactory than methods previously described in the li
re.' A solution of 24.2 grams (one molecule) of bromine in 40 c
loroform was allowed to flow slowly into a warm solution of 25 g
le molecule) of m-nitroacetophenone dissolved in 200 cc of the :
vent, the operation being carried on in the sunlight. The first red
the successively added portions of the bromine solution was iramedi
icbarged, the final color of the reaction mixture being yellow, .
itilling the chloroform, taking up the residue in ether and washing
ute sodium carbonate solurion, the dried ethereal solution ga\'e a
llow, crystalline residue, which was identical in every respect with 1
imonobromacetopheaone described by Korten and Scholl.' This me
ve a theoretical yield of 37 grams, which, when purified from a nm
equal portions of ether and ligioin (6o°-8o°), gave a melting-poi:
°. The same general procedure may also be followed in the prepan
ffi-nitrodibromacetophenone. Into a boiling solution of 5 grams
>lecule) of m-nitroacetophenone, in chloroform, were added slowly
tms (two molecules) of bromine dissolved in 25 cc. of the same sol
; operation being carried on in the direct sunlight. The quickly d
zed reaction mixture was then treated as above, giving finally a
9.8 grams of a yellow oil, which solidified on cooling to light yi
'stals melting at 61°' (Engler found 59°). Applying the above ge
ithod to the prepamtion of dibromacetophenone, 20 grams (one mole
monobromacetophenone, prepared according to the direction of Mob

' Hunnius, Ber., 10, 200S. Korten and Scholl: Ibid., 34, 1909.

*Loe.eii.

' Ber., 18, 324<3.

* Ibid., 15, 3464.

MI^A-NITROBENZOYL CARBINOU 407

were dissolved in 100 cc of boiling chloroform and to this solution were
added slowly 16 grams (one molecule) of bromine in 30 cc of the same
solvent. A final yield of 27.0 grams (theory =« 27.9 grams) of the crude
substance was thus obtained.

Preparation of m-Nitroacetophenone Acetate. — In preparing the
M-acetate of m-nitroacetophenone the same general method was followed
as previously indicated by one of us in the preparation of the o»-acetate of
acetophenone. It was found in the experiment herewith described that
the temperature had to be carefully controlled owing to the tendency of
the reacting substances to form tar. A mixture of 15 grams (one mole-
cule) of m-uitrobromacetophenone, dissolved in 60 cc. of glacial acetic
add, and 7.56 grams (1.5 molecules) of powdered fused sodium acetate^
was heated tmder a reverse condenser on an oil bath, the temperature
being brought gradually to 105^, at which point it was maintained for
two hours. Sodium bromide began to separate out at 90-95®, the larger
portion coming down, however, at 105°. Finally the temperature was
raised to 115® for one hour, after which the reaction mixture was poured
into six volumes of water. The separated reddish-brown oil soon solidified
on standing, after which the crystals were filtered and dissolved in ether.
To this solution of the acetate was added that which was obtained by
extracting the filtrate six times with ether. The combined solutions were
washed with dilute sodium carbonate, after which the dried ether was
evaporated to one-fifth of its volume, whereby the acetate began to
crystallize out. By successive evaporation of the mother-liquor, a yield
of i3«3 grams (theory =13.7 grams) of the w-nitroacetophenone acetate
was finally obtained. The substance (m. p. 51°), purified from a mixture
of ether and ligroin, gave the following analysis:

Calculated for C^^HgOgN: C, 53.81 ; H, 4.03; N, 6.30. Found: C, 53.89,
53.80; H, 4.10, 4.10; N, 6.70.

Preparation of Meta-NitrobenzoylcarbinoL — Twenty-five grams of m-
nitroacetophenone acetate were hydrolyzed by means of 500 cc. of boiling
water containing i cc. of dilute (1:1) sulphuric acid, the time of hydrolysis
being thus reduced from nine to four hours without apparently increasiag
the yield of tar. After filtering and cooling, 16 grams (theory 20.5 grams)
of the light yellow crystalline carbinol were obtained (m. p. 92.5-93**),
this yield comprising the product also obtained by the evaporation of the
second filtrate to one-fourth of its volume under diminished pressure.
The aqueous solution of the carbinol reduces ammoniacal silver nitrate
and Fehling's solutions. The analysis of the substance, purified from
benzene, was as follows:

Calculated for CgH^O^N: C, 53.03; H, 3.86; N, 7.73. Found: C, 53.20,
52.97; H, 3.88, 3.83; N, 8.50.

The following special experiment was performed: 0.20 gram of w-nitro-

408 WM. LUDYD EVANS AND BENJ. T. BROOKS.

benzoylcarbinol, dissolved in i cc. of glacial acetic acid, was ad
0.5 cc. of acetyl chloride for one hour at 50°. The oil, obtained
ing the reaction mixture into 25 cc. of water, soon crystallized
0.18 gram of a product which when purified from ether-ligroin 1
to be identical with the acetate above described.

m-Nilrobemoyl Carbinol, CupHc Hydtoxide and Sodium Hyi
m-Nitrobenzoyl carbinol, like benzoyl carbinol and acetol, redt
ing's solution in the cold and also like these two latter substan
with alkaline cupric hydroxide the corresponding a-hydroxy
m-nitromandelic acid. The point of dissociation of the salts o:
and secondary alcohols being much lower than the free alco'
reaction undoubtedly proceetls as follows:

NO,.C,H,CO.CH,ONa —^ NO,.C,H,.CO.Ch/ +HON
the m-nitrobenzoylmethylene thus formed is then oxidized tc
benzoylformaldehyde, which in turn, in the presence of the alkt
goes a benzilic acid rearrangement to m-nitroraandeiic acid.^ The
of behavior of m-nitrobenzoyl carbinol to acetol and benzoyl c
shown by the following experiment: To a solution of 2.0 grams (
cule) of fft-nitrobenzoyi carbinol in 300 cc. of water were added
(two molecules) of crystalline copper sulphate in 20 cc. of w
the addition to the mixture of 3.31 grams (7.5 molecules) of sc
droxide dissolved in 20 cc. of water, an instantaneous reduction t
oxide took place in the cold. After standing 24 hpiirs the yel
was filtered and washed till the total volume was 800 cc. .
with dilute sulphuric acid (i : 1) and extracting 6 times with etli
of r.93 grams of m-nitromandelic acid was obtained (theory 2,1
A study of the action of copper salts on m-nitrobenzoyl carbii
being made in this laboratory in the hope of obtaining the iut
. aldehyde. In preliminary experiments with ciipric sulphate a
obtained which did not react with Fehling's solution even after
boiling and which, with phenylhydrazine, gave a beautiful yell
crystalline product.

m-Nilrobenzoyl Carbinol and Potassium Permanganate. — Th
difference in the behavior of benzoyl carbinol and m-iiitrobenzo]
towards potassium permanganate and alkalies is shown in the
experiment in which the substituted carbinol gave the corr
benzoic acid exclusively and no trace of the keto-acid as does
substituted substance. A mixture of 2.62 grams (three mol
potassium permanganate and 0.66 gram (three molecules) 1
hydroxide in 100 cc. of water was added to a solution of i gram <

' Nef, Ann. Chem., 318, 138.

' Compare Evans, Am, Chera. J., 35, 125. Also Denis, Ibid., 3^ 584.

META-NITROBENZOYL CARBINOL. 409

benzoyl carbinol in 150 cc. of water, the resulting temperature being 28®.
The color of the first additions of the permanganate was instantly dis-
charged, the reaction being accompanied by a gradual separation of the
oxide of manganese. After standing four hours, the excess of perman-
ganate was reduced with a few drops of alcohol, after which the aqueous
filtrate was worked up in the usual manner. There was obtained 0.86
gram of the pure m-nitrobenzoic acid (theory =0.92 gram). A solution
of 5.22 grams of potassium permanganate in 100 cc. of water when added
to a solution of 2 grams of m-nitrobenzoyl carbinol in 300 cc. of water,
reacted instantly when carried out as in the previous experiment, a final
yield of 1.78 grams (theory =1.84 grams) of pure m-nitrobenzoic add
being obtained — a result in harmony with the studies on benzoyl carbinol.
f-Lactic acid and r-mandcHc acid are alike in their general behavior
towards alkaline potassium permanganate in that both give acetylformic
add and benzoylf ormic acid, respectively. ^ In striking contraist to this gen-
eral behavior is that of w-nitromandelic add, which with these agents gives
w-nitrobenzoic add exclusively as is proved by the following experiment:
A mixture of 2 grams of m-nitromandelic add and 0.84 gram of potassium
hydroxide in 50 cc. of water was added to a solution of 3.20 grams (two
molecules) of potassium permanganate and 1.12 grams (two molecules) of
potassium hydroxide in 100 cc. of water. After standing twenty-four hours,
a yield of 1.6 1 grams (95 per cent, of theory) of pure w-nitrobenzoic add
was obtained. From this behavior of m-nitromandelic add towards alkaline
potassium permanganate, it was concluded that m-nitrobenzoylfomric
add should also be completely oxidized to m-nitrobenzoic add. That
this expectation was fully realized is shown by the following experiment:
To a solution of 4.03 grams (one molecule) of m-nitrobenzoylformic add
and 1.03 grams (1.^5 molecules) of sodium hydroxide in 100 cc. of water
was added a solution of 3.45 grams of potassium permanganate in 100 cc.
of water. After standing at room temperature for twenty-four hours the
reaction mixture )rielded 3.29 grams (theory = 3.45 grams) of m-nitro-
benzdc add.

ffirNitrohemoyl Carbtnolf Potassium Ferricyanide and Potassium Hv'
droxide. — m-Nitrobenzoyl carbinol in its beha\dor towards alkaUne potas-
sium ferricyanide is different from benzoyl carbinol in that it gives m-nitro-
benzoic add in addition to its corresponding hydroxy add, as the follow-
ing experiment shows: One gram of m-nitrobenzoyl carbinol dissolved
in 150 cc. of water was added to a mixture of 10.91 grams (six molecules)
of potassium ferricyanide and 2.48 grams (eight molecules) of potassium
hydroxide dissolved in 100 cc. of water, the resulting temperature being
28.5**. After standing over night the solution was addified with dilute

1 Compare Ulzer and Sddd, Monatsh. Chem., 18, 138. Also Denis, Am. Chem.
J-. 38, 575.

:0 WU. t4X>YD EVANS AND BENJ. T. BROOKS.

ilpfauric add and extracted six times with ether. There was thus ob
ijied 0.93 gram of add material (m. p. 85-95°) which, upon dissohiiii

15 cc. of hot water and cooling suddenly, gave 0.27 gram of a flak;
ledpitate (m. p. 138-139°) which proved to be m-nitrobenzoic add, th
]ueous filtrate containing m-nitromandelic add. A study of the b«
ivior of m-nitromandelic add towards alkaline potassium ferricyanid
as made as follows: To a solution of 2.0 grams (one molecule) of m-nitit
andelic add and 3.98 grams of potassium hydroxide (seven molecules

50 cc. of water were added 16.71 grams of potassium ferricyanide, th
suiting temperature being 24". After standing twenty-four hours an
orking up as above a yield of 1.75 grams (m. p. 85-90°) of add materii
as obtained which gave no reaction with phcnylhydrazine' thus showin
le absence of m-nitrobenzoylfomiic add. In view of this result, tli
illowing experiment needs no further explanation: To a solution of 1,3
■ams (one molecule) of tn-nitrobenzoylfomiic add and i.oo gram <
idium hydroxide, in 100 cc. of water, were added 13.47 grams of potai
um ferricyanide and 3.26 grams of potassium hydroxide in 100 cc. i
ater. After standing at room temperature for 24 hours, the reactio
ixture yielded i.io grams (theory=i.i4 grams) of m-nitrobenzoic adi
he reaction mixtures when addified with dilute sulphuric acid gave tl
russian blue test with ferric chloride.

m-Nitrobemoyl Carbinal and Silver Oxide. — wi-Nitrobenzoyl carbin
hen acted upon by silver oxide either in the presence or absence 1
kalies gives m-nitrobenzoic add, a result which is in harmony with tt
^on of the same reagents on benzoylcarbinol and with the experinien:
' Denis on acetol,' It is perfectly evident that m-nilrobenzoyl carbim
assesses a possibility of dissociating in two directions, viz. :

(1) NO,.C,H,.CO.CH,OH —^ NO,C^..CO.Ch/ +HOH
id (2) NO,.C,H«.CO.CH,OH »* N0,.C,H,.CHO + H,CO.

Consequently, in the absence of alkalies, it is possible for tn-nitn
>nzoic add to arise from an oxidation of the dissodatcd particles i
juation (2).* One gram (one molecule) of tn-nitrobenzoyl carbim
issolved in 150 cc. of water was added to 2.5 molecules of freshly pn
pitated and well washed silver oxide, prepared from 4.68 grams of dlvt
itrate. On standing nine hours at room temperature a slight reductio
as observed, accompanied by a brown coloration of the carbinol solutioi
he temperature was raised to 70° for one hour and finally to 100° till tli
rolution of carbon dioxide ceased, determined by its action on bariui

' Fefarlin, Ber., 33, 1576.

* Am. Cbem. J., 38, 579.

■Compare Evana, Am. Cbem. J., 35, 119. Nef, Ann, Chem., 335, 369. Als
ling, Ann. chlm. phys. (&), Si 5>9-

BiETA-NlTROBBNZOYi; CARBINOL. 4II

hydroxide solution. Treating the reaction mixture with an excess of
aimnoniuin hydroxide, carefully acidifying the filtrate from the metallic
sihtT and working up in the usual way, there was obtained 0.80 gram

I..' ory=o.92) of m-nitrobenzoic acid. The action of w-nitrobenzoyl
cail.h.ol on silver oxide in the presence of alkalies was carried on as fol-
lows A solution of i gram of the carbinol in 150 cc. of water was added
slowK to four molecules of freshly precipitated and well washed silver
oxide (from 4.68 grams of silver nitrate) suspended in 50 cc. of an aqueous
solution of sodium hydroxide containing 1.13 grams, the resulting tem-
perature being 29°. At the end of three days the reaction mixture gave
0.74 gram (80 per cent.) of w-nitrobenzoic acid and 2.19 grams (theory =
2.56 grams) of metallic silver. The following special experiment shows
that the presence of w-nitrobenzoic add could not be attributed to the
intermediate formation of m-nitromandelic add and its subsequent oxida-
tion. Two grams of w-nitromandelic add and 0.85 gram sodium hj^-
droxide dissolved in 100 cc. of water were added to an emulsion of 4.71
grams of freshly predpitated and well washed silver oxide, prepared
from 6.94 grams of silver nitrate. After standing at room temperature
for 24 hours and working up the reaction mixture In the usual way, i .90
grams of the original add were recovered. A similar result was ob-
tained by allowing the reaction mixture to stand six weeks.

m-Nitrobemoyl Carbinol and Mercuric Oxide. — The recent experiments
of Denis* confirm the observations of Nef and Kling; namely, that acetol
with mercuric oxide in the presence of alkalies gives r-lactic add instead
of acetic add, as one would expect from the course of the analogous re-
action in the aromatic series. w-Nitrobenzoyl carbinol is analogous to
the behavior of benzoyl carbinol towards alkaline mercuric oxide as the
following experiments show: A solution of i gram of w-nitrobenzoyl
carbinol in 150 cc. of water was added to 2.5 molecules of freshly pre-
cipitated and well washed mercuric oxide (from 3.74 grams of mercuric
chloride) and 1.54 grams of sodium hydroxide, reduction taking place
instantly. After standing 24 hours the reaction mixture )delded 0.68
gram of i»-nitrobenzoic add. Repeating the same experiment but in
the absence of alkali, there was obtained 0.84 gram of m-nitrobenzoic
add. In the latter experiment, however, the reaction mixture after
standing 24 hours was kept on a boiling water bath for six hours, during
which time a large amount of barium carbonate was predpitated from the
attached barium hydroxide solution. A spedal experiment with w-nitro-
mandelic add and alkaline mercuric oxide showed that this substance is
amilar in its behavior towards these agents as it is towards silver oxide
and caustic alkalies. One gram (one molecule) of w-nitromandelic add
and 1. 10 grams of sodium hydroxide dissolved in 100 cc. of water, when
* Loc. cit.

added
oxide

reduct

I.oni
ptnus I
which
lightwi
becaus
VBiiets
formit
that ii
derivat

The
produG
destnu
mote Q
with p
compo

ThR
the gn
the oil
sold re;
dissolv
methot
dissolv
works
rubber
in am]
self at
cables,
icgaidi

When
by the
petfun

IX>NG LSAIf PINE OIL. 413

understood and repeatedly stated during the last three years in what may
be called the commercial literature, that its chief constituent is terpineol.
In the scientific literature however, I can find only one single reference
to its composition. Walker,^ from its distillation and from its fonnation
of a hydrochloride melting at about 50*^, surmised that it might contain a
terpineol. His experiments, however, were made on a small amount of
material produced on a laboratory scale only.

The commercial long leaf oil, as it comes on the market, is either clear
and water white, containing 3 or 4 per cent, of dissolved water, or it may
have a very faint yellow color and be free from dissolved water. The
specific gravity ranges from 0.935 to 0.947, depending on freedom from
lower boiling terpenes. A good commercial product will begin distilling
at about 206° to 210® and 75 per cent, of it will distil between the limits
2ii**-2i8®, and 50 per cent, of it between 213-217°. A sample having
a density of 0.945 at 15.5° showed a specific rotation of about [a]^° — n°,
and an index of refraction of N^ 1.4830. In fractional distillation of the
oil the specific gravity of the various distillates rises regularly with increas-
ing temperature, becoming steady at about 0.947 at 217 ^

If the oil consists essentially of terpineol, CioHj.O, it should be easy to
convert it into terpin hydrate, CioHjoOj+HjO, by the method of Tiemann,
and Schmidt.' The conversion was found to proceed easily when the
oil was treated with 5 per cent, sulphuric acid, either with or without ad-
mixture with benzene. If agitated continuously, the reaction is com-
plete within 3 or 4 days. If, on the other hand, the mixture is allowed
to stand quietly, the formation of terpin hydrate extends over several
months and produces most beautiful large crystals, which, without recrys-
tallizing, melt at 117-118°. When recrystallized from ethyl acetate they
melt at 118°. Yield, about 60 per cent, of the theoretical This forms
such a simple, cheap and convenient method of making terpin hydrate
that it will doubtless supersede the usual manufacture from turpentine,
alcohol and nitric add, and instead of terpin hydrate serving as raw ma-
terial for the manufactture of terpineol, as heretofore, the reverse will be
the case.

Terpineol Nitrosochloride. — This compoimd was made in the usual
manner from amyl nitrite, glacial acetic add and hydrochloric acid. The
yield was good and the product, after crystallization from ethyl acetate,
proved to be very stable. Melting point 101-103°.

Terpineol Nitropiperidide. — ^This compoimd was made from the nitroso-
chbride and piperidine according to Wallach*s method. The once recrys-
tallized product melted at 158-159° and formed well-shaped crystals
from methvl alcohol.

' Massachusetts Institute of Technology Bulletin, September, 1905.
*Bcr., 28, 1781.

414 W. D. RICHARDSON.

Terpineol nitranilide was made from the nitrosochloride and aniE
dissolved in alcohol On dilution of the reaction mixture with a lit
water good crystals of the nitranilide were obtained.

Dehydration Products. — ^Wallach' found that when 25 grams of t
pineol were heated with dilute sulphuric acid, there resulted 16 ca t
pinene and cineol boiling at 177-180", 4 cc. boiling at 181-185°, ^^^ ■
cc. boiling at 185-187°, containing terpinolene. This experiment w
repeated, using a sample of bng leaf pine oil, of which nearly the whi
distilled at 216-218°. 100 grams of this oil heated for one hour wi
400 cc. of dilute sulphuric add (1:2) and then distilled with steam a
dried over sodium, gave 8g cc. of oils volatile with steam. On redistil
tion the following fractions were obtained: 5 cc. at 165-175°, 56 1
at 175-180°, 6 cc. at 181-185°, and 15 cc at 186-192°. The largi
fraction had a specific gravity of 0.860 and evidently consisted essentia
of terpinene. It will be seen that these results agree quite closely m
those obtained by Wallach from pure terpineol.

In conclusion, the specific gravity, index of refraction and boiling poi
of the long leaf pine oil, its absorption of bromine in quantity correspoi
ing to a dibromide, the formation of a nitrosochloride, nitropiperidi
and nitranilide, the formation of terpinene by dehydration and the ea
conversion and large yield of terpin hydrate leave little doubt that the 1
sentia! constituent of long leaf pine oil is a terpineol and is probal
the optically active levo modification of «-terpineol (A'-ZJ-menthenoK!
boiling at 217-218°.

This seems to be the first recorded occurrence of terpineol in any me;
tier of the pine family. Even in this case it was obtained not from t
living tree but from lightwood, i. e., from portions of the tree which h
been cut at least three years and were very resinous. It is the comm
belief of those familiar with lightwood that its resinous content increa:
regularly from the time it is cut from the tree or from the death of the tr
If any such action occurs the appearance of terpineol here would set
quite natural, as its formation from pinene requires only the additi
of one molecule of water accompanied by splitting the tetra ring of pinei

The investigation of this oil is being continued.

164 Fkoht ST.. New Voiik City,

Deeemlwr. 1907.

TRAHSPAKEHT SOAP— A SUPERCOOLED SOLnTIOH.

BT W. D. RlCHiltOSOM.

Received January ij, 1908.

The alkali-metal salts of the fatty adds — commonly called soaps
have been usually considered crystalloids by the older authorities and
others in recent years as colloids.*

1 Wallach, Ann., 330, 225.

' Leimdorfer, Seifcnsied. Ztg. 190C, Nos. 24-19; Merkleo, Cbem. Ah*. 1907, a*
IiVwlcowitBd), J. Soc Chem. Ind., a^ 590 (1907).

TRANSPARENT SOAP. 415

The criteria of colloids are: (i) slow or no diffusion through colloidal
membranes; (2) failure to raise the boiling-point or lower the freezing-
point of solvents; (3) electrical migration; (4) coagulation phenomena;
(5) absorption phenomena; (6) hysteresis; (7) impenetrability for other
collQids; (8) optical inhomogeneity; (9) separation by salts ("salting
out"). Of these criteria at least (2), (5), (7), and (9) hold good for soap.

It is probably true that many facts in regard to the formation of soap
and also many properties of finished soaps can be explained on the as-
sumption that they are colloids forming in solution reversible hydrosols
in the classification of Zsigmondy/ but it is equally true that they crys-
tallize, if not quickly, at least slowly and definitely. In all likelihood
the ordinary pure soap of commerce consists chiefly of a compact mass
of soap crystals embedded in a non-crystalline soap phase. A thin sec-
tion of such a soap transmits light through crossed Nicols and while this
phenomenon can be explained upon other assumptions, the facts of opacity
and the presence of straight line boundaries indicate the presence of
masses of crystals. Possibly, the best view to take of the matter would
be to consider soap as a substance with distinctly crystalline tendencies,
which on accotmt of the size of its molecules, its viscosity and its solu-
bility, which varies considerably with the temperature, exhibits colloidal
properties.

In cold water many of the alkali metal salts of the saturated higher
fatty adds are but sparingly soluble, whereas at temperatures under 100°
and ordinary pressures, solutions are readily formed consisting of 70
to 80 per cent, of soap and 30 to 20 per cent, of water. These solutions
are quite transparent, but upon cooling they become opaque. The opacity
of some soaps is in part due to the presence of an emulsion of soap and
fatty matter, and this is well illustrated in the case of cold-made soaps,
m which saponification is purposely left incomplete; but in the instance
mentioned the opaque mass is formed of pure soap and water.

A soap may, however, be of such a composition that when it soHdifies,
instead of becoming opaque, it remains clear and such a soap is known
in the trade as transparent soap. A familiar example of such a product
is a cooled alcoholic solution of soap. This, upon solidification, assumes
the appearance of a jelly and if not perfectly transparent is at least
translucent. A solidified dilute water solution of soap is also frequently
translucent; but whether the jelly be made with alcohol or water, soap
crystals of rather imperfect shape usually form in time. It may be said
in passing that transparency in soap is not a mark of purity or even of
definite composition. Probably a better soap can be made of the ordi-
nary opaque sort. The best known brand of transparent soap on the
market contains 20 per cent, of rosin, calculated on the basis of total

^ Zaigaumdy, Zur Erkenntnis der Kolloide.

4l6 W. D. RICHARDSON.

adds, and rosin must be considered as an inferior soap-making materia
The demand for a transparent soap must be attributed to a whim i
that portion of the public to whom transparencj' is synonymous wil
purity.

In commercial tmnsparent soaps there are usually present one or mo
substances which appear to act as retarders, preventing crystallLzatio
Among these substances are glycerol, ethyl and methyl alcohols, cai
sugar and alkali-metal salts of rosin. Also certain soaps (such as t1
sodium soap of castor oil) seem to have this power. While some of the
substances, at least under certain conditions, maintain thi
parent state by means of a considerable solvent powei
they appear to act simply as retarders, preventing or h
lization. In mentioning retarders, the action of certain
as protein substances boiled with lime water, upon the sei
Paris might be cited. One part of retarder to one tho
may retard the setting time an hour or two; large quan
it an infinitely long time. Such seems to be the action of
stances mentioned upon the crystallization of soaps. 1
another way also (since some are hygroscopic) by preven
desiccation under ordinary circumstances and, therefon
the soap from assuming the supersaturated condition. Thus some tiau
parent soaps which would never form crystals when exposed to the ai
do so when kept in a defecator over calcium chloride or sulphuric ad
When removed from the desiccator and placed in a moist chamber i)
soap swells, absorbs water and the crystals in time dissolve. The san
soap, returned to the desiccator, may not form crystals the second tim
It is ordinarily stated in text -books that there are two ways of prepa
ing transparent soap; one by dissolving dry soap in alcohol, distilling c
the alcohol and moulding the syrupy residue; the other by "filling
molten soap with sugar, syrup, etc. As a matter of fact, there are m
two but many ways of manufacturing transparent soap, but the pri
dple in all cases is essentially the same, namely adding to a soap sol
tion substances wliich will form a jelly and retard crystallization. Pu
soap, when dissolved in alcohol and the alcohol distilled oS entire
or in part, will form not a transparent soap, but an opaque one. If tl
20 per cent, of rosin were left out of the brand for which the first procc
mentioned above is claimed and if no glycerol were present, it is que
tionable whether or not a transparent soap would be the result. Oi
formula for transparent soap coiomonly given in the literature is:

Tallow 100 parts

Cocoanut oil 100 "

38° B£. caustic soda soluti<»i 150 "

Glycerol 40 "

Alcohol So "

Explanation of Plate.
A, B and C, cakes of transparent soap containing naturally occiirring soap crys-
tals ; A and B contain many disc-shaped crystals, C fewer. D, E, F and H contain
crystals which have been formed by seeding molten soap with crystal fragments from
A and B. In the lower part of D is an area where crystals have not formed, probably
owing to absence of nuclei, G is a bar in which a rod of soap containing crystals,
cnt from C by means of a cork Iwrer. has been inserted and from which crystals have
grown, although they do not show well in the illustration.

TRANSPARENT SOAP. 417

This formula does not make a transparent soap but a decidedly opaque
and crystalline one. One hundred parts of castor dil and the necessary
caustic soda may be added besides, the glycerol and alcohol may be trebled,
but the product is still opaque. If now rosin soap is added or cane sugar,
or both, a perfectly transparent soap results.

As transparent soap comes from the kettle it is quite clear and trans-
parent; after solidification in frame or mould it is usually, though not
always, less clear. After partial desiccation it becomes quite trans-
parent again to the eye, when viewed with ordinary light, although it
probably diffuses light to some extent. Immediately after manufacture
transparent soap does not transmit light between crossed Nicols but
with age this property increases until finally, when viewed even in thin
sections, a soft bluish light is transmitted by such soaps of the most varied
compositions. It is the usual rule with transparent soap that after
aging it reaches a condition after which no great change ensues; the
bars remain transparent and firm and do not lighten or darken in color.
But occasionally soap crystals form in a bar which is in all visible respects
like the normal bar; further, these crystals may form in a bar of identical
composition with one free of crystals; the bar with crystals and the one
without may indeed have come from the same batch of soap or the same
frame. The crystals when first visible are about one^half millimeter
in diameter; they may enlarge until a diameter of fifteen millimeters
is reached. When one millimeter in diameter they appear under the
microscope in section as fine needles radiating from a center, which is
somewhat opaque, into a clear matrix. Between crossed Nicols the ap-
pearance is striking. The central portion transmits light in a marked
manner and appears red and yellow; surrounding the central portion
in a circle of bluish light; surrounding this and reaching almost to the
crystal-needle ends is a black ring. Just outside the crystals is a bright
bluish ring; this is in the transparent region and beyond this is the uniform
dim bluish area characteristic of the non-crystalline soap. The outer
bright, bluish ring would appear to indicate a condition of unusual strain
in the soap. The crystals usually become noticeable in from four to
eight weeks and they grow in size and numbers for an indefinite time,
depending on conditions. Sometimes they appear in as short a time
as one week and again they may not become visible for from eight to
ten months. In some cases only a few crystals develop, in others the
bar finally appears as a mass of soap crystals lying in a clear matrix.
Attempts have been made to separate the crystals from the matrix by
solution in hot and cold alcohol and water and other solvents, but un-
successfully, although the cr3rstals differ to some extent in their physical
behavior from that of the transparent portion of the soaps. The crys-
tals have been separated by hand in small quantities by excision, and

4l8 W. D, RICHARDSON.

their analysis indicates their compo^tion to be tliat of ordinary p
soap. The amount of material thus separated was small and proba
impure, owing to the fine needle form of the crystals and their ramif]
tendency. The fatty acids were separated from them as well as li
the clear matrix surrounding them and melting-point determinati
made upon the adds, lliose from the crystals melted at 43-44°, n
those from the clear portion melted at 36-38" C. These figures esi
lish little beyond the fact that it is the soaps of the harder fatty a
which form the crystals and that these soaps are probably compt
of saturated adds, for the melting-points of unsaturated fatty adds
low.

The facts indicated above, that crystals form in one bar of soap
not in another when both are from the same batch and that the in
duction of soap crystals leads to the formation of crystals in othen
non-crystalline bars, would lead to the conclu^ou that nuclei of si
sort are one of the essentials leading to crystal formations. Ther
another fact of interest in this connection, namely that crystals are
quently formed abundantly in the interior of a bar whereas the suri
to a depth of from two to six millimeters is free from crystals. It is
parent that this condition is connected with the progress of desiccai
in the cake, for the line of demarcation between well dried surface
moister interior corresponds closely with the boundary between c
talline and non -crystalline soap. The surface of soap dries quiddj
a depth of about one millimeter, but thereafter the desiccation is
tremely slow, inasmuch as the surface skin so formed protects the iate
against the loss of moisture. The dried portion is very hard and to
and consequently possessed of great viscosity and rigidity. In sue
medium, even though it be supercooled, crystal formation is almost
not quite, impossible.

I have used the term "supercooled" in referring to transparent soaps rai
than "supersaturated," because the substance is in the solid conditio)
amorphous and may start to crystallize at once upon solidification f
the kettle without any evaporation of the solvent. While there 1
be a slight impropriety in the use of the term, it appeared fully as us
in describing the properties of the soap as the term "supersaturated."

Now the characteristic of the supercooled or supersaturated co
tion is its instability as regards crystallization. The means suffic
for bringing about equilibrium are shock, or the introduction of a
or isomorphous crystal. Viscosity or rigidity in an amorphous substs
may render crystallization extremely slow or prevent it altogether,
example, the devitrification of glass is usually a very slow process.

If a soap crystal be introduced into a bar of transparent soap an
the solvents are not present in excess (that is, if the soap is indeed in 1

TRANSPARENT SOAP. 419

dition for ciystallization) radiating crystals will start to grow from the
one introduced and will spread to an indefinite extent through the soap,
if sufficient time be allowed. If soap crystals be ground in a mortar
and ever so small a weight of them be introduced into transparent soap
as it is about to solidify, the soap will in time become a mass of crystals.
If, on the other hand, the same weight of soap crystals be dissolved in
a small quantity of alcohol, and this solution added to the molten trans-
parent soap, and if the soap be then filtered as a special precaution against
the introduction of nuclei, no crystals will form. In one series of ex-
periments holes were made by means of a cork borer in bars of transpar-
ent soap, which contained no crystals, and into these holes, plugs of soap
cut from a crystalline bar with the same cork borer were pressed. Around
the plugs so introduced masses of crystals formed and are still forming,
although some of the experiments are now a year and a half old. In
the parts of the bars remote from the crystalline plug no crystals have
formed at all. Not all nuclei induce the formation of crystals. For
instance, in one series of experiments particles of crystalline siUca were
incorporated in molten soap, and these, during one and one-half years'
time, developed no crystals, although soap of identical composition,
when seeded with soap crystals, produced a large crop. Experiments
on various substances as nuclei are now in progress.

The influence of shock or physical strain in the production of soap crys-
tals is shown by the tendency of the crystallization to proceed in certain
planes which follow the lines of strain produced in the soap by the im-
press of the die used in stamping the bars. This produces crystals, in
the form of circular discs (of radiating needles) rather than spheres;
also the crystals near the surface frequently cluster around a line or edge
where the die has struck, rather than against a plane surface of the soap.

It will be seen that no one explanation will accoimt for the formation
of crjrstals in transparent soap under all conditions. The inertia or
hysteresis attending change of state may be accentuated or retarded
in this case as in others by a variety of circumstances. The principal
influences and conditions affecting the formation of crystals may be grouped
as follows:

(i) Supersaturation or the supercooled condition presupposing a con-
dition of strain.

(2) Absence of sufficient quantities of retarders to prevent crystalliza-
tion altogether.

(3) Desiccation: provided the soap is not supersaturated at the time
of manufacture, desiccation is necessary to bring about conditions suita-
ble for crystal formation. As desiccation proceeds there is apparently
an increase in strain produced in the soap as shown by the transmission
of light between crossed Nicols.

(4) Nuclei appear to be the principal exciting: cause of crystal fon
on when other conditions are favorable. These nuclei are soap a
Us or particles and possibly other substances. Experiments are r
I progress in regard to this point.

{5) Not too great viscosity or this may overcome the cry5talli2
>rces. Rapid desiccation may entirely prevent crystal formation.

(6) Shock produced by blows or pressure act favorably on cry
)rmation in many cases.

In conclusion, crystal formation, while it does not reduce in any
ree the detergent power of a soap, renders it more or less unsalabb

public which buys according to the appearance of goods. The exj
lentat work detailed above was undertaken in order to examine
;niatically the causes of crystal formation in transparent soap anc
rovide a remedy.

BOTES.

Rapid Detetmtnation of Petroleum Naphtha in Turpentine. — To
:rmine the amount of petroleum naphtha in a suspected saniph
irpentiue, 10 cc. are carefully measured into a 50 cc, carbon tube, w!

graduated into tenths of a cubic centimeter.

Thirty cubic centimeters of aniline are now added and the mixture
ntly shaken for five minutes, and left to settle until the Uquid has
>me perfectly clear, when, if there is any petroleum naphtha pre
L the turpentine.it will be thrown out of solution and float in a 1
a top, and the percentage can be readily ascertained. I get exce
:sults from this method, but care must be exercised that the an
oes not contain any water. Henry C, Feb

Determination of Sodium and Potassium in Silicates. — ^We 1
lund the following method of determining sodium and potassium in
nd sihcates which can be decomposed by sulphuric and hydrofli
dds, rapid and satisfactory.

One gram of clay is decomposed by means of sulphuric and hy
uoric acids, and the excess of sulphuric add expelled in a hot air b
he residue is then dissolved in water and powdered barium hj-d
le added to the boiling liquid to alkaline reaction, llie solutio
Ecanted and filtered and the residue boiled again with water and t
ughly \vashed. Carbon dioxide is passed into the filtrate in ex
le solution evaporated to 50 cc., 25 cc. of alcohol (96 per cent.) ad

NOTES. 491

and the solution filtered and the residue washed with 50 per cent, alco-
hol. A measured excess of N/io hydrochloric add is then added to
the filtrate and the solution boiled to expel the carbon dioxide, Utnius
being used as an indicator and more acid being added if necessary, to
give a permanent add. reaction after boiling. The titration is then fin-
ished with N/10 sodium hydroxide. The solution is evaporated to
dryness in platinum, dried at 110° and finally at very faint redness
and the residue of potassium and sodium chlorides weighed. The amount
of each metal can be calculated on the following principle:
Let a = No. of cc. of N/io HCl used less the No. of cc. of N/io NaOH.

b =aX A 0.00585 = weightof NaClequivalenttosumofNaCl-h KCl.

c = weight of NaCl -f KCl formed less the weight of NaCl corre-
sponding to the weight of NaOH used.

X = weight of Na.

y =5 wdght of K.

Then b = 5?:^ X + ^^ y.
23-05 39- 1 5

^ ^ 58^ X -f ^-^ y.
23-05 39- 1 5

y = 2.432 (c-b).

23.05 / 74.6 \
X = ^^ (c-^^ y) = 0.3937 c-0.75 y.

Three samples of clay gave by the above method:

By I«awrence Smith's method. By the above method.

< * '" » * * »

K. Na. K. Na.

A 0.76 0.14 0.77 0.14

B 0.33 0.53 0.33 0.45

C 0.18 1.39 o 14 1 42

J. E. Thomsen.

Laboratory of Joseph Dixon Crucible Co.,
Jersey City, N. J.

The Determination of Total Nitrogen Inchiding Nitrates in the Pres-
erice of Chlorides, — Asboth, Jodlbauer and Scovell have modified the
Kjeldahl nitrogen method with the view of making possible the detenni-
nation of nitric nitrogen simultaneously with organic nitrogen. In the
presence of common salt, however, these modifications, and also the
method of the Official Agricultural Chemists (wliich is essentially! Sco-
vell's) are inapplicable; for the sulphuric acid used in the method acts
upon chlorides and nitrates producing hydrochloric and nitric acids and
before the latter is reduced to ammonia by the reducing agents present,
the following reaction occurs:

HNOs + 3HCI «-► 2H2O + CLj + NOCL

433 REVIEWS.

The pickling solutions used for curing meats contain usually sal
saltpeter and sugar; and after use they contain also the class of substanc
knovm as meat bases and various proteins. Cured meats themselv
also contain all these compounds. The Kjeldahl method and its mot
fications are inapplicable, therefore, to the determination of total i
trogen in cured meats and pickling solutions. Various methods we
tried in this laboratory to arrive at the amount of total nitrogen in the
two products and the following was devised for the purpose: (i) c
tennine nitric nitrogen by the Schloessing- Wagner method; (2) in a
other portion iletermiue nitrogen excluding nitrates by adding to t
substance in the Kjeldahl flask 10 cc. more or less of saturated ferro
chloride solution and boiling with dilute sulphuric acid until nitrat
are destroyed. Then proceed with the determination of the remaini:
nitrogen by the Kjeldahl or Kjeldahl-C5unning method. The sura
(r) and {2) gives Ihe total nitrogen. A test solution was made cental
ing ten grams ammonium chloride, ten grams potassium nitrate and ?
grams sodium chloride in 1000 cc. By the Kjeldahl method inoi
ficd to include nitrates the following quantities of total nitrogen we
found in three determinations in aliquot parts of this solution: 0.09
g., 0.0883 g- ii"d 0.0834 g. calculated: o.io(x> g. By the method d
scribed above the following quantities were found: nitric nitrogen 0.03
g. (calculated 0.0346 g,); other nitrogen by Kjeldahl method after 1
moval of nitrates, 0.0653 g. (calculated 0.0654 g) ; total nitrogen foui
o.iooo g. (calculated o.iooo g.). Mr. E. F. Scherubel assisted in tl
work. W. D. Richardson.

REVIEWS.

REVIEW OF ANALYTICAL WORK DONE IH i()o6.

In this review the only change from the plan of previous ones is tf
American work has been included. The writer's acknowledgment
his indebtedness to the Ckemisches ZeTttralblalt for general grouping
subjects and for abstracts is due again and is here made. He has ma
use occasionally also of abstracts in the Journals of the London Che
ical Society and Society of Chemical Industry.
General Analysis.

Apparatus. — A new Orsat apparatus was proposed by Bendema
(/, Gasbel. , 49, 583, from Z. Ver. Ing.) for analysis of the new power ga:
which contain something like 30% of carbon monoxide, 12% of hydrog
and only traces of methane. Two cuprous chloride pipettes are usi
and where considerable amounts of oxygen are to be absorbed eitl

REVIBWS. 413

phosphorus or a second pyrogallol pipette. About 30 cc. of gas are
taken for the combustion. The combustion pipette is best covered with
a water jacket connected with the water mantle of the burette. Lux
(Ibid., 49y 475) described the Raupp gas calorimeter as better than the
Junker. It consists chiefly of a copper cylinder whose lower part is solid,
the hollow upper part carrying a tiiermometer divided into tenths. Un-
der the copper body is placed at a certain time the gas flame, whose
height has been previously determined, and the time necessary for the
thermometer to rise 10® is noted. The apparatus is standardized by
means of gases of known heating value, so from a table and from the
measured time the heating value can be calculated. McDowall (Chem.
News, 94, 104) recommended the use, instead of the ivory scale on bal-
ances, of a brass one which should move by a horizontal screw under
the box engaging a tooth attached to the scale. After the pans are re-
leased the scale may be moved till the pointer swings equally on either
side of the zero. An apparatus was described by Weimem (J, russ, phys.-
chem, Ges. 38, 228) for determining the solubility of solids in liquids.
It consists essentially of two glass cylinders held together by an oblique
tubular connection projecting downward inside the lower cylinder. The
liquid is saturated with the solid in the upper cylinder which has a large
stirrer that is used also to force the Uquid through the side arm contain-
ing a wadding tampon and so into the weighed glass receiving vessel.
The lower cylinder has an arrangement by which the cover of the weigh-
ing vessel may be placed before the vessel is removed from the cylinder.
It is then weighed, the liquid evaporated, and weighed again.

Combustion and Heating Value. — ^A good deal of work was published
on the various combustion methods noted in previous reviews. Car-
rasco and Plancher {Gazz. chim. itaL, 36 II, 492) gave more details con-
cerning their method of internal electrical heating in the use of which
priority was claimed by Morse and Gray (Am. Chem. /., 35, 451). Denn-
stedt (Z. angew. Chem., 19, 517; Z. anal. Chem., 45, 26) and Dennstedt
and Hassler (J. Gasbel, 49, 45) gave more details with regard to Denn-
stedt's simplified combustion method. The second of these three arti-
cles is in reply to the criticisms of Hermann which were maintained by
the latter (Z. anal. Chem., 45, 236). The last of the three contains the
modifications of the method necessary for the analysis of coals. Holde
(Ber., 39, 161 5) and Holde and Schliiter (Mitth. kgl. Materialprufungsamt
Gross Lichterfelde West, 24, 268) gave the results of some experiments
with the Dennstedt and Heraeus furnaces, mentioning as some of the
difficulties of Dennstedt's rapid method those of obtaining the correct
agreement between oxygen addition and rapidity of combustion, the
occasional failure of the platinum quartz to glow as a criterion for the
proper carrying out of the method and the necessity of constant watch-
ing. Dennstedt (Ber., 39, 1623) replied, saying that the commonest
error in the rapid method is the too rapid volatilization of the substance.
This should be run with an oxygen current of about 60 cc. per minute
instead of running the oxygen current according to the rapidity of volatil-
ization. Von Konek (Ibid., 39, 2263) added his favorable experiences
with the Heraeus furnace. Marek (/. pr. Chem. [2] 73, 359; 74, 237)
recommended the use of a 5 cm. layer of copper oxide or copper oxide
asbestos in combustion analyses instead of the layer of ordinary length.
This proposal was criticized by Deimstedt (Ibid., [2] 73, 570).

V number of metbods for detennining the halogens in organic compound
re proposed. That of Berry {Chem. News, 94, 188) is a combioatlo
those of Piria and Schiff (ignition ^th sodium carbonate and lime
i the thiocyanate method of Volhard. The cooled mass is dissolve
dilute nitric acid (1:4), kept cool, excess of tenth-normal silver n
te added, the silver balide filtered and the filtrate titrated with tentl
mal potassium thiocyanate, using ammonium iron alum as ind
or. With substances containing iodine, sodium carbonate alme :
d lor the ignition. Moir {Proc. Chem. Soc., 23, 261) heated the sul
ace in a nickel crucible with 10 drops of water and pure caustic potas
a water bath, stirring with a platinum wire, then gradually decon
ed the product with 0.5 to i gram of finely pulverized potassiui
manganate, evaporated and drove out the precipitated carbon h
ition. The cooled crucible was brought into a warm dilute solutic
primary potassium sulphite, the solution acidified with acetic ad
n filtered into a silver nitrate solution. The silver halide was dete
led as usual. Or the cooled crucible might be, put into water, tl
ition acidified with acetic add till the manganate is converted inl
manganic acid, the latter destroyed with barium peroxide, the f
^d solution neutralized with primary sodium carbonate and thf
ated. Robinson {Am. Chem. /., 35, 531) recommended a filling ■

combustion tube as usual with copper oxide, placing also in it a ca
Ige filled with lead chromate, like Morse and Taylor's {Ibid., 33, 60
ingement for the combustion of sulphur-bearing compounds. Vai

and Scheuer {Chem.-Ztg., 31, 67) proposed to weigh out o.a to o
m of substance in a dry 150 to 200 cc. distilling flask, setting a se
tory funnel in place in the neck of the flask. The side arm was coi
ted with a Volhard flask in such a way that it did not dip into tl
er nitrate. 30 to 50 cc. of concentrated sulphuric add were adde

stopper of the separatory funnel closed, the liquid warmed gradual
I a weak current of air passed through the apparatus either from tl
inning or at the end of the experiment, according to the ease wi'
ch the gases are evolved. Iodine was driven out of the side tu!

warming. To insure the formation of silver iodide and not ioda
3gen-free filter paper or metallic copper was placed in the distillu
k to increase the amount of sulphur dioxide formed. In the V<
d flask silver halide and sulphite were formed, gathered into a beak(
ited with water and about 50 cc. of concentrated nitric add, boili

sulphur dioxide was driven out or all silver sulphite was changi
^Iphate, diluted till the sulphate dissolved and the halide determini

weighing or by titration of the silver in solution. Nitrogen mi
determined at the same time. Bianchi {Boll. chim. farm., 45, 82
icized this method, saying that it was not in general usable for (
ic chlorine derivatives; he did not pass on its value for bromine ai
ne derivatives. If it does work for these it may be simplified 1
ig the Volhard volumetric method. The Kjeldahl nitrogen dete
lation may be made at the same time all right.
'as. — Franzen {Ber., 39, 2069) proposed the use of sodium hydi
)hite for the absorption of oxygen, i gram absorbing 64 cc. of t
Na,SjO, + HjO + O = 2NaHS0,. 50 grams of the hydrost
:e are dissolved in 250 cc. of water and 40 cc. of sodium hydros

REVIEWS. 425

solution (500 grams in 700 grams of water) and the solution used in a
Hempel pipette for solid substances filled with iron wire gauze, i cc.
of this solution absorbs 10.7 cc. of oxygen. It has the advantages of
being cheaper than pyrogallol ; it may be used in weakly alkaline solu-
tion and has the same absorption at various temperatures; it may be
used with gases containing carbon monoxide (ammoniacal cuprous chlo-
ride cannot) and it may be used at lower temperatures and in the pres-
ence of substances that hinder the oxidation of phosphorus. For Bunte
burette determinations a weaker solution is used which has to be shaken
for three minutes. Haber (Z. angew. Chem., 19, 1418 and Z. Elektro-
chem,, 12, 519) with Lowe had made by Zeiss a gas refractometer con-
sisting of a prism telescope, a glass prism for the air or gas and a mirror.
The observer looks at a luminous spot within the telescope. On looking
into the instrument a dark shadow is seen which falls upon the scale
in a position corresponding to the gas in the prism. The method has
the advantage that the refractive index of the gas sample as compared
with air remains the same as long as both are subjected to the same pres-
sure and temperature changes. If chimney gases be led through the
prism an increase of 0.9% carbon dioxide causes the image to be dis-
placed I division of the scale. The instrument can be adapted for pho-
tographic registration or the shadow can be made visible at a distance.
The apparatus is sensitive to 0.600,0003 in the refractive index, corre-
sponding to a change of 0.2-0.25% carbon dioxide in chimney gases.
Methane, hydrogen, hydrochloric and hydrocyanic acids can be deter-
mined in air with the same delicacy. The delicacy is twice as .great
for acetylene and hydrogen sulphide, 2i times as great for sulphur di-
oxide and nearly 10 times as great for pentane and benzene vapor. No-
wicki (Oesterr, Z. Berg.-Huttenw., 54, 6) recommended oxidation with iodine
pentoxide as the most satisfactory way to determine carbon monoxide,
subsequently determining the carbon dioxide in various ways. Hy-
drogen sulphide is also easily oxidized. Acetylene acts on iodine pentox-
ide above 85®, but the carbon monoxide oxidation begins at 45° and
is finished at 88°. The pentoxide heated by itself begins to decompose
at 165® and the reaction is completed at 300°. Gautier and Clausmann
{CompL rend., 142, 485; jBw//. soc. chim, [3] 35, 513) made a similar recom-
mendation with regard to carbon monoxide.

Soil. — Hall, Miller and Marmu (Proc. Chem. Soc, 22, 103; /. Chem-
Soc., 89, 595) found that the commonly used moist oxidation with chro-
mium trioxide yields always 10 to 20% too low values because the oxi-
dation of the carbon does not proceed completely to the dioxide. They
inserted a short layer of copper oxide. Murray {Chem. News, 93, 40)
used the following apparatus for the mechanical analysis of soils. An
Erlenmeyer flask of about 200 cc. capacity without a flanged neck (neck
of 2 to 3 cm. diameter) is bound to a piece of glass tubing of the same
diameter as its neck by means of stout rubber tubing. 5 grams of fine,
air-dried earth suspended in ammonia solution are put into the flask,
it and the tube are filled carefully with water, the tube opening is closed
and the apparatus turned upside down, the mouth being brought into
a dish filled with water. Immediately under the tube opening is placed
a small porcelain dish which is replaced by others at stated times. The
portion deposited in each dish is evaporated and weighed. In order

436 RBvmws.

to get a common basis for the expresdon of results Murray recommetids
uniting the particles that fall through i cm. in i, then 5, 10, 20, 100 and
400 seconds.

Water. — Bruhns (Z. anal. Chem., 45, 473) stated that barium car-
bonate reacts alkaliiie and is attacked by dilute oxalic add, hence the
determination of carbon dioxide in water by precipitation with barium
hydroxide and back titration with tenth-nonnal oxalic acid without
filtering leads to errors. He used a tube carrying above a stopcock a
cylindrical vessel of 100 to 300 cc, capacity and placed just above the
stopcock glass wool or wadding and on top of that paper pulp, thus mak-
ing a filter. The measured sample was put into this vessel and protected
from air by a 3 to 5 mm. layer of benzene. The barium or caldum
hydroxide solution was run in from a burette whose tip reached below
the benzene layer. The mixture was carefully stirred, let stand to cleai
and the liquid run through the filter. The filtrate was caught undei
a 3 to 5 mm. layer of benzene and was titrated in small portions. Tenth-
normal oxalic acid was used for titration and neutrahzed phenolphthakis
for indicator if the water contained no magnesium. If ammonium chlo-
ride had to be added to keep magnesium in solution Utmus or methyl-
orange was used. It is not advisable to titrate under the benzene layer
because add can be occluded by it. Buisson {Compt. rend. 143, 289
and /. pkartn. chim. [6] 34, 289) studied the determination of ammo-
lua by Nessler's reagent. His analyses of the predpitate from ammo-
nium chloride and the reagent lead to the formula Hg,N,I„ agreemg
with the work of Franfois. The predpitate is soluble in potassium
iodide, hence the reaction is reversible and some ammonia escapes de-
tection (in one case 21%). The determination of ammonia carried
out by detennining the mercury in the predpitate is therefore inexact.
Drawe {Chem.-Zlg., 30, 530) determined nitric add by evaporating 100
cc, of water first with residue-free hydrochloric add, then several times
with water, then determining the chlorine in the solution of the residue.
He subtracted from the total number of cc. of tenth-normal silver nitrate
solution the number of cc. used in titration of the same amount of watei
sample and the number of cc. of tenth-normal acid combined in the deter-
mination of carbonate hardness in 100 cc. The difference was calculated tc
nitric add. Kiihn {Arbl. kais Gesundheitsaml., 23, 389) determined
minimal amounts of lead by adding to 4 or 5 liters of the water a solu-
tion of 25 cc. of acetic add and 500 cc. of sodium sulphide solution (8: 500)
mixed just before using. The coagulation of the colloidal lead sulphide
was aided by the addition of 100 grams of sodium tutrate and the Uquid
v&s shaken with 2 grams of short fibered asbestos. The predpitate was
filtered on asbestos, using suction, the lead sulphide oxidized by hy-
drogen peroxide containing a Uttle nitric add and the lead sulphate
dissolved in sodium acetate solution. The rest of the method is that
of Diehl andTopf {Dingier' s pol. J., 246, 196; Z.anal.Ckem.,26, 137, 277)
It is accurate to i mg. of lead in i liter of water. Phelps (This Jour-
nal, 28, 368) determined small amounts of copper by evaporating 1 litei
of the water to about 75 cc. (for o.i to i gram of copper), and adding 2
to 5 cc. of sulphuric add and electrolyzing, using the dish as anode, with
a current of ND,,, = 0.3 ampere for 4 hours or over night with gentie
stirring. The cathode was removed without interrupting the current

RKvmws. 427

and dipped into boiled nitric add. This solution was evaporated to
diyness, the residue taken up in water, put into a 100 cc. Nessler tube,
filled to the mark and lo cc. of potassium sulphide solution (equal vol-
umes of 10% caustic potash and saturated hydrogen sulphide solutions)
added. The copper sulphide color appeared and was compared with
the color in a similar tube of 10 cc. of the potassium sulphide solution
diluted with water and standard copper sulphate solution added (0.2
cc. at a time) till colors were the same. For i liter of water taken, i cc.
of the copper sulphate solution is equivalent to a copper content of 0.2 :
loooooo. Raschig (Z. angew, Chem.y 19, 334) determined sulphuric
add by adding to the sample one-twentieth of its volume of a concen-
trated benzidine solution, stirring and allowing to stand 15 minutes.
If there were no precipitate the water contained 1.5 mg. of sulphur triox-
ide per liter or less. The precipitate was washed by suction, using very
little water, and titrated with tenth-normal sodium hydroxide (i cc. ==
4mg. sulphur trioxide). There should be added for the benzidine loss
1.5 mg. Iron does not interfere if i to 2 cc. of a 1% hydroxylamine hy-
drochloride solution be added before the benzidine precipitation. Scriba
(Z. physik. chem. Unierricht, 19, 298) stated that a paper strip dipped in
a solution of i gram of ferrous ammonium sulphate in 20 cc. of water,
dried and rubbed with pulverized potassium ferricyanide will give a
deep bhie spot with the smallest amount of water.

Volumeiric. — ^Acree and Brunei (Am. Chem, /.,36, 117) prepared a stand-
ard solution of hydrochloric acid by filling a clean liter flask nearly full
of water, running through a single holed rubber stopper a glass tube
with a long capillary reaching nearly to the bottom of the flask, weigh-
ing this to o.ooi gram with another flask as tare, then passing into it
a current of dry hydrochloric acid gas till the increase in weight is a little
more than i gram molecule. The flask is cooled to room temperature
before the last weighing. The solution is then made up to the mark
and the extra water added from a burette. A further standardization
is unnecessary. Solutions of other gases obtainable dry, as hydrobromic
and hydriodic acids, hydrogen sulphide, sulphur dioxide in any solvent,
may be thus prepared. They gave also an improvement over the or-
dinary gravimetric method for standardizing hydrochloric or sulphuric
add solutions. About 4.12 grams of twice recrystallized primary so-
dium carlx>nate are neutralized with the necessary amount of add (us-
ing methyl orange). The end point is reached when a weak rose-red
color persists after some standing in a vacuum. The solution is evap-
orated to dryness in a weighed platinum dish and the residue heated
to constant weight. From the weight of the sodium chloride or sul-
phate and the volume or weight of solution used it is easy to calculate
the strength of add. The method can be used with all acids, giving
sodium salts that can be dried and weighed. Richardson (/. Chem.
Ind.y 26, 78) standardized sulphuric add by neutralizing 5 cc. of dilute
weighed add with filtered saturated barium hydroxide solution, using
pbenolphthalein. The neutral solution was evaporated on the water
bath and the barium sulphate ignited and weighed. Riegler (BtUL
assoc. chim. stu:r. disU, 24, 528) recommended ammonium triiodate
[NH4H,(I03)3] as an original standardizing material. For tenth-nor-
mal solution 3.025 grams of the salt are dissolved in loo cc. of boiling

rater and this diluted to i liter. To standardize sodium thiosulphati
o cc. of water are put into a flask with i gram of potassium iodide, i ix
{ hydrochloric acid (d. 1.3) and 10 cc. of the above triiodate solutioi
nd the mixture titrated as usual with thiosulphate. sNH^HiCIO)),-!
Na^jO, = 3N&^p, + sNalO, + Nal + aHjO + sNH^IO,. The tri
Mlate acts as a dibasic acid and so can be used directly for alkalimetr
nth luteol, Congo red 01 diazonitranilineguaiacol as indicator. Tb
riiodate neutralized by sodium hydroxide is no longer affected by so
ium thiosulphate so the base to be determined can be treated will
n excess of triiodate and the excess titrated back with thiosulphate
tandardization of tenth-normal acid may also be effected gas volumetri
ally by letting the triiodate act on hydrazine sulphate and detenrnu
ig the nitrogen (Z. anal. Chem., 42, 677). Wagner, Rink and Scbultz
Ckem.-Zlg., 30, 1181} suggested the second method of Acree and Bnme
3r the standardization of acids and bases. They stated also that wher
here are tables showing the relation between refraction and concentra
ion the Zeiss immersion refractometer may be used to determine quan
itatively the reaction product. They gave a table for the relation be
ween refraction and concentration of nitric add.

Ahlum {Proc. Chem. Soc, 22, 63; /. Chem. Soc, 8q, 470) gave avolu
letric determination of free acid in the presence of iron salts, the iron bein,
recipitated with monosodium phosphate, the phosphate filtered ou
nd the filtrate titrated with sodium hydroxide. The add formed a

result of the iron predpitation, for example, FejCl. + 2NaH,P0, =
FePO< + sNaCl + 4HCI, is corrected for. Rupp {Z. anai. Chm.
S, 687) stated that permanganate oxidations run more rapidly ani
igorously in alkaline than in add solution, hence it is advisable in som'
ases to oxidize in alkaline solution, then to addify and titrate back tb
xcess of perman^nate according to Raschig {Ber., 38, 3911). Fonni
nd nitrous adds are cases given as examples. The formate solution
> warmed in a glass stoppered flask with considerable excess of pennan
anate standardized against sodium thiosulphate and with 0.5 giaii
f pure dry sodium carbonate for 15 to 30 minutes on the water bath
iluted after cooling with about 75 cc. of water, 25 cc. of dilute sulphurii
dd are added and i to 2 grams of potassium iodide, then the hberate(
)dine is titrated with tenth-normal thiosulphate (i cc '^ 0.0023 P^"
t formic acid = 0.0023 gram of nitrous add anion).

Brandt (Z. anal. Chem., 45, 95) found that the violet color of diphenyl
irbohydrazide (observed by Cazeneuve {Ckem.-Ztg., 24, 684; Bull, soc
'tim. [3] 25, 758] ) could be used to detect the end point of the bichro
late iron titration if a suifident amount of hydrochloric add were pres
nt with the manganese sulphate solution containing phosphoric add 0
le Reinhard method. There should be at least 60 to 80 cc. of hydrocbbiii
dd[d. 1.12] added to i-slitersof water containing 100 cc. of the manganesi
alution (6 kg, of sulphate, 33 liters of dilute sulphuric add [1:3] and 3 liter
f phosphoric add [d. 1,7] diluted to 60 liters), the iron solution and thi
idicator (about 5 cc. of 0.1% solution). Then on titrating a red-viole'
ilor is obtained, i>ecoming a mixed color as the green chromium salt in
reases till finally a sharp change to pure green is obtained at the end
Without the add the decomposition of the coloring matter is too energelii
□d takes place before the iron is oxidized. Iron in amounts less thai

RBVIEWS. 429

a2 gram are to be titrated without the manganese sulphate solution
which tends to weaken the oxidizing action on the coloring matter. With
kss than 0.1 gram ferric chloride must be added. Corsini (Giorn, chim.
farm,, 55, 200) recommended the use of tropaeolin OO in alcoholic solution
as indicator in the determination of free mineral acids in bromatology
instead of methyl violet. The color change from yellow to red-viofet
is sharper and more delicate than that of the latter, for example, for
sulphuric acid 1:20000 (1:10000 with methyl violet) and for hydro-
chloric acid 1.5 to 2:2000 (2 to 2.5:1000). Fenton (Proc. Camb. Phil.
5oc., 13, 298) stated that the previously described condensation deriva-
tive of methylfurfuraldehyde of the composition CnH804 may be used
as an indicator for strong acids and bases. A paper prepared from aque-
ous alcoholic solution gives with primary amines in acetic acid solution
an intense green color destroyed by mineral acids; with urea in the pres-
ence of strong hydrochloric acid a blue color; with alkalies a violet-blue
color destroyed by acids and disappearing on strong dilution. If the
substance be melted at 120° with urea a colorless base is obtained giving
a blue color with acids. Another acid indicator is obtained by boiling
the compound with ^-naphthylamine in alcohol solution which gives
with weak acids an intense green color. Salm (Z. Elektrochem., 12,
99, and Z. physik. Chem., 57, 471) stated that by dilution of hydrochlo-
ric add it is possible to prepare a series of standard solutions of' varying
hydrogen ion concentration. Solutions near the neutral point are better
prepared by mixing tenth-normal monosodium and disodium phosphate
solutions. For each one of these concentrations an indicator can be
found which will give a color change. By comparison with suitably
chosen standards the hydrogen ion concentration of solutions may be
accurately determined, for example, for oxalic acid K = 0.09 (o.i),
tartaric o.ooii (0.00097), fumaric o.ooii (0.00093) ^^^ camphoric
0.000,025 (0.000,0225). The figures in parentheses are from Ostwald's
electrical conductivity measurements. The dissociation constant of
an indicator base or acid is equal to the hydroxyl or hydrogen ion con-
centration of a solution half dissociated. With two-colored indicators
like litmus such a solution is formed at the color change. With singly
colored ones like phenolphthalein this solution is obtained by adding
acid to the completely reddened liquid till the color intensity has half
fallen away. The following dissociation constants are given:

Indicator acids. Indicator bases.

Methyl orange K = 4.6 X io"~* Cyanine K = 4.2 X lo""

^-Nitrophenol 2.3X io"~"^ Dimethylaminoazobenzene. . . . i 45 X 10 **

Rosolic add i.iX io~"

Alizarin 8.8 X io"~*

Phenolphthalein 8.0X 10""*®

Indicators and color changes for definite hydrogen ion concentrations are :

Concentration of H^. Indicator. Color change.

^ I Mauvein Yellow-green

X X lo^^N) " Green-green blue

1 X io""*Ni " Green blue-blue

I X io—*N I •• Blue-violet

iXxo"~*NJ Congo red Blue- violet

a-naphtolbenzoin

Tropaeolin

Trinitrobenzene

Safranine

Color chaiis*-

Violet-scarlet

Brown-red

Bright brownish red-ro

Colorless-roac

Gfecn -green blue
Green yellow-orange
Colorless-orange
Yellow (red )-rose
Rose red-violet

Schoorl {Chem. Weekblad, 3, 719, 771, 807) defined color indicate
I those coloring matters which possess distinctly different colors accor
g to whether the aqueous solution into which they are put contai
ee hydrogen or hydroxyl ions. In the first case they show "aci(
ilor, in the second, "basic" color. On transition from one color to tl
her they show "transition" color. An add delicate indicator is 01
lat shows transition color in aqueous solution that contains hydros
ns, as phenolphthalein, tropaeolin OOO and curcumin; alkali delica
dicators-are the opposite like /j-nitrophenol, lacmoid, Congo and meth
'ange and neutral indicators are those whose transition color she

aqueous solutions containing both ions, as rosolic add and litmi
he different behavior of indicators in pure water is explained easi
1 the basis of the difference in their sensibility quotients H: OH; t
ihavior of solutions set at transition color on warming is explained 1
le change in that same quotient as a result of the increased dissod
jn of water molecules. The following results are given in a cou[

tables :

ConcCDtration of Color Ukrn bj n

. . TrSMition Color io pore w«ter tion wllblransjli

Indicator.

leDolphtbaleiii
opaeolin OOO

H+ OH-

Colorless

Yellow

YeUow

Ydlow

Unchanged

Orange-red

Violet

aearycUow

Intense yell

Violet

Blue

Red

R<^

Yellow

YeUow

IQ— * Bright red

' io~' Brown -yd low

>solic add io"~' lo"* I Orange-red

tmus 10 ' io~' I Violet

Nitrophenol . io~* io~' 10' Clear yellow

icmoid io~* I9~' m' Violet

ingo io~* 10^" 10* Violet

sthyl orange. . io~* lO—" lO* Orange
ifferent values are obtained whether add be titrated with alkali,
ce versa. This can be helped by titrating to the same end color
>th cases, for example, to clear red with phenolphthalein. Vainest
ined with methyl orange and phenolphthakin are not the same; t
fference can be decreased by titrating to the transition color in ea
.se and neglected by titrating to the color of the indicator in water.
Optical. — Fredenhagen (Ann. Pkysik [4] 20, 133) dedded that t
lief series of the potassium and sodium lines are oxide spectra wh
le secondary series are due to the metals. The green thallium line
so an oxide line. He tried the alkalies and some other elements in t

REVIEWS. 43 1

ordinary, the carbon monoxide-oxygen and the hydrogen-chlorine flames.
In the second flame the same spectra were observed for the alkalies as
in the BunseQ flame. They gave no spectra in the hydrogen-chlorine
flame, but when the chlorine was cut out the spectra appeared. Cal-
cium and cuprous chloride gave chloride spectra in the hydrogen-chlo-
rine flame. Riesenfeld and Wohlers (Ber., 39, 2628) found that an ap-
proximate quantitative determination of the alkaline earths can be
made from the brilliancy of the lines of their spectra. With equal amounts
the red and green calcium, the orange and blue strontium Unes will ap-
pear equally bright but the two green barium lines are noticeably darker.
With more than twice as much of any one, its lines will appear dis-
tinctly brighter. For detection of calcium and strontium the spectral
analytical way is more delicate than the chemical, but for barium
the chemical is better. A combination solution for the standardiza-
tion of the spectral apparatus in one series of measurements may be
made up of 50 cc. of water, 10 cc. of 10% hydrochloric acid, i drop of
10% sodium hydroxide, 10 grams of potassium chloride, 3 grams of
strontium chloride, i gram of calcium chloride and 0.8 gram of lithium
chloride. This gives the Ka, Lia, Na D, Caa, Sr8 and K/? lines.

Horn {Am. Chem. /., 35, 253) and Horn and Blake (Jbid., 361, 95, 576)
stated that the assumption usually made in colorimetric work that the
delicacy of all solutions which are more dilute than those obviously un-
suited because of depth of color is practically the same, is not true. It
is variable and the ease with which a determination can be carried out
varies with the concentration, though a simple relation between them
does not exist. Measurements with potassium bichromate solutions
showed that the delicacy is greatest at concentrations of 0.004 to 0.008
normal calculated to gram atoms of chromium. The smallest amount
(rfchromium that can be detected in distilled water is 0.000013 gram while
0.000001 gram causes an easily told diflFerence at concentrations of max-
imum sensibility. The difference between depth of color of two colored
solutions is much more easily detected than the difference between a
colored and a colorless one. For colorimetric determinations the con-
centrations of greatest delicacy should be fotmd experimentally and the
conditions of experiment exactly laid down. The different amounts
of potassium bichromate and of copper sulphate which cause a notice-
able difference in color were determined. The reciprocals of these val-
ues plotted as ordinates with square roots of dilutions as abscissae gave
a curve with two distinct maxima for the bichromate and one for copper
sulphate. The copper sulphate curve corresponded to the first part
of the bichromate curve. The delicacy of colorimetric analysis for cop-
per is about thirty times as delicate as for chromium. The delicacy for cop-
per sulphate is inversely proportional to the concentration of copper.
Within certain concentrations the percentage errors are constant in all
colorimetric analyses. The maximum of delicacy with ammoniacal
ooppfer sulphate lies at a concentration of i gram atom of copper in 4996. i
liters. In more concentrated solutions a change of 5 per cent, called
forth a marked change in color. About the same relation was found
to hold for ammonia-free copper sulphate (8 per cent.). The delicacy
appears to be independent not of color tone but of the nature of the sub-
stance in solution.

433 KBvmws.

DeVecchi (Z. wiss. Mikrosk., 23, 312) recomtneDded i and 5 perce
methyl alcohol solutions of photoxyline as embedding material in 1
croscopic work, Visser {Ckem. Weekblad, 3, 74.3} stated 'that instt
of weighing the nitron precipitates in Busch's determination of nit
add they can be compared in height with precipitates from solutii
of known content. Instead of no mg. per liter he found in this *
100 mg. Nitrous, chloric, perchloric, oxalic and salicylic adds %
salts with nitron characteristic under the microscope. So also d
saccharin after conversion with some drops of dilute alkali into o-sul
aminobenzoic add.

Analysis of Inorganic Compounds.

Noyes {Techn. Quarterly, 16, No. 2; 17, No. 3; Chem. News, 93, 1
146, 156, 179, 189, 205, 216, 226, 239, 250, 262) has published part
his system of qualitative analysis including practically all of the elemei
It is worked out with great care with numberless test analyses, checks i
notes on the various procedures.

Metalloids, Oxygen, Sulphur. — Bancroft and Hamilt (/. PkysioU
34, 306) determined the oxygen dissolved in physiological salt soluti
by boiling the liquid under very low pressure (about 3 mm, of roercu
and catching the gas in a 3 mm. wide tube. In such a tube 0.3 cc
gas will cause a bubble 100 mm. long. The analysis conosts in not
the difference in length of bubble Ijefore and after absotption with p5
gallic add. Mathewson and Calvin (Am. Chem. J., 36, 113) gave a metl
for the determination of hydrogen peroxide or of ferrous sulphate or ot
reducing agents in which ferrous ammonium sulphate and ammonium 1
phate in about equal amounts are treated with 5 cc. of phosphoric a
and diluted to 50 cc. Then 5 cc. of a titanium potasdum sulphate
lution are added and the titration carried out immediately with ab
0.15 normal hydrogen peroxide to yellowcolor. Results when the met!
is reversed for hydrogen peroxide are a httle lower than those of
permanganate method which is to be expected because permangati
ondizes the organic matter of the hydrogen peroxide as well. Vn
sodium nitrite the results are too high because the reaction is very s
near the end point.

Berger {Compl. rend., 143, 1160) determined free sulphur by pom
over an amount of substance containing 0.1-0.2 gram of free siilpl
10 cc. of fuming nitric add, adding 0.5 to i gram of potassium brom
evaporating the liquid after a few minutes to dryness, fuming the-
idue down 2 or 3 times with a few cc. of hydrochloric add, taking up v
water and precipitating as usual with barium chloride.

McFarland and Gregory (Chem, News, 93, 201) detected snip
in crude iron by mixing 5 grams of sample intimately with 0.5 gran
tartar, wrapping in fitter paper and igniting for 15 minutes in a mu
cooling, breaking up the mass, putting it in an evolution flask with 1
tube dipping into an ammoniacal cadmium chloride solution. E
ing hydrochloric add {2 parts add, i part water) is put on the mixt
The solution containing the suspended cadmium sulphide is addi
with hydrochloric add and titrated directly with normal iodine solut
Reinhardt [Stahl u. Eisen, 26, 799) gave a similar method, treating
iron sample with hydrochloric acid in an atmosphere of hydrogen :
catching the hydrogen sulphide in an ammoniacal cadmium solut

REVIEWS. 433

The precipitate was washed with ammonia, shaken with iodine
solution, hydrochloric acid added, and the iodine titrated back with
sodium thiosulphate and starch. Gyzander (Chem. News, 93, 213) took
up 0.2 gram of pyrites with a mixture of 5 cc. of hydrochloric and 15 cc.
of nitric adds, evaporated, evaporated again with water and 5 cc. of
concentrated hydrochloric acid, took up the residue with 100 cc. of water,
I cc. of concentrated hydrochloric acid and 3 cc. of hydroxylamine hy-
drochloride solution (i ounce in 500 cc. of water). After the iron was
reduced the solution was heated to near boiling, 10 cc. of a cold 10 per
cent, barium chloride solution added dropwise and the barium sulphate
determined as usual. Hintz and Weber (Z. anal. Chem., 45, 31) took
up their 0.5 gram of pyrites similarly, but removed the iron with ammo-
nia, filtered and washed till filtrate and washwater amotmted to 450 cc.
This was neutralized, using methyl orange as indicator, i cc. of hydro-
chloric add (d. 1. 17) added, the whole heated to indpient boiling and
predpitated with 24 cc. of 10 per cent, barium chloride solution diluted
to 100 cc. heated to boiling and added all at once with vigorous stirring.
The washed iron predpitate was dissolved in hydrochloric add and pre-
dpitated again with ammonia, the filtrate and washwater from this treated
again with barium chloride, any sulphate formed being added to the
main quantity. Raschig (Z. angew. Chem., 19, 331) determined sul-
phur in pjnites by predpitation, after its oxidation, as benzidine sul-
phate, stating that the method is more accurate than Hintz and Web-
er's barium sulphate predpitation.

Bruhns (Z. anal. Chem., 45, 573) determined small amoimts of sul-
phuric add, especially in waters, by treating 150 cc. of the sample in a
200 cc. flask with 5 cc. of a barium chromate emulsion (29.45 grams of
potassium chromate and 20 grams of primary potassium carbonate in
750 cc of water with 48.86 grams of crystallized barium cjiloride in 250
cc. of water), pouring off the supernatant liquid and diluting the resid-
ual emulsion to 500 cc. i cc. of strong hydrochloric acid (3 : 2) was
added and the whole allowed to stand with some shaking. Thirty minutes
after the solution had colored yellow by solution of barium chromate
in the hydrochloric acid it was made weakly alkaline with dilute ammo-
nia, diluted to 200 cc. and filtered through a dry filter. After shaking,
100 cc of the filtrate were treated in a glass stoppered flask with some
potassium iodide and 5 cc. of hydrochloric acid (3 : 2) and after 30 min-
utes titrated with sodium thiosulphate and starch. Using twentieth-
normal thiosulphate which has the value of thirtieth-normal here, the
number of cc. used multiplied by 1.78 gives the number of mg. of sul-
phuric add in 100 cc. of water. The result must be corrected by the
subtraction of 0.15 cc. of thirtieth -normal thiosulphate. Johnson (This
Journal, 28, 1209) determined carbon bisulphide in commercial ben-
zene by conversion into copper xanthogenate, treating 75 cc. of ben-
zene with I cc of saturated alcoholic caustic potash for each o.i gram
of bisulphide, adding a weighed amount of bisulphide (0.06 to 0.14 gram)
and shaking for 15 to 20 minutes. The potassium xanthogenate was
dissolved in water by shaking in a separatory funnel and the well washed
benzene extracted with 3/4 of the original amoimt of caustic potash.
The extracts and washwater were diluted to 500 cc, an aliquot portion
weakly addified with very dilute acetic acid and treated with copper

434 MtvlBWS.

sulphate in not too great excess. This was allowed to stand for li hours
with repeated shakmg, filtered and the precipitate washed, dried and
ignited in porcelain. The results gave a CuOiCS, ratio varying from
I ■ i«593 to 1 : 1.825, an average of i : 1.750 or about 90 per cent- of Macag-
no's value i: 1.931. Seyewitz and Bloch (Btdl. soc. chim. [3] 35, 293)
determined hydrosulphurous acid in hydrosulphites and their compounds
with formaldehyde by means of their reduction of silver halides, 2AgCl +
4NH3 + Na2SA-2HaO = 2NaCl + 2(NHJjS08 + 2Ag. Sulphites, bi-
sulphites and hyposulphites do not do this. Hyraldite, NaHSO2.CHjO.2HjO,
decomposes smoothly with ammoniacal silver chloride at 80®. About
4 times the theoretical amount of silver chloride should be used.

Halogens. — ^Jannasch and Zimmermann and Jannasch alone (Ber., 39,
196, 3655) separated the halogens by adding to a solution of the mixture
of their compounds in 120 to 150 cc. of water, 15 cc. of acetic acid and
at least 3 cc. of 30 per cent, hydrogen peroxide and distilling off the io-
dine in 20 to 25 minutes in a current of steam. It was absorbed in hy-
drazine sulphate, concentrated ammonia and water. The contents of
the absorption vessels were united, acidified with 30-40 cc. of nitric
acid and precipitated cold with silver nitrate. The separation of chlo-
rine and bromine is dependent upon concentration and upon presence
of a determined excess of sulphuric acid. With much acid hydrogen
peroxide sets bromine free even in the cold. It is separated by means
of a current of carbon dioxide by mere warming on a weakly boiling
water bath. Rupp and Horn (Arch. Pharm., 244, 405) gave a volu-
metric determination of iodine in the presence of chlorine and bromine
ions. 0.2 to 0.4 gram of the substance was dissolved in 50 cc. of water,
acidified with 25 cc. of dilute sulphuric acid and then about 3 grams of
oxalic acid were added, i per cent, potassium permanganate was added
till the liquid shaken upon flask wall showed a distinctly violet color, the
mixture allowed to stand 3 hours with occasional shaking, i gram of
potassium iodide was then added and the liberated iodine titrated with
tenth-normal thiosulphate. In the presence of chlorine and bromine
ions the end point of the permanganate addition fails. 10 cc. of i per
cent, permanganate may be added when from 8 to 25 cc. of thiosulphate
are required. Then by making a Volhard determination the data are
at hand for iodine and chlorine ion separation, iodine and bromine or
iodine from chlorine and bromine ions. Ville and Derrien (BulL soc,
chim. [3] 35, 239) detected fluorine in food products by means of the
fact that the absorption spectrum of methemoglobin is changed by
sodium fluoride. The red band X633 disappears and Menzie*s band
(X612) appears. 100 cc. of red wine were evaporated i or J to remove
alcohol, the volume made up again with water, 50 cc. of the liquid shaken
with I gram of manganese dioxide, filtered, 25 cc. of the filtrate treated
with 0.1 gram of dioxide, i to 1.5 cc. of defibrinated blood diluted with
4 volumes of a i : 100 solution of potassium oxalate added, shaken, fil-
tered and tested. Or 100 cc. of the dealcoholized wine might be treated
with 5 cc. of a solution of i part egg albumen and 7 parts of i per cent,
potassium oxalate solution, the whole boiled up and filtered after cool-
ing, 25 cc. treated with i to 1.5 cc. of the reagent blood and tested. Wines
with 0.08 to 0.1 gram of sodium fluoride give Menzie's band.

Nitrogen y Phosphorus, — Bom water {Chem, Weekhladj 3, 30) gave a

REVIEWS. 435

simplified method for determining nitric add in nitrates. He placed
in an 800 cc. Erlenmeyer flask about 0.5 gram of nitrate, 200 cc. of bro-
mine, 5 cc. of alcohol and 50 cc. of caustic potash (d. 1.3), also about
2.5 grams of finely cut aluminum wire. The flask was connected through
tube and condenser with a vessel containing standardized sulphuric
add. It was heated slowly till the reaction started and to boiling only
after the gas evolution weakened. Busch (Z. angew. Chem., 19, 1329)
and Busch and Schneider (Z. ges. Schiess.-u. Sprengstoffw.j i, 232) de-
termined the nitrogen content of nitrocellulose by warming 0.2 gram
of sample in a flask with 5 cc. of 30 per cent, sodium hydroxide and 10 cc. of
3 per cent, hydrogen peroxide on a water bath and then boiling over
free flame to complete solution. Forty cc. of water and 10 cc. more of 3
per cent, peroxide were added, the mixture warmed to 50° and 40 cc.
of 5 per cent, sulphuric acid allowed to flow into the bottom of
the dish from a pipette. The predpitation of the nitric add was then
made with nitron acetate. Busch (Ber., 39, 1401) determined nitrous
add by oxidation with warm neutral 3 per cent, hydrogen peroxide,
addification with 2 per cent, sulphuric acid and precipitation of the
nitric add formed as nitron nitrate (Ibid.^ 38, 861). Nitrous and ni-
tric acids together may be determined by titrating the nitrous acid in
an aliquot portion with permanganate and in another oxidizing it with
hydrogen peroxide and weighing both adds as nitron nitrate. Reich-
ard (Chetn. -Zig., 30, 790) stated that dry arbutin gives with as little as
0.0001 gram of nitric acid a yellow color. The color is fairly stable, espe-
cially if concentrated hydrochloric acid be used instead of concentrated
sulphuric in the test. Forty per cent, caustic potash colors the solution
reddish yellow, ammonia weak violet. Berberine and free nitric acid
only on warming give a reddish brown substance, in great concentra-
tions almost black, if hydrochloric acid be used to liberate the nitric
add. With sulphuric acid the reaction appears in the cold. It is more
delicate than the first.

Artmann and Skrabal (Z. anal. Chem., 46, 5) determined ammonia
iodometrically, treating the substance containing it with bromine and
alkali, shaking in a closed flask, adding solid potassium iodide, addif3ring
and titrating the iodine with thiosulphate. Madri (Gaz. chim. ital., 36 I,
373) critidzed the method of Roberto and Roncali (Uindustria chim-
tea J 6, 178) for the determination of hydrazine with permanganate. They
stated that on heating, 5 molecules of nitrogen were set free for 4 mole-
cules of permanganate. Madri said their own results never agreed with
their own equation or with the earlier work of Peterson (Z. anarg. Chem.,
5i 3). In acid solution the equation is, according to Peterson,
i7N,H,.HjSO, -t- 13O = 13H3O -f- 7(NH4)2S04 -h loN, + H3SO,. Without
the add the grade of oxidation of the hydrazine increases but does not
reach the stage of Roberto and Roncali. Hence the reaction cannot
be used for the determination of either permanganate or hydrazine.
Potassium bichromate oxidizes hydrazine completely, as has been al-
ready pointed out by Purgotti {paz, chim, ital., 26 II, 559).

Aronstein (Chem. Weekblad, 3, 283, 493) detected white phosphorus
in the presence of much phosphorus sesquisulphide by leading hydro-
gen or carbon dioxide mixed with air over the substance to be tested.
Phosphorescence is given to the gas, disappearing as more air is blown

and appearing as the air content becomes again less. The pure s
lisulphide shows a similar behavior only at about 80° or above, 1
st is delicate to 0.2 per cent, of phosphorus, if its absolute mass isi
iS than 0.04 mg. He criticized Van Eijk's lead acetate method e
ated that rubbing the carbon bisulphide extract residue in the d:
IS delicate only to 1.4 per cent, of phosphorus in the sesquisulphi
in Eijk (Ibid., 3, 367, 404, 623) stated that a test tube containing
g. of phosphorus- free sesquisulphide would light up above 70° wl
le warmed at same time in the same water bath containing 0.02
nt. of phosphorus in the same amount of sesquisulphide would d(
ider 60". A better test is to distil the preparation with lead aceti
e sesquisulphide is decomposed and an illumination is observed n
02 mg. of phosphorus or more. To exclude red phosphorus it is 1

extract the mass with carbon bisulphide and to test the residue fi
■aporation of the bisulphide for white phosphorus. If this extract
sidue be gently rubbed in a dark room, illuniination is obtained in
esence of white phosphorus. The test is delicate to 0.004 ™K-
esence of turpentine is disturbing to the phosphorus tests. .\i
^in's test is said to be less dehcate than these. Schenck and Sch
ler., 39, 1522) detected small amounts of white phosphorus by pi
g a current of air over the warmed substance and leading this to
otective cylinder of an electroscope. Phosphorus sesquisulpl
uses scarcely any conductivity up to 75° and none at 50°, but a fi
>n of a mg. of the white phosphorus will. Temperatures of 35°
." are the most suitable. The Umit of sensibility is about 0.004 '
emtns (Arb. kais. Gesundkeitsamt, 24, 26n) tested the red phosphorus
■mmerce for white or vellow bv extracting 5 grams of the sample y
,0 cc. of benzene on a boiling water bath for 4 hour by using a ret
ndenser. After cooling the filtered solution i cc. of it was trea
th I cc of ammoniacal silver nitrate solution (1.7 grams of niti

100 cc. of ammonia [d. 0992 ]) and shaken. If the color is weak ;
w there is no white phosphorus. The color is reddish or dark brc

there is a precipitate in its presence. In i cc. of benzene o.oi 1

white phosphorus can be detected. Mauricheau-Beaupr^ (Ctm
nd., 142, 1206) gave a qualitative test for phosphorus. A piece
iss tubing 5 to 10 mm. in diameter is brought into the upper oxii
g part of a hydrogen or acetylene flame. In the presence of pt
lorus the glass is not only etched but gains in weight, while with
losphorus it loses weight and there is no etching. The element r
terwards be determined in the etched part. Phosphine in acetyl
the ratio i : loooo may be detected. Hydrofluoric add must be
nt, Fricke (_Stahl u. Eisen, 26, 279) determined phosphorus in i
id steel by dissolving the sample in nitric acid, oxidizing the phosphc
th permanganate, dissolving the precipitated manganese dioxide
imonium chloride solution, evaporating to 30 or 40 cc., making wea
id by nearly neutralizing with ammonia and precipitating warm v
alybdate solution. The precipitate Is filtered, washed with cold wi
1 iron is removed, dissolved in standard sodium hydroxide solul
cc, = 0.00025 gram phosphorus) and the solution titrated back with
uricacidofthesamestrengthusingphenolphthalein. 2[(NH,),PO,,izM<
46NaOH = 2(NHJ^P0, + (NH,),MoO, + 23Na, MoO, + 22H

RBvnsws. 437

2P«46NaOH = 23H2SO4. Jannasch and Heimann (Ber., 39, 3625)
stated that phosphoric acid can be quantitatively volatilized from its
salts if an intimate mixture of phosphate and carbon be made first by
treating phosphate and sugar solution in a distilling flask with sulphuric
acid. The phosphorus is volatilized in a current of chlorine.

Arsenic, Antimony. — Bemtrop (Ckem. Weekblad, 3, 315) determined
the arsenic content of a mirror by oxidation at 60® with potassium bi-
chromate and sulphuric acid. ^KJCrfij + 20H3SO4 + 6As ■= 5KJSO4 +
20H2O + 5Crj(S04)8 + sAsjOj. The excess of bichromate is titrated
back with potassium iodide and thiosulphate. The results are some-
what too low because of the formation of some arsenious oxide in the
preparation of arsine in spite of efforts to keep air out and the reten-
tion of some arsine in the evolution flStsk, but chiefly because the liquid
of the flask remains markedly arsenic-bearing. If this liquid be run
again for arsenic and the result added to that of the main determination
the method is good. Strzyzowski {Pharm. Post, 39, 677) determined
arsenic in animal objects, etc., by heating carefully in a porcelain cru-
cible, I gram of magnesia, 10 cc. of the liquid (5 to 10 g. of half solid mat-
ter or I gram of solid broken up and rubbed with 10 cc. of water) and
0.5 to I cc. of concentrated nitric add on an asbestos plate, then over
a free flame till after breaking up the residue is pure white. This is taken
up with 10 cc. of water and 5.5 cc. of 50 per cent, sulphuric add, filtered
from the caldum sulphate and carbon, and the filtrate brought by wash-
ing with i2i per cent, sulphuric add to 20 to 25 cc. and the magnesium
arsenate in solution investigated in the author's Marsh test apparatus.
Rosenthaler (Z. anal. Chem., 45, 596) determined arsenic add by the
reverse of the iodometric method for arsenious add. 2H8ASO4 -f 4KI +
4HCI = AsjOj + 4I + 4KCI -f sHjO. The reaction is complete at the
end of 10 to 15 minutes and the liberated iodine is titrated as usual.
Both adds of arsenic may be determined by obtaining the arsenious
add amount first and after oxidation the total amount of arsenic acid.

Low (This Journal, 28, 17 15) gave a technical determination of ar-
senic and antimony in ores in which the sample is taken up with pri-
mary potassium sulphate, tartaric and concentrated sulphuric acids.
The two sulphides are precipitated by hydrogen sulphide from this
diluted and filtered solution, then they are dissolved in potassium sulphide
solution. This solution is evaporated with primary potassium sulphate
and concentrated sulphuric acid and finally heated till the sulphur and
most of the free add are gone, the cooled residue taken up with water
and concentrated hydrochloric add and the arsenic precipitated with
hydrogen sulphide. The filtrate is evaporated again with primary sul-
phate and sulphuric add, the cooled residue taken up with water and
hydrochloric add, diluted and titrated with permanganate. The oxalic
add value of the permanganate solution multiplied by 0.9532 gives the
antimony value. The arsenic predpitate is dissolved in water contain-
ing ammonium sulphide, the solution evaporated with primary sulphate
and add, and the cooled melt taken up with water and boiled. The
solution is made weakly alkaline with ammonia, then acid with hydro-
chloric acid, cooled to room temperature, primary sodium carbonate
added and the solution titrated with iodine. Mateme {BuU. soc. helg.
ckim.,, 20, 46) separated arsenic, antimony and tin by treating the sul-

•*^- •J'^'^f* ■■\-'-' ;-^.

, -■ .-.' • •*••.'. •• » t^
^ . » ■ ■ t * •■ . • * ■ ■ "

.'• v ..

!: C'-*;- ; -: ■- •••J

•a .■•■•• i ■ ' > . . •

" • • 1. •.•!■.

Jt ^ .' * ■ '^ - ^

r ■ . ', ■ ... <■ ; . .

' * ■' ■•■••■»

• .'.

*- «

5 "^ ' . ••;• /

4 •

' ' ,

\ ■- •

: {

• »■

> ...

< '.

■.V" - ... • '4

438 REVIEWS.

phide mixture (containing bivalent tin without tetravalent) with a hot
-• f " -.^ : 2 per cent, borax solution, filtering and washing with gradually decreas-

ing strength of borax solution; the filtrate contained the arsenic. The
precipitate was treated with boiling 5 per cent, sodium carbonate solu-
tion and filtered hot; the filtrate contained antimony. The tin in the
residue was dissolved in lo per cent, sodium hydroxide. If stannic tin
^ *•_•'*.;}/ ^ were present instead of stannous, the arsenic was removed by digestion

with cold 2 per cent, borax solution and the residue boiled with 5 per
^ cent, sodium hydroxide solution, dissolving out the tin; the antimony

in the residue could then be dissolved in hot 5 per cent, sodium carbon-
• ~wN ..A ate solution. Treatment of these solutions with hydrochloric acid or

with tartaric acid and ammonium chloride gives the sulphide in each
, case. The borax solution is made up of 500 cc. of 2 per cent, borax

: and 20 cc. of 10 per cent, ammonium chloride, the soditun carbonate

solution of 500 cc. of 5 per cent, sodium carbonate and 10 cc. of 10 per
cent, ammonium chloride, and the tartaric acid solution of 300 cc of
20 per cent, add and 150 cc. of 10 per cent, ammonium chloride. Czer-
wek (Z. anal. Chem., 45, 505) separated antimony from tin in alloys by
dissolving in a warm mixture of tartaric and concentrated nitric adds
with a little water and precipitating the tin from this solution heated
to incipient boiling by 45 per cent, phosphoric add. The predpitate
was filtered, washed with dilute ammonium nitrate solution, dissolved
in warm ammonium sulphide, repredpitated with dilute sulphuric add,
filtered, washed again with dilute ammonium nitrate, dried and ignited
to tin dioxide, oxidizing with nitric acid. The filtrate was neutralized
with ammonia, ammonium sulphide added, the solution addified with
acetic add and the antimony trisulphide filtered warm. The predpi-
tate was washed with dilute ammonium nitrate, dissolved in a large
porcelain crudble with ammonium sulphide, the solution evaporated,
the residue oxidized with fuming nitric add and ignited to antimony
tetroxide. With other metals present the tin phosphate was washed
with 150 to 200 cc. of hot normal nitric add, dissolved in ammonium
sulphide, or sodium sulphide if copper were present, and the tin predpi-
tated as above. In the tin filtrate other metals were removed with
ammonia and ammonium sulphide and the antimony predpitated by
**•.. * • ; .*• . ' . acetic add.

. ". . V \,; • .• r Carbon, Boron, Silicon. — Aupperle (This Journal, 28, 858) gave a

-..V ' ... volumetric method for the determination of carbon in iron and steel

V •^' . which rests on the titration of barium hydroxide into which the carbon

': '^ ; >-... ' :. ; . dioxide from the combustion of filings of the sample in oxygen has been

. ;. ' , . passed. He stated that the barium hydroxide may be titrated with add

*: * ■ , * ^^ t^^ presence of barium carbonate without losing carbon dioxide if

the add be run deep into the solution by means of a long capillary tube
for any dioxide set free is absorbed again by the hydroxide above. (Cf.
Bruhns, p. 426 Super.) Rosenthaler and Tiirk (Arch. Pharm. ,'244, 517) and the
former alone {Ibid., 244, 535) investigated the percentage of dissolved
substance absorbed from i per cent, solutions of it in different solvents
by 5 times its weight of different kinds of charcoal in the cases of codeine,
caffeine, salicine, picrotoxine, gallotannic, gallic and oxalic adds, po-
tassium oxalate, indigotine and glucose. The amount and rate of ad-
sorption are greatest in the case of animal charcoal, great also with "flesh"

RKVIEWS. 439

and less so with *'plantblood" charcoals, and are small with "blood,"
"limewood" and "sponge'* charcoals. They are greatest in aqueous
solutions, then in descending order come ethyl alcohol, methyl alco-
hol, acetic acid, acetone and chloroform. There is relatively less adsorbed
in concentrated than in dilute solutions. Adsorption and decoloriza-
tion are little dependent on the temperature. The more readily a sub-
stance is adsorbed the more difficult it is to dissolve it out again. The
decolorizing power of charcoals is dependent on their adsorbing power.
Charcoal to be used for decolorizing should be carefully purified and used
in as small amount as possible. Solutions in solvents other than water
and as concentrated as possible are advisable. Substances readily ox-
idizable must not be decolorized with animal charcoal. The percent-
age of a substance adsorbed increases with its molecular weight. Color-
ing matter in a solution usually has a high molecular weight and its con-
centration in the solution is usually small, both being conditions favor-
able to adsorption.

Castellana (Gaz. chtm, ital., 36 I, 136, 232) stated that the green flame
test for boric acid is obtained if the substance be mixed with potassium ethyl
sulphate and heated with the flame till the first vapors rise and then
setting fire to these. 0.5 mg. of boric add may be detected. The char-
acteristic odors of their esters are obtained with a considerable number
of organic acids if carefully heated dry with potassium ethyl sulphate.
He stated in reply to Velardi's (Ibid.y 36 I, 230) criticism that copper
chloride, phosphites and hypophosphites do not interfere with the boric
acid color and that the turmeric paper test is not superior to his in deli-
cacy. Fendler (Z. Nahr.-Genussm., 11, 137) gave a modification
of the turmeric paper test, comparing the color obtained under
certain conditions with a variety of standard colors obtained with known
amoimts of boric acid. Low (This Journal, 28, 807) found that the
turmeric test is extraordinarily sharp if the paper be dried not at 100®
but at ordinary temperatures or 40° to 50° in a vacuum desiccator.
In I cc. 0.000,001 gram of boric acid may easily be detected. In the quan-
titative determination it is not possible to drive all of the boric acid out
of water solution with methyl alcohol because the smallest amount of
water will hold back considerable acid. All the acid may be driven
over by the use of some water-extracting substance like calcium chloride.

Hinden (Z. anal. Chem,, 45, 332) said that the taking up of silicates
after evaporation with hydrofluoric add may be accomplished by evap-
orating 4 to 6 times with hydrochloric add, the bases being converted
into chlorides. One gram of substance is moistened in platinum with a
little water and evaporated down with 10 to 15 cc. of concentrated hy-
drofluoric add, the residue taken up with hydrochloric add (1:1), 10
cc. of hydrofluoric acid again added, evaporated and the evaporation
repeated about 6 times with 10 to 20 cc. of hydrochloric add. Com-
plete decomposition is not to be obtained in this way with barium and
lead-bearing glasses; here the recommendation is made to filter off the
insoluble residue and to treat again with the two adds. Schucht and
Moller (Bef.,39, 3693) titrated hydrofluosilidc add with sodium hydrox-
ide in the presence of methyl orange, adding first an excess of neutral
cafcium chloride solution. HjSiFg + 3CaCl, + 6NaOH - 3CaF, -f
6NaCl + H,SiO, + 2H,0.

Melals, General. — Tarugi and Marchionneschi (Boll. diim. jar.
, 629) gave some particulars regarding the use of thioacetic acid reco
;tided by Schiff and Tarugi {Gaz. chim. iial., 24, 551) in qualitat
alysis. They stated that it works welL Under pressure in sea
xs at 90° it gives complete precipitation of sulphides much m
idily than hydrogen sulphide, and in acid concentrations such t:
der normal circumstances no sulphides would form. The act
thioacetic add in closed vessels at 90° is equal to that of hydro]
phide at 14.34 atmospheres, Daitz (Z. anal. Chem., 45, 92) a
ed the ammonium sulphide group separation of fioetticher {lb
, 99), saying that in the separation of cobalt, nickel, iron and man
ie from aluminum, zinc and chromium by excess of sodium hydr
and bromine a good deal of nickel and some cobalt go into soluti'

0 that on treatment of the first four as hydroxides with concentra
drochloric acid, evaporation, addition of excess of ammonia, heat
boiling after strong shaking and addition of 2 to 3 cc. of hydroj
oxide and filtering, much nickel and some cobalt are left in the n
; while a good deal of iron and manganese go into the filtrate wh
luld contain only complex nickel and cot^lt salts. Jannasch s
imann (/. pr. Chem. [2] 73, 473, 488) gave some more metal sepa
OS by distillation in a current of dry hydrochloric add gas. Tin i
. over away from cadmium at a temperature not above 320°. I
th is separated from cadmium by distilling at temperatures betwi
>° and 350°; bismuth and silver may be easily separated. Antimi:
tils from lead mixture between 150° and 250°. Antimony a
iper, cadmium or silver may be separated. The temperati
d in no case exceed 350=. ' The authors upheld their method agai
edheim's (Z. attcU. Chem., 44, 465) critidsm.

ilkalies. — Hiibener (Chem.-Ztg., 30, 58) in detecting sodium sulpb
thiosulphate made use of the fact that the sulphur dioxide liberal
m the thiosulphate requires twice as much iodine for its oxidat
loes the thiosulphate itself. Na^Oa.sHjO + I = iNajSp, + Nal
,0. Na,SA-5H,0 + H^O^ - SO, + S + NajSO, + 6H,0. SO
+ HjO = 2HI + H^O,. One determination is made direct, w
ine, another by passing the sulphur dioxide evolved in an atmospb
:arbon dioxide through an excess of iodine and titrating back. Th'

1 values by suitable calculation give the amounts of sulphite a
>sulphate in the sample. He found in a sample supposedly 90 ]
t. pure, 92,87 per cent, thiosulphate, 3.11 per cent, sulphite and 2

cent, sulphate, the last figure representing sodium sulphate cor
adingto the barium sulphate difference of the totally oxidized sa

and that calculated to correspond to the sulphite and thiosulpha
mimboeuf (Ann. chim. anal, appl., 11, 130) determined potassii
>ride in potassium bromide by predpitating both as silver salt a
ghing. 2 grams of pure potassium bromide should give 3.1;
ms of silver bromide while 2 grams of potassium chloride should wei
i23 grams as silver chloride. He assumed i per cent, of potassii
iride would increase the weight of the silver bromide by o,oo6c
n and gave a table for the chloride content of the bromide. Pajei
z. chim. Hal., 36 II, 150, 298) stated that the solubility of potassii
x>tassium and sodium persulphate is increased by the presence

REVIEWS. 441

sodium sulphate and decreased by potassium sulphate, the solubility
coefficients per 100 cc. at 12^ being for saturated solutions of sodium
sulphate 5.982, of primary sulphate 8.72, of potassium sulphate 0.792,
of primary sulphate 0.329. There is direct proportionality to the greater
or less solubiUty of the sodium or potassium salt. The solubility of
the persulphate in sodium salt solutions is a function of their concentra-
tion of sodium. Hence the greater solubility in the presence of sodium
salts is probably due to chemical action. KjSaOg + Na^04 = NajS^Og +
K^4. On this grotmd he criticized Tarugi's {Ibid,^ 34 I, 324) method
for the determination of potassium. Schlicht (Chem.'Ztg.y 30, 1299)
observed that sodium phosphomolybdate solution gives a yellow pre-
cipitate when heated with potassium salts acidified with nitric acid and
makes a good test for potassium.

Beryllium, Magnesium, Alkaline Earths. — Glassmann, {Ber., 39, 3366,
3368) gave a quantitative separation of beryllium and aluminum, neu-
tralizing the hydrochloric or sulphuric add solution of the oxides approx-
imately with sodium carbonate and adding excess of sodium thiosul-
phate, then boiling till the odor of sulphur dioxide disappears and heat-
mg for i hour on the water bath. Beryllium remains in solution as
sulphite or basic sulphite. The aluminum hydroxide and sulphur pre-
cipitate is washed and ignited. The excess of thiosulphate in the fil-
trate is decomposed with hydrochloric add and the beryllium predp*
itated with ammonia or according to Glassmann's method with potas-
sium iodide and iodate, which according to Friedheim (Ibid., 39, 3868) was
first described by Joy in 1864 and later by Zimmermann in 1887. Par-
sons and Barnes (This Journal, 28, i5%9)^parated beryllium from alum-
inum and iron by neutralizing the chloride solution as nearly as possible
with ammonia, treating the cold solution with 10 grams of primary so-
dium carbonate, heating the mixture as rapidly as possible to boiling
and boiling for i minute. The beaker is set in cold water to cool, the
piedpitate filtered and washed with hot water and dissolved on the
filter in as little hydrochloric add (i : i) as possible, diluting to 100 cc.
in the original beaker. This solution is neutralized with ammonia and
the predpitation repeated. The filtrate and washings are neutralized
with concentrated hydrochloric add, the predpitate dissolved, the
solution boiled to drive out carbon dioxide and the beryllium predpi-
tated by ammonia as the hydroxide, this washed with ammonium ace-
tate solution, dried, ignited and weighed as the oxide. The aluminum
hydroxide predpitated is dissolved in hydrochloric add, repredpitated
with ammonia, ignited and weighed. The iron separation is the same.

Grirabert (/. pharm. chim. [6] 23, 237) carried out the Schlagden-
hauffen reaction for magnesium by treating 10 cc. of the solution with
5 cc. of 10 per cent, potassium iodide and 2 to 3 drops of a concentrated
sodium hypochlorite solution. In the presence of magnesium a red-
dish predpitate looking like ferric hydroxide is obtained. The test
is delicate to 1:2000; the solution must never be add. Bellier (Ibid.,
[6] 23, 378) treated the magnesium solution with a solution of iodine
in potassium iodide and then dropwise with dilute sodium hydroxide.
With more than 0.02 per cent, of magnesium a relatively abundant red-
dish brown predpitate is obtained ; with 0.005 P^r cent, a reddish brown-
yellow color. A delicacy of i : 20000 is claimed for this modification.

442 REVIEWS.

Ammonium salts, adds and alkalies prevent the reaction completely;
lime lowers its delicacy somewhat. The precipitate is perhaps a mix-
ture of magnesium oxide and iodine. Berju {Chem,-Ztg., 30, 823) de-
termined small amounts of magnesium indirectly by weighing the phos-
phoric acid of magnesium ammonium phosphate as P2O«.24Mo03.

L6b {Ibid.f 30, 1275) found that barium dioxide could not be titrated
with permanganate in the presence of sulphuric acid because the ba-
rium sulphate apparently occluded some of the substance, but it might
be in the presence of hydrochloric acid and manganese sulphate. The
results of the method were compared with those of an iodometric one,
the barium dioxide in hydrochloric acid solution being treated with potas-
sium iodide solution and the iodine titrated back with thiosulphate.
A neutral or weekly ammoniacal solution of the alkaline earth was
treated by Benedict (This Journal, 28, 1596) with its volume of 5 times
normal hydrochloric add, then 2 to 3 cc. of saturated potassium iodate
solution. No precipitate indicates the absence of barium, an imme-
diate one shows considerable and a slow one little barium or consider-
able strontium. The filtrate is tested with a little more than an equal
volume of saturated ammonium sulphate solution and heated to boil-
ing; a white permanent precipitate indicates strontium. Another por-
tion of the original solution is allowed to stand with twice its volume
of saturated potassium iodate solution for i to i minute after shaking,
then the filtrate is tested with i its volume of ammonium oxalate for
calcium. Caron and Raquet (Btdl. soc. chim. [3] 35, 106 1) precipitated
barium as chromate from the alkaline earth mixture, then after making
the filtrate alkaline again with an\monia threw out the strontium with
alcohol. The calcium was tested for with potassium ferrocyanide so-
lution, i

Iron, Aluminum. — Komar (Chem.-Ztg., 30, 15, 31) obtained the salt
FeH(SOj2«4H20 by evaporation of a solution of ferric sulphate (pre-
pared by the oxidation of a solution of ferrous sulphate in sulphuric acid
by means of nitric acid or by electrolysis) from a sulphuric acid content
of 400 cc. of the monohydrate per liter to a concentration of 45° to 50°
Baum4. The compound is partly easily soluble in water, partly difificultly.
The at first cloudy and finally clear green solution reacts weakly acid
and does not reduce permanganate. By heating at 90 ° to 1 00 ° the compound
destroys paper and smells of sulphuric acid ; on gently heating in a cm-
cible sulphur trioxide and ferric oxide are obtained. Iron and zinc
may be separated by conversion into sulphates, dissolving these in sul-
phuric acid (400 cc. of monohydrate per liter [d. 1.3 to 1.4]), evapo-
rating this solution to dryness and igniting the residue to constant weight
over the burner. The zinc sulphate decomposes only at about 700°
and may be extracted with water. Rupp and Horn (Arch. Pharm^,
244, 571) modified Rupp's {Ber.y 36, 164) method for the titration of
ferrous salts with alkaU hypoiodite, using caustic potash in place of so-
dium potassium tartrate as the hydriodic acid binding agent. Ferrous
iron is instantly oxidized to ferric by the measured excess of tenth-nor-
mal iodine in the presence of normal or 5 per cent, caustic potash. The
solution is then acidified with acetic or better with oxalic acid and the
iodine excess titrated with thiosulphate.

Moody (i4w. /. Sci. [4] 22, 483; Z. anorg. Chem., 52, 286) gave an

REVIEWS. 443

iodometric determination of basic alumina and free sulphuric acid in
ahiminum sulphate and alums. The iron is determined in aliquot por-
tions before and after reduction with zinc, and any zinc by electrolysis
from acetate solution. A portion is boiled with potassium iodide-iodate
mixture in a suitable apparatus and the iodine collected in a receiver
containing potassium iodide. After slightly acidifying with sulphuric
add the iodine is titrated with thiosulphate. The precipitate formed
in the flask contains besides alumina, ferric and zinc oxides which are
determined in the usual ways, i molecule of alumina requires 6 atoms
of iodine, i of ferric oxide 6 of iodine, i of ferrous oxide 2 of iodine, 5
of zinc oxide 8 of iodine, i of ammonitun i of iodine and i of sulphuric
acid 2 of iodine. The decomposition of zinc sulphate is abnormal.
isZnSO^ 4- 20KI + 4KIO3 + i2H,0 = 3Zn^(p}i)^0^ + i2KaS04 + 24I.
The total iodine less the sum calculated to correspond to the single
sulphates gives the iodine difference ; if this be positive the mixture con-
tains free add, if negative, free alumina.

Cobalt, Nickel, Manganese, Zinc. — ^Alvarez {Ann. chim. anal. appL,
II, 445; Chem. News, 94, 306) stated that the blue color pointed out by
Donath in 1901 as obtained when solid caustic potash or soda or very
concentrated alkali solution was added to cobalt solutions is obtained
also when barium hydroxide, calcium chloride or other water-extracting
substance is added, i drop of i : 100 cobalt solution added to boiling
concentrated alkali solution will give the reaction, which takes place in
the presence of nickel Grossmann and Schiick (Ber., 39, 3356) gave
a new test for nickel, treating a solution of dicyanodiamine with a Uttle
hydnxhloric acid, heating to boiling, adding the nickel salt, then caustic
potash solution and obtaining a yellow crystalline precipitate of nickel
dicyanodiamidine (Ni[C2H50N4]2.2H20) in needles arranged in star
shapes. They are immediately soluble in potassium cyanide, but not
in boiling caustic potash solution, and are sparingly soluble in water
and ammonia. Cobalt forms no analogous compound. Reichard {Chem.-
^%M 30i 556) stated that if powdered dehydrated nickel salts of mineral
adds are heated with an equal amotmt of fully dry methylamine hydro-
chkride in porcelain the color becomes deep dark blue. This color dis-
appears on cooling, leaving a dirty gray-yellow, solid mass which soon
deUquesces. It becomes blue again on heating and decolorizes on
cooling. The color is shown with 0.1 mg. of nickel Cobalt salts sim-
ilarly heated yield deep blue oily drops which do not lose their color
on cooling.

Funk (Z. anal. Chem., 45, 562; observed that iron and manganese sul-
phides dissolve very easily in dilute adds, but in the presence of ammo-
nium salts nickel sulphide and even more easily cobalt sulphide dissolve,
too. At ordinary temperature nickel and cobalt sulphides are not dis-
soh'ed by a little dissodated add like formic but in separations they
dissolve and some iron sulphide remains. Manganese sulphide dissolves
except for traces. Jannasch and Gottschalk (J. pr. Chem. [2] 73, 497)
predpitated manganese from ammoniacal solutions by means of oxy-
gen rich in ozone. Small amounts of manganese can be predpitated
by slow passage of the ozone through the ammoniacal solution. Large
amounts are managed by adding the manganese solution dropwise to
100 cc. of strong ammonia through which a vigorous ozone current is

444 REVIEWS.

passing. The precipitate is hydrated manganese dioxide. Manganese
may be separated from sodium, calcium and zinc in this manner by one
precipitation. Magnesium, nickel, cadmium and copper each require
a repetition of the precipitation in order that the manganese may be
free from them; the precipitate is dissolved in hydrochloric acid contain-
ing hydrogen peroxide. The separation of manganese and cobalt failed.
Tarugi {Gaz. chim. ital,, 36 I, 332) gave a test for manganese and a new
method for the formation of glycerose. Manganese hydroxide dissolves
in glycerol and such a solution colors itself red by oxidation through
air or more quickly through oxygen or sodium hypochlorite. The color
intensity depends only on the amount of manganese present. 0.00001
gram of manganese can be detected. Cobalt and copper interfere only
with amotmts of less than i per 1000. Glycerol is converted into glycer-
ose through sodium hypochlorite by i drop of i : looo cobalt chloride
solution. 50 cc. of glycerol, 2 cc. of 5.9: 1000 cobalt chloride and 10 cc.
of 50 per cent, caustic soda put all at once into 150 cc. of sodium hypo-
chlorite (7 per cent, active chlorine) gave on cooling in ice, 18 per cent
of glycerose.

Bertrand and Javillier (CompU rend., 143, 900; BtUL soc, chim. [4]
'f 63) &Lve a method for the precipitation of zinc, treating a solution
containing zinc and a sufficient amount of lime with excess of ammonia
and heating to boiling, the calcium zincate coming out in microscopic
crystals. It is difficultly soluble in excess of lime and may be used for
quantitative work. The zincate is mixed with carbonate; the precip-
itate is dissolved in hydrochloric acid, the solution evaporated to dr}--
ness, the residue taken up with some water, the lime precipitated with
ammonium oxalate and the filtrate evaporated and ignited with sul-
phuric add to zinc sulphate. Less than i part of zinc in 500,000 of so-
lution can be detected. A zinc test was given also by Bradley (see under j
copper). j

Mercury y Silver. — Rupp (Ber,, 39, 3702) gave a volumetric determina-
tion of mercury, adding to the mercuric salt solution (about 0.2 gram
in 25 to 50 cc.) I to 2 grams potassium iodide so that the precipitate
first formed dissolves, making alkaline with caustic potash or soda, then
adding with shaking, a mixture of 2 to 3 cc. of 40 per cent, formaldehyde
and 10 cc. of water. The mixture is acidified with acetic add to distinct
odor, an excess of tenth-normal iodine solution (25 cc.) added, the me-
tallic mercury brought into solution by shaking and the excess of iodine
titrated with tenth-normal thiosulphate. Utz {Pharm, Post, 39, 785)
modified his 1905 method of determining sublimate in dressing mate-
riab to conform to this mercury determination of Rupp's. Seidell (This
Journal, 28, 73) determined mercury and iodine in antiseptic soaps by
treating the soap sample with 150 cc. of 95 per cent, alcohol and 3 to 5
cc. of concentrated hydrochloric acid, warming the mixture and adding
gradually small amounts of water till the whole is in solution. Mercury
is predpitated as sulphide, filtered into a Gooch crudble and washed
with 95 per cent, alcohol. The filtrate freed from fat may be shaken
in a separatory funnel with chloroform and a few drops of nitrous acid
and the iodine in the chloroform determined by titration with thiosul-
phate. Goldschmidt (Z. anal, Chem., 45, 87) stated that silver is pre-
dpitated quantitatively as a black powder, if cobalt foil be put into

RBVIEWS. 445

boiling silver salt solutions and that it may be weighed. Gold is like-
wise thrown out of boiling solutions by nickel as a brown powder.

Copper^ Cadmium^ Bismuth. — Bradley {Am. J. Sci. [4] 22, 326) ob-
servwi that the blue color of "logwood hematoxylin'* and copper salts
is a copper test of extraordinary delicacy, i : 1000,000,000. Zinc nitro-
pnisside is crystalline and may be detected under the microscope even
in the presence of amorphous precipitates, the reaction being much more
delicate than the common precipitation tests for zinc. Rhead (Proc.
Chem. Soc.f 22, 244; /. Chem. Soc, 89, 1491) determined copper with
the aid of standard titanium trichloride solution in the presence of potas-
sium thiocyanate. Cupric salts are reduced and the copper precipi-
tated in the presence of sulphuric or hydrochloric acid as cuprous thio-
cyanate. A ferrous salt is added to sharpen the end-point. The
cupric salt oxidizes an equivalent amoimt of the ferrous salt and the red
color of ferric thiocyanate appears. The color disappears at the end
of the reaction. The titration must be carried out below 30° and as
rapidly as possible. Nitric add must be absent. Ferric iron and cupric
copper may be determined together and the iron subtracted after sepa-
rate determinations. The titanium trichloride is standardized by means
of a ferric salt solution obtained by the oxidation of a ferrous salt with
permanganate.

Goldschmidt (Z. anal. Chem., 45, 344) observed that cadmium is quan-
titatively precipitated from boiling salt solutions in aluminum dishes
in the presence of traces of chromium and cobalt nitrates. The catalyz-
ing agent is aluminum. Other metals can be used for the quantitative
determination by catalysis. Moser {Ibid., 45, 14) found that bismuth
precipitated as phosphate would carry down some cadmium in the sep-
aration of bismuth from copper and much cadmium and that the cad-
mium is hard to remove. It is not easy to make a second precipitation
of the bismuth phosphate because of its insolubility. The method is
gpod for the determination of bismuth alone but as a separation has no
advantages over the ordinary one.

Uranium^ Vanadium, Molybdenum, Tungsten. — Finn (This Journal,
28, 1443) separated uranium and vanadium, after solution of the mineral
sample in sulphuric acid, by precipitating twice with excess of sodium
hydroxide solution, boiling each time, acidifying the united filtrate and
washwaters with sulphuric acid, adding ammonium phosphate and
making alkaline with ammonia. The filtrate containing vanadium is
addifi^ with sulphuric acid, reduced with sulphur dioxide till the so-
lution is blue and titrated hot with permanganate. The uranium
precipitate (UO2NH4PO4) is dissolved in sulphuric acid reduced with
zinc and the filtra^te titrated with twentieth-normal permanganate at
60°. The iron factor multiplied by 1.631 gives vanadium pentoxide, by
0-9159 vanadium, by 2.567 uranous uranic oxide and by 2.133 uranium.
Gilbert (Z. offentl. Chem., 12, 263) determined molybdenum in glance by
extracting the trioxide with ammonia, after roasting in air, and ignit-
ing to constant weight. The small amout of molybdenum left in the roasted
ore is obtained by fusing with potassium and sodium carbonates, tak-
ing up with add, reducing with zinc and titrating with permanganate.

Von Knorre {Stahl u, Eisen, 26, 1489) modified his earlier method
for determining timgsten in steel. The steel is dissolved in hydrochloric

cid with exclusion of air and without filtering the .
Kd with sodium carbonate ; after cooling, lo cc. of appit
lOTtnal sulphuric acid or alkali sulphate and 40 to 60 i
olution are added. The precipitate of tungsten, bens
jid sulphate is filtered, washed with dilute benzidine i
lited in platinum. The iron bearing tungsten trioxide
usion with sodium carbonate, the melt extracted with
ron oxide filtered out and the solution acidified with h;
usmg methyl orange). After addition of 10 cc. of suj
ungsten trioxide is precipitated with benzidine solution,
is before, ignited and weighed as trioxide. Watts (I
ind Metallurgist, July, 1906; Chem. News, 95, 19) dete
icid in natural and concentrated tungsten ores by tali
rith 50 cc. of hydrochloric and 15 cc. of nitric acid, bo1
leating to near boiling for 4 hours in a covered beaker.
rated to 10 to 15 cc., 50 cc. of hot water and 5 cc. of h
Ldded. Tungsten trioxide is separated from silica by
lontaining some ammonium chloride. The ammonia
ivaporated and the trioxide ignited and weighed.

Gold, Platinum Group, Tin. — Donau (Monatsh., 37,
I new method for the determination of the metals, espi
)alladium, through conductivity measurements. The ]
)Ut into a U conductivity measuring tube, reduced by c
ind the maximum effect of the carbon monoxide mea
:rease of conductivity (d) is affected somewhat by ad
).000476 + 0.277I + lol' {1 = initial conductivity). Ea<
[z mg. per 100 cc.) with definite initial conductivity (>
lifferent for the different amounts of hydrochloric acid
I certain increase in conductivity (x.io~'). ' The x y :
mknown, an empirical function is assumed and an equa
:he method of least squares, z ■= 1.76 a: — o.02iJxy
3.001I x'y + 0.0Q0738 xy'. The gold content can also I
^phic representation. The pal&dium determination
:onductivity is only insignificantly affected by the fr
md is nearly proportional within certain limits to the pa
The numberofmg. of palladium per 100 cc. is given by mu
io~* by the increase in conductivity. The mean error he
Maxson {Am. J. Sci. [4] 21, 270; Z. anorg. Ghent., 49, :
iactory colorimetric determinations of gold in small ami
liis red colloidal solution by mixing gold chloride and sa
icetylene solutions. The content of such solutions
jravimetrically and different concentrations were pre;
lilution. First a Gallenkampf then a Penfield colorir
The least determinable quantity was 0.000,01 gram; thi
was 0.0008 vrith an error of 0.000,06. Petersen {Z.
342) instead of the usual separation of gold, platinum
irsenic in the hydrogen sulphide group, precipitated a
the group and some cobalt and nickel from weakly ac
line turnings. After warming for 1 hour the predpit
ind warmed with dilute hydrochloric add; cadmium, t:
i>alt dissolved. The residue was washed, then boiled v

RSviJ^ws. 447

add; mercury, lead, bismuth, copper, nickel dissolved leaving gold,
platinum, antimony and some antimonic acid. This residue was ig-
nited witii I to 2 parts of ammonium nitrate and 5 parts of ammonium
chbride in porcelain; antimony volatilized as chloride. The residual
gold and platinum were dissolved in aqua regia, and detected through
concentrated ammonium chloride or sulphur dioxide or alkaline hy-
drogen peroxide. Arsenic, antimony and zinc must be tested for in spe-
cial portions. The original solution was tested for zinc by precipitating
with sodium carbonate, dissolving the precipitate in hydrochloric acid,
passing in hydrogen sulphide and decomposing with excess of sodium
acetate; white zinc sulphide was precipitated. Orloff (Chem.-Ztg.f 30,
714) observed that hydrogen peroxide seems to dissolve osmium and
osmium hydroxide to a marked degree to osmium tetroxide. Histolog-
ical specimens blackened by osmic acid are completely decolorized by
hydrogen peroxide. From mixtures of the platinum metals obtained
by reduction with zinc or magnesium the peroxide dissolves only os-
mium. Silver iodide rapidly blackens with palladium chloride or bro-
mide, forming a mixture of palladium iodide and silver chloride. So-
luble alkali and alkaline earth iodides form insoluble precipitates with
salts of other platinum metals, hence a potassium iodide solution may be
used only with certain precautions to separate palladium from the otners.
But freshly precipitated silver iodide changes only palladium chloride
to black iodide. Thompson and Miller (This Journal, 28, 11 15) deter-
mined melting points, cooling curves, microstructure, densities and elec-
trical conductivities of platinum silver alloys containing about 10, 20,
3p, 40, and 50 per cent, platinum. They concluded that it is not pos-
rible to separate platinum from gold, iridium, etc., by alloying with sil-
ver and d^solving in nitric acid, that platinum alloys of more than 20
per cent, platinum cannot be completely separated by concentrated
sulphuric acid, and that the irregular results obtained from treating
these alloys with nitric acid are caused apparently by the existence of

£tinum silver compounds. They analyzed the alloys by treating
m hot with concentrated sulphuric acid in two portions, diluting,
fitering and washing out the silver, then igniting the residue in porce-
bm. This residue was taken up in aqua regia and evaporated nearly
to dryness with nitric add. The solution was diluted and the silver
precipitated with sodium chloride solution. The precipitate was
prashed free from chlorine, dissolved in ammonia and re-
[Rrecipitated with nitric acid. From the solution in sulphuric acid the
silver was precipitated after dilution either as sulphide or as chloride.

Reichard (Pharm, Centrh., 47, 391) gave a new reaction for tin, treat-
ng a little finely pulverized uric acid with some drops of stannic chloride
ohition, then adding concentrated sodium hydroxide to the mass drop-
wse with stirring till nearly all is dissolved and heating. A gray to
ntense black fleck is formed. Stannous compounds do not give the
leaction, neither do arsenic or antimonic add. Lead and cadmium do
lot give it. Copper hydroxide gives a black on heating without the
me add, owing to the formation of copper oxide. Mercuric chloride
pves a reddish brown compound. Bismuth gives the same reaction
IS tin but the predpitate is insoluble in sodium hydroxide. 0.0001
pcam of tin may be detected. Nitric and hydrochloric adds destroy

44^ t(^vi]SWS.

the black residue only slowly and incompletely, sulphuric add immc--'
diately.

Miscellaneous. — Rimini (Atti accad, Lincei [5] 15, II, 320) stated that-
both Riegler and Ebler have overlooked his work on hydrazine vohn*
metric methods. He modified his iodometric method by carrying out^
the determination in alkaline solution, thus avoiding the separation oii
iodine. sN^H^.H^SO, -f 2KIOs + 6K0H = 3N2 + 2KI + sK^SO^ +
12H2O. Instead of Ebler's mercury method he proposed to add to a
concentrated solution of hydrazine sulphate after neutralizing, using
methyl omnge as indicator, a certain excess of half-normal sodium hy-,
droxide, then the mercury salt solution and then to heat. The solution was'
brought to definite volume and the excess of sodium hydroxide deter-
mined in an aliquot portion or in the filtrate with half -normal add. l
N,H,.HjSO, + 2HgCl2 4- sNaOH = 4NaCl + NaHvSO^ + Hg^ -h N, - '
5H2O. Pannani's method for potassium persulphate gives too high re-
sults. Rimini added to a neutral solution of persulphate a solution of
hydrazine sulphate neutralized with caustic potash, then shook this up
with titrated caustic potash and determined the excess of alkali after
5 minutes. 2K2Sa08 + NaH^.H-^SO^ + 5KOH = N, 4- sKjSO^ -f 5H,0.

Puschin (/. russ, phys.-chem, Soc, 38, 764) gave a quantitative sep-
aration of tin from manganese, ferrous iron and chromium by electrol- ;
ysis. Bivalent iron and manganese may be separated like cobalt and :
nickel from tin (Ibid.y 37, 828). Trivalent manganese and chromium \
may also be separated easily from tin. A satisfactory separation of i
ferric iron and tin was not obtained.

Analysis of Organic Compounds.

General f Hydrocarbons ^ Derivatives, — Alvarez (Chem. News, g4, 297;
Ann, chim. anal, appl., 12, 9) observed that hydrated sodium peroxide,
the color reagent for polyphenols, gives characteristic colors with some
other compounds when 0.2 to 0.3 gram of peroxide and 0.05 to o.i gram
of substance are treated with 5 cc. of alcohol, then after 4 to 6 minutes
with 15 cc. of water. Emodine gives an intense light red, changed to yellow
with acetic acid ; chrysarobin a wine color, yellow with acetic acid, i , 2-
dihydroxyanthraquinone blue-violet» yellow with adds (blowing on
the blue- violet solution causes it to turn red at edges) ; alizarin violet,
orange with acids; 1,2,4-trihydroxyanthraquinone intense red -violet
in alcohol, cherry-red in water; chrysophanic acid cherry-red in alcohol,
lighter red in water; rosolic acid intense purple; purpurin alizarin-red;
anthragallol almost black dark blue ; dihydroxyquinone brown in al-
cohol, red in water, and ellagic acid brown-black in alcohol, yellow in
water. Graefe {Chem. Rev. Felt.- u. Harz-Ind., 13, 30) stated that car-
bon tetrachloride surpasses all other ordinary solvents in the power to
dissolve paraffin, for example at 20° of a paraffin of melting point 53.5°
I cc. of ethyl acetate, dissolves i.i, of acetone 1.2, 96 per cent, alcohol
1.9, ether 83.4, petroleum ether 200, ligroin 244, chloroform 246, benzene
285 and carbon tetrachloride 317 mg. Carbon tetrachloride also gives
good results in the fractional precipitation of different paraffins. Ni-
cloux (Compt. rend. J 142, 163; Bull. soc. chim. [3] 35, 321) determined
small amounts of chloroform in air or in blood or an aqueous liquid by
heating 60 cc. of alcoholic chloroform solution with 10 cc. of 10 per cent.
alcohoHc caustic potash for 30 to 45 minutes. The reaction CHCl, +

RBvmws. 449

4KOH - 3KCI + KHCO3 4- H,0 is quantitative. Fifteen cc. of water are
added after cooling, the solution is neutralized with sulphuric acid, using
phenolphthalein, and titrated with silver solution and potassium chroma te
as indicator. The error is i to 2 per cent. To determine chloroform
'm air the air is drawn through 2 wash flasks containing 96 per cent, alco-
hol at the rate of 2 liters per hour. In blood the five-fold volume of
80 to 95 per cent, alcohol acidified with 0.25 gram of tartaric acid is added,
then 10 cc. of 95 per cent, alcohol, J of the liquid is distilled off and the
rest of the determination carried out as above.

Rdchard (Pharm. Cenirh., 47, 309) observed that phenanthrenqui-
none dissolved in concentrated excess of primary sodium sulphite solu-
tion and warmed gave on cooling a colorless crystal mass which on stand-
ing for some weeks about 2/3 liquefied again. At the same time there
formed at the upper edge of the crystal mass a red to red-brown ring, chang-
ing soon to gray and fiinally increasing in volume till the whole mass
was colored gray, while the liquid was colored clear green. Dupr^ (The
Analyst, 31, 213) determined moisture in a substance containing other
volatile material by putting the substance in layers in a tube i cm. wide
and 12 long between layers of fine sand and calcium carbide, connecting
the tube with a nitrometer filled with saturated sodium chloride solu-
tion. On warming on a water bath acetylene was obtained. One cc. of
acetylene corresponds to 0.001725 gram of water. The method can

be used on a substance like cordite. Raikow and Urkewitsch (Chem.-
Ztg., 30, 295) observed that solid pulverized caustic potash gives a brown
color with nitrotoluene and nitrobenzene, weaker with the latter. Pul-
verized caustic soda does not act at ordinary temperatures with nitro-
benzene (distinction from caustic potash) but gives a yellow-brown
color with nitrotoluene. The reaction is more deUcate when gasoline
is used as a solvent (i cc. for 0.5 gram of caustic soda). In i cc. of
gasoline solution, 0.0025 cc. of nitrotoluene can be detected. To detect
tolaene in benzene both are converted into nitro compounds and tested.
An approximate quantitative determination can be made by compar-
mg the brown color with that caused by known amounts of nitrotoluene
in gasoline solution. Alkali hydroxides in this reaction are effective
in the order of the atomic weight of their metals, lithium hydroxide
weakest, then those of sodium, potassium and rubidium.

Alcohols, Phenols, — Blank and Finkenbeiner (Ber., 39, 1326) deter-
mined the methyl alcohol in formaldehyde solutions by putting i gram
of the solution into 50 cc. of twice normal chromium trioxide and 20 cc.
of 98 per cent, sulphuric add, diluting after 12 hours to i liter, adding
a little potassium iodide to 50 cc. of this solution and titrating back
with tenth-normal thiosulphate. Methyl alcohol -32 (0-6)100/48 per cent.,
where a = 0.8 gram (the amount of oxygen used — 0.016 X number
of cc. of thiosulphate used) and b = the oxygen needed for the oxidation
of the formaldehyde. The latter is calculated from the formula

"? rz ; , Fendler and Mannich (Arb. Pharm. Inst,

30 X 100

Unvo. Berlin III, i) determined methyl alcohol in spirits by carefully

heatmg 10 cc. of the liquid to boiling in a 50 cc. flask provided with a

doubly bent tube for condenser, i cc. is distilled off in 4 to 5 minutes,

4 cc. of 2 per cent, sulphuric acid added and then gradually i gram of

450 RSVIKWS.

pulverized potassium permanganate with cooling and shaking,
mostly reddish liquid is filtered and heated for 20 to 30 seconds at
boiling, cooled and i cc. of the colorless liquid mixed with 5 cc. of
centrated sulphuric add, then with 2.5 cc. of a fresh solution of
phine hydrochloride (0.2 gram in 10 cc. of concentrated sulphuric ad
With 0.5 per cent, methyl alcohol the liquid becomes in 20 minutes i
tense violet or violet-red in color. Voisenet (Bull. soc. chim. [3]
748) applied his recently discovered reaction of formaldehyde with
trite containing ^gg albumen {Ibid. [3] 33, 1 198) to the detection of me
alcohol. An amount of the alcohol corresponding to i cc. absohite
diluted to 50 cc. with water, 5 grams of potassium bichromate and
cc. of 20 per cent, (by weight) sulphuric acid are added, shaken, let
for an hour at the ordinary temperature, then distilled so as to ob
30 cc. in I hour. These 30 cc. contain all the acetaldehyde and are throml
away. The following 20 cc. must contain the methyl alcohol and aic
tested for it. Aldehydes with phenol character give a similar violet
color which is, however, quite stable toward reducing agents; the formal-
dehyde color is not. Gascard (J. pharm. chim. [6] 24, 97) determined
the molecular weight of alcohols and phenols by heating the dried sub-
stance in a long-necked flask with 2 to 3 times the theoretical amount
of benzoic acid anhydride for 24 hours in a water or oil bath, the flask
being covered. In most cases boiling calcium chloride solution (cdd;
saturated) will do for the bath. 10 to 20 cc. of ether are run into tkj
flask, then 5 cc. of water and 2 drops of phenolphthalein and the liqiiidl
titrated with normal potash. Molecular weight P == p -I- 1000 / (N-n),l
where p = weight of substance, N = number of cc. of caustic potash
n = number of cc. of caustic potash found in the blank determinatioii
with everything but the alcohol or phenol. In the case of a polyal-
cohol the result is to be multiplied by the number of alcohol groups
present. Where the benzoic acid ester is difficultly soluble in ether, benzene
or chloroform are used . ROH -f (CeHjCOj) ,0 = CeHg.COOR + C^HrCOOH.
The free benzoic acid is titrated. Klason and Carlson (Ber., 39, 738)
determined organic hydrosulphides and thioacids by titration with io-
dine and with alkalies. 2RSH -f I2 = RjSj -h 2HI. Only thiocyanic
acid is indifferent to iodine. Aromatic hydrosulphides are strong enough
acids to give neutral salts with alkalies in alcoholic solution and as a
result they can be titrated with alkali and phenolphthalein in alcohol
solution. With the aliphatic hydrosulphides the indication is not
sharp and with thioglycolic acid it fails, even in alcoholic solution.
Rosen thaler (Arch. Pharm. ^ 244, 373) used Nessler's reagent as a test
for hydroxyl groups and found that, except benzhydrol, octyl and cetyl
alcohols, all compounds that he tried containing primary and secondar}'
alcoholic hydroxyl groups, on boiling for i minute gave reduction.
These three did on boiling for some hours with a return condenser.
Compounds with tertiary alcoholic hydroxyl do not reduce Nessler's re-
agent. Of compounds with phenol character — phenol, salicylic acid,guaia-
col, thymol, resorcinol, phloroglucinol, and orcinol give no reduction, xy-
lenols and creosol an unimportant, hydroquinone, pyrocatechol and
gallic acid an energetic reduction. The Sacchse liquid reduces, but not
the Knapp solution. Nessler's reagent may be used to test amylene
hydrate for fermenting amyl alcohols and citric add for tartaric.

REvmws. 451

Carobbio {Boll, chim, farm., 45, 365) tested for resordnol by letting
I to 2 cc. of the suspected ether flow down the wall of the test tube onto
I cc. of zinc chloride with enough ammonia to give a clear solution. With
OX) I to 0.001 gram of resordnol the place of contact turns yellow, then
green and after a few minuter blue. Traces require considerably longer.
Alcoholic hydrochloric acid added carefully forms a weakly red layer
between the ring and the ether, spreading through the latter on shak-
ing. By this reaction, o.oi part of resordnol per 1000 can be recognized
in a few minutes. Hydroquinone gives the yellow ring, changing
soon to a brown-red. Pyrocatechol and adrenaline give a garnet-
red ring. Desmouli^re (/. pharm. chim. [6] 23, 244, 281, 332) deter-
mined glycogen by finding the amount of glucose pre-existing in the or-
gan in question and that formed by hydrolysis. Two portions of the
organ are taken, one of 10, the other of 40 grams. The 10 gram portion
is digested with 0.15 gram of pepsin in the presence of water and a little
sulphuric acid for 6 hours at 48° to 50°. A little more sulphuric acid
is added and the whole heated to ii5°-i20° in an autoclave for 1.5 hours.
The cooled contents of the autoclave are heated with slight excess of
mercuric nitrate, then dilute caustic soda added to neutral or weakly
alkaline reaction, the whole diluted and filtered. 100 cc. of the filtrate
are shaken with 4 or 5 grams of zinc dust to separate the mercury, fil-
tered, 50 cc. of tMs treated with caustic soda till the zinc precipitated
dissolves again, water added to 55 cc. and the glucose determined grav-
imetrically or by titration. In the calculation the volume of the pre-
cipitate obtained in the purification should be subtracted from the vol-
ume of the liquid, here 200 cc. The glucose pre-existing in the solution
is determined in the 40 gram portion by extracting with water, the
uiiited extracts are purified with mercuric nitrate and caustic soda, the
volume brought to 1000 cc., the filtrate treated with zinc dust and the
determination carried out as above. Ottolenghi (Atti. accad. Lincet,
[5] 15 I, 44) stated that the reaction of Neuberg and Rauchwerger for
the detection of cholesterol (with rhamnose and concentrated sulphuric
add) is not characteristic of cholesterol but is given by phytosterol from
many sources. Neuberg (Z. physiol. Chem., 47, 335) admitted this to
be true. Windaus (Chem.-Ztg., 30, loii) detected small amounts of
cholesterol in the presence of phytosterol by the difficult solubility of
cholesterol dibromide in a mixture of ether and acetic acid. 100 cc.
of a 50 per cent, mixture dissolves only 0.6 gram, 40 cc. ether and 60 cc.
add only 0.25 gram of cholesterol dibromide. Addition of a Uttle water
lowers the solubility while much water causes the phytosterol dijbro-
mide to separate out in an oily form and to crystallize with more difficulty.
Cholesterol dibromide may be obtained almost quantitatively by dis-
solving cholesterol in a little ether and adding a solution of bromine in
acetic add (5 grams in 100 cc). The use of petroleum ether gives a cho-
lesterol salt of different melting point.

Aldehydes, Ketones. — Meth (Chem.-Ztg., 30, 666) stated that one of
Rimini's {Ann. farmacoL, 98, 97) tests for formaldehyde, that of the
red color with ferric chloride, phenylhydrazine hydrochloride and hy-
drochloric or sulphuric add, occurs also with acrolein even though with
different shade and less delicacy. Rimini's other test with phenylhy-
drazine hydrochloride, sodium nitroprusside and sodium hydroxide giv-

452 RSVIBWS.

ing with formaldehyde a blue color gives nothing with acrolein. Scboorl
(Pharm. Weekblad, 43, 11 55) criticized Blank and Finkenbeiner*s (Bcr.
39, 1326; see under alcohols) method for the determination of fonnal-
dehyde. He added to 3 grams of the formaldehyde solution a mixtuxt
of 50 cc. of normal alkali and 50 cc. of neutral 3 per cent, hydrogen per
oxide, warming on a water t^th, cooling and adding phenolphthafein.
Hence it must take 16 to 17 cc. of normal add to neutralize. H&issey
(/. pharm, chim. [6] 23, 60) determined small amounts of benzaldc-
hyde by distilling 50 cc. from the liquid resulting from the splitting
the glucoside, treating this with 50 cc. of a solution of i cc. of freshhf
rectified phenylhydrazine and 0.5 gram of acetic acid in 98.5 cc. of water,
heating for 20 to 30 minutes on a water bath, then after 12 hours'
ing, collecting in a Gooch crucible, washing, drying and weighing the
phenylhydrazone. This weight multiplied by 0.54081 gives the weight
of benzaldehyde. Wallis (Pharm, J, [4] 22, 162) determined chloral-
hydrate by dissolving in 10 cc. of alcohol, digesting for 3 hours on a water
bath with 10 cc. of a volumetric sodium hydroxide solution, neutraliz-
ing with sulphuric acid and titrating "with standard silver nitrate sohi-
tion. CClj.CHCOH), + sNaOH = 3NaCl + 2HCOONa + 3H,0.

Auld (/. Chem, Ind,, 25, 100) determined acetone by adding to a hot
10 per cent, caustic potash solution of the weighed liquid dropwise a so-
lution of 200 grams of bromine and 250 grams of p6tassium iodide in i
liter of water, heating for 1/2 hour at 70°, and distilling off the bromo-
form. The distillate is then heated with 50 cc. of alcohol and enough
caustic potash to make a 10 per cent, solution, using a return conden-
ser till all bromoform is decomposed. 3CHBr8 + 3KOH -f CjH^OH -
3CO + CjH, + 9KBr + 7HjO. The distillate is titrated with siher
nitrate solution. To 58 parts of acetone, 240 parts of bromine corre-
spond. JoUes (Ber., 39, 1306) determined acetone by the addition of a
3- to 4-fold excess of titrated primary sodium sulphite titrating back after
30 hours* standing with iodine solution. CH5.CO.CH5 -f NaHSO, =
CHj.CCOH) (SOaNa) .CH3.

Acids, — Carletti (Boll. chim. fartn.y 45, 449) observed that furfuml
gives colors with the aromatic amines, at any rate with aniline, only ifl
the presence of organic acids, not with aniline salts of mineral adds nor
after the addition of a mineral acid to aniline originally combined with
an organic add. Two solutions are prepared; (a) 5 grams of pure aniline
are treated with 25 cc. of concentrated acetic add and diluted to 100
cc. with water, and (b) i gram of freshly prepared furfural is made up
to 100 cc. with 95 per cent, alcohol. 50 cc. of wine or vinegar are de-
colorized with animal charcoal, 25 cc. of 95 per cent, alcohol added, 10
cc. of tliis solution shaken with 5 drops of (a), then 5 of (b). With no
mineral acid there is a distinct reddish color reaching a maximum after
30 minutes. But with i/iooo of free hydrochloric, nitric or sulphuric
acid the solution keeps its original weak green color. The reaction fails
only with wines to which gypsum has been added. Guamieri (Staz.
sper. agrar. Ifal., 38, 906) detected benzoic and salicylic adds in to-
mato preserves by diluting and addifying the extract with sulphuric
acid (1:3), extracting with 1/2 volume of a mixture of equal parts of
ether and petroleum ether, repeating with ether, evaporating after the
addition of a little weakly ammoniacal water, filtering, adcUng 2 to 3

RBVIEWS. 453

grams of freshly ignited carbon, when the odor of benzoic add is evi-
dent. The cooled test treated with a drop of i per cent, ferric chloride gives
a reddish brown precipitate in the presence of benzoic add. A violet
color changing to blood-red on shaking shows salicylic and a brown color,
green on shalang, tannic add. In the latter case the tannic add is pre-
cipitated as iron tannate and salicylic add detected in the filtrate by the
violet color. The iron tannate is dissolved in hydrochloric add, diluted
and shaken with ether. On slow evaporation of the ether, crystals of
benzoic add appear and may be identified by the aniline blue reaction.
Herzog (Festschrift Adolf LiebeUy 440; i4nn. 351, 263) gave a test for lactic
add, treating the silver salt with iodine. 2R.CH0H.C00Ag + 12 =
R.CH0H.COOH -h RCHO -f CO2 + 2AgI. The acetaldehyde may be
detected with sodium nitroprusside and piperidine. Amino adds were
carefully converted in the cold into the a-oxyadds by means of silver
nitrite and their silver salts treated as above with iodine. GlycocoU,
alanine and other higher homologues can thus be detected. Schloss
{Beitr. chem. Physiol, u. Pathol., 8, 445) detected glyoxylic add by means
of indole and skatole. With no ring formed at contact zone between so-
lution and skatole (0.2 gram in 100 cc.) there is no glyoxylic add present.
If a red or brown ring is formed, the following test should be tried. 20
cc. of liquid are decolorized with animal charcoal, i or 2 cc. of dilute
sulphuric acid added to a portion of the colorless filtrate and let stand
for 10 minutes in a water bath at 50°. The skatole reaction should be
set up in another portion, adding about i cc. of skatole solution and pour-
ing concentrated sulphuric add down to the bottom. The first portion
is tested with indole. A sharp red ring after at most 2 or 3 minutes points
to the presence of glyoxylic acid. The skatole reaction alone is not pos-
itive. Glyoxylic add to 0.000,01 gram can be thus detected in wine. Sullivan
and Crampton (Am, Chem. /., 36, 419) detected tartaric acid or tar-
trates by means of the cr)rstalline structure of the caldum salt. 50 cc.
of the concentrated salt solution containing not more than 30 per cent,
of dry substance are rendered alkaline with caustic potash, a few drops
of 20 per cent, potassium acetate added, acetic acid to acidification and
then 10 cc. of 30 to 40 per cent, caldum chloride solution. The mix-
ture is stirred for i to 2 minutes and allowed to stand for 12 to 15 hours
at room temperature. The calcium tartrate crystallizes in rhombic
prisms or pyramids recognizable under the microscope. No other or-
ganic add gives a caldum salt of similar form. Citric and oxalic add
predpitate, malic does not, but in the presence of these acids the cal-
dum tartrate is thrown out in needles and plates. Tocher (Pharm, J,
[4] 23, 87) used a couple of reactions for the detection of dtric, tartaric
and malic adds. Tartaric on heating with concentrated sulphuric add
gives a black mass, dtric a yellow solution and malic a dark solution.
Tartaric gives with cobalt nitrate a red solution becoming colorless with
excess of caustic soda, deep blue on boiling, the color fading out on cool-
ing. Citric gives a deep blue solution, and a predpitate if the neutral
solution be boiled with caldum chloride solution. Malic gives also a
blue solution but no predpitate with caldum chloride and gives on heat-
ing with dilute sulphuric add and potassium bichromate the odor of
ripe fruit. Ulpiani and Parrozzani (Atti accad. Lincei, [5] 15 II, 517)
gave a rapid determination of dtric add in lemon juices, determining

454 REVIEWS.

first the approximate total acidity of the juice with normal sodium hy-
droxide, then putting 50 cc. in a 200 cc. flask with enough nonnal so-
dium hydroxide to neutralize about i/io of the total acidity. This
value represents the maximum proportion of tartaric and oxalic adds
which are then precipitated after the addition of about 17 grams of cal-
cium chloride and 5 grams of animal charcoal and boiling. The mix-
ture is cooled, made up to volume and filtered. The following deter-
minations are made in the filtrate: (i) 50 cc. are titrated with normal
sodium hydroxide till a faint permanent turbidity appears, and (2) an-
other 50 cc. are boiled and titrated with the caustic soda to turbidity.
The difference in sodium hydroxide between (i) and (2) represents 2/3
of the free and combined citric acid in the juice. Kastle (Public Health
and Marine Hospital Sendee U, S. Hygienic Lab. BuU., No. 26, 31) stated
that small amounts of sacccharin heated with small amounts of a re-
agent (5 cc. of phenol and 3 cc. of concentrated sulphuric add) for 5 min-
utes at 160° to 170®, the mass dissolved in a little water and made al-
kaline with twice normal caustic soda, gives a purple or deep red color
according to the amount of saccharin. The limit is 0.025 mg. Sali-
cylic and benzoic acids do not interfere, neither do cumarin or ethyl
/>-sulphobenzoate. Vanillin becomes yellow, then red in the cold; heated
to 160° to 170° it is first blood-red, then almost black; after the addi-
tion of caustic soda it is deep dark red. With other phenols substituted
for ordinary phenol in the test the following colors are obtained:

Phenol Saccharin at 160° to 170® Vanillin at 100**

Pyrocatechol Green Dark blue to green

Hydroquinone Dark red-brown with blue fluo- Dark red-brown

rescence

Resordnol Salmon color with strong green- Red with weak green fluorescence

yellow fluorescence

Tricresol Purple-red Deep purple-red

Phlorogludnol Wine-red Yellow

Thymol Clear blue Clear red

Cumarin at 100° gives colorless (with tricresol) to orange-yellow (with
phloroglucinol) compounds. Stanek (Z. physiol. Chem,, 47, 83, 334;
Z. Zuckerind, Bohmeny 31, 316) found that choline precipitates from add
or alkaline solutions with potassium triiodide while betaine precipitates
only from acid. To 25 cc. of at most a 5 per cent, solution of the mix-
ture of both hydrochlorides 5 per cent, primary potassium or sodium
carbonate is added and precipitation made with the triiodide; the
choline periodide is filtered, washed and the nitrogen determined. The
filtrate is concentrated to about 25 cc, then about 10 per cent, sul-
phuric acid added, the liquid saturated with sodium chloride, and potas-
sium triiodide added as long as a precipitate forms. This is filtered,
washed with saturated sodium chloride solution and the nitrogen de-
termined. The action of other nitrogen bases with potassium triiodide
is tabulated and the method applied to the determination of choline and
betaine in vegetable substances.

Derivatives of Carbonic and Uric Adds, Heterocyclic Compounds. —
Ackermann (Z. physioL Chem., 47, 366) recommended benzene sulpho-

REVIEWS. 455

guanidine as a means of detecting guanidine in the presence of arginine.
3 grams of guanidine carbonate warmed in 30 cc. of water are shaken
with 6 cc. of 33 per cent, caustic soda and 4 cc. of benzene sulpho-chlor-
ide when, on cooling, white crystals of benzene sulphoguanidine separate
out. They may be recrystallized from boiling water and boiling alco-
hol, and melt at 212*^. In 100 cc. of water 0.02 gram is soluble. Ar-
ginine does not give the corresponding compound. Gumming and Mas-
son {Chem. News, 93,5, 17. From Proc. Soc. Chem. Ind, Vierona, 1903,
July-August) gave a new method for the determination of cyanates in the
presence of carbonates. KCNO -f 2HCI -f HjO = KCl + NH^Cl -f
CO3. The titration is first made in the cold for carbonates, with
Congo red or methyl orange as indicator, then an excess of acid is
added, the solution boiled and the acid excess titrated back. A second
determination of the cyanic acid may be made by boihng the solution
containing the above reaction products with excess of caustic potash
and determining by titration the amount of the latter required to break
up the ammonium chloride. Herter and Foster (/. Biol. Chem., i,
257; 2, 267) gave an indole determination, treating a dilute solution
(i : 100,000 of water) made weakly alkaline with caustic potash with a
drop of 2 per cent, ^-naphthoquinone sodium monosulphonate, yielding
a blue or blue-green color. With more indole, well formed crystals of a
bluish color are obtained. The green color fails at a dilution of i : 1,024,000.
The compound is soluble in chloroform with red color and this solution
may be used for colorimetric work. Skatole is separated from indole
by distillation and determined by a colorimetric method which rests
on the blue color caused by Ehrlich's reagent. Konto (Z. physiol, Chem.,
47, 185) detected indole in putrid albumen by distilling off 1/3 of an al-
coholic solution, making the distillate alkaline with caustic soda and
distilling, making this distillate acid with sulphuric and distilling. To
I cc. of this are added 3 drops of 4 per cent, formaldehyde and an equal
volume of concentrated sulphuric add allowed to flow onto it. With
a trace of indole the whole solution becomes a magnificent violet-red.
The reaction is visible with i part of indole in 700,000 of water, but not
in 800,000. Ronchfese (J, pharm, chim. [6] 23, 336) determined uric
add with titrated iodine, 100 cc. of urine being treated with 15 cc. of
ammonia and 15 grams of ammonium chloride, the precipitate filtered
and washed with a solution of 150 cc. of ammonia and 150 grams of am-
monium chloride per Uter, brought into solution in 300 cc. of water with
some acetic acid, 20 cc. of saturated borax and primary potassium car-
bonate solution are added and the solution titrated. The number of
grams of uric acid in i liter of urine is equal to the number of cc. of tenth-
normal iodine multiplied by 0.084 plus o.oi. i molecitle of uric acid
requires 2 atoms of iodine. Sperling (Z. Oesterr. Apoth. Ver., 44, 51)
treated 2 to 3 cc. of a i : 100 aqueous solution of phenyldimethylpyraz-
olone with 2 drops of fuming nitric add and after the appearance of the
green color, added carefully 5 cc. of concentrated sulphuric add. A
cherry-red ring was obtained at the contact zone, which spread through
the whole liquid on shaking. The reaction is characteristic for anti-
pyrine and its derivatives except amidopyrine. Other substances treated
as above gave the following results:

456

REvmws.

2CC. of ftoiution Addition of
of strength, nitric acid caused

Salicylic add i : 500

Qtilnine sulphate i : 100

'* hydrochloride. . . i : 200

Cocaine ** i : 100

Codeine " i : 100

Phenol 1 : 100

Reaordnol i : 100

Antipyrine salicylate. ... i: 200

Migranine i : 100

Tussol 1 : 100

Pyramidone i : 100

Addition of
sulphuric acid gave.

gold-yellow color

((

wine-yellow color
yellow-brown color
green color

colorless
orange-yellow
red-brown and douding
vigorous gas evolution
cherry-red zone

violet

wine-yellow color

Weehuizen (Pharm. Weekblad, 43, 1105) gave a number of microscopic
reactions for pyramidone in i : 100 solution, with iodine-potassium iodide,
bromine-potassium bromide, potassium cadmium iodide, mercuric chlo-
ride and sodium palladous chloride and in strength i : 400 with Mayer's
solution. The precipitates are all crystalline and are characteristic.

Albumens. — ^Amy and Pratt {Am, J, Pharm., 78, 121) detemuned
casein in milk by adding to 10 cc., 20 cc. of tenth-normal iron alum so-
lution (48.1 grams per liter) in the cold, filtering, washing, acidifying
with hydrochloric acid, adding potassium iodide and titrating the fil-
trate with tenth-normal sodium thiosulphate. Bordas and Touplain
(Compt. rend., 142, 1345) made use of the insolubility of the chief albu-
mens like egg, fibrin and casein and of gelatinous substances like dias-
tases and peptones in pure or suitably diluted aqueous acetone for thdr
determination. The mass must be neutral or at most only slightiy al-
kaline or add. 10 grams of butter, for instance, were treated with pure
acetone, the residue washed with aqueous acetone, the residue contain-
ing the casein and ash. Levene and Rouiller (/. Biol. Chem., 2, 481)
determined trjrptophane in the products of hydrolysis of egg albumen.
The liquid to be tested is made to contain 5 per cent, sulphuric add and
treated gradually with Hopkins and Cole's reagent (10 parts of mercuric
sulphate and 90 of 5 pel' cent, sulphuric add). Bromine water is added
till the red color of the liquid over the predpitate begins to disappear
The liquid is filtered after 24 hours and the mercury predpitated with
hydrogen sulphide. The filtrate is heated, made up to a definite vol-
ume, 15 cc. of this are treated with 2 cc. of amyl alcohol and then bro-
mine solution added with vigorous shaking till the purple color of the
alcohol solution disappears. In the presence of tyrosine the mercury
precipitate with the reagent is treated with 5 per cent, sulphuric add
till the t)n'osine reaction disappears. In the presence of cystine (i)
the cystine-tryptophane solution is titrated with bromine, (2) an add
determination is made in an ahquot portion of the solution, (3) the amount
of bromine corresponding to cystine is calculated and subtracted from
the value in (i). The bromine solution is standardized against trjrpto-
phane and cystine solutions.

Carbohydrates. — Bang (Biochem. Z., 2, 271) determined sugar after
its reduction of a copper solution by titration of the copper oxide remain-
ing with hydroxy lamine solution. The copper solution was made by
dissolving 250 grams of potassium carbonate, 200 of potassium thio-
C3ranate and 50 of primary potassium carbonate in about 600 cc. of water

REVIEWS. 457

■

at 50** to 60®, cooling to 30® and allowing 12.5 grams of purified copper
sulpbate pentahydrate dissolved in about 75 cc. of water to flow in, fil-
tering after 24 hours and diluting to i liter. The hydroxylamine solu-
tion contained 6.55 grams of the sulphate and 200 grams of potassium
thiocyanate dissolved and made up to 2 liters with water. 10 cc.
of the sugar solution were boiled quietly for 3 minutes with 50 cc. of the
copper solution, cooled quickly and titrated to colorless. The linut of
the method lies at about o.i mg. of sugar. The copper reduced by the
sugar precipitates as cuprous thiocyanate and the rest of the copper
does also after reduction by the hydroxylamine. Browne (This Jour-
nal, 28, 439) analyzed sugar mixtures by determining the total content
of reducing sugar as glucose and the polarization. If ^ = per cent,
of sugar A, y of B and a the relation of reduction capacity of sugar A to
that of glucose, b that of sugar B to that of glucose and R the per cent,
of reducing sugar as glucose ax + by = R and ax + ^y ^ P, a and
j8 being the polarization factors of A and B respectively and P the pelar-

imeter value, x = -—= — ^ and v = - — . With these formulas

ab — afi ^ b

the author has determined amounts of common reducing sugars like
glucose, fructose, galactose, xylose and arabinose. The reduction con-
stants of these sugars toward glucose were determined experimentally
to be for fructose 0.91 5, galactose 0.898, xylose 0.983 and arabinose 1.032.
Chavassier and Morel {CompL rend., 143, 966) stated that 10 cc. of a rea-
gent (i gram of m-dinitrobenzene dissolved in 100 cc. of alcohol and 35
cc. of 33 per cent, caustic soda added) with 20 cc. of i per cent, aqueous
sugar solution gives with maltose, lactose, galactose, glucose or arabin-
osej inside ofJi5 minutes, with levulose inside 2 to 3 minutes, a violet
color, while solutions of saccharose and glycogen are not changed. Al-
dehydes and ketones without the alcohol group give a red color which
can cover the violet. Uric add gives the same color as reducing sugars.
Glassmann (Ber., 39, 503) gave two indirect volumetric methods for
the determination of grape sugar. In one the sugar solution is added
to a boiling excess of the Liebig-Knapp solution (alkaline mercuric cy-
anide) or to Sachsse's solution (alkaline potassium mercuric iodide).
The precipitated mercury is filtered off and dissolved in concentrated
nitric acid. This solution is diluted, treated with i to 2 cc. of saturated
iron ahim solution and the amount of 30 per cent, nitric add necessary
for complete decolorization, then the solution is titrated with hundredth-
normal ammonium thiocyanate. (a) CHjOH.(CHOH),.COOH -^ 3Hg(CN),
6KOH»COOH.(CHOH),.COOH+4H20-h6KCN + 3Hg (6) CHjOH.(CHO
H)^.COOH -h 3K,Hgl4 + 6K0H = COOH.(CHOH),.COOH + ^Ufi
+ 12KI 4- 3Hg. One cc. of the thiocyanate solution corresponds to
oxx)i,ooi,5 gram of mercury. In the other method the grape sugar
solution is boiled 10 minutes with a measured excess of alka-
line mercuric cyanide, filtered from the mercury, the filtrate put
into a Hempel nitrometer and the nitrogen evolved on the ad-
dition of hydrazine sulphate measured (Ebler, Z. anorg. Chem,,
47> 377)- The difference between the amount of mercury taken
and the amount corresponding to the nitrogen gives the amount pre-
cipitated by the grape sugar. Arnold (Ber., 39, 1227) observed that
these methods of Glassmann cannot be used with urines containing

45^ RSVIBWS.

creatinine, for this substance reduces alkaline mercuric cyanide in the
cold and alkaline potassium mercuric iodide on warming. Lewinsky
(Berl. klin Wochschr., ^^ 125) showed that such substances as peptone,
Liebig's beef extract, hydrolyzed casein, gelatin and glycogen in solu-
tion concealed the presence of varying amounts of glucose and glucxrae-
amines. For instance in a solution containing 2 per cent, peptone 0.5
per cent, sugar gave no reduction with Fehling's solution and i per cent,
gelatin concealed 0.005 P^r cent, sugar. Grape sugar can be approx-
imately determined in fruit according to Lyon (This Journal, 28, 998)
by obtaining the amount of soluble substances and the inversion polari-
zation. The former is obtained from the density of the solution. If
a = per cent, of soluble substance, h = polarization, x = per cent, of
glucose, y = per cent, of cane and invert sugar then x -{- y = a and
1 .75-r — o.34y = 6or x = (0.34 a — b)/2 .09. Both numbers in the formula
change a unit in the second decimal for a change of 2° in temperature.
Pellet (BtUL assoc. chim, sticr, dist., 23, 1015) observed that alcoholic
digestion is of no value in sugar beet investigations and that errors come
even from filtering in the alcoholic extraction. Cellulose (of the paper)
absorbs sugar in alcohol solution and in the alcoholic digestion the
cell substance may absorb oxygen, as has been observed by Hoglund
and Crocker respectively. An absorption of sugar by cellulose does
not take place in aqueous solution. Schoorl and Van Kalmthout {Ber.,
39, 280) gave some color reactions of the most important sugars, criti-
cizing some of Pinoff' s (Ibid., 38, 3308). Ten cc. of an alcohol-sulphuric
acid mixture (750: 200), 0.05 gram of sugar and 0.2 cc. of 5 per cent
alcDholic a-naphthol solution give a red-violet in i minute if fructose or
cane sugar were used, with d-glucose or milk sugar after 15 minutes (in-
stead of 30). If the sugar solution be diluted with 10 cc. of alcohol,
the color comes in 5 minutes with the first two sugars and with last two
within 1/2 hour (instead of not at all). The ^-naphthol reaction is given
by Pinoff as characteristic for fructose but after 20 minutes fructose
gives only a slightly darker yellow color than cane sugar, while after
30 minutes both glucose and milk sugar give a weak yellow color. Pi-
noff's reaction with resorcinol is a not specially fortunate modification
of the Selivanow test. With potassium bichromate and ammonium
chloride Pinoff said only the ketoses (fructose) gave a precipitate, but
within 1/2 hour cane sugar gives at least a small precipitate. Tlie for-
mation of calcium fructosate is the only certain test for fructose. Berg's
reaction for the detection of aldoses in opposition to ketoses and non-
reducing polyoses (oxidation of the aldose to hydroxy acid with bro-
mine and yellow color of this with ferric chloride) is not reliable because
other carbohydrates color ferric chloride yellow, even though weakly.
Talon (Ann, chim, anal. appL, 11, 244) found that the esters of glucose
are always formed in the presence of alcohol and traces of acid, that they
increase in amount with the temperature and the time and that they
have no reducing power. So alcohol, even methyl alcohol, is to be avoided
in tests or determinations of sugars. Glycerol esterifies like a mona-
tomic alcohol on glucose and fructose but its action at the same dilu-
tion is less. Weehuizen {Pharm, Weekhlady 43, 1209) observed that
glucose, milk and cane sugar, starch and cellulose give with skatole a
beautiful stable violet color and with indole a brown-red , if warmed at

REVIEWS. 459

the same time with strong hydrochloric add. The skatole reaction is
deKcate to i : 300,000.

Baur and Polenske (Arb. kais. Gesundheitsamt^ 24, 576) separated starch
and glycogen by dissolving o.i mg. of the mixture in 30 cc. of water
and treating with 11 grams of finely divided ammonium sulphate. The
starch is filtered and washed with ammonium sulphite solution (11 grams
per 30 cc.). Glycogen is precipitated out of the 60 cc. of filtrate on di-
lution witii water to 500 cc. and addition of 500 cc. of alcohol. JoUes
{Ber,, 39, 96; Monatsh. 27, Si; Z. anal. Chem,, 45, 186) applied his
titration with primary potassium sulphate to the determination of
pentoses, changing them first to furfural by distilling with hydrochloric
add and steam, taking then an aliquot portion of the distillate, neu-
tralizing, treating with a measured excess of sulphite and titrating back
after 2 hours with iodine. C4H3O.CHO + HKSO3 = C.HjO.CHOH.SOjK.
Pfliiger (Pfluger's Arch., 114, 231) determined glycogen by warming 100
grams of the organ in question for 3 hours in a boiHng water bath with
100 cc. of 60 per cent, caustic potash. After cooling, the solution is di-
bted to 400 cc. and precipitated with 96 per cent, alcohol. The glycogen
precipitate is allowed to stand, then washed with alcohol and ether, then
dissolved in hot water, the aqueous solution made up to a definite vol-
ume and the glycogen determined by polarization. It may be inverted
and the sugar determined.

Alkaloids. — Georges and Gascard (/. pharm. chim. [6] 23, 513) stated
that the color of a neutral or weakly alkaUne solution of morphine with
iodic acid (yellow to reddish yellow and yellow-brown if made weakly
alkalme with ammonia) may be used for the colorimetric determination
of morphine. The iodic add color is sharpest at dilutions from i : 500
to 1 : 5000; the iodic add-ammonia color at dilutions greater than i : 2500.
The first color reaches a maximum in 1/2 minute and remains unchanged
for 15 minutes. The alkaline color is at a maximum after 2 to 3 min-
utes. Mai and Rath {Arch. Pharm., 244, 300) gave another colorimet-
ric determination of morphine, using Marquis's reagent (2 drops of 40
percent, formaldehyde and 3 cc. of sulphuric add). One cc. of a i : 1000
solution of morphine is evaporated in glass on a water bath and the resi-
due stirred with i cc. of the reagent, the deep violet colored solution
put in little tubes of about 10 mm. width and after washing out the dish,
diluted with 4 cc. of sulphuric add. The limit of delicacy is about 0.000,03
gram of the alkaloid. Radulescu {Bull. Soc. Sciinie Bucuresci, 14, 602)
added to a clear solution containing i : 300,000 or more of morphine, a ker-
nel of sodium nitrite and an add, making the solution alkaline before
the end of the gas evolution with concentrated aqueous caustic potash.
The solution is pale red to deep ruby- red according to concentration.
The color disappears on acidification, reappears on making alkaline and
is not taken up in ether, chloroform, carbon bisulphide or benzene. On
long standing or on boiling, the add solution loses its property of becoming
colored by alkalies. Herder {Arch. Pharm., 244, 120) observed that the
ease of solubility of alkaloid precipitates in alkali and alkaline earth
mercuric iodides decreases with rise of atomic weight of the metal and
that differently from the precipitate in aqueous solutions which in 30 per
cent, chloralh ydrate is almost always crystalline. The limits of sensi-
bility of pure alkaloids toward caesium mercuric iodide in aqueous (I)

460 RBvmws.

and in 3 per cent chloral hydrate (II) solutions and toward barium i
curie iodide in aqueous (III) and in chloral hydrate (IV) solutions
riven as follows:

I. II. III. IV.

Berberine i:300,cx)o 1:65,000 1:500,000 i:8o,oc

Hydrastiae and can&dine. . 1:30,000 1:1000 1:38,000 1:1400

Strychnine 1:200,000 1:40,000 1:710,000 i;43,«

Brudne 1:40,000 1:10,000 1:41,000 i: ii,o(

Quinine 1:100,000 1:30,000 i:io,ooo »: iS.ot

Cinchonine 1:90,000 i: 10,000 i; 100,000 i: ii,a

Coniine i: 1300 bdowi: 1000 1: 1300 belowi: tooo

The alkaloids were detected and localized in sections of a numtx
plants. Coniine gives no precipitate in aqueous solution. The se(
treated with barium mercuric iodide solution is washed quickly with w
and laid in a 0.5 per cent, solution of potassium bichromate in 3c
cent, chloral hydrate and acidified with a few drops of hydrochloric
A yellow to yellow-brown solution indicates the alkaloid. Jonescu 1
pharm. Ges., 16, 130) observed that by Thoms's method of pr
itation with potassium bismuth iodide and decomposition of the
dpitate with alkali not only atropine and hydroscyamine but qui
caffeine and antipyrine may be nearly quantitatively precipitated. I
I gram of quinine used the author obtained 0.9405 gram, of caf
0.9546 and of antipyrine 0.9273 gram. Monti {Gaz. chim. ital., 36
477) criticized the reaction of Alvarez for aconitine and gave one 0
own. The alkaloid (0.000Z to 0.001 gram) was treated in pore
with 2 to 4 drops of sulphuric add (d. 1.75 to 1.76) heated for 5 or 6
utes on a boiling water bath, then a small amount of resordnol a
and the whole warmed further. A yellow-red color appears, atta
a maximum after 20 minutes. Thus o.oooi gram can be detected. R
ard {Chem.-Ztg., 30, 109 on picrotoxine; Pharm. Ztg., 51, 168, 591 01
caine, 532 on quinoidine, 817 on opium alkaloids; Pharm. C«
47, 247 on morphine, 347 on cocaine, 473 on berberine, 623 on theb
727 on coddne, 1028, 1048 on narceine) has again pubUshed a :
of articles on alkaloids.

Coloring Materials, Oils. — Green, Yeoinan and Jones (Mon.
[4] 20 I, 181) gave a method for the systematic analysis of colors on
mal fibers. The method includes two operations; in the first the <
is removed from the fiber and its chemical behavior studied, in the
ond it is reduced and the course of its reoxidation studied. Given
coloring matter alone it is reduced with zinc dust and oxidized
air and with chromium triojdde; if on the fiber it is reduced with hj
sulphite and oxidized with air and potassium persulphate. Their cl
fication follows: I. Decolorized by hydrosulphite and regenei
in air: azines, oxazines, thiazines, indigo. II. Decolorized by h^
sulphite, not regenerated in air but by chromium trioxide: triphi
methane colors. III. Destroyed by hydrosulphite: nitroso and n
azo colors. IV. Not changed by hydrosulphite: pyrones, acrid
quinoleins, thiazoles, certain anthracene colors. V. Changed to bi
products by hydrosulphite and regenerated in air or by potassium
sulphate: most anthracene colors. Knowing the class, chemical 1
tions and shade the choice is limited to a few closely allied colors. Gi

REvmws. 461

(Chem,-Ztg., 30, 298, 299) gave the following main differences between
Hgnite pitch and other pitches.

Residue afler
extraction with
Melting point. benzene. Sulphur. Iodine value.

Lignite pitch 86® 0.0 2.14 93.7

Coal tar pitch 91-92° 46.0 0.31 50.0

Wool fat pitch 32** 0.0 0.00 36.9

Stearin pitch 43° 0.0 0.67 40.4

Petroleum pitch 1 33** 2.0 i . 17 49 .4

Petroleum pitch II 73** 3.5 i .09 70.3

Petroleimi pitch III 126° 4.0 i.oo 103.5

Lignite "goudron" 52® 0.0 1.88 66.5

Wood pitch 195® 42.0 0.00 140.0

A test for phenols which serves to distinguish lignite pitch from other
pitches is given. A little of the asphalt is pulverized and boiled with
some normal caustic soda and filtered. A little of a solution of a few
drops of aniline in i cc. of hydrochloric add and 10 of water with some
sodium nitrite solution added in the cold is added to this alkaline liquid.
A red or reddish brown color or precipitate appears according to the
phenol content. Without phenols the liquid is only yellow. Valenta
{Ibid,, 30, 266) used methyl sulphate in the detection and determina-
tion of tar oils. It dissolves the hydrocarbons of the benzene series
or mixes with them in all proportions while hydrocarbons with open
carbon chain and petroleum distillates do not dissolve in the cold and
even esters are difficultly soluble. A measured amount of the oil sam-
pk is poured into a 100 cc. glass stoppered measuring cylinder and 1.5
to 2 times as much methyl sulphate added. The whole is shaken and
the volume of the lower liquid after separation read, the difference giv-
ing the amount of oil dissolved. The tar oils may be precipitated from
solution by caustic potash. A gravimetric determination is possible
if the caustic soda be neutralized, the whole shaken with alcohol and
the oil weighed after driving off the latter. Rebs (Protokoll i. Stzg. Komm.
Bekdmpf. Missverstdnd., Herst, etc,, Farben u, Malmaterialien Num-
berg, 1906, 35) determined turpentine resins or abietic acid in resins of
various sorts, oil and copal lacs, pitch and paper by treating 10 grams
of substance with 20 to 25 cc. of 10 per cent, alcoholic potash, warming,
decomposing the soap, after cooling, with dilute hydrochloric add and
filtering, washing and drying the separated resin. The pulverized resin
is extracted with 50 cc. of warm petroleum ether, the abietic add pre-
dpitated from the filtered solution as ammonium salt, separated and
the ammonia driven off by gentle heat. The remaining resin-like mass
gives the resin content of the substance. For less accurate work the
substance may be extracted with benzene and the saponification omitted.
Vaubel (Z. offentl, Chem., 12, 107) simplified his last year's method for
determining turpentine oil in the commerdal product by adding to i
to 2 grams of oil in chloroform 100 cc. of water, 5 grams of potassium
bromide and 10 cc. of concentrated hydrochloric add or a corresponding
amount of sulphuric add and then titrated potassium bromate solution till
a permanent bromine color appeared. Genuine oils of turpentine have
a bromine absorption of 220 to 230 while that of substitutes falls as low
as 16 sometimes. Paulmyer (La Savonnerie Marseillaise, 6, No. 62;

463 RBVtBWS.

SHfensiederztg. 33, 286) made use of the "critical sohition temp
hire" of various fatty oils for detecting adulterations in cocoanut
A cloudy mixture of fatty adds warmed with acetic add will dear
suddenly and the temperature at which it does so is constant for t
add (critical solution temperature). With pure cocoanut oil fatty a
the liquid clears at 33°. Warmed a little above this and allowed to >
with stirring it will cloud at 33°. The value for various acids is:

Cocoanutoil 33° Rape oil 107° "Palmkemfir". . 49°

Earthnutoil 90° Linseed oil 7«° Stearic add 94° (conunerd

Sesam^oil 89° Olive oil 93° Oleic add 98" "

Nigeroil 85* Cotton oil 81.5°

CastoroU 13.5° "Mafentalg". . 88°

With mixtures of the fatty adds the difference in solubility temp
tures is proportional to the quantities of single adds present. Twiti
(This Journal, 28, 196) found that a fat with excess of water and
than I per cent, of naphthalenestearosulphonic add will be almost c
pletely saponified in 8 to 10 hours. This add tried on a sohibk gh
ide, triacetin, showed about the same hydrolyzing power as hydroch
add, but the latter has practically no action on an insoluble glyct
like a common fat while the organic add works ahnost as well as 1
soluble ester. The capadty of these snlpho fatty adds to dissolv
well in fatty adds as in water and to make fatty adds soluble is of v
in the separation of solid and liquid fatty adds.

Butler, Milk. — Robin {Compi. rend., 143, 512) detected cocoa fat
margarine in butter by means of the facts that the fatty adds of c
fat are nearly completely soluble in 56.5 per cent, alcohol at 15° ^
those of butter are only partly and those of margarine difficultly soh
and that the content of water-soluble fat in butter is much larger
in the others. The butter sample is heated with alcoholic ca
potash and the liquid so diluted with water after cooling that a 56.^
cent, alcohol results. A blank determination with the same am
of alcohohc caustic potash is carried out, and both are titrated
half-normal hydrochloric acid in 56,5 per cent, alcohol. The diffei
gives the amount of acid needed for the separation of the fatty 1
in the first flask and the soap there is decomposed with this amoui
add, the liquid diluted to 150 cc. with 56.5 per cent, alcohol, cook
15° and filtered. In 50 cc. of the filtrate is found by titration with ti
normal caustic potash the amout of fatty adds soluble in 56.5 per
alcohol. Another 50 cc. are evaporated to 15 cc., the water-inso
fatty adds filtered out, dissolved in a mixture of alcohol and ethe
trated, thus giving the amount of fatty adds soluble in 56.5 per
alcohol but insoluble in water. The difference between these two
ues gives water-soluble acids.

Butter. Mttgatiae. Cocc

Adds soluble in alcohol 11. 67-14. 83 2.67 46

" insoluble in water 551-8.31 256 44

" soluble " " 5.9a- 6.66 o.ii i

Water insoluble; water soluble 8,3-11.7 a3i-7 m5

Alcock (Pkarm. J. [4] 23, 28) detected formaldehyde in milk by
ing to 2 cc. the same volume of 20 per cent, caustic potash, shaking 1
hen adding an excess of strong hydrochloric add and wanning card

RBvmws. 463

A cxjagulum is obtained, colored more or less deep violet according to
the amount of formaldehyde, while the Hquid below remains colorless.
Bonneina {Phatm. Weekblad, 43, 434) found that the mean freezing point
of milk is — 0.555° ^^^l that it is influencd by dissolved crystallizable
substances. The amount of water added can be determined by means of

the change in freezing point. D ^ "'^^f ^f '^ orw = °'^^^J^^^'^-88.5,
** ** '^ w 4- 88.5 D

where w is the weight of water added in grams and D is the freezing
point in degrees below zero. This formula applies to milk with an aver-
age of 88.5 per cent, water. The freezing point rises a few hours after
milking because of the formation of ammonia by bacteria and by which
some phosphates are precipitated. The subsequent lowering is due to
the formation of lactic acid which dissolves some of the precipitated
phosphates. Comanducd (Rend, accad, scienze fisiche e matem, Napoli,
1906, April) observed that a certain amount of pure milk requires al-
ways the same amount of potassium permanganate to oxidize the or-
ganic matter in it. Watering can be detected in this way. i cc. of
cow's milk requires 50 to 52 cc. of tenth-normal permanganate, goat's
milk 44 to 46, sheep's 43 to 48, ass's 55 to 58, woman's 53 to 60, these
oxidation indices sinking if the milk is adulterated. Cow's milk with
10 per cent, of water requires 44 cc. permanganate for i cc, 20 per cent.
39. With half the cream removed the oxidation index fell also to 44
to 46. Cow's milk with an oxidation index of less than 50 is therefore
suspicious. Lelli (Arch, farmacol., 5, 645) detected primary sodium
carbonate in milk by means of aspirin. Milk, an equal volume of water
and a little aspirin are heated to 60*^, the opaque solution filtered, and
the filtrate treated with a little 10 per cent, ferric chloride solution. In
presence of the carbonate an abundant red-yellow precipitate is formed.
0.5 per cent, of carbonate can be detected. Trillat and Sauton (Compt.
rend., 142, 794; 143, 6; Bull. soc. chim. [3] 35, 906, 1207) gave a deter-
mination of the albuminous substances of milk and the casein of cheese,
making use of the property of formaldehyde of rendering these albumin-
oids insoluble without changing their weight. 5 cc. of milk are diluted
with 20 cc. of water, the hquid boiled 5 minutes, 5 drops of commercial
formaldehyde added, the hquid boiled for 2 or 3 minutes longer, allowed
to stand for 5 minutes, treated with 5 cc. of i per cent, acetic acid, shaken
and the pulvenilent precipitate collected on a tared filter. The pre-
cipitate is washed with water, extracted with acetone, dried at 70° to
80° and weighed. The casein determination is similar.

Blood, Glucosides. — Carlson (Z. physiol. Chem., 48, 69) stated that
3 per cent, hydrogen peroxide is to be preferred to ozone in turpentine
oil for the guaiac blood test, as it gives a sharper and more certain re-
action. Schumm (Ibid., 50, 374) disagreed with this. Horoskiewicz
and Marx (Berl. klin. Wochschr., 43, 11 56) observed that 10 to 15 per
cent, quinine solution is a good extracting agent for old blood flecks,
giving a brown yellow solution with a characteristic absorption band
between C and D (wave lengths 628 and 596) instead of the original oxy-
hemoglobin bands. Eight per cent, quinine solution mixed with the blood
in ratio 2 : 4, heated slowly and gradually to boiling and treated after
cooling with 2 to 3 drops of fresh ammonium sulphide solution will give
with normal blood a dirty brown-green, but if the blood contain carbon

464 RBVIBWS.

monoxide it stays carmine-red. Neisser and Sachs (Ibid., 42, 44; 43
detected human blood by the use of specific sera. If human serutr
mixed with an antiserum obtained by treatment of rabbits with hui
serum a mixture is obtained which is capable of interaction. If a se:
hemolysin be added, it must remain inactive. Only human and a
sera cause cessation of hemolysis in this case ; all others have shown tb
selves inactive. Blood spots 3 months old on linen gave the react
Rabbit serum (0,25 cc.) is placed in the liquid to be tested for hui
blood and mixed with the antiserum (prepared from rabMts tres
previously with human senim), the mixture allowed to stand for i 1
at 37°, then i cc. of 5 per cent, sheep blood emuhdon added and
stand again at 37° will give proof of the presence of human bloot
hemolysis fails to appear. It is prevented by the interference of o.(
cc. of human serum and even 0.000,001 and 0.000,0001 cc. can be dete
by distinct differences in the hemolytic action. The method is at 1
as delicate as the detection by precipitation. Piorkowslci (
pharm. Ges., 16, 226) observed that the hydrocele liquid gives coag
tion with human but not with cow's milk. Every clear human \
Uquid shows this reaction. Cow's milk reacts similarly with cattle sei
The hydrocele liquid gives soon with human blood a red predpi
while other bloods dissolve; only homologous bloods are precipita
heterologous are dissolved. Sardia and Caffart (Compt. rend., 143,
found t^t if a drop of blood solution be evaporated slowly on a mi
scope slide, then treated successively with a drop of chlorine water,
of pyridine and one of ammonium sulphate solution, many chlorht
tin crystals, rhomboidal rods of varying sizes isolated or in groups 0
intense brown-red or bright red color, appear. At the same time
intense red hemmochromogen crystals form in varying numbers in :
tuft and brush shapes. Van Itallie {Pharm. Weel^lad, 43, 27, 33;
Ber. pharm. Ges., i6, 65} found that the blood of men and apes ■
tained catalyzers even after being heated to 63° for 1/2 hour. ;
of blood (i; 1000) were so heated, then cooled to 15° and mixed
3 cc. of I per cent, neutral hydrogen peroxide solution when the b
of men and apes liberated oxygen while that of horses, cattle, sn
goats, sheep, rabbits, rats, hares, hens, doves, bony fishes and frogs
not. Wonian'sniilktoo,giviugunheated 24.8cc.of oxygen, when heat
the same way for 15 minutes gave 18 cc., 45 minutes 7.5 and 60 min
4.0 cc. of oxygen. The conclusions to be drawn from the above ■
opp<»ed by Arnold {Apotk. Ztg., 21, 220) and upheld by Van It
(/Wet., 31, 230).

Bread, Flour. — Pohl (Z. angew. Chera., 19, 668) determined ale
in freshly baked bread by distilling, after addition of water to pre
burning, in portions of about i kg. The united distillates from a'
4.5 kg. were several times redistilled, each time after saturation witl
dium chloride, only a part being distilled. The last time the disti
was saturated with caknum chloride and a part distilled off. I
the density determination in the first 50 cc. of this the alcohol was
culated. It was then salted out with potash, distilled and conve
into ethyl iodide for identification. 0.0744 and 0.0830 gram were f(
in white bread prepared from leaven, 0.0508 and 0.0547 gram in
from compressed yeast. Marion {Ann. cbim. anal, appl., 11, 134)

REvmws. 465

termined gliadin by digesting 10 grams of flour with 50 cc. of alcohol
(73** French) in a closed flask at 40*^ to 45**, shaking frequently, cooling
to 15° to 20°, decolorizing with 0.8 gram of animal charcoal, filtering
and polarizing. The percentage of gliadin is equal to the reading in min-
utes multiplied by 0.0722. Shaw (This Journal, 28, 687) tested bleached
flour by boiling i kg. of it for 4 hours with 95 per cent, alcohol, evapo-
rating the filtrate and wash alcohol nearly to dryness, and extracting the
residue with a mixture of equal parts of alcohol and ether. This was
evaporated to a syrupy consistency, allowed to dry in a thin film on the
inside of the dish and a drop of a solution of diphenylamine in sulphuric
add caused to flow over the surface. The drop called forth a blue color
in bleached flours, none in unbleached, the color being due to higher
oxides of nitrogen. Gastine (CompL rend,, 1421 1207) gave a new method
for the analysis of flour and the detection of rice in wheat flour. The
flour was moistened with a solution of coloring matter, the mass dried
carefully, heated for a few minutes at 110° to 130 ^^ and then investigated
under tiie microscope in a drop of cedar oil or Canada balsam. The
hilum of the starch kernel, at least of some forms of starch, shows in the
form of red points ('Tunktierungen")- The polyhedral starch kernels
of rice give a plain, relatively large reddish hilum, the wheat starches
seldom show one. Various blue, green, brown and orange coal tar col-
ors were used, the most suitable concentration being 0.05 gram per 100
grams of 33 per cent, alcohol. One to 2 per cent, of rice flour can be de-
tected with certainty. Maize and budcwheat starch granules behave
Hke that of rice, and many other starch granules like those of the
wheat.

Wine. — Billon {Ann, chim, anal, appl.y 11, 127) observed that the to-
tal alkalinity of wine ash as ordinarily determined and the alkalinity of
the ash of salts precipitated from wine by an alcoholic ether mixture
always lie close together but the difference is never negative; for 100 cc.
calculated as sulphuric acid it is about o.oi and never exceeds 0.02 gram.
But if the wine contain free sulphuric acid the alkalinity of ash of salts
precipitated by the mixture remains unchanged while the total alka-
Hnity is lowered and the difference becomes negative. This is the best
test for free sulphuric add in wine. With wines containing much cal-
cium sulphate it is necessary to add potassium chloride to take care of
the tartaric add. The same author (Rev. intern, falsific.y 19, 57) gave
also a method for the determination of glycerol in wines. 50 cc. of an
unsugaredwine are concentrated to 15 cc, 10 per cent, caldum hydroxide
added in slight excess, and the mixture evaporated to sirupy thickness.
The cooled residue is washed with absolute alcohol and then with abso-
hile ether completely into a 100 cc. flask and filled with the
latter to the mark. All except glycerol is predpitated. An ali-
quot portion of the filtrate is evapoiUted, dried and weighed;
the residue is pure glycerol. With more than 10 per cent, sugar
an amount of slaked lime equivalent to the sugar content is added
to the 15 cc. concentrate, the whole evaporated to sirupy thickness
and the residue extracted 7 or 8 times with 10 cc. of alcohol each time,
cooled and made up to 100 cc. An aliquot portion of the filtrate is evap-
orated, the residue extracted with alcohol and ether and the deter-
mination carried out as above. Cordier (BiM. sciences Pharmacol.^ 13, 79)

466 RBvnsws.

tested wines for poisonous substances by lowering the alcohol content
to 8 per cent., adding 25 grams of sugar per liter, introducing a colony
of wine yeast and letting the liquid stand in a well stoppered bottle for
8 days. The pressure developed and the gas evolution give with some
practice without further test the grade of activity and the increase of the
yeast. If the wine contain any poisonous substance, it cannot be worked
into a sparkling wine. Primary potassium sulphate hinders the test
when present in considerable amount only, but traces of fluorides and
other poisonous substances are enough. Mycoderma vini and M . aceti
appear to be still more sensitive. Ross and Mestrezat {Ann. chim, anal.
appLy II, 41) determined the volatile acids of wine by distilling 20 cc.
under diminished pressure, repeating the distillation twice, adding 20
cc. of water each time. With lactic acid present, about 5 cc. must be
left in the flask each time, otherwise the wine can be distilled almost
dry without danger of decomposition. The volatile adds may be ti-
trated in the distillate using phenolphthalein or determined by differ-
ence between total acidity and non-volatile add content. Vitali {Vor-
trag VL Intemat. Kong, angew. Chem. Rome, 1906; BolL chim, farm.,
45f 701) found that toluene extracts only salicylic acid from aqueous
and alcoholic solutions. The evaporated toluene extract may be taken
up with water and tested with ferric chloride or a solution of copper
sulphate so dilute that it is colorless. If the solution tested by the lat-
ter substance be evaporated to dryness the residue is green, which is not
true without the add. The test may be used on wines and foodstuffs.
Schmidt (Z. Nahr.-Gentissm., 11, 386) distinguished between fermen-
tation vinegar and *Vinegar essence" by making the sample al-
kaline with caustic soda, evaporating an amyl alcohol extract, adding
water and a little sulphuric add and treating with iodine -potassium
iodide solution. If the mixture is unchanged the vinegar is "essence";
if a cloud or predpitate appears, it is fermentation vinegar. Very small
amounts of fermentation vinegar are tested as above in 100 cc. of dis-
tillate from sample.

Fibers in Mixed Weaves. — Lecomte (/. pharm. chim. [6] 24, 447)
gave a method which makes it possible to distinguish and to count with
a magnifying glass the threads of different fibers in mixed weaves. It
consists in diazotizing the silk and wool which contain each an amino
group and to couple on a phenol group in alkaline solution. Wool con-
tains also sulphur and so by simultaneous action of an alkaline plum-
bite solution, lead sulphide is formed, concealing the azo color. Plant
fibers contain neither of these substances and remain unchanged. One deci-
meter of decolorized weave is put into 30 cc. of 10 per cent, nitric add 30 cc. of 5
per cent, aqueous nitrite solution added during 3 minutes gradually and with
stirring, allowed to stand 10 minutes, washed for 2 minutes with water,
pressed out and cut into 2 pieces. Over one is poured 40 cc. of /?-naphthol
plumbite solution (50 grams of caustic soda in 500 cc. of water, 25 grams
of lead acetate in 300 cc. of water slowly added and 5 grams of ^-naphthol
added to the clear liquid) ; the other piece is treated with 40 cc. of resor-
cinol-plumbite solution (2 grams of resorcinol in place of the naphthol).
The weaves are washed after i hour at not above 20° with running water
for 15 minutes, placed for 5 minutes in 100 cc. of 0.5 per cent
hydrochloric add, washed again for an hour with running water,

REVIEWS. 467

pressed out between white filter papers and driqd in the dark. Silk
will be colored rose-red by the ^-naphthol solution, orange-yellow by
the resordnol, the wool black by both, while the plant fibers will remain
white.

A portion of the work for this review was done in the library of the
department of chemistry of Cornell University, through the courtesy of
Professor L. M. Dennis, for which the writer desires to express his thanks.

Dnivbksitt op Nbbraska,

LINCOLIV, NBB.

mXER-RELATIONS OF THE ELEMENTS. *

By Hbrbbrt N. McCoy.
Received January i, 1908.

The history of chemistry from the first to the sixth decade of the 19th
century shows all too clearly how unsuccessful were the attempts of Dal-
ton and of his contemporaries and successors to fix the atomic weights
of the elements by means of arbitrary rules regarding the numbers of
atoms which imite with one another, even when these rules were supple-
mented by knowledge of chemical behavior. Consistent results were ob-
tained only by the aid of the auxiUary hypothesis of Avogadro ; which, pro-
posed in 181 1, almost unappreciated until after its reanimation by Can-
nizzaroin 1858, to-day is the comer-stone of the magnificent structure of
which the atomic hypothesis is the foundation. Between the physical
and chemical properties of the elements and their atomic weights, fixed
by means of the hypothesis of Avogadro, there are, as is well known to
every chemist, the most fundamental relationships. Of these relation-
ships none is perhaps more significant than that discovered in 18 19 by
Dulong and Petit. This law supplemented by the relations discovered
by Neumann and by Kopp means, simply, that for soHd substances,
the atoms of all elements, either free or combined, have approximately
the same capacity for heat. A few months ago Lewis showed that atomic
heats at constant volume are much more nearly constant than the ordi-
nary values, which refer to constant pressure.

During the fifty years following Dulong and Petit's discovery many
other relations, more or less quantitative, were found between the proper-
ties of elements and their atomic weights; all these relationships were
correlated, in 1869, in the Periodic Law of Lothar Meyer and Mendel-
&£F, which may be concisely summarized in the statement that, in general,
the properties of elements are periodic functions of their atomic weights.
The comprehensive nature of the law is best appreciated when we re-
member that these properties include chemical nature, valence, atomic
volume, hardness, thermal expansion, crystalline form, conductivity
for heat and electricity, melting-point, boiling-point, spectral wave-
length and ionic mobility, as well as other properties of less importance.
What could illustrate more forcibly the validity of this great law than
the startling coincidence in properties of the element, germanium, dis-
covered by Winkler in 1886, and the hypothetical eka-silicon described by
Mendel6efF sixteen years earlier, or the equally accurate predictions of
the same famous chemist concerning eka-boron and eka-aluminium.

' Read at the Chicago meeting of the American Chemical Society.

468 REVIEWS.

Very early in the history of the atomic hypothesis, in 1815, to be ex-
act, Prout, a London physician and dilettante in chemistry, observed
that the currently accepted atomic weights were, for many elements,
exactly or very approximately whole numbers, the atomic weight of
hydrogen being taken as unity. Prout contended that all atomic weights
should be whole numbers and suggested as the explanation that the ele-
ments were composed of hydrogen in various stages of condensation.
These views were vigorously opposed by many who were best qualified
to judge; Berzelius, especially stoutly denied that the fractional values
could be wholly due to experimental error. Later, Stas, whose monu-
mental work on atomic weights was largely inspired by a desire to settle
this vital question, convinced himself and also the chemical world in
general that Prout's idea was fallacious; at least as to the atomic weights
being whole numbers. And this conclusion is strongly confirmed by
all the beautiful work of more recent years upon atomic weights. Yet
the fact that the values foimd on the basis of oxygen equal 16, are, in so
many cases, so very near whole numbers, must have some far-reaching
significance. Calculations by Strutt, based on the theory of probability,
have shown that the chances are only about one in a thousand that any
set of numbers, assigned purely at random, would differ from whole
numbers by so little as do the atomic weights.

Prom the critical consideration of all the evidence of the sort to which
reference had been made, one may safely say that there was abundant
reason for the belief that the elements are but various modifications of
a primitive substance; btU^ the case was not yet proven.

Far from being novel, the idea of the unity of matter, with the attend-
ant possibility of transmutation, antedates the whole science of chemis-
try. True it is, of course, that the ancient and medieval views had other
bases than the facts which we have been considering; was not the trans-
mutation of a base metal into gold a far simpler change (apparently)
than many chemical and metallurgical transformations with which men
were even then familiar? To the alchemist the argument was convincing
and the goal alluring; and with enthusiasm worthy of a twentieth century
** promoter" he often, in bombastic and ambiguous terms, described as
accomplished that which existed only in his imagination.

Before the close of the iSth century, however, attempts at transmuta-
tion were recognized as futile and one thing appeared certain, even at
that time; the elements were not to be disintegrated or altered by such
chemical processes as those which served to decompose their compounds.
Most of the foregoing facts and arguments are well known to every chem-
ist, since they constitute important chapters in the history of our science.
I have merely touched upon them in order to lend perspective to that
which is to follow.

During the past eleven years enormous advances have been made
along entirely new lines, in the knowledge of the interrelations of the ele-
ments and the nature of matter. This new knowledge had its origin
in the discovery by Becquerel, in 1896, of the radiations emitted by ura-
nium. It was found by Becquerel that all uranium compounds send out
rays capable of affecting a photographic plate through light-proof paper
and also of enormously increasing the electrical conductivity of air, by
ionizing it. Schmidt, and independently Mme. Curie, found that all

REVIEWS. 469

compounds of thorium produce similar rays. Scientists will never for-
get the intense interest taken in the discovery by the Curies, of radium,
a substance which possessed the properties of uranium and of thorium
augmented more than a milHon-fold. There were also new properties:
powerful physiological effects, evolution of light and even of heat, it hav-
ing been found by Curie and Laborde that the temperature of a tube of
radium is always perceptibly above that of its surroundings. Here then
was a most marvelous result; the continuous and seemingly undiminished
production of energy by a portion of matter, which appeared to suflFer
no chemical change. It even seemed as if a source of perpetual motion
had been found.

It was soon clearly established that the activity of radioactive sub-
stances was not due to the excitation of any known radiation. Some
scientists, however, including Lord Kelvin, Becquerel and the Curies,
imagined, as the source of the observed energy an unknown cosmic radia-
tion, which was intercepted and transformed by the radioactive body:
the elevation of temperature of radium above its surroundings being,
according to Kelvin, analogous to that of a piece of black paper in a bot-
tle exposed to sunlight.

The solution of the mystery of radium is largely due to Rutherford
and his colleagues. There were two main problems (i) the nature of
the radiations, and (2) the nature and cause of radioactivity. The radia-
tions were found to be of three sorts, called the a-, /?-, and ;r-rays. The
nature of the /9-ra)^ was first established. The photoactivity is chiefly
due to these rays; they readily penetrate light-proof paper; are easily de-
flected by a magnetic or an electric field, and in such a direction as to show
that they are negatively charged. In this respect, as well as in many
others, the ^-rays closely resemble cathode rays, which latter have been
shown to consist of particles of negative electricity, called corpuscles.
These corpuscles, which are shot out from the cathode with i/io of the
velocity of light, have inertia, due to the moving electric charge; and
having inertia, they have mass. The mass of a corpuscle is very small;
about I / 1000 that of an atom of hydrogen. The ^-rays seem to differ
from the cathode rays only in having greater velocity. The ^-rays are
apparently very penetrating X-rays ; and just as the latter are produced
by the cathode rays, the former are produced by the ^-rajrs. ^

The a-rays are quite different: they are unable to penetrate a single
sheet of writing paper; but they are the chief cause of the ionization
of the air and the elevation of temperature. Rutherford has shown
that the a-rays consist of positively charged material particles, each hav-
ing a mass 2 (or 4) times that of an atom of hydrogen and moving with
1/3 the velocity of light. It was now readily understood that the kinetic
energy of the rays was the proximate cause of the observed evolution of
heat by radium.

The counterpart of the phenomena of the rays was foimd in the change
which occurred at the same time in the active substance itself. Crookes
had observed that by treating uranium nitrate solution with an excess
of ammonium carbonate a minute insoluble precipitate remained; this
precipitate, called uranium X, was found to possess practically the whole
of the photoactivity of the uranium from which it had been separated.
Rutherford and Soddy precipitated thorium with ammonia and obtained

470 REVIEWS.

from the filtrate a trifling residue called Thorium X, which possessed a
large share of the original activity. The activity of the ThX decreased
gradually with time; in 4 days the activity was reduced to 1/2, in 8 days
to 1/4. The loss of activity occurred according to the exponential law,
or, as the physical chemist would say, the change was that of a first order
reaction; in each imit of time a fixed fraction of the remaining activity
disappeared. To some scientists it did not appear remarkable that the
radioactive principle of uranium or thorium could be separted by a
chemical process; in fact the original activity of these elements was "ex-
plained" as due to the presence of these highly active "impurities."
Nor did the loss of activity by UX or ThX, considered by itself, seem
extraordinary; it might be likened to loss of temperature by a cooling
body.

But a most extraordinary thing was discovered by Rutherford and
Soddy, in the further study of this same thorium experiment. The
thorium from which ThX had been removed had, at first, only about
1/3 of its original activity; but in just that measure in which the separa-
ted ThX lost its activity, that of the thorium itself returned ; imtil, after
a period of several weeks, the thorium had entirely regained its original
activity, while the ThX had become inactive. A repetition of the treat-
ment with ammonia now yielded a new portion of ThX equal in amount
to that first extracted! The behavior of UX was similar. Ruther-
ford and Soddy then announced their now celebrated Disintegration
Hypothesis: the radioactive atom is a complex system, made up of cor-
puscles and a-particles, in rapid orbital motion; of such systems some,
in the course of time, became unstable and disintegrate, the corpuscles
and a-particles resulting from such atomic disintegrations constitute
the rays, the kinetic energy of which existed previously in the atom.
The residual portion of the atom, after the escape of a corpuscle or an
a-particle, constituted an atom of a new substance (e. g. ThX), which
atom might have even less stability than the original.

In all, over 20 distinct radioactive substances are now known. These
constitute three series, in each of which each member produces, by its
radioactive disintegration, the following member. The evidence in the
case of each change is just as clear as in that of thorium and thorium X.
One case i:i particular, that of the formation of radium from uranium,
is of especial interest. It has been shown in two different ways that
radium and uranium are always associated in minerals, and, moreover,
in a perfectly fixed proportion, as demanded by the hypothesis that the
one is the product of the other. Further, uranium minerals freed from
radium by chemical processes, slowly but surely reproduce the latter
substance.

The chemical and physical properties of radium and its salts clearly
show it to be a member of the alkaline earth family; its atomic weight
is according to the latest determination of Mme. Curie, 226.5, assuming
the elements to be bivalent. The formation of radium by uranium
was then a clear case of the transmutation of one element into another
whatever might be thought of such changes as that involving the produc-
tion of ThX.

l]The properties of the a-particle led Rutherford and Soddy to suggest,
in 1902, that it might be an atom of helium. In 1903 Ramsay and Soddy

REVIEWS, 471

were able to show that radium emanation actually does produce helium
in. the course of its spontaneous disintegration; this observation has,
since then, frequently been confirmed. The formation of helium seemed
to be another real transmutation ; since the element radium imdoubtedly
produced the emanation which yielded in turn another element, helium.

If now uranium gives ultimately radium, what is the final product of
the disintegration of the latter? Bolt wood believes it to be lead, since
he finds that all uranium-radium ores contain notable amounts of lead.
In the disintegration of an atom of radium five distinct a-ray changes
occur. If each produces one a-particle, with atomic weight of 4, the
final product of radium should have an atomic weight of 226.5 — 20 = 206.5,
whereas the atomic weight of lead is 206.9.

There is now no shadow of doubt that the rate of radioactive change
is entirely independent of the form of chemical combination of a radioac-
tive substance; temperature is also without influence; a given trans-
fomiation occurs at precisely the same rate at the temperature of liquid
hydrogen and at white heat. It thus appears that radioactive change
is a natural process which is entirely beyond man's control, and impor-
tant as is the establishment of this fact, the discovery falls far short of
being a complete solution of the problem of the transmutation of the
elements.

A further step in this direction seems to have been taken within the
past few months, however, by Sir William Ramsay. The alchemist in
his attempts to alter the elements, could apply only the forces of moder-
ate heat and chemical affinity. The chemist of the 19th century could
command enormously greater ranges of temperature as well as power-
ful electrical forces; all these means were likewise without avail. And
now Ramsay has tried a new force, that of radioactive radiations: corpus-
cles and particles of matter, both electrically charged, projected with al-
most inconceivable velocity. The results of the experiments of Ram-
say and Cameron may be stated very briefly; radium emanation acting
on water produces neon; while with a solution of copper nitrate it pro-
duces argon; and even the copper itself appeared to be attacked, for
when the latter element was removed by means of hydrogen sulphide,
the filtrate was found to contain lithium. A blank experiment, in which
the radium emanation was omitted but all other conditions remained
precisely the same, yielded no trace of lithium. The formation of Uthium
from copper has been observed four times by Ramsay and though, as he
writes in a private communication, **he was loth to believe," yet he *Tiad
to chronicle the results after four repetitions." The importance of these
facts, provided they are not due to some spurious cause, needs no com-
ment. In this connection it occurred to me that radium emanation,
or in general radioactive radiations, might act on copper in the solid
state; if so, uranium-radium minerals which contain copper should con-
tain lithium also. For another purpose I had, more than a year earher,
separated a sample of pitchblende into its principal constituents. There
was a considerable quantity of copper; there was also a solution which
could contain only alkali and ammonium salts. It was but the work of
an hour to examine this solution for lithium ; in the spectrum the charac-
teristic red line of lithium was next in brilliancy to that of sodium ! Later,
from one to two grams of four other uranium minerals, including sam-

472 REvmws.

pks of pitchblende, camotite and gummite, trom ^laely sepdrated locali-
ties, were examined for copper and lithium: all contained lithium and all
Iwit one contained copper. The presence of lithium in a mineral free
from copper may be explained upon the supposition that the transmuta-
tion of the latter metal had been completed.

The view that the atom is a complex system made up of negatively
charged corpuscles and positively charged a-particles has been made
the subject of extensive mathematical studies by J. J. Thompson.
Thompson's hypothetical atom is a system composed of rings or shells
of corpuscles in a uniform sphere of positive electrification. The elec-
tropositive or negative nature of atoms, their chemical affinity as well
as their dual valence toward hydrogen and oxygen are all most beauti-
fully and at the same time simply explained by the behavior of such
imaginary systems, the properties of which vary periodically with increas-
ing number of corpuscles and so fulfil another indispensable requirement
of any hypothetical atom. Numerous other properties of matter also
receive a satisfactory explanation in terms of the new hypothesis: among
such are the conductivity of metals for heat and electricity, bright line
spectra and the 2^eman effect. That the corpuscular theory is the
last word on the nature of matter, none will contend: Thompson him-
self writes, *'The theory is not an ultimate one; its object is physical
rather than metaphysical. From the point of view of the physicist, a
theory of matter is a policy rather than a creed ; its object is to connect
and coordinate apparently diverse phenomena, and above all to suggest,
stimulate and direct experiment."

The view just given of the nature of the atom accounts also for its
radioactivity. A system of revolving corpuscles of negative electricity
would slowly radiate energy, at an almost infinitesimal rate it is true;
but the final result would be to produce in the system a condition of in-
stability which would cause its partial disintegration and rearrangement.
Is then all matter undergoing evolution? That is the question which
cannot be positively answered at present. And yet there is much evi-
dence that many elements are faintly radioactive. Such activity may
be either intrinsic or due to minute amounts of known highly active
substances. Radioactive matter is easilv extracted from the air by
means of a negatively charged wire exposed out of doors, as first shown
in 1901 by Elster and Geitel. The rate of change of this active matter
indicates that it is chiefly the excited activity, Ra A, B and C, resulting
from the disintegration of radium emanation. A simple experiment
which I recently tried shows the correctness of this view. Air was lique-
fied and about a liter of the Uquid was allowed to evaporate, and the last
portion was examined for radioactive emanations. A considerable ac-
tivity was found, the rate of change of which plainly indicated the pres-
ence of radium emanation together with small amounts of other active
matter. The excited activity, which is deposited everywhere by the
radium emanation contained in the air, is certainly the cause of a por-
tion of the observed activity of common things. Nevertheless, at least
one common element, potassium, seems from the studies of Campbell
to possess slight intrinsic activity.

On the other hand, it is found that a-particles having less than a cer-
tain minimum velocity are unable to produce ionization of the air. It

CORRBCTION.

473

is therefore quite possible that some atomic transformations may occur
which do not show radioactive phenomena, the initial velocity of the
a-rays being, in such cases, below the minimum. Such change might
occur at rates far greater than that of uranium and still be impossible of
detection in the course of a lifetime by gravimetric or even probably
by spectroscopic methods, since in the case of uranium the rate of change
is so slow that no more than o.ooi per cent, disintegrates in 10,000 years;
and yet the ionization method of recognizing a-radiations is so surpass-
ingly delicate that one may detect with certainty the activity due to a
single milligram of uranium in any form of combination!

In conclusion, it may be said that while the work of the 19th century
produced abundant and varied evidence that between the elements there
exists the most intimate interrelationships, the researches of the past
few years of this new century have shown the fundamental significance
of these relationships and lead us to the conclusion that the elements
may no longer be considered immutable ; that matter is probably of but
a single sort, of which our commonest elements represent the more stable
forms, which have resulted from a process of natural evolution.

Ukiversitt of Chicago.

CORRECTION,

Plot and footnote omitted by mistake from article on "The Corrosion of
lion and Steel," by W. H. Walker, A. M. Cederhohn and L. N. Bent, in
September number of this Journal, 1907. The plot should be inserted to

lOOr

)8 Zrem Fe.

74 NBW BOOKS.

ixxwnpany page 1260 and the footnote' added to "2" at bottoi
age 1255.

Plot showing the relation between speed of corrosion of iron in water am
artial pressure of the oxygen in the carbon-dioxjde-free atmosphere.

BXPGRIUBNTAI. DATA.

No. I. No. i.

FcrceDtage oxygen. GfStni Iron diuolvcd. PcrFcntBEC oiygcn. Grrmi iron diHo:

18. a 0.018 ao.o 0.017

31.8 0.021 37.1 0.019

38.0 0.031 533 0.041

64.5 O.OS7 68,7 0055

ja.i 0.064 97a o.oSj
97.0 0.086

' Concerning the passivity imparted to iron by chromates, see A. S. Cusl

lulletin No. 30, OflScc of Public Roads, U. S. Department of Agrictilture.

NEW BOOKS.

Ln i^ementory Study of Chemistry. By William McPherson and Woliai
WAKSS Hehdbrson. Revised edition, viii -1- 434 pp. Ginn & Con
Boston. (No date on title page.) Price, $1.35.

This is an important book for it is manifestly destined to be w
ised in high schools and small colleges and thus to have much infli
ipon education in chemistry. In the preface the authors say '
lave made a consistent effort to make the text clear in outline, si
n style and language, conservatively modem in point of view and
)Ughly teachable." It is a pleasure to be able to offer congratula
Ipon the good measure in which these aims have been attained.

The elementary facts of descriptive chemistry, chosen with ai
ible judgment, are presented clearly and interestingly. Even in
after part, treating of the metals and their compounds, materia!
o condense itself to a tedious catalogue of substances and prope
nterest is well kept up by judicious interpolation of applications
hort digressions. The simplicity and directness of the language
ts hold on the attention remind one of Professor Remsen's texts

The treatment of the theories is less praiseworthy. The atomic tl
s stated on page 62, immediately following the law of multiple pr
ions. The reviewer believes this theory should be reached cautii
ind laboriously through not only the laws of definite and multiple
lortions by weights, but also Gay Lussac's law of combining vol
,nd Avogadro's molecular theory. These latter subjects are not
idered until pages 194 and 226 respectively.

The authors do not insist enough upon the uncertainty inhera
11 theories. On the contrary they repeatedly make the serious
ake of using theory as solid, rock-bottom fact upon which to t

NEW BOOKS. 475

As one illustration of this we may cite the definition on page ii6: "The
valence of an element is that property which determines the number
of the atoms of another element which its atom can hold in combina-
tion." If there is one place where theory has absolutely no business,
where if present it does maximum harm, that place is in a definition.
The dissociation theory also is brought in early (p. 99), on an inadequate
experimental foundation, and is utilized as fundamental fact. Thus
on page 107 we have the definition: "An acid is a substance which pro-
duces hydrogen ions when dissolved in water or other dissociating liq-
uids."

Much time and thought must have been given to composing the numer-
ous compact definitions, but even if they were not, many of them, open
to the above criticism, it is doubtful if they are wholly desirable. Such
brief statements are convenient from a teacher's point of view, furnish-
ing knowledge in tablet form, which he has but to prescribe, two or
three tablets daily five times a week, and go off about other business.
Thus drugged the student makes a fine show when put through his paces
by his trainer, but let an unfamiliar voice ask the questions in differ-
ent form or, not satisfied with an accurate recital of the well committed
phrase, press for further information, and great is the resultant distress,
confusion or irritation as the case may be.

A book is not made modem by including the dissociation theory, the
conception of chemical equilibrium, etc., alone. The real test is in the
way theories are utilized. If they are taught didactically (by far the
easiest way to teach them), the student is almost sure to attribute to them
the finality of mathematical demonstrations, and the book is "conserv-
ative," i. e.y old-fashioned. If they are brought out as the best sug-
gestions we have yet been able to make regarding branches of knowledge
still in a state of flux, debatable and in no sense certain, the book is mod-
em. The good old-fashioned book sends to the universities students
much surer of many fundamental propositions than their teachers ven-
ture to feel, but with atrophied thinking faculties. The more difficultly
teachable, modem book develops alert and critical but humble individ-
uals with an almost painful appreciation of the limitations of human
knowledge and a realization that the most far-reaching, most difficult,
least understood problems of science lie at the beginning, in elementary
chemistry. Judged by the above criteria the book under review is more
"conservative" than "modem." It must be acknowledged that only
a minority of teachers believe that it is better to give even immature
high school students the undisguised truth, interlarded though it must
be with many a "we don't know," and '.'we can only surmise;" and pos-
sibly this minority is mistaken. But it is also barely possible that the

47^ NEW BOOKS.

high value set on "teachableness" is due, sometimes, as much to consid-
eration for self as for students.

But the reviewer does not wish to be misunderstood, the good points d
the book overbalance those which do not happen to coincide with his own,
perhaps peculiar, notions, and the proof of his appreciation is that he
intends to recommend it as one of the two or three best elementary texts
known to him. S. Lawrence Bigelow.

Die Kathodenstrahlen. By G. C. Schmidt. Prof. Phys. Univ. Konigsberg. S0>
ond editon. Braunschweig: F. Vieweg und Sohn. 1907. 127 pp. Price,
Mark, 3.60, bound.

This monograph, which forms No. 2 of the collection. Die Wissen-
schaft, gives a clear and concise account of our knowledge of the elec-
tric discharge in evacuated vessels. The book is intended for the non-
spedalist; the use of mathematics is almost wholly avoided; yet, by means
of well chosen illustrations and ingenious analogies, the reader is easily
led to an accurate understanding of this most fascinating subject. The
topics treated include the nature of light and the luminous ether; the
cathode rays, their production and behavior, together with an excel-
lent critical discussion of the various hypotheses regarding their nature;
the nature of the electron or corpuscle, its velocity, charge and mass;
the Zeeman effect; the canal rays, etc. This little book is a welcome
addition to the semi-popular literature of the corpuscle, the primitive
unit of which all matter seems to be built up. Herbert N. McCoy.

The MicroBcopy of Technical Prodocts. From the German of Dr. T. F. Hanausek.
Translated by Andrew L. Win ton, Ph.D., with the collaboration of Kate G. Barber,
Ph.D. New York: John Wiley & Sons. 1907. 8vo, xii + 471 pages, 276 illustra-
tions. Price, $5.00.

This book xwhich has enjoyed a well established reputation in the
original is now presented to English readers in the work of Dr. Winton.
While not dealing with chemical methods of identification, except in-
cidentally, it nevertheless must possess no little interest for those chem-
ists who are engaged in various lines of expert testing work in which
recourse to the microscope is often absolutely necessary. The portion
of the book which will be found the most useful to analytical chemists
are the chapters on the starches, stems and roots and fruits and seeds.
These are clearly written and illustrated.

The rapid extension of the national and state food and drug laws makes
the kind of knowledge contained in this book especially valuable at the
present time. The translator is at the head of the Government Food
and Drug Laboratory in Chicago and has had a long experience in the
line of work discussed in the book. From this practical experience
he has been able to make more than a translation of it, as the numerous
notes attest. The illustrations, essential in a work of this character,

NEW BOOKS. 477

are especially good. On the whole, the book can be heartily recom-
mended. . J. H. Long.

Introduction to the Theory and Practice of Qualitative Analysis by Solution. By F. W
Martin, Ph.D., Lynchburg, Va. J. P. Bell Co. 1907. pp. 64. Price, $0.75.

In this small guide to qualitative analysis the first three chapters
are devoted to general matters related to the theory of solutions, such
as osmosis, vapor pressure, ionization, chemical cquiHbrium, hydrol-
ysis, etc. The fourth and fifth chapters deal with the classification
of the bases and acids into groups. A few reactions of the members
of each group are given. Chapters 6 and 7 give the systematic proce-
dure for the identification of acidic and basic ions. Chapter 8 contains
a list of 27 exercises to be carried out according to the directions given
in the preceding chapters.

The appendix contains a list of reagents with brief directions for
making them. It is not intended that the treatise should be used with-
out the personal instruction of the teacher. As the author says in the
preface, "He (the instructor) is the one indispensable feature of a labora-
tory." Edward H. Keiser.

Electro-AnalyBis. By Edgar P. Smith. 4th edition revised and enlarged, with 42
illustrations. Philadelphia: Blakiston's Son & Co. 1907. Price, $2.50 net.

Since the last edition of Professor Smith's book was published in 1902,
the rotating electrode has been introduced into electro-analysis, with
the result that in many cases the time required to complete an electroly-
sis has been reduced from hours to minutes. The present work is the
first to give full details of the conditions under which satisfactory results
may be obtained with tliis new tool, and is thus the only really ** up-to-
date" treatise on electro-analvsis in existence.

Many of the new methods have been worked out in the laboratory
of the author, and their accuracy and speed are attested by numerous
trial analyses; particularly interesting are those in which both anion and
cation are determined in a single operation with nierairy cathode and
silver anode.

In many instances, no doubt, the slower methods will still remain
in use — to leave the current on over night is a very easy method of mak-
ing an analysis — ^and the author has done wisely in retaining the de-
scriptions of the older methods side by side with those of the new. One
hundred and thirty pages have thus been added to the book; if in a fu-
ture edition it should prove necessary to economize space, it might be
a good idea to cut out the pictures on pages 64 and 97 in which a milli-
ammeter worth thirty dollars or so is represented as tilted against an
accumulator cell filled with sulphuric add. There are laboratories
where a student could get himself into trouble by attempting to carry
this suggestion into practice. W. Lash Miller.

478 NBW BOOKS.

Chnrch'i Laboratory Goido. By Edward Kimcb. D. Van Nostrand Co., N. \
ed„ (906. pp, ■349. Price, $2 50.
This book is lirgely devoted to the quantitative analysis of ag
tural prodiiets. I'he first edition was published in 1864 and foim
tensive use in Great Britain, India. Italy, Japan, etc. The prest-n
tion has been carefully revised and largely rewritten and brougl
to date. Part I consists of a general introduction in chemical im
lation. Part II is devoted to qualitative analysis, especially in re
to those elements of agricultural interest. Part III, comprising
2UO pages, deals with the quantitative analysis of agricultural
rials, such as soils, manures, farm crops, cattle foods, dairy pro
bread, etc. The book is clearly written, well arranged and, in
respects, well up to date. We notice that, iu commoii withmos
ropean works, the Babcock method of determining fat in milk a;
products {almost exclusively used in America) is entirely ignon
the writer. As a rule, European writers on chemical methods of
cultural analysis do not prepare works that are entirely satisfi
for the use of American students in our agricultural colleges. It
difficult to see why this should be so. L. L. Van Sli

Dairr Laboratory Guide. By Chari-ES W. Melick. D. Van Nostrand Co.
1907. pp. 129. Price, $1.35.

This book is intended for use in short dairy courses covering t
few months of work. It treats from the standpoint of laboratory
tice a large number of subjects comiected with milk and its mai
tions in connection with the processes of butter and cheese makir
preparation of other dairy products. The treatment given to eacli
is brief and necessarily superficial but is undoubtedly intended to b
plemented by class work with lectures. We question the wisdi
using chemical formulas in a book of so elementary a characte)
students who have studied chemistry Uttle. if at all. The state
are for the most part clear and accurate. The book has no ini
serious fault in any book, however small. We notice that the i
adheres to the old theory of the cause of tuottles in butter, attri
it excluavely to the action of salt and ignoring the essential part ]
by the presence of buttermilk. On the whole, the book, if pr
used, can be recommended for the purposes intended.

L. L. Van Sl-

Annuaire pour I'An 1908. Public par 1e Bureau des Longitudes. 16 mo,, g$c

Paris: Gauthier-Villars. Price, i franc 50 centimes (by mail, frs. i

Nearly one-half of the volume consists of astronomical and geoma
data, one-quarter is physical tables, nearly one quarter chemical
and an appendix contains special articles on the distances of the

RECENT PUBLICATIONS. 479

stars (G. Bigourdan), solar researches (Deslandres), the observatory of
Montsouris (E. Guyon), and obituary notices of M. Loewy and Chas.
Tr6pied. It is impossible to obtain anywhere else so much physical
and chemical data for such a small price, and the data given are in the
main reliable. The criticism made on previous volumes still holds true
on this, viz,, that data obtained by famous French scientists continue
to be given even where some of them have been proved in error and
replaced by better observations by scientists outside of France: e. g.,
Regnault's determination of the vapor tension of mercury below loo®.

J. W. Richards.

RECENT PUBLICATIONS.

Arrhenius, S. Immunochemib. Anwendungen der physikalischen Chemie
auf die Lehre von den physiologischen Antikorpem. Leipzig: 1907. 203 ss. M. 7.

AuTBNRiSTH, W. Quantitative chemischb Analyse, Massanalyse, Gewichts-
analyse, und Untersuchungen aus dem Gebiete der angewandten Chemie, zum
Gebrauche im chem. Laboratorien. Zweite vdllig umberarb. Aufl. Tubingen: 1907.
gr. 8. 380 ss. M. 9.40.

Bbard, J. T. Mine Gases and Explosions. Text-book for schools and col-
lies and for general reference. New York: John Wiley & Sons. 1908. 380 pp.
68 fig. 12 mo. $3.

Lincoln, Azariah T., and Walton, James H. Exercises in Elementary
Quantitative Chemical Analysis for Students op Agriculture. New York:
The Macmillan Co. 1907. 518 pp. i2mo. $1.25.

Low, Albert H. Technical Methods op Ore Analysis. 3rd Ed. revised
and enlarged. New York: John Wiley & Sons. 1908. 12 + 344 pp. 8vo. $3.

Maire, F. -Modern Pigments and their Vehicles: their properties and uses,
considered mainly from the practical side, and how to make tints from them. New
York: John Wiley & Sons. 1907. 266 pp. i2mo. $2.

MCllbr, a. Allgemeine Chemie der Kolloide. Leipzig: 1907. M. 10.

Namias, R. Theoretisch-praktisches Handbuch der photographischen
Chemib. I. Band. Photographische Prozesse mid orthochromatische Photographic.
Nach der 3. italienischen Auflage tibersetzt von A. Valerio und C. Stiirenberg. Halle:
1907. 406 ss. M. 8.

Nbssler, J. Bereitung, Pflegb und Untersuchung DBS Weines. 8 Auflage,
neubearbeitet von K. Windisch. Stuttgart: 1907. gr. 8. 508 ss. mit 134 Figuren.
M. II.

NBI7BURGBR, A. Handbuch der praktischen Elektrometallurgie. Munchen
und Berlin: 1907. M. 14.

Nissbnson, H. UND Pohl, W. Laboratoriumsbuch Pt>R DEN MetallhOtten-
chemikBR. Halle: 1907. gr. 8. M. 3.

OsTWALD, A. LEhrbuch DER chemischen Pathologie. Leipzig: 1907. gr.
8. 614 ss. M. 14.

Passon, H. Die Hochopenschlacke in der Zementindustrie. Wiirzburg:
1907. M. 7.

480 RBCBNT PUBLICATIONS.

Picket, B. Dib Entwickslunc dbr Thboribn und dsk Vbrpahsensi
BEi DEN Herstblldng der plOssigen Lupt. Weunar: 1907. M. 3, 30.

RiGBi, A. Die Bewecung dek Ionen bei dbr slektkischbn Entu
Deutsch von Max Ikle, Leipzig: 1907.' M. 1.

Schmidt, J. Die Alkaloidcbbmie in dbn Jahbbn 1904-1907. Sta
1907. gr. 8. 146 ss. M. 7.

Treadweli., F. p., und Mever, V. Tabbllbn zur qualitativbn Ah
6, vennehrte uad verbesserte Auflage; neubeatbeitet von F. P. Treadwell.

Vakgss, J, Nahruncsuittblchbmib. lUustrienes Lexikon der Niilini
Genussmittel sowie Gebrauchsgegenstande. Leipzig: 1907. gr. S, 30; !
3 farbigen Tafeln u. 118 Figuren. M. 10.

Waldeck, Kari,. Streifzuge durcb die Blbi — und SilbbrhOtten dbs
ir^RZES. Halle: 1907, M, 3, 40.

Wbdekind, E. Zur Stbreochbmie dbs fOnfwbrticbn Stickstoi',
g&nzl. umgeaib. u. fortgefGhite Auflage unter Mitwiikung von E. FraUicb.
lig: 1907. M. 4, 10.

Vou XXX. Aprh., 1908. No. 4.

THE JOURNAL

OF THE

American Chemical Society

(CONTBIBUnONS PROM THB RBSSARCH LABORATORY OF PirVBICAL CHEMISTRY OP THB

Massachusetts Institute of Technology No 23).
A SYSTEM OF QUAUTATIVE AITALTSIS FOR THE COMMON

ELEMEITTS.'

PAST m: ANALYSIS OF THB ALUMIlffUM AND IRON GROUPS, INCLUDING
BERYLLIUM, URANIUM, VANADIUM, TITANIUM, ZIRCONIUM

AND THALLIUM.
Bt Arthxtr a. Notbs, William C. Brat, and Bllwood B. SPBAm.

Received January a8, 1908.

Introduction.

This article is a continuation of a preceding one which appeared in an
earlier number of this Journal in which were presented the first two
parts of this system of qualitative analysis, dealing respectively with
the preparation of the solution, and the analy^s of the silver, copper,
and tin groups.^ For the 'purposes of this investigation, for the gen-
eral considerations underlying it, and for various conventional matters
relating to its presentation, the reader is referred to the introduction to
the preceding paper.

Although the final form of the scheme of analysis of the groups here
considered has been worked out during the past year by the authors
of this article, much of the preliminary experimental work, especially
that relating to the rarer metals, was carried on by others in
this laboratory. It is unfortunately, scarcely practicable to indicate
in just what respects each of these investigators has contributed to the
final result; but we wish to express in a general way our great indebted-
ness to Messrs. Howard I. Wood, Bart B. Schlesinger and Charles Field,
3rd, for the assistance which their work has been to us.

* Copyright, 1908, by Arthur A. Noyes.
■ This Journal, ag, 137 (1907).

I A. A. NOVBS, W. C. BRAY AND E. fl. SPEAK.

The present publication deals with the analysis of the predpi
3duced by ammonium hydroxide and sulphide in the filtrate from
drogen sulphide precipitate. In addition to the seven com
meats (nickel, cobalt, iron, manganese, zinc, chromium and a
im) considered in almost all schemes of quaUtative analyas
ve included six of the especially important rarer elements, nav
ryllium, uranium, vanadium, thallium, titanium and drcon
le portions of the procedure and of the notes referring to these
ments are, however, marked with asterisks, so that they ma
idily omitted by any one interested only in the common elemen
The general features of our scheme for the analyas of this soli

II be most readily comprehended by an examination of the tal
tline presented in Tables VII to IX. The considerations which

1 to the adoption of this procedure will be discussed in the next cfat
titled "General Discussion," and the detailed process and tht
inations of Jt will be presented in the following one entitled "
lure and Notes." Later chapters, as in the preceding publica
U be devoted to the "Test Analyses and to Confirmatory Ex
mts and References."

General Discussion.
(i) With respect to the original precipitation of these elem
lemes of quaUtative analysis differ as to whether ammonium
:>xide and ammonium sulphide be used successively with a filtr
tween, or whether they be added together so that all the element
atained in a single precipitate. The former of these method:
e serious disadvantage that the separation with ammonium
Qxide of the trivalent elements, aluminum, chromium and 1
in, from the bivalent elements, nickel, cobalt, manganese and
lile satisfactory enough when certain combinations of these eleu
: present is not so for other combinations. Thus even's large qua
zinc may be quantitatively precipitated by ammonium hydn
len a larger proportion of chromium is present; and manganese
any case be partially precipitated by that reagent owing to its o:
m by the air to the manganic state, and it will be completely
)itated when phosphate is present in the solution. It is ther
cessary to provide for the detection of zinc and manganese both ii
droxide and in the sulphide precipitate; and thus the scheme is
cated rather than simplified by precipitating separately with '
igents.

We have therefore adopted the plan of a single precipitatioi
Qultaneous addition of both reagents, provision being made,
er, for observing the effect of the addition of ammonium hydn
>ne for the sake of the indications which it may funush. If ca:

SYSTEM OF QUALITATIVE ANALYSIS. 483

taken to avoid an unnecessary excess of both the hydroxide and sul-
phide, all the elements in question are completely precipitated by these
reagents even in moderately dilute solution, with the exception of a
little of the nickel and a variable proportion of the vanadium. The
nickel can be removed from the filtrate by boiling. Even a large quantity
of vanadium remains completely in solution when it is present alone;
but on the other hand it may be almost completely precipitated, prob-
ably as a hypovanadate or vanadate, when certain other elements of
these groups are present. Many experiments were made in this labora-
tory by Mr. Charles Field, 3rd, to devise a practical method of reducing
vanadium to a vanadous salt (corresponding to the oxide ViOg), in which
state it is completely precipitated by ammonium hydroxide. In ac-
cordance with the results of Gooch and Curtis^ hydriodic add (best in
the form of a mixture of ammonium iodide and hydrochloric acid) was
found to be the only available agent, but even with the aid of this
reagent reduction and precipitation were never quite complete and
sometimes did not take place at alL For this reason it was not con-
sidered worth while to use this reagent, especially since the vanadium
can be removed from the ammonium sulphide filtrate by acidifying,
adding ferric chloride and making alkaline with ammonia.

(2) The ammonium sulphide precipitate in all the schemes of analysis
known to us is first treated with cold dilute hydrochloric add, in order
to separate nickel and cobalt from the other elements. In spite of the
general use of this process, we have become convinced that it does not
fulfil the requirements of exact qualitative analysis; for not only is it
true, as is generally known, that a considerable quantity of nickel and
cobalt dissolves in a mixture of i volume of HCl (1.12) with 5 volumes
of water when there is a large residue containing these elements, but
our experiments have shown that a moderate quantity of either of them
(up to at least 5 mg.) may completely dissolve and thus escape detection
when it was originally disseminated through a large predpitate of iron
sulphide. We have therefore eliminated this treatment as a method of
separation; and, after adding hydrochloric add at first to decompose
such of the sulphides as it will act upon and to get an indication as to
the presence of much nickel or cobalt, nitric add is added, so as to bring
all the elements into a single solution. Inddentally it may be mentioned
that our experiments support the view that the fact that nickel and
cobalt sulphides, though not precipitated by hydrogen sulphide from a
slightly add solution, yet dissolve difficultly in a much stronger add,
is due to an abnormally slow rate of solution of these sulphides, which
are in fact relatively soluble substances, at least in the freshly pre-
dpitated state. For these experiments have shown that nearly all of a
» Am. J. Sci. (4), 17, 45 (1904)-

484 A. A. KOYBSi W. C. BRAY AND B. B. SPEAR.

portion of precipitated nickel sulphidej^dissolves when treated with
successive portions of cold dilute HCl (i volume HCl (1.12) with 5
volumes of water) even when the add is kept saturated with hydrogen
sulphide, and that solution continues to take place even after con-
sidemble nickel (30 mg. in 30 cc) has passed into solution. When,
therefore, the surface exposed to the add is greatly incxeased, either,
by the residue being a large one or by a small residue being left in a
finely divided state by the dissolving out of iron sulphide, a considerable
quantity of nickel and cobalt passes into solution in a comparatively
short time. It is interesting to note that the reverse reaction, the
predpitation of nickel sulphide by hydrogen sulphide in add sohitioD,
also takes place very slowly, for Baubigny^ has observed that in the
presence of acetic add, or of very small amounts of sulphuric or hy-
drochloric adds, the predpitation is a slow but continuous process.

(3) Having now all the elements together in solution, the next step
in our process is to divide them into two main groups by the addition
of sodium hydroxide and peroxide, followed by subsequent boiling.
This method has been previously applied by other authors to the separa-
tion of certain of the common elements, but not, we believe, as a general
means of subdivision. These reagents cause the complete predpitation
of iron, nickel, cobalt and thallium, as hydroxides of the trivalent form,
and of manganese, titanium and zirconium as hydroxides of the quad-
rivalent form. We shall designate all these elements so predpitated
as the '4ron group." All the remaining elements, namely, aluminum,
beryllium, zinc, chromium, uranium and vanadium, remain in solution
in the form of sodium salts of the corresponding adds, namely, as alumi-
nate, zincate, chromate, peruranate and vanadate. We shall designate
all these elements so dissolved as the ''aluminum group." The separa-
tion of the two groups by this process is entirely satisfactory, at any
rate, from the standpoint of qualitative analysis, with the single ex-
ception that when only 5 or 10 mg. of zinc are present, this may be
carried down completely when elements of the iron group (especially
manganese) are present in large quantity. This makes it necessary to
provide for the detection of zinc in the analysis of the predpitate when
it is large, but this is not attended with special difficulty. The use of
sodium peroxide has the distinct advantage over that of sodium hy-
droxide alone, that chromium, uranium and vanadium are taken com-
pletely into solution whereby not only a division of these elements be-
tween the predpitate and filtmte is avoided, but also the carrying down
of zinc into the predpitate is made less common and less considerable.

(4) Since, owing to the possible presence of phosphate, oxalate, or

» Baubigny, ComfH, rend,, 94, 963, 1183, 1251, 1417, 1473, and 1715 (1882), and
95, 35 (1882).

SYSTEM OP QUAUTATIVE ANALYSIS. 485

hypovanadate in the original solution, the alkaline-earth elements
may be precipitated by ammonium hydroxide and sulphide, sodium
carbonate is added with the hydroxide and peroxide, in order to ensure
the complete precipitation of these elements (more especially barium)
with the iron group. The presence of phosphate and carbonate does
not affect at all the separation of the elements of the aluminum and
iron groups from each other.

(5) The separation of the elements of the aluminum group from each
other is very simple when only the common elements, chromium, alumi-
num and zinc, are to be provided for, and the process recommended below
for this case offers no original features. It consists in precipitating the
aluminum hydroxide from the solution with ammonium hydroxide
after acidifying with nitric add, the chromate in the filtrate with barium
chloride after acidifying with acetic add, and the zinc with hydrogen
sulphide in the filtrate from the barium chromate.

(6) The presence of beryllium does not involve any complication,
since it goes with the aluminum in the process just referred to, and can
be separated from it as described bdow. When, however, uranium and
vanadium are to be provided for, this process is entirely inadequate,
for upon the addition of ammonium hydroxide after addif3dng, vana-
dium divides between the filtrate and predpitate wherever uranium is
present, owing to the insolubility of uranyl vanadate; uranium itself
will divide owing to the presence of H^Oj formed on addifying the
sodium peruranate solution; and vanadium, when present in large
amount, again divides upon the addition of barium chloride to the
acetic add solution, owing to the slight solubility of barium vanadate ;
finally vanadium interferes with the test for zinc with HjS in acetic
add solution, since a predpitate of sulphur is always formed and some-
times one of black vanadium sulphide. Moreover, the uranium pre-
cipitate obtained with ammonium hydroxide will in general be mixed
with aluminum, beryllium and vanadium, so that the difficult part of
the separation still remains to be accomplished. After much experi-
menting, guided by the conception that under proper conditions of
alkalinity it might be possible to separate the more basic elements,
zinc, aluminum and beryllium, from those present as constituents of
add radicals, chromium, uranium and vanadium, it was finally found
that this could be accomplished in a hot solution of sodium hydrogen
carbonate, provided care be taken to prevent loss of carbon dioxide by
heating the solution in an open vessel only to 90°, or better in a closed
bottle to 100®. Under these conditions the separation is a fairly satis-
factory one. A small amount of uranium may, however, be carried
down almost completely when a large amount of aluminum or beryl-
lium is present, making it necessary to test for uranium in the pre-

486 A. A. NOYES, W. C. BRAY AND E. B. SPEAR.

cipitate. Moreover, when uranium and vanadium are simultaneously
present, each in large quantity (about loo mg.), some uranyl vanadate
precipitates, but a large quantity of both elements remains in the solu-
tion, so that their detection is not interfered with.

(7) The separation of the zinc from the aluminum and beryllium
in the precipitate is readily effected by dissolving it in hydrochloric
acid and adding a small excess of ammonium hydroxide. For the
separation of the aluminum and beryllium from each other we studied
what seemed to be the two most promising methods mentioned in the
literature. The first of these was that described by Parsons and Barnes,^
which consists in boiUng for a short time a solution of the two elements
to which enough sodium hydrogen carbonate is added to make a 10
per cent, solution, whereby aluminum is precipitated and beryllium
dissolved. We found, however, that though this method is satisfac-
tory for the detection of beryllium when a moderate amount of aluminum
is present, yet with a large quantity of aluminum (say 100-500 mg.)
2-5 mg. of beryllium are almost completely retained in the precipitate,
which may cause it to escape detection, and which at any rate leads
to an incorrect estimate of its quantity. The second method investi-
gated was that proposed by Havens* which consists in saturating a solu-
tion of the chlorides of the two elements in a mixture of ether and strong
hydrochloric acid with hydrogen chloride gas, whereby aluminum is
precipitated and beryllium remains in solution. It was found that this
method gives entirely satisfactory results; even 0.5 mg. of aluminum
is precipitated, as AICI5.6H2O, provided care be taken to use a
sufficient proportion of ether and to saturate completely with the gas;
even 100 mg. of beryllium remain wholly in solution, and 0.5 mg. is not
carried out with a large quantity of aluminum. This method was, there-
fore, adopted for the separation, it being supplemented by a confirmatory
test for beryllium based on the process of Parsons and Barnes.

(8) The separation of the chromium, vanadium and uranium, which
are present together in the filtrate from the sodium hydrogen carbonate
precipitate, also required much investigation. It was soon decided
that there was more promise of effecting a separation of the first two
of these elements in the state of chromate and vanadate, in which they
already exist, than in a lower stage of oxidation ; and it was found that
the lead salts differed sufficiently in solubility in nitric acid to enable
0.5 mg. of chromium to precipitate while retaining 100 mg. of vanadium
in solution. A separation based on this fact was therefore adopted.
The excess of lead added is subsequently removed by saturating the
filtrate with hydrogen sulphide. To avoid the addition and removal

* This Journal, 28, 1589 (1906).
' Z. anorg. Chem., 16, 15 (1898).

SYSTEM OF QUALITATIVE ANALYSIS. 487

of lead when chromium is absent, a preliminary test for chromate with
hydrogen peroxide is introduced. The uranium is separated from the
vanadium (after oxidation to the vanadic state) by precipitating it as
uranyl ammonium phosphate in acetic acid solution — a method that was
found to give satisfactory results for the limiting case of a large pro-
portion of uranium and a small proportion of vanadium, and also in the
converse case. Vanadium is tested for by making the filtrate strongly
alkaline with ammonia and saturating with hydrogen sulphide, where-
by a violet-red solution of a vanadium sulpho-vanadate is formed.

(9) We will next consider the analysis of the precipitate produced
by sodium peroxide, which contains the manganese, iron, nickel, cobalt,
thallium, titanium and zirconium, the alkaline-earth elements, and pos-
sibly phosphate. The main problems connected with this were the sepa-
ration of the manganese from the other elements, that of the alkaline-
earth elements from phosphate and that of the titanium and zirconium
from the iron and from each other.

(10) In almost all schemes of qualitative analysis it is thought
sufficient to test a portion of the precipitate for manganese by fusing
it with sodium carbonate or by boiling it with lead dioxide and nitric
add, without isolating the manganese. These color tests give, how-
ever, but little idea of the quantity of the element present. Moreover,
aside from this objection, the large number of the elements contained
in the precipitate in this scheme makes their separation necessary.
The one reaction of manganese which seemed in every way suited for
this purpose is that frequently employed in iron and steel analysis con-
sisting in the conversion of the manganese into the dioxide by the action
of chloric acid and concentrated nitric acid. For this is not only a
behavior highly characteristic of this element ; but, since the separation
is carried out in a strongly acid solution, it might be anticipated that the
other elements, which are not oxidizable to insoluble peroxides, would
not be retained in the precipitate to an important extent. Our ex-
periments have shown that this is, in general, the case; but one ex-
ception has been discovered. It has been found, namely, that titanium,
which is quadrivalent like manganese in the dioxide, when present
even in considerable quantity (up to 50 mg.) may be completely pre-
cipitated with a large quantity of manganese (500 mg.), and that a large
proportion, though not all of the zirconium is likewise carried down.
The method is, however, otherwise so satisfactory that we have adopted
it, special provision being made for this unusual case in a way that
need not be described here. The procedure consists in dissolving the
whole sodium peroxide precipitate in strong hydrochloric acid, in evapora-
ting with excess of nitric add, adding concentrated nitric acid and

488 A. A. NOYBS, W. C. BRAY AND B. B. SPEAR.

potassium chlorate, heating, and filtering ofif the manganese dioxide
on an asbestos filter.

(ii) Owing to the fact mentioned above that zinc is carried down
in considerable quantity by manganese in the Na,0, precipitation,
experiments were made mth the view of previously removing man-
ganese by introducing this chloric add procedure at the beginning of
the analysis of this group. It was found, however, that vanadium,
which is not carried down in the sodium peroxide procedure, is, like
titanium, precipitated in large quantity with the manganese in the
chloric acid procedure. It was found also that some zinc (1-4 mg.)
may be carried down completely in the Na^O, precipitate by iron,
nickel and cobalt, so that it would be still necessary to provide for the
detection of zinc in the iron group. For these reasons, it is evidently
best to have the sodium peroxide precipitation precede the treatment
with chloric add.

(12} The filtrate from the chloric add predpitate is first tested for
phosphate. When it is not present, the iron, thallium, titanium and
zirconium are separated from the other elements by the addition of
ammonium hydroxide. When phosphate is present, in order to separate
it from the alkaline-earth elements, the basic acetic predpitation is
employed, ferric chloride being first added, if necessary. The pro-
vision here made for the case that phosphate is present is thought to
have many advantages over the methods ordinarily employed in schemes
of qualitative analysis, where the phosphate is removed by tin in nitric
add solution, by ferric chloride and barium carbonate, or by ferric
chloride and ammonium acetate in the first stages of the analysis of the
group. Of these three processes, the basic acetate is much more rapid
and simple of execution ; but it does not give a separation which is at all
satisfactory when applied to a solution containing all the elements of
the aluminum and iron groups; thus chromium and zinc may in certain
combinations of elements be found either in the predpitate or filtrate,
and manganese also divides, unless great care is taken to make the
precipitation in a large volume at the proper add concentration, hi
the scheme of analysis here presented, this basic acetate separation has
been introduced only after these troublesome elements have already
been removed , for the presence of phosphate involves no complications
in the preceding steps of the process. Under these drcumstances it is
no more difficult to secure accurate results in this separation than in the
predpitation with ammonium hydroxide. Indeed, by adopting this
process for all cases the compUcation of the alternative procedure and
the special test for phosphate might be removed; but since the opera-
tions require a somewhat longer time, it has seemed best to retain the

SYSTBM O? QUALITATIVE ANALYSIS. 489

ammonium hydroxide precipitation for the case that phosphate is not
present.

(13) The iron might be separated from the titanium and zirconium
by the well-known method of boiling a solution of the elements kept
slightly add with sulphurous acid; but it is difficult to secure com-
plete precipitation of the titanium and zirconium and at the same
time to prevent the canying down of iron. Besides this, the operation
is a long one, involves large dilution, and makes it necessary to use
hydrofluoric acid in redissolving the precipitated hydroxides. On the
other hand, the removal of the iron by the method of Rothe,* which
consists in shaking it out of a strong hydrochloric acid solution by
means of ether, is extremely simple and rapid, gives a perfect separation,
and leaves the titanium and zirconium in solution. We have therefore
unhesitatingly employed this method in our scheme of analysis. More-
over, in this process the thallium, which is present as thallic chloride,
is extracted together with the ferric chloride by the ether; it can be
readily detected, after evaporating oflF the ether and reducing the ferric
and thallic salts with sulphurous acid, by the formation with potassium
iodide of the characteristic yellow precipitate of thallous iodide.

(14) For the separation of the titanium and zirconium from each
other we have adopted the process of Hillebrand,' which consists in
adding sodium phosphate to a slightly acid solution containing hy-
drogen peroxide, whereby the zirconium is precipitated and the titanium
remains in solution. The latter is shown to be present by the color of
the solution, but it can also be precipitated as phosphate by destroying
the hydrogen peroxide with sulphurous acid after filtering oflf the zir-
conium compound.

(15) The analysis of the ammoniacal solution containing nickel,
cobalt and perhaps zinc and the alkaline-earth elements, is carried out
abng the conventional lines. The first three elements are precipitated
by the addition of ammonium sulphide, and the zinc is extracted, if
the precipitate is large, by treating it with cold dilute hydrochloric
add, or if it is small by dissolving the predpitate completely and treat-
ing again with sodium peroxide. Instead of separating the nickel and
cobalt from each other, it was found to be shorter and more conclusive
to divide the solution of the sulphides into two parts and to test one
portion for nickel by adding potassium cyanide and sodium hypo-
bromite and the other portion for cobalt with potassium nitrite. The
conditions for securing the best results in the nickel test were fully
studied. Finally, in the filtrate from the ammonium sulphide pre-

'Rothe, Stahl imd Eisen, X2, 1052 (1892), 13, 333 (1893).
•Bull. U. S. Gcol. Survey, No. 176, p. 75 (1900).

490 A. A. NOVnS, W. C. BRAY AND E. B. SPEAR.

cipitate the alkaline-earth elements are precipitated with ammonium
carbonate as usual.

(i6) Though no provision is made in the system of analysis for the
separate detection of any of the rare-earth elements, yet a process has
been described for determining whether any of them are present and for
removing them when they are found to be in the solution. This pro-
cess consists in evaporating the add solution of the original ammonium
sulphide precipitate, adding to the residue hydrofluoric add and filter-
ing. This converts the rare-earth elements completely into insoluble
fluorides, and enables them to be separated from all the other elements
of the aluminum and iron groups (except from a ver>' large quantity of
aluminum). When alkaline-earth elements are present, these are also
precipitated as fluorides, wholly or in part, and are separated from
the rare-earth elements by decomposing the fluorides with sulphuric
add, diluting, filtering off any alkaline earth sulphates that separate,
and precipitating the rare-earths in the filtrate with ammonium hy-
droxide. This process of isolating the rare elements has been worked
out in this laboratory by W. C. Arsem and H. I. Wood, for use as a
group separation in the "System of Qualitative Analysis Including Nearly
all the Metallic Elements."*

(17) Final confirmatory tests are given for almost all the elements;
and much experimenting has been done on some of these tests in order to
make them delicate and reliable; thus, this is true of the color tests for
aluminum and zinc made by igniting the oxides with cobalt nitrate, of
that for chromium with hydrogen peroxide and ether, of that for nickel
with H3S in an alkaline tartrate solution, and of that for vanadium
made by adding HjO, to an add solution ; also of the precipitation test
for uranium with potassium ferrocyanide. No satisfactory tests have
as yet been found for beryllium or zirconium. Many, but not all, of
these confirmatory tests will be found superfluous and will be omitted
by the experienced analyst, except in cases where a very small pre-
cipitate or one of doubtful character is obtained; but they will, we
believe, be useful to those unfamiliar with this scheme, and they serve.
in the case of students, the educational purpose of making them ac-
quainted with additional reactions of the elements in question.

(18) It may be thought an objection to this scheme of analysis that
it involves a number of manipulative operations unusual in qualitative
analysis, such as heating in a closed bottle, saturating the acid-ether
solution with hydrogen chloride gas (in the separation of aluminum and
beryllium), filtering through an asbestos filter (in the separation of
manganese), and shaking out vnth ether in the separation of iron from

* Two parts of this "System" have already been published. Technology Quarteri),
16, 93-131 (1903): 17, 214-257 (1904).

SYSTEM O^ QUAlrlTATlVB ANALYSIS. 49 1

titanium and zirconium. These operations, when they have been once
executed, are found to be little if any more troublesome than the ordi-
nary operations of precipitation and filtration. They are, moreover,
mostly employed only in connection with the detection of the rarer
elements, where the difficulties in finding any satisfactory method are
so great that a little additional trouble is an insignificant factor. And
finally, from an educational standpoint, they introduce the student to
new kinds of processes, thus enlarging his knowledge and diminishing
the force of the objection that the ordinary study of qualitative analysis
is too limited in its scope.

Tabular Outline.

In the tables below, the enclosure of a symbol in brackets shows that
the element may divide itself between the residue and the solution in
the operation immediately preceding.

Procedure and Notes.

Procedure 51. — Boil the filtrate from the H,S precipitate (P. 21)*
till the H,S is expelled. Transfer it to a flask, add NH4OH (0.96) until
the mixture after shaking smells of it, and then 2-4 cc. more. Note
whether there is a precipitate. Add ammonium monosulphide slowly
(or if nickel is likely to be present pass in HjS), until, after shaking,
the vapors in the flask blacken a piece of filter-paper moistened with
lead acetate solution. To coagulate the precipitate shake the mixture
or heat it nearly to boiling. Filter, and wash the precipitate, first with
water containing about i per cent, of the (NH4)3S reagent and then
with a little pure water. If the filtration is slow, keep the funnel covered
with a watch glass so as to prevent oxidation. To the filtrate add a
few drops (NH4),S, boil the mixture for a few seconds, or longer if it is
dark colored (till it becomes colorless or light yellow) ; filter if there is a
precipitate, uniting it with the preceding one. (Precipitate, P. 52;
filtrate, P. 71, or first by *P. 51a, if vanadium is to be tested for.)

Notes.^(i) The H,S is boiled out, and the effect of the addition of NH^OH alone
is noted because it often gives a useful indication as to what elements are present.
To save time the expulsion of the H,S may be omitted when this indication is con-
adered unimportant. Ammonium monosulphide is used, rather than polysulphide,
in order to prevent as far as possible the dissolving of NiS, and in order not to intro-
duce solphur into the precipitate, or polysulphide into the filtrate; for this gives to
the filtrate a deep yellow color, and causes in the subsequent ancdysis separation of
sulphur on standing or on heating. Excess of the monosulphide is avoided, for the
same reasons, since it rapidly oxidizes in the air to polysulphide. By passing H^
into the ammoniaca] solution instead of adding (NHJ^, the dissolving of NiS is en-
tirely prevented ; therefore, though the operation takes a little longer, the use of H,S
is to be preferred when nickel is likely to be present. The mixture is shaken in order

' For this and similar references to former procedures, see the previous article.
This Jouknai., 39, 137.

. NOYKS, W. C. BRAY AND E. B. SPEAR.

S 4

f^ S

' §'

•?<

S-Sjll

£nj|

■sS

■s^

'^'

'i^.

■S.i

^■i

s>!?!

Sl^s

■So s ^

Jli

Sill

\ ill

SYSTBM OP flUAIJtATlVB ANALYSIS.

3s ;

'Mi

A. A. NOYBS, W. C. BRAY AND Q. B. SPBAR.

B" si i
I si a-

s . o ° *

' 111

1^0"

SYSTEM Olf QUALITATIVE ANALYSIS. 495

to coagulate the precipitate, and make it filter more readily. Heating also promotes
the coagulation of the precipitate, and it is recommended when the precipitate does
not coagulate and settle quickly on shaking. The filtrate is boiled for a few moments
to ensure the complete precipitation of Cr(OH)„ or longer to ensure that of NiS, whose
presence is indicated by a brown or nearly black color of the filtrate. Finally it is
directed to wash with water containing a httle (NHJ^, and to keep the filter covered,
in order to avoid the oxidation of the sulphides by the air, by which soluble sulphates
may be formed.

(2) Under the conditions of the procedure, which provides for a small excess of
NH4OH in the presence of NH4CI, aluminum, chromium, iron, titanium, zirconium,
and beryllitun are completely precipitated as hydroxides and uranium as ammonium
diuranate, (NH4)2XJx07. All of these precipitates are white, except those of chromium,
uranitmi, and iron; Cr(OH), is grayish-green, and (N 1^4)211207 is yellow. The color
of the precipitated hydroxide of iron varies with the state of oxidation of the iron,
pure ferrous salts yielding a white precipitate, and ferric salts a reddish-brown one,
while mixtures of them 3rield green or black precipitates. In the alkaline mixtures
the precipitate is rapidly oxidized by the oxygen of the air and undergoes correspond-
ing changes in color. Manganous salts are also oxidized rapidly, with the result that
brown Mn(OH), separates. Under the conditions of the procedure zinc and nickel
when present alone, are completely dissolved; the same is true of cobalt except when
it is present in large quantity; but zinc is precipitated when chromium is also pres-
ent. The ammoniacal solution of nickel is blue and that of cobalt of a reddish color,
which darkens rapidly on exposure to the air owing to oxidation. If a smaller ex-
cess of NH4OH is used than is directed, some zinc hydroxide, as well as cobalt hy-
droxide, may remain tmdissolved when large amounts of these elements are present;
but this has no effect on the subsequent analysis. If, however, a much larger excess
of NH4OH is employed, a few milligrams of aluminum and chromium may be dissolved,
the latter giving a pink colored solution.

(3) The presence of a considerable quantity of NH^O, such as is formed by the
neutralization of the add already in the solution, serves to prevent the precipitation
of Mg(0H)2 (and of MnCOH),), and also to lessen the amount of A1(0H), dissolved
by the NH^OH.

(4) The influence of an excess of the NH4OH and of the presence of NH^Cl on the
solubilities of the variotis hydroxides is explained by the mass-action law and ionic
theory as follows: In order that any hydroxide, say of the type MO^H,, may be pre-
cipitated, it is necessary that the product Cm'C'oh of the concentrations of the ions
M'^^ and OH'~ in the solution in question attain the value known as the solubility-
product. This varies, of course, with the nature of the hydroxide; but for all the ele-
ments of the iron group and also for magnesitmi it has so small a value that, even in a
solution containing in 50 cc. only one milhgram of the element, and a slight excess of
NH4OH, the product CmC'oh exceeds it, and precipitation results. When, however,
much NH4CI is also present, this greatly reduces, in virtue of the common ion effect,
the ionization of the NH4OH and therefore the OH"" concentration in the solution, so
that now for certain elements the product CmC'oh does not reach the solubility value,
even when Cm is moderately large (say 500 mg. in 50 cc.). This is true of magnesium
and manganese; but in the cases of aluminum, chromitun, and ferric iron the solubility
of the hydroxides in water is so slight that even in NH4CI solution the solubility is
not appreciable.

If these were the only effects involved, the greater the excess of NH4OH added,
the less would be the solubility of any hydroxide; but other influences come into play
with certain of the elements. These influences are of two kinds. The first of these

49^ A. A. NOYBS, W. C. BRAY AND B. B. SPBAR.

is shown by zinc, nickel, and cobalt. In the case of these elements, just as with d-
ver and copper, the excess of ammonia combines with the simple cathion M'*"^ fonn-
ing complex cathions of the types M(NH,),++ and M(NH,)4"*"^, thereby removing
the simple cathion from the solution and making it necessary for more of the hydrox-
ide to dissolve, in order to bring back the value of CmC'oh to that of the solubility-
product. In such a case, the presence of NH4CI increases the solubility sdll farther,
since it greatly decreases Cqh and slightly increases Cnh^oh cmd Cnh^ owing to the
common ion efiPect on the ionization of the NH4OH. Chromium also forms similar
ammonia complexes, but in much smaller proportion.

The second effect is exhibited in the case of A10,H,. This hydroxide is a so-called
amphoteric substance, — i, «., is one which behaves both as a base and as an add in
consequence of its being appreciably ionized both into 3(0H'~) + Al'^''"^ and into
H+ + AlOjH,"" (or A10,~ and H,0). With the H arising from the latter form d
ionization, the 0H~ coming from the excess of NH^OH combines to form H^ thns
causing more A10,H, to dissolve until the value of Caios-Ch again attains that of the
solubility-product. Since Caiot^h = const, in any solution saturated with AlOA
and since the equilibrium equations Ch-Cqii = Kw and CNH4C0H = K/^nh^oh
must be satisfied, it follows from combination of these equations that CAlOi/CoH==
const, and that CaiOi = const. X Cnh40h/Cnh4' The first equation shows that the
quantity of aluminum dissolved is proportional to the concentration of hydroxide-ion in
the solution, and that therefore it would be much greater in a solution of a largely
ionized base like NaOH than in that of a slightly ionized base like NH^OH. The
second equation shows that the solubility in solutions containing NH^OH and NH^Q
(or other ammonium salt) is proportional to the ratio of the concentration of the base
to that of the salt ; so that the presence of ammonium salts tends to neutralize the sol-
vent action of an excess of the hydroxide. All these conclusions are in accordance
with the facts. BeO^, behaves in the same way as A10,H„ forming the cation Be'^'*'
and the anion BeO,"^ or HBeO,"~.

(5) It follows from the statements in the preceding notes that if the NH4OH produces
no precipitate it proves the absence of as much as one milligram of aluminum, iroo,
beryllium, uranium, titanium, and zirconium ; also of chromium, if the mixture is heated
to boiling after the addition of NH^OH. Care must be taken not to overlook a small
precipitate which might otherwise escape detection on account of its transparency.
The mixture should therefore be well-shaken and allowed to stand 2 or 3 minitta
in order that the precipitate may collect in flocks. This treatment also oxidizes the
iron when present in small quantity, and thus enables it to be more readily detected;
for its precipitation in the ferric state is more complete.

(6) When phosphate is present, magnesium, calcium, strontium, barium, and man-
ganese may be partially, or even completely precipitated by NH4OH. It is that-
fore necessary, when phosphate is present, to provide for the detection of the alka-
line earth elements in the analysis of this precipitate. The normal phosphates and
the monohydrogen phosphates of these elements are difficultly soluble in water, hot
dissolve readily in adds, owing to the formation in solution of the much more soluble
dihydrogen phosphates or of free phosphoric add, for example, according to the equa-
tions:

(Ca++),(PO,^, + 4H+a- = 2Ca++(Cl-), + Ca++(H,PO -),
Ca++(H;P04— ), + 2H+a— = Ca++(C1— ), + H.PO,.

Upon the addition of sodium or ammonium hydroxide to such a solution the hydrogen-
ion in equilibrium with the H,P04~ and H,P04 is removed, and these dissociate into
HP04°° and P04^, thus causing predpitation of the corresponding salts. When other
elements, like iron, forming more insoluble phosphates are also present, it is evident

SYSTEM OF QUALITATIVE ANALYSIS. 497

that they will combme with the phosphate radical, thus leaving the alkaline earth
elements in solution. On the other hand, when a soluble carbonate, or a large excess
of a strong base, is also present, the alkaline earth phosphates will be partially con-
Terted into carbonates or hydroxides, leaving phosphate in solution.

(7) The presence of any other add radical which forms with the alkaline earth
dements salts soluble in dilute adds but insoluble in ammonia may also cause their
pccdpitation at this point. Such radicals are fluoride, borate, oxalate, and hypo-
vanadate. The fluoride will ordinarily have been removed in the evaporation with
adds in the preparation of the solution. The borates of the alkaline earth elements,
though somewhat difficultly soluble, are not suffidently so to cause them to be pre-
dpitated, except when present in very large quantity. Oxalate and hypovanadate,
even if present, do not make any change necessary in the usual process of analy^s;
for, in the course of it, vanadate and much of the oxalate are separated from the al-
kaline earths in P. 52, and the remainder of the oxalate is destroyed in P. 61.

(*8) Vanadium when present alone in moderate quantity may remain in solution,
hot when present in large quantity is partially predpitated by NH4OH as a dark gray
hydnxzide, VO(OH),. This compound corresponds to the state of oxidation (oxide VO,)
to which vanadic add (oxide Yfi^) is partially or completely reduced by H^. It
is an amphoteric substance which forms with adds soluble blue salts such as VOCl,,
vanadyl dichloride, and VOSO4, vanadyl sulphate, and with bases hypovanadates such
as Ka,V,0^ and Na^V^O,. When other elements of this group are also present the
vanadium may be completely predpitated with them, since their hypovanadates
and vanadates are in general difficultly soluble substances.

(*9) When a base is added to a uranyl salt (for example UOjCl,), uranyl hydroxide,
UO|(OH)„ is first formed, but this is an amphoteric substance, and it is converted
by the excess of base into salts of diuranic add, H,U,0„ of which even the alkali
salts are insoluble.

(10) (NHJ^ predpitates ZnS, MnS, NiS, CoS, and TljS, and converts Fe(OH),
into FeS, Fe(OH), into Fe^ and (NHJjUjOy very slowly into UO,S. The hydrox-
ides of aluminum, chromium, titanium, zirconium, and beryllium are not affected
by the (tlELJ^. Whether a hydroxide predpitate is converted into a sulphide pre-
dpitate or whether the reverse reaction takes place depends on the rdative solubil-
ities of the two compounds and on the rdative concentrations of sulphide-ion and hy-
droxide-ion in the solution. Since in the solution the concentration of the sulphide-
ion greatly exceeds that of the hydroxide-ion, even difficultly soluble hydroxides
would be converted into more soluble sulphides, provided that the difference in
solubility were not too great. In the case of hydroxides, like Al(OH)„ which are not
so changed, their sulphides are so much more soluble that they do not form in aqueous
sulphide solutions.

(11) The sulphides of iron, nickd, and cobalt are black; T1,S and UO^ are dark
brown; ZnS is white; and MnS is flesh colored, but turns brown on standing in the
air owing to oxidation to hydrated Mn,0,.

(12) When nidk:d is present alone, or in large proportion in the (NH4)jS predpi-
tate, several miUigrams of it usually pass into the filtrate making it dark colored,
and some NiS also passes through the filter with the wash water. In this case it is
useless to try to remove the NiS by filtering again ; but it can be coagulated by boil-
ing for sevoal minutes. This behavior of nickd, as stated above, can be avoided
altogether by passing H^ into the NH4OH solution to predpitate the sulphide, in-
stead of adding (NHJ^. The formation of this brown solution depends upon the

of ammonium polysulphide, for nickd may be completely predpitated by

498 A. A. NOYSS, W. C. BRaY AND ^. B. SP^AK.

ammonium monosulphide in the absence of air. The nature of the brown
is not known.

♦Procedure^Sia.— To the filtrate|from the (NH4)aS precipitate (P. 51)
add 5 cc NH4OH (0.90) and completely saturate the solution with
HjS. (Pink or violet-red color, presence of vanadium.)

If no red color results, boil the solution for several minutes in a casse-
role to expel most of the NH^OH and (NHJ^S, filter off the sulphur,
and treat the filtrate by P. 71.

If the solution assumed a pink or red color, acidify it with HCl (1-12),
heat it to boiling, and filter. Boil the filtrate in a casserole to expel
H3S, add 0.5-5 cc. 10 per cent. FeCl, solution, and NH4OH (0.96) until
the mixture after shaking smells of it, and filter. (Filtrate, P. 71.)
Heat the HCl precipitate (with the filter, if necessary) with 5-10 cc
HNO3 (1.20) until the black precipitate is dissolved, and filter. (If
the red coloration produced by HjS was slight, evaporate the filtrate to
about 2 cc.) Add to the solution a few drops of 3 per cent. HjOj. (Orange-
yellow or orange-red color, presence of vanadium.) Dissolve the NH4OH
precipitate by pouring a small portion of HNO5 (i'2o) repeatedly through
the filter, and test for vanadium with HjO^ in the same way, first adding
a little water to the HNOj solution if it has a red color owing to the
presence of much ferric nitrate.

Notes, — (i) An ammoniacal solution of a vanadate or hypovanadate quickly br
comes yellowish red when KJS is led into it, and this color slowly deepens as more
H^ is absorbed ; but the characteristic, brilliant violet-red color is obtained only when
the solution is completely saturated with H^S. The presence of ammonium salts tends
to prevent the formation of this red compound, but their influence is overcome by
the addition of a large excess of NH^OH. These facts make it probable that this red
compound is an ammonium sulpho vanadate, from which the sulphur is readily split
off as SH~" ion, owing to hydrolysis. Under the conditions of procedure, 0.2 mg.
V can be easily detected, the solution then having a pink color.

(2) Upon the addition of HCl the sulpho salt is immediately decomposed, witk
formation of a black precipitate of V^S^ or VaS,. This precipitation is far from com-
plete under these conditions, only about half the vanadium being thrown down. Mcce
or less sulphtu will also be precipitated , but the dark color of the sulphide is apparent
even when less than 0.5 mg. V is present. Acetic add may be used instead of HQ,
but the proportion of vanadium precipitated as sulphide is not much increased.

(3) When a three- to fivefold excess of a ferric salt is present and NH^OH is added,
the remainder of the vanadium is precipitated (probably as hypovanadate) together
with the FeCOH),.

(4) The confirmatory test for vanadium with H,0, depends upon the formatiofi
of pervanadic add, HVO4. o.i to 0.2 mg. V may be detected in a volume of 5 cc.,
provided that this solution is strongly add. A very large excess of H,0| is to be
avoided, for this decreases the intensity of the color, and may even decolorize the
solution completely if very little add is present, probably owing to the fonnation of
a colorless compound of pervanadic add with hydrogen peroxide.

(5) Molybdenum, if not completely predpitated by H^ in P. 21, and tungstCD,
if it passed into the original add solution (owing to the presence of phosphate or ar-

SYSTEM OF QUALITATIVE ANALYSIS. 499

aeoate) in P. 3, may also be present in the filtrate from the (NH4),S precipitate. Upon
saturating with H^ molybdentmi also gives a deep red color which would obscure
the test for vanadium and might be mistaken for it ; but tungsten gives no color. Upon
acidification with HQ, tungsten divides between the pedpitate and filtrate, but
molybdenum is thrown down completely as black MoS,. (If the cold solution is acidi-
fied with acetic add, the sulphomolybdate remains undecomposed for some time
ttd the MoS, is not completdy predpitated even on boiling.) In the final confirma-
tory test with H,Of tungsten gives no color; and molybdenum, even when a large
quantity is present, gives only a pure ydlow color but not an orange or red one.

Procedure s^. — Transfer the (NHJjS precipitate (P. 51) with the
filter if necessary to a casserole ; add 5-20 cc. HCl (1.12), stir for a minute
or two in the cold, and then boil the mixture for two or three miijiutes ;
if a black residue still remains, add a few drops HNO, (1.42) and boil
a^ain. Dilute with a little water, filter off the sulphur residue, and
evaporate the filtrate to a small volume to remove the excess of acid.

♦In order to detect rare earth elements, and to remove them if present,
treat this solution by *P. 52a.

*If uranium and vanadium are to be tested for later by *P. 58a-/t,
evaporate twice with a little HNO, (1.42), to destroy HCl.

Dilute the solution to 10 or 20 cc. ; make alkaline with NaOH solution,
avoiding a great excess; add 10-20 cc. more water if so large a pre-
cipitate separates that the mixture becomes almost gelatinous. Cool
by placing the casserole in cold water, and add 0.5-3 gram solid Na^O,,
in small portions with constant stirring. Then add 5 cc. 10 per cent.
NajCOj solution; boil for two or three minutes to decompose the ex-
cess of Na^O,, cool, dilute with an equal volume of v^ater, filter vsrith the
help of suction and wash with hot water. (Precipitate, P. 61; filtrate,
?• 53f or if uranium or vanadium is to be tiested for, *P. 58a.)

I Notes. — (i) All the hydroxides and all the sulphides, except NiS, CoS, usually
Ifiasolye readily in cold HG. If, therefore, there is considerable black residue after
idding the HCl, it shows the presence of nickel, or cobalt (or possibly vanadium);
i very small black residue may, however, be due to FeS enclosed within sulphur. The
fKt that there is no such dark colored residue does not, however, prove that
mdcel and cobalt are absent, for a considerable quantity of them (even 5 mg.) may
finolve completely in the HCl when large quantities of other elements, especially
iron, are also present.

(2) The fact that NiS and CoS dissolve so much less readily in dilute adds than do
the other sulphides of this group seems to be due not to a lesser solubility in water,
hat to an unusually slow rate of solution, for nickel and cobalt are not precipitated
by H|S even from a much more weakly add solution, and their sulphides obtained
hy predpitation with an alkaline sulphide continue to dissolve in dilute adds with-
ont reaching a limit determined by the concentration of the H^S and the nickd-ion
or ooboH-ion in the solution, as would be the case if the phenomena were that of the
soltthiHty of a difficultly soluble sulphide.

(3) The (NH^)^ predpitate is first treated with HCl, partly in order to furnish
the indication just referred to of the presence of nickel or cobalt but mainly because
much more free sulphide and sulphate would be formed^by oxidation if HNO, or aqiui

500 A. A. NOYBS, W. C. BRAY AND E. B. SPEAR.

regia were used at the start. The presence of the sulphate in considerable qituitity
in the solution interferes with the subsequent test for chxomate (with Ba(NO|),iD
P. is or with Pb(NO0, in *P. 586). If NiS, CoS, or VgS, is present in the readv.
HNO, must, however, be subsequently added, to ensure the solution of these nl-
phides. *The HG is destroyed by evaporation with HNO„ since chloride interfieni
with the test for chromate with Pb(NO,), in *P. 586.

(*4) If the (tiJl^)JS precipitate be allowed to stand for a long time before tretting
it with add, or if the mixture be heated for a long time after the precipitation, TiO(OH),
and ZrO(OH), may remain in part undissolved even by the boiling concentrated add%
owing to the fact that the hydroxides at first precipitated become partially dehydrited,
in which state they are very difficultly soluble. If this happens at this pomt or in
dissolving hydroxide precipitates obtained later in the analysis, the residue may be
dissolved in a little HF in a platinum dish, and the HP then expelled by evaporadng
two or three times with HNO, (or HG) nearly to dryness.

(5) By NaOH, iron, manganese, nickel, cobalt, ^titanium, ^zirconium, and Mi-
nium are completely precipitated and do not dissolve in moderate excess; wfaile alu-
minum, chromium, zinc, ^vanadium, and ^beryllium, remain in solution or disBoliK
when a sufficient excess is added, owing to the fact that their hydroxides are ampbo-
teric substances (see P. 51, Note 3), and form with the NaOH soluble alnmioatc
(NaAlO,), chromite (NaCrO,), zincate (Na,ZnO,), vanadate (Na,VOJ, and beryl-
late (Na,BeO,). Thallium in the thaUous state also remains in solution since VXM
is a readily soluble substance. When zinc and chromium are simultaneoudy pm-;
ent, they are precipitated in the form of a double compound (ZnCr^O^). Cfaionuna
would also be completely precipitated, owing to hydrolysis and the formation of t^
less soluble solid hydroxide, if the NaOH solution were boDed before adding Na/V
It will be observed that NaOH precipitates manganese, nickel and cobalt while NH,OH
does not. This occurs with manganese because of the far greater concentration d
hydroxide-ion in the NaOH solution, and with nickel and cobalt partly on this aoooiBt
and partly because there is no complex formation, as there is with NH^OH. Mn(OH]^
is white, but rapidly ttims brown, owing to oxidation to Mn(OH),; Ni(OH), is liglit
green; Co(OH), is pink, but from cobalt solutions a blue basic salt is first precipitated
in the cold. If a large excess of NaOH be added, a little Co(OH)s dissolves yiddiqi
a blue solution, doubtless forming a salt such as Na^CoO,. This is to be avoided sioti
then the cobalt will not be completely oxidized and precipitated upon the subseqnedt
addition of Na,0^

(6) By the addition of Na,0» Fe(OH), is changed to dark red Pe(0H)„ BCn(OH]L
to brown hydrated MnO„ Co(OH), to black Co(OH)„ Ni(OH), partially to black
N](OH)„ and thallium is precipitated as dark red T1(0H)„ all of which are iasMt
in excess of NaOH and remain, together with TiO(OH), and ZrO(OH)» in the preopi-
tate. Chromium and uranium, which after the addition of NaOH are present as
soluble sodium chromite or as insoluble sodium diuranate (Na^UsOr), axe converted
by Na^O, into chromate (NajCrO^) and peniranate which are soluble compounds and
remain in solution, together with the zinc, beryllium, and vanadium, which are stiB
present as zincate, beryllate, and vanadate. The separation is more satisfactory,
especially in the case of uranium when a large amount Na,0, is used

(7) Even a cold solution of Na,0, decomposes rapidly with evolution of oxyscn,
and this decomposition takes place with explosive violence when the solution is bot
The peroxide is therefore added in small portions to the cold solution. A A^
evolution of gas continuing after the mixture has been well-stirred is an indicatifla
that sufficient peroxide has been added. The mixture is finally boiled in order to d^
stroy the excess of Na,0, and to cause the complete precipitation of titanimn, sooe

SYSTEM OF QUALITATIVE ANALYSIS. 50I

of which first passes into solution, probably as the pertitanate. The solution is cooled
because T1(0H), is somewhat more soluble in the hot solution than in the cold one.
Even in the cold when thallium is present alone about 0.5 mg. will usually pass into
the filtrate but precipitation is complete when elements of the iron group are also
present. The solution is diluted before filtering in order to avoid the disintegration
of the filter paper; it is often advantageous to support the filter by folding it together
with a small hardened filter.

(8) The Na,CO, is added to cause the complete precipitation of the alkaline-earth
dements, whose hydroxides, especially that of baritmi, are somewhat soluble even in
the presence of NaOH. ZnCO,, though insoluble in a dilute solution of Na,CO, alone,
dissolves when much NaOH is present, owing to the nearly complete conversion of the
zinc-ion into zincate-ion by the reaction Zn'^^ -f 4011"" = ZnO,"" + 2H2O. The
Na^, also serves to decompose the chromates of the alkaline-earth elements; if it is
not added, chromium may remain in the precipitate and escape detection. It is un-
necessary to add the Na^CO, when the alkaline-earth elements are known to be absent.

(9) Phosphate or oxalate, if present, divides itself in this procedure between the
precipitate and solution in a proportion which depends on the nature and quanti-
ties of the metallic elements. (See P. 51, N. 6.) Their presence does not cause any
of the elements to precipitate which would not otherwise do so, in spite of the slight
nlubility of aluminum, zinc, and beryllium phosphates. This is due to the fact that
the cathions of these elements (A1+++, Zn+"^, Be"^+) are present in the NaOH solu-
tion only at an extremely small concentration, owing to their combination with the
OH ion to form anions (A10,~", ZnOj'^, BeO,=).

(10) Even when less than i mg. Cr is present as chromate, it imparts a distinct
ydlow color to the alkaline solution, so that when a colorless solution results, it proves
the absence of this element. Uranium in small quantity also gives a yellow color
to the solution, which is noticeable with i or 2 mg. ; a moderate amount (about 10
mg.) gives a red color, which is intense with still larger quantities. Alkaline solu-
tions containing vanadium are colorless after boiling.

(11) This separation with NaOH, Na^O^, and Na,CO, is a very satisfactory one,
except in the case of zinc. This element, when present in small quantities, is com-
pletely carried down in the precipitate when much iron, nickel, or cobalt, or especially
nanganese, is present. Provision for the detection of zinc in the precipitate must
therefore be made.

(xa) If Na,0, is not available, sodium hypobromite^ NaBrO (prepared by mixing
NaOH and bromine, as described in P. 70 N. 3), may be used as the oxidizing agent,
hot it is not quite so satisfactory as Na,0,, for it does not oxidize Cr(OH), so readily,
and it is apt to oxidize some of the manganese to NaMnO^ (especially if there is not
a sufficient excess of NaOH present).

*Proccdiire 52a. — ^To detect rare earth elements, and to remove them,
if present, transfer the add solution (P. 52) to a platinum dish (after
destroying HNO„ if any has been added, by evaporating in porce-
lain once or twice with HCl (1.20)), and evaporate just to dryness.
Add to the residue 45 per cent. HF solution, little by little, until the
i^due has been dissolved, or until about 25 cc. have been added,
stirring with a platinum rod after each addition, and finally boiling
gently under a hood for a minute or two if there is still a residue. Collect
the residue on a filter supported on a platinum ring or in a celluloid

502 A. A. NOYBS, W. C. BRAY AND B. B. SPBAR.

funnel; collect the filtrate in a platinum dish; wash the icsidiic
thoroughly with water.

To the filtrate add 3-5 cc. HCl (1.20), and evaporate it just to dry-
ness; add a little HCl and evaporate again, and repeat these opeiations
a third time. Dissolve the residue in a little water, adding HCl i
necessary; transfer to a casserole, and treat the solution by the kst
paragraph of P. 52, first evaporating with HNO, to destroy HCl if
chromium, uranium and vanadium are to be tested for by *P. 5&»-^

Transfer the residue insoluble in HP to a platinum crucibk; if the
filter has been added, ignite to destroy it. Add 2-5 cc. HjSO^ (1.20),
heat until sulphuric add fumes are given off, cool, add 10-20 cc. water,
heat and filter. Treat the precipitate (which can consist only of CaSOg,
SrSO^ or BaS04) with concentrated Na,CO, solution by P. 6. To the
H2SO4 solution add ammonia until alkaline. (White pieciiMtate,
presence of rare earth elements,) Filter and test the filtrate for calcinm
and magnesium. Wash the precipitate thoroughly, dissolve it in a
little HCl, evaporate just to dryness in a platinum dish, dissolve in a
little water, and add 5-10 cc. HF. (White precipitate, presence of
earth elements.)

Notes. — (i) Among the so-called rare earth elements are included thorium, oerinai,
lanthanum, praesodymium, neodymium, yttrium, 3rtterbium, and a number of odxf
similar elements. These elements, like the alkali earth elements, are cfaaracterixBl
by the insolubility of their fluorides; but differ from the alkali earth elements in tbit
their hydroxides are precipitated by ammonia, and that their sulphates are not pR- 1
dpitated by sulphuric add. It is upon these facts that the separation described
in the above procedure is based. If there is no residue after the treatment with HF,
it shows of course that the rare earth elements are absent ; if, on the other hand, there
is a residue, it does not necessarily show thdr presence, except when it is known tlot
alkaline earth elements cannot be present. (See P. 51, N. 6 and 7.) The further
treatment described in the procedure serves to eliminate the latter.

(2) Owing to the fact that aluminiun and chromium fluorides are somewhat diiB-
cultly soluble in HF, care must be taken to use a rather large quantity of Una add
when a residue remains. It is also obvious that the fluoride residue must be thoma^
washed, if the formation of a predpitate with ammonia is to be considered a ooft-
dusive indication of rare earth elements. The final treatment of the NH4OH {R-
dpitate is recommended to eliminate the possibility of error from these soaroes.

(3) The HF solution, which may contain all the dements of the alnminam and
iron groups, is evaporated repeatedly with HCl to expel the HF, which is especiallT
apt to be retained when titanium and zirconium are present.

The Aluminum Group.

Procedure 53. — Acidify the alkaline solution (P. 52) with HNO,
(1.42), avoiding a large excess; add NH^OH (0.96) tmtil the mixtuie
after shaking smells of it, and then add 2-3 cc. more. Heat almost to
boiling in order to coagulate the precipitate, filter, and wash thoroughly
with hot water. (White, flocculent precipitate, presence of alumifnm

SYSTEM OF QUALITATIVE ANALYSIS. 503

(or *berylUum) ; colorless solution, absence of chromium.) (Precipitate,
P. 54, or to detect beryllium, *P. 58^ and h; filtrate, P. 55.)

Notes. — (i) The alkaline solution is acidified with HNO„ instead of with HCl, be-
cause the latter acid might reduce chromic add, especially if a large quantity were
added, or if the acid solution were heated. A moderate excess of NH4OH must be
added in order to keep the zinc in solution, which it does because of the production
of Zn(NH3)^+'^(N03"~)2; but a large excess is to be avoided, since it dissolves A1(0H)„
owing to formation of NH^'^AlOa . The zinc is dissolved even when carbonate, phos-
phate, or oxalate is present.

(2) Since alumintun and silica are very likely to be present in the NaOH and Na^O,
used as reagents, and since they may be taken up from the dishes, a blank test for
these impurities should be made whenever new reagents are employed for the first
time, by following P. 52 and 53 and comparing the NH4OH precipitate with that
obtained in any regular analysis. It is also well at the same time to test for zinc by
acidifying the NH4OH solution with acetic add and following P. 57.

Procedure 54. — Dissolve the precipitate (P. 53), or a small portion

of it if it is large, in 5 cc. HNO, (1.20). From the appearance of the

precipitate estimate the number* of milligrams of aluminum which have

been dissolved, and to the solution add about one-fourth as much

cobalt as cobalt nitrate, using, however, not less than 0.2 mg. Evaporate

almost to dryness in a casserole, add a drop or two of water, and soak

up the solution in a small piece of filter-paper. Make a small roll of the

paper, wind a platinum wire around it to form a spiral, and incinerate

the paper in a small flame, finally heating the residue strongly. (Blue

residue, presence of aluminum.)

Notes. — (i) This confirmatory test for aluminum should always be tried when the
NHpH precipitate is small, for general reasons, inasmuch as the precipitation by
NH4OH of an element whose hydroxide is soluble in NaOH is not very characteristic
Oead, antimony, tin, and beryllium showing a similar behavior), but also especially
to guard against mistaking SiOgH, for AlCOH),, for the former substance, if not en-
tirely removed by proper dehydration in the process of the preparation of the solu-
tion (in P. 3, Part I), will appear at this point. A gelatinous precipitate which does
not dissolve in HNO, indicates siUca; it may be tested for by P. 5.

(2) The test described in this procedure depends upon the formation of a blue com-
pound, whose formula is not definitely known; but it is doubtless a compound of the
two oxides CoO.xAljO,, and may be simply cobalt aluminate, CoCAlO,),. It may be
formed in various ways; but the process described in the procedure seems to be the
most suitable one for making the test for aliuninum reliable and delicate. It is of
the utmost importance to have the aluminum present in excess; for, otherwise, the
blue color is obscured by the black oxide of cobalt. In order that a small enough
amount of cobalt may be added, it is convenient to use a very dilute solution of cobalt
nitrate, say one that contains one-tenth of a milligram of cobalt per cubic centimeter.

(3) When the test is properly made, the ash retains the form in which the filter
paper was rolled, and the whole mass, or a large part of it is colored blue. The pres-
ence of an equivalent amount of phosphate does not spoil the test. When sodium
or potassium salts are also present, the ash fuses together, and the test is very unsatis-
factory. For this reason the sodium salts present should be all washed out of the
NH^OH precipitate before dissolving it in HNO,. No other element gives a blue color

504 A. A. NOYBS, W. C. BRAY AND E. B. SPEAR.

to the ash , but certain elements, especially iron, obscure the test, so that it can be
applied only after other elements have been removed in the regular process of analysis.
0.5 mg. Al may be easily detected, and even 0.2 mg. after a little practice.

(4) Another very good confirmatory test for aluminum consists in dissolYxng the
NH4OH precipitate in HCl (1.12), adding one and one-half volumes of ether, and
saturating the mixture in the cold with HCl gas, as described in *P. 58 g. Under
these conditions aluminum separates in the form of the cr3rsta]line compound AlCla.6H/);
but no other element of the aluminum group is precipitated, except chromium, which
if present in quantity greater than 10-20 mg. gives a violet precipitate. MoieoYer,
silica does not interfere with this test. By this process, moreover, beryllium it quan-
titatively separated from aluminum, and may be tested for in the filtrate.

Procedure 55. — Acidify the NH4OH solution (P. 53) with 30 per cent,
acetic acid solution, avoiding an excess of more than 2 cc.

If the solution is colorless, treat it by P. 57.

If it is at all yellow, add about 10 cc. 10 per cent. BaCl, solution,
allow the mixture to stand for at least five minutes and filter. (YeOow
precipitate, presence of chromium.) (Precipitate, P. 56; filtrate, P.

57.)

Notes. — (i) The presence of less than 0.5 mg. chromitun as chromate in a volume
of 50 cc. makes the solution distinctly yellow, and the addition of Bad, is thcrefoR
unnecessary when the solution is perfectly colorless. It is to be avoided, since BaSO,
may be precipitated and has then to be removed by filtration. In doubtful cases
the color of the solution should be compared with that of water. The color test is,
of course, not deh'cate by artificial light.

(2) Since some sulphate may be present, the formation of a white precipitate with
BaCl, does not prove the presence of chromium. Whether the precipitate is pore
white or yellow should therefore be carefully noted. The yellow color of a smal!
BaCrO^ precipitate is most apparent when the precipitate has settled or when it fass
been collected on the filter. If there be sufiident sulphate present to obscure the
yellow color of a little BaCr04, the confirmatory test for chromium described in the
next procedure should be tried.

Procedure 56. — Pour repeatedly through the filter a 5-10 cc portum
of a mixture of i volume of HNO, (1.20) with 9 volumes of water; to
the cold solution in a test-tube add about 2 cc of ether and i cc 3 per
cent. H2O2, and shake. (Blue coloration of ether layer, presence of
chromium.)

Notes. — (i) This blue compound which is formed by the action of H,0, on chromic
add is one of the perchromic adds. It has the formula H,Cr07, and appears to be
an addition product of HjO, and a higher oxide of chromium. It is a very unstabk
substance; by its decomposition oxygen is evolved and the chromium is reduced to
a chromic salt. Its decomposition is greatly accelerated by an excess of H,0» by the
presence of much add, and by raising the temperature. It is therefore important
not to add too much Hfi^* ^°d to use dilute add, as directed in the procedure. If,
in dissolving the predpitate, the filter be heated with HNO„ the paper causes the
chromate to be reduced to a chromic salt; but when the cold add is merdy poured
through the filter this reduction does not take place. If a green solution shoiild be
obtained, the chromitun must be reoxidized with Na,0, in alkaline solutioii before

SYSTEM OF QUALITATIVE ANALYSIS. 505

making the test. Under proper conditions 0.2 mg. Cr may be detected, but the test
may fail with a much larger amount if the directions are not followed.

Procedure 57. — Warm the acetic acid solution (P. 55 or *P. 58/) to
50® or 60®, saturate it in a small flask with H^S, cork the flask and allow
it to stand for five or ten minutes if no precipitate separates at once.
(White flocculent precipitate, presence of zinc,) Filter through a
double filter (two filters folded together), wash once with a little water.
Reject the filtrate.

To confirm the presence of zinc, pour a 5-10 cc. portion of HNO3
(1.20) two or three times through the filter containing the HjS pre-
. dpitate. To the solution add an amount of cobalt as cobalt nitrate
equal to about one-fourth of the amotmt of zinc estimated to be present,
using, however, not less than 0.2 mg. cobalt. Evaporate in a casserole
almost to dryness to expel the add, neutralize with 10 per cent. Na^CO,
solution, and add about 0.5 cc. in excess. Evaporate to dryness, ignite
gently until the purple color due to the cobalt disappears, and allow the
casserole to cool. (Green color, presence of zinc.)

Notes. — (i) ZnS precipitates more rapidly, and in a somewhat more flocculent form,
from a warm solution. Very small quantities of zinc (less than i mg.) may be missed
unless a short time be allowed for the predpitate to coagulate; but, since sulphur may
then separate, the appearance of a white turbidity is not sufiident proof of the pres-
ence of zinc. The precipitate may be allowed to settle, in order that the amount
of zinc present may be better estimated. A double filter is used, since the ZnS is apt
to pass through the filter.

(2) The immediate formation of a white flocculent predpitate with H,S in acetic
add solution is so characteristic as to be a sufficient test for zinc. Manganese is the
only other element of this group that forms a light colored sulphide; and this, owing
to its greater solubility in water, does not predpitate from an acetic acid solution.
The confirmatory test described in the last paragraph of the procedure is, however,
useful when only a small non-coagulating predpitate, which may be sulphtu-, results,
or when owing to the presence of a small quantity of other elements the predpitate
is dark colored.

(3) The green compound obtained in the confirmatory test is doubtless a compound
of cobalt and zinc oxides, perhaps cobalt zincate CoZnO,. The conditions under
which the zinc and aluminum compounds of cobalt are formed are very different.
As we have seen, the aluminum compotmd is formed only at very high temperatures,
and the test is not at all delicate in the presence of a salt of an alkali-dement. On
the other hand, the zinc compound is obtained at comparatively low temperatures,
and the presence of an alkali is essential. Excess of cobalt must, of course, be avoided,
for the black cobalt oxide completely obscures the green color. A larger proportion
of cobalt than is recommended in the procedure may be added without danger, but
the test is very satisfactory even when a large excess of zinc is present.

♦Procedure 58a. — To the fiiltrate from the NajOj predpitate (P. 52) ,
add HNO3 (1.20), keeping the solution cool, until it reacts slightly add
and any predpitate just redissolves upon shaking. (See note i.) Dilute
the solution to 100 cc. and transfer it to a strong 200 cc bottle. Add
soHd NaHCOj, a little at a time, until the mixture after shaking no

5o6 A. A. NOYBS, W. C. BRAY AND E. B. SPBAR.

longer turns blue litmus paper red at once; finally add 1.0-1.5 grams
solid NaHCOj (weighed out roughly). Close the bottle with a tightly
fitting cork, wire it in, wrap a cloth around the bottle, place it in a
vessel of warm water, and boil the water gently for twenty or thirty
minutes. Cool the bottle to at least 50° (best by slowly adding cold
water to the bath), remove the cork, filter at once, and wash, usmg
suction if the precipitate is large. (Filtrate, *P. 586; precipitate,*?.

58/.)

Notes. — (i) If up to this point in the analysis there has been no indication of any
of the elements that are to be tested for in the alkaline solution, time may often be
saved by determining before treating with NaHCO, whether any of them are present
by proceeding as follows: to one-fourth of the solution, which has been acidified
with HNO„ add NH4OH (0.96) in small excess, note whether a precipitate fonns^
and then add a few drops of (KH.^)^ solution. Even if there is no precipitate, vana-
dium may be present; to test for this, add 3 cc. NH4OH (0.90) and completely satu-
rate with H^. (See *P. 51a.) If the results of these tests show that any of these
elements are present, treat the remainder of the HNO, solution by the r^ular pfo-
cedure. If, on the other hand, the results are negative, no further treatment is neces-
sary.

(2) The alkaline solution is kept cold during the neutralization and an excess of
HNO, is avoided because chromate in the presence of H^O, and an add is rapidly
reduced to a chromic salt, especially when the solution is hot. This reduction, if
complete, would prevent the detection of chromium in the subsequent test. H,0,
will, to be sure, not be present, since NajO, is very rapidly destroyed by the boiliag
of the alkaline solution in P. 52, except when uranium is also present; in this case
the peruranate which is formed by the treatment with Na,0, is not decomposed upon
boiling, but breaks up into a uranyl salt and H,0, upon acidification.

(3) The success of this separation depends upon securing the proper concentra-
tion of the NaHCO,. Since the NaHCO, may be at first used up in precipitating
zinc, aluminum, and beryllium as well as in neutralizing the free add, the weighed
amount of NaHCO, is added only after the solution ceases to react distinctly add.
A much larger concentration of NaHCO, than i.o to 1.5 per cent, would not prevent
the complete predpitation of zinc or of aluminum, but it would interfere with that
of beryllium. Thus with a volume of 100 cc. the predpitation of the beryllium
is complete when the concentration of the NaHCO, is i per cent., and i mg. can ns-
ually be detected when it is 2 per cent., but 3 mg. remain dissolved in a 3 per cent
solution, about 15 mg. in a 5 per cent, solution, and about 150 mg. in a 10 per cent,
solution. A smaller concentration of NaHCO, and a smaller volume than 100 cc
are avoided, in order to prevent as far as possible the predpitation of uranyl vanadate,
which may otherwise occur when large quantities of uranium and vanadium (about
100 mg. of each) are simultaneously present. The presence of phosphate or oxalate
(or chromate) does not cause the predpitation of uranitmi, nor otherwise interfere
with the analysis. When only small amounts of the dements of this group are pres-
ent, the separation can be made in a smaller volume, care being taken that the con-
centration of the NaHCO, rather than the quantity taken be that prescribed

(4) If it is desired, the separation may be made in an open flask ; but in this case
the solution must not be boiled and the NaHCO, solution should not be stronger than
I per cent. The process is then best carried out by digesting the mixture on a miter
bath for 20 to 30 minutes in a flask covered with a watch glass. Under these condi-

SYSTEM OI^ QUALITATIVE ANALYSIS. 507

tioos neither zinc nor beryllium dissolve in significant quantity; but one to two milli-
grams of aluminum may be completely dissolved. The mixture must not be boiled,
for a larger amount of aluminum may then dissolve (as much as five milligrams on
one minute's boiling).

(5) When a large amount of aluminum or beryllium is present, two to five milli-
grams of uranium may be carried down almost completely in the NaHCO, precipi-
tate, so that in this case uranium has to be tested for in the analysis of the precipi-
tate.

(6) In a dilute solution of Na+HCO, sattu-ated with CO,, the hydrogen ion and
hydrozyl ion concentrations are both very small and are nearly the same as in pure
water. A10,H„ basic BeCO, and ZnCO, are completely precipitated because the
solubility of these substances is not much increased through removal of the OH
or CO,~ by combination with H+ with formation of H^O or HCO, (Ic&ving the cathions
Al'*"^"'", Be+'^orZn"'"'*in the solution) or through removal of the metal ions by com-
bination with the 0H"~ with the formation of A10,~", BeOj^ or ZnO,'^. The higher
hydroxides of chromium, and vanaditun (chromic and vanadic adds) are soluble
polybastc adds, which are so much more highly ionized than H,CO, that they displace
it from its salt NaHCO„ forming mainly sodium hydrogen chromate or vanadate.
The fact that uranium is not predpitated, even though Na^VjO^ is difiicultly soluble,
may be due to the formation of a complex sodium uranyl carbonate.

^Procedure 58b.— To the NaHCO, filtrate add HNO, (1.20) until
the solution is distinctly add, avoiding an excess of more than i cc.
(Colorless solution, absence of chromium.) Unless the solution is as
colorless as pure water, test one-fourth of it for chromium by adding to
the cold solution in a test-tube 2 cc. ether and 0.5-1.0 cc. 3 per cent.
HjOj solution. (Blue coloration of the ether layer, presence of
chromium.) To this mixture add about 5 cc. HNO3 (1.42). (Red
coloration of water layer, presence of vanadium.)

If chromium is not present, treat the HNO, solution or the part of
it not tested with HP, by *P. 58c.

If chromium is present, exactly neutralize the remainder of the
HNO, solution with NaOH, add 2 cc. HNO, (1.20) and then 20 cc.
20 per cent. PbCNO,), solution; allow the mixture to stand for fifteen to
twenty minutes, and filter. (Yellow precipitate, presence of chro-
mium.) Saturate the filtrate with H,S, filter off and reject the pre-
cipitate; boil the filtrate for two or three minutes to expel H3S and to
coagulate any sulphur that may separate; filter off and reject the pre-
cipitate. In order to oxidize vanadyl salts to vanadate, add bromine
water (or if much is required, liquid bromine) until the solution has a
permanent reddish color. Boil in a casserole until the bromine is
expelled and treat the solution by *P. 58c.

Notes.— (i) The NaHCOg filtrate should not be heated after acidifying because
some H^O, may still be present (even though most of it is decomposed in the heating
with NaHCO,), which would cause the reduction of chromate, and might thus pre-
vent the detection of even several milligrams of chromium. (See *P. 58a, N. 2.)

(2) It is desirable to determine in advance, ty n'ald'rg a jTch'n 'ran test if rcccs

508 A. A. NOYES, W. C. BRAY AND E. B. SPEAR.

sary, whether or not chromium is present, for, if it is absent, the addition of the lead
salt, and the subsequent removal of the lead with H^ may be omitted. If lead be added,
its removal is necessary before the tests for uranium and vanadium can be made.
If the add solution is perfectly colorless, an amount of chromium exceeding 0.5 milli-
gram may safely be pronounced absent. Uranium and especially vanadium in mod-
erate quantity (20 to 50 mg.) also give yellow solutions. A yellow solution is there-
fore tested for chromate with H^O,. In regard to this test, see P 56, N. i. The
test can be made in only a small portion (one-fourth) of the solution , for it is so deli-
cate that the presence of o.i milligram chromium in this portion can be detected.
The portion in which the test is made is rejected, because chromium, if present, is re-
duced to the chromic state by the H^O, in the add solution, and would therefore not
be predpitated by the lead nitrate.

(3) When vanadium is present even in moderatdy small quantity, the water layer,
on the addition of H^O,, becomes orange-ydlow to orange-red in color, owing to the
formation of pervanadic add. This test for vanadium becomes more delicate when
the solution is made strongly add as is directed; a distinct color is then obtained even
when only 0.5 mg. vanadium is present in the portion tested, corresponding to about
2 mg. in the whole solution. (Compare *P. 51a, N. 4.) The test is not essential
since vanadium is always tested for later; it is introduced here, where it can be made
in a moment's time, as an additional confirmation of the presence or absence of that
element.

(4) The separation of chromium from vanadium and uranium by Pb(N0,)2 depends
on the rdatively small solubility of lead chromate in dilute HNO3 in the presence
of a large quantity of Pb(NO,),. Under the conditions described in the prooedm:e,
over 100 mg. of vanadium yidd no predpitate and only o.i to 0.3 mg.
of chromium usually remains in solution, though this amotmt is somewhat increased
when a very large quantity of NaNO, has been introduced. To secure these resuhs,
however, care must be taken to use the prescribed quantities of HNO, and Pb(N0,)9
for lead vanadate is also a difficultly soluble substance, and would be quantitatively
predpitated in the presence of a much weaker add, such as acetic add in the presence
of ammonium acetate. Lead uranate would also be predpitated from a neutral or
slightly alkaline solution.

(5) The presence of much chloride or sulphate would cause the predpitation of
white PbCl, or PbSO^, which might obscure the ydlow color of a small quantity of
PbCr04 and prevent the estimation of the amount of chromium present. These anions
will, however, not be present in harmful quantity, if in dissolving the original (NHJgS
predpitate (P. 51) the adds are used in the way prescribed, and if in addifying the
solutions in *P. 58a and 6, HNO, and not HCl is used, as directed.

(6) The lead which is added to predpitate the chromate must be removed before
testing for uranium with Na3HP04 (in *P. 58c), since it would give a predpitate of
Pb,(P04),. It is precipitated with H^S, rather than with (NHJ^O^, because in the
latter case enough lead (about i mg.) still remains in solution to give a predpitate
with Na^PO^.

(7) By the HjS, vanadic add is reduced to vanadyl nitrate VO(NOs), slowly in the
cold but more rapidly on heating the solution, so that sulphur may be predpitated
on boiling the filtrate to expd H^S. This must be re-oxidized to vanadic add, for
otherwise vanadyl ammonium phosphate would be predpitated with the uranium
phosphate in *P. 58c. The oxidation by bromine does not take place instantaneously;
and a considerable excess of bromine must therefore be used, and the mixture alloved
to stand for a few minutes.

^Procedure 58c. — Make the solution obtained in fP. 586 neutral

SYSTEM OF QUALITATIVE ANALYSIS. 509

with NH4OH, add 5 cc, 30 per cent, acetic add, 1-2 grams solid
(NHJjSO^ (or NH.NOg), and then 2 grams soUd Na2HP04.i2HaO,
and heat to boiling. (White precipitate, presence of uranium.) Allow
the mixture to stand ten to fifteen minutes to coagulate the precipitate,
filter, and wash the precipitate with a 5-10 per cent, solution of
(NHJjSO^ (or NH.NOa). (Precipitate, *P. 58^; filtrate, *P. 58^.)

Notes. — (i) The precipitation of white uranyl ammonium phosphate, UO2NH4PO4,
is a delicate test provided the solution be made only moderately add with acetic acid
and a sufficient excess of Na^PO^ be added. The separation from vanadium as
vanadic add is a very satisfactory one; for vanaditun, even when present in large quan-
tity (100 mg.), does not predpitate, nor is a small quantity carried down by uranium;
on the othei* hand 0.5 mg. uranium gives a distinct predpitate. The test is a some-
what more delicate one if made in a smaller volume, say 40 cc. ; in this case the same
quantities of reagents may still be used, and the separation is perfectly satisfactory.

(2) The formation of a predpitate at this point is not suffident evidence of the pres-
ence of uranium, for aluminum, beryllium, and lead will separate as phosphates if
they have not been completdy removed in previous procedures, and vanadium will
predpitate as vanadyl ammonium phosphate if the oxidation by bromine was incom-
plete. On the other hand a slight turbidity may correspond to an appredable amount
of uranium (0.2 to 0.5 mg.). Therefore the confirmatory test for uranium (*P. $Sd)
should always be tried.

(3) When the solution is made alkaline with NH4OH a pale yellow predpitate of
ttnmyl ammonium vanadate may separate, and may not dissolve when the acetic
acid is added. This, however, has no effect on the separation, since, on boiling, it is
converted into the less soluble tu-anyl ammonium phosphate and vanadic add.

(4) In the absence of ammonium salts uranium is predpitated as tu-anyl hydrogen
phosphate, UO^HPO^, which often separates as a very finely divided predpitate that
runs through the filter. The ammonium salt is added to cause the uranium to pre-
dpitate as uranyl ammonium phosphate, which coagulates and filters more readily.

^Procedure sM. — Dissolve the Na^HPO^ precipitate (*P. 58c) by
pouring a small portion of hot HCl (1.12) repeatedly through the filter,
evaporate the solution nearly to dryness, add about 10 cc. nearly
saturated NaCl solution, pour into a test-tube, cool, and add 5 cc. 10
per cent. K4Fe(CN)e solution. (Dark red precipitate or coloration,
presence of uranium.)

Notes. — (i) The predpitation of the dark red uranyl ferrocyanide (UOj)jFe(CN)e
is a very characteristic and also a very delicate test for uranium, provided care be taken
to avoid any excess of add. The solution must, however, be distinctly add, for other-
wise the predpitate may not form, owing to the small concentration of the uranyl
ion, UOj"'"'". Uranyl ferrocyanide tends to form a colloidal solution, but the pres-
ence of NaCl and HCl soon causes it to coagulate.

Procedure sSe. — Neutralize the filtrate from the NajHPO^ precip-
itate (♦P. 58c) with NH4OH (0.90) and add at least 5 cc. more. Saturate
the solution completely with HjS by passing the gas through it in a
small flask for ten or fifteen minutes. (Pink or violet red color, presence
of vanadium.) Pour the solution through a filter; make it distinctly

5IO A. A. NOY«S, W. C. BRAY AND «. B. SP^AR.

acid with acetic add (or HCl) and heat nearly to boiling. (Darkpre-
cipitatCi presence of vanadium.)

Notes. — (i) In regard to the formation of the sulphovanadate of vanadium in the
strongly alkaline solution, the partial precipitation of vanadium sulphide on acidi-
fying, and the action of H,0, on vanadic add, see the Notes on *P. 51a.

*Proccdure sSf.— Heat the NaHCO, precipitate (*P. 58a) with
10-30 cc. HCl (1.06) in a casserole, or dissolve it by pouring the add
two or three times through the filter, boil the solution to expel CO,,
add NH4OH (0.96) until the solution after shaking smells of it and
then 2-3 cc. in excess. Heat the mixture nearly to boiling, filter and
wash the predpitate, using suction if it seems desirable. (Predpi-
tate, *P. 58g; filtrate, addify with acetic acid and test for zinc by P. 57.)

Notes. — (i) The zinc remains in the ammoniacal filtrate as a complex ammonia salt,
chiefly Zn(NH,)4++G,"~. Aluminum and beryllium are predpitated as hydroxides.
The predpitate will also contain any uranyl vanadate that was predpitated in the
NaHCO, separation.

♦Procedure sSg. — Dissolve the NH^OH predpitate (*P. 58/) by
pouring a hot 5-15 cc. portion of HCl (1.12) repeatedly through the
filter, using another portion of add, if necessary. Add a volume of
ether equal to one and a half times that of the solution. Pass into
the mixture in a small flask HCl gas until a single layer results, and
until fumes of HCl are copiously evolved, cooling the flask in nmning
water during the progress. Cover the flask and let the mixture stand
for fifteen minutes even if no predpitate has separated. (White crystal-
line predpitate, presence of aluminum.) Filter through an asbestos
filter (see P. 61) or an ordinary filter supported by a small hardened
filter folded with it, after first moistening the filter with a mixture of
two volumes of HCl (1.20) and three of ether previously saturated with
HCl gas; wash the predpitate once with this mixture. During the
filtration and washing keep the funnel covered with a watch glass to
prevent evaporation of the ether. (Precipitate, P. 54 to confirm the
presence of aluminum; filtrate, *P. 58^.)

Notes. — (i) Aluminum chloride, AlCl^.6Kfi, is only slightly soluble in concentrated
HCl solutions and the predpitate is complete when ether is added and the mixture
saturated with HCl. 0.5 mg. of aluminum can be easily detected in 30 cc The
test is therefore a ddicate one. It is also very characteristic; for no other dements
of the aluminum group is predpitated by this treatment, except chromium when it
is present in moderate quantity.

(2) The ethereal solution of HCl, unlike the concentrated aqueous solution, does
not disintegrate filter-paper rapidly, and the filtration can almost always be made
with an ordinary filter supported by means of a hardened filter. This filtration is
apt to be slow, and it is often advantageous to filter through asbestos with the help

of suction.

(3) When the NH^H predpitate is small (corresponding to less than 30 mg. Al),
it need not be treated by this procedure, but it may be dissolved in a little HCl, and
treated directly by *P. 58*.

SYSTEM OF QUAUTATIVE ANALYSIS. 5II

♦Procedure sSh. — Evaporate the filtrate from the HCl precipitate
(*P. sSg), first on a water-bath under a hood until the ether is expelled,
and then over a flame, almost to dryness tmtil nearly all the HCl is
expelled; add a little water, make alkaline with NH4OH, avoiding a
large excess, and heat nearly to boiling. (No precipitate, absence of
beryllium.)

If there is a precipitate, dilute to 30 cc, add enough solid NaHCO,
to make a 10 per cent, solution, heat to boiling, boil for one minute,
cool, pass in H^S for a few seconds, and filter after a few minutes if
there is a precipitate. Acidify the filtrate with HCl, boil for two
or three minutes to expel CO,, make the solution alkaline with NH4OH,
and heat nearly to boiling. (Whjte flocculent precipitate, presence of
beryllium, yellow precipitate, presence of uranium.)

Treat this precipitate or a portion of it corresponding to about 20
mg. of beryllium by *P. 58d. (Dark red precipitate, presence of
uranium.) Filter and to the filtrate add NH4OH. (White pre-
dpitate, presence of beryllium.)

Notes. — (i) The filtrate from the HG precipitate may contain besides beryllium,
anmium or vanadium (carried down in the treatment with dilute NaHCO, in *P.
58a), alumimmi (if care was not taken in precipitating and filtering aluminmn chloride),
and a little iron introduced from the reagents). This solution is first tested with
NH4OH to determine if any further treatment is necessary. The treatment with the
hot concentrated NaHCO, solution serves to precipitate the aluminum completely
and to dissolve the beryllium and uranium. This method of separation of aluminum
and beryllium is a satisfactory one, when, as in the present case, only a small amount
of aluminum is present ; a large quantity of aluminum, however, retains almost com-
pletely a small quantity of beryllium. The NaHCO, solution must not be boiled
for a long time, because, owing to the escape of CO, and the formation of free NaOH,
the solution becomes alkaline enough to dissolve some aluminum. The reason for
the great solubility of beryllium hydroxide in concentrated NaHCO, solutions is not
known. H^ is passed in to remove the iron, since a small amount of it remains dis-
solved in the concentrated NaHCO, solution; the filtrate is sometimes dark green
even after repeated filtration, due to the presence of colloidal iron sulphide; to precipi-
tate this, add a small amount of a ferrous salt (say 2 to 4 mg. Fe), shake, allow the mix-
ture to stand several minutes, and then filter, finally passing more H^S into the filtrate
to make sure that the iron has been all removed.

(2) The treatment with K4Fe(CN)0 described in *P. 58 <f gives a satisfactory sepa-
ration of beryllium from uranium (and vanadium) provided the amount of beryl-
lium in the solution does not exceed 20 milligrams. When more beryllitmi is pres-
ent a gelatinous precipitate separates; on this account it is directed to use only a por-
tion of the NH4OH precipitate when it is large, but even in this case a small amount
of uranium can still be detected, owing to the delicacy of the ferrocyanide test. Vana-
diom, if present owing to its having been precipitated as uranyl vanadate in the treat-
ment with dilute NaHCO, (*P. 58a), remains with luranitmi and is precipitated as
greenish yellow vanadyl ferrocyanide, but does not obscure the dark red color of the
unminm predpitate. It is not necessary to provide for the detection of vanadium at
this point, since a large quantity always dissolves in the NaHCO, treatment.

512 A. A. NOYES, W. C. BRAY AND E. B. SPBAR.

The Iron Group.

Procedure 6i. — Transfer the NajO, precipitate (P. 52) to a casseiok
together with the filter if necessary, add 5-30 cc. HCl (1.12), boil gently
till the precipitate is dissolved, filter to remove the paper, and evaporate
the filtrate to i or 2 cc. To decompose the HCl add about 5 cc. HNO,
(1.42), and boil as long as oxides of nitrogen are given off. Add 5-20
cc. HNOg (1.42), heat to boiUng, add about 0.5 gram of solid KCIO,
and boil gently, adding more KClOj in small portions if a large pre-
cipitate forms. (Dark brown or black precipitate, presence of manr
ganese.) Boil gently for a minute or two, and filter through an asbestos
filter, made by pouring a suspension of washed asbestos over a com-
pact wad of glass wool in a glass funnel. (See note 4.) Heat the
filtrate to boiling, add more KCIO,, boil, and filter through the same
filter if more of the precipitate separates. Wash two or three times
with HNO, (1.42) which has previously been freed from the oxides d
nitrogen by warming with a Uttle KCIO,. Evaporate the filtrate to
about 5 cc, but not further, dilute to 20 or 30 cc, and filter the solution
if it is turbid. (Precipitate, P. 62 ; filtrate, P. 63).

Notes. — (1) HCl is used for dissolving the NajO, precipitate rather than HNQr
Pure concentrated HNO, docs not dissolve hydrated MnO, except in the presence
of filter paper, whereby the HNO, is reduced to lower oxides; the action is more rapid
with HCl (for the MnOj is thereby quickly reduced to manganous chloride with evolu-
tion of chlorine.)

(2) By HCIO, in HNO, solution (but not by HNO, alone) manganous salts are rap-
idly oxidized to hydrated MnO, with formation of chlorine dioxide (CIO,), which es-
capes as a yellow gas. The HCl must previously be completely removed by evapora-
tion and boiling with HNO„ since the oxides of nitrogen resulting from its action
on the HNO, would continuously dissolve the precipitate.

(3) The separation of manganese in this way from the other metals of this group
is entirely satisfactory with the following exceptions. A small quantity of iron (up
to I mg.) may be completely carried down with a large quantity (500 mg.) of manga-
nese. The same is true in a much higher degree of titanium, of which even 50 mg.
may be entirely precipitated with 500 mg. of manganese. Much zirconium is also
carried down, but never quite completely. Provision is therefore made (in ♦?. 62a)
for the detection of these three elements in the precipitate (if it is large) as well as in
the filtrate. If it is not thought necessary to test for zirconium in the predpitale,
the certain detection of titanium can be much more quickly accomplished just before
precipitating the manganese, by diluting the HNO, solution, from which the HQ
has been removed, with once or twice its volume of water, and adding 3 cc of 3 P«^
cent. H O, solution. An orange-yellow or orange-red color shows the presence of
titanium. (See *P. 656, N. i.) In the absence of HCl iron does not interfere with
this test, nor does even a large quantity of nickel or cobalt prevent a distinct change
of color from being seen.

(4) In filtering the MnOj a cylindrical glass funnel with a small delivery tube is
usually employed in quantitative analysis. An ordinary conical funnel is, however,
satisfactory, provided the wad of glass wool is made compact and enough asbestos

SYSTEM OF QUALITATIVE ANALYSIS. 513

is used. Filtration will, to be sure, be slow if the glass wool is packed very tightly
or if the asbestos mat is very thick, but in that case suction may be applied.

(5) On evaporating the HNO3 filtrate, titanium and zirconium oxides may sepa-
rate out, especially if the solution be evaporated almost to dryness. For this reason
the solution is evaporated only to about 5 cc, but even then a small white precipi-
tate will sometimes be obtained when these elements are present in large amount.

Procedure 62. — Transfer the whole of the HClOj precipitate (P. 61)
it it is small (containing less than 5 mg. Mn), or 5-10 mg. of it if it is
large, to a casserole; add i or 2 grams solid PbOj, and about 10 cc.
HNO3 (1.20); boil for about two minutes in a casserole covered with a
watch glass; pour the mixture into a test-tube, and allow the PbO, to
settle. (Violet red solution, presence of manganese.)

If the precipitate is large, dissolve the remainder in hot HCl (1.12)
in a casserole, or by pouring a 10-15 cc. portion repeatedly through the
filter, and boil the solution to expel chlorine. To one-tenth of this
solution add 5 cc. KCNS solution. (Deep red color, presence of iron.)
Treat the remainder of the HCl solution by *P. 62a to recover any
titanium and zirconium that may be present.

Notes. — (i) This confirmatory test for manganese is usually superfluous since the
precipitation of MnOj by HCIO, is highly characteristic. In order that the PbO,
test may be satisfactory, the HNO, used must be fairly concentrated and the boil-
ing continued for two or three minutes.

♦Procedure 62a. — To the HCl solution (P. 62), without evaporating
it, add NH4OH (0.96) until the mixture is barely alkaline, avoiding
the addition of more than two or three drops in excess; heat nearly to
boiling for two or three minutes, filter at once, and wash with hot water.
Reject the filtrate. Dissolve the precipitate in a little hot HCl (1.12)
(without diluting it with water), reserve it, and unite it with the main
HCl solution to be treated in *P. 65a.

Notes. — (i) In order to avoid as far as possible the precipitation of manganese
by NH4OH, the OH"" concentration is kept small by avoiding an excess of NH^OH
and by having a large quantity of ammonium salt present. Under these conditions
the oxidation by the air to the manganic state is slow. If, however, much NH^H
be added, oxidation takes place rapidly and much manganese may be precipitated
as Mn(0H)3, yielding a brown precipitate. Even at best a little manganese will come
down, but a moderate amount does not interfere with the subsequent tests for titanium
and zirconium.

Procedure 63. — Add about one-tenth of the HNOj solution (P. 61)
to three or four times its volume of ammonium molybdate reagent,
and heat to 60-70°. (Yellow, finely crystalline precipitate, presence of
phosphate.) If there is no precipitate, or only a very small one, treat
the remainder of the HNO, solution by P. 64; otherwise by P. 65.

Notes.— (i) Phosphate is tested for at this point because a different treatment is
necessary when it is present in significant amoimt, in order to separate from it alka-
line-earth elements and to provide for their detection. When phosphate is not pres-

y A. A. NOYBS, W. C. BRAY AND B. B. SPEAR.

, iroQ, thallium, titanium, and zirconium can be separated from nickel, o

1 the allcaliue-earth elementa by NH,OH (as in P 64); but, when conadi
Kphate is present, the alkaline-earth elements would be partly or wholly pi
ed in combination with it. (See P. 51, N. 6.)

3) In order that the phosphate test may be delicate and may appear inunedi

vge proportion of the molybdate reagent must be used and the solntioa mi

rmed. The predfritate of ammonium phosphomolybdate is of complicate

lewhat variable compositian; it contains ammonium phosphate and mo

i, approximately in the proportion (NH,),PO,;i2MoO,.

Procedure 64.— If phosphate is absent, make the HNO, sol

, 6a) strongly alkaUne with NH^OH (0.96) using an excess of 3-

ark red precipitate, presence of iron.) Filter, and wash the

litate, using suction if the precipitate is large, and sucking it a

possible.

Treat the filtrate by P. 66.

Dissolve the precipitate in HCI (1.12), warming if necessary, t

re not to dilute the acid by wash water. To about one-tenth c

ution add 5 cc. KCNS solution. (Dark red color, presence of '

reat the remainder of the' solution by *P. 65a.

'ioles. — (1) If titanium and zirconium are to be tested for, it is important

re the NHjOH precipitate in HQ of a specific gravity 1.12 (14 per cent, HC

tvoid dilution, for the separation in *P. 65a depends on the concentration of tl:

3) The red color obtained on adding KCNS is due to the formation of un-i

ic thiocyanate, Fe(CNS),. This test may be made in the presence of

1, for the add HCNS is also a highly dissodated odd, which is therefore 0

ced from its salt. Much HNO, must not, however, be present; for, by its

KCNS, NO, may be formed and this also gives a deep red color with KCNS.

t for iron is an extremely delicate one ; and if only a faint color is obtained, th'

d in the process must be tested for iron.

Procedure 65. — When phosphate is present, test one-tenth o

■JO, solution (P. 63) for iron, by evaporating it just to dryness, at

2 cc, HCI (1.20), evaporating again to decompose the HNO,, dil
5 or 10 cc and adding 5 cc. KCNS solution. (Permanent red

;sence of iron.) To the remainder of the solution add NHjOH I
til the precipitate formed by the last drop does not redissoh
iking. If, owing to the addition of too much NH(OH the sol
comes alkahne or a large precipitate separates, make it disti
id with acetic add. Add 5 cc. of a 50 per cent, solution of ammc
;tate, and, unless the mixture is already of a brownish red coloi
per cent. FeClj solution drop by drop until such a color is prod
Id enough water to make the volume about 100 cc, boil in a 2;
sk for five minutes, adding more water if a very large preci)
larates, and let the mixture stand for a minute or two. Filter
11 hot, and wash vrith hot water. Add 3-5 cc. more ammc
etate solution to the filtrate, boil it again, and collect on a sep

SYSTEM OF QUALITATIVE ANALYSIS. 515

filter any further precipitate. Make the filtrate alkaline with NH4OH,
adding an excess of 2-3 cc, filter off and wash any precipitate, uniting
it with the main precipitate. (Filtrate, P. 66; precipitate, reject or
dissolve in hot HCl (1.12), and treat by *P. 65a.)

Notes, — (i) With regard to the test for iron with KCNS and the necessity of re-
moving the HNO„ see P. 64, N. 2.

(2) This method of separation depends on the facts that, upon boiling an acetic
add solution containing much acetate, ferric iron and titanium are completely pre-
cipitated, and thallic thallium and zirconium nearly so, in the form of a basic acetate
or hydroxide; and that all the phosphate present combines with these elements when
they are present in excess, and therefore it then passes completely into the precipitate,
leaving the bivalent elements in solution. This behavior of the phosphate is due to
the fact that the solubility in acids of the phosphates of the trivalent and quadrivalent
elements is much smaller than that of the phosphates of the bivalent elements.

(3) If upon adding the ammonium acetate the solution becomes of a reddish color,
it shows that iron is present in quantity more than sufficient to combine with the phos-
phate; for a cold solution containing ferric acetate is of a deep red color. If, on the
other hand, a colorless solution results (either with or without a precipitate), it shows
that there is no excess of iron, and FeCl, is therefore added, which causes the precipi-
tation of FePO^ as a yellowish white precipitate. Upon boiling, the excess of iron
separates completely as a dark red gelatinous precipitate of basic ferric acetate, leav-
ing the supernatant liquid colorless, except when nickel or cobalt is present.

(4) The solution is diluted to at least 100 cc., owing to the large volume of the
precipitate, and it must be heated in a capacious flask owing to its tendency to boil
over.

(*5) Zirconitmi may not be completely precipitated under the conditions of this
procedure, 1-5 mg. sometimes remaining in solution, especially when consid-
erable acetic add is present. To ensure its complete separation, the filtrate is made
alkaline with NH«OH.

(*6) The precipitation of thallium in this procedure is not quite complete, but there
is no danger of losing even 0.5 of a milligram when iron is also present, as is always
the case.

♦Procedtire 65a. — To the HCl solution (P. 64 or 65, and *62a) which
should be of specific gravity of 1.11-1.12, add 10-20 cc. more HCl
(1.12), transfer the (cold) solution to a separating funnel, add an eqtial
volume of ether, shake vigorously several times (preferably after in-
verting the funnel and opening the cock), and then allow the two layers
to separate. Draw off the layers separately, and rinse out the funnel
with a little ether. Return the aqueous layer to the funnel and treat it
with ether as before; if necessary, repeat this treatment once or twice
until the ether layer remains colorless. (Water solution, *P. 656;
first ether solution, *P. 65^; remaining ether solutions, reject.)

Notes. — (i) If the directions are followed, 97 to 99 per cent, of the FeG, present
passes into the ether layer in each extraction. It is evident from this statement,
that even when 500 mg. of iron are present, substantially all of it will be removed
in three extractions. But it is important that the concentration of the HCl solution
in contact with the ether layer lie within the narrow limits of 20 to 22 per cent. HD,
corresponding to a specific gravity of i . 10 to i . 1 1 at 1 5 ^ This concentration is realized

5l6 A. A. NOY^, W. C. BRAY AND E. B. SPEAR.

in the procedure, even though a little stronger add is used, for some of the HCl passes
into the ether layer. The extraction of the FeCl, is less complete both with stronger
and weaker HCl solutions; thus with HG containing initially either i8 or 25 per cent
about 94 per cent, of the PeCl, passes into the ether layer; while with 8 per cent. HCl
only 4 or 5 per cent, of the FeCl, was extracted in each shaking. Almost all the thal-
lium, which is present as T1C1„ also passes into the first ether extract.

(2) The following are probably the principles involved in this ether extraction.
Since iron passes into the ether layer only in the form of PeCl,, the quantity of it ex-
tracted by the ether increases, in accordance with the distribution law, the larger the
proportion of un-ionized anhydrous FeCl, in the water layer. This proportion is,
however, increased by increasing the concentration of HCl both in virtue of the re-
duction of the ionization by the common-ion effect and of the reduction of the hydrol-
3rsis of the ferric salt by the free acid. It is doubtless true that the strong add has
also a dehydrating effect, thereby increasing the anhydrous FeCl, in the water layer.
As the HCl becomes very concentrated, however, another effect, opposite in charac-
ter, comes into play; namdy ether dissolves in large quantities in the aqueous layer,
and HCl and water dissolve in large quantity in the ether layer, thus making the two
layers more nearly alike, and doubtless decreasing the distribution-ratio for the FeCl,
between the ether and water layers. With respect to this explanation, it should be
added that it is uncertain to what extent complex adds (HFeQ^ etc.) may be involved.

(3) Since the color of the ether layer is a sensitive indication of iron, the treatment
with ether may be discontinued as soon as a nearly colorless ether extract is obtained.
When titanium is present the water layer may remain distinctly yellow, owing to the
presence of hydrogen peroxide as impurity in the ether; such a color does not, there-
fore, show that iron is still present.

(4) Titanium and zirconium remain completdy in the water layer. When mucb
zirconium is present some of it may be predpitated out, as chloride; but this remains
suspended in the water layer.

(5) Phosphoric add does not interfere with this separation, and the iron can there-
fore be extracted even after the basic acetate procedure.

♦Procedure 65b. — Heat the HCl solution (*P. 65a) on a waterbath
until the ether is expelled, add i cc. H2SO4 (1.20), evaporate almost to
dryness until the H3SO4 begins to fume, adding i cc. more H,S04 {1.20)
if the residue is solid. If the residue is dark colored, owing to organic
matter, add a few drops HNO3 (1-42) and evaporate again imtil the
HjS04 begins to fume. Cool, add 5 cc. water, 10 cc. 3 per cent. HjO,
solution, and then 10 cc. 10 per cent. Na3HP04.i2H20 solution. (Orange-
yellow to orange-red solution, presence of titanium; white, flocculent
precipitate, presence of zirconium.) Let the mixture stand for at least
an hour, filter, and wash the precipitate. (Filtrate, if colored, *P. 65c;
precipitate, note 3.)

Notes. — (i) By the addition of HjO, titanium is converted into a sulphate corre-
sponding to the higher oxide TiOj, the red color being due to the cathion. This color
test is an extremdy ddicate one, even o.i mg. Ti imparting a distinct, yellow color
to the solution.

(2) By the NajHP04, zirconium is predpitated as a basic phosphate, Zr(OH)P04.
Its predpitation is slow; but, if nothing has separated after half an hour, it is safe
to conclude that less than 0.5 zirconium is present, provided care has been taken not
to use more H3SO4 than is directed in the procedure. If the titanium had not been

SYSTEM OF QUALITATIVE ANALYSIS. 517

oxidized by the addition of Hfi^, it would also give an entirely similar precipitate
of Ti(0H)P04; but from a solution containing H^O, in excess when titanium is alone
present none of it separates even on standing several hours. When zirconium and
titanium are present together, a small proportion of the titanium is carried down
with the zirconium and the phosphate precipitate may then have a distinct, yellow
color.

(3) Some of the rare earth elements, such as thorium, may also be precipitated
as phosphate at this point, if these elements have not been proved absent or removed
in *P. 52a. In such a case the presence of zirconium may be proved by pouring a
portion of dilute HF several times through the filter (supported in a celluloid funnel
or a platinum ring), evaporating with H^S04 in a platinum dish, diluting, and adding
MH4OH. The rare-earth phosphate would be left undissolved by the HF.

(4) Besides titanium and zirconium, the solution from *P. 65a will sometimes
contain uranium, manganese, and cobalt (carried down in the precipitates in P. 52,
*P. 62 a, and P. 64, respectively); but not in sufficient amoimt to interfere with the
zirconium and titanium tests. The only other element of the aluminum and iron
groups that gives a similar color with HgO, in acid solution is vanadium; but this,
aside from the fact that it should not be present, would give no precipitate with
Na^PO^ (not even on the addition of NaaSO, in *P. 65c).

Procedure 65c.— To the HjOj solution (*P. 656), if colored, add
powdered NajSO^ little by little until the solution is decolorized, and
let the mixture stand twenty or thirty minutes. (White flocculent
precipitate, presence of titanium.)

Notes. — (z) This test serves to confirm the presence of titanium and to enable the
quantity of it to be better estimated. HjO, and the sexivalent titanitmi compound
are rapidly reduced by the action of H^0„ even in the cold, and titanitun then
precipitates as Ti(0H)P04. As in the case of zirconium, the precipitation takes place
slowly. When there was only a faint color with HjOj, no precipitate will be ob-
tained; but 0.5 mg. of titanium is easily detected, if the concentration of the sul-
phuric add does not exceed i cc. H^SO^ (1.20) in 25 cc. solution.

♦Procedure 65d. — Evaporate the first ether extract (*P. 65a) on a
waterbath, and dissolve the residue in 3-5 cc. H2SO4 (1.20) and 3-5 cc.
water. To the cold solution in a test-tube add 2-3 cc. i per cent. KI
solution and powdered NajSOg a little at a time until the iodine color
has permanently disappeared. (Yellow precipitate, presence of
thaUium.) Filter, using preferably a hardened filter when the pre-
cipitate is small. (Collect a little of the precipitate on a clean platinum
wire and introduce the wire into a colorless gas flame. (Momentary
green flame, presence of thallium.)

Notes. — (i) The precipitation of thallium as Til is practically complete, provided
that the volume of the solution is small, say less than 15 cc. The presence of iron
does not prevent a good blank being obtained, nor does its presence in large quantity
(500 mg.) prevent the detection of 0.5 mg. of thallium. Til is readily oxidized to the
soluble Til, by iodine; therefore an excess of sulphite must be present.

(2) Any quantity of thallium in excess of 10-15 ™g- is precipitated in P. 11, except
when the solution there treated with HD contains the thallium in the thallic state.

(3) The green flame test is a very delicate and characteristic test, but, on account

5l8 A. A. NOYES, W. C. BRAY AND B. B. SPBAR.

of the volatility of the thallium oompound, the green color is seen only at the moment
in which the wire is introduced into the flame. A hardened filter is recommended
when the precipitate is small, because the precipitate can be more readily collected
on the wire on account of the smooth surface of the filter.

Procedure 66. — Into the ammoniacal solution (P. 64 or P. 65) pass
HjS gas until the mixture after shaking blackens lead acetate paper
held above it. (Black precipitate, presence of nickel or cobalt,) Filter,
and wash the precipitate with water containing a very little (NHJjS.
(Precipitate, P. 67; filtrate, P. 81.)

Notes. — (i) In precipitating NiS, the use of H^ has the advantage that the nidcd
is all thrown down at once, while with QiH^)^ some of it usually remains in the so-
lution, giving it a dark brown color. If fotmd more convenient, (tlR^)^ can of course
be used, the filtrate being boiled to throw down the unpredpitated nickel, as described
in P. 51.

(2) The filtrate is in general tested for the alkaline earth elements, for these may
be precipitated with the aluminum and iron groups when phosphate or certain other
add radicals are present, as discussed in P. 51 N. 6 and 7.

Procedure 67. — Transfer the HjS precipitate (P. 66) with the filter
to a casserole, and add 10-30 cc of a cold mixture of i volume HG
(1.12) and 5 volumes of water. Digest in the cold for five minutes,
stirring the mixture frequently, and filter.

Treat the residue by P. 68.

Boil the HCl solution until the HjS is completely expelled, add 10
per cent. NaOH solution until the mixture is slightly alkaline, transfer
to a casserole, cool, and add 0.5-1 gram Na^O, a small portion at a time.
Boil for several minutes to decompose the excess of Na^O, and cool the
mixture; filter off the precipitate, and treat it by P. 68, uniting with it
the residue already obtained in the HCl treatment. Make the solution
add with acetic add, warm it to about 60^, and pass in H,S for two or
three minutes. (White flocculent predpitate, presence of zinc.) Apply
to the predpitate the confirmatory test as described in P. 57.

Notes. — (i) This treatment with dilute HCl serves to extract almost completely
the zinc which may be present in this precipitate, owing to its having been carried
down in the Na^O, precipitate as described in P. 52, N. 11. A small proportion of
the nickel and cobalt present (5 to 20 per cent.) always dissolves in the dilute HD,
and the subsequent treatment with Na^O, serves to separate them from the zinc. This
separation is satisfactory when, as in this case, the nickel and cobalt are present in
such small quantity that only an insignificant quantity of zinc is carried down with
them. When, therefore, the H^S precipitate is small, it may, instead of being treated
with dilute HCl, be dissolved at once in aqua reqia and the solution treated directly
as described in the last paragraph of the procedure.

(2) This procedure must always be followed in order to determine whether or not
zinc is present in the substance, except in the case that a satisfactory test for it has
already been obtained in P. 57, or in the case that the original Na,0, precipitate (P.
52) was small.

Procedure 68. — Transfer the residue insoluble in dilute HCl, and th^

SYSTEM OF QUALITATIVE ANALYSIS. 519

Na^O, precipitate (P. 67), with the filters to a casserole, add 5-15 cc.
HCl (1.12) and a few drops HNOj (1.42), warm until the black pre-
cipitate is dissolved, and filter off the paper. Bvaporate the solution
nearly to dryness to expel most of the acid, add about 5 cc. water, and
then NaOH solution drop by drop tmtil the mixture is neutral, or until
a permanent precipitate just forms. Test one-half of this mixture for
cobalt by P. 69, and the remainder for nickel by P. 70.

Procedure 69. — ^To one-half of the neutral solution (P. 68) add 15 cc.
30 per cent, acetic add, and then 50 cc. 30 per cent. KNO3 solution;
dilute to 100 cc, and allow the mixture to stand at least half an hour if
no precipitate forms sooner. (Yellow, finely divided precipitate, pres-
ence of cobalt.) Filter, and wash with KNOj solution. If the pre-
cipitate is very small, indnemte the filter. Introduce a portion of the
predpitate, or of the ash, into a borax bead made in the loop of a plati-
num wire, and heat strongly, adding more of the predpitate or ash if no
color is obtained. (Deep blue color, presence of cobalt.)

Notes. — (i) The yellow precipitate is potassium cobalti-nitrite, KJCo(fiO^)^ which
in solution dissociates into K~^ and the complex anion Co(NO,)s^. The precipitate
is somewhat soluble in water but very difficultly soluble in a concentrated KNO, solu-
tion, owing to the common ion effect of the potassium ion. In the formation of this
substance the cobaltous salt is oxidized to the cobaltic state by the nitrous add dis-
placed from its salt by the acetic add, the cobaltic salt combining as fast as formed
with the potassium nitrite, according to the equations:

Co+^(NOj-), + 2HNO, = Co+++(NO,— ), -h H,0 + NO and
Co+++(NO -), + 3K+NO,- = KJ^o{NOX

(2) The formation of the KjuoiJtiO^)^ predpitate takes place slowly, but even when
very little cobalt is present (o.i to 0.2 mg.) a distinct test is obtained within ten min-
utes; but the complete predpitation of a large amount of cobalt requires several hours,
90 that the method is ill-adapted for the removal of cobalt before testing for nickel.
Moreover, when nickd is present, some of it is carried down with the cobalt, and this
is true even when the total amount of nickel is small. For these reasons the test for
nickd is made in a separate portion.

(3) Nickdous salts are not oxidized by nitrous add, and are not predpitated by
KNOj except in a very concentrated solution, when a dark ydlow to dark red pre-
dpitate of potassium nidcdous nitrite, K4Ni(N02)«, may separate. By making the
volume large, as directed in the procedure, there is no danger of the predpitation of
nidcd.

Procedure 70. — To the remainder of the neutral solution (P. 68) add
10 per cent. KCN solution, a few drops at a time, until all or nearly all
of any precipitate that may form at first redissolves; then add 0.5-3
cc more (according to the amount of the KCN precipitate). Heat to
50^ or 60® in an open casserole, with frequent stirring, for five minutes,
or longer if the solution has not become light-colored. Filter off and
reject any small precipitate that may remain. To the filtrate, pref-
eiably in a test-tube, add freshly prepared, concentrated NaBrO

520 A. A. NOYES, W. C. BRAY AND E. fl. SPEAR.

solution (see note 3) until a piece of filter paper moistened with
and starch solutions, when dipped into the solution, is colored blu
brown. Allow the mixture to stand five to ten minutes, and fi
(Brown to black precipitate, presence of nickel.)

Wash the precipitate ; dissolve a small portion of it, if it is latg
2-3 cc. HNO, (i-2o); add 3-5 cc. 10 per cent, tartaric acid solui
neutralize with 10 per cent. NaOH solution, and add 3 or 4 cc. 11
cess. Pass in H,S for about one minute, filter out any precipitate
may form and saturate the filtrate with H^. Filter again if there
precipitate. (Brown coloration, presence of nickel.)

Notes.^ii) The reactions involved in the first test for nickel are as follows: '
a little KCN is added to the neutral solution, precipitates of (gieeo) Ni(CN), and
brown) Co(CN), result except when only small amounts of these elements are pr
The addition of more KCN causes the precipitate to dissolve owing to the fom
of soluble complex cyanides, as K,+[Ni{CN),]= and Kj+ [Co (CN ),)==. The
plex nickel salt is stable in the air, but the cobalt salt is very readily oxidized ai
ing to the equation:

iKj+[Co(CN),]== + O + H,0 = 2K,+ [Co{CN),]= + jK+OH-
(Potassium cobaltocyanide) (Potasaum cobalticyanide)

The first action of the NaBrO is to decompose the excess of KCN, chiefly with forn
of KCNO. It then oxidizes the nickel to the nickelic state, whereupon the nickel
mediately precipitated as brownish-black Ni(OH), by the NaOH present. Tl
bait, though already in the cobaltic state, is not precipitated as Co(OH)„ bt
the complex ion, [Co(CN),]=, is so slightly ionized into its constituent-ions (D
and 6 CN~) that the concentration of the C0 + + + does not suffice to produo
the OH"" present the solubility-product for the Co(OH)„ and that of the CN~
small that it is only very slowly oxidized by the NaBrO.

(z) In executing this procedure the following precautions should be observei
very large excess of the strong KCN solution over that required to redissolve tb
dpitate should not be added, for the excess must be destroyed by the NaBtO I
the nickel can be oxidized and precipitated by it. Vet there must be sufGcieot
added, not only to combine with all the cobalt, but to furnish a moderate exo
order that the oxidation to the cobalticyanide may take place rapidly. Care
also be taken to heat the solution long enough in the air to complete this oxid
before the NaBrO is added, for otherwise the latter reagent after destroying tb
KCN will oxidize the decomposable cobaltocyanide with precipitation of Co('
just as it does the nickelocyanide; the completion of the oxidation by the air
dicated by the disappearance of the dark color in the solution. Finally one
make sure (by applying the iodide-starch test) that an excess of NaBrO ovo
required to oxidize both the cyanide and the nickel has been added. If thesi
cautions are observed, there is no difficulty in securing a precipitate with 0.1 v
nickel nor in cauang 300 tng. of cobalt to remain entirely in solution.

(3) The hypobromite reagent is prepared by adding liquid bromine to a k
volume of to per cent. NaOH solution until the solution becomes distinctly red, <
to the presence of excess of bromine; and then adding half as much more NaOH
tion. This solution may be filtered through a hardened filter. It decomposes
rapidly, with formation of bromate and bromide, and also with evolution of oi
and should therefore not be used when more than a few days old. Instead □

SYSTEM OF QUALITATIVE ANALYSIS. 52 1

reagent, bromine water and NaOH solution may of cotirse be used. The concentrated
hypobromite solution is recommended because a much smaller volume of it is required
to decompose the large excess of cyanide used in the procedure.

(4) When an alkaline tartrate solution containing a small amotmt of nickel (even
o.i to 0.2 mg. in 20 cc.) is saturated with HjS a clear brown solution is obtained.
With somewhat larger amounts of nickel (10 to 20 mg.) the liquid is opaque but runs
through a filter very readily. The condition of the nickel in this solution is not known.
The presence of the tartrate serves merely to prevent the precipitation of Ni(OH),
by the NaOH solution, owing to the formation of a complex salt containing the nickel
in the anion. The brown color does not appear tmtil the alkaline solution is nearly
saturated with H^S, so that care must be taken to use an excess of H^S.

(5) This confirmatory test for nickel is not interfered with by moderate amounts
of other elements of this group, such as cobalt and iron (which, however, should not
be present at this point), for on leading HjS into an alkaline tartrate solution contain-
ing these elements, they are completely precipitated as sulphides and may be filtered
o£f, yielding a filtrate which in the absence of nickel remains clear when saturated
with H^, or becomes dark brown when it is present in even small amounts.

Test Analyses.

Numerous analyses were made to test the efficiency of the process
above described. Nearly all of those which were made after the pro-
cedure assumed final form are reproduced in the tables below. Al-
most all the analyses relating to the common elements alone were
made before the procedures for the rare elements had been worked out,
but the process for the analysis of mixtures containing only the common
elements was at that time substantially the same as given above. In
these tables the numbers in each vertical column show the weights in
milligrams of the various elements which the solution submitted to
analysis contained. The results of the tests for each element are shown
by the letters following these numbers. That the result was satis-
factory is indicated by the letter S ; that is, when the element was present,
that the test for it, however small, was unmistakable, and therefore
conclusive ; and when the element was absent, that a good blank test was
obtained. When the test was very small, especially in comparison to
the quantity of the element present, though still unmistakable, this is
sometimes indicated by the symbol S-. When in the presence of the
element the test failed, or in its absence a result was obtained that
might be thought to indicate its presence, the letter F is used. When
the result was doubtful or inconclusive, owing to the appearance or
small size of the precipitate, this is indicated by the letter D. When
the test was not tried, if the element was present a dash is used in
place of a letter; or, if the element was not present, dots are inserted.

Common Elements.

The following analyses (Nos. 11 4-1 19) were begun at P. 51; P. 61
was introduced before P. 52.

A. NOYGS, W. C. BRAY AND E. B. SPEAR.

114. iij. lie. 117. 118-

,.ioS jS iS' sooS 500S

Sooi

Uyses 130-125 were begun at P. 61.

No. ■». III. 131.

500S 500S 100 S

dyses 125-127 were made by P. 69; analyses 128-130 by P.

L. Mo. 116. iiT. "S. lag. ijg.

> oS 0.5 S 150- 150- 150-

i 25t>- 250- oS 0.3S 0.15S

: following analyses (Nos. 131-143) were begun at P* 52.

iP-57.
1 P. 67-

IP.S7... JF
1P.67... ^[s
500 s

: following analyses (Nos. 144-155) were begun at P. 52 ; P. f
ed.

iP-57.-
I P. 67. .

Soo S

SYSTEM O^ QUALITATIVE ANALYSIS.

523

T. A. No.

Cr.

[ in p. 67

PO,.
Fe..
Co..
Ba..

150.

15».

152.

2 S

• ■ • •

0 s

3 _

f s

5 s

0 s

100 S

ic» S

■ • • «

100 S

icx) S

0 s

100 S

500 s

2 s

I F

• • • •

153.

{

I54-
2 S

• • • •

500 s

100 s

159.

I S"
100 s
i<x> -

Mixtures Containing the Rarer Elements.
Analyses 156-16 1 were begun at *P. 58a.

T. A. No. 156. 157. 158.

Cr o S o S 2 S

U 100 S o S 100 S

V oS 100 S 100-

Analyses 162-165 were begun at *P. 58/.

T. A. No. 162. 163.

Al 500 S 500 S

Be o S 2 S"

Analyses 166-170 were begun at *P. 58a.

T. A. No. 166. 167. 168.

£iSi. .... ....

Al 200- 2CK>-

Be . . . '. 50-

in *P. sSc-d. ... r F r S- f F

in *P. 58*

x6o.

• ■ ■ «

i(x> S
I S

164.

I s

o S

{- '{- '{-

i68a.

• • • •

• • ■ •

50 s

169.
200-

(

155

100

161.

• • • •

I s
100 s

165.

I s

ICX) S

170.
200 S

2<X) S

F
S

Analyses 171-176 were begun at P. 64; analysis 176 at P. 52.

T. A. No.

171.

Al.

Mn.

Fe.

Ti .

Zr.

Tl.

500 S
o S
o S
o S

172.

500 S

I s
I s

173-

• • • «

174.

175-

• • •

0.5 s
5 S

500 S 5cx> S

0.5 S

176.

2<X>-
2CX) S
2CX> S

I S

The following analyses (Nos. 177-18 1) were made by Mr. R. D. Gale,
who had had no previous experience in connection with the process.
The first analysis was with a known mixture, the remaining four were
with unknowns. Mr. Gale received no assistance while making an
analysis (except in T. A., No. 179), but all the precipitates and solutions
were preserved in order that the causes of mistakes might afterwards be
investigated. The process was practically in its final form with ex-
ception of the procedures relating to the separation of chromium, ura-
nium and vanadium (*P. 586-^), which were revised on account of the
results of these analyses; moreover, uranium was not tested for in *P.
58fc. Test analyses 156-16 1 and 166-170 were made at a later date
to test these procedures in their final form.

A. NOYES, W. C. BRAY AND E.

T. A. No.

tl P. S7. .
I1P.57..

loln.—(\) The prcKDce at aluminum in the letgcnU mndc the rtsult doubtful.

2) A distinct pT«i|ritdtc of Al(OH)i *niH obtained in P. S3, but not more tban with tlir in

3) Such of HQ wu added in P. S3 and the Hlution boiled, u a renilt of which the ct
1 wai reduced. On analyiing another mixture, and IviDi HNO), a latisfartoiv test toi ehic

i) The abaeace of manj^neae was Lonfirmed by fusing some of the iron salt on platinum foi
fXh and KNO): no (rcen color vai obtained,

I) A amuU prerlpitatc of ZnS was obUined in F, S7, lAich did not contain more than i m
C. B.,P. 52. N. II,

S> A aniall predtntate of ZoS vma obtained in P. S7, conesprnding to Z or 3 ing. Zn.

7} In thii analyaia a larse scceia of alkali nae added before the NaiO, in P. S2. and the 1:

B) Thi» uulytia wm reimted Fic?rpt that the acid aolution in P. 42 was poured skiwlv '1
eu of hot alkali befoie adding NajOi: a distinct t«t for line was then obtained in P. S7.
9} A distinct test for calcium vtai obtained (except powbly in T. A. No. 148); the Inni a

10) This analysia was repeated except that no Na,CO] was added in P. S2: the tnt lor chic
ed,

II) SeeC. E.,*P. 58*. N. I.

12) The predidtatc of Be(OH)t obtwhied in *F. 58* was luice and cncrespended to 2 ««. B<
wln£ that much amallef amounts coutd be easily detected,

13) The basic acetate procedure (P. 65) was used Instead of P. M: good tests for thalliuD
oined both in the precipitate and in the filtrate. (Compare C. E., P. 65, N. 6.)

14) Chramium was lost in •?. 58n, owing to the pmcDce of HiO, (from the peruniiste.
P. 32. N. 6 and *P. Sab. N. 2) and the addition of a large exceu of acid.

15) A shght blue coloration was obtained with 11^ in *P. SSI: corresponding to not mor
ing. Cx. No lead nitrate was added.

16) The directions in *P. 3Sc weie faulty: too little phosphate was added to nedpitatc I

17) A good teat for aluminum was obtained in *P. 58|i, but the confirmatory test (P. M)
IBJ The test for iron in P, 65 was a failure: the iron waB found with the manganese in P
l?) Satisfactory tests for Iron were obtained in P, 62 and In P. 64.

20) (NHi)^ wa^ used instead of H2S in precipitating the sulphides in P. 66, and some nidi
: in the brown solution. Nickel and cobalt were also tost in P. 67 in the tiestment with
I, owing to the rejection of the NajO, predpitote,

21) Only B few miUigrams of vanadium were found in the Rltrate from P. 51.

22) About twice as much zinc was found in P. 5V as in P. 67.

23) Owing to the inulubiUty of Zr(OH)PO. it is scarcely pomihle fo< lirconiom and phospl
ur together in the analysis of the aluminum and iron groups. The zirconium was undoi
: bl dissolving the predintates obtained in P. 32-65.

Although all possible combinations of small and targe quantities ol

SYSTEM 0]P QUALITATIVE ANALYSIS.. . 525

•

elements have not been investigated, the test analyses are sufficiently
varied to justify general conclusions as to the reliability of the method of
separation and detection of the elements.

An examination of these analyses shows that i mg. of any of the seven
common elements in various combinations with one another and in the
presence of the rarer elements was detected in almost all cases, even when
the element was associated with a large quantity of other elements. Zinc
frequently escaped detection in the analysis of the aluminum group (in
P. 57), but was then always found in the analysis of the iron group
^ P. 67) when tested for (see T. A. 131-153, 178, 180). One mg. is, how-
ever, near the limit of detectabiUty when the elements of the iron group
are present in large quantity. Two mg. of chromium were lost in two
cases (T. A. 118, 181) owing to the reduction of the chromate; these anal-
yses served to emphasize the necessity of added precautions, which have
since been provided for in the directions. One mg. of iron in the presence
of a large quantity of manganese would have been overlooked (in T. A.
180) if it had not been tested for in the solution of the chloric acid pre-
cipitate, as directed in P. 62. In the case of aluminum, owing to its be-
ing contained in the reagent, it was sometimes difficult (see T. A. 116, 117,
119, 131-136, 144-149) to determine with certainty whether or not i mg.
was present

It will be seen from T. A. 142-151 and 180 that the presence of much
phosphate does not interfere with the detection of i mg. of any of the
common elements of these groups. It does, however, cause a. larger
quantity of zinc to be completely precipitated by sodium peroxide (com-
pare T. A. 141 with 142 and 143). It is shown by 144-146, 150, 151 and
180 that 2 mg. of barium and i mg. of magnesium are detected in the
process, showing that adequate provision is made for the recovery of the
alkaline-earth elements when carried down with these groups, through
the presence of phosphate.

The analyses with the rarer elements (T. A. 156 to 181) show that the
process led to the detection of i mg. of each of these and even of 0.5 mg.
of beryllium and titanium in all cases except the following: In T. A. 180
zirconium was missed, owing to the simultaneous presence of much phos-
phate; but, even in the absence of phosphate, i mg. is not far from the
Hmit of detectability in a complete analysis. In T. A. 166, 168, 170 and
178 uranium was missed in the usual place (in *P.58c and d) owing to the
fact that it was carried down with aluminum or beryllium in the treat-
ment with dilute NaHCOg (*P. 58a), but T. A. 168 and 170 and a number
of special experiments with mixtures of beryllium and uranium show that
in such cases uranium can be detected in *P. s^h.

J6 A, A. KOYBS, W. C. BRAY AND E. B. SPBAR.

CONmuiATORV EXPEUUBHTS AND RKFSKSNCBS.
G. D., Section i: PrtcipHation of Zinc with Chromium on adding NH,OH.
g. Zn as ZnO, and loo mg. Cr as KCr(SO.), were dissolved in 4 cc Ha (
id 96 cc, water; 10 cc. NH^OH (0.96) were added a little at a time to the
ixture (giving an excess of about 3 cc. NH,OH); after 3 or 4 minutes the pn
te was filtered off, the filtrate, which had a very faint [Hnk color was heated [0
g; and the precipitate that separated (estimated to contain i or 3 mg. Cr) wa
red oS. H^ was finally passed into the alkaline filtrate: no precipitate sepor
.us showing that the rinc had been completely carried down with the Ct(OH)^
'ccipitates were united and dissolved in HCl. and the solution treated by P. ;
■; the dnc was precipitated as ZnS in P. 57.

For the preparation and properties of ZnCr.Oj, see Chancel, Compt rmd., 43
856): Viard, Bu:i. de la 10c. Mm. (3), a, 331 (1889).

Action of NHJDH and (NH,)^ on Nickel Solalions, and Precipitation of Ni
jiiing Ihe Bravn Soluiion.—See C. B., F. 51, N. 12.
The Test for Vanadium with H^.—See C. E., P. 510, N. i.
The Slow Reduction of Vanadic Acid to Hypovanadic Acid by H^. — 50 mg.
,V0, were dissolved in 4 cc. HG and 100 cc. water; the mixture was salui
ith H,S in the cold. After about 10 mintues the mixture was filtered several 1

remove the sulphur that had separated, and the blue filtrate was heated to
g: a large precipitate of sulphur separated, showing that reduction was moie

higher temperatures. The mixture was filtered and the hot filtrate again
ted with H,S; a large precipitate of sulphur separated. After 10 minutes the
ire was boiled, filtered and again saturated with H^: sulphur separated, sbc
at the reduction was still incomplete. The treatment of the hot solution witl
ia repeated, the sulphur filtered off, the mixture was boiled in a casserole to
^, and I g. KI was added: no iodine was liberated, showing that the reductii
'povanadic acid was complete.^The first experiments were repeated widi
NO, (1.30) in 100 cc: the results were the same.

Reducttan of Vanadic Acid to Hypcmanadic Acid by HCl (r.Jo). — See Goocb
utis. Am. J. Set. (4} 17, 41 (1904). The reaction is slow in the cold but lap
ating. When a moderate amount of vanadium is present, the reduction L
>solutely complete imless a more concentrated HCl solution is used than that obt
1 evaporating. Compare C, E, 'P. 58d, N. i.

Acli^ of NHfiH on a Hypovanadate.^ioo mg, V as Na,VO, were treated iritl
r P. 21 to reduce it to the hypovanadic state; the solution was evaporated to t
1 cc. and made alkaline with NH^OH: a dirty gray precipitate separated
g. V as Na,VO, were boiled with concentrated HO (r.20) to reduce the vanat
le mixture was diluted to 100 cc. and made alkaline with NH,OH; a small
edpitate, with a greenish shade, separated. Several cc. NH,OH were ai
e precipitate dissolved.

Precipitation of Vanadium in P. 51 or by NHfiH alone when Iron or Zinc is Pt
,500 mg. Fe as FeSO, and 100 mg. V as Na,VO, were treated with H^ by I
'aporated to 40 cc. and treated by P. 51 ; the filtrate was tested for vanadiu
', 51a: less than i mg. V was found.— The experiment was repeated with jo»
; and 10 mg. V: no vanadium was found in the filtrate.
500 mg. Fe as FeCI, and 100 mg. V as Na,VO, were dissolved in 40 cc. conta
cc, HCl (1,12), the solution was made alkaline with NH,OH, and (without ai
IH,)^) the mixture was filtered, and the filtrate was tested for vanadium b:
a; only a trace of vanadium was found, — The experiment was repeated will
g. Fe as FeSO, and 100 mg. V as Na,VO,; the mixture was allowed to stanc

SYSTEM Olf QUALITATIVB ANAI.YSIS. 527

shaken frequently in order that the iron might be completely precipitated: no vanadium
was found in the filtrate. — ^50 mg. Fe as FeCl,, 15 mg. V as hypovanadic add and
several grams NH4CI were dissolved in a little HQ and 100 cc. water; the solution
was made alkaline with NH^OH, cmd the filtrate was evaporated to dryness, ignited,
dissolved in a little NQ and tested for vanadium by P. 51a: only a trace of vanadium
was found. — ^Therefore vanadium is almost completely carried down when a three-
to fivefold excess of iron is present.

300 mg. Zn as ZnCl, and 10 mg. V as H,V04 were treated with HjS by P. 21, evap-
orated to 30 cc. and treated by P. 51; the ^trate was tested for vanadium by *P.
51a: 3 or 4 mg. V were found. The ZnS precipitate was analyzed by the regular
procedure: the remainder of the vanadium was found.

Reduction of Vanadic Acid by HI and its Subsequent Behavior with the Group Rea-
gents.— According to Gooch and Curtis {Am. /. Science (4) 17, 45, 1904), the first
stage of the reduction, that to V3O4, took place rapidly in a dilute HI solution, but
the second stage, that to VjO,, takes place slowly and only in hot concentrated solu-
tions. When a mixture containing 60 mg. VaO„ i g. KI and an excess of HCl was
evaporated to 2 cc., about 97 per cent, of the vanadium was reduced to a salt cor-
responding to VgO,.

In our experiments i mg. V as Na,V04 was dissolved in 15 cc. HCl (1.20) in a
50 cc. round bottom flask; i g. NHJ was added, and the mixture was evaporated
carefully to 2 or 3 cc., a capillary ebullator tube being used to prevent btunping.
The mixture was diluted with 10 cc. water, and NH^OH (0.96) was added tmtil the
solution after shaking smdled distinctly of it: a dark green predpitate formed. A
few drops colorless ammonium sulphide were added: the predpitate did not dissolve.
It was filtered off and dissolved in a little HNO,; a large excess (5 to 10 cc.) of NH4OH
(0.90) was added and the mixture satturated with H^: the solution became red, show-
ing the presence of vanadium. The filtrate from the NH4OH predpitate was tested
for vanadium in the same way: only a very small amount was found. — The experi-
ment was repeated, except that the evaporated mixture was diluted to 20 cc : more
vanadium was found in the filtrate than in the predpitate. — The experiment was re-
peated, the mixture being diluted to 40 cc.: all the vanadium was found in the fil-
trate.

The experiment was repeated several times with 100 mg. V, the mixttu-e being di-
luted to 20 cc. after the evaporation: on adding NH^OH a dark colored (brown)
predpitate separated, but it dissolved completdy or in large part on adding a few drops
of colorless (NH^)^.

The experiment was repeated with 50 mg. V; after the first evaporation 15 cc.
HQ (1.20) and i g. NHJ were added, and the evaporation repeated; the mixture
was diluted to 50 cc: a large dark greenish colored predpitate separated on the ad-
dition of NH4OH, and it did not dissolve on adding several drops ($lli^^. The fil-
trate was evaporated nearly to dryness; an excess of 5 cc. NH^OH (0.90) was added,
and the mixture saturated with HjS: not more than 0.5 mg. V was foimd in the fil-
trate. The last experiment was repeated with 5 and with 10 mg. V: in the experi-
ment with 5 mg. only a very slight predpitate was obtained on adding NH4OH and
it quickly redissolved; with 10 mg. a predpitate was obtained but about half the
vanadium was fotmd in the filtrate. Therefore in a volume of 50 cc. the predpita-
tion is fairly satisfactory for large amotmts of vanaditun but not for small amounts.
The above results were not perfectly reprodudble, but each of them was obtained
several times.

A number of attempts were made to reduce 50 to 100 mg. V as Na,V04 by evapo-
rating as described above with xo to 15 cc. ptu-e concentrated HI, and to predpitate

i28 A. A. NOYES, W. C. BRAY AND E. B. SPBAS.

he vanadium with NH^OH after diluting to 40 ot 50 «- 1 but the piedpitatioi
nuch less complete than after the reduction with NH,I and HCl, probaUy c
o the larger proportion of iodine found to be retained in the solution in tbe ft
ase.— The experiments were repeated with HI that had been exposed to the si
ontained much iodine: the precipitation was still less complete.

Rale 0} OxidaHon of Trivaleni Vanaiiium in tiie Presence of NHfiH, and of {Nl
—50 mg, V was reduced twice by evaporation with 15 cc. HCl (i.ao) and 1 g.
o I or 3 cc. ; the residue was diluted to 50 cc, and made alkaline with Nf
rhe mixture was divided into two parts; one was filtered at once and the second
lalf an hour; both filtrates were tested for vanadium by "P. 510: a very poo
ras obtained in the first filtrate, and a good one in the second. The experimcD
epeated, except that the second portion was filtered after 3 hours: (be first S
;ave a slight test for vanadium; the second contained z to 4 mg. V. Thereto]
ixidation takes place slowly in the presence of NH,OH.

The experiments were repeated except that several drops colorless (NH,)^
Jbo added: the filtrates obtained at once, after 1/3 hoiu, and after 3 hours gavi
light tests for vanadium. Therefore the addition of a little (NH,)^ does not
he vanadium hydroxide to dissolve, and the rate of oxidation is slower in the
noe of (NHJ^ than in that of NH.OH alone.

Ibe oxidation, by the oxygen of the air of the trivalent vanadium after di
» therefore much too slow to account for the abnormal results, such as the :
nmplete precipitation of 50 to 100 mg. V in one experiment, and the non-prei
ion in another performed under apparently the same conditions.

G. D., Section a: Action oj HCl on NiS or CoS. — Herz (Z. aaorg. Chem., i'
ind 38, 343, 1901) treated sulphides which had been kept on filters for several
ifter precipitation with 0.5 normal HCl (i vol. HCl (i.u) and 13 vols, water
ound that the evolution of H^ could not be detected by odor or by action on
noistened with a lead salt. On the other hand, he found H^ to be evolved
upidly by the action of 0.5 normal HCl on freshly precipitated sulphides. Hi
Judes that there are two allotropic forms of the sulphides, one that is soluble
lormal HCl, and one that is not. As GlixelU (Z. anorg. Chem., 55, 197, 190;
Minted out, this evidence of decreased solubility is not very conclusve, for t
LCtion with the old sulphides may simply be much slower. The following expai
ihow that NiS, at any rate while freshly precipitated, is not a mixttue of two sul;
liffeiing markedly in solubility,

300 mg. Ni as NiS, precipitated in the cold by passing H^ into NH,OH sol
vas digested in the cold for a known time with about 30 cc. dilute HCl (1 vol
;i.i3) and 5 vol. water). The HO was first saturated with H^, and H^ was 1
Itfough the mixture in a sm^ll Bask during the treatment. The mixttue was fi
md the experiment repeated several times with the NiS residue. The amoi
lickel dissolved in each treatment was estimated by adding NH^OH and predpi
Jie nickel as NiS by passing in H^. In the first treatment of 15 min. about :
Ai dissolved; in each of the next four treatments of 15 min. each, steadily deer
(mounts of Ni dissolved, and in the 5th treatment the amoimt was estimated
[O mg. Ni. The 6th and 7th treatments were each for half an hour: somewhal
lickel dissolved than in the 5th treatment of 15 minutes. The 6th treatmei
or 4 hours: nearly as much dissolved as in the first treatment (say 18 mg.),
light, without passing the current of H^S, about 20 mg. Ni dissolved. In a last
nent for 6 hours, 35 or 30 mg, dissolved. The final residue contained only
)o rog. Ni. — A similar series of experiments was performed, except that the Hi
lot saturated with H^ and no H^ was passed through the mixture: the result

SYSTEM OF QUALITATIVE ANALYSIS. 529

sinukr, except that the NiS dissolved more slowly owing to the fact that the mix-
ture was very little stirred.

100 mg. Ni as NiS (precipitated by H,S in NH4OH solution) were digested with 30
cc dilute HCl (i vol. HD (1.12) and 3 vols, of water) for 5 minutes: about 5 mg.
Ni dissolved. — The experiment was repeated with HCl (1.12): the amount of Ni that
dissolved was not much greater than in the preceding experiment. — The last experi-
ment was repeated with i mg. Ni: only a small proportion of the nickel dissolved.
AcHon of HCl on NiS when Other Sulphides are also Present, — 500 mg. Fe as FeSO
and 5 mg. Ni as NiSOf were dissolved in 50 cc. water containing 2 cc. HJSO^ (1-84)1
and the solution treated at about 40^ with NH4OH and (NH^)^ as described in P.
51: the filtrate was clear and light colored^ showing that it contained no nickel. The
precipitate was dried by suction, and treated in a casserole with 30 cc. cold dilute
HQ (i vol. HCl (1.12) and 5 vols, water), with frequent stirring, for 5 minutes: only
a very small black residue remained, which was found to contain about 0.5 mg. Fe
but no nickel. — ^The experiment was repeated, except that ferric chloride was used
instead of ferrous sulphate: the residue contained i or 2 mg. Fe and a mere trace of
nickel. — Both experiments were repeated, except that the sulphides were precipitated
at about 90^ by passing H^S into NH4OH solutions. In the experiment with ferrous
iron the residue insoluble in dilute HCl contained nearly i mg. Fe and less than i
mg. Ni; in the experiment vnth ferric iron the residue contained about i mg. Fe and
nearly all the nickel. — The last experiment was repeated with 500 mg. Fe as PeCl,
and 2 mg. Ni as NiSO^: the residue insoluble in cold dilute HG contained a little iron
and nearly all the nickel. — The first of the above experiments was repeated with 500
mg. Fe as FeS04 and 10 mg. Ni as NiSOf; the residue insoluble in dilute HCl and also
the filtrate were tested for iron and nickel : the residue contained about i mg. Fe and
5 mg. Ni; the HD solution contained also about 5 mg. Ni. — The experiment was re-
peated except that the sulphides were precipitated at about 90^ by passing H^ into
an NH4OH solution: the residue insoluble in dilute HD contained about 0.5 mg. Fe
and 3 or 4 mg. Ni.

500 mg. Mn as MnS04 and 5 mg. Ni as NiS04 were dissolved in 2 cc. H,S04 (1.84)
and 50 cc. water, and the solution was treated with NH4OH and (NH4),S as described
in P. 51; the filtrate was dear, showing the absence of nickel. The precipitate was
digested with 30 cc. cold dilute HD (i vol. HD (1.12) and 5 vols, water), with frequent
stirring, for 5 minutes: the small dark colored residue on analysis was estimated to
contain 3 or 4 mg. Ni; the HCl solution contained only about 0.5 mg. Ni.

The experiment was repeated with 500 mg. Zn as ZnS04 and 2 mg. Ni as NiSO :
after 10 minutes treatment with the dilute HCl a very large residue remained; the
HCl solution contained only about 100 mg. Zn, and no Ni. The 2nd treatment with
HCl lasted over night: about 300 mg. Zn and no Ni dissolved. The third treatment
lasted two days: the residue was small and dark colored, and contained over i mg.
Ni and little or no Zn; the HCl solution contained about 100 mg. Zn and a httle Ni.
The experiment was repeated with HCl (1.12): a black residue remained.

These experiments show that small amounts of NiS may be dissolved completely
when distributed throughout a large precipitate of iron sulphide, but that they are
not readily dissolved in the presence of ZnS or MnS. They also show that ZnS dis-
solves 8k)wly in dilute HD just as NiS does.

G. D., Section 6: PrecipUaUon of Vanadium by Alkaline Hydroxides when Uranium
is Prisenl. — ^For the insolubility of uranyl vanadate, see v. Klecki, Z. anorg, Chem.,
5, 381 (1894); Camot, Compt, rend., 104, 1850 (1887); also C. E. *P. 58a, N. 3.

50 mg. V as NaaV04 and about 250 mg. U as U02(N0,), were dissolved in 3 or 4
oc HNO, (1.20) and 30 cc. water; the solution was made alkaline with NH4OH (0.96):

530 A. A. NOYBS, W. C. BRAY AND E. B. SPBAR.

a large yellow precipitate separated. This was filtered ofiF and the solution was tested
for vanadium by *P. 58e: not more than i mg. was found. In this experiment there
was just sufficient uranium to form UO^H^VO^. The precipitate obtained in the
preceding experiment was dissolved in a little HNO,, diluted to about 15 cc. and ponied
into about 20 cc. 10 per cent. NaOH solution: a large yellow precipitate separated.
This was filtered off and the filtrate tested for vanadium by *P.58«: only 5 to 10 mg.
V w^e found.

Separation of Zinc and Vanadium by Hydrogen Sulphide. — HjS was passed into
a solution containing H^VOf and acetic acid: the solution slowly became blue and
sulphur separated, which rapidly became dark colored owing to the precipitation of

some VA-

I mg. V as Na,V04 and 50 mg. Zn as Zn(NO,), were dissolved in 30 cc, a little acetic
add was added, and an excess of about 3 cc. NH^OH (0.90). The dear solution was
saturated with H^. A heavy predpitate of ZnS separated at once, and the solution
slowly became deep red in color, which color was very distinct after filtering. — ^The
experiment was repeated with i mg. Zn and 50 mg. V: a predpitate of ZnS was ob-
tained, which was confirmed by the last paragraph of P. 57.

Action of NHfiH on Uranium Solutions containing Hfi^ — 100 mg. U as UO,(N0,),
were treated by P. 52. The red (peruranate) solution was made add with HNO^
Without heating, the add solution (which contained H,0^ see C. E., P. 52, N. 6) was
made alkaline with NH4OH: a large yellow predpitate of (NH^),!!,©, separated, but
the filtrate was yellow. The filtrate was addified with HNO„ boiled for two or three
minutes, and again made alkaline with NH4OH: a large ydlow predpitate separated,
which was nearly as large as the first NH^OH predpitate. The colorless filtrate was
again addified, boiled, and then made alkaline with NH4OH: no predpitate sepa-
rated. Therefore turanium, if present, will divide in P. 53.

To a nitric add solution containing 200 mg. U as UOjCNO,), and 40 mg. V as H,V04
in 40 cc. were added about 20 cc. 3 per cent. H^O,; the mixture was made alka-
line withNH^OH: no predpitate separated, showing that the predpitation of
UO,NH4V04 and of (NH 4)5,11,07 is prevented by the presence of an excess of H,0^

Action of BaCl^ (^ Uranate Solutions. — 100 mg. U as UOjCNO,), were treated by
P. 52 and 53. The yellow filtrate from the NH4OH predpitate, which contained about
half the uranium, was treated by P. 55: a moderatdy large predpitate separated.
After standing over night the predpitate was filtered off and the filtrate tested for
uranium by adding HNO„ boiling and adding NH4OH: a small yellow predpitate
separated, which was estimated to contain 2 mg. U. Therefore barium uranate (or
peruranate) may be predpitated nearly completdy in P. 55.

Action of BaClt on Vanadate Solutions. — 100 mg. V as Na3V04 were treated by P.
52 and 53. See C. E., P. 52, N. 10. The NH4OH solution, which had a volume of
about 40 cc., was made just add with acetic add and BaCl, solution was added in the
cold as described in P. 55 : no predpitate separated from the yellow solution in 5 min-
utes.— The experiment was repeated with 300 mg. V: no predpitate appeared when
the BaCl, was first added, but in 10 minutes a large orange predpitate of barium vana-
date had separated. This was filtered off and the filtrate allowed to stand over night:
a large predpitate formed. The predpitate was collected on a filter and ?rashed
with water: much of it dissolved, showing that barium vanadate is fairly soluble in
water. A portion of the filtrate which was still slightly ydlow was tested for vana-
dium by *P. 58e: much vanadium was fotmd, showing that the predpitation of barium
vanadates take place slowly, and that large amotmts of vanadium remain in the fil-
trate in the cold. — A solution containing 100 mg. V as Na3V04 and a little NH4OH
in 40 cc. was made add with acetic add; 10 cc. 10 per cent. BaCl, solution were added,

SYSTEM O^ QUALITATIVE ANALYSIS. 531

and the mixture was boiled for a minute or two: a large nearly white precipitate sepa-
rated, and the solution became perfectly colorless. The precipitate was filtered off
and the filtrate tested for vanadium by '*'P. sSe: the filtrate was found to contain
not more than a few milligrams of vanadium, thus showing that a large proportion
of the vanadium had been precipitated on boiling. — ^The last experiment was repeated
except that the acetic add solution was made alkaline with NH4OH before boiling:
the filtrate was found to be nearly free from vanadium. — The last experiment was
repeated with a mixture of 100 mg. V and 10 mg. Zn as nitrate: only about 3 mg. Zn
were found in the filtrate, showing that a large part of it had been precipitated with
the vanadium, probably as zinc vanadate.

Action of NHJDH on SoluUons containing Zinc and Vanadium as Vanadate, — 500 mg.
Zn as nitrate and 100 mg. V as NajVO^ were treated by P. 52 : only a very small pre-
cipitate remained and the solution was colorless. The filtrate was acidified with HNO,
(1.20), an excess of about 2 cc. being added; and the solution was made just alkaline
withNH40H: a large precipitate separated. 4 cc. NH4OH (0.96) were added: the
precipitate dissolved completely. The NH4OH solution was neutralized with acetic
add: a large precipitate separated but it dissolved when a small excess of acetic add
was added. The acetic add solution was boiled for a minute or two: a large predpi-
tate separated leaving a dear solution, which, however still contained considerable
zmc and vanadium.

Separation of Zinc and Uranium by NHJDH. — i mg. Zn as Zn(NO,), and 100 mg.
U as UO,(NO,), were dissolved in 4 cc. HNO, (1.20) and 100 cc. water; the mixture
was made alkaline with NH4OH and an excess of about 3 cc. added ; the yellow predpi-
tate was filtered off and the solution tested for zinc with H^: a small nearly white
ptedpitate separated. — The experiment was repeated with 50 mg. Zn and 100 mg.
U: the filtrate contained nearly all the zinc. The predpitate was washed, dissolved
in HNO„ diluted to 100 cc. and treated again with NH4OH : the filtrate contained only
about 2 mg. Zn. The NH4OH predpitate was again treated in the same way: the
filtrate contained no zinc.

Action of NHfiH on Solutions containing Uranium and Chromate. — ^4 mg. Cr as
ILpxO^ and 100 mg. U as U0,(N0,)2 were treated as described in the preceding para-
graph: the filtrate from the NH4OH predpitate contained nearly all the chromium;
the second NH4OH filtrate was colorless. — ^The experiment was repeated with 50 mg.
Cr and 100 mg. U: the filtrate after the second NH4OH predpitation was nearly color-
less, thus showing that little or no chromium was carried down with the uranium.
G. D., Section 13: The Separation of Titanium and Zirconium by Boiling with Sul-
pkurous Acid. — ^With regard to the unsatisfactory nattu-e of the process usually em-
pbyed, see Hillebrand, Bull, U. S. Geol. Sur., 176, 72-3 (1900). A fairly satisfactory
quantitative separation of titanium from iron is obtained by boiling a solution of a
volume of about one liter containing acetic and sulphuric adds. This large volume
is impracticable in qualitative analysis, and H. I. Wood and B. £. Schlesinger (Theses
M. I. T. 1901, 1902) devised the following method: "Heat the HCl solution (volume
10 to 20 cc) to boiling, remove the flame, add at once carefully NH4OH until litmus
paper is turned distinctly blue; add 30 to 40 cc. R^Oj solution, and place on the water-
bath for 15 to 20 minutes." This method was carefully tested by us. The results
were satisfactory with titanium or zirconium alone, and small quantities of these de-
ments were easily detected. But when iron was also present, some of it was carried
down. In an experiment with 3 mg. Fe, 100 mg. Ti and 100 mg. Zr as chlorides, no
test for iron was obtained in the filtrate, thus showing that this amount of iron had
been completely carried down. Also when 500 mg. alone (as PeCl,) of Pe were pres-
ent, tome of the Pe(OH), predpitated by NH4OH did not dissolve in the H^, solo-

32 A. A. NOYES, W. C. BRAY AND E. B. SPEAR.

on; and evea when ferrous iron was used, some Fe(OH),
uring the heating on the waterbath. Moreover, the predpitation of titaniin
rconium was usually incomplete when much of these elements were present,
ipedally if the solution was not made distinctly alkaline with NH,OH. This a

therefore a very unsatisfactory one.

G. D.. Section 17: Precipitalvm 0} BeryUiuM Phosphate as a ConfitmaUiTy 1
ee B. E. Schlesinger, Thesis, M, I. T. 190s; Classen, Ausgewahlte Meth. d. fi
hemie, 5th Ed., i, 713-

1 mg. Be as BeCI, was in each of three experiments dissolved in water cont
little HCl ; 1 cc. 10 per cent, citric add and 2 cc. 10 per cent, ammonium pha

>lutions were added, the mixture was made strongly alkaline with ammonia, and
oiled gently for about 5 minutes in a small flask : a small white crystalline prec
sparated. — The experiment was twice repeated with 2 mg. Al as AlCl,: no prec
:parated. — The experiment was rejjeated with 0.5 mg. Be: no predpitale n
I experiments). Therefore this confirmatory test is hardly delicate enough,
ally since beryllium has a very low atomic weight.

Terf for Zirconiunt with Turmeric Paper. — The following procedure recomn
y B. E. Schlednger (Thesis, M. I. T., 1902) was tested: "Dissolve the pho
redpitate in HF (i: 5), evaporate to one or two drops, and dip a piece of tn
aper into it; heat the paper at 100° until dry." (Pink color, presence of Hrco
i agreement with the results of Schlesinger, i mg. was found to be the limit
stability. Even with 3 mg. the test was not a striking one. Moreover, til
iso caused a pink coloration of the turmeric paper, and the test was more d
lan in the case of zirconium. The blank test with strong HF was not perfectly
ictory. The results obtained when HCl, instead of HF, was used were pnu

P. SI, N. 1: PiKipitation of Small Anumnis oj Various Elements by Amnu
5 and 1 mg. of Al, Be, U, and Zr as nitrate, of Fe as FeSO,, of Or as Cr,(SI
i as TiCl, were dissolved in separate experiments in 4 cc. HQ (1.12} and 30 cc
1 small conical flasks; the solutions wrere neutralized with NH^OH (0.96) and
cc excess added ; the mixtures were shaken several times and then allowed tc
few minutes: in each case the solution was transparent after the NH,OH was
ut a distinct Bocculent predpitate was observed after the shaking and sti
he predpitates were more difficult to see with chromium, aluminiun, and ur
lan with the other substances, but even in these cases 0.5 mg. was easily de
The experiments with altuninum, uramum, and chromium were repeated in a '
: 100 cc. containing 4 cc. HCl (1.12): i mg. Al was near the hmit of detect
: a cold solution, but 0.5 mg. could be easily detected if the solution were
early to boiling and allowed to stand. With uranium the limit of detectatnli
to 3 mg. in a cold solution, and with chromium it was 2 mg. in a cold solutit
ith 0.5 mg. Cr the solution became turbid on heating the NH,OH solution t

ig-

2 mg. Cr as KCr(SO.), were dissolved in 4 cc. HCl (i.ii) and a little water; th
on was boiled, diluted to 100 cc, cooled, and made alkaline with NH.OH

very small predpitate was seen after shaking and standing. This was filter
id the nearly colorless filtrate was heated to boiling: a findy divided, light
redpitate then separated, showing that predpitation of Cr(OH), by NH^OH
>ld is incomplete.

SohlHlHy of Zinc and Nickelous Hydroxides in Solutions containing NHti
H,OH. — To a number of solutions containing 500 mg. Zn as ZnCl, and va
uounts of HCl in 30 ca of solution, known amounts of NH^OH (0.96) were grs

SYSTEM OF QUALITATIVE ANALYSIS. 533

added, 1.0 cc. HCl (1.12) being equivalent to about 1.5 cc. NH4OH (0.96). The fol-
lowing table shows the amount of NH4OH that had to be added (i) before a large
precipitate was formed, (2) before a distinct odor of NH4OH persisted after shaking,
and (3) before the precipitate dissolved to give a colorless solution.

NH4OH (0.96) required

HO (1.12)
present.

to give ft
large precipitate.

to gire a
distinct odor.

to dissolve
the precipitate.

2 CC.

3CC.

8 CC.

II CC.

4 cc.

7 CC.

II CC.

14 cc.

6 cc.

II cc.

14 cc.

16 cc.

8cc.

no precipitate.

17 cc.

• • . .

The experiment was repeated in a volume of 100 cc. containing 4 cc. HCl (1.12): the
result was practically the same as in the corresponding experiment with an initial
volume of 30 cc.

The experiments were repeated with 500 mg. Ni as NiClj, the initial volume being
30 cc : the solution became blue when an excess of NH^OH was added, but no precipi-
tate separated even when only 2 cc. HCl were used. In an experiment with no add,
a large green precipitate resulted, which dissolved on adding about 6 cc. excess NH4OH
after the odor of NH^OH was persistent.

Precipitation of Cobalt Hydroxide. — The foregoing experiments were repeated with
500 mg. C0CI3 and 4 cc HCl (1.12) in 30 cc: when the NH4OH was added slowly a
moderately large green precipitate was always obtained, which did not dissolve on
adding a large excess of NH^OH; the supernatant solution was reddish brown. The
experiments were repeated with 4 and 6 cc. HCl (1.12), the ammonia (3-4 cc. excess)
being added all at once: no precipitate was obtained, but the solution weis red colored,
and darkened on standing owing to oxidation. The experiments were repeated in
a volume of 100 cc containing 4 cc HCl (1.12): the results were the same, except
that the amount of green precipitate formed was greater than in the smaller volume
and that a larger excess of NH4OH was necessary, in order to prevent the separation
of a precipitate when the excess of NH4OH was added all at once.

Solubility of Aluminum Hydroxide in Excess of Ammonia. — To a solution contain-
ing 10 mg. Al as nitrate and 4 cc HCl (1.12) in 100 cc were added in the cold 17 cc
NH4OH (0.96); t. «., an excess of about 10 cc. and the mixttu-e was shaken; the pre-
cipitate of Al(OH), was filtered off after about five minutes, and the filtrate was boiled
for several minutes to expel the excess of NH^OH: a precipitate of Al(OH), separated
which was estimated to contain about 3 mg. Al. . The experiment was repeated, ex-
cept that an excess of only 2 or 3 cc NH^OH (0.96) was used: the filtrate on boiling
became turbid, but the amount of aluminum that separated did not exceed 0.5 mg.

P. 51, N. 3 and 4: SolubUity of Mg{OH\ in NH^Cl Solutions.— The solubility of
MgCOH), in water is about 2 X lo""^ mols. per liter. For evidence that the non-pre-
dpitation of MgCOH), in moderately dilute solutions containing an ammonium salt
depends solely on the driving back of the dissociation of NH^OH, see Lovdn, Z. anorg,
Ckem., II, 404 (1896); Treadwell, Z. anorg. Chem.y 37, 326 (1904); Herz and Muhs,
Z. anorg. Chem., 38, 138 (1904).

The following experiments show that enough ammonium salt is produced by the
neutralization of the 4 cc HCl (1.12) originally present to prevent the precipitation
of Mg(OH), even when 500 mg. are present. A solution containing 500 mg. as MgCl,
and 4 cc HCl (1.12) in 100 cc was neutralized with NH4OH (0.96) (6 cc), and more
NH4OH was added. No precipitate formed when 40 cc in all had been added, but
the addition of 10 cc more gave a precipitate. — The experiment was repeated with
an initial voltmie of 30 cc instead of 100 cc. : 30 cc. NH4OH produced no precipitate,

A, A, NOYES, W. C. BRAY AND B. B. SFBAK.

even after several minutes, but 40 cc, did. — The experiment yias repeated iritli
HCl and a volume of 30 cc.: 15 cc NH,OH (0.96) produced no preci[Htate, btit
did.

CompUx ZitK Ammonia Co(ftio».— Zn(NH,),+ •■. See He«, Z. ani>rg. Cke\
315 (i9oo);Gau9, Z. anorg. Chem., a;, 136 (1900)1 Buler, Ber,, 36, 3400 (1903);
dorff, Z. anorg. Chem., 41, 132 ('904).

Comf/Ux Nickel Ammonia Calhion.—m(,tili,\^-'-. See Dawson and McO
Chem. Soc., 77, 1239 {1900); KonowalofI, Chem. CenlraWIaa, 1900 I, 646; anc
dally Bonsdorff, Z. anorg, Chem., 41, 13; (1904).

Behavior of Chromium lowards Ammonium Hydroxide. — To a solution com
in 30 cc. 3 or 4 cc. HCl (i . 1 1) and 350 nig. Cr as CrCI, freshly prepared by boiling
with concentrated HCl, was added NH,OH (0,96) until after shaking the od
distinct. The mixture was divided into two parts; one part was filtered al
to the other part 10 cc. NM.OH (0.96) were added, and it was immediately £
the former had only a very faint pink color, and on boiling not more than c
Cr precipitated as Cr{OH),; the latter was distinctly pink, and on boiling 1 to
Cr precipitated as Cr(OH),.— The experiment was repeated, except that an
of 2 cc. NH,OH was added after the odor of NH.OH was distinct, and that u
tion 5 g. solid NH,C1 were added to one-half of the mixture, and both portioi
filtered at once: each filtrate was faintly pink, and on boiling that containing
cesa of NH,C1 there resulted a precipitate of Ct(OH), estimated to contain (
Cr, while on boiling the other filtrate no precipitate separated, and the solnl
mained pink.— The last experiment was repeated, except that the mixtures *
lowed to stand 20 hours before filtering: the filtrate containing no excess of
was faintly pink but did not contain more than i mg. Cr. Tlie second filtra
highly colored and on boiling 5 to 10 mg. Cr precipitated as Cr(OH),, but the ]
tation was not quite complete. These results show that both NH,OH and
increase the amoimt of the pink chromium compound formed, and therefon
that this compound is a complex ammonia salt and not a chromite. For tb
that these solutions arc unstable, even at room temperatures, see Hen, Z.
Chem., 3r, 357 {1901).

P. S'l N- <i' PrecipHation of Phosphates of Calcium, Barium, Magnesium and .
nese by Ammonium Hydroxide. — To solutions containing 2 mg. PO, as Na^P
4 cc, HCl C1.12) in 30 cc., were added in separate experiments varying amo
manganese, barium, and magnesium, as chloride and of calcium as nitrati
solutions were then neutralized with NH,OH (0.96) and about 2 cc. excess
making S cc. in all, after which the mixtures were shaken and allowed to si
and 3 mg. Mn gave small precipitates after a minute or two, but i mg. gave :
cipitate; 10 and 10 mg. Ca gave large precipitates at once but 5 mg. gave no ]
tate even in half an hour. 100 mg. Ba gave a large precipitate at once, and
gave no precipitate in half an hour, i, 3 and 5 mg. Mg. gave precipitates on si
—rTbe experiment was repeated with 10 mg. PO, and varying amounts of bariu
mg. Ba gave a large precipitate at once, but 40 mg. gave no precipitate, or
very small one, in half an hour. — The experiment was repeated with 1 mg. Ca an
ing amounts of phosphate: with 30 mg. PO^ a precipitate was obtained at one
70 mg. after several minutes, and with 10 mg. no precipitate appeared in half a
On repeating the experiment with i mg. Ca a precipitate was obtained with
PO, after a few minutes but not with 20 mg.

P- 5', N. 7: SdubaUy of the Borates of the Albaiine Earth EUmtnt
mg. BO, as H,BO, and 30 mg. Ca as chloride were dissolved in 4 cc. HCl (i.i
36 cc. water, and NH,OH (096) were added until the mixture after sfaaki

. SYSTKM OF QUALITATIVE ANALYSIS. 535

sxnelledof it: no precipitate separated. — The experiment was repeated with 200 mg.
Ba as chloride; no precipitate separated. — ^The experiment was repeated with 300 mg
Ba; a small precipitate formed, but it dissolved on adding a little concentrated NH4CI
solution. — ^The experiment was repeated with 500 mg. Mg as chloride: no precipitate
resulted.

P. s^, N. g: Action of Ammonium Hydroxide on Uranyl Solutions. — See Kern. /.
Am Chem. Soc., 33, 701-5 (1901). — i and 3 mg. U as UOj(NO,), were dissolved in
sqjarate experiments in 20 cc. water, and a few drops NH4OH (0.96) were added:
no precipitate separated on standing, nor on boiling. A small quantity of solid
NaNO, was added to the cold solutions: pale yellow precipitates separated at once,
showing that the uranium had been present in the colloidal form. — The experiment
with I mg. U was repeated, except that a little NH^Cl instead of NaNO, was
added: a yellow precipitate separated.

P. SJf N. 9: Uranyl Salts and Diuranates. — See Dittrick, Z, physik. Chem., 29,
449-90 (1899); Kern, /. Am, Chem. Soc, 23, 686-726 (1901).

P. $j, N. 10: Completeness of Precipitation of the Sulphides of the Iron Group. — In
each of the following experiments the quantity of the element given below was dis-
solved in 4 cc HCl (1.12) and 96 cc. water; to this solution in the cold were added
9-10 cc. NH4OH (0.96), and then (NH4),S solution drop by drop until an excess was
present. The mixttires were well shaken and then filtered, generally through a double
filter. The results were as follows: With 5 mg. Mn as MnCl,, a light colored pre-
ctpitate was formed at once; the filtrate was dear after two filtrations. With 0.5
mg. Mntis MnCl,, the result was the same. With 0.25 mg. Mn as MnCl,^ the solution
became turbid in two or three minutes. The filtrate in the first two experiments
was evaporated almost to dryness; HNO3 (1.42) was added; the mixture was evapo-
rated to 5 cc., KCIO, was added, and the mixture boiled: no precipitate of MnO, sepa-
rated, showing that the precipitation of MnS had been complete.

With I mg. Zn as Zn(NO,)2, ^^^ mixture became turbid at once, and was readily
filtered. With 0.5 mg. Zn, the result was the same. The filtrate in the first experi-
ment was evaporated to 5 cc., made alkaline with NH4OH and (NH4)2S was added :
only an insignificant turbidity appeared, showing that the precipitation of ZnS was
practically complete.

With 0.5 mg. Co as CoCNOj),, a black precipitate formed at once, which was readily
filtered ofif. With o.i mg. Co, the solution became dark colored at once.

With 0.2 mg. Fe as PeS04, the solution became dark colored at once and the pre-
cipitate was easily filtered off. With 0.5 mg. Fe as F6SO4, the result was the same.
The filtrate was evaporated to 5 cc. ; NH4OH and (NH4)jS were added : the solution
remained colorless.

With 0.3, 0.5 and i.o mg. Tl as T1,S04, the solutions became dark brown at once
and yielded clear filtrates, which when evaporated almost to dryness and tested for
thallium by *P. 65^, gave no precipitate of Til.

With 0.5 and i.o mg. U as U02(N03)2, distinct flocculent precipitates could be seen
after the addition of (NH4)2S; the filtrates were clear, and the precipitate, when col-
lected on the filter, was yellow in color. With 0.3 mg. U as U02(N08)2, a slight color-
ation was seen on the filter, but the precipitate could scarcely be seen before filtration.
With 5 mg. U as U02(N03)2, the result was the same. The filtrate was evaporated
almost to dryness, and was tested for lu-anium by *P. 58^ with K4Fe(CN)6: no brown
coloration was observed. The last experiment was repeated except that the filtrate
was evaporated to 5 cc. and NH4OH and (NH4)2S added: no precipitate formed, show-
ing that the precipitation as sulphide in 100 cc. had been complete.
Nature of the Uranium Precipitate Produced by Ammonium, Sulphide. — 20 mg. U

536 A. A. NOYES, W. C. BRAY AND E. B. SPEAR.

as UO,(NO,), were dissolved in 4 cc. HCl (1.12) and 96 cc. water, about 9 cc NH/)H
were added and (N 114)28 drop by drop in the cold, until a distinct excess was present.
Tlie precipitate was flocculent and had a bright yellow color. After two hours it was
light brown and in six hours dark brown in color. — ^The experiment was repeated
except that a much larger excess of (NH4),S was added: the precipitate darkened
more rapidly. It is evident that (NH4),U,07 is first precipitated contrary to the state-
ments in some text books, and that the conversion of this substance into 110,8 takes
place slowly.

The Existence of Ferric Sulphide (Fe^,).— That Fe^Sj is formed when (NHJ^ is
added to suspension of Pe(OH), in dilute NH^OH has been proved by Stokes, /. Awl
Chem. Soc.f 29, 304 (1907). When, however, an add solution is first saturated with
H^ and then made alkaline with NH4OH, FeS and not Fe,S, is formed.

P, 51, N. 11: Dark Color of Sulphide Precipitate or Evidence of Iron, Nickel, or Co-
balt.— To solutions containing 500 mg. Zn as nitrate, 4 cc. HCl (1.12) and 30 cc. H,0
were added in separate experiments i and 2 mg. Ni as nitrate, i and 2 mg. Co as ni-
trate, I mg. Fe as FeS04 and i mg. Fe as FeCl,. The mixtures were neutralized with
NH4OH (0.96) and 3 or 4 cc. excess added, making 10 cc. in all : A large white pre-
cipitate of Zn(0H)2 remained which was estimated to contain at least 100 mg. Zn.
Colorless (NH4)2S was then added as described in P. 51 and the color of the precipi-
tates was compared with that obtained with 500 mg. pure zinc; the precipitates con-
taining I mg. Fe were nearly black vdth a greenish tinge; those containing 2 mg. Ni
and Co were much darker than the pure ZnS, but those with i mg. Ni and Co were
only very slightly darker. — The experiments were repeated except that 4 cc. more
NH4OH were added in order to dissolve the Zn(OH),: the results were substantially
the same.- -A solution containing 500 mg. Ca and i mg. Ni as nitrates, i g. PO4 as
(NHJjHPO^, 4 cc. HCl (1.12) and 35 cc. H,0 was made alkaline with NH^OH (0.96)
and I cc. colorless (NH4),S was added: the large white precipitate of phosphate be-
came dark colored as soon as the sulphide was added.

P. sif N. 12: Behavior of Nickel towards Ammonium Monosulphide and Polystd-
phide. — See Lecrenier, Chem. Ztg., 13, 431, 449 (1889); Anthony and Magri, Gaa.
chim. ital.f 31, II, 265 (1901). By boiling the brown solution i^ the absence of air
the last-named authors have prepared NiS^. The composition of the brown solution
is unknown; it may be ammonium sulpho-nickelate, (NH4)2NiS| 4. xi or colloidal
nickel persulphide.

That in absence of air nickel is completely precipitated as NiS and the filtrate is
colorless, thus proving that the brown solution is due to the presence of polysulphide,
has been shown by Lecrenier, Chem.-Ztg., 13, 431, 449 (1889) and ViUiers, Compt.
rend., 119, 1263 (1894). The presence of NH4OH tends to prevent the precipitation
of NiS, and it is possible to obtain colorless (or bluish) solutions, containing a small
excess both of (NH4)2S and of nickel. In the presence of air we have found it alnxist
impossible, when working with large amounts of a pure nickel salt, to prevent some
nickel from passing into the filtrate, giving a brown solution. The amotmt of nidcd
in the filtrate increased with the excess of ammonium sulphide used and with the
length of exposure to the air.

Some of the brown solutions obtained in these experiments were boiled in small
fiasks for 3 to 10 minutes, and then filtered: in each case the filtrate ¥ras colorless after
a single filtration.

Precipitation of Nickel in Ammoniacal Solution by Hydrogen Sulphide. — ^In several
experiments 500 mg. Ni as NiCNO,), were dissolved in 30 cc. water and 4 cc. HG (i.u);
NH4OH (0.96) was added tmtil the odor could be detected after shaking and then
3 cc. more, and H,S was led into the mixture for 15 minutes: the precipitates were al-

SYSTEM OF QUALITATIVE ANALYSIS. 537

ways granulaTp and the filtrates clear and colorless; when the precipitates were washed
with water containing either RgS or a little (NHJ^S the wash water was invariably
dear.

♦P. ^la, N. i: Test for Vanadate with H^S in Alkaline SoltUion. — To separate solu-
tions containng 3 g. NH4CI and 5 cc. NH4OH (0.90) in a volume of 1 10 cc. were added
0.1,0.3 and 0.5 mg. V as vanadate, and the mixtures were saturated with H3S: in the
experiments with 0.3 and 0.5 mg. V the solutions quickly became dark yellow, then
reddish yellow and finally pink; in that with o.i mg. V the solution finally became
faintly pink, but the test was a poor one. Therefore, the limit of detectability under
these conditions is o.i to 0.2 mg.

The experiment with 0.5 rag. V was repeated except that, instead of 5 cc. NH^OH
(0.90) I, 2.5 and 10 cc. respectively were used: with i cc. and with 2.5 cc. NH4OH
the pink color was scarcely noticeable; with 10 cc. a good color was obtained as with
5 cc. NH4OH, but only after passing in H^ for a longer time. These experiments
show that a fairly large excess of NH4OH is necessary.

20 mg. V as Na3V04 were added to 20 cc. NH4OH (0.96) and HjS passed in: the color
qidckly became brown and slowly turned red, becoming after 5 or 6 minutes a deep
dicrry-red. The solution was divided into two parts. To the first was added an equal
volume of water: the red color disappeared in 2 or 3 minutes, but appeared again
on resaturating with H^S. To the second part of the red solution was added an equal
volume of NH4OH (0.96) : the color faded slowly (in 4 or 5 minutes) but reappeared
on passing in H^S. These experiments show that the solution must be saturated
^th HjS.

1 mg. V as NajVO^ was dissolved (a) in 20 cc. NH4OH (0.90), (b) in 20 cc. NH4OH
(0.96), (c) in 5 cc. NH4OH (0.96) and 15 cc. water, (d) in 2 cc. NH4OH (0.96) and 38
cc. water, (e) in 2 cc. NH4OH (0.96) and 98 cc. water. Each solution was saturated
with H^: a good color was obtained in every case, showing that in the absence of
ammonium salts the test may be obtained in any concentration of NH4OH.

A solution containing i mg. V as Na3V04 and 3 cc. HNO3 (1.42) in 10 cc. was neu-
tralized with NH4OH (0.90) and i or 2 cc. in excess added : on saturating with H^S a
very faint color was obtained. — The experiment was repeated, except that the acid
solution was first evaporated to dryness and ignited: a very good test was obtained,
showing that the presence of ammonium salts interferes with test for vanadium.

0.5 mg. V as Na,V04 was added to some HCl (1.20) and the mixture was evapo-
rated twice nearly to dryness to reduce the vanadic acid to hypovanadic acid ; 4 cc.
HG (1.12) were added and 96 cc. water; the mixture was neutralized with NH4OH
(0.90), an excess of 5 cc. was added, and H^S passed in for 10 to 15 minutes: the solu^
tion darkened quickly, became reddish yellow and finally pink ; the color was exactly
the same as in the experiment described above with 0.5 mg. V as Na3V04. — The ex-
periment was repeated with 25 mg. V, which was reduced to hypovanadic add by
kmg continued treatment of the hot solution with HjS: the solution darkened and
then became deep red very quickly; the color was the same as in the experiment
described above with 50 mg. V as Na8V04.

I mg. V as Na3V04 was dissolved in 18 cc. water, 2 cc. 10 per cent. NaOH solution
were added, and the mixture saturated with HjS: a deep red color was finally obtained
which was almost the same as that obtained when NH4OH was used. The experi-
ment was repeated with 20 cc. undiluted NaOH solution: the final color was the same,
bat the solution remained colorless for a long time while the HjS was being passed
through it. Therefore the formation of the red compound does not depend on the
presence of NH4OH or NH4+ ion.

♦P. $ia, N, 2: Action of Acids on Sulphovanadale. — 0.5 mg. V as sodium vanadate

53^ A. A. NOYES, W. C. BRAY AND E. B. SPEAR.

was dissolved in loo cc. containing about 3 g. NH4CI ; 5 cc. NH4OH (0.90) were added,
and the mixture saturated with H^S; the pink solution was filtered, and then acidi-
fied in the cold with acetic acid, stirred, and filtered: a small black precipitate contain-
ing sulphur was obtained. This was dissolved by boiling with a little HNO, (1.20);
the solutions were evaporated to about 2 cc. and i to 2 cc. 3 per cent. H,0, were added:
an orange-yellow color resulted, showing the presence of a small amount of vanadium.
— ^The experiment was repeated with sodium hypovanadate: the results were the same

The experiment was repeated with 25 mg. V: on boiling the filtrate from the precipi-
tate of vanadium sulphide it became blue and sulphur separated. To it 50 mg. Fe as
FeCl, were added and then an excess of NH4OH ; the mixture was filtered, the filtrate
was evaporated to dcyaess, the re^due ignited and dissolved in a very little HNO„
and a few drops H^O, solution were added: no color appeared, showing that the vana-
dium in excess of o.i mg. had been carried down with the Fe(OH),. The sulphide
and hydroxide precipitates were dissolved separately in HNO„ H3O, was added to-
gether with and 20 to 30 oc water, and the intensities of the colors compared:
it was estimated that about 10 mg. V were precipitated as sulphide, and the
remaining 15 mg. V with the Fe(OH),. — The experiment was repeated with 50
mg. V, a larger amount of FeCl, being used: the results were similar, thus showing
that tmder these conditions less than half the vanadium is precipitated as sulphide.
— In other experiments in which the volume was only 20 to 30 cc., and the relative pro-
portion of ammonium salt to NH^OH was smaller, 80 to 90 per cent, of the vanadium
was fotmd to be precipitated as sulphide by acetic add. Even when NaOH was used
instead of NH4OH, it was not possible to precipitate the sulphide completely.

The experiment with 50 mg. V described in the last paragraph was repeated, except
that HCl was used to acidify the cold solution: it was estimated that roughly one-half
of the vanadium was precipitated as stdphide and the other half with the Fe(OH)|.
— The last experiment was repeated, except that the mixture containing HCl was boiled
for about i minute: somewhat less than half the vanadium was fotmd in the HCl pre-
cipitate.

*P. 5ia^ N. j: Complete Precipitation of Vanadium by Ammonium Hydroxide in
the Presence of Iron. — See preceding section, and C. E., G. D., Section i.

*P. 51a, N. 4: Pervanadic Acid. — ^See Scheuer, Z. anorg. Chem., 16, 284 (1898);
Pissarjewsky, Z. phys. Chem., 43, 171 and 173 (1903), and 40, 368 (1902).

Action of Hfi^ on Vanadic Acid. — In a series of 6 test tubes, each of which contained
I mg. V as Na,V04, were placed o, 0.5, 2, 5, 10, and 20 cc. H,S04 (1.20); each solu-
tion was diluted to 20 cc. and 2 cc. 3 per cent. HjO, added: no change was observed
in the tube with no add; the solution became orange to orange-red in the remaining
tubes, and the intensity of the color increased greatly as the concentration of the add
increased. The exj)eriments were repeated with HNO,, (1.20): the results were the
same. All the tubes were allowed to stand over night: no changes in the colors were
observed. Ether was added to a number of the solutions and the mixtures shaken:
the ether layer remained colorless.

To solutions containing i mg. V as Na,V04 dissolved in i cc. HNO, (1.20) were added,
(a) I cc. 3 per cent. HjOj and 8 cc. water, (6) 3 cc. H.Oj and 6 cc. water, and (c) 9
cc. HaO^: the first two solutions were orange-red, but the third solution was practically
colorless, thus showing that a large excess of HjO, spoils the test. To the third solu-
tion were added 3 cc. HNOj (1.20): the orange-red color reappeared, thus proving
again that the test is more easily obtained in the presence of a large excess of add.
To this solution was then added 7 cc. HJO,: the color became much fainter.

The first series of experiments with varying amounts of HNO, (1.20) and a total
volume of 20 cc. was repeated, except that o.i mg. V as NajVO^ was used instead

SYSTEM OP QUAI^ITATIVE ANALYSIS. 539

of I mg.: all the solutions remained colorless, showing that o.i mg. V cannot be de-
tected in 20 cc — This series of experiments was repeated with 0.3 mg. V: a distinct
yellow color could be seen on looking down the tubes in the experiments with 5, 10,
and 20 cc. HNO^ a light color with 2 cc. HNO„ and none with 0.5 cc. — This series
was repeated with 0.5 mg. V: with 0.5 cc. HNO3, the solution was colorless, with 2
oc it was slightly yellow, and in the remaining experiments a slight but distinct orange-
yellow color was observed. Therefore the limit of detectability of vanadium with
H,0, in HNOg solution is about 0.5 mg. V in 20 cc, and the solution must contain
at least i volume HNO, (1.20) to 3 volumes of water.

♦P. 5ra, N. 5.* Tungstic Acid Dissolves in Acids when Phosphate or Arsenate is Pres-
ent.— Tech. Quart., 16, 122 (1903).

Partial Precipitation of Tungsten on Acidifying a Solution of Sulphotungstate. —
See Tech. Quart,, 17, 253-5 (1904)-

AcHon of H^ on Tungstate Dissolved in Excess of NHfiH. — 50 mg. W as (N}i^)^0^
and 3 g. NH4CI were dissolved in 100 cc. water, 5 cc. NH4OH (0.90) were added, and
the mixture was saturated with H^S: the solutions remained nearly colorless for over
10 minutes, but finally became slightly yellow, probably owing to the formation of
polysnlphide.

Action of Hfi^ on Tungstic and Molyhdic Acids. — See Tech. Quart,, 17, 251 (1904).
— 20 mg. Mo as (NH4)^o04 were dissolved in 5 cc. HNOj (1.20) and 2 cc. HjO, added:
a lemon-yellow color resulted. On account of the absence of an orange tint there is
no difficulty in distinguishing between this color and that of pervanadic add

Action of H^ on Molyhdate Dissolved in Excess of NHJDH. — A solution contain-
ing 6 mg. Mo as (NH4),Mo04 and 4 cc. HCl (1.12) in 30 cc. was treated by P. 51: no
precipitate formed with NH^OH, nor with (NH4)aS. To the nearly colorless filtrate
were added 5 cc. NH^OH (0.90), and HjS was led through the solution in a test tube
for 20 minutes: after 3 minutes the solution was dark yellow, after 5 minutes it had
a reddish color, after 10 minutes it was brilliant red, and the color did not change
in the next 10 minutes. The color was not the same as in the case of vanaditun, but
might be mistaken for it if a comparative test were not made. — 50 mg. Mo as
(NH4)^o04 and 3 g. NH4CI were dissolved in 100 cc. water, 5 cc. NH^OH (0.90) were
added, and the solution was saturated with H^S: the colors obtained were the same
as in the preceding experiment. — ^The last experiment was repeated with i mg. Mo:
the solution finally became deep orange in color.

Action of Acids on Sulphomolybdate. — 50 mg. Mo as (NH4)2Mo04, 3 g. NH4CI, and
5 oc. NH4OH (0.90) were dissolved in 100 cc. water, and saturated with HgS. The
red solution was filtered and then made distinctly acid with acetic acid: the color
remained nearly the same; on filtering, a very small black precipitate was obtained
which contained less than i mg. Mo. The solution was allowed to stand for one hour:
it remained clear. After 3 hours a small precipitate had separated which contained
only 2 or 3 mg. Mo. The mixture was then boiled for 5 minutes and filtered: about
2/3 of the molybdenum precipitated as M0S3 and the filtrate was still deep orange.
The filtrate was evaporated to a small volume and 10 cc. HCl (1.20) added: a large
black precipitate of M0S3 separated, but the solution was found to still contain i or
2 mg. Mo.

The foregoing experiment was repeated, except that HGl was added to the cold solu-
tion: a large black precipitate separated at once. The filtrate was evaporated to 30
cc and tested for Mo by *P. 43^: none was found. — ^The experiment was repeated
except that the mixtture was boiled after adding HCl: the result was the same, not
more than a trace of molybdenum being found in the filtrate.
P. 52, N. j: The Formation of Sulphate on Dissolving NiS in Acids. — In two experi-

540 A. A. NOYBS, W. C. BRAY AND E. B. SPEAR.

ments 300 mg. Ni as freshly precipitated NiS were treated by P. 52, the sulphur was
filtered off, and BaCl, was added to the filtrate: small precipitates of BaSO^ separated,
each of which was estimated to contain 5-10 mg. Ba. — In another series of experi-
ments the NiS was first treated in the cold for several minutes with mixtures of 5 cc.
HNO, (1.20) with 20 to 20 cc. water and the residue was dissolved by evaporatiog
the solution to concentrate the HNO,, and then adding a few drops HCl (1.20): p^^
cipitates of BaSO^ resulted, estimated to contain 20-50 mg. Ba. In each of these
experiments and especially when the precipitate was first treated with HNO, consid-
erable sulphur separated and the NiS enclosed in it did not dissolve readily.

P. 52, N. 4: Action of HCl on Titaniun^ and Zirconium Hydroxides. — Add solutions
containing 100 mg. Ti and 100 mg. Zr as chlorides in volumes of about 100 cc. were
treated in the cold with NH^OH; the precipitates were filtered off and treated with
20 cc. cold HCl (1.12): they dissolved completely in a minute or two. — The experi-
ments were repeated, except that the solutions were heated to boiling, and the NH4OH
was added to the hot solutions: on pouring 20 cc. HCl (1.12) repeatedly through the
filters containing the hydroxide precipitates, almost all of the TiO(OH), dissolved,
but only a small portion of the ZrOCOH), (8-10 mg. Zr). On pouring a 20 cc. por-
tion of hot HCl (1.12) repeatedly through the filters, the remainder of the titanium,
but only about half the zirconium dissolved. The remainder of the ZrO{OH), was
boiled in a casserole with HCl (1.12) for several minutes: it dissolved complctdy.

P. 32f N. 5: Complete Precipitation of Titanium and Zirconium in the Sodiunt Perox-
ide Procedure. — i mg. Ti as TiCl4 was treated with NaOH and Na^Oj by P. 52, the
mixture being boiled for about 4 minutes after the Na^O, was added: a white precipi-
tate separated. The filtrate was made acid with HCl; one-half of it was tested for
titanium by adding HjO,: no color appeared, showing that the titanium had been
completely precipitated. The other half of the filtrate was tested for HjO, by adding
excess of TiCl4: no color appeared, showing that the Na^O, had been completely de-
composed.— The experiment was repeated with 500 mg. Ti: the filtrate contained
I or 2 mg. Ti. The precipitate was treated with HCl (1.12): nearly all dissolved in
the cold and the remainder on warming. The solution was reddish yellow, showing
that the precipitate contained some TiO,. — The HCl solution obtained in the last
experiment (containing nearly 500 mg. Ti) was treated with NaOH and Na,0, by
P. 52; the mixture contaiiung Na202 was boiled for less than i minute: the filtrate
contaned at least 5 mg. Ti. — These experiments show that the amotmt of titaniom
that remains in the filtrate may be greatly lessened by long continued boiling.

I mg. Zr as ZrCl^ was treated with NaOH and NajO, by P. 52: a distinct precipi-
tate separated. — The experiment was repeated with 100 mg. Zr; the large precipitate
was filtered off, and the filtrate was tested for zirconium by acidifying with HNO„
evaporating almost to dryness, and adding Na^HPO^ solution: only a trifling precipi-
tate sepacrated, thus proving that the precipitation of the zirconium hsA been com-
plete.

Action of NaOH on Uranyl Salts. — A solution containing 5 mg. U as UOjCl,
and a little HCl in 15 cc. was neutralized with 10 per cent. NaOH solution,
and an excess of 2 or 3 cc. was added: a small yellowish precipitate separated. This
was filtered off and the filtrate was tested for uranium by acidifying and adding NH4OH:
a small precipitate separated, which was estimated to contain 2 or 3 mg. U. The
experiment was repeated, except that the mixture containing NaOH was boiled for
a minute or two: nearly all of the precipitate which had formed in the cold was re-
dissolved. Therefore the uranium is not completely precipitated by a small excess
of NaOH. — 50 mg. U as U02(N08)2 vrere added to 30 cc. 10 per cent. NaOH solution,
and the mixture was heated to boiling, cooled, and poured through a hardened fUter;

SYSTEM OI^ gUALlTATIVE ANALYSIS. 54 1

half of the filtrate was tested for uranium as above: only a small precipitate of
(NH4),U,0, separated, showing that the uranium had been nearly completely pre-
cipitated by the alkali.

Composition of Sodium AluminaiCf Zincaie, Beryllate and Chromite. — For the evi-
dence that sodium aluminate in solution has the formula, Na ' AlOj*", see Noyes and
Whitney, Z. physik. Chem,, 15, 694 (1894); Hantzsch, Z. anorg. Chem., 30, 296 (1902).
Hantzsch has shown by means of conductivity measurements that a solution of the
empirical composition Na,A10, contains mainly NaOH and NaAlO, and some col-
loidal A1(0H),.

Hantzsch, Z. anorg. Chem.^ 30, 298, 303 (1902), concluded from conductivity meas-
urements that a 1/200 molal solution of sodium zincate, even in the presence of a
sevenfold excess of NaOH, is almost completely hydrolyzed into NaOH and Zn(0H)3
and that the latter is present as a colloid. Fisher and Herz, Z. anorg. Chem., 31,
355 (1902), confirmed this by dialysis experiments. Kunchert, Z. anorg. Chem., 41,
343-8 (1904), working with a larger excess of alkali, and using Bodlander's electro-
motive force method, proved, however, that the solutions contained chiefly ZnOa"^ and
someHZnO,"". Forster (Z. Elektrochem., 6, 301, 1899) has prepared solid NaHZnO,,.

Hantzsch (Loc. ct/.) concludes that HjBeO^ is a very weak acid, weaker than HAIO^,
but much stronger than H^ZnO,.

From dialysis and conductivity experiments Fisher and Herz (Loc. cit.) conclude
that in alkaline chromite solutions chromium is present almost solely as colloidal
Cr(OH),.

Behavior of Cobalt Hydroxide towards NaOH. — See Donath, Z. analyt. Chem., 40,
*37 (1901). — ^An add solution containing 100 mg. Co as nitrate in 30 cc. was neutral-
ized with 10 per cent. NaOH and an excess of 15 cc. added; 2 g. Na^O, were added a
little at a time to the cold solution, the mixture was boiled for two or three minutes,
cooled, and filtered: the filtrate had a deep blue color. It was acidified, made alka-
line with NH4OH, and HjS was passed into it: a precipitate separated which was esti-
mated to contain 5 to 8 mg. Co. — The experiment was repeated, except that no ex-
cess of NaOH was added: the filtrate was colorless and no cobalt was found in it. — ^This
last experiment was repeated except that 4.5 g. Na^O, were used instead of 2 g. : again
the filtrate contained no cobalt. Therefore the blue (soluble) cobalt compound is
formed only by the action of concentrated alkali on a cobaltons salt, and the cobalt
is completely precipitated if it is first oxidized to the cobaltic state by Na^O, in a
weakly alkaline solution.

The first experiment with the large excess of NaOH was repeated with 100 mg. Ni
instead of Co: no nickel was found in the filtrate.

P. 52, N. 6: Formation of Peruranates. See P. Melikow and L. Pissarschewsky,
Z. physik. chem., 28, 556 (1899). A large number of salts are known in the solid state;
t. g.. U0^.2Na,0^8H,0; U04.2BaO,.8H,0; U0^.2Ba02.ioH20;etc. The salts of the
alkali elements are soluble in water, the others insoluble. These salts may be regarded
as compounds of pertu'anic anhydride, UO4, with peroxides of the other elements.
They were prepared by the action of HgO, on solutions of uranyl salts in the presence
of the hydroxides of the other elements.

A HNO, solution containing 100 mg. U as UOaCNOa)^ in 25 cc. was treated by P.
52: the solution was yellow when acid, and a yellow precipitate separated on adding
NaOH; on adding Na^Oj (3 g.) and boiling, a deep red solution resulted. This solu-
tion was acidified with HNO3: it became yellow. To a portion of it was added a little
Tid^ in HCl solution : an orange-red color resulted, proving the presence of free HjO,.
Therefore the uranium is oxidized by Na^Oa to a soluble "peruranate," which is stable
in the strong alkali, but apparently decomposes readily in the acid solution with forma-

542 A. A. MOVES, W. C. BRAY AND E. B. SPBAR.

ion of H,0, and a uranyl salt. — The expetinient was repeated except that only
''5 S- 1^3,0, was added: a yellow uranium precipitate remained, but the solutic
:ained a large proportion of the uranium.

P. 52, N. 7: Complete Decomposilion of Nafi, by Boiling. — 2 g. Na,0, were
;o a neutral solution containing 0.5 g. NaCl in 30 cc., the mixture was heated 1
ng, boiled for one minute, cooled, acidified with HNO„ and tested with TiQ, s
or H,0,; tlie solution remained colorless. — The experiment was repeated wi
■ng. Mn as MnCl,, the MnO, being filtered off before the filtrate was made add:
:ained no HjOt — The experiment was repeated with 100 mg. V as Na,VO,: thi
was the same. See the preceding paragraph as to the effect of uranium

Precipitalion of Thallium by Sodium Peroxide, — i, 3, and ao mg. Tl as T
were treated in separate experiments by P. 52: brown precipitates separated :
:ase on adding NaOH, and remained after the treatment with Na,Or The so.
»hich were filtered off almost immediately after diluting, were found to contaii
1.5, 1, and 1 to 3 mg. respectively. — 1 mg. Tl as Tl^, was treated by P. 52:
dpitate .separated on adding NaOH; there was a shght coloration on adding
lUt no precipitate remained upon boiling; when, however, the solution was st
a cool, a good precipitate separated in 10 or 15 minutes, and the filtrate wa:
.0 contain only 0.2 to 0.4 mg. Tl.— A mixture containing i mg. Tl as TlCl, s
Fe as FeCl,, 200 mg. Mn as MnCI, and 200 mg. Al as AlCl, was treated by P.
iltrate was evaporated nearly to dryness and tested for thallium by *P. S^d:
lupitate of Til separated.

P. 52, N. S: Necessity of Adding Na^CO, la PrecipitaU Barium. — i and 2 ;
IS BaCl, were treated by P. 52, the final volume being about 30 cc.: on boilinf
:ipitate which was proved to contain barium appeared in both experiments,
iras very slight in that with i mg. — These experiments were repeated, exce
10 Na^Oj was added: no precipitate appeared on boiling.

1 mg. Cr and too mg. Ba as chlorides were treated by P. 52, except that no
iras added: the filtrate was colorless, and the solution obtained on dissolving t
dpitate in HNO, was yellow, showing that the chromium had been prediHt
3aCrOj. — The experiment was repeated with 1 mg. Cr and 100 mg. Ba, exce
>Ja,CO, was added: the filtrate was yellow, showing that the addition of Na,CO,
Jiromium to pass into the filtrate even when a large amount of barium is ]

Solubiiily of Zinc Carbanale in NaOH. — 50 mg. Zn as nitrate were dissolve
:c. water and an equivalent amoimt of Na,CO, added: a white precipitate se[
i cc. more to per cent. NaOH solution were added: the precipitate dissolved
nore NaOH and 10 cc. 10 per cent. Na,CO, solution were added, and the i
leated to boiling: no precipitate separated. More Na^CO, was added: a pre
inally formed.

P. 5i, N. g: SolvbUily of the Phosphates oj Aluminum and Zinc in NaOl
□g. Zn and too mg. Al as phosphates were dissolved in separate experiment
;c. water and a very little HNOj (1.20); 10 per cent. NaOH was added slow)
ipitates separated but dissolved when an excess of atxiut 4 cc. had been added

Partial Decomposition of Phosphates on Treating with Sodium Hydroxide. —
Ig. SO mg. Ca, and 20 mg. (ferric) Fe, all as freshly precipitated phosphates weri
eparately with about 30 cc. 4 per cent. NaOH solution for 3 or 4 minutes in ■
asseroles; the mixtures were filtered, the precipitates being washed with dilute
Tie filtrates were made strongly add with HNO„ and the precipitates were di
n HNO,. Each solution was evaporated to a small volume and treated with
c. ammonium molybdate solution. The amoimt of phosphate in each soluti
stimated by the amount of yellow precipitate that had separated out after

SYStEM OI^ QtrAMtATlvn ANALYSIS. 543

hours: in the case of magnesium only a very small proportion of the phosphate remained
in the precipitate; in the case of the iron about one-fifth remained in the precipi-
tate; and in the case of calcium the phosphate was about equally divided between
precipitate and filtrate.

Behavior of Caicium Borate and Oxalate towards Sodium Hydroxide. — A mixture
containing 100 mg. Ca as CaCl, and 80 mg. BO, as H3BO, was treated by P. 52 ; the
precipitate was tested for borate by adding to it in a casserole H2SO4 (1.84) and alco-
hol, igniting the alcohol and stirring vigorously: no green color appeared. The fil-
trate was acidified with H2SO4, evaporated, and tested in the same way : a large amotmt
of borate was fotmd. A trial experiment showed that 5 mg. BO, could easily be de-
tected in this way.

A mixture containing 500 mg. Pe as FeCl,, 500 mg. C^O^ as oxalic add, and 100 mg.
Ca as CaCl, was treated by P. 52 ; the filtrate was analyzed for oxalate by acidifying,
adding NH^OH and CaCl,: a very large precipitate of calcium oxalate separated.
—The experiment was repeated with 300 mg. Ca and 500 mg. Cfi^ (but no iron) : the
result was the same. The precipitate was tested for oxalate by dissolving it in HCl
and adding NH4OH: a small precipitate separated, showing that a large proportion
of the oxalate had passed into the filtrate.

Decomposition of Oxalic Acid by Acids. — 10 and 100 mg. oxalic add were treated
separatdy by P. 61, about i g. KCIO, being added to the concentrated HNO, solution.
The solutions were fijially evaporated to almost 5 cc, diluted to 30 cc, neutralized
with NH4OH, and tested for oxalate by adding 10 cc. of 10 per cent. CaCNO,), solu-
tion: no predpitate separated in dther experiment. — ^The experiment with 10 mg.
oxaUc add was repeated, except that no KCIO3 was added to the HNO3 solution:
a predpitate containing about 3 mg. oxalic add resulted. — 10 mg. oxalic add were
added to 15 cc. HCl (1.20), the mixture was evaporated to 2 or 3 cc, diluted and tested
for oxalic add as above: a predpitate of caldum oxalate separated that was estimated
to correspond to 3 or 4 mg. oxalic add. — The experiment was repeated with aqua
regia instead of HCl (1.20): somewhat more oxalic add was decomposed, but at least
2 mg. remained. — These experiments prove that the decomposition of oxalic acid
is very rapid in a hot mixture of HNO3 and HCIO3, much slower in aqua regia, and
still slower in UO (x.20) or HNO, (1.42).

P. 5^, N. 10: Color of an Alkaline Solution of Sodium Chr ornate and Peruranate. —
I mg. Cr as KCrSO^ was treated by P. 52, the final volume being about 30 cc: the
yellow color of the solution was easily seen.

I mg. U as UO,(NO,)j| was treated by P. 52, the final volume being about 15 cc. :
the solution was distinctly yellow. — The experiment was repeated with 5 mg. U ; the
solution had a reddish tinge. It was diluted to 30 cc: the reddish color changed
to yellow.

Color of Vanadium Solutions. — 100 mg. V as NajVO^ were added to a little concen-
trated HCl, and the mixture was evaporated to a small volume: the solution was at
first orange-yellow but turned greenish blue on evaporation with ' HCl, showing re-
duction of the vanadium to the quadrivalent state. Excess of concentrated HNO3
was added and the mixture again evaporated to a small volume: the solution rapidly
became yellow and the concentrated solution was deep red, doubtless owing to the
presence of free vanadic add (See bdow). On diluting to about 20 cc, it became yd-
low. The solution was neutralized with 10 per cent. NaOH solution: a small ydlow
predpitate separated during the neutralization but dissolved on the addition of a
little more NaOH; the alkaline solution was of a deeper yellow color than the dilute
HNO, solution. Na^O, was added: no change was observed. The mixture was boiled
for I minute: the solution became colorless. The solution was cooled and made add

544 ^' ^' NOYBS, W. C. BRAY AND E. B. SPBAR.

with HNOa: it became yellow, the color being deepest at the neutral point. To a por-
tion of this add solution were added 50 mg. Ti as TiCl^: no change in color was ob-
served, showing that the Na^O, had been completely decomposed and that no per-
vanadate had remained in the alkaline solution after boiling. DQUberg, Z, physik,
Chem., 45, 172 (1903), considers that the yellow color is due to H4V«0,7=, an ion of the
tetrabasic hexavanadic add, ll^JO„ (see C. B., *P. 58a, N. 6).

P. 52, iV. 11: Division of Zinc in the Sodium Peroxide Treatment. — 500 mg. Fe as
FeS04 and 10 mg. Zn as ZnCl, were treated by P. 52, the final volume being about
30 cc. ; the filtrate was tested for zinc by P. 53 and 57 : only about 3 mg. Zn were found,
showing that the remainder had been carried down with the FeCOH),. — ^The experi-
ment was repeated with 5 mg. Zn as ZnCl,: only a trace of zinc was found in the fil-
trate.— ^The last experiment (with 5 mg. Zn) was repeated except that a large excess
of NaOH was added before the Na^O,: i to 2 mg. Zn were found in the filtrate. The
predpitate of Fe(OH), was analyzed by P. 64, 66 and 67: the remainder of the zinc
was found.

500 mg. Mn as MnCl, and 10 mg. Zn as ZnCl, were treated by P. 52, and the filtrate
was tested for zinc by P. 53 and 57 : no zinc was found. — ^The experiment was repeated
with 20 mg. Zn: a very small predpitate of ZnS was obtained in P. 57 which contained
less than 0.5 mg. Zn. On analyzing the predpitate by P. 61, 64, 66 and 67 the zinc
was found. Therefore nearly 20 mg. zinc may be completdy carried down when
500 mg. Mn are present. — The experiment was repeated with 10 mg. Zn, except that
the add solution was poured into a fairly concentrated NaOH solution in P. 52 : about
0.3 mg. Zn was found in the filtrate, showing that a little zinc remains in the filtrate
in this case.

For the fact that zinc is carried down with nickd and cobalt, see T. A., No. 140-143.

Separation of Vanadium from Manganese. — ^A mixture containing 250 mg. Mn as
nitrate and i mg. V as Na3V04 was treated by P. 52; the filtrate was tested for vana-
dium by *P. s^e: a very good test was obtained.

^* 5J» ^' '•' Solubility of Zinc Phosphate, Carbonate, and Oxalate in AmmoniuM
Hydroxide. — 100 mg. Zn as ZnSO^ along with an equivalent amount of sodium phos-
phate, sodium carbonate or oxalic add were dissolved in separate experiments in
about 30 cc. cold water and about 2 cc. HNO3 (1.20). The solutions were neutral-
ized with NH4OH (0.96) and an excess of about 3 cc. added: dear solutions were ob-
tained in each case.

P. 54, N. i-j: Confirmatory Test for Aluminum. — ^See Knoevenagel, PrakHcum
des anorg. Chemikers, p. 160.

0.5 mg. Al and 0.2 mg. Co as nitrates were treated by P. 54: the ash retained the
form in which the filter paper was rolled and had a brilliant blue color. — The experi-
ment was repeated with 0.2 mg. Al and o.i mg. Co: the residue was distinctly blue.
— ^The experiment was repeated with o. i mg. Al and o. i mg. Co : no blue color was ap-
parent.

Effect of Other Elements upon the Confirmatory Test for Aluminumr. — 0.5 mg. Al
and 0.2 mg. Co as nitrates were treated by P. 54, except that i mg. Fe as Fe(NO,)s
was also added: the residue was brown. — ^This experiment was repeated with i mg.
Al: the residue was partly blue. — The experiment was repeated with 2 mg. Fe and
I mg. Al: the blue color could scarcely be distinguished.

The experiment with 0.5 Al was repeated in the presence of i mg. Be as Be (NO,),:
the residue was blue. — The experiment was repeated with 5 mg. Be: a satisfactory
test for aluminum was obtained. — ^The experiment with i mg. Al was repeated in
the presence of 5 mg. Be and also of 10 mg. Be as nitrate: a fair test for aluminum
was obtained in the presence of 5 mg. Be but none with 10 mg. Be. In the latter
case the paper and ash disintegrated.

SYSTEM OF QUALITATIVE ANALYSIS. 545

I mg. Al as nitrate was treated by P. 54, in the presence of i, 2, and 5 mg. U as
UOjCNOa), respectively: a slight test for aluminum was obtained in the first experi-
ment, a very poor one in the second, and none in the third. — ^The experiment was re-
peated with 0.5 Al and i mg. U: the test was very poor.

The experiment was repeated with i mg. Al and 5 mg. V as NagVO^: the residue
was distinctly blue.

0.5 mg. Al as Al,(P04), and 0.2 mg. Co as nitrate were treated by P. 54: the residue
was blue.

I mg. Al as nitrate was precipitated by adding to it NaHCO, solution and boiHng.
The precipitate was filtered off, washed once and treated by P. 54: the ash fused
together into a small mass and no blue color was apparent. — i mg. Al and about 5
mg. Na as nitrates were then treated by P 54: the result was the same, showing that
the failure in the first experiment had been due to the presence of a sodium salt.

50 mg. SiOj as NajSiO, were dissolved in about 2 cc. HNO, (1.20) and 30 cc. water,
heated to boiling, and made alkaline with NH^OH: the solution slowly became tur-
bid, and a gelatinous precipitate separated. This was collected on a filter, washed,
and a portion of hot HNOa(i.2o) was potued through the filter two or three times:
the silica did not dissolve. The HNO, solution was tested for aluminum by P. 54,
as mg. Co as nitrate being added : the residue was black. — The experiment was repeated
except that 2 mg. Al as nitrate were also present: a fair test for aluminum was ob-
tained.

P. $6, S. 1: Test for Ckromate with Hydrogen Peroxide. — For the constitution and
properties of the perchromic add, and of other perchromates, see Riesenfeld, Bcr.,
38, 1885, 3380, 4578, and 4068 (1905).

In a series of test tubes, each of which contained 0.3 mg. Cr as K^CrO^, were placed
o, 0.5, 2, 5, 10, and 20 cc. HNO, (1.20); each solution was diluted to 20 cc. ; 3-5 cc
ether and 2 cc. H^O, (3 p>er cent.) were added, and the solutions were shaken: in
the solution containing no add, and in the solutions containing 5, 10, and
20 cc. add, no blue color was obtained, but the remaining two solutions became blue
at once, and the blue compound was extracted by the ether on shaking. In the solu-
tion containing 2 cc. HNO3, the color in the ether layer in contact with the solution
disappeared in about half an hour on standing; the color was more brilliant and re-
mained for a longer time in the experiment with 0.5 cc. HNO,. — The experiment was
repeated with HjSO^ (1.20) instead of HNO,: no cQlor was obtained with no add or
with 10 and 20 cc. of it; the most brilliant and most permanent color was obtained
in the experiment with 0.5 cc. add. — The experiment was repeated with 30 per cent,
acetic add: in this case the best tests were obtained in the solutions containing 10
and 20 cc. of add; with the more ddicate solutions only a faint blue color was ob-
tained.

Another series of experiments was performed with solutions that contained 0.3
mg. Cr as K^CrO^, and 2 cc. HNO, (1.20) in about 20 cc. ; 3 cc. ether and varying amounts
of H,0, (3 per cent.) were added: with 10 cc. HjOj no test was obtained, with 5 cc.
a slight one, with 2 cc. a good test, but with 0.5 cc. H^Os a much better one.

In ail of these experiments in which the blue perchromic add was formed, the blue
color disappeared completely in less than i hour. In some of them the solution was
warmed to 50 or 60**: the blue color disappeared very rapidly. These experiments
prove that perchromic add is a very unstable substance even in the absence of an
excess of H+ or of H^O, and is still more so at higher temperatures.

Delicacy of the Confirmatory Test for Chr ornate. — ^To a solution containing 0.5 cc.
Cr as K^04 in 100 cc. was added a little ammonium acetate solution, 2 cc. acetic
add, and x g. BaCl,: a ydlow predpitate separated at once. This was collected on

54^ A. A. NOYBS. W. C. BRAV AND E. B. SPEAR.

a ^tei, and treated by P. 56; upon the addition of H^, and ether, a distinc
color resulted. The experiment was repeated with 0.3 mg. Cr: a slight hut d
precipitate and color were obtained.

Redaction oj Chromic Acid by Filler Paptr. — 10 mg. Or as K,CrOj were piedf
as lead chromate, and th6 mixture filtered. The predi^tate, nith the filter, was
in a casserole: to cc. water, and 5 cc. HNO, (i.io) were added, and the mixtu
boiled. The solution turned green in about i minute, thus showing that red
takes place readily.

P. 57, N. 3: Ignition Test for Zinc with Cobalt NOrale. — 0,1, 0.3 and 10 mg.
Zn(NU,), were treated as described in the second paragraph of P. 57, 0.3 nig.
nitrate being used in each experiment: a distinct green color resulted in the 1
ment with o.i mg, Zn, a deeper color with 0.3 mg. and an intense color with
— I mg. Zn and a mg. Co as nitrates were treated as described in the last par
of P. 57: the residue was green. The readue was then heated strongly: it
black, showing that when an excess of cobalt is present the teat is more delicau
mixture is not heated strongly. — The experiment was repeated with 2 mg. Cr
mg. Zn: a green color resulted which remained after the residue had been
strongly.

A solution containing 5 mg. Zn and 0.5 mg. Co as nitrates was evaporated 1
ness in a casserole and the mixture was ignited, first gently and then strong
residue was black owing to the presence of cobalt oidde.— 0.3 mg. Zn and o
Co as nitrates were treated by the last paragraph of P. 57, except that 5 cc. of
cent. Na,CO, solution were added in excess: only a very faint green color re
— The experiment was repeated except that only 3 drops Na,CO, solution were
in excess: the green color was mixed with black. — The experiment was repeatei
an excess of i cc. and also with an excess of 0.5 cc Na,CO, solution: a good gree
was obtained. — These experiments show that some Na^O, must be added, bt
a moderate quantity,

A solution containing 0.3 mg. Zn and 0.2 mg. Co as chlorides and an exce
was treated by P. 57 Oast par.), no HNO, being added: the tendue had no i
color, sbo^ng that HCl cannot be subsdtuted for HNO^

•P. 5*0, N. 3: Composition of Ae Zinc and Beryllium Precipitalts Produetd
dium Hydrogen Carbonate. — 500 mg. Zn as nitrate were treated by ♦P. 580,
a total volume of 100 cc. The precipitate was carefully washed with water, and
for carbonate in the usual way by treating with acid in a flask and passing I
evolved through Ba(OH), solution. A large precipitate of BaCO, was obtained
corresponded to at least aoo to 300 rag, Zn. — ^The experiment was repeated wi
mg. Be: a large amount of carbonate was also found in the beryllium precipit^
precipitate of BaCO, being about twice as large as in the experiment with zinc
sons states, J. Am. Chem Soc., aS, 557 (1906}, that the precipitate obtained in t
ammonium carbonate solution is a basic carbonate of beryllium.

PrecipitaUon oj Aluminum in Hot Sodium Hydrogen Carbonate Solutions in a
Bottle. Soiatioaa containing 3 mg. Al as nitrate in 100 cc. were placed in p
bottles, and 2, 5 and 10 mg. NaHCO, added: each of the three solutions quitUy I
turbid in the cold. The bottles were heated at about 95° in a water bath f
an hour: a flocculent precipitate formed in each case in less than 5 minutes, a
tied in less than 20 minutes. The bottles were cooled and the precipitates
off; the filtrates were evaporated with excess of HKO, and tested for alnminu
NH,OH: little or no precipitate separated, shovring that the pred|Htation of tl
minum had been practically complete in each case. — The experiments were n
with I, 1/4, 1/16, and 1/32 g. NaHCO,: the results were the same catcept tl

SYSTEM OI^ QUAi,lTATIVE ANALYSIS. 547

precipitation was not complete with 1/32 g. NaHCO,. — ^The experiment was repeated
with 1/64 g. NaHCO,: no precipitate was obtained. — ^These experiments were repeated,
except that the solution was thoroughly sattu'ated with CO, gas before adding the
NaHCO,: the results were nearly the same.
A solution containing 500 mg. Al as nitrate in 100 cc. was placed in a bottle and

4 g. NaHCO, added: there was a violent evolution of CO, gas, a small precipitate
formed, and the solution still reacted add to litmus. One g. more NaHCO, was added :
a very large precipitate separated, very little gas came off, and the solution after shak-
ing did not turn blue litmus red. The mixture was heated in the closed bottle at
95® in a ¥raterbath for half an hour, cooled, and filtered: the filtrate was fotmd to
contain no aluminum. Therefore 500 mg. Al are completely precipitated when about

5 g. NaHCO, are added in 100 cc. of solution.

PredpitoHon of Zinc Carbonate in Hot Sodium Hydrogen Carbonate Solutions. —
Four solutions, each containing 2 mg. Zn as nitrate dissolved in 100 cc, were potued
into strong 200 cc. bottles; to these were added i, 2, 3 and 5 g. respectively of solid
NaHCO,; the bottles were corked and shaken tmtil the NaHCO, dissolved; they were
then heated in a waterbath at 90 to 95^ for half an hour: all the solutions became
slightly turbid in the cold and after heating for 5 minutes precipitates could be seen
in suspension. After the half hour's heating the bottles were allowed to cool to about
40^, and the solutions containing i and 5 g. NaHCO, were filtered and tested for zinc
by evaporating with HNO„ adding NH^OH and acetic acid and passing in H,S: only
a trace of zinc was found in each filtrate, thus showing that the precipitation was prac-
tically complete in both i per cent, and 5 per cent. NaHCO, solutions. — The experi-
ments with I and 2 g. NaHCO, were repeated except that the solutions were thor-
oughly saturated with CO, gas at room temperature before the NaHCO, portions
were added: the results were the same.

The experiment was repeated with a solution containing 500 mg. Zn as nitrate in
100 cc., 2 g. NaHCO, being added: no precipitate of ZnS separated. The experiment
was repeated with 500 mg. Zn and 1.5 g. NaHCO,: about 5 mg. Zn were found in the
filtrate, showing that 1.5 g. NaHCO, is not quite sufficient to precipitate 500 mg. Zn
completely.

PrecipiiaUon of Beryllium in Hot Sodium Hydrogen Carbonate Solutions in a Closed
Bottle. — A solution containing i mg. Be as nitrate in 100 cc. was placed in a 200 cc.
bottle, I g. NaHCO, was added, and the mixtture was digested in the tightly stoppered
bottle at about 95^ in a waterbath for half an hour: the solution became turbid in less
than 5 minutes and a precipitate separated in 10 minutes ; on standing in the cold for
an hour the precipitate seemed to redissolve to a small extent. The experiment was
repeated with a solution that was saturated with CO, gas before the NaHCO, was
added: the result was the same. — These experiments were repeated with i mg. Be
and 2 g. NaHCO,: the mixture that had been saturated with CO, was distinctly turbid
in 15 minutes while the other was only very slightly turbid after half an hour. The
solutions cleared on standing for an hour in the cold. — ^The last experiments were
repeated with 2 mg. Be and 2 g. NaHCO,: distinct precipitates were obtained corre-
sponding to about I mg. Be but these redissolved to a considerable extent in the cold.

These experiments were repeated with solutions containing 3 to 10 g. NaHCO,
in 100 cc. of solution and varying quantities of ber}'llium: in the 3 per cent. NaHCO,
solution 3 mg. Be gave no precipitate, but a larger amoimt did so; in the 5 per cent,
solution, about 15 mg. Be just remained in solution in 100 cc. ; and in the 10 per cent,
solution, about 75 mg. Be gave a small precipitate in 50 cc, while 50 nig. Be gave
only a negligible precipitate.
Solubility of Uranyl Vanadate in Sodium Hydrogen Carbonate Solutions. — Solutions

54^ A. A. NOYES, W. C. BRAY AND K. B. SPBAR.

containing loo mg. U as UOjCNOg), and 60 rag. V as Na,V04 were mixed, a few drops
HNOg (1.20) were added to dissolve the precipitate, the mixture was diluted to 100 cc.,
2 g. NaHCO, were added, and the mixture was digested in a closed bottle at 95** for
half an hour: no precipitate separated. The mixture was cooled down, 40 mg. V as
Na3V04 were added, making 100 mg. in all, and the mixture was again heated at 95^ for
half an hour : no precipitate separated in the hot solution, nor in the cold on standing
several days. — ^The experiment was repeated with 100 mg. U and 100 mg. V, except that
the mixture was first treated with NaOH and HjOj by P. 52, and that 1.5 g. NaHCO,
were used instead of 2 g. : the result was nearly the same. — 100 mg. U as UO,(NO,)p
100 mg. V as NagVO^, and 3 g. NaNiOg were treated by *P. 58a, 1 g. excess of NaHCO,
being added : a precipitate formed on adding NaHCOg and a considerable one remained
after heating the mixture at 100° in a closed bottle for 30 minutes. The precipitate
and filtrate were analyzed for uranium and vanadium by *P. s^c-e: the filtrate was
found to contain 35-40 mg. of uranium and all but 10 or 15 mg. of the vanadiam.

The last experiment was repeated except that the mixture vras warmed in a flask
at 95° (without allowing it to boil) for 30 minutes: the precipitate that remained was
smaller, and contained only 10 to 20 mg. U and 2 to 5 mg. V. — The last experiment
was repeated except that the mixture was boiled for i minute in an open flask: the
precipitate dissolved completely, and no precipitate formed on cooling even in 24
hours, showing that uranyl vanadate is more soluble in slightly alkaline carbonate
solutions than in those from which the CO2 is prevented from escaping.

An acid solution containing 50 mg. V as NagVO^, and 50 mg. U as UO,(NO,). was
diluted to 100 cc. and neutralized by adding a little powdered NaHCO,. Then 0.5
g. NaHCO, was added, and the mixture was heated in a presstu-e bottle at 95^ for half
an hour: on heating a large precipitate remained. — ^To a solution containing 100 mg.
V as NagVO^, 20 mg. U as U02(N08)2 and a known excess of HNO, (1.20) in 50 cc.
(instead of 100 cc.) was added just sufficient NaHCO, to leave i g. undecomposed
NaHCO,. The mixture was heated in a closed bottle at 95° for half an hour: no pre-
cipitate separated. — To this solution after cooling, was added 20 mg. more U and
the mixture was again heated at 95°: a large precipitate separated in the cold afiid
remained on heating. — ^The experiment was repeated with 100 mg. U and 10 mg.
V: no precipitate separated. 10 mg. more V were added and the heating repeated:
a large precipitate separated in the cold and remained on heating. — ^The last experi-
ment was repeated with 100 mg. V and 100 mg. U and the precipitate and filtrate
were analyzed for uranium and vanadium by *P. 58c to e: the filtrate was found to
contain 20 or 30 mg. of uranium and all but a few mg. of the vanadium.

N on- Precipitation of Uranyl Chromate^ Phosphate^ and Oxalate in the NaHCO^ Pro-
cedure.— ^To a solution containing 100 mg. U as U0,(N0,)2 200 mg. Cr as K^CrO^,
and a known (small) amotmt of HNO, (1.20) in 100 cc. was added enough NaHCO,
to leave i g. of the imdecomposed carbonate ; the mixture was heated in a closed bot-
tle at 95^ for half an hour: no precipitate separated. — ^A solution containing 100 mg.
U as U02(N03)2 and 100 mg. PO4 as NaHPO^ was treated by P. 52 and then by *P.
58a; no precipitate separated in the NaHCO, procedure. After cooling an additional
250 mg. portion of PO4 was added, and the heating repeated: no precipitate separated.
— The experiment was repeated with 100 mg. V as NagVO^ and 200 mg. oxalic add:
no precipitate separated.

*P. 38a, N. 4: Precipitation of Aluminum^ Zinc, and Beryllium in Hot Sodium Hy-
drogen Carbonate Solutions in Open Flasks, — 2 mg. Al as nitrate and 3 g. NaNO, were
dissolved in 100 cc. water ; a little NaHCO, was added until the mixture after shaking
just ceased to react acid to litmus paper; i g. more was added ; the mixture was heated
to about 90° and digested at this temperature in a covered flask on a waterbath for

SYSTEM OF QUALITATIVE ANALYSIS. 549

half an hour: the precipitate slowly dissolved, except a few small flakes. Tlie mix-
ture was filtered while hot, and the filtrate tested for aluminum by acidifying with
acetic acid, adding Na^HPOf and heating to boiling: almost all the aluminum was
found in the filtrate. — This experiment was repeated twice with 2 mg. and with i
m%: the results were the same. — The experiment was repeated twice with 20 mg. Al:
the filtrate contained less than i mg. Al.

The experiments with 2 and 20 mg. were repeated, except that CO, gas was passed
through the mixture: the results were nearly the same, except that the filtrate may
have contained somewhat more aluminum, showing that there is no advantage in using
COjgas.

The experiment with 20 mg. Al was repeated except that the mixttire was boiled
for one minute: the filtrate contained 3 or 4 mg. Al. — ^The experiment was repeated
except that the mixture was allowed to cool before filtering: the filtrate contained
about I mg. Al. — ^The experiment was repeated with 4 mg. Al : the precipitate disap-
peared completely on boiling, but a precipitate separated out on cooling. — The ex-
periment was repeated with 6 mg. Al : a very small precipitate remauied on boiling i
minute.

I mg. Zn as nitrate and 2 g. NaNO^ were dissolved in 100 cc, i g. excess NaHCOj,
was added, and the mixture boiled in a fiask for one minute: the precipitate did not
dissolve and was proved to contain zinc by P. 53-57. — The experiment was repeated
with 20 mg. Zn; the filtrate being tested for zinc by acidifying with HNO,, boiling,
adding NH^OH and passing in H^S: no precipitate formed, showing that the zinc
was completely precipitated.

10 mg. Be as chloride and 2 g. NaNOg were dissolved in 100 cc. water, NaHCO;,
was added until the mixture after shaking just ceased to react acid to litmus paper ;
I g. more was added; the mixture was boiled for i minute, and filtered while hot;
the filtrate was acidified with HNO„ eva|X)rated to about 15 cc. and tested for beryl-
lium by adding NH,OH: the filtrate contained only alx)ut o.i mg. Be. — ^The ex-
periment was repeated except that the mixture was allowed to cool before filtering:
the filtrate contained about 0.3 mg. Be. — ^The experiment was repeated with i mg.
He: the precipitate remained on boiling for one minute, and did not disappear on
standing in the cold for several hoiu's.

These experiments were repeated with solutions containing from 1.5 to 5 g. NaHCOn
in 100 cc. of solution and varying quantities of beryllium: in a 1.5 per cent. NaHCOj
with I mg. Be little or no precipitate remained on boiling i minute; in a 2 per cent,
solution no precipitate remained with 2 mg. Be, but one remained with 4 mg. ; in a 3
per cent, solution 40 mg. Be dissolved almost completely, but 50 mg. gave a large
precipitate. A comparison of these results with those obtained in a similar series
of experiments in closed bottles (C. E., N. 3, above) shows that considerably more
beryllium dissolves in concentrated NaHCOj solutions when the COj is allowed to es-
cape.

♦P. ^Sa, .W 5; Precipitation of Uranium with Aluminum or lieryllium in Hot Dilute
Sodium Hydrogen Carbonate Solutions. — See T. A., No. 171 to 175. In an analysis
with 200 mg. A\ and xo mg. U, starting at *P. 58a, i or 2 mg. U were found in *P.

*P. 58a, N. 6: Dissociation Relations of Carbonic Acid.- Sec McCoy, .4m. Chem. /.,

»9. 437 (1903)-

Dissociation Relations of Chromic Acui. — See Spitalsky, Z. anorg. Chem., 54, 265
(1907) and Sherrill, /. .4m. Chem. Soc, 29, 1641 (1907). The work of the latter shows
that HjCrO^ dissociates in steps, first into HCrO^"" and then into CrO^=, that the
HCrO,"~ is a very weak acid, and that the latter even in dilute solution is converted

550 A. A. NOYES, W. C. BRAY AND E. B. SPEAR.

by dehydration in large measure but by no means wholly, into dichromate-ion (Cr,Oy^).

Condition of Vanadic Acid in Solution. — See DtUlberg, Z. physik. Chem., 45, 129-181
(1903). He considers that in a solution of the conjposition of Na,V04, which reacts
strongly alkaline, the salt is Isirgely hydrolyzed according to the reaction 2Na,V04 +
HjO = 2NaOH + Na^VjOy, but his evidence is not convincing. On adding HCI
slowly the results indicate the presence of a tetrabsisic add, H^VaOi,. This is a strong
acid, two of the hydrogens splitting off nearly completely in dilute solutions, and the
third one to a considerable extent. The yellow color is probably due to the ion
HVeOi7=. The transition of one form of vanadic add into another takes place readily,
the final equilibrium state bdng reached fairly quickly. Conductivity and freezing-
point determinations make it probable that the formula of sodium metavanadate
is (Na+),V,0 =.

♦P. jSb, N. i: Partial Decomposition of //,0, in the NaHCO^ Treatment, — A mixture
containing 100 mg. U as UOjCNOj), and 100 mg. V as Na,V04 was treated by P. 52 and*P.
58a and then was acidified with HNO,: the add solution before adding the NaHCO, was
red owing to the presence of per vanadic add, indicating the presence of much H,Oj;
on neutralizing with NaHCOj it became yellow ; and the final HNO, solution was also
pale ydlow. The solution was tested for H^O, with a titanium solution: a distinct
color was obtained. These results show that much but not all of the H^O, had been
decomposed.

♦P. 5<?6, N. 2: Detection of Chromic Add in the Presence of Uranium and Vanadium
by Hfi^, — A mixture containing i mg. Cr, 100 mg, IT, 100 mg. Zn and 100 mg. Al as
nitrates, and 100 mg. V as NajV04 was treated with NaOH and NasOj by P. 52: a deep
red solution resulted and there was no residue. The solution was made add with
HNO3, bdng cooled carefully during the process. Just enough add was added to
dissolve the precipitate that formed. To about one-fourth of the solution in a test-
tube was added a little ether and about i cc. HjO,: the ether layer became blue. The
water layer was red before HjOj was added, showing the presence of pervanadic add.
— The experiment was repeated with 2 mg. Cr, 100 mg. U and 100 mg. V : the result
was the same. After testing for chromium the remainder of the solution was treated
by *P. 58a, the solution was cooled, made just acid with HNOj, and one-tenth of it
tested for chromic add with HjOji a good blue color was obtained. The remainder
of the solution was made somewhat more strongly add and evaporated to half its
volume; the test for chromic add was then made with half the solution: no blue color
was obtained, showing that the chromic add had been reduced on boiling.

Action of HJJ^ on Uranium Salts. — To 10 mg. U as UO,(NOa), in 20 cc. were added
I cc. HNO3 (1.20), I cc. 3 per cent. HjO,, and a little ether: the yellow color of the solu-
tion did not change, and the ether layer remained colorless.

*P. 5<y6, N. 4: Precipitation of Chromate by Lead Nitrate. — Several solutions each
containing 0.5 mg. Cr as K^CrO^, 2 g. NaNO,, and i g. NaHCO, in 100 cc. or 70 cc.
were made distinctly acid with HNO3, shaken to expd CQ,, and neutralized exactly
with NaOH ; varying amounts of HNO, (1.20) were added, and finally known volumes
of a 20 per cent, lead nitrate solution. The results are shown in the following table:

predpitate in i or 2 min.
precipitate, in 3 or 4 min.
no precipitate in 30 min.
predpitate in 2 or 3 min.
predpitate very slowly,
predpitate in 3 to 5 min.

Initial
volume.

Volume of
HNOsCi.ao).

Volume of lead
nitrate solution.

100 cc.

1 . 5 CC.

10 CC.

100 CC.

2.5 CC.

20 cc.

100 cc.

5.0 CC,

30 cc.

70 cc.

1 . 5 CC.

10 cc.

70 cc.

2 . 5 cc.

10 cc.

70 cc.

2.5 cc

20 cc.

SYSTEM OF QUALITATIVE ANALYSIS. 55 1

The last two experiments were repeated with 6 g. NaNOj (instead of 2 g.): only very
small precipitates were obtained in half an hour, even with 20 cc. PbCNOg), solution,
showing that the presence of much NaNO, prevents the precipitation of PbCr04.

Behavior of Vanadate and Uranate towards Lead Nitrate. — Several solutions, each
containing loo mg. V as NagVO^, 2 g. NaNOg and i g. NaHCOg, in 100 cc, were
acidified ^ith HNO,, shaken, and then neutralized exactly with NaOH; 1.5 to 3 cc.
of HNO, (1.20) and 10 to 30 cc. of a 20 per cent, lead nitrate solution were added and
the mixture allowed to stand 30 minutes: with 3 cc. HNO, there was no precipitate
even with 30 cc. Pb (NO,), ; with i .5 and 2 cc. HNO3 there was none with 10 cc. Pb (NO,), ;
but a small one with 20 cc., which was however very slight when 2 cc. HNO, were used.
The experiments were repeated with an initial volume of 70 cc: the results were nearly
the same, except that the precipitates when formed were somewhat larger.

A solution containing 100 mg. U as UOa(NO,)2, 2 g. NaNO,, and i g. NaHCOg in
100 cc. was neutralized with HNO3, and an excess of i cc. HNO3 (1.20) was added ;
finally 20 cc 20 per cent. Pb(N0,)2 were added: no precipitate separated in several
hours.

Precipitation of Aluminum, Beryllium or Lead by Sodium Phosphate in *P. 58c. —
0.5, 1, and 2 mg. Al as nitrate, in solutions containing 2 g. NaNO,, 2 g. NH4NO3, and 5
cc. 30 per cent, acetic add in a total volume of 100 cc were heated to boiling : floccu-
lent white precipitates separated in each case, the results showing that this is a very
good method of estimating small amotmts of aluminum.

The experiments were repeated with 0.5, i and 2 mg. Be as chloride: flocculent
white precipitates resulted except in the experiment with 0.5 mg. The filtrate in the
experiment with 2 mg. was estimated to contain about 0.5 mg. Be.

The experiments were repeated with i and 2 mg. Pb as nitrate: a distinct precipi-
tate was obtained with 2 mg. but only a very small one with i mg. Pb.

Oxidation of Hyporoanadic Add by Bromine. — 50 mg. V as Na3V04 and i g. Pb as
Pb(NO,), were dissolved in 2.5 cc HNO, (1.20) and 100 cc. water; the mixture was
saturated with H,S in the cold, filtered, and the filtrate was boiled: the cold solution
had a blue color, but on boiling sulphur separated and the color became deeper. The
sulphur was filtered off, bromine water added imtil the bromine odor was distinct
after shaking, and the mixture was boiled: the color was still blue. A small excess
of bromine was again added: the solution was still blue. — The experiment was repeated
except that several drops liquid bromine were added, and the mixture shaken, and
allowed to stand two or three minutes; the bromine was then boiled off: the solution
was yellow. The filtrate was treated by *P. 58c.' no precipitate separated.

Behavior of Vanadyl Salts and Vanadates towards Sodium Phosphate. — 50 mg. V as
NajVO^ and i g. Pb as Pb(NO,)2 were dissolved in 2.5 cc HNO3 (1.20) and 100 cc.
water; the mixture was saturated with H,S, filtered, boiled and filtered again. The
blue filtrate was treated by *P. 58c, 2 g. (NHJ^SO^ and 2 g. Na^HPO^ being added :
on warming, a large, flocculent, bluish white precipitate separated. — The experiment
was repeated except that the vanadyl salt was first oxidized with liquid bromine as
described in P. 586; no precipitate separated in *P. 58c.

♦P. 5^c, N. i: Precipitation of Uranyl Ammonium Phosphate. — See Kern,/. Chem.
Soc., 23, 705-10 (1901).

Detection of Small Amounts of Uranium. — 0.3, 0.5 and i mg. U as U02(N08)2 were
dissolved in separate experiments in 100 cc water containing 5 cc. 30 per cent, acetic
add and 3 g. NaNOj, and treated by P. 58c.' white flocculent precipitates separated
in each case, but that with 0.3 mg. was very small. The precipitates were filtered
off and tested by P. 58^.' in the experiment with 0.3 mg. U the ferrocyanide test failed,
but was very satisfactory in the other two cases, — ^The experiment with 0.3 mg. U

552 A. A. NOYES, W. C. BRAY AND E. B. SPEAR.

was repeated except that 0.5, 4, and 6 g. respectively (NH4),S04 (instead of 2 g.) woe
added : very small precipitates of uranyl ammonium phosphate were obtained in cadi
case, but the ferrocyanide test failed. This shows that there is no advantage in add-
ing more than 2 g. (NH4),S04. — The experiments with 0.5 and i mg. U were repeated,
except that no ammonium salt was added: with i mg. the result was satisfactory,
but with 0.5 mg. the confirmatory failed.

The first experiments with 0.3, 0.5 and i.o mg. U were repeated in a volume of 40 cc.
instead of 100 cc. : a distinct phosphate precipitate resulted in each case, and the ferro-
cyanide test was satisfactory, even with 0.3 mg. With 0.2 mg. a small phosphate
precipitate formed but the confirmatory test failed. — ^The experiment was repeated
with 1.0 mg. U, except that 0.3 g. Na^P04.i2H30 was added instead of 2 g.: the
confirmatory test failed, showing that a large excess of Na2HP04 is necessary.— The
experiment with 0.3 mg. U was repeated except that 10 cc. acetic add was used m-
stead of 5 cc. : the confirmatory test failed, showing that a large excess of acetic add
makes the phosphate precipitation less complete.

Separatum of Uranium and Vanadium by Phosphate: — 100 rag. U as UOjCNOj),!
50 mg. V as Na,V04, and i g. NH4NO3 were dissolved in a few drops HNO, and 30 cc
water. The mixture was neutralized with NH4OH : a large pale precipitate of uranyl
ammonium vanadate separated. 10 cc. 30 per cent, acetic add were added: the pre-
dpitate did not dissolve. 5 cc. 10 per cent, ammonium phosphate were added, the
mixture was heated to boiling, allowed to stand for 15 minutes, and filtered. The
filtrate was tested for vanadium by '"P. 58^.* a large quantity was found. The pre-
dpitate was washed with dilute NH4NO, solution, dissolved in dilute hydrochloric
acid, and the phosphate precipitation repeated : less than i mg. V was now found m
the filtrate. The phosphate predpitate was treated in the same way: less than ai
mg. V was found in it. — The complete experiment was repeated except that the mix-
ture containing the uranium ammonium vanadate predpitate and the acetic add was
heated to boiling, cooled, and allowed to stand over night before the phosphate was
added: the result was the same, showing that the vanadium passes into the filtrate
even when it is first predpitated in combination with the uranium.

*P. s8c, N. 4: Precipitation of Uranyl Hydrogen Phosphate. — See Kern, /. Am. Chem.
Soc, 23, 705 (1901).

*P. ^Sdt N. i: Behavior of Uranyl Salts towards Potassium Ferrocyanide. — Known
amounts of K4Fe(CN)o solutions were added to solutions containing 0.5 mg. U as
UOjCNOj)^, I cc. HCl (1.12), and 10 cc. nearly saturated NaCl solution: with 4 cc
of a 3 per cent, ferrocyanide solution a brown color appeared only after 40 minutes,
with 10 cc, a brown predpitate separated in about 10 minutes, while with 10 cc of
a 10 per cent, ferrocyanide solution a predpitate separated in a minute or two, ^w-
ing that a large excess of K4Fe(CN)« is needed to make the test ddicate. — ^The ex-
periment was repeated except that 0.5 cc. HCl (1.12) was added, instead of i cc,
10 cc. 3 per cent, ferrocyanide solution being added: a brown predpitate separated
in a minute or two. — This experiment was repeated with o.i and 0.05 cc. HCl (1.12)'
brown predpitates separated at once. — ^The experiment was repeated without the
addition of any add: no brown color appeared. — The experiment was repeated, add-
ing 5 cc. HCl (1.12): no brown predpitate separated in i hour, but the solution he-
came blue and a blue predpitate separated slowly on account of the decomposition
of the ferrocyanide. — These experiments show that a little add must be present, hot
that much add makes the test less delicate.

o.i and 0.2 mg. U as UOa(NO,)j were treated by *P. 53d, a few mg. PO4 as NaJlPO,
being added: a good test was obtained with 0.2 mg., but only a very poor one after

SYSTEM OF QUALITATIVE ANALYSIS. 553

several minutes with o.i mg., showing that o.i mg. is about the limit of detect ability,
and that phosphate does not interfere with the detection of 0.2 mg. U.

The experiment with 0.5 mg. U and 0.5 cc. HCl (1.12) was repeated, except that
water was added instead of NaCl solution: a dark red color resulted, and no precipi-
tate separated out in i hour, showing that NaCl is needed to coagulate the colloid.

Behavior of Vanadyl Salts and V anodic Acid towards Potassium Ferrocyanide. —
5 mg. V as Na,V04 after evaporation with HCl (whereby it is reduced to vanadyl chlor-
ide) were treated by *P. 58^; a greenish yellow precipitate separated. The precipi-
tate was filtered off and the filtrate tested for vanadium by *P. 58^; only a very faint
pink color resulted, showing that the precipitation of the vanadium by the ferrocy-
anide is nearly complete. — ^The experiment was repeated except that only two drops
of HCl were added and the solution was not evaporated: no precipitate separated in
half an hour, showing that a small quantity of vanadium in the form of vanadic add
is not precipitated by K^FeCCN)^.

♦P. 58e: Non-interference of Lead with the Vanadium Test, — A solution containing
2 mg. Pb as Pb(NO,)„ and 0.5 mg. V asNa3V04and 2 g. NH4NO, was treated by P. 58*
and the black precipitate of PbS formed was filtered off: on saturating completely
with H3S, the filtrate had the characteristic violet-red color due to vanadium.

*P' 5*gf ^^- '•* Separation of Aluminum and Beryllium. — See Havens, Z. anorg,
Chem., 16, 15 (1898).

Precipitation of Aluminum as AlCl^.dH^O, — 0.5 mg. Al as AlCl, was treated by *P.
58^, the total volume of the acid ether mixture being about 15 cc.: on saturating with
HCl gas no precipitate could be seen, but on standing about 15 minutes a distinct crys-
talline precipitate separated out.

100 and 500 mg. Al as AICI3 were treated by *P. 58g; the filtrates, which had a total
volume of about 70 cc., were evaporated to about 20 cc. and made alkaline with NH4OH :
no precipitate separated, but in the second experiment there was a faint turbidity
corresponding perhaps to 0.1 mg. Al.

Behavior of Beryllium, Uranium and Chromium in the Hydrochloric Acid Ether
Process. — 100 mg. Be as BeCl, were treated by *P. 58g, the final volume being about
30 cc.: a small precipitate remained. This was filtered off, washed twice, dissolved
in a little water, and proved to contain aluminum but no beryllium by boiling in a
10 per cent. NaHCO, solution as described in *P. 58A.

500 mg. Al and 2 mg. Be as chlorides were treated by *P. 58g in a volume of 60 cc.
and the filtrate was treated by *P. 58A; a good test for beryllium was obtained and
the amount of beryllium in the filtrate was estimated to be i or 2 mg., showing that
very little beryllium was retained by the A1C1,.6H20.

100 mg. U as UOjClj were treated by *P. 58g, the total volume being 50 cc. : a clear
yellow solution was obtained.

20 mg. Cr as K^CrO^ were boiled with HCl (1.20) to reduce the chromium to the
chromic state, and the green solution was treated by *P. 58g, in a volume of about
30 cc.: the green color quickly disappeared and a violet precipitate separated. This
was filtered off after several hours and the filtrate tested for chromium by evaporating,
adding NH^OH, and boiling: a precipitate estimated to contain 5-10 mg. Cr separated.

*P. 58h, N. i: Separation of Beryllium and Aluminum in Strong Sodium Hydrogen
Carbonate Solution. — See Parsons and Barnes, /. Am. Chem. Soc, 28, 1589 (1906).

200 mg. Al as nitrate were dissolved in 20 cc. water, and added to a warm solution
of 10 g. NaHCO, in 80 cc. water, the mixture was heated to boiling in a flask, boiled
for I minute, and filtered; the filtrate was acidified with HNO^, evaporated to about
20 cc. and made alkaline with NH^OH : no precipitate separated on warming gently,
nor on standing.

554

A. A. NOYES, W. C. BRAY AND E. B. SPEAR.

loo mg. Al and i nig. Be as nitrates were dissolved in 50 cc. water and 6 g. NaHCO,
were added ; the mixture was heated to boiling, boiled for i minute, and filtered while
still hot ; the filtrate was made add with HNO„ evaporated to about 10 cc and made
alkaline with NH4OH : a very small precipitate separated which was estimated to con-
tain 1/20 mg. Be. — ^The experiment was repeated with 2 mg. Be: the result was the
same. — The experiment was repeated with 5 mg. Be: scarcely i mg. Be was found m
the filtrate. — ^Therefore small amotmts of beryllium cannot be completely separated
from 100 mg. Al by this method.

The precipitate obtained in the experiment with i mg. Be was treated by ♦?. 58g-fc;
the beryllium present was estimated to be nearly i mg., showing that this separation
with HCl and ether is more satisfactory than that with 10 to 12 per cent.
NaHCO,.

Beharoior of Iron in Strong Sodium Hydrogen Carbonate Solution, and its Precipita-
tion as Sulphide, — 2 and 5 mg. Fe as FeCl, were treated by the last paragraph of ♦?.
58A.* small precipitates of Fe(OH), were formed in the NaHCO, solutions, but the pre-
cipitation was incomplete, for after acidifying the filtrates and adding NH«OH, pre-
cipitates of Fe(OH), were obtained, each of which was estimated to contain over i mg.
Fe. — The experiment with 5 mg. Fe was repeated except that 5 cc. NaOH were added
to the NaHCO, filtrate: no predpitute resulted in the cold, but about 0.5 mg. Fe as
Fe(OH), predpitated on boiling. The predpitation was still incomplete for about
0.5 mg. Fe was found in the solution on addifying and adding NH4OH.

0.5 mg. Fe as FeS04 and in a second experiment as FeCl,, was boiled with 30 cc
10 per cent. NaHCO,, for i minute, the mixtures were poured through filters, cooled,
and H^ was passed in for about 3 seconds: the solutions became dark colored at once.
After about 5 minutes the mixtures were filtered: the filtrates were clear, with a faint
greenish shade corresponding to an insignificant amoimt of iron; more H^ caused
no further darkening. — The experiment with d.s mg. Fe as FeCl, was repeated, ex-
cept that the NaHCO, solution was not cooled before passing in H^: a dark green
solution was obtained which ran through the filter, showing that it is better to add
the H,S in the cold. To this solution (which was now cold) was added 2 or 3 mg.
Fe as FeS04: on filtering after several minutes the filtrate was nearly colorless, and
more H^ gave no predpitate nor color.

♦P. 5<yA, N. 2: Behavior of Uranyl Salts in Strong Sodium Hydrogen Carbonate Solu-
tion, and on Passing in //,S. — 50 mg. U as UO^^ (and in another experiment as
UO,(NO,),) were dissolved in a little water and added to a solution containing 5 g.
NaHCO,, the final volume being 50 cc.; the mixture was boiled for 5 minutes: no pre-
dpitate separated. — ^The experiment was repeated with 5 mg. U as UO/21, in a volume
of 30 cc. and 10 cc. 10 per cent. NaOH was added to the 10 per cent. NaHCO, solution
after boiling: no predpitate separated.

To a mixture containing too mg. Be and 20 mg. U dissolved in 50 cc. 10 per cent.
NaHCO, was added i g. NH4CI, and the mixture was boiled: no predpitate separated.

A mixture containing 100 mg. Be and 10 mg. U (but no iron) dissolved in 30 cc.
10 per cent. NaHCO, solution was saturated completely with H,S gas: no predpitate
separated.

The Separation of Uranium andlBerytlium by Potassium Ferrocyanide. — 15 and 25
mg. Be as chlorides in HCl solution were evaporated almost to dryness, 10 cc. satu-
rated NaCl solution added, the mixture was cooled and 5 cc. 10 per cent. K4Fe(CN)(
solution were added : dear solutions resulted which were blue colored, owing to the pres-
ence of a small amotmt of iron in the beryllium; after standing several hours a gelat-
inous light colored predpitate had separated in the experiment ti*ith 25 mg. Be. — The
experiment was repeated with mixtures of i mg. U as UO,(NO,), and with 5, 10, 15

SYSTEM OF QUALITATIVE ANALYSIS. 555

and 20 mg. Be: a dark red color appeared at once in the experiments with 5 and 10
mg. Be, and on standing dark red precipitates settled out. In the experiments with
15 and 20 mg. the color was somewhat obscured owing to the presence of the iron.
—The experiment with 20 mg. Be and i mg. U was repeated, except that the iron was
first removed from the berylUum solution by P. 52: a good test for uranium was then
obtained.

P. 61, N. t: Solubility of Manganic Peroxide in HNO^ in the Presence of Filter Paper.
- 500 mg. Mn as Mn(NOJ, were treated by P. 52; the MuO(OH), precipitate was
boiled with 30 cc. HNO, (1.20) in a covered casserole for several minutes: very little
of the precipitate dissolved. The experiment was repeated except that the filter, as
well as the precipitate, was treated with the nitric acid: on boiling, the filter disin-
tegrated and the precipitate dissolved completely in 10 or 15 min. — ^The last experi-
ment was repeated with HNO, (1.42): nearly all of the precipitate dissolved after 5
minutes' boiling, and the remainder on standing over night. The solution obtained
in the last experiment was evaporated to a small volume and diluted to about 15 cc.;
the paper was filtered off, and the filtrate was treated by P. 61 : the manganese was
completely precipitated.

P. 62, N. 2: PreripiiaHon of Manganese by Chloric and Nitric Acids, — See Harmay,
/. Ckem. Soc, 23, 269 (1878); Ford, Trans. Inst. Min. Eng., 9, 397.

P. 62, N. 5; Separation of Other Elements from Mar^ganese by Chloric and Nitric
Acids. — See T. A., No. 117, 118.

Iron. — 100 mg. Mn as MnCl, and i mg. Fe as FeS04 were treated by P. 52 and P. 61
and the filtrate was tested for iron by P. 64: a good test was obtained. — 500 mg. Mn as
MnCI, and i mg. Fe were treated by P. 52 and then by P. 61; the precipitate and fil-
trate were tested for iron by evaporating with excess of HCl, diluting, and adding
KSCN solution: a distinct test for iron was obtained in the filtrate but much more
iron was found in the precipitate than in the filtrate.

Titanium. — ^A mixture containing 500 mg. Mn and i mg. Ti as nitrates was treated
by P. 61; the filtrate was evaporated and made alkaline with NH4OH: no precipitate
separated, proving that the titanium had been completely carried down with the
MnO,. — ^This experiment was repeated with 5 and with 50 mg. Ti: the result was the
same in each case. — The experiment was repeated with 50 mg. Mn and 50 mg. Ti;
the precipitate was analyzed by *P. 62a, and about half the titanium found in it;
the filtrate was evaporated almost to dryness: some of the titanimn separated during
the evaporation as a white precipitate which did not dissolve readily in HCl, but did
in HF. — ^The experiment was repeated with 50 mg. Ti in the absence of manganese:
no precipitate separated on adding KCIO,, nor on evaporating to 5 or 10 cc.

Zirconium. — ^A mixture containing 500 mg. Mn and 2 mg. Zr as nitrates was treated
by P. 61; the filtrate was evaporated and made alkaUne with NH4OH: only a very
small precipitate separated, showing that not quite all of the zirconium had been car-
ried do^n. — ^The experiment was repeated with 50 mg. Zr: the filtrate was foimd to
contain 15 or 20 mg. Zr. The manganese precipitate was treated by *P. 62a; the rest
of the zirconium was recovered.

Vanadium. — A mixture containing 250 mg. Mn as nitrate and i mg. V as Na,V04
was treated by P. 61 : the filtrate was treated by *P. 58^ to test for vanadium: no trace
of vanadimn was found. For the fact that vanadium is not carried down by manganese
in the Na,0, procedure see C. E., P. 52, N. 11. — ^The experiment was repeated with
250 mg. Mn and 10 mg. V: a mere trace of vanadium was found. — ^The experiment
was repeated with 50 mg. Mn and 10 mg. V : not more than i or 2 V mg. were found
in the filtrate. — ^The experiment was repeated with 20 mg. Mn and 10 mg. V: the fil-
trate contained not more than 2 or 3 mg. V.

556 A. A. NOYES, W. C. BRAY AND E. B. SPEAR.

Uranium.— A, mixture of 250 mg. Mn as Mn(NOa), and i nig. U as UOj(X(),);W^s
treated by P. 61 : the filtrate was tested for uranium by *P. 5Sc/; a good test for uranium
was obtained.

Thallium. — 15 mg. Tl as Tl(OH)3 were treated by P. 61: no precipitate se|»arated.
200 lug. Mn as MnCl, and i mg. Tl as TICI3 were treated >)y this procedure. ITie fil-
trate was tested for thallium by the regular procedure: a good test was obtained. The
precipitate was dissolved in HCl (1.12) and treated by *P. b^a-d: a very small pre-
cipitate of Til was obtained, which was estimated to contain not more than o.i mf.
Tl, showing that thallium is not carried down with the manganese.

The Test for Titanium Jtfith H^)^ in the Presence of Iron, Cobalt, or Nickel.- -500 mg.
Fe as FeClg were evaporated with a large excess HNO3 (i-42) to 5 cc; the dark red
solution was diluted to 40 cc: it became almost colorless. The solution was divided
into two parts, and to one of these 0.5 mg. Ti as TiCl4 solution was added; to both
parts 2 or 3 cc. H^Oj were added: in both a deep yellow color very clearly indicated
the presence of titanium. — The experiment was repeated, except that the HNO3 solu-
tion was diluted to about 15 cc.: the test for titanium was distinct, although the ferric
solution was not quite colorless.

500 mg. Co as Co(N03), were evaporated with excess HNO^ (1.42) to 5 cc. ; the red
solution was diluted to 15 cc. and divided into two jiarts, to one of which i m^. Ti
was added: 2 cc. 3 per cent. HjOj were added: the solution containing titanium ac-
qiured a reddish color of a distinct yellow tinge.

The experiment was repeated with 500 mg. Ni as Ni(N03),: the green color of the
nickel solution containing titanium changed to olive when the HjO, was added.

P. 62, N. i: Confirmatory Test for Manganese with Lead Dioxide. — To i g. PbO,
and 10 cc. HNOg (1.20) in a casserole were added in separate experiments 0.5, 0.3,
0.1 and 0.02 mg. Mn as MnCl,. Tlie mixtures were boiled gently for alx>ut 2 minutes,
in covered casseroles, and then poured into test tubes: after the PbO, had settled,
the color of KMn04 could be clearly seen, even in the last experiment. — The e3[peri-
meut was repeated without adding MnCl,: a perfectly colorless s<:>lution was obtained.
— ^The series of exfjeriments was repeated, except that HNO3 (^-42) was used: the color
of KMn04 W35 easily distinguished in each case, but was not so pronoimced as with
the more dilute acid. Moreover, on standing the pink color faded slowly in the con-
centrated HNO3.

To determine whether the test would be satisfactory when the manganese was ini-
tially present as MnOj, i mg. Mn as MnClj was treated by P. 61, and the preajmate
collected on an asbestos filter. About i/io of it (1. e., o.i mg. Mn)'was treated by
y. 62: the i^rmanganate color was very distinct.

*P. 62af N. i: Separation of Titanium and Zirconium from Manganese hy .4»«-
monia. — To a mixture containing 50 mg. Mn and 50 mg. Ti as nitrates and 4 cc. HCl
(1.12) in 50 cc. was added NH4OH (0.96) very carefully tmtil the mixture was barely
alkaline to litmus paper: a white precipitate formed. 2 drops more NH4OH (0.96) were
added, and the mixture was heated on a steam bath for 10 minutes: the precipitate
remained white. The solution was filtered. The filtrate was tested for titanium
by adding more ammonia: no more precipitate sejjarated at once, showing that the
precipitation of the titanium was complete. The precipitate was dissolved in hot
HCl, the process was repeated, and the filtrate was tested for manganese by adding
(NH4)S,: a precipitate estimated to contain 3 or 4 mg. Mn separated. The titanium
precipitate was again dissolved and treated in the same way: no manganese was found
in the filtrate, showing that the manganese had been completely removed in twt) sepa-
rations.— ^The experiment was repeated except that the 2 drops of ammonia in excess
were not added: only about 2/3 of the titanium precipitated. — ^The experiment »^5

SYSTEM OF QUAUTATIVE ANALYSIS. 557

repeated, except that NH4OH was added imtil the odor after shaking was distinct:
the NH4OH precipitate was browu, showing the presence of a large aniotmt of man-
ganese.— The la55t exfjeriment was repeated, except that 15 g. NH4CI were also added:
the NH^CJH precipitate was white, and contained only about i mg. Mn. — These ex-
periments prove that a satisfactory separation is obtained only when the hydroxide-
ion concentration is very small.

A solution containing 500 mg. Mn and i mg. Ti as nitrates, and 10 cc. HCl (1.12)
in 60 cc. was made barely alkaline to litmus paper by means of NH4OH (0.96), an ex-
cess uf two drops NH4OH was added, and the mixture was heated for 10 minutes on
a steambath: the precipitate was dark colored. The precipitate was dissolved in hot
HCl and the separation repeated: a small white precipitate resulted, which was prac-
tically free from manganese. It was dissolved in HCl, and a little 3 per cent. H,0,
was added: the solution l)ecaine deep yellow, showing the presence of titanium. — The
experiment was re|>eated with 500 mg. Mn and 2 mg. Zr as nitrates: the results were
the same. The final white precipitate was dissolved in hot HCl, the solution was evap-
oraied to a few drops, and a piece of turmeric paper was dipped in it and dried over
a small llame: it turned pink, proving the presence of zirconiimi.

Separation of Zirconium from Manganese by Ammoniutn Acetate. — A solution of
500 mg. Mn and 5 mg. Zr in 10 cc. HCl (1.12) was just neutralized with NH4OH; 2 cc.
30 per cent, acetic acid were added, the mixture was diluted to 50 cc, 6 cc. 50 per cent,
ammonium acetate were added, and the mixture was boiled: the precipitate was brown,
showing that several milligrams of manganese were present and that the sei)aratioii
WHS unsatisfactory.

Complete Precipitation of Thallium by Ammonium Hydroxide. — 20 mg. Tl as T1^04
were treated with HNO, and HCIO, by P. 61, and then with NH4OH by P. 64:
a dark red flocculent precipitate separated. This was filtered off, the filtrate was
evaporated nearly to drjmess and tested for thallium with KI by *P. 65^.' no pre-
cipitate of Til separated, showing that thallium had been completely precipitated
by the NH4OH. Since thallous hydroxide is soluble, it follows also that thallous
compounds are completely oxidized by HNO, and HCIO3 in P. 61.

P. 64, N. 2: Acti^m of Nitric Acid on Potassium Sulphocyanate. — In each of a num-
ber of test-tubes i cc. HNO3 (i-42) was placed; varying amotmts of water (from o
to 15 cc.) and finally 5 cc. 10 per cent. KSCN solution were added: in each case a red
color appeared slowly in the cold, more quickly on warming slightly, and more quickly
in the more concentrated HNO, solutions. The red solutions were boiled in casse-
roles: the color disappeared quickly, and the solutions remained colorless on cooling.
More KSCN was added: the color reappeared quickly. — ^The experiments were re-
peated with 0.1 mg. Fe as FeCl^: a red color appeared at once in the cold, but on boil-
ing, this color also disappeared in a minute or two. To one of these colorless solutions,
after cooling, was added KSCN: the red color appeared at once. To another was
added 0.1 mg. Ke as FeClg: the solution remained nearly colorless. — These experi-
ments show that on boiling with HNO^, the KSCN is completely decomposed, and
therefore that the red color due to nitrous acid cannot be eliminated by boiling.

P. 65, S. 2, 5 and 6: Precipitation of Titanium^ Zirconium^ and Thallium in the
Ammanium Acetate Procedure, — 0.5 mg. Ti as TiCl4 in 100 cc. was treated by P. 65:
a small precipitate separated, which was proved to contain titanitmi by *P. 656. — ^The
experiment was repeated with 10 and with 500 mg. Ti; the filtrates were tested for
titanium by evaporating with a httle H^Of to fuming, cooling and adding H/),: no
color appeared.

A mixture containing 500 mg. Fe as FeClj and 20 mg. Zr as ZrCl4 was treated by
P. 65: a large amount of zirconium was found in the precipitate, but on adding NH4OH

558 A. A. NOYES, W. C. BRAY AND B. B. SPEAR.

to the filtrate a white precipitate was obtained which was estimated to contain 3 tu
5 nig. Zr. — ID mg. Zr as nitrate in HNO, solution were treated by P. 65, except that
no FeCl, was added; NH^OH was added only until a very small precipitate of m-
coniiun hydroxide separated and the solution was still add when the ammonium
acetate was added ; the filtrate from the basic acetate predptate was made alkaline
with NH4OH to test for zirconitun: a white predpitate separated, which was estimated
to contain i or 2 mg. Zr. — ^The experiment was repeated with 1 and 2 mg. Zr, except
that NH4OH was added until the solution was neutral; the filtrate was made add,
evaporated to 20 cc. and then made alkaline with NH4OH : no predpitate separated.—
The last experiment was repeated with 5 nig. Zr as nitrate and 20 mg. PO4 as NaJHP04:
no zirconium was found in the filtrate. — For the fact that zirconium is not predpi-
tated in the presence of considerable acedc add, and that it then prevents the com-
plete predpitation of titanium, see Hillebrand, BuU. U. S. Geol. Survey, 176, 72
(1900).

20 mg. Tl as TlCl, were treated by P. 65, except that no FeCl, was added : a brown
predpitate separated on boiling and also on adding NH4OH ; the former was two or
three times as large as the latter, thus showing that at least 5 mg. Tl were not pre-
dpitated till the NH4OH was added. The filtrate was evaporated nearly to dryness;
H^S04 was added, and the soludon when cold was tested for thallium with KI and
Na^O, by *P. 65 rf; a small predpitate of TU was obtained which was estimated to
contain 0.5 to i mg. Tl. — The experiment was repeated with i mg. Tl: no predpitate
residted on boiling, nor on adding NH4OH. The filtrate was proved to contain the
thallium. — ^Thc experiment with i mg. Tl was repeated, FeClj being added as directed
in the procedure: no precipitate separated on adding NH4OH to the filtrate. Both
predpitate and filtrate were tested for thallium : about half of the thallium was found
in the predpitate and the remainder in the filtrate. — The last experiment was repeated
with o 5 mg. Tl: a small but distinct test was obtained for thallium both in the pre-
dpitate containing the iron and in the filtrate.

♦P. 65a, N. I to j: Extraction of FeCl^ by Ether. — See Rothe, Stahl und Eisen, 12,
1052 (1S92); 13, 333 (1893); Langmuir, /. Am. Chem. Soc, 22, 102 (1900); Kern, J.
Am, Chem. Soc, 23, 689 (1901).

To determine the best concentration of HCl to use in this extraction, a series of ex-
periments was perfoniied as follows: 500 mg. Fe as FeCl, were dissolved in 30 cc.
HCl of known specific gravity and this solution was treated several times with about
35 to 40 cc. ether in a separating ftmnel as described in the procedtu'e. 'J^e amount
of FeClj extracted in each treatment was estimated by evaporating the ether extract
to dryness, adding HCl, and predpitating with NH4OH. The results are given in
the following table. Since the ether used contained initiaUy no HCl, the amount of
HCl in the water layer is somewhat decreased in each treatment. The proportion
of FeCl, extracted in each treatment was 4 to 5 per cent, with 8 per cent. HCl
(sp. gr., 1.04); 95 to 96 per cent, with 18 per cent. HCl (sp. gr., 1.09); 99 per cent
with 22 per cent. HCl (sp. gr., i.ii); and 94 per cent, with 25 percent. HG (sp.
gr., 1. 125).

\^th the 22 per cent, add (sp. gr., i.ii) the first extraction took out nearly all the
FeCl, (495 mg. Fe), the second extraction nearly all of 5 mg. Fe remaining, and the
third only a fracdon of a milligram. A similar result was obtained when 500 rag. Fe
as FeCl, were predpitated with NH4OH, the predpitate dried between filter papers,
and dissolved in HCl (1.12). This result is in agreement with that of Rothe, who
recommends an add of sp. gr. i.ioo to 1.105, ^^^ he shook his ether with HCl of this
strength before making the separation.

SYSTEM OI^ QUAUTATIVE ANALYSIS. 559

Molsculor Formula of Ferris Chloride in Ether Solution. — See Beckmann, Z. physik.
Ckem,, 46, 860 (1903).

ExbracUon of ThaUic Chloride by Ethet.— 15 mg. Tl as Tl(OH), were treated by *P.
65a; the solution of TlCl, in HCl (1.12) had a yellow color, but this layer was
colorless after it was shaken with ether; the ether layer had a yellowish color. The
first ether extract contained nearly all the thallium, the second ether extract contained
less than i mg., and the water layer after the two extractions was practically free from
thallium.

*P. 6sa, N. 4: Behavior of Titaniumf Zirconium, and Uranium in the Ether Treat-
ment.—-A solution of 50 mg. Ti as TiCl4 in one experiment, and of 50 mg. Zr in an-
other, in 30 cc. HCl (i.ii) was shaken with an equal volume of ether; the ether layer
was evaporated to dryness in a casserole on a waterbath, HCl (1.20) was added and
heated, the solution was made alkaline with NH^OH : no precipitate separated in either
case, showing that no titanium or zirconiuiu had dissolved in the ether layer. In
the case of titanium the water layer became reddish yellow owing to the presence
of H2O, in the ether, and in the case of zirconium a white precipitate (ZrCl4?) separated
in the water layer, but when the ether was expelled by evaporation a clear solution
resulted. A white precipitate was also found to result with titaniiun in an experi-
ment in which 300 mg. PO4 were also present.

For proof that uranium is not extracted, see Kern, J. Am. Chcm. Soc., 23, 689 (1901).

*P. 6^, N. 5: Extraction of Iron by Ether in the Presence of Phosphate, — 500 mg. Fe
as FeCl, and 300 mg. PO^as ammonium phosphate were dissolved in 30 cc. HCl (1.09)
and treated wth ether (35 cc.) as described in the procedm-e: about 95 per cent, of
the iron was extracted in each treatment and the phosphate remained in the water
layer.

♦P. 656, N. i: Nature of the Compound of Hydrogen Peroxide with Titanium. —
In regard to the colored solution, and the preparation of sohd TiO„ see Schonn, Z.
analyt. Chem., 9, 41 (1870); Classen, Ber., 21, 370 (1888); Levy, Ann. chim. phys. (6)
25* 463 (1892); Melikoff and Pissarjewsky, Ber,, 31, 953 (1898).

Blr. Chas. Field, 3rd, working in this laboratory, performed the following migration
experiment. A colored solution was prepared by rotating for three hours pure TiO„
which had been dried over VJD^, with a normal HNO, solution which contained i
mol. H,0, per liter. The resulting solution was 0.087 molal with respect to TiO,.
A large U tube which was partly filled with normal HNO, solution was placed in a
thermostat at 25^ and the titanimn solution was carefully introduced through a tube
at the bottom of the U. The surfaces of contact of the two solutions were sharply
maxked. After a current had passed for 8 hours, one of the surfaces of the red solu-
tion had moved 4 cm. upward toward the cathode, and the other botmdary had
moved away from the anode a nearly equal distance. This proves that the color is
due to a cathion containing titanimn.

Delicacy of the Hydrogen Peroxide Reaction for Titanium.— 0.1 mg. Ti as chloride
was treated by *P. 656; a distinct yellow color appeared on adding the H,0, solu-
tion.

*P. 63b f N. 2: Precipitation and Separation of Titanium and Zirconium as Phos-
phate.—Set Hillebrand, Bull. U. S. Geol. Survey, 176, 75 (1900).

0.5 and I mg. Zr and a mixture of i mg. Zr and 100 mg. Ti as chloride were treated
by *P. 656; flocculent precipitates formed in every case within 10 minutes but were
more distinct after half an hotur. — 100 mg. Ti alone were treated by ♦P. 656; no pre-
cipitate separated in several hours.

10 rag. Zr as chloride were treated by *P. 656; the precipitate was filtered off after
half an hour, and the filtrate made alkaline with NH4OH : a small precipitate estimated

56o A. A. NOYISS, W, C. BRAY AND E. B. SPEAR.

to contain about 0.3 nig. Zr separated. The result was the same when the filtratioo
wsis made after 20 hours. — The experiment was repeated, except that only 3 or 4 cc.
phosphate solution were used: the filtrate contained 0.5 to i.o mg. Zr after half an
hour and after i hour. — ^The experiment was repeated, using about 15 cc. phosphate
solution; only o.i to 0.2 mg. Zr remained in the filtrate after 2 hours.

To a niuiiber of solutions, containing i mg. Zr as chloride and varying amounts
of HJ1SO4 (1.20) in 15 cc, were added 5 cc. 7 per cent. Na,HP04.i2H,0 solution: wth
I and 2 cc. of add the solutions became turbid at once, and fiocculent precipitates
settled out within 20 minutes; with 5 cc. acid the solution remained clear for about
10 jninutes, but after i hour there was a distinct precipitate; with 10 cc. acid the solu-
tion remained clear for a longer time and only a minute precipitate separated in 1.5
hours.

Precipitation of Titanium with Zirconium Phosphate. — 10 mg. Zr and 10 mg. 11 as
chlorides were treated by *P. 656; the precipitate when collected on a filter was dis-
tinctly yellow and this color remained after washing with water for an hour. About
10 cc. 15 per cent. HF solution were poured through the filter in a celluloid funnel
and the solution was treated again by *P. 65b: the phosphate precipitate was wlute,
and the titaniimi in the solution was estimated from the color to be about i mg. — 0.1
mg. Ti and 100 mg. Zr as chlorides were treated by *P. 656; the solution became dis-
tinctly yellow on adding H,0„ and the filtrate from the phosphate precipitate was
also yellow, showing that titanium is not completely carried down by the zircuniuxn.
(100 mg. Zr alone gave no color with HjO^.)

*P. 6^b, N. 3: Precipitation of Thorium as Phosphate. — 2, 5, and 50 mg. Th as ni-
trate were treated by ♦?. 656.- white gelatinous precipitates separated on the addi-
tion of Na^PO^, the precipitate being small and forming slowly in the experiment
with 2 mg. Til. In the experiment with 5 mg. a 10 cc. portion of HF (i volume 45
per cent. HF to 2 volumes water) was poured several times through the filter; the solu-
tion was evaporated with H^SO* to fuming, cooled, diluted and excess NH^OH added:
no precipitate separated at once and only a very small one on standing an hour.

♦P. 656, N. 4: Behavior of Manganese, Cobalt, Uranyl and Vanadyl SalU in Ae
Tests for Titanium and Zirconium. — 100 mg. Mn as MnCl, were treated by ♦?. 6sb~c:
with H,0, no color resulted, and no precipitate formed when Na^HPO^ was added,
nor on standing i hour. After the addition of just sufficient powdered Na^Oj to re-
duce the H,Og (determined by te.<>ting portions of the solution with Ti solution), no
precipitate formed, but when about i g. more NagSO, was added a large precipitate
separated. — ^The experiment was repeated separately with 100 mg. Co as CoCl, and
with 5 mg. U as UO,(NO,),: there was no change of color with H,0, and no precipi-
tate with Na,HP(\ on adding enough Na^O, to destroy the H,0,. — ^The experiment
was repeated with 100 mg. U as U02(N03),: the solution was distinctly yellow before
the HgO, was added, and a large white precipitate formed on adding just sufficient
Na^O, to decompose the HjO,.

10 and 100 mg. V as Na4V04 were treated in separate experiments by ^P. 65fr-c:
with the 10 mg. the color obtained on adding H^Os was similar to that obtained vnih
3 to 5 mg. Ti; with the 100 mg. however the color was of a much redder shade than
that with titanitun; on the addition of an excess of NagSO, the color changed at once
to blue, showing the presence of a vanadyl salt, and no precipitate had separated
in either case after several hours in the cold or on boiling.

♦P. 63c, N. i: Behavior of Titanium in Acid Solutions towards Sodium Phosphate.—
To a solution containing 0.5 mg. Ti as chloride and 2 cc. H^O^ (1.20) in 10 cc. were
added 10 cc. 7 per cent. NajHP04.i2H20 solution: the solution remained ckar for
10 minutes, but had become distinctly turbid in half an hour. Several solutions,

SYSTEM OJf QUALITATIVE ANALYSIS. 56 1

containing 10 mg. Ti as chloride and varying amounts of H^O^ (1.20) were treated
in the same way; the flocculent precipitates were filtered off after 10 minutes, and the
filtrates were made alkaline with NH4OH: in the experiment with 2 cc. acid the fil-
trate contained 0.5 to o.i mg. Ti; in that with 3 cc. acid, 1 to 3 mg. Ti; in the experi-
ment with 5 cc. add, 3 to 4 mg. Ti.

10 mg. Ti as chloride were treated by ♦?. 6sl>-c, and after half an hour the mixture
was filtered, and the filtrate made alkaline with NH4OH: a very small precipitate
was obtained containing about 0.5 mg. Ti. — Tlie experiment was repeated with 0.5
mg. Ti* the solution became distinctly turbid on decolorizing with H^Oa. Half of
the turbid solution was heated to boiling: a distinct, flocculent precipitate was ob-
tained. The other half was allowed to stand in the cold for half an hour: the pre-
cipitate in this case also became somewhat fiocculent.

♦P. 63d, N, i: Test for Thallium with KI. — 0.5, 0.2, and o.i mg. TI as Tlj^04 were
treated by *P. 6$d, the total volume being about 10 cc: a distinct yellow finely di-
vided precipitate of Til was obtained in each experiment, even in that with 0.1 mg.
TI. For proof that 500 mg. ferric iron does not interfere with the test, see T. A., No.

174-^.

*P. djd, N. j: Flame Test for Thallium. — 0.5, 0.2, and 0.1 mg. TI weie precipitated
by ♦?. 65^ as TU. The precipitates were collected on small filters and washed twice
with a very little water. The moist filter was removed from the funnel, a looped
platinmn wire was drawn across its surface to collect a little of the precipitate, and
introduced into a colorless gas flame: with 0.5 mg. TI, the momentary green color
was generally seen, but it was sometimes obscured by the yellow flame, due to sodium
and to small fibers of paper; with 0.2 and 0.1 mg. TI the green color could not be de-
tected with certainty. — ^The experiments were repeated, except that the Til precipi-
tates were collected on hardened filters: the green flame was much more brilliant with
0.5 and 0.2 mg. TI than in the corresponding experiments with ordinary filter paper;
it could, however, scarcely be seen with 0.1 mg. TI.

P. 69, N. i: Potassium Cohaltic Nitrite. — See Fisher, Pogg. Ann., 74, 115 (1848);
Sadtler, Am. J. Sci. (2), 49, 196 (1870); Rosenheim and Koppel, Z. anarg. Chem.,

17, 35 (1898).
P. 69, N, 2: Precipitation of Cobalt with Potassium Nitrite. — 0.1 and 0.3 mg. Co

as CoCl, were treated by P. 69: in each experiment the solurion became distinctly
turbid within 5 minutes. — The experiment was repeated in the absence of cobalt:
the solution remained perfectly clear. — For the detection of 0.5 mg. Co in the pres-
ence of 250 mg. Ni in P. 69, see T. A., No. 127.

500 mg. Co as nitrate were treated by P. 69: the mixture was shaken well, allowed
to stand, and filtered after about half an hour, and again allowed to stand : a considerable
precipitate again separated. — ^The experiment was repeated except that the mixture
was heated ou a waterbath to 50 or 60^ with frequent shaking for half an hour;
it was allowed to cool and filtered: no precipitate separated in the filtrate even on stand-
ing over night.

Separation of Nickel from Cobalt with Potassium Nitrite. — .soo mg. Co and about
5 mg. Ni as nitrates were treated by P. 69 ; after standing 20 hours the mixture was
filtered; the filtrate was evaporated almost to dryness with HCl, and made alkaline
with NaOH: only a very small green precipitate separated corresponding to not more
than 2 uig. Ni. Half of the cobalt precipitate obtained was treated by P. 70: a very
good test for nickel was obtained. — ^The experiment was repeated except that the mix-
ture was heated on a waterbath for half an hour: less nickel was foimd in the filtrate
than in the preceding experiment.

Precipitation of Potassium Nickelous Nitrite. — 250 mg. Ni (previously freed from

562 A. A. KOYltS, W. C. B&AV AND U. B. SP^AR.

cobalt by a KNOg treatment) were treated by P. 69, except that^the volume was cut
down to 50 cc. without however altering the total amounts of reagents used: a dis-
tinct reddish colored precipitate separated within 20 minutes which was proved to
contain nickel but no cobalt by the borax bead test. — For the action of HNO, on nickel
salts and the formation of K^NiCNO J^ see Lang, /. f}rakt. Chem., 86, 299 (1862) ; Hampe,
Lieb» Ann., 125, 346 (1863); and Reichard, Chem.-Ztg., 28, 479, 885, 912 (1904).

P. 70 f N. 2: Delicacy of Hypobromite Test jor Nickel. — 150 mg. Co free from nickel
were treated by P. 70: no precipitate was observed, not eveu on filtering. Vat ex-
periment was repeated except that 0.15 mg. Ni as NiCl, was added: the solution be-
came dark colored on adding excess NaBrO, but no precipitate collected ; it was easily
seen, however, on the filter. — 0.2 mg. Ni as NiCl, was treated by P. 70: a distiuct pre-
cipitate was obtained on filtering.

Separation of Nickel frotn Cobalt by Hypobromite. — 200 mg. Co and 0.5 mg. Ni as
chlorides were treated by P. 70; the precipitate was tested for cobalt in the borax
bead: no blue color was obtained.

A large utmiber of experiments were performed to determine the proper conditions
for making this separation. In working with cobalt free from nickel, it was found
that a precipitate of Co (OH), always formed when the NaBrO reagent (or bromine
water and NaOH) was added very soon after the addition of KCN, and that the neces-
sary interval of time was greatly shortened by increasing the excess of KCN added.

The following experiments show that excess of NaBrO is essential to the precipita-
tion of nickel, i mg. Ni as Ni(NO,), was treated by P. 69; the NaBrO solution was
added in small portions and after the addition of each portion the solution was tested
with the starch KI paper: as long as this paper remained colorless no precipitate of
Ni(OH)j separated, but after the precipitate formed the paper became blue when dipped
into the mixture. — ^The experiment was repeated with 10 mg. Ni. As long as the pre-
cipitation was incomplete the paper remained colorless, or only a small brown ring
was formed on the paper, but after complete precipitation all of the paper immersed
in the solution became brown or blue.

P. 70, N. 4: Action of H^ on Alkaline Tartrate Solutions Containing Nickel or Co-
balt.— Villiers, Compt., rend, xiq, 1263 (1894); 120, 46 (1895), foimd thpt when H^
was passed into a NaOH containing freshly precipitated Ni(OH), (but no tartrate)
the hydroxide was quickly converted into black nickel sulphide, but that a portion
of the nickel passed into solution giving a deep brown color, proving that the pres-
ence of tartrate is not essential for the formation of the brown solution. — 0.5 mg.
Ni as nitrate was treated by both parts of P. 70: a deep brown solution was obtained
on saturating the alkaline tartrate solution in a test tube with HjS. — o.i and 0.2 mg
Ni as nitrate were dissolved in a little HNO3 ^"^ treated by the second paxagraph
of P. 70, about 5 cc. of 10 per cent, tartaric acid and 5 cc. excess of NaOH being added:
clear dark yellow solutions were obtained on saturating with H^. — ^The experiment
was repeated with i mg. Ni: on passing in H^ the solution remained nearly colorks
for about i minute, but finally a clear dark brown solution resulted. — ^The experiment
was repeated with 20 mg. Ni: on saturating with H^S the liquid in the test-tube was
opaque and almost black in color. The liquid was filtered : very little precipitate re-
mained on the filter. It was allowed to stand several hours: a black precipitate sepa-
rated but the filtrate was still black and opaque.

20 mg. Co as CoCl, in a little dilute HNO, were treated by the second paragraph
of P. 70: a black precipitate separated as soon as the H^ was led in and the cobalt
was completely precipitated within i minute. The mixture was filtered, the filtrate
was saturated with HjS, and the test-tube corked and set aside: the solution remained
colorless for several hours. A similar solution was exposed to the action of the air

OXALATE oi^ The rare earths. 563

in an open flask: it became dark yellow in about i hour owing to oxidation of the sul-
phide and consequent formation of* polysulphide. — The experiment was repeated
except that i mg. Ni as NiNO, was also present; the excess of 10 per cent. NaOH was
4 or 5 cc.; H^ was led into the solution for about i minute and the CoS filtered off :
the filtrate was nearly colorless. This was saturated with H^: it became brown,
the color being such as to indicate that very Uttle of the nickel had been carried down
with the cobalt. — This experiment was repeated except that the H^ was led through
the solution for 5 minutes before the CoS was filtered off: the filtrate was of a lighter
brown than before, indicating that over half the nickel had been cairied down with
the cobalt. — The experiment was repeated except that the CoS was not filtered off
till after 10 minutes: the filtrate was almost colorless, and remained so on saturating
again with H^. — The experiment was repeated, the CoS being filtered off after half
an hour: the filtrate was colorless and contained no nickel. — In a similar series of ex-
periments in which a smaller excess of NaOH than 4 to 5 cc. was added, the filtrate
was Hght brown after i minute, and nearly colorless after 5 minutes, thus showing
that there is more danger of losing nickel when the excess of alkali is small, in which
case the brown solution is fonned more quickly. — This result that NiS is deposited
on CoS after the separation of the latter was confirmed by several experiments.

Similar experiments were made with a mixture of 20 mg. Fe as FeCl, and i mg.
Ni, and nith one of 20 mg. Mn and i mg. Ni: good tests for nickel were obtained in
both cases.

[Contribution prom ths Chbmical Laboratory op Harvard Collegb.]

THE CARRTHTG DOWN OF SOLUBLE OXALATES BY OXALATES

OF THE RARE EARTHS.

Bt Grboort Paul Baxtbr and Hbrbkrt Wilkbns Daudt.

Received January 23, 1908.

In a recent investigation^ it has been shown that neodymium oxalate,
when precipitated in neutral or nearly neutral solution by means of
ammonium oxalate, carries down considerable quantities of this salt,
and that the amount carried down increases with increasing concen-
tration of molecular ammonium oxalate at the moment of precipita-
tion. Furthermore, it was shown that neodymium oxalate has no ten-
dency to carry down molecular oxalic acid, and that occlusion of ammo-
nium oxalate may be prevented by diminishing the molecular concen-
tration of the latter salt with a strong acid before precipitation. Other
rare earth oxalates were found to exhibit a like tendency to occlude
ammonium oxalate. Since it seemed probable that the carrying down
of sodium and potassium oxalates^ would vary with conditions of pre-
cipitation in a similar manner, the following investigation was under-
taken to test this point.

The method employed was to precipitate the rare earth oxalate under
different conditions, and to analyze the precipitated oxalate by deter-

* Baxter and Grifiiii, This Journal, 28, 1684 (1906).

' The well-known fact that the oxalates of the alkalis are carried down by the
oxalates of the rare earths was first noted by Sheerer. Pogg. Ann. [2], 56, 496 (1842).

564 GREGORV P. BAXTER AMD HERBERT W. DAUI>T.

raining the ratio of metallic oxide to C2O3. The atomic weight of the
metal being known, the excess of C2O3 could be calculated and hence
the purity of the precipitate. • Although all the rare earth oxalates con-
tain crystal water, even after drying at an elevated temperature, the
amount of this water is immaterial for the purpose in hand.

Since none of the rare earth specimens were pure, it was necessary
to determine the average atomic weight of each sf)ecimen. This was
done by the oxalate method first proposed by Stolba. ^ A hot solution
of oxalic acid containing a small quantity of nitric acid was slowly added
with constant stirring to a hot nitric acid solution of the rare earth until
precipitation was complete. After the precipitate had been washed
ten times by decantation with hot water, it was collected upon a porce-
lain Gooch crucible provided with a disk of filter paper in place of an
asbestos mat, and was dried in an electric air-bath at about 125° for
twenty-four hours. Shortly before being weighed out for analysis, the
dried precipitates were thoroughly mixed by grinding in an agate mortar
to insure uniformity in water content,^ and all portions of the same ma-
terial were weighed out at the same time, in order to avoid error from
hygroscopicity.

The per cent, of oxide in the oxalate was determined by igniting weighed
portions of about one- half gram in platinum crucibles, while the ratio
C3O3: oxalate was found by dissolving weighed amounts of the oxalate
in hot 2 N sulphuric acid and titrating the oxalic acid with standard
potassium permanganate. From the ratio MjO, : 3C2O3 the average
atomic weight of the sample was calculated, the following atomic weights
being assumed: H = 1.008, O = 16.00, C = 12.00.

The permanganate solution was standardized with oxalic acid which
had been allowed to come to constancy over sulphuric acid of the specific
gravity 1.35. This oxalic acid had been three times recrystallized,
with centrifugal drainage. Although the permanganate solution changed
in concentration ver\' slowly, re-standardization was carried out fre-
quently.

In order to gain some idea of the extent to which the alkali oxalates
are carried down by the rare earth oxalates under conditions most fawr-
able for this effect, each material under investigation was precipitated
by pouring a hot nearly neutral solution of the nitrate of the rare earth
into a hot solution of from three to four times the equivalent quantit}*
of each alkah oxalate. In this precipitation the solution of the rare earth
nitrate was on an average about four-tenths normal and the solution
of the alkali oxalate about six-tenths normal.

Next the same material was again precipitated in as nearly as possibk

' Sitzber. bdhm. Ges., Dec, 1878; also Chem. News, 41, 31 (1880).
• Gibbs, Proc. Amer. Acad., a8, 262 (1893).

OXALATES OI^ THE RARE EARTHS. 565

the same way, except that about twice the equivalent quantity of nitric
acid was added to the solution of the alkali oxalate before the precipita-
tion. Although the dissociation of the first hydrogen of oxalic acid is
ver}' considerable/ that of the second hydrogen is small.' Hence, in
the presence of a high concentration of the hydrogen ion the concentra-
tion of the oxalate ion and therefore that of the molecular alkali oxalate
must be very low, so that little carr\4ng down of the alkali oxalate is
to be expected.*

With solutions as concentrated as the above it is of course probable
that a small amount of soluble oxalate would be "included" in cells of
mother liquor in the highly crystalline precipitates. This may be the
reason for at least a portion of the smaU amount of alkali oxalate found
even in precipitates formed by methods calculated to reduce occlusion
to a minimum.

As in the previous investigation, the first element studied was neodym-
ium. The sample used was not quite pure, its absorption spectrum
showing, besides the bands of neodymium, traces of those of samarium
and praseodymium. Other elements giving no absorption spectrum
may have been present. The average atomic weight of the sample was
found by analysis of the oxalate as described above.

Atomic Weight.

I. II. III. IV. Average.

Per cent, of Nd,0, 53.87 53.81 53.77 53-82

Per cent, of CjOj 34.29 34.36 34.30 34.37 34.33

Ratio M,0,: sCjO,- 1 .5677. M = 145.3.

Although the average atomic weight of the sample is only slightly
higher than the most probable atomic weight of neodymium, 144.5,*
this result does not indicate with exactness the purity of the sample,
for the atomic weights of the known impurities, praseodymium and
samarium, lie on opposite sides of that of neodymium. The specimen
undoubtedly consisted chiefly of neodymium, however.

A portion of the same material was next precipitated by pouring a
hot nearly neutral solution of the nitrate into a hot solution of about
three times the equivalent quantity of potassium oxalate. Portions
of the carefully washed and dried precipitate, when held in the Bunsen
flame, indicated the presence of considerable quantities of potassium.
A fruitless attempt to expel all the potassium by prolonged ignition in
the flame of a blast lamp, showed that some other method was necessar^^

* Ostwald, Z. physik. Chem., 3, 281 (1889).
*^Wd., 9, 553 (1892).

' In the previous paper the effect of increasing the hydrogen ion concentration
was erroneously imputed to the formation of molecular oxalic acid instead of the acid
oxalate ion.

* V. Welsbach, Sitzb, Akad. Wiss, Wien, 112, 1037 (1904).

II.

III.

Average.

46.76

46.70

46.74

37.61

37.66

29.82

7-84
18. xo

566 GREGORY P. BAXTER AND HERBERT W. DAUDT.

for the determination of the neodymium oxide. Accordingly, weighed
amounts of the precipitate were first dissolved in either sulphuric or
nitric acid, and, after the solution had been neutralized with freshly distilled
ammonia, ammonium oxalate was added until precipitation appeared
complete. More ammonia was then added until the solutions were
slightly ammoniacal. After standing some time, the precipitates were
washed several times with hot water, filtered, ignited and weighed. Oc-
cluded ammonium oxalate was, of course, volatilized during ignition.
The oxalic acid was determined as before described.

Neutral Precipitation with Potassium OxaItATB.

Per cent, of Nd,0, 46. 76

Per cent, of C,0, 37 . 71

Per cent, of C,Og equivalent to 46.74 per

cent, of Nd,0,

Excess per cent, of C,0,

Per cent, of Kj^fi^ carried down

The experiment with potassium oxalate was then repeated with sim-
ilar solutions except that the potassium oxalate solution before pre-
cipitation was made acid with about twice the equivalent amount of
nitric acid. It has already been shown that neodymium oxalate shows
no tendency to carry down oxalic acid.^ This new precipitate of neodym-
ium oxalate gave no visible flame test for potassium, and analysis
showed only 0.30 per cent, of potassium oxalate to have been carried
down. The per cent, of neodymium oxide was found by ignition of
weighed portions of the oxalate, finally with a blast lamp to expel traces
of potassium. This method is later shown to give accurate results in
the case of a precipitate formed with sodium oxalate.

The result of this experiment, in which the carrying down of potas-
sium oxalate is a little less than two per cent, as large as in neutral so-
lution, is in accord with the prediction, and also with the behavior of
neodymium oxalate with ammonium oxalate previously observed, the
excess of oxalate found not being greater than could be accoimted for
on the basis of inclusion.

When sodium oxalate in neutral solution was used as precipitant,
the neodymium solution being added to a large excess of oxalate, the
dried precipitate gave scarcely any flame test for sodium. The same
surprising result was obtained in two successive repetitions of the ex-
periment. One of the precipitates, upon analysis, proved to contain
only a few tenths of a per cent, of sodium oxalate. The analyses for
neodymium oxide were made by ignition of the oxalate with a blast
lamp.

^ Baxter and Grifi&n, Loc. cit.

OXALATES OF THE RARE EARTHS. 567

In strongly acid solution with sodium oxalate as precipitant, exactly
the same proportion of sodium oxalate was found in the precipitate.
The neodymium oxide was determined by ignition. In one of the anal-
yses the neodymium oxide, after being weighed, was dissolved in nitric
acid and the neodymium was precipitated with ammonium oxalate.
The weight of the oxide obtained by ignition of this latter precipitate
agreed essentially with that of the original, showing that small quan-
tities of the occluded alkalies may be volatilized completely by ignition.

Since the solubility of the oxalates of the rare earths in nitric acid is
a variable one, and since our material was known to be a mixture, at
the end of the experiments with neodytmvLtn, the average atomic weight
of the material which had been through the preceding operations was
redetermined, and was found to have diminished 0.3 to 145.0. It is
to be noted that the effect of this diminution is to exaggerate slightly
the occlusion. With sodium oxalate in acid solution for instance, the
per cent, of sodium oxalate carried down, calculated upon the basis of
the lower atomic weight, is 0.3 instead of 0.4.

Occi^usiON BY Neodymium Oxalate.

K9Cs04. Na2C«04.

Per cent. Per cent.

Neutral predpitation 18. i 0.4

Acid precipitation 0.3 0.4

Lanthanum upon examination was found to behave similarly to neodym-
ium. The material used was essentially free from elements whose solu-
tions absorb in the visible region, and its atomic weight, determined as
in the case of neodymium to be 139. i, was found to be very close to the
probable value of this constant, 138.9.

Since the carrying down of ammonium oxalate by lanthanum oxa-
late was not investigated in the previous research, this point was taken
up here. The precipitate formed by adding a nearly neutral solution
of lanthanum nitrate to a large excess of ammonium oxalate proved to
contain a considerable amount of ammonium oxalate. When, how-
ever, the ammonium oxalate solution was acidified with twice the equiv-
alent quantity of nitric acid, only a trace of ammonium oxalate was found.

Similar results were obtained with potassium oxalate as precipitant,
this oxalate being carried down in considerable quantities from neutral
solution and very slightly from strongly acid solution. In the analyses
of the precipitate from neutral solution the lanthanum oxide was deter-
mined by igniting weighed portions of the oxalate, leaching the oxide
with water and filtering the wash water through a tiny filter, and finally
igniting both oxide and filter paper. In analyzing the precipitate from
add solution the lanthanum oxide >vas determined by ignition only.
When sodium oxalate was used as precipitant, only small quantities

568 GREGORY P. BAXTER AND HERBERT W. DAUDT.

of this substance were found in the precipitates of lanthanum oxalate
either from neutral or from acid solution.

Occlusion by Lanthanum Oxalate.

(NH4),C,04 K,C,04. NtiCaO^.

Percent. Percent. PereeoL

Neutral precipitation % 5.4 2.3 0.5

Add precipitation 0.2 o. i 0.8

It is noticeable that the carrying down of both potassium and am-
monium oxalates by lanthanum oxalate is very much less than by neodym-
ium oxalate under nearly the same conditions of precipitation.

The third material examined was of somewhat complex nature. A
solution of gadolinite earths, which had been treated with potassium
sulphate, was fractionated with magnesium oxide until about half the
earths remaining had been precipitated. The magnesium oxide frac-
tions, which seemed identical as far as spectroscopic evidence was con-
cerned, were combined and fractionally crystallized from concentrated
nitric acid until the greater part of the neodymium and praseodymium had
passed into the mother liquors. A portion of this material, which consisted
largely of samarium, was used in the following work. Subsequent pro-
longed fractional cr\'stallization showed the presence of gadolinium, dyspro-
sium, europium and holmium. This material was investigated in exactly
the same way as in the cases of neodymium and lanthanum, by precipita-
ting the oxalate in both neutral and acid solution with a large excess
of ammonium, potassium and sodium oxalates. The average atomic
weight of this material was found to be 149.3.

Occlusion by Samarium Oxalats.

(NH4)sCs04. K^a04. Jgtifi^*.

Percent. Percent. Percent

Neutral precipitation 5.9 20. 7 5.9

Acid precipitation 0.0 0.0 o. i

From the preceding table it can be seen that the canning down of
the precipitant in neutral solution and almost absolute purity of the
precipitate from acid solution is in accord with the behavior of the ele-
ments previously studied — ^lanthanum and neodymium. It is further
to be noted that the quantity of sodium and potassium oxalates found
in the precipitates from neutral solution is markedly greater in all cases
than was found with neodvmium and lanthanum oxalates. This is
in accord with the fact that elements of the yttrium and erbium groups
in general show marked tendency to form soluble double oxalates with
the oxalates of ammonium and the alkalis.

Finally, a sample of yttria was examined in a similar fashion. This
sample was very crude, its atomic weight being found to be 102.7.

First, precipitation with potassium oxalate in strongly acid solutioo
was investigated. Considerable quantities of potassium in the preap-

OXALATieS OF THE RARE EARTHS. 569

tate were indicated by a strong flame test. Some difficulty was ex-
perienced in the determination of the yttrium oxide in the oxalate, ow-
ing to failure of all attempts to reprecipitate the oxalate completely
from either neutral or slightly acid solution. The method finally adopted
was that of leaching the ignited oxalate with hot water and collecting
the small amount of suspended yttrium oxide upon a tiny filter paper,
as previously described in the case of lanthanum. Even after filtration
the decantate was cloudy, but repeated filtration through the same
filter paper removed all but negligible amounts of the yttrium oxide.
The washed oxide was dried upon the steam-bath, ignited and weighed,
and its weight was added to the weight of oxide upon the filter paper
after ignition. The filtrate was alkaline to phenolphthalein. Over
seventeen per cent, of potassium oxalate was found.

Since the material had been precipitated in acid solution several times
between the original determination of the atomic weight and the acid
precipitation with potassium oxalate described above, the change in
atomic weight produced by partial solubility of the oxalates in nitric
add was determined, and found to be considerable, the new value for the
atomic weight being 104.7. Hence the calculated percentage of potassium
oxalate carried down in the above case is probably slightly too low.

The fraction of material used in the above experiments was now mixed
with a new portion of the original substance, and the average atomic
weight of the mixture was found to be 103.2.

This new material was now precipitated with potassium oxalate in
neutral solution under as nearly as possible the same conditions as before.
The precipitated oxalate contained potassium oxalate in slightly greater
quantities than when formed in acid solution.

Examination of the precipitates formed with sodium oxalate in both
neutral and acid solutions showed that in both cases very small quan-
tities of the precipitant were carried down.

Since 3rttrium oxalate was found to carry down large quantities of
potassium oxalate even in strongly acid solution, an experiment was
performed to determine whether this was the case with ammonium oxa-
late also. Baxter and Griffin have already found that yttrium oxalate,
when precipitated from neutral solution, may carry down as much as
16.5 per cent, of ammonium oxalate. The precipitate of oxalate from
acid solution gave a strong test for ammonia when treated with caustic
soda, and was found to contain a somewhat lesser amount of occluded
ammonium oxalate than the precipitate formed in neutral solution.

Occlusion by YiTRroM Oxalate.

(NH4)aC,04. KjCjOv NasCs04.

Percent. Percent. Percent.

Neutral precipitation 18.0 i.o

Add precipitation 13.6 17.5 0.9

570 GREGORY P. BAXTER AND HERBERT W. DAUDT.

Finally, in order to show conclusively that oxalic acid itself is not car-
ried down by yttrium oxalate, a precipitate was formed by adding a solu-
tion of yttrium nitrate to a concentrated solution of a large excess of
oxalic acid. Analysis of the precipitate showed not only that no carry-
ing down of oxalic acid takes place but also that the average atomic
weight of the material had increased to 106.1. The result of this rise
in atomic weight, as previously stated, is to make the carrying down
of precipitant appear less than it really is.

Since in all cases previously described the precipitates were formed
in hot solution, in order to determine the effect of temperature upon
the occlusion, precipitations with neodymium solutions were made at
ordinary and at boiling temperatures with solutions otherwise identical
In one case a cold solution of neodjnnium nitrate was added to a cold
saturated solution of a large excess of ammonium oxalate, and in a sec-
ond case similar solutions were precipitated boiling hot. The precip-
itate from cold solution gave no test for ammonia when treated with
sodium hydroxide and analysis of the precipitate gave no evidence of
occlusion, while in the second case 2.1 per cent, of ammonium oxalate
was found.

Similar experiments were then carried out with a cx)ld saturated solu-
tion of potassium oxalate and with a similar solution at boiling tem-
perature. The precipitate formed at the lower temperature was found
to contain lo.o per cent, of potassium oxalate while the one formed at
boiling temperature contained nearly double this proportion, 17.5 per
cent.

The effect of high temperature is very marked both with ammonium
oxalate and with potassium oxalate. The smaller quantity of ammo-
nium oxalate found even in the hot precipitation is due at least in part
to the fact that the solutions of ammonium oxalate were more dilute,
owing to the lesser solubility of this salt at ordinary temperatures.

The very considerable extent of the carrying down of the oxalates
of the alkalis and ammonium by all the rare earth oxalates investigated,
points to the formation of double salts as the cause of the phenomenon
rather than to ordinary solid solution. Although the foregoing experi-
ments may not indicate the limiting values of the cx^clusion under the
conditions most favorable and least favorable for the phenomenon, it
is interesting to tabulate the molecnilar ratios between the occluded and
occluding oxalates.

A glance at the following table shows that in no case does the carry-
ing down of a soluble oxalate exceed the proportion of one molecule
of alkali oxalate to one of rare earth oxalate. One might conclude from
this fact that stable insoluble double salts containing more than one

OXALATE OF THE RARE EARTHS. 57 1

molecule of alkali oxalate to one of rare earth oxalate do not exist,
although the grounds for such a conclusion are not by any means final.

MoxxcuLAR Ratio of OcctuDBD Oxalate to Rarb Earth Oxalate.

Na,C,04.

K,C504.

(NH4),Ca04.

Acid.

Neutral.

Acid.

Neutral.

Acid.

Neutral.

Nd,(C,OJ,..

. . 0.02

0.02

O.OI

0.79

O.OI*

0.71*

La,(C,OJ,. . .

.. 0.04

0.02

0.00

0.09

O.OI

0.27

Sni,(C,OJ,. .

. . 0.00

0.30

0.00

0.94

0.00

0.32

Y,(C,OJ,. . . .

.. 0.04

0.04

0.69

0.73

0.65

0.82»

It is somewhat difficult to explain the unexpected behavior of yttrium
oxalate, when precipitated from acid solutions, the occlusion being only
slightly diminished thereby. The oxalates of the earths of the yttrium
and erbium group, however, show considerable tendency to form solu-
ble double oxalates with the oxalates of the alkalis and ammonium, indi-
cating a more marked tendency toward double salt formation than is
possessed by the oxalates of the neodymium group. Since even in the
presence of a high hydrogen ion concentration the oxalate ion concen-
tration and hence that of alkali oxalate must be appreciable, the carry-
ing down of alkali oxalate is still possible where the tendency in this direc-
tion is strong. The fact that the occlusion by yttrium oxalate is not
greater in neutral solution may be explained on the hypothesis previously
stated that there is no tendency to form insoluble double oxalates con-
taining more than one molecule of alkali oxalate to one of yttrium oxalate.

The following general conclusions seem to be justified from the fore-
going results:

(i) The oxalates of the rare earths show marked but varying ten-
dencies to carry down the oxalates of the alkalis and ammonium.

(2) This tendency increases with increasing concentration of molecular
alkali oxalate at the moment of precipitation.

(3) Potassium and ammonium oxalates are carried down to a much
greater extent than sodium oxalate. Precipitation with sodium oxa-
late in most cases gives precipitates only slightly contaminated with
this substance even in neutral solution.

(4) The carrying down of the soluble oxalates is greater at high than
at low temperatures.

(5) By conducting the precipitation in the presence of a quantity of
a strong acid considerably more than equivalent to the alkali oxalate,
thus very much reducing the concentration of molecular alkali oxalate,
the carrying down is in many cases almost wholly prevented. In the
case of yttrium, the diminution in occlusion is slight.

(6) In order to produce as pure as possible a precipitate of a rare earth
oxalate by means of an alkali oxalate or ammonium oxalate, precipita-

* Prom the results of Baxter and Griffin, Loc, cit.

572

VICTOR LENHER.

tion should be conducted in cold dilute solution in the presence of a quan-
tity of a strong acid considerably more than equivalent to the oxahte
added.

We are greatly indebted to the Welsbach Light Company for some
of the rare earth material.

Cambridob, Mass.,
January 20, 1908.

YTTRIUM EARTHS.

[first paper.]

By Victor I«bnhbk.
Received December 38, 1907.

The- methods which we have at our disposal for the separation of the
earths of the yttrium group may be classified under the following heads:
(i) Fractional precipitation; (2) Fractional crystallization; (3) Fiae-
tional decomposition of such salts as the nitrates by heat.

Under fmctional precipitation, we have methods which depend largely
on the differences in basic properties, such as the fractional precipita-
tion by ammonia, magnesia, etc. The speed by which separations are
effected by use of this principle depends largely on how quickly the sys-
tem can be brought into equilibrium.

In the methods of fractional crystallization we must depend necesr
sarily on the differences in solubility of various salts and as a rule with
the mixtures which are found in the rare earth minerals; the solubilities
of a given salt of the various metals are not widely different. On this
account separation by the crystallization of the nitrates or double nitrates
is not rapid, while with the chromates accurate conditions must be ob-
served, in which case this method gives splendid results.

The decomposition of the nitrates by heat is slow, but can, by patience,
be carried out with success. The basic nitrate method of Welsbach^
which can be applied to the yttrium group is a combination of this method
and that of fractional precipitation. It is more rapid and successful
than either method alone.

The successful use of any of the methods for separating the metak
of the yttrium group depends largely on the ratios of the various con-
stituents present in the mixtures, as well as on the character of the ele-
ments to be separated. We note, for example, that Dennis and Daks*
in their study of the yttrium earths from sipylite find that magnesia,
as a precipitating agent, causes little change in the atomic weights and
absorption spectra, while James' was more successful in using this method

* Monatshefte, 5, 508.

" This Journal, 24, 428.

• Ihid., 29, 495.

on gadolinite earths. James used the nitrate solution while Dennis
and Dales used a chloride solution. The author in working in a nitrate
solution with yttrium earths from monazite has found that various frac-
tions acted quite differently toward magnesia, some showed marked
differences in atomic weight and absorption spectra after treatment
with magnesia, while others showed little change. It has been observed,
moreover, that in order to work this method with any appreciable de-
gree of success the magnesia must be freshly ignited. The various de-
grees of success with different methods can also be illustrated by the
Welsbach method^ of crystallizing the oxalates from an ammoniacal
solution or by the James method^ in which the "oxalate-carbonates**
are crystallized from an ammonium carbonate solution of the oxalates.
This method with gadolinite earths in the hands of James yielded first
yttrium and successively fractions with higher atomic weight to ytter-
bium. In the author's hands, it has worked similarly with the yttrium
earths from samarskite but on applying the same method to certain
oxalates from monazite, atomic weight determinations showed that the
elements with heavier atomic weight appeared first while the more solu-
ble portion yielded fractions whose atomic weight was far below the
more insoluble portions. In other words, we here have the same method
producing opposite results with different mixtures of earths.

From time to time, it has been proposed to use salts of organic acids.
The oxalates can be crystallized from either ammoniacal or ammonium
carbonate solution, yielding a fairly rapid method of fractionation, or
the oxalates can be crystallized from nitric acid solution yielding frac-
tions of different atomic weights. The ethyl sulphates and the acetyl
acetonates have been used by Urbain and others as means of separation
in this group. The formates have been repeatedly used for fractiona-
tions. Salts of a number of organic acids have been prepared, but Httle
has been done in the application of the derivatives as means of separa-
tion. Such salts as the tartrates, citrates and succinates have been pre-
pared, but little has been attempted in the way of separation.

The tartrates and citrates of the yttrium earths appear as white gelat-
inous precipitates when a neutral salt of potassium, sodium or am-
monium is added to a solution of the yttrium salt. In a similar manner,
insoluble derivatives are formed with neutral salts of fumaric, maleic,
tartronic, malic and malonic acids. The neutral succinates of the al-
kalis or ammonium deport themselves in a very interesting manner
with the neutral nitrates of the vttrium earths.

When neutral ammonium or sodium succinate is added to a neutral
nitrate solution of the yttrium earths and the solution allowed to stand,

^ Ifonatshefte, 27, 935.
» This Journal, 29, 495.

574 VICTOR LENHER.

a finely divided crystalline precipitate of the succinates appears. This
insoluble precipitate forms slowly, in fact, in the cold a few hours are
necessary to insure complete precipitation. On the other hand, when
the solution is hot or boiling, complete precipitation is effected in much
less time, from ten minutes to half an hour being sufficient time for
complete formation of the insoluble succinates. The ready formation
of this finely divided precipitate and the fact that the reaction is
far from instantaneous appears to us as promising to be a satisfacton-
method for fractionation. The fact that it forms as slowly as it does,
would indicate that there should be plenty of time for equilibrium to
be established and the physical character of the salt and its insolubility
enables it to be quickly filtered and readily washed.

That the yttrium earths form succinates was shown by Berlin in 1835.^
He showed that with sodium succinate the yttrium earths form a fine
crystalline powder. Ekeberg in 1802' thought that the yttrium earths
were not precipitated by the alkaline succinates while beryllium was,
which was contrary to the results found by Berlin and to the work of
Cleve and Hoglund* who showed that ammonium succinate precipitates
yttrium but not erbium from nitrate solution, but out of a mixture pre-
cipitates both.

In the thorium cerium group, BerzeHus showed in 1829* the forma-
tion of an insoluble succinate of thorium. This reaction has been later
studied by Kaufmann^ and Schilling.* The use of an alkaline succinate
has been recommended as a means of separation of iron from the gado-
linite earths by Gadolin, Vauquelin, Berzelius, Berlin, and Hermann,
after Klaproth had shown that iron would be first precipitated from such
a solution.

The yttrium earths from samarskite have been studied with the view
of testing the applicability of the succinates as a means of separation
in this group.

Treatment of Samarskite.

Fifteen pounds of samarskite, containing very little gangue minerals
were treated with concentrated hydrofluoric add according to the method
of J. Lawrence Smith. The mineml dissolved with effervescence. The
columbium and tantalum passed into solution while the earths appeared
as insoluble fluorides. These insoluble fluorides were thoroughly washed
with water by decantation, after which they were dried and treated
with concentrated sulphuric acid. After the first copious evolution of

» Pogg. Ann., 43i xo8.

« Gilb. Ann., 14, 247 ; Ann. chim. phys., 43, 228.

» Ber., 6, 1468.

* Pogg. Ann., 16, 385.

* Dissertation Univ. Rostock, 1899.

* Dissertation Univ. Heidelberg, p. 141, 1902.

YTtRIUM ^ARtHS. 575

hydrofluoric add, the mass was wanned and finally heated until the heavy
fumes of the sulphuric add came off. The pasty mass was allowed to
cool; when it was extracted with water and the insoluble residue, which
consisted of more or less of the oxides of columbium and tantalum, in-
soluble sulphates and a little undecomposed mineral was thoroughly
washed with water by decantation. This sulphate solution of the earths
and uranium was nearly neutralized with soditun hydroxide. Oxalic
add was then added, and the oxalates predpitated. These crude oxa-
lates were ignited to oxides, dissolved in nitric add, and treated with
a hot saturated solution of potassium sulphate, with the addition of
the solid salt. The soluble yttrium double sulphate solution was pre-
dpitated with oxalic add, the oxalates were roasted to oxides, from
the nitric add solution of which it was again treated with potassium
sulphate. Three such treatments with potassium sulphate from the
nitrate solution were found necessary to completely remove the more
insoluble group of earths and after three such treatments the didymiums
could not be detected in a strong solution of the nitrates by means of
thdr characteristic absorption spectra nor could cerium be detected by
means of hydrogen peroxide.

Fractionation of the Succinates.

With R. C. Bbnnbr.

About one htmdred grams of the oxides were dissolved in nitric acid
and the sHght excess of free add neutralized in ammonia. This neutral
nitrate solution was diluted to a liter, brought to boiling and a saturated
solution of neutral sodium succinate added in portions of loo cc. By
this means twelve fractions were obtained. After the addition of each
portion of the sodium succinate, the solution was boiled fifteen minutes,
after which the succinates were filtered and washed with 400 cc. of hot
water. The fractions thus obtained were dried and ignited to oxide,
small portions being taken for the determination of the atomic weight
and the study of the absorption spectra.

SBRI9S I.

Weight of fraction as oxide.

Grams. Atomic weight in RfOs.

1 22.2 114. 2

2 15.8 III. 83

3 lo.o 110.06

4 8.8 1 1 1 . 30

5 90 108.5

6 7.1 107.2

7 7.0 105.6

8 6.1 104.4

9 7.0 lOI.O

10 6.4 97.0

II 4.0 95.8

" 30

....

576 VICTOR LENHER.

The fractions thus obtained were combined and refractionated, se-
lection being made of the portions whose atomic weights were dose to
each other; thus, for the second series, 2, 3 and 4 were combined and
5, 6, 7 and 8 were united while the lightest material, 9, 10 and 11, was
similarly combined.

These oxides were dissolved in nitric acid, the excess of nitric add re-
moved by evaporation and in a dilution similar to that in Series I were
fractionated by addition of sodium succinate in fractions as before. By
again combining fractions of close atomic weight in three such series
of fractionations the most soluble succinate fraction gave a nearly white
oxide, whose nitrate solution showed only very weak absorption bands.
The atomic weight of the element in the most soluble portion was 93,
and the fraction corresponds to yttrium containing small amounts of
samarium, europium and holmium as shown by the absorption spectra.

On the other hand, the third fraction at the other end of the series
or the most insoluble portion gave a yellow earth whose atomic wdght
corresponded to 139. This oxide dissolved in nitric acid to a pink solu-
tion and showed absorption bands and was doubtless a mixture of 3rttrium
with terbium, holmium, europium, samarium and erbium. Further
study of these earths is being continued.

It has been considered well worth the while in making these studies
on the samarskite earths and on the work which is in progress on the
yttrium earths from monazite to determine the atomic weight in each
fraction and also to study the absorption spectra.

The determination of the atomic weight for control purposes can be
carried out sufficiently accurately by the estimation of the oxalic add
and of the earth oxide in the oxalate. While it is true that this method
has some serious defects yet very good results can be obtained if it is
properly handled.

Baxter* has recently shown in splendid detail that when the oxalates
of neodymium, praseodymium, yttrium and certain other of the rare
earths are precipitated in neutral or nearly neutral solution, they ex-
hibit a strong tendency to carry down ammonium oxalate. This earn-
ing down of ammonium oxalate by the insoluble oxalates of the yttrium
earths is very pronounced and is difl&cult to prevent without going to
the opposite extreme and making the solution too add, in which case
we have the factor of the solubility of the oxalate in add appearing and
the obviously incomplete precipitation of the oxalate. In the first case,
the result due to the presence of additional oxalic add in the predpi-
tated oxalate would cause the atomic weight to be too low, while if con-
siderable free nitric acid is present, the more soluble yttrium oxalate
in incompletely precipitated and the result is to obtain a high atomic
^ This Journal, 28, 1684.

MODIPIBD SPBCTROSCOPIC APPARATUS.

577

weight, due to the greater solubility of yttrium oxalate in nitric acid
tban of those earths of higher atomic weight.

The most accurate method for the precipitation of the oxalate of the
yttrium earths which has come to the attention of the author has been
to use a gram or less of the nitrate in very slightly add solution and to
use a dilution of about 500 cc. The oxalate is precipitated from the boil-
ing solution by means of a dilute solution of pure oxahc acid.

In conclusion, the author wishes to acknowledge his appreciation of
the courtesy of Mr, H. S. Miner, of the Welsbach Co., of Gloucester,
N. J., who has placed at our disposal a large quantity of rare earth resi-
dues from Carolinian nionazite and who has been able to secure for us
a quantity of rare minerals for the study of the chemistry of the metals
of the yttrium group.

VBRSITT OF WiSeOt

Madison, wis.

[CoNTRmuTioN PBou TH8 CtiBHiCAL LABOXATORr OP Harvaxd Collsgb.]
MODIFIED SPECTROSCOPIC APPARATUS.

ReedTcd January 31, 190S.

In examination of absorption spectra of dilute solutions in long tubes,
the faintness of the spectra owing to the necessarily great distance of

K^»e I

of light
from the

. slit is fre-

J quently a

disadvan-

"^ tage. A

form of container which partially obviates the diffi-
culty is easily constructed of the shape shown in
Fig. I from a T of glass tubing of suitable diameter.
The light passes through the tube B lengthwise and
is focused upon the slit S by the solution in the
tube A, which acts as a cylindrical lens, thus very
much increasing the light mtensity. If the tube B
is long, the length of path of the outside and
middle rays of the beam within the tube is essen-
tially the same, so that absorption is neariy equal
in all parts of the beam. Hence this form of ap-
paratus does not possess the disadvantage of a simple
cylindrical vessel in which the outside rays pass
through a relatively shorter length of solution.

ESS

Figui

578 WILUAM li. D^HN.

Fig. 2 illustrates a very convenient form of fulgurator for the ex-
amination of the spark spectra of a number of different solutions at one
time. Such a process is frequently much retarded by the inconvenience
in cleaning the ordinary forms of fulgurating apparatus between the
examination of each two solutions. Two glass tubes, AA, into one end
of each of which platinum wires have been sealed, are fused together
in a nearly parallel position by means of a short piece of glass rod, B.
One of the wires is bent in the form of a U so that the end is directly
below and parallel to the wire in the other tube. The end of the lower
wire may be covered with a glass capillary, C, in the usual way. The
apparatus is dipped into the solution to be examined tmtil the capillary
is completely filled with solution. This system can be readily trans-
ferred from one vessel to another and can easily be rinsed into the ves-
sel in which it has been used. If a rod is used in joining the tubes to-
gether the tubes may be brought so near without danger of short-cir-
cuiting that the apparatus is narrow enough to be inserted into a large
sized test-tube. If the tubes are joined through a tube there is some diffi-
culty from this source.

Cambrxbob, Mass.,

January 26, 1908.

[Contribution prom thb Univsrsity op Washington.]
SIMPLE DEMONSTRATIONS OF THE GAS LAWS.

Bt William M. Dbhn.
Received January 31, 1908.

The experiments usually given in textbooks to demonstrate Charles's
law and Boyle's law involve pieces of apparatus so heavy or so com-
plicated that they are imsafe or too time-consuming to be put into the
hands of begiimers in chemistry. That a knowledge of these laws and
of the effect of aqueous vapor on gases should be developed early in
chemical instruction can scarcely be denied, but almost no laboratory
course for beginners gives time or attention to these demonstrations.
If considered at all and apart from the study of physics, their demon-
stration is given in the chemical lecture and inevitably large numbers
of students fail to develop a working knowledge of the individual lavs
or a rational conception of their joint application in the formula:

760(273 + 0'

With the apparatus described below, involving use of the moving

drop of mercury,^ all of these effects of heat, pressure and aqueous

vapor may not only be demonstrated and calculated within one hour

by the stttdent, but the pieces of apparatus represent small initial cost

* This Journal, 29, 1052.

SIMPLE DBHONSTRATION5 OF THS GAS LAWS.

579

and minima of liability of breaking. Furthermore, these forms place
in the hands of students instruments that not only admit of great ac-
curacy but of direct visible demonstration of the laws. Finally these
instruments not only avoid the necessity of making weighings but may
in a simple manner demonstrate the joint effect of heat, pressure and
aqueous vapor, as embodied in the above-written gas formula.

Charles's Law. — Part I consists of a calibrated bulb. A, and a
graduated stem, B, whose internal diameter is less than 3 mm. Having
been cleaned properly and filled with air, which must be dry or low in
aqueous vapor, the mercury drop is adjusted to a position in the bulb
end of the stem.' A rubber tube of convenient length is placed over
the end of the stem and the apparatus, except the end of the rubber
tube, is immersed in a pneumatic trough or a convenient form of water

:.fe

bath whose tempemture is near that of the room.^ After adjusting
the mercury drop, the temperature (() of the bath and the volume of
the air confined in the apparatus (y) are read. Hot water is then poured

' Of coune if changes of volume at lower temperatures are to be studied, the
mereury drop is adjusted near the open end of the stem. However, when working
Bt lowered temperatures, it should be remembered that the air contained in the ap-
paratus mnit be dry.

'Tlie most simple method for the student is to place the instrument and a ther-
ffiometer flat on the bottom of a pneumatic trough and to fill up with the necessary
quantity of tap water. Pot the second readings, a beaker of hot water is prepared
and added to the tap water. Indcasei cd 30-30° are usually sufBctent.

580 WILLIAM M. DEHN.

into the bath and, after complete readjustment of the mercury-piston,
the increased temperature (/') and the increased volume (v') are read
The mathematical relation,

v:v' : : (273 + 0 :(273-^ 0» (H)

is shown by the data to be correct or is approximated very closely.*

Boyle* s Law, — For this experiment Part II (Fig. i) is attached to Part 1
and the whole apparatus is suspended on the ring of an ordinary tripod.
After adjusting the mercury drop near the -rubber cotmection of Parts
I and 11,^ the atmospheric pressure (p) and the volume of the contained
air (v) are read. Air from the lungs is then blown into the instrument
at the point M, the mercury rises in the manometer, as shown in the
figure, while the mercury drop moves along toward the bulb end. The
stopcock is then closed and the volume of the contained air (if) is read
and the height of the mercury column in the manometer is measured
and added to the atmospheric pressure as the increased* pressure {ff).
The mathematical relation,

if :v' :: f/' :p\ (IIIj

is shown to be correct or is approximated very closely.*

Aqueous Vapor. — The instrument (Part III of Fig. 2) and the meth-
ods used to determine aqueous vapor are fully described in the previous
contribution.* Since the volume occupied by the aqueous vapor in air
is actually determined in this experiment, its relation to the atmospheric
pressure as a partial pressure (a) is easily developed and the true pressure
of a gas containing watery vapor is seen to be p — a.

Almost invariably students meet with difficulty both in embodying
in the gas formula the above-derived mathematical statements of the
laws but also in conceiving that they really are embodied in the formula.
These diflficulties may be removed (i) by constant practice in successive

^ If the absolute temperature is to be calculated, x in the derived equation,

X -■

'i/

shows from the experimental data dose approximation to 273. Of couise, the ir-
crement of volume for i ° C, is equal to ^/x and should approximate V^^

' Since it is more convenient to use separate instruments for the two experiments,
no readjustment of the mercury drop need be made by the student. The proper
initial adjustments may be made by the instructor and the same instrument may then
be used by the students for series of duplicate experiments.

' When the mercury drop starts near the bulb end, air may be drawn from tbe
apparatus and thus the effect of reduced pressure may be shown. In either case the
height of the manometer needs to be only a little greater than 100 mm., since this is
about the average pressiu-e of air blown from or drawn into the lungs.

* The average error in a class of twenty freshmen was found to be only o.i per
cent, or an error of one part in a thousand.

» This Journai*, 29, 1052-55.

SIMPlrE DEMONSTRATIONS OP THE GAS LAWS. 58 1

application of the respective laws, (2) by development of the formula*
from the above-derived mathematical statements of the laws, and (3)
by experiments involving the simultaneous effects on gases of changed
temperature, pressure and aqueous vapor pressure.

Joint Effects on Gas. — The apparatus depicted in Fig. 2 is employed
to show the simultaneous effects on gas volumes of changes of tempera-
ture, pressure and aqueous vapor. When the aqueous vapor pressures
(a and a') at different temperatures are taken from tables, the following
method is employed: The instrument is filled with ordinary air and
set up in the waterbath in the manner shown in the figure. A
small measured quantity of water is run in; the volume occupied by
the moist enclosed air (v), the temperature of the bath (/), the baro-
metric pressure (/?) and the aqueous pressure (a) are read. The water
of the bath is gradually replaced by warmer water, until the mercury drop
has moved to the limits of graduation on the manometer end of the scale ;
the mercury in the manometer indicates the resulting increased pres-
sure. Should greater increased pressure be desired, sustained blowing
at m produces another rise of mercury in the manometer and a simul-
taneous repulsion of the mercury drop along its scale. A further increase
of temperature may now be made, provided the manometer can sus-
tain an increase of pressure. The final volume (i;'), temperature of
the bath (/'), internal pressure (/>') and aqueous pressure (a') are read.
The data show that

or that

(p — a)f
from which, if the initial volume had been that of a dry gas at 0° C. and
760 mm., the usual gas formula may be derived.

^ The following is given as an example of development of the gas formula. The
effect on a gas of a change of temperature is shown in equation (II). Let 1/ be the
volume at o® C. (f) and any tmdetermined pressure. Solving, we have

. ^ ^,(273^^^) ^ T>373
" 273 + < "273 + /-

The effect on a gas of a change of pressure is shown in equation (III). Let v^ be the
volume at 760 mm. (^ and at o^ C, studies of changes of volume at different pres-
sures may be made at this or any other temperature. Solving, we have

Bquating the values of if, we have

1/^760 1/273

^/ _ ^^^73 ^y(/>— 0)273
760 (273+0 "■760 (273+0
in which p — a is the true pressure exerted by the gas.

582 C. C. TUTWII^KR.

Should it be desired to determine experimentally all of the factors
(i, e., a and a') of the above equation (IV), the following method may
be employed : The apparatus is first filled with dry air. After adjust-
ing the mercury drop, the volume of the contained dry air, the tempera-
ture of the bath (/) and the barometric pressure (p) are read. A small
measured quantity of water is then run in and the new volume (v) is
read. From the data obtained, the aqueous pressure (a) is calculated.
After v'y />' and V have been determined by the method described above,
the water of the pipette (P) is replaced by concentrated sulphuric add
and a measured quantity of the acid is run in. After the internal pres-
sure (/>') is restored by blowing at M, the volume of contained dry air
is read and the original aqueous vapor pressure (a') at p' and /' is cal-
culated.

Sbattlb. Wasbimoton.

AS IMPROVED HYGROMETER FOR DETERMUnNG THE MnilHDM
TEMPERATURE OF GAS IN DISTRIBUTIOir MAINS.

By C. C. Tutwilbr.
Received January 17, 1908.

In the distribution of illuminating gas, it has been found by repeated
experiments that the gas will leave the works storage holder saturated
with water vapor and also with the vapors of unfixed hydrocarbons,
which latter contribute largely to the photogenic value of the gas. This
condition of saturation is due to the fact that the gas reaches a temper-
ature in the storage holder which is lower than any temperature to whicb
it has been previously subjected up to this point, and it is therefore satu-
rated with water and hydrocarbon vapors at the temperatiue of the holder.
It has also been found that when the gas enters the relatively colder
distribution mains some of these vapors will be dropped, the amount
remaining saturating the gas at the lower temperature.

These \'apors are unavoidably present. They are not vapors of the
oil used in making the gas but are high temperature products of the
closed ring series of hydrocarbons formed by the heat necessary to break
up the oil into permanent gas. The manner in which these hydrocarbons
are dropped out of the gas and again reabsorbed and the effect upon the
candle power of the gas is an interesting study which need not be discussed
at present, as it has only an indirect bearing upon the subject of this paper.
It may not be out of place to say, however, that the aim of the gas en-
gineer is to eliminate from the gas as tnany of the vapors of low tension high
boiling hydrocarbons as possible and to retain those whose tension will
permit of their being carried to the burner under all conditions of tem-
perature and pressure met with during distribution.

AN IMPROVED HYGROMETER. 583

It is evident that if the gas could be delivered to the consumer at a
temperature equal to or greater than that obtaining in the storage holder,
there would be no loss of light-producing hydrocarbons in the mains,
and the candle power of the gas delivered at the burner, would be the
same as that shown at the outlet of the storage holder. While this
might seem to represent an ideal method of distribution, it is not prac-
ticable from an economic standpoint, as the temperature to which the
gas is subsequently cooled in the mains during some seasons of the year
is far below the outlet temperature of the storage holder and the cost
of reducing the gas to the temperature of the mains at these times would
more than offset any advantage gained. It has been found that the
gas in the distribution system quickly reaches the temperature of the
earth surrotmding the mains, and where the mains are exposed, it may
even be cooled to the surrounding atmospheric temperature. The at-
tendant loss of hydrocarbon vapors and therefore loss in candle power
is considerable, depending upon the so-called "permanency" of the gas
and the degree of cold to which it has been subjected. Gases which owe
a large percentage of their photogenic value to so-called unfixed hydro-
carbon vapors will suffer a greater loss in this respect than those that
contain a relatively greater amount of fixed hydrocarbons.

When, therefore, it is required to deliver to the consumer a gas of a
uniform candle power throughout all seasons of the year, it is necessary
to turn into the distribution mains a gas sufficiently high in candle power
to take care of subsequent losses in the mains, which losses will vary
from day to day.

The problem at once presents itself as how best to determine what
this initial candle power should be, since any excess over and above
that actually necessary, means financial loss to the gas company on
account of extra enrichment or on the other hand may subject it to
penalization if it should fall below that required by contract.

Since the loss in candle power is due to reduction in temperature of
the gas in the mains, it will appear at once that the first thing to be de-
termined in the above problem is the minimum temperature to which
the gas will be cooled after it leaves the holders, and after having found
this temperature to raise or lower the candle power of the holder gas
to such an extent that when it is cooled down to the minimum temper-
ature it will still have a candle power equal to that which must be fur-
nished to the consumer.

It was found impossible to determine the minimum temperature by
means of thermometers placed in the mains, as the gas continually changed
in temperature, owing to its passage through more or less exposed por-
tions of the mains, or on account of its being subjected to other condi-
tions tending to change its temperature, such as rate of flow or nature

584 C. C. TUTWILER.

of the ground through which the main passed. For these reasons the
location of the point of minimum temperature by means of thermome-
ters was found to be exceedingly difficult and unreliable and the attempt
was abandoned. A method was finally evolved which depended, for
its successful operation, upon the following data:

In the course of a series of experiments made in the Philadelphia Gas
Works, it was found that if a gas was cooled down in contact with its
condensate to a lower temperature than any to which it had been pre-
viously subjected and then allowed to warm up in contact with its con-
densate to within a few degrees of the original temperature, or, to its
original temperature out of contact with its condensate, the water which
it would then contain as determined by means of calcium chloride, would
just saturate it at the minimum temperature to which the gas had been
cooled. An explanation of this is found in the fact that if oil and water
are simultaneously deposited the oil will form a film upon the surface
of the water, preventing its being again picked up when the temperature
is raised so long as any oil remains. It appeared that advantage might
be taken of this action and the minimum temperature which the gas
had reached in the mains up to any given point be ascertained by de-
termining its water dew point. For this purpose, an ordinary wet and
dry bulb psychrometer was employed, the instrument being hung in a
bell jar sealed in mercury, through which a current of the gas to be tested
was continually passing. The dry bulb thermometer indicates the tnie
temperature of the gas, while the wet bulb thermometer registers the
temperature of evaporation, which is usually several degrees below the
temperature of the gas. In saturated gaseous atmospheres, the ther-
mometers will read alike and diflference will be recorded in proportion
to the dryness of the gas. The dew point may be obtained by means
of these observations from Glaisher's table by multiplying the diflference
between the reading of the two thermometers by the factor opposite
the dry-bulb reading and subtracting the product from the dry-bulb
reading.

This instrument was given a thorough trial and though it furnished
some very valuable data, its lack of portability and the knowledge that
under the best conditions its indications were known to be only approxi-
mately near the truth, caused us to continue our eflForts to devise a more
satisfactory apparatus.

It was suggested at this time by Mr. Chas. O. Bond, chief photometri-
cian of The United Gas Improvement Company, that the instrument
devised by Regnault* known as the condenser hygrometer might be
adapted to our needs.

The essential parts of this instrument comprise a thin-walled glass

* Ann. chim. phys. [3], 15, 129.

AN IMPROVED HYGROHBTBR. 585

vessel quite similar in size and shape to an ordinary 15 cc. test tube,
provided with a delicate thermometer and having means for passing
air through a small amount of ether contained in the tube. Upon ex-
posing the tube to an atmosphere containing water vapor and reducing
the temperature of the tube by volatilizing the ether, moisture will finally
be deposited on its outer wall. When this occurs the temperature is
read on the thermometer, which reading is the temperature at which
the atmosphere would be just saturated with the moisture contained
therein, or in other words, its "dew. point." The apparatus as designed
by Regnault was intended to be used for the determination of the dew
point of the atmosphere only, but its principles were successfully utilized
in an apparatus which was found to be ap-
plicable to the determination of the dew
point of any gas. This was accomplished
by providing means for surrounding the
vaporizing tube with the gas to be tested by
the use of an outside jacket through which
the gas was made to Sow and providing
scrubbers for the removal of hydrocarbon
and water vapors, the use of which will be
explained later on. In the first apparattis
designed, air was forced through the ether
by means of a rubber hand-pump, and it
was found that at times when the dew point
was very low, moisture would be deposited
in the tube from the air and so cloud the
ether that the deposit of dew on the outside
of the tube could be seen only with great
difBculty. This trouble was finally overcome
by using a current of the gas being tested
to volatilize the ether, and besides correcting
the trouble, the cumbersome hand-pump was
thus done away with. This and other im-
provements tending to compactness, resulted
in the improved ap[>aratus about to be de-
scribed.

The construction and operation of the
apparatus is as follows: The interior glass
vessel (A) known as the condensing tube,
the thermometer (T) and the small tube
(B) reaching to the bottom of the con-
densbg tube by means of which the ether
is volatilized are quite similar to the essen-

586 C. C. TUTWILER.

tial parts of Regnault's apparatus, and if this portion of the apparatus
is detached from the jacket by unscrewing the collar (C), it may be used
in the same way as the Regnault apparatus for obtaining the dew point
of the atmosphere. In order, however, to adapt it to our piurpose, the
glass jacket (D) was provided, through which a stream of the gas to be
tested could be made to flow, surrounding the condensing tube in its
passage to the burner. The course of the gas from the time it enters
the inlet (I) imtil it issues at the burner (E) on top of the instrument,
is as follows:

The cock (F) at the base of the instrument is so constructed that the
gas upon reaching the same can be made to flow straightway into the
jacket (D) or made to first pass through one or other of the two scnib-
bing vessels (G and H). In either course it finally passes into the jacket
(D) and surrounds the condensing tube (A). It then passes through
the hole (J) in the screw cap covering the jacket, and into the tube (B)
which runs to the bottom of the condensing tube. The gas after leav-
ing the condenser tube passes into the cap (K) and is finally burned
at the burner (E). If a few cubic centimeters of ether or other volatifc
liquids, such as pentane, are placed in the condensing tube the gas bubbling
through will rapidly volatilize it, thereby reducing the temperature of
the tube and causing a deposition of dew upon its outer wall as soon as
the point of saturation is reached. As soon as this occurs, the thermom-
eter is read and the dew point thus ascertained. It is obvious that the
instrument may be placed on any convenient gas bracket and that a
few moments will suffice for determining the minimum temperature to
which the gas has been cooled, up to that point.

It has been found that the determinations made with the instrument
under all ordinary conditions are very reliable. Cases arise, however,
when in order to get a correct indication of the minimum temperature,
it is necessary to make use of a rubber scrubber to remove some of the
hydrocarbon vapors prior to testing the gas and more rarely to use a
calcium chloride drying tube to remove a portion of the water vapor.
For example, if the gas after having been cooled .to its minimum tem-
perature in the mains should pass through a section of main which for
some reason wsls vrarmer and any Hquid hydrocarbon was present in the
main at that point, the gas would saturate itself at the higher temper-
ature with hydrocarbon vapor and therefore the instrument would show
the dew point of the vrater vapor which would correctly indicate the
minimum temperature. If, however, the gas is first passed through
the scrubber (G) which contains finely divided rubber, the hydrocar-
bon vapors vsrill be removed to such an extent that the amount remain-
ing vnll not saturate the gas before the dew point corresponding to the
vrater vapor is reached. Such conditions are rarely met with in the dis-

DBTBRMINATION OF LEAD IN ORES, ETC. 587

tribution mains. It is well, however, as a precautionary measure to
take both hydrocarbon and water vapor dew points and if the hydro-
carbon vapor dew point is found to be higher than the water vapor dew
point, the latter should be taken as representing the minimum tempera-
ture.

Again, it is possible that the gas after having reached its minimum
temperature may warm up in contact with water vapor which has been
introduced into the gas, as for example in purifiers where steam is ad-
mitted for manufacturing reasons. Under such conditions the calcium
chloride scrubber (H) must be used and the dew point of the hydrocar-
bon vapors taken as the minimum temperature.

It is a very simple matter to test the gas in all three ways, t. e., direct,
through the rubber scrubber, and through the calcium chloride scrub-
ber, and it has been found advisable in order to get a correct idea of
what is taking place or what has taken place in the mains to frequently
check the direct readings with readings made after scrubbing the gas.

Such observations also enable the gas engineer to judge whether
the gas is being scrubbed by tar or heavy drips deposited in the mains
or whether it is picking up hydrocarbons from the mains. If the gas
has been cooled to a low temperature as it might be in passing through
an exposed main as under a bridge and afterwards warmed up in the
ground, as previously stated, we would expect the hydrocarbon vapor
dew point to be higher than the water vapor dew point. If the hydro-
carbon vapor dew point is lower than the water vapor dew point, it would
indicate contact of the gas with a deposit of tar or hesivy drip oil.

The practical application of this instrument in gas distribution prac-
tice is well defined. Owing to its low specific heat, the gas flowing from
the works quickly reaches the temperature of the surrounding earth.
The gas engineer knowing by this instrument to what extent the hydro-
carbon vapors have been dropped and what the consequent fall in candle
power will be, is capable of anticipating the reduction in candle power
by raising the candle power of the gas going into the holder accordingly.

Laboratory op tbb ^

UifrrsD Gab Improvkmbnt Company, Philadblpbia,

December 18, 1907.

TECHNICAL METHOD FOR THE DETERHmATION OF LEAD m

ORES, ETC.

By A. H. IrOW.
Received January 28, 1908.

The following scheme is the result of many attempts on my part to
improve the methods wherein lead is separated as oxalate and subse-
quently titrated with permanganate. I have used the method as de-
scribed below for several months and have fotmd it more satisfactory

588 A. H. lyOW.

for technical work than any other with which I am acquainted. Dupli-
cate assays usually check within one-tenth of one per cent. A test of
five portions of pure lead, weighed so as to approximately represent
3, 12, 2o, 45 and 6o per cent, of lead in an ore, and put through the en-
tire process, showed errors of — o.oi, — o.o6, -f o.oi, +0.05 and +0.05
per cent., respectively, the average error being 0.036 per cent. The
permanganate was standardized on about 0.200 gram of lead.

Chromate-Oxalate Method. — ^Take 0.5 gram of ore and treat in a 6-oz.
flask by the usual methods to obtain the washed lead sulphate, etc,
on a 9 cm. filter. Dissolve the lead sulphate on the filter by stirring
it up repeatedly with a jet of hot sodium acetate solution contained
in a wash-bottle, receiving the filtrate in the original flask. Prepare
the solvent by diluting a cold saturated solution of commercial sodium
acetate with an equal bulk of water and adding 40 cc. of 80 per cent
glacial acetic acid per liter. To test if the extraction of the lead sulphate
is complete, the flask may be replaced by a small beaker, the washing
continued and a little potassium dichromate added to the filtrate. If
any lead chromate is produced, the mixture may be added to the dear
solution in the flask. Agitate the extmct in the flask and heat it if
necessary to redissolve any separated precipitate. Add 10 cc. of a 5
per cent, solution of commercial potassium dichromate, heat to boil-
ing and boil gentl}'^ for a few minutes to render the precipitate basic
and easily filtered. The change is shown by its becoming reddish yellow
in color. Filter hot, wash out the flask with hot water and then wadi
the precipitate only once, simply to clean the upper edge of the filter.
Place a wide-necked fimnel in the flask, open the filter and spread it against
the wall of the funnel and wash off the lead chromate with a jet of hot
oxalic acid solution, using from 25 to 40 cc. Rinse down any adhering
chromate in the flask with hot water. The oxalic acid solution consists
of a cold saturated solution of commercial acid i part, water 3 parts.
Heat nearly to boiling in a wash-bottle. To the mixture in the flask
add grain alcohol and then boil until the chromic add is all reduced and
the lead converted to oxalate. Remove from the heat, add 30 cc. of
cold water and cool thoroughly, best under the tap or in cold water. When
cold, filter through an 11 cm. filter, wash out the flask thoroughly with
cold water and then wash filter and precipitate 10 times with cold water.
Place 5 cc. of strong sulphuric acid in the flask, dilute first with a littk
cold water and then with hot water to about 125 cc. Add the filter
and precipitate and titrate to the usual pink tinge with standard potas-
sium permanganate solution. . The solution used for iron will serve,
although too strong for the best work. Theoretically, the oxalic add
(C204H,.2HjO) value of the permanganate multiplied by 1.642 will give
the lead value, but owing to slight losses as sulphate, oxalate, etc, and

BLBCtROLYTlC DETERMINATION OP BISMUTH. 589

the fact that the lead oxalate is not perfectly pure, the factor 1.669 ^U
give a closer approximation. On this basis, for 0.5 gram of ore taken
for assay, the solution should contain 1.5 185 grams of potassium per-
manganate per liter, in order that i cc. may equal i per cent. lead. It
is best to standardize on about 0.200 gram of pure lead dissolved in a
little 1:2 nitric add and put through the entire process.

Notes, — ^The lead oxalate formed in the above process is not pure
white, but yellowish, and still contains about i per cent, of lead chro-
mate. It appears to be sufficiently uniform in its nature to give accurate
results.

Instead of dropping filter and precipitate into the flask for titration,
as descpbed above, a neater method of procedure is as follows: Place
the flask tmder the funnel and pour through and over the filter about
75 cc. of hot dilute add, containing 5 cc. of strong sulphuric add, and
then wash the filter and residue well with hot water, so that the final
bulk of the filtrate will be about 125 cc. .Titrate the hot liquid as before.
This method takes a little longer and gives practically the same results
as the simpler way, but it has the advantage of a permanent end-point,
there being no organic matter present to slowly decolorize the pink tinge.

Caldum does not interfere with the method, nor does antimony. Bis-
muth in small amounts is without material influence. Ten per cent,
of bismuth added to a mixture containing about 23 per cent, of lead
raised the result 0.36 per cent., most of the bismuth being removed as
sulphate and chromate.

Dbhvbr, Colo.

[CONTRIBirnONS FROM THB HaVEBIBYSR LABORATORIES OF COLUMBIA UNIVBRSITy,

No. 150.]
THE ELECTROLYTIC DETERMINATION OF BISMUTH.

Bt p. J. Mbtzobk and H. T. Bbans.
Received January 39, 1908.

Much has been written about the determination of bismuth in the
electrolytic way and many electrolytes have been proposed. Without
attempting any detailed discussion of the numerous publications which
have appeared on the subject, the difficulties with which one almost
invariably has to contend may be briefly summed up as follows :

When a bismuth solution is electrolyzed there is deposited not only
metallic bismuth on the cathode, but frequently there is a simultaneous
deposition of peroxide on the anode and it has been suggested that both
anode and cathode be weighed for each determination. Again, the de-
posited metal is nearly always black and spongy, and in cases where ac-
curate results have been obtained it has been necessary to exercise the
greatest care in washing and dr3dng in order to prevent loss mechanic-

590 F. J. BIBTZGER AND H. T. BEANS.

ally. A method which has been recommended especially for compara-
tively large quantities of bismuth is the amalgam method originally
proposed by Vortman.^ In this method a known quantity of mercury
salt is added to the bath and the deposit of bismuth amalgam weighed.
From this weight the bismuth is obtained by difference. The quan-
tity of mercury recommended is four times that of the bismuth present

The method described in this paper presents none of the difficulties
encountered in previous methods; the metal is deposited in compact,
adherent form and can be washed and dried without any possible chance
of loss, nor is there ever any deposition of peroxide on the anode when
the electrolysis is completed.

It has been found that by the addition of acetic acid to a bismuth
nitmte solution precipitation by hydrolysis may be completely prevented,
even though the solution be subsequently largely diluted. It is also
possible to obtain this result by first eliminating all free nitric add by
the addition of sodium hydroxide to alkaline reaction and then redis-
solving the precipitated bismuth hydroxide by means of acetic add,
the whole opemtion being carried out in the cold. It has also been
observed that the addition of boric acid to the bath has a decided effect
on the character of the deposit and in addition it serves to bring about,
readily, complete solution of bismuth hydroxide without the addition
of very large quantities of acetic add.

The apparatus employed is a slightly modified form of that originally
proposed by Gooch and Medway,^ using a rotating cathode. The plati-
num thimble employed as cathode was connected to the shaft by means
of a rubber stopper wotmd with fine platinum wire for contact and had
an available surface of forty square centimeters. In all cases it was
driven at the mte of about seven hundred revolutions per minute.

As a basis for the experiments a bismuth solution was prepared by
dissolving chemically pure bismuth nitrate in water containing 25 cc.
cone, nitric add per liter. The solution was then carefully standard-
ized by gravimetric methods.

The method of procedure was in all cases as follows: To a known
quantity of the standard solution phenolphthalein was added, then
sodium hydroxide solution, drop by drop, to alkaline reaction. The
predpitate formed was redissolved in acetic acid and then two grams of
boric add were introduced. The solution was diluted, heated to 70-80"
and electrolyzed. The working conditions are shown in the following
table:

* Ber., 24, 2749.

" Am. J. Sci. [4], 15, 320.

KLBCTROLYTIC DOTERMINATlON OF BISMUTH.

591

Tabids I.

5
6

7
8

9
10

13
14
15
16

17
18

19
20

21
22

23
24
25

s»*

20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20

30
30
30

30
30
30
30

s§

2
2
2
2

none

2
none

2
nonp

2
none
none

2
none

250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o
250 o

2-0.15

2-0. 175

2-0.15

2-0.17

2-0.15

2-0. 125

2-0.2'

2-0.17

2-0.175

2-0.16

2-0.17

2-0. 175

2-0.16

2-0.175

2-0.15

2-0.16

2-0.175

I -2.8

9 -2.75
9 -2.8

9-2.75 63**-77°
°-78°

9 -2.7
9 -2,65

85-2 . 74

8 -2.6

9 -2.8

85-2.75
8-2.72

8 -2.64

8 -2.4

8 -2.7

8 -2.6

8-2.55
77-2.55

75-2 . 37

8 -2.45
75-2 . 5

7 -2.42
75-2 . 55
72-2.45

75-2.50
76-2.55

a
S

4**-79''
5^-79"

2^5^

4"-83^

2^-77^

7^9^

5^-78**

6*»-78*»

6^-79^

7*^-88**

5M8**
3^-77"
6^7**
3M8''
o**-8o**
5**-8i**
o*»-78**
o**-78*»

3*^-79^
4^-77"
4°-79^
8*'-79**

^-80*^

• — ■
«

Bi found
BiTor.

•A

o- 03956

0.0391 — 0.00046

0.0989

0.0990 +0.0001

»V.

0.0989

0.0990 -fo.oooi

I'A

0.0989

0.0989 ±0.0000

0.0989

0.0989 ±0.0000

0.0989

0.0987 — 0.0002

0.0989

0.0989 ±0.0000

ly*

0.0989

0.0986 — 0.0003

ij<

0.0989

0.0984 — 0.0005

0.0989

0.0992 +0.0003

0.0989

0.0992 +0.0003

0.0989

0.0991 +0.0002

0.0989

0.0990 +0.0001

0.1978

0.1981 +0.0003

0. 1978

0.1979 +0.0001

0.1978

0.1978 ±0.0000

0.1978

0.1980 +0.0002

iV.

0.1978

0.1984 +0.0006

0.3956

0.3962 +0.0006

0.3956

0.3953 — 0.0003

0.3956

0.3955 — 0.0001

0.3956

0.3956 ±0.0000

0.3956

0.3957 +0.0001

aV.

0.3969

0.3968 — O.OOOI

0 3969

0.3969 ±0.0000

In Experiments 19-25 inclusive, the amount of acetic acid given does
not completely redissolve the precipitate, but after the addition of the
boric acid the solution becomes perfectly clear. In Experiments 19,
21 and 25, in which no boric acid was used, the slightly opalescent solu-
tion was electrolyzed, the solution, however, becoming perfectly clear
during the period of the electrolysis.

The change in voltage during the electrolysis serves as an excellent
indicator in the determination. The voltage remains constant until
all but the last trace of bismuth has been deposited, when it begins to
rise rapidly to a maximum at which it again remains constant. The
electrolysis should be continued five or ten minutes beyond this max-
imum point. In the table given above the voltages recorded show the
initial and final values. Table II being a record of Experiment 16 above,
makes this a little more clear.

' In this experiment the current was maintained constant.
' The figures in the second column represent the point to which the current drops
at the end of the experiment.

592 F. J. HBTZGER AND H. T. BEANS.

Table II.

Time.

p. M. NDio* Voltage. TemperttoR.

2:50 0.2 1.8 73®

3:05 0.2 1.8 75®

3:20 0.2 1.8 76®

3:35 0.2 1.8 78*

3:43 0.19 2.0 78®

3:45 o.iS 2.x 78**

3:50 0.175 2.4 78*

3:55 0.175 2.5 78*»

4:00 0.160 2.55 78®

4:05 0.160 2.55 78®

The action of acetic acid on the deposited metal was found to be so
slow that it is not necessary to siphon ofiF the liquid at the end of the
experiment. All that is required is to quickly substitute a beaker of
water without interrupting either the current or rotator. The thimble
is then detached, washed with alcohol and ether, allowed to dry and
weighed after standing on the balance for ten minutes. To avoid any
errors which might arise from changes in atmospheric conditions, the
weight of the thimble was taken after dissolving oflF the deposited metal
by immersing in nitric acid, washing and drying, etc., the same as above.

In all of the experiments given in Table I, the available cathode sur-
face was forty square centimeters, and it was naturally supposed that
a stronger current could be employed with a larger electrode, and in this
way the time factor might be reduced. A large electrode having an
available surface of one hundred square centimeters of the same form
as the one previously used was procured with which we obtained the
following result :

I 8; "Si s & S * s -i

« $i* » > Z > HHpq a n

26 30 2 3500.5-0.17 1.9-2.5 78°-8o** iVa 0.3969 0.3972 +0.0003

In working with this larger electrode the current should be reduced
to 0.2 ampere at the time when the voltage begins to rise.

As seen in the table the results are accurate without the use of boric
acid. The metal is deposited in bright, compact form up to the time
when the voltage begins to rise, t. e,, when only very small quantities
remain in solution. From this point, if there be no boric acid present,
the last traces of metal deposit over the surface of the bright coating
in the form of a granular layer less firmly adherent and somewhat darker.
With boric acid present the deposition is uniform to the very end of the
experiment, giving a smooth, compact, adherent deposit having the char-
acteristic color of the metal.

SBPARAtlON Oi^ IRON I^ROM MANGAN^^. 593

In not a single experiment was there any evidence of peroxide forma-
tion on the anode at the close of the determination. Occasionally there
was a slight tinge of jrellow on the anode at the very beginning of the
electrolysis, but this, however, soon disappeared. The solution after
each determination was examined for bismuth and none could be found.

The effect of boric add on the deposition of other metals is being
studied.

AKALYTICAL I«ABOKATOKIBS, ,

January, 1908.

A separahon of iron from mangaitese.

By Richard B. Moorb and Ivy Miller.

Received December 30, 1907. *•

If to a solution of ferric chloride containing free hydrochloric add,
pyridine, in slight excess, be added, the iron is completely pi^'dpitated
as hydroxide. Aluminium, chromium, and zinc, under such conditions,
are not completely predpitated, while manganese, nickel, and"6obalt
remain in solution. This method can therefore be used to separate iron
from the last three metals. The present paper deals only with the sepa-
ration of manganese from iron by means of pyridine. /

When pyridine is added to a neutral solution of a manganous salt no
predpitate is obtained. On warming, the solution slowly oxidizes and
the manganese begins to come down. The rate at which oxidation takes
place, however, is about one-third as fast as the rate at which a man-
ganous solution oxidizes when treated with ammonium chloride and
ammonium hydroxide under similar conditions. If the manganous
solution is made add with a little hydrochloric add before addition of the
pyridine to slight excess, the solution may then be heated for ten minutes
irithout oxidation. Consequently, pyridine possesses considerable ad-
vantages over ammonium hydroxide in the separation under question,
especially when a large amount of manganese is present. Practically,
its only disadvantage is its cost. A very fair separation can be made
with one predpitation.

On washing ferric hydroxide predpitated by pyridine, no iron at first
appears in the filtrate. As the excess of pyridine is washed out of the
predpitate, however, colloidal iron passes through the filter. On wash-
ing the predpitate with pyridine water (1:500) the iron is completely
retained by the filter.

In the following experiments Merck's pyridine, "Medicinal grade,*'
was used. It was redistilled before using.

Pure iron wire was dissolved in lo cc. of 4 iV hydrochloric acid. A
few drops of concentrated nitric add were added and the solution

594 RICHARD B. MOOR^ AKD IVY MILI<^R.

warmed to oxidize the iron. It was then diluted to loo cc., wanned,
and pyridine added from a burette. The precipitate was washed with
pyridine water (i : 500) :

Iron taken. Iron found.

Gram. Gram.

No. I 0.0806 0.0807

2 0.0823 0.0824

3 0.0798 0.0810

In No. I, the pyridine was added until the iron was just precipitated.
In No. 2 there was an excess of 0.5 cc. of pyridine, and in No. 3 an excess
of I cc.

Solutions of manganous chloride were then mixed in different propor-
tions with ferric chloride. The amoimt of free acid was the same as was
taken above. The precipitation takes place equally well in the cold or
after gently warming. In the following series the manganese in the fil-
trate was not estimated, the object being to see what kind of a separa-
tion was made by a single precipitation. One-half a cubic centimeter
in excess of pyridine was used in each case :

Iron taken.

Iron found.

Manganese present.

Gram.

No. 1

0.0797

0.0798

O.X405

" a

0.0790

0.0792

0.0857

•' 3

0.0791

0.0792

0.0719

" 4

0.0798

0.0802

0.0610

In the next series both the iron and manganese were estimated in order
to see what effect the presence of the pyridine has on the precipitation
of the manganese. The manganese cannot be completely precipitated
by addition of bromine water to the hot solution. . In addition, the forma-
tion of pyridine bromide interferes. In No. i the manganese was esti-
mated in the usual manner by potassium carbonate. In numbers 2, 3
and 4, c. p. sodium hydroxide was used instead of the carbonate, the pre-
cipitation being made in porcelain basins:

No. I
" 2

" 3
" 4

Iron.

BCanganeae.

Taken.
Gram.

0.0858
O.II36
O.I162
0.1268

Pound.
Gram.

0.0862

O.II33

O.I155

0.1264

Taken.
Gram.

0.0698

0.0667

0.2289

O.II45

Pound.
Gram.

0.0691
0.0666
0. 2285
O.II40

The conditions imder which aluminum can be separated from man-
ganese and aluminum and iron from beryllium by means of pyridine
are under investigation.

CHBBCICAL IfABGKATORT,
BUTLBR COLLBOB, INDIANAPOLIS.

SODIUM BBNZYL CYANIDE WITH CINNAMIC EStBR. 595

THE ACnOR OF SODIUM BENZYL CYANIDE WITH CINNAMIC ESTER.

Bt S. Avbrt and G. R. McDolb.
Received January ^9, 1908.

Introduction.

Victor Meyer and his pupils* have studied the character of the methylene
hydrogen atoms in benzyl cyanide. They found, in brief, that these
atoms resembled in some respects the corresponding atoms in aceto-
acetic ester and in malonic ester. The reactions studied related in gen-
eral to the condensation with aldehydes, and to the substitution of alkyl
radicles. Later Zelinsky and Feldmann* condensed benzyl cyanide
and methylene iodide, forming the nitrile of symmetrical diphenylglutaric
acid. Michael* has shown that sodium acetoacetic ester, as also sodium
malonic ester, forms addition products with esters of the tmsaturated
acids. This reaction has been used by Auwers,* Perkins,* Avery,* and
others in preparing alkyl glutaric acids. Having had occasion to try
to effect the synthesis of certain alkyl glutaric acids the writers decided
to study the action of benzyl cyanide on cinnamic ester. It was assumed
that an addition product would be formed, which on saponification and
the splitting off of carbon dioxide would give a diphenylglutaric acid.
The results obtained have shown that the reactions which took place
were not strictly analogous to Michael's as the following comparison
shows.

Following the analogy of Michael's reaction, we should expect:
C,H5CHNaCN+C,H,CH=CH— COOR = NaCH— COOR

C^Hj— C— H

C,H,-i— CN

In reality, however, the reaction was accompanied by the saponifi-
cation of the ester and the two principal products formed were

CH,— COONa

C,H,-C-H

CeHj— CH— CN
and a product to which we will provisionally ascribe the formula

> Ber., 20, 534; Ihid., 21, 1291 ; Ann., 250, 118.

• Bcr., 22, 3290. ^
» J. pr. Chem. [2], 35, 352.

• Bcr., 24, 1936.

• J. Chem. Soc. (London), 69, 1472.

• Am. Chem. J., 20, 509; Ibid., 28, 48.

596 ^ S. AVQS.Y AKt> G. It. MCDOUS.

CH,CO

I \

CH,-C-H \
C,H,— C— CN O.

A '

CeH,— C— H

CjHaCO

This formula is in harmony with all of the facts observed except the
following: It appears to be michanged by boiling with sodium car-
bonate or ammonia, either in aqueous or alcoholic solution. Although
it is converted into a salt by boiling with alcoholic potassium or sodium
hydroxide, the acid liberated from the salt has resisted all attempts to
reconvert it into an anhydride.

Of special interest is the formation, easily and in good quantity, of
the compound : CHjCOOH

C^Hj— C— CH

CeHj— CHCN

This on reduction may be expected to yield /?,^-diphenyW-amiiK>-
valeric acid, which on heating should form the corresponding piperidone
derivative. This reduction will occupy our attention in the near fu-
ture.

Experimental.

P^Y'Diphenyl-y-Cyanbutyric Acid. — ^Twenty grams of benzyl cyanide
were mixed with 30 grams of cinnamic ester, and solid sodium methyhte
equal to 4 grams of sodium was added with stirring. The mixture be-
came hot and the action was apparently completed in a few minutes.
When the action had ceased the product was allowed to stand on the
waterbath for two hours. It now appeared as a viscous semi-crystal-
line yellow mass. On adding hydrochloric acid and stirring, it changed
to a very viscous oily mass which on thoroughly acidifying and washing
out the sodium chloride, gradually assumed a semi-crystalline conditkm.
This was dissolved in hot benzene and on cooling hair-like needles sepa-
rated out. On filtering oflF the mother liquor and condensing, a second
crop of crystals was obtained. With careful manipulation the yield
amounted to nearly 80 per cent. Almost equally good results were ob-
tained when solid sodium hydroxide was used instead of sodium methykte.

During the reaction the escape of alcohol vapors was noticed.

The acid obtained was recrystallized from benzene. It melts at 161.5*,
is easily soluble in alcohol, ether and hot benzene, difl&cultiy sohibk
in hot and almost insoluble in cold water. Glistening white needles
from benzene or dilute alcohol.

SODIUM BENZYL CYANIDE WITH CINNAMIC ESTER. 597

The analysis gave: €,76.67 11,5.84 N, 5.19

Calculated for C„Hi»0,N: 0,76.98 H, 5.66 N, 5.28

Titration, 0.2 loi gram required 7.55 cc, N/io sodium hydroxide.

Theory for one acid H in CiyH^sOaN requires 7.52 cc.

The silver salt is a white insoluble powder, which gave 29.01 per cent.
Ag. Calculated;. 29.03 per cent.

a,fi-Dtphenylgluiaric Acid. — This acid was first prepared by us by adding
sodium (i atom) dissolved in ethyl alcohol to benzyl cyanide (i mole-
cule). After the action had ceased a molecular quantity of cinnamic
ester was added. The action proceeded with the liberation of consid-
erable heat. The residting mass was viscous and light brown in color.
This was acidified with hydrochloric acid and the sodium chloride washed
out. The mass resisted all attempts to recrystallize. Attempts were
made to saponify with potassium hydroxide, in one instance the com-
pound above mentioned, the half nitrile acid was obtained in a very
impure state. As a whole the saponification with potassium hydrox-
ide was unsatisfactory and a product free from nitrogen could not be
obtained by this method. The viscous mass was sealed in a Carius tube
with concentrated hydrochloric acid and heated for five hours to 150**.
The mass was extracted with boiling water several times and the remain-
ing mass reheated and the process repeated. The aqueous extractions
were condensed and a crystalline product was obtained which showed,
under the microscope, crystals having both cubical and needle-like shapes.
Varying products were obtained which melted from 195° to 215*^. By
careful recrystallization out of 50 per cent, alcohol the pure diphenyl-
glutaric add was obtained which melted at 223-4^. The crystals are
needles. By working over the mother liquors from these recr)rstalliza-
tions, crystals of a cubical appearance were isolated. It was impossible
to get very satisfactory melting points of any of the lower melting prod-
ucts but by repeated recrystallizations a substance was obtained melting al-
most constantly at 205°. This compound resembled the diphenylglutaric
so cbsely that it was for a time thought to be a stereoisomer. Analysis
for carbon and hydrogen agreed very closely with that of the glutaric
acid, melting at 223-4*^, and it formed an anilic acid which, though it
melted about 30 ^^ lower, corresponded very closely in other respects
with that obtained from the glutaric. However, very careful work
showed the presence of a small amount of nitrogen in the supposed
isomeric glutaric acid (m. 205 *') and when this compound was placed in
the sealed tube with hydrochloric acid and heated, the compound obtained
was in every case the diphenylglutaric acid, melting at 233-4® and free
from nitrogen. The cjranide group was either saponified or converted
into insoluble combinations.

The diphenylglutaric acid can be obtained in quantity and without

59^ S. AVERY AND G. R. MCDOLE.

the formation of objectionable by-products by heating the jJ,^-diphenyl-
y-cyanbutyric acid with concentrated hydrochloric add in a sealed
tube to 150° for three hours.

The analysis of a,,9-diphenylglutaric acid obtained as first mentioned
gave: C, 71.86; H, 5.33. Calculated for CijHyfi^: C, 71.79; H, 5.67.

Titration, 0.1142 gram of substance required 8.01 cc. N/io NaOE
Theory for Ci7Hie04, 8.03 cc. N/io NaOH.

The silver salt gave 43.34 per cent Ag. Theory for Cj^H^O^Ag, 43.35
per cent.

Analysis* of the diphenylglutaric acid obtained by the saponificatioo
of the ^,pdiphenyl-7-cyanbutyric add gave: C, 71.59; H, 5.69. Cal-
culated for C^HjeO^: C, 71.69; H, 5.67.

Titration, 0.1084 gram of substance required 4.05 cc. N/ioNaOH.
Calculated for two acid hydrogens in C17H15O4, 3.99 cc. N/io NaOH.

The Anhydride, — Attempts were made to prepare the anhydride both
by heating the acid above its melting point and by treatment with acetyl
chloride and acetic anhydride. In all cases a brown product resulted
which resisted our attempts to obtain it in a crystalline form.

Aniltc Acid. — ^The anilic add was prepared by heating some of the
glutaric acid with acetyl chloride, and boiling off the excess of acetyl
chloride. It was then treated with the calculated amount of aniline
dissolved in alcohol. The anilic add separates out of dilute alcohol
in flat needle-like crystals having a beautiful mother of pearl luster when
dry. Soluble in ether and alcohol. Purified by recrj^tallizing out of
alcohol. Melting point, 230-2®.

Titration, 0.1537 gram of substance required 4.25 cc. N/ioNaOH.
Theory for CajHaiONj, 4.28 cc. N; 10 NaOH.

The nitrogen determination gave 4.16 per cent. N; calculated, 3.93
per cent.

Other Prodticts formed by ike Action of Sodium Benzyl Cyanide on Cin-
namic Ester. — Dry sodium methylate representing 4 grams of sodium
was mixed with 20 grams of benzyl cyanide and the mixture heated to
140*^. This was now added to 30 grams of hot dnnamic ester and the
heating continued with constant stirring at the above temperature for
ten minutes. The mass was cooled, treated with strong h3rdrochloric
acid, well washed with water and then boiled with 95 per cent, alcohol
Partial solution took place accompanied by the separation of white crys-
tals, which increased on cooling the mixture. These were filtered off,
washed with a little alcohol, then a small quantity of benzene, dried
and boiled with sodium carbonate solution, again filtered and dried.
Yield, 9 grams.

After having tried all of the ordinary solvents without satisfactory

* Analysis by C. J. Frankforter.

SODIUM BENZYL CYANIDE WITH CINNAMIC ESTER. 599

results, amyl alcohol was found to be well suited for the purpose of re-
crystallization. The pure compound forms long obliquely pointed plates,
sometimes resembling needles, which melt at 231-3°. Insoluble in am-
monia, sodium carbonate solution, and petroleum ether; difficultly solu-
ble in alcohol, ether and benzene, soluble in hot acetic acid and amyl
alcohol. It is unchanged by heating In a sealed tube with concentrated
hydrochloric acid for 5 hours. It is, however, acted upon by a strong
solution of alcoholic potash. The analysis corresponded to the formula
CjjHjiOjN. From all data at hand it corresponds to the anhydride of
the acid resulting from the union of two molecules of cinnamic ester
with one of benzylcyanide.

The following formula expresses this constitution but is put forth
only tentatively:

The analysis gave:* C, 78.93; H, 5.59; N, 3.64. Calculated for
C„H„0,N: C, 79.04; H, 5.34; N, 3.55.

CHaCO

I \
C.H,— C— H \

C.H5— C— CN O

CeH,— C— H

CHjCO

The molecular weight was determined by the freezing point method,
using benzene as a solvent. The substance is so slightly soluble in this
substance that the determinations were not very satisfactory. The
mean of the two determinations gave 388 as the molecular weight. Cal-
culated for CjeHjiOjN, 396.

The Acid Obtained from the Supposed Anhydride. — ^When the compound
last considered is heated for 10 minutes with very strong alcoholic potash
it is dissolved and a sodium salt is obtained. This, upon acidifying, yields
the corresponding acid. The acid was purified by dissolving in ether
and evaporating the ether off in the presence of benzene. This was al-
lowed to crystallize and recrystallized again out of 50 per cent, alcohol.
The substance crystallizes in rectangular plates and under some con-
ditions in crystals resembling cubes suggesting the impurity encoun-
tered with the a,/9-diphenylglutaric acid. Soluble in ether and alcohol,
sparingly soluble in benzene, and boiling water. Melts at 213°.

The analysis gave: C, 75.38; H, 5.56; N, 3.69. Calculated for
C«HaO,N: C, 75-54; H, 5.56; N, 3.39.

Titration, 0.2530 gram of substance required 12.05 cc. of N/io NaOH.
Theory for CaeHjsO^N, 12.2 cc. N/io NaOH.

* Analysis by C. J. Frankforter.

6oo

S. AVERY AND FRED. W. UPSON.

Silver salt is an insoluble white powder, darkening slightly upon ex-
posure to light. It gave 34.40 per cent. Ag; calculated, 34.24 per cent.

We have attempted to saponify the cyanogen group of this compound
and thus obtain the tricarboxylic acid but it has resisted all attempts.
Neither potassium hydroxide nor hydrochloric acid in a sealed tube
gave the desired result. We expect to investigate the character of this
nitrogen atom and the salts of the acid more fully, as well as the exact
constitution of the anhydride from which it was formed.

Crbbcical Laboratory,

University of Nberaska,

Lincoln, Nbb.

THE SYITTHESIS OF CERTAIN AROMATIC SUCCIinC ACIDS.

By S. Avbry and Prbd W. Upson.
Received January 39, 1908.

In the preceding article by one of us and McDole it is pointed out
that in the synthesis of a,^8-diphenylglutaric acid apparently two stereo-
isomers were obtained. It was conceived, however, that one of these
might be the isomeric succinic acid formed according to the reaction:

CeHj— CH=CHCOOR + CeH^— CHNa— CN«C,H,— CHNa— CH— COOR

C,H,— CH— CN
This ester on saponification would jrield the unknown benzyl-phenyl-
succinic acid.

For the purpose of comparing this acid with the product described
in the preceding article it was decided to effect its synthesis in such a
way as to leave no doubt in regard to its constitution.

Accordingly 25 grams of sodium benzylmalonic ester were condensed
with 2 1 grams of bromphenylacetic ester. The heavy brown oily product
thus obtained was heated in a bomb with hydrochloric add but no cr\v
talline product was obtained, although a large amount of carbon dioxide
was liberated. The oil was then boiled with potassium hydroxide solu-
tion, in which it partially dissolved. On separating from the undissolved
portion and acidifying, crystals mixed with an oil came down. This
mixture was extracted several times with hot water, which on cooling de-
posited crystals in a fairly pure state. These were filtered off and washed
free from oily and resinous matter with a mixture of chloroform and
petroleum ether. After recrystallization from a mixture of ether and
chloroform, pure white, very fine, needle-like crystals melting sharply
at 176*^ were obtained. The acid is difiicultly soluble in hot, almost
insoluble in cold water; soluble in ether and alcohol; insoluble in chloro-
iorm and petroleum ether and difiicultly soluble in benzene.

Titration. — 0.0615 gram acid required 4.25 cc. N/io sodium hydroxide;
calculated for 2 acid hydrogens in Ci7H|e04. 4.32 cc.

CERTAIN AROMATIC SUCCINIC ESTERS. 6oi

The silver salt is a white amorphous precipitate. It gave 43.13 per
cent. Ag; calculated, 43.30 per cent.

The yield of the acid was very unsatisfactory and a considerable amount
was lost in repeated recrystallizations to obtain an absolutely pure product
from which to determine the melting point. As the analytical data
above given indicated that the desired substance had been obtained,
it seemed unnecessary to pursue the subject further, especially as in the
meantime the nature of the apparently low melting glutaric acid had
been determined.

While the work in this and the preceding article was in progress Hig-
son and Thorpe* published an article describing the synthesis of alkyl
succinic acids by the condensation of ethyl sodiumcyanoacetate with
aldehyde cyanohydrines.

Quoting from their article: "The condensation between ethyl sodio-
cyanoacetate and either a ketone or an aldehyde-cyanohydrine pro-
ceeds in nearly all the cases investigated very smoothly at the ordinary
temperature * * * * The reaction may be represented by the
following general equation :
CHNa(CN).COOEt + HO— C.R.(CN) -> C02Et.C.(CN).C.(CN).R-f-H20.

R Na R

4: 4c * 4: The alkyl derivatives of succinic acid can then be prepared
from these ethyl salts by hydrolysis with hydrochloric acid according
to the equation :

C02Et.CH(CN).CR2.(CN) -> C03H.CH2.CR2.C02H^
It seemed, therefore, of interest to see if benzyl cyanide could be made
to condense in a like manner with aldehyde cyanoh3^drines in the synthesis
of similar products. Accordingly, the first condensation attempted
was between benzaldehyde cyanohydrine and sodiumbenzylcyanide
which should give the nitrile of diphenylsuccinic acid. Thirty grams of
benzaldehyde were treated with a saturated solution of acid sodium
sulphite, and then with 29 grams of potassium cyanide, yielding 47.7
grams of the cyanohydrine. This was condensed with 49 grams of so-
diumbenzvlcvanide. Reaction :

H Na

I I CeHj— C— CN

QH^— C— CN + CeHs-C— CN = I -f NaOH.

CeHg— C— CN

OH H

The heavy black tarry mass resulting from the condensation was heated

over the steam bath for several hours with a slight excess of sulphuric

* J. Chem. Soc., 89, 1455.

602 S. AV^RY AND FRED. W. UPSON.

add. On cooling, the mass became thick and showed the presence of
crystals. These crystals, insoluble in benzene, were washed free from
the oily, foreign substance with benzene, and then purified by recrystal-
lization from glacial acetic acid. The first melting point on what ap-
peared to be the pure substance was 188-190°. Several recrystaUi-
zations raised the melting point to 232*^. It is recorded by Knoeve-
nageP that there are two isomeric modifications of this nitrile. One
melts at 160° and may be changed to the other isomer melting at 239-
240° C. After treating the compound in the manner described by
KnoevenageP the isomer melting at 239-240° was obtained.

A nitrogen determination gave 12.28 per cent. N. Calculated for
CieHiaNj, 12.09 V^^ cent.

The nitrile was saponified by heating in a bomb for five hours with
hydrochloric acid. The resulting crystals were purified out of alcohol
and melted at 229°.

Titration. — 0.10475 gram required 7.73 cc. N/io sodium hydroxide;
calculated for two add hydrogens in CieHi404, 7.75 cc.

Since both the acid' and the nitrile* are known and are fully described
in the literature, and since the analytical data given clearly indicate
that the desired end has been reached, no further examination of this
succinic add was made.

In order to test this reaction, using an aldehyde of the aliphatic series,
25 grams of isobutyl aldehyde were converted to the cyanohydrine and
condensed with sodiumbenzylcyanide. It was assumed that the nitrik
of phenylisopropylsuccinic acid would be formed according to the equa-
tion:

I (CH8),CH— CH— CN

(CH3)2CH— CH— CN + CeHj— CH-CN = | + NaOH.

I QjHg— CH— CN

The light brown oily condensation product was heated on the water
bath for some time with a slight excess of sulphuric add. On standing
several days and with occasional stirring the mass showed evidences
of forming crystals. When it had entirely solidified the crystals were
purified by crystallization, first from alcohol and then from a mixture
of petroleum ether and chloroform. White, feathery crystals, melting
at 126°, were obtained.

A nitrogen determination gave 7.21 per cent. N. Calculated for
C13H14N2, 14.16 per cent.

* Ber., 25, 289.
' Ibid., 25, 295.

■ Ann., 258, 87; Ber., 14, 1802; Ber., 23, 117.

* Ibid., 25, 289.

CERTAIN AROMATIC SUCCINIC ESTBRS. 603

A titiation at this point gave a similar result. 0.0856 gram required
3.82 cc. N/io sodium hydroxide; calculated for one acid hydrogen in
C^HisOaN, 4.38 cc. N/io sodium hydroxide.

These results seemed to indicate that the nitrile was partially saponi-
fied. Some of the nitrile was boiled with dilute sodium hydroxide and
the crystals that came down on acidification were purified by crystalliz-
ing from a mixture of alcohol and a little petroleum ether.

Titration. — 0.0756 gram required 3.45 cc. N/ 10 sodium hydroxide. Cal-
culated for one acid H in CjaHisO^N, 3.48 cc.

Apparently one nitrile group had undergone saponification during
the progress of the reaction. In this connection it may be noted that
in the various c)ranogen compounds studied, the readiness with which
saponification took place was greatly influenced by the neighboring
groups. Thus in the intermediate compound, (CHj),CH — CH — CH (i)

C.H,-CH— CN (2)
(i) is apparently very easily saponifiable, while (2) saponifies with more

CHj— COOH

diflSculty. The compound C^Hg — CH contains a cyanogen

CeH— CH— CN
group that saponifies with great difficulty, while the compound
C^5— CH— CHj— COOH

CjHj — C — CN contains a cyanogen group that appears to

C,Hj— CH— CHj- COOH

resist the ordinary methods of sappnification. The influence of neigh-
boring groups, especially the phenyl group, in retarding saponification
will be studied in detail later.

Some of the half nitrile of the phenylisopropyl succinic acid was con-
verted to the dibasic acid by heating in a bomb with hydrochloric acid.
The resulting mixture was extracted with hot water .from which fine,
pure white crystals of the phenylisopropyl succinic acid were deposited
on cooling. These were recrystallized from benzene containing a little
petroleum ether. The acid has a white powdery appearance in the mass
but under the microscope shows very minute colorless plates, melting
at 178 ^

The acid is very soluble in ether, alcohol and acetic ether, moderately
soluble in benzene, chloroform and hot water, and almost insoluble in
petroleum ether and cold water.

TUration. — 0.0868 gmm of the acid required 7.80 cc. N/io sodium
hydroxide. Calculated for 2 add H. in Cj3Hie04, 7.88 cc. N/io sodium

604 HARRY SNYDER.

hydroxide. Combustion, 0.1238 gram substance gave 0.3005 gram
CO3. This analysis gave : C, 66.2; H, 6.6. Calculated for C^HnO^: C,
66.1; H, 6.8.

Coiiclusions. — (i) The data here given show that succinic derivatiws,
made by reactions that leave no doubt as to their composition, are different
compounds from the isomers described in the foregoing article. Hence
these latter are in all cases glutaric derivatives.

(2) Sodium benzyl cyanide resembles sodium cyanacetic ester in its
action with aldehyde cyanhydrines. As in the case of the deportment
of benzyl cyanide with cinnamic ester, so here also there is a tendency
toward partial saponification. Having determined the deportment
of benzyl cyanide in this respect, since we were using benzyl cyanide
in the synthesis of alkyl glutaric and succinic acids when the article
by Higson and Thorpe appeared, it is not our intention to make any
closer approach to the field of work of Thorpe and his pupils.

Chemical I«aboratory,
The University op Nebraska, I^ikcoln.

INFLUENCE OF FERTILIZERS UPON THE COMPOSITION OF WHEAT.

By Harry Snyder.
Received January 23, 1908.

• There are a number of factors which are known to materially influence
the composition of wheat as (i) seed, (2) soil, (3) climatic conditions,
and (4) storage. These have been studied by a number of investiga-
tors, and in general it can be said that while the composition of wheat,
like that of all seeds, is fairly constant it is possible by increasing the
fertility of the soil, by seed selection, control of the soil moisture, sx-s-
tematic cultivation or irrigation, and by control of the chemical changes
incident to storage to favorably influence its composition and nutri-
tive value.

In order to study the influence of fertilizers upon the composition
of wheat, sixty samples grown at 12 different localities in Minnesota
and fertilized with different kinds of fertilizers were analyzed. At each
of the twelve different places where the fertilizer tests were made a uni-
form piece of land was selected and five quarter acre plots were staked
off. Each plot received similar treatment as to cultivation and seed-
ing. On one of the plots no fertilizer was used and on the remaining
plots, complete, potash, superphosphate and nitrogen fertilizers were
applied. The grain from each of the plots was harvested and threshed
separately, and bushel samples were shipped to the Chemical Labora-
tory of the Minnesota Experiment Station for analysis and milling and
technical tests.

Influence on Physical Qualities of Grain, — In many instances the fer-

COMPOSITION OI^ WH^AT. 605

tilizers exerted some special influence upon the growth of the crop, e, g.,
nitrogen used alone retarding maturity, and minerals used alone
hastening maturity. In some cases size and character of the kernels
were influenced by the fertilizers. Larger, better filled, and better col-
ored grain generally, resulted from their use, particularly where the
fertilizers perceptibly increased the yield. In eight trials the phosphate
fertilizers increased the weight of the grain per bushel and in two trials
the weight was the same as when no fertilizer was used. In five trials
potash increased the weight per bushel and in no case was the weight
decreased by the application of potash. In some cases the nitrogen
fertilizer increased and in other cases decreased the weight. In general,
the heaviest weight and best quality of wheat was produced on the fer-
tilized plots. In some of the tests, the phosphate, and in others the pot-
ash fertilizer exercised the greatest influence upon the quality of the grain,
as to weight per bushel and uniformity of kernels. Nitrogen alone did
not exert as great an influence toward improvement of the kernels as
the mineral elements alone; in a few instances, however, nitrogen alone
improved the glutenous character and general appearance of the grain.
From the tests made upon the different soil types of the state it would
appear that fertilizers may improve the quality of the grain, but the
kind of fertilizer element as potassium, or phosphorus required for
purposes of improvement depends entirely upon the individuality of
the soil on which the wheat is grown. Improvement in quality of
the grain follows as a result of increase in the fertility of the soil, and
a soil must be built up in the elements it lacks and these must be ascer-
tained by experiments. In many localities where these experiments
were made the climatic conditions were unfavorable, but it was noted
that with the more liberal supply of plant food in the fertilized plots
the quality of the grain as to weight per bushel, plumpness, maturity,
and imifomiity of kernels was better than on the unfertilized plots. This
would indicate that during unfavorable seasons crops produced upon
soils of low fertility are more susceptible to the adverse climatic condi-
tions than crops grown upon soils of high fertility.

Inplubncb op Fbrtiuzbrs Upon thb Composition op Wheat. — Average Com-
position.

Protein Ether Crude Nitrogen-

No. of Moisture. NX6**. extract. fiber. Ash. free extract.

Kind of fertilizer. samples. Percent. Percent. Percent. Percent Percent. Percent.

Nitrogen 12 10.03 1363 2,15 2.74 1.58 69.87

Potash 12 10.24 13.02 2.10 2.65 1.62 70.37

Phosphoric acid 12 10.39 12.65 2.19 2.73 1.73 70.31

Complete (N,K,0 and PjOj) 12 10.15 13. 17 2.22 2.76 1.69 70.01

No fertilizer 12 10.16 1304 2. 11 2.72 1.64 70.33

In all of the individual tests except one, the highest percentage of
nitrogen was secured from the wheat grown upon the plots receiving

6o6 HARRY SNYDER.

either nitrogen alone or the complete fertilizer of which nitrogen fonned
a part. A similar result was secured in 1905 from a more limited num-
ber of trials. Increasing the supply of nitrogen in the soil slightly in-
creased the amount of nitrogen in the grain. As previously noted, this
increase in nitrogen alone, unless associated with the mineral elements,
may result in a poorer quality of grain, for while nitrogen alone increased
the crude protein content of the gmin, to secure improvement in quality
as well, the nitrogen must be associated with the other essential elements
of plant food. The results indicate that in many cases it is possible
to increase the protein content of wheat one per cent, or more through
the use of fertilizers, also to secure an improvement in quality, although
the average increase in protein was small.

The influence of the nitrogenous fertilizers upon the form of the ni-
trogen in wheat was also studied. A number of investigators have
reported the presence of nitric nitrogen in plants. King and Whitson,
in the Eighteenth Annual Report of the Wisconsin Experiment Sta-
tion, page 220, state that in the case of oats in the **milk stage" grown
on soil very rich in nitrates, 2.64 per cent, of nitrogen was obtained by
the ordinary Kjeldahl method, but when the method was modified to
include nitrates, 3.12 per cent, was obtained. They also report nitric
nitrogen in com and potatoes.

For tlie purpose of determining the amount of nitric and other forais
of nitiogen in the wheat fertilized with nitrate of soda, three samples
were selected; wheat grown on University Farm, in the central western
and in the southwestern part of the state. The results of the analyses
are given in the following table:

Forms of Nitrogbn in Whbat Fbrthjzbd wtth NrrRAxe op Soda.

Wheat Wheat from Wheat from

from Uni- central western southwestern
versityfarm. part of state, part of state.
Percent. Percent. Percent

Nitrogen (Kjeldahl process) i . 90 i . 77 2.07

Alhuminoid nitrogen i . 78 i .69 1 .98

Nitrogen (modified to include nitrates) 2.01 i . 86 2 . 22

Excess of modified over ordinary Kjeldahl

process (nitrates?) o. 11 0.09 o. 15

Nitrogen as nitrites 0.00006 0.00004 0.00008

The difference between the nitrogen obtained by the modified and
by the ordinary Kjeldahl process can not all be considered as nitric nitro-
gen but when the nitrate of soda was used as a fertilizer a small amount
of nitric nitrogen was found by qualitative tests to be present in the
wheat. The albuminoid nitrogen and the modified nitrogen determina-
tions show that not all of the nitrogen in the plant was present as pro-
teins. Qualitative reactions of all of the flours made from the wheats
fertilized with nitrate of soda and organic nitrogen showed the presence

COMPOSITION OI^ WHEAT. 6o*J

of nitric nitrogen and also traces of nitrites. Some of the wheats grown
upon the plots where there was no nitrogen in the fertilizers gave the
same reactions, while others did not. The nitrogen content of the wheat
was slightly increased by applications of nitrogenous fertilizers to the
soil, but as previously stated the bread-making qualities of the flour
from such wheat are not necessarily improved. Similar results
have been reported by Hall, of the Rothamsted Station; ''Again, as
we have seen, 'strength* is generally associated with a high nitrogen
content, yet the wheats grown on some of the Rothamsted plots, where
so large an excess of nitrogenous manure is applied that even the grain
becomes more nitrogenous, instead of becoming stronger only gets in-
credibly weaker." — Journal of the Board of Agriculture, September,
1904, page 332. It is quite evident that the form as well as the amount
of nitrogen must be taken into consideration in studying the bread-mak-
ing qualities of flour.

Influence on Bread-making Value, — Forty-one of the sixty samples
of the wheat were milled at the experimental flour mill of the Minnesota
Experiment Station. The protein content of the flour was determined,
and technical bread-making tests were made by an experienced baker,
accustomed to making tests for a large flour mill.

The wheat from three of the nine places, grown upon the plots fer-
tilized with phosphates, produced flour that made the best bread. From
two of the places the wheat fertilized with nitrogen made the best bread ;
from two fertilized with potash, and from two the complete fertilized
wheat.

In thirty of the forty-one tests the fertilizers which gave the largest
yields per acre produced wheats of the highest bread-making value,
while in ten of the tests the best quality of flour was secured from the
fertilized wheats which did not show the largest yield per acre. While
yield and bread-making quality are both improved by the use of fertil-
izers they are not necessarily both improved to the same extent by the
same fertilizer.

There appears to be no constant relationship between the per cent,
of protein in the grain and flour, and the bread-making value, and while
it is possible to increase the amount of protein in flour by the use of
nitrogenous fertilizers the bread-making value of the flour is not propor-
tionally increased. In many instances the increase in nitrogen content
imparted a negative value.

The experiments taken as a whole show that not only the yield of
wheat, but also the bread-making value can be enhanced by increasing
the fertility of the soil, and that there is a very close relationship be-
tween the amotmt of available plant food in the soil, and the quality of the

6o8 CHARLES D. HOWARD.

wheat produced upon that soil and its bread-making value. Credit is
due Mr. L. O. Bemhagen for assistance rendered in the analytical work.

Agricultural Experiment Station,
St. Paul, Minn.

THE PRECIPITATION METHOD FOR THE ESTIMATION OF OILS
IN FLAVORING EXTRACTS AND PHARMACEU-
TICAL PREPARATIONS.

By Charles D. Howard.
Received February 6, 1908.

The polariscopic method for the estimation of essential oils in com-
mercial extracts is of but limited application, the oils of lemon and orange
being the only ones that can be accurately determined in this maimer.
For the estimation of such oils as peppermint, clove, wintergreen, and
many others, two procedures are open: (a) the application of methods
for the estimation of the most important constituents of these oils, such
as menthol, eugenol, or methyl salicylate, and (6) the method by pre-
cipitation as suggested by Mitchell^ and now adopted with modifica-
tions as official by the A. O. A. C.

In most instances the first procedure, as applied to extracts, is ob-
viously capable of affording but little better than a general idea as re-
gards strength or quality, while results by the second process invoh^
the application of a large and variable correction. Thus, in the case
of lemon oil MitchelP found that results near to the truth were obtain-
able only in the presence of a relatively large proportion of oil — a 6 per
cent, extract, for instance, showing 4.80 per cent, recoverable, while
a 2.50 per cent, extract afforded by this procedure less than one-half
of the oil actually present.

With a less proportion of oil the error becomes still greater, and when
we consider in addition that this error apparently varies not only with
the quantity of oil, but with the kind, it is evident that for the examina-
tion of many of the miscellaneous extracts and essences now on the mar-
ket— many of them containing as they do but one or two per cent, of oil—
the method as at present carried out is of but very limited value. This
fact will be appreciated by any who has attempted to examine some
of the cheaper grades of peppermint essence.

By the modified method here proposed the writer has obtained most
excellent results. The procedure has the advantage that no correction
whatever for oil retained in solution is necessary, and moreover, with
the single exception of almond extract, it affords equally accurate re-
sults in the case of alcoholic solutions of almost any one of the large

* This Journai^, 21, 11 32 (1899).
' Loc. cit.

PRECIPITATION METHOD FOR THE ESTIMATION OF OII^S. 609

class of essential oils. The advantages of simplicity and rapidity of
execution may also be claimed, it being possible to carry through a series
of several determinations in ten minutes.

Procedure, — ^To 10 cc. of the extract, pipetted into an ordinary Bab-
cock milk bottle, are added in the following order, 25 cc. of cold water, i
cc. hydrochloric add of 1.2 specific gravity and 0.5 cc. chloroform. The
mouth of the bottle is then closed by the thumb and vigorously shaken for
not less than one minute. By this means all of the oil is dissolved by the
chloroform, while the latter, in saturating the water, apparently serves
to displace any appreciable trace of oil otherwise retainable by the al-
cohol-water mixture. The bottle is now whirled in the centrifuge for
one and one-half to two minutes and the resulting clear supernatant
liquid is removed to within 3 or 4 cc. by the insertion of a glass tube of
small bore connected with an aspirator. To the residue i cc. of ether
is added and the contents of the bottle well agitated. Holding the latter
at a slight angle it is plunged to the neck in a boiling waterbath, and,
giving a gentle rotary motion, is maintained at this temperature for
exactly one minute. This step is best carried out by removing one of
the small rings from a water- or steambath and holding the bottle in the
live steam. The ether serves the purpose of steadily and rapidly sweep-
ing out every trace of chloroform — a result that would be otherwise
attainable only with considerable difficulty and loss of oil. The latter
by this procedure has been found to be inappreciable. Finally the bot-
tle is cooled and filled with water at room temperature so as to bring
the oil into the graduated stem of the bottle, and after centrifuging for
one-half minute the reading is taken to the highest point of the menis-
cus; the reading multiplied by 2 gives the per cent, of oil.

If it be desired, a special milk bottle with stem of smaller bore, or a
skim milk bottle provided with a straight introductory tube, may be
employed. It has been found, however, that with the ordinary Bab-
cock bottle there is no difficulty in securing check results to one-tenth
per cent.

Modification for the Heavier Oils, — In the case of oil of wintergreen
it was found to be impossible to secure a compact readable column of
the oil by means of salt solution. This procedure also proved to be
not satisfactorily applicable for almond extract. For the estimation
of these oils use was at first made of a specially devised form of the so-
called Hortvet tube, in which the bore of the stem is so reduced that
the graduated portion contains 2 cc. instead of 5 cc, the subdivisions
having therefore the same values as in the case of the Babcock bottle.
Using this form of tube, in the case of almond extract the most satisfac-
tory results were obtained by employing double the quantities of chloro-
form and ether specified. Working in this manner it was found pos-

6 10 CHAIUU^S D. HOWARD.

sible to recover practically loo per cent, of the oil from a i per cent,
extract, but with stronger extracts 80 to 90 per cent, only proved re-
coverable.

In most cases, however, the use of the ordinary Babcock bottk with
a suitable heavy liquid will be found to be preferable. For this pur-
pose, except with the oils of cinnamon and cassia, the use of salt solu-
tion is out of the question. Trials were made of the applicability of
diluted sulphuric acid, diluted glycerol and of sugar solution. The
two latter were found to serve well for the lighter oils, but with winter-
green a gravity of not less than 1.2 is requisite and this involves a too
high degree of viscosity. On the other hand, diluted sulphuric add
(1-2) was found to answer admirably. While it might be objected
that the acid would tend to decompose some of the oils — notably those of
clove and cinnamon — and thus afford low results, yet it was found that
if agitation is avoided. and the temperature of the add mixture does
not exceed 25°, no readable error is involved.

As a result of a large number of trials of this method, it was found
that (except in the case of almond extract) not only could practically
the theoretical amoimt of oil be recovered from i, 3, 5 and 10 per cent,
strengths of alcoholic solutions, but that also in the case of lemon and
orange extracts results thus obtained very generally agreed to Vu P^
cent, with those obtained polarimetrically, if the factor designated hy
Mitchell and Leach (3.4) was used.

It is well recognized that results by the polarimetric method of ex-
amining lemon and orange extracts are not to be implicitly relied
upon as indicating in all cases the true proportion of oil present. The
use of small quantities of cane sugar in the preparation of these extracts
is apparently somewhat more common than has been generally sup-
posed. Without doubt, so-called "washed" or "distilled" oil is being
used to some extent in the making of extracts, and furthermore the
addition of high polarizing orange turpenes for the purpose of increas-
ing the rotatory power is a perfectly practicable form of adulteration.

The precipitation method, therefore — aside from merely providing
material for a refractometric examination — affords a direct and vahiabfc
check on the polarimetric results. For instance, if the results by pre-
cipitation are materially lower than those obtained polarimetrically,
there is ground for the suspicion that other than a "straight" (ril has
been used ; if on the contrary, they are higher and the oil is present in
moderately large quantity, such would be proof -positive either of adulter-
ation with a foreign oil, or else of the use of an oil that had undergone
marked deterioration, while if the polarimetric reading was but sHght,
or zero, any precipitated oil might be assumed to represent citxal or
one of the other lemon oil substitutes.

TWO TESTS OF RED LEAD. 6ll

The results presented below were obtained on solutions of definite
strength prepared with 90 per cent, alcohol. Results obtained with com-
mercial extracts are also submitted.

strength. Oil recovered.

Extract. Per cent. Per cent.

Lemon 5.0 5.0

i.o i.o

Peppennint 1.0 1.0

30 30

5.0 5.0

Clove 1.0 1.0

10. o 10.2

Cassia 1.0 1.0

Wintergreen 1.0 i.i

2.0 2.0

5.0 51

Bitter almonds 1.0 1.0

30 2.5

Citral (pure, optically inactive) 3.0 3.0

" 0.5 0.5

Commercial Extracts.

Variety. Oil by precipitation. Oil by polarizatior

Lemon 4.8 4.80

4-6 463

, 4.0 4.10

50 500

440 450

4-8 470

Peppermint 3.8

5-6

12.4

Cinnamon 3.0

Checkerberry 4.0

ia.5

Rose 0.6

Nb'W Hampshire Laboratory op Hygiene,
CoKCORD. New Hampshire.

«
tt

It
«<

A COMPARISON OF TWO TESTS OF RED LEAD.

By ET7GENE K. DUNLAP.

Received March 14, 1908.

The tests generally used for the determination of the amount of free
litharge, PbO, in red lead are the lead acetate test and the lead peroxide
test. The latter is made in two ways, by the gravimetric and volumetric
methods.

We will first consider the lead acetate test. This test depends upon
the solubility of lead oxide in a solution of lead acetate. A weighed
quantity of red lead is taken, and to it is added, in a beaker, a like quan-
tity of lead acetate crystals dissolved in about 150 cc. of hot water. After

6l2 EUGENE E. DUNLAP.

stirring, the solution is brought to a boil, and allowed to boil from ten to
thirty minutes. It is then filtered through a tared filter, the residue
washed thoroughly, dried, and weighed. The loss of weight between the
amount of red lead taken and what remains is the weight of free
litharge.

The lead peroxide test can be used in two ways, the gravimetric, which
is a direct, and the volumetric, which is an indirect method. We will
first consider the gravimetric method. This consists in treating the red
lead with nitric acid in a warm solution. The peroxide formed is col-
lected on a tared filter, washed thoroughly, dried and weighed. The
amount found is then calculated to PbgO^. The loss between the original
sample taken and the amount of PbjO^ found is the weight of the free
litharge.

The volumetric method consists of changing the red lead into lead per-
oxide, heating gently and adding a known quantity of N/s oxalic add
solution, then boiling. This decomposes the peroxide of lead. While
still hot, N/5 permanganate solution is added until the color remains for
a few seconds. The volume of N/5 permanganate solution required is
deducted from the amount of N/5 oxalic acid solution taken, the result-
ing volume of the N/5 oxalic acid solution is calculated to lead peroxide
and from that of red lead, PbjO^.

So much for the tests themselves. Now red lead as it is made com-
mercially, no matter by what process, is pig lead carried through the
various stages of oxidation until we come to the point where oxidation
practically ceases. The oxygen which is used for oxidation is taken from
the air at the temperature at which the lead is worked in the furnaces.
In some cases the first part of the oxidation is obtained from oxygen
given up by chemical compounds in order to produce new chemical
compounds. Therefore from the methods of producing red lead,
we can readily see that it is impossible to obtain a pure compound,
PbgO^. Red lead is therefore not a pure compound, but consists, outside
of very slight quantities of impurities such as lead sulphate, silica,
oxide of iron, etc., of various oxides of lead, namely protoxide,
PbO; sesqtdoxide, PbjOjj and minimum, Pb304. Now the sesquioxide",
PbjOg, is insoluble in a solution of lead acetate, while on the other
hand it is soluble in nitric acid. This is one of the fallacies of
both tests, but sesquioxide of lead is not harmful in red lead in connec-
tion with its physical properties, and should therefore not be taken into
consideration as an impurity.

On the other hand, litharge, or the protoxide of lead, if present in con-
siderable quantity, is objectionable and especially deleterious to the
physical properties of red lead. It is on account of this impurity that
red lead is tested, as it is beneficial not only as a protection to the

TWO TESTS OF RED LEAD. 613

buyers, but also to the manufacturer, who loses profit if a considerable
quantity of protoxide of lead is present.

In order to compare more closely the tests of lead acetate, and lead
peroxide, I prepared some red lead as pure as I could get it, by boiling
with lead acetate solution and very dilute acetic acid, washed it thor-
oughly, put it through a No. 21 silk bolting-cloth and dried it. After it
was thoroughly dry I made an analysis of it by lead sulphate method and
obtained an average result of 99.95 per cent, pure red lead.

Part of this sample with the lead acetate test gave average results of
99.98 per cent. Other samples of this red lead were put through the per-
oxide tests, using different conditions in each case and gave the following
results :

No. I sample. — 10 cc. nitric acid, sp. gr. 1.42, 150 cc. of hot water =
99.16 per cent.

No. 2 sample. — 30 cc. nitric acid, sp. gr. 1.42, 150 cc. of hot water ==
94.64 per cent.

No. 3 sample. — 50 cc. nitric add, sp. gr. 1.42, 150 cc. of hot water =
89.02 per cent.

No. 4 sample. — 50 cc. nitric acid, sp. gr. 1.42, 150 cc. of hot water
and boiled = 77.66 per cent.

The above shows that lead peroxide is aflFected under certain conditions
and that it is soluble in the stronger acid and also upon boiling. There-
fore, this test is practically useless in the test for purity of red lead and
should not be used. The volumetric method depends upon the formation
of peroxide, but gave lower results in all the above cases.

Now red lead which is made from the protoxide of lead produced in
the formation of nitrite of soda, will, after it comes through the various
processes, contain a small quantity of sodium hydroxide and sodium
nitrite. The peroxide test in this case, is even more useless than in the
case of red lead made by the processes which carries on the oxidation
from pig lead to red lead direct in the furnace. The nitrite of soda
present acts upon the peroxide of lead and gives even greater solubility
than if added after the peroxide is formed.

In the case of the lead acetate test upon such samples of red lead we
first neutralize the caustic soda with very dilute acetic acid, and then
proceed with the test. Of a quantity of tests made upon red lead con-
taining caustic soda and less than 005 per cent of sodium nitrite, the lead
acetate test gave results which varied very little, but in the case of the
peroxide test the results were quite varied, some showing a diflFerence of
about 10 to 12 per cent.

It seems, therefore, that the peroxide test is much less reliable than
that with lead acetate.

2016 North Blbvbkth Strbbt,
Pbiladblpbia, Pa.

6 14 NOTES.

NOTES.

Notes on the Separation of Silica and Alumina in Iron Ores. — ^A read-
ing of G. W. Dean's "Notes on the Determination of Silica and Ahimina
in Iron Ores,"^ and "The Determination of Silica in Iron Ores Contain-
ing Alumina,*'^ led to a comparison of the three methods outlined in
the above-mentioned articles with the sodium carbonate fusion method.

A brief statement of the four methods may not be out of place ; they
are as follows:

(i) Sodium carbonate fusion. One gram of the sample was treated
with concentrated hydrochloric acid and evaporated to dryness twice,
then dissolved in hydrochloric acid, the solution filtered and the
residue fused with sodium carbonate. The fusion was treated with

hydrochloric acid and evaporated twice to dryness and the silica deter-
mined as usual.

(2) Double dehydration of ore in hydrochloric acid, solution in same,
ignition of the insoluble residue in platinum and re-solution in hydro-
chloric add, and silica determined.

(3) Ignition of ore in porcelain with sulphur, solution in hydrochloric
acid, and silica determined.

(4) Ignition of ore in porcelain without sulphur, solution in hydro-
chloric acid containing i gram of stannous chloride in 225 cc, and silica
determined.

Alumina was precipitated as phosphate.

The ores used were those encountered in routine work and were as
follows: (a) Brown hematite-limonite mixture, (6) brown hematite,
(c) red-brown hematite mixture.

The following figures are averages of numerous determinations:

Ores.
U^. (*V (Q.

» * t < * » < • »

Methods. SiOj. AUOs. SiOs. AlsOg. SiO*. AlfOi.

(i) 770 3" 5.52 1-44 13-31 400

(2) 8.90 2.49 6.01 1. 15 13.76 3.70

(3) 892 2.37 6.00 1.07 14.07 3.44

(4) 8.71 .... 6.17 .... 14-17

The determination of alumina subsequent to (4) was, of course, im-
possible.

The following figures show the amounts of alumina soluble in hydro-
chloric acid before the ignition in method (2), and the amounts liber-
ated by the ignition :

* This Journal, 29, 1208.
« Ibid., 28, 882.

NOTES. 615

AlfiOs.

Ore. No. i. No. 2.

(«) - .-.. 0.97 1.53

(b) 0.62 o. 59

Three different ignition temperatures for method (2) were tried on
ore (a). A low red heat, barely sufficient to redden the crucible bot-
tom, was applied till the filter was carbonized and about half consumed ;
this gave results as follows: Silica 8.92, alumina 2.48. A moderate
red heat, sufficient to redden the crucible (15 g.) clear to the top, and
applied till the filter was entirely consumed, gave : Silica 8.88 and alumina
2.51. Igniting to bright redness gave figures much higher on silica and
lower on alumina and so discordant as to be totally worthless.

The sodium carbonate fusion being considered a standard method for
the determination of silica, it is evident from the above comparisons that
ignition methods (2), (3) and (4) are not universally applicable to the
accurate determination of silica and alumina in iron ores.

T. George Timby.

Stbvbnson, Minn.

An Apparatus for the Quantitative Electrolysis of Hydrochloric Acid, —
A desirable piece of apparatus for a lecture experiment is one that is
easily put together for operation, certain to give correct results without
any time-consuming preparation or preliminary adjusting. In the opin-
ion of the writer, the following piece of apparatus for demonstratmg
the volume relation of the hydrogen and chlorine obtained in the electrol-
ysis of hydrochloric acid fulfils the above requirements of "desirable
apparatus."

The special points of difficulty to be overcome in an apparatus for
this purpose are (i) the mixing of the cathol3rte with the anolyte; (2)
the solubility of the chlorine in the water over which it is to be collected.
In the apparatus here described, the mixing of the catholyte with the
anolyte is practically entirely prevented by surrounding the anode with
a porous cup. The second difficulty is eliminated by discharging the
chlorine into the bottom of a tall, large cylinder, and collecting the equal
volume of air forced out of the top of the cylinder. These ideas are
embodied in the following design.

The electrolysis is carried on in a cylindrical glass jar, i; 2 is a cylin-
drical porous cup ; j, the carbon anode ; and 4 the cathode of sheet plat-
inum. The glass jar and the porous cup are both closed with rubber
stoppers, through which are inserted the carbon anode, the platinum
wire connection for the cathode, and the delivery tubes. The chlorine
delivery tube is fitted with a rubber stopper to the lower opening of

6i6 NOTES.

the tall cylinder p, and the outlet tube for air is fitted with a mbbet
stopper to the top of the cylinder. A loose cotton phjg jo, is placed
in the cylinder to* retard the mingling of chlorine and air. A suitable
two-way stopcock interposed as shown at j will be found very con-

venient. By means of it the chlorine may be diverted from enteiiog
the cylinder, or chlorine may be blown out of the cylinder without ais-
connecting the apparatus.

The air (or chlorine) and the hydrogen delivery tubes terminate in
two small glass tubes, the ends of which are drawn to ^oall openings to
discharge the gases in small bubbles. These tubes are fitted into a wooden
block by means of which both terminab can be placed under the col-
lecting tubes simultaneously. The rubber tubes which connect the
terminals with the main portions of the delivery tubes are made as short
as possible, and are coated with shellac, to prevent diffusion, particu-
larly of the hydrogen. Two Inverted burettes may serve as collecting
tubes.

To start the apparatus, the anolyte is saturated with chlorine by add-
ing some crystals of potassium chlorate to it. The mixture may be used
immediately. The anolyte and the catholyte should fill the vessels
to practically the same level. After connecting up properly, electroh-as

NOTES. ' 617

need be carried on only a minute or two (during which time the chlorine
is preferably diverted from entering the cylinder) and then the apparatus
is ready for the demonstration.

When put away, the electrolysis vessel, etc., should be filled with
distilled water.

1 am indebted to Dr. E. P. Schoch, of this laboratory, for the funda-
mental notions of this design. J. B. Lewis.

Thb Univbrszty op Texas,
School of Chemistry.

A Supposedly New Compound from Wheat Oil. — ^While engaged in
investigating the properties of bleached flours* at the University of Ne-
braska the writer had occasion to extract about 100 cc. of wheat oil from
unbleached flour by means of ether. This oil on standing for a short
time was observed to deposit a considerable number of small, white
crystals. Some of these crystals were removed from the oil by suction
and washed on the filter with ether in which they are not readily soluble.
The crystals so obtained were oily to the touch and melted to a color-
less liquid at 93-94°. By recrystallization from absolute alcohol this
melting point was raised to 96.5°.

That the compoimd contained nitrogen was proven by the usual tests.

Some attempts were made to saponify the compound by boiling with
10 per cent, alcoholic potash but the melting point remained unchanged.

At this point the investigation was broken off, owing to the fact that
it was not directly concerned with the bleaching of flours. There was
not obtained sufficient of the compound for a complete investigation
although it is hoped by the writer to prepare larger quantities in the
near future. Ross A. GorTner.

Chemical I«aboratory.

University of Toronto,

February i6, 1908.

Determination of Phosphorus in Ash Analysis, — In our article in the
March number of The Journal, attention was called to the fact that
when the ash of cereals is burned at too high a temperature or fused, the
method of determining phosphoric acid by extracting the ash with hot
nitric acid gave an apparent loss of the phosphorus, although no appre-
ciable loss in the ash occurred. More recent investigations show that the
loss is not entirely due to volatilization of the organic phosphorus as was
supposed, but to a conversion of the phosphorus to a form which is not
precipitated by ammonium molybdate.

The following results show that even boiling the ash with strong nitric
acid for an hour is not quite suflBcient to recover all of the phosphorus.

* Alway and Gortner, This Journal, 29, 1503 (Oct., 1907).

6l8 • REVIEWS.

Whenever the ash has been burned at too high a temperature it is neces-
sary therefore to determine the phosphoric acid in the ash by means of
the Neumann method (digestion with 5-10 cc. of a mixture of equal parts
concentrated sulphuric and nitric acids).

These results show that neither phosphoric acid nor ash is appreciably
volatilized on high ignition, but that to recover the converted phosphoric
add from the ash, the Neumann or an equally eflScient method must be
used.

Results.

Sample.
A....
B....

Ash

Phosphoric acid.

redness.

(I)

Bright
redness.

(2)

Hot

nitric acid.

(I)

Hot Boiled i hr. Neumann
nitric acid. Nitric acid. method.
(2) (2) (2)

2. II

2.08

1. 10

0.42 0.99 1.09

2.18

2.16

1. 16

0.48 1.07 1. 16

Sherman LEAvn"r,
J. A. LeClerc.

Laboratory of Vbobtablb Physxoi^ooical Chbmistrt.
Bureau of Chbmxstry, Washington.

REVIEWS.

REVIEW OF nrORGAinC chemistry for 1907.

By Jas. Lewis Howb.
Received January 13, 1908.

The most important paper which has appeared in morganic chemistry
during 1907 is the lecture which was delivered before the German Chemical
Society by Werner, on "The Problems of the Constitution and Con-
figuration of Inorganic Compounds' ' (Ber. , 40, 15). The subject was treated
under five heads, ^2:. : constitution of the metal-ammonia salts; constitu-
tional relations between the metal-ammonia salts and complex salts; rela-
tions between the metal-ammonia salts and the hydrates; the special rela-
tions of complex inorganic radicals; polynuclear metal-ammonia salts.
The paper, which unfortimately does not admit of abstraction within the
limits of this review, gives the latest and most comprehensive views of
the author, which are winning general acceptation, as offering the only
rational theory of the constitution of complex inorganic compomid&
The application of Werner's view to many classes of compounds is as yet
far from clear or satisfactory, but an important beginning has been made
which bids fair to be second only in importance to the doctrine of valence
itself.

Two other important papers from a theoretical standpoint have been
by Pfeiffer and by Werner on the theory of hydrolysis and the theory of
bases (Ber,, 40, 4036, 4133). According to the ideas of these chemists
hydroxo-compounds, that is, those containing the undissociated hy-
droxide group, when treated with acids give salts, not by substitution
as generally assumed, but by addition, and the salt formed is primarily
an *aquo' salt, though it may secondarily go over into an anhydrous

REVIEWS. 619

salt by the loss of water. Appl)dng this idea in its simplest form we
would have for the neutralization of cuprous hydroxide the following:
CuOH 4- H+.Cl- = Cu (OHa) + + C1-. Applying this to the theory of hydrol-
ysis, we have the reaction, Me(OH2)X = MeOH + HX, or Me(OHj)+ :^
MeOH + H+ , in which hydrolysis is conditioned by the tendency of the aquo-
metal ion to break up into a hydroxo-compound and the hydrogen ion. Ac-
cording to the theory of Arrhenius, on the other hand, we have the hydrolytic
reaction : MeX + HaO = MeOH + HX, or Me+ + HjO = MeOH + H+. Here the
dissociation of water is the determining factor in hydrolysis, while in the
theory of Pfeiffer, it is of little influence. This may be, however, taken
account of in the reaction which conditions the basic nature of the hy-
droxo-compound: MeOH-f-HjO ::^ Me(OH2)+-hOH"-. Here the water
mokcule is dissociated in the formation of the aquo-metal ion, the hy-
droxyl ion being left and imparting the basic reaction to the solution.
The strength of the basic character of the hydroxo-compound is thus
conditioned upon the tendency of the metal to form aquo-metal ions,
and upon the dissociation of water. According to Arrhenius the basic
character is independent of the dissociation of water and only dependent
upon the dissociation tendency of the metal hydroxide. In Pfeiffer's
theory hydrolysis is practically a measure of the tendency of the aquo-
metal ion to break up, and of the hydroxo-compound to imite with the
hydrogen ion. The application to amphoteric hydryxides is best seen by
an example, as of zinc hydroxide :

Zn(OH), (base) + 2H+=Zn(0Ha)a++, and

Zn(OH), (acid) -f (0H-) = Zn(OH),-".

In his paper on the theory of bases this idea is further developed by
Werner and bases are divided into two classes: the anhydro bases which
unite with water to form a hydrate which is dissociated in aqueous solu-
tion into a complex cation and a hydroxyl anion; and aquo-bases, which
comprehend all products formed by the addition of water which dissociate
in water giving hydroxyl anions. A number of different classes of anhy-
dro bases exist, as oxygen, nitrogen, phosphorus bases, etc. While the
theories of Pfeiffer and Werner have been worked out chiefly in the effort
to explain the metal-ammonia bases, they bid fair to throw much Ught
on the more general reactions of inorganic chemistry. The above is but
a bare outline and the original papers will well repay a careful perusal.

That which may prove to be the most important inorganic research of
the year is one which has been presented as yet only in two preliminary
papers by Sir WilHam Ramsay and Cameron (J. Chem. Soc, 91, 1266,
1593)* By the action of a large quantity of radium (87.7 mg.) as bromide
and sulphate on water, a mixture of hydrogen and oxygen was obtained.
On explosion, a small quantity of hydrogen and emanation remained.
An important characteristic of this emanation is its rapid decrease in
volume, which seems to point to a chemical change, perhaps into a di-
atomic gas. A calculation of the life of radium ik)ints to a period of only
236 years, a lower value than that found by Rutherford and others.
The attempt to account for the excess of hydrogen led to most remarkable
results. With the thought that if the emanation acted on the salt of a
heavy metal, this would be liberated instead of hydrogen, a solution of
copper sulphate was submitted to the action of radium. No copper was

620 REVIEWS.

deposited, but on removal of the copper, lithium was recognized in the
solution by the spectroscope. Numerous experiments were carried out,
with every possible precaution against error, but each time with similar
results. A small quantity of sodium seems also to be formed. When
lead nitrate is used in the place of copper sulphate, only sodium seems
to be present. Ah examination of the evolved gases showed that from the
emanation alone helium is formed, in the presence of pure water, neon,
while in the presence of copper sulphate the gaseous product is argon,
with no trace of either helium or neon. The suggested explanation of
these remarkable phenomena is that the a-particles are, contrary to the
view of Rutherford, not identical with heUum, but when they collide
with the atoms of the emanation, if nothing but the emanation is present,
helium results. If, however, the emanation is mingled with the heavier
water molecules, the decomposition does not go so far, and neon is the
result. In the presence of the still heavier copper atoms, the decom-
position remains at argon. By the action of the a-particles atoms seem
to be decomposed into other atoms of the same group, thus copper is
broken up into sodium and lithium. The authors also find that a sohi-
tion of thorium nitrate produces continuously minute but clearly de-
tectable quantities of carbon dioxide. The importance of these experi-
ments in opening up new lines of research can hardly be overestimated.
In this connection may be noted the work of Boltwood (Am, /. Scu (4),
22, 537 ; 24, 370) on tte relation of radium to uranium in camotite. From
the fact that the amount of radium obtained in a given time from uranium
is less than theoretically would be formed if radium is an immediate
product of uranium, the formation of an intermediate product seemed
probable. This was found in camotite, and at first supposed to be identi-
cal with actinium, but further investigation showed that it could be
separated from actinium, and that it differs from it in several important
particulars. It seems without doubt to be a decomposition product ot
uranium and the progenitor of radium. Boltwood proposes for it the
name ionium.

It has been suggested by Coblentz (Jahrb. Radioaktiv. u. Elektranik,
3, 397) that it should be possible to determine whether water is present
in crystallized compounds as water of crystallization or as so-called water
of constitution, by the presence or absence of the absorption bands of
water in the injfra-red spectrum. The results obtained from a number of
minerals and compounds gave results quite in accordance with the or-
dinarily received ideas on the subject, but there were also results which
were anomalous, such as the presence of water of crystallization in cane
sugar, though not in d-fructose and rf-rafl&nose. Similarly absorption
bands for the hydroxyl group were found in talc but not in serpentine.
The method seems to offer possibilities of throwing light upon molecular
structure, but needs much further investigation.

The subject of isomorphism continues to attract some attention, es-
pecially in its connection with molecular volume. Gossner (Z. KrysL,
43f 130) concludes that isomorphous substances, while in general not
possessing the same molecular volume, do not differ to any considerable
extent. If there is much difference in molecular volume, the series of
mixed crystals will not be continuous. In another paper (Ber., 40,
2373) he compares the members of the series of double alkali sulphates

REVIEWS. 62 1

of nickel, cobalt, copper and zinc, and also the fluosilicates of the same
metals. In the latter case the nickel and zinc salts have nearly the same
molecular volume and form a continuous series of mixed crystals, while
the cobalt and copper salts differ considerably in molecular volume and
there is a break in the series of mixed crystals from 10-30 per cent, of
the cobalt salt. In considering the isomorphism of the elements, Tam-
raann (Z. anorg. Ckem., 53, 446) holds that Mitscherlich's rule that
similarly constituted compounds are isomorphous cannot apply, because
we know nothing regarding the constitution of the elementary molecule,
and he concludes that those elements which are chemically analogous are
isomorphous. In geneml, elements of the same group in the periodic
system form mixed crystals and not compounds, but occasionally ele-
ments of different groups form a much more marked series of mixed
crystals than those of the same group. This is particularly apt to be the
case when the elements in question are chemically similar and have high
melting points. Indeed, the temperature of crystallization seems to be
even more important in determining the power of elements to form mixed
crystals than is chemical analogy. As a rule, elements of high melting point
separate out as mixed crystals, while those of low melting point crystallize
as the pure metal. If the melting point of both the metals in a binary
mixture is high, a complete series of mixed crystals may be expected to
separate.

Fused salt mixtures have been largely studied from a thermometric
standpoint, but Shemchushny (/. Rtiss Phys. Chem, Ges., 38, 1135) has
recently begun an investigation of the microscopic structure of fused salt
mixtures, by etching with an appropriate solvent the polished surface
of the solid melt. From the study of several series of binary mixtures,
such as those of potassium chloride-potassium chromate, he concludes
that the micro-structure of salt mixtures differs little from that of metal-
lic alloys.

An interesting case of the action of light, which seems to be recognized
for the first time, is described by Alefeld (Z. wiss. Phot, 4, 364; Chem.-
^^•> 30, 1087, 1 127). In his earlier experiments, a solution of colo-
phonium was spread thinly on a glass plate and dried for fifteen minutes
at 100°. The plate was then exposed under a photographic negative for
half an hour to direct sunlight. On heating the negative more highly,
a clear positive appears on the plate. This action was strongest on ex-
posure to blue light, and careful experiments showed that it was not
due to the action of heat. The experiments were later extended and it
was found that practically any solution could be used, those giving the
best results which leave the most highly colored ash. The solutions
may be dried sufficiently before exposure to the light to have the negative
placed directly against the plate, but if dried too hard, no change is effected
by the light. The action seems to be due to the migration of the mole-
cules from the shaded portions of the plate to those exposed to the light.
The best results are obtained by the use of such varnishes as are used as
menstrua in porcelain painting. Forty-five different elements were
tested and found to be susceptible to this action of light. It is thought
that practical application of this phenomenon can be made in the trans-
ference of photographs and other designs to porcelain and glass.

That manganese is perhaps the best excitant in luminous paint has

622 REVIEWS.

long been known. It now appears from the investigations of Karl {CompU
rendu, 144, 841) to play a similar part as regards triboluminescence. If
zinc sulphide is heated in an electric furnace with one-sixth its weight of
manganese nitrate to 1200°, the resulting crystalline mass, after powder-
ing and washing, shows a remarkably strong triboluminescence, visibk
even in the daylight. The mass, however, does not exhibit the phenom-
enon of phosphorescence at all. The manganese nitrate can be re-
placed by other manganese salts, and even by the oxides of tin, silicon,
zirconium, titanium, etc., and triboluminescence ensues, but it does
not seem possible to replace the zinc sulphide by even zinc oxide.

Quite an extensive paper by von Hasslinger on the nature of metalSc
and electrolytic conductivity has appeared in the Monaishefte (28, 173).
Considering the criteria by which metallic elements are distinguished
from the non-metallic, he notes that increase in metallic properties ac-
companies increase in atomic weight, and it also accompanies an increase
in temperature. Thus, sulphur becomes almost black and far less trans-
parent at high temperatures, and those forms of carbon which are maA
like metals are formed at high temperatures. It may be assumed as
probable that all the elements could be brought to the same degree of
metallic character by properly choosing the temperature for eadh. At
absolute zero, on the other hand, all substances would become non-metallic,
and non-conductors of electricity. The distinction between metallic
and electrolytic conductors is rendered more difficult from the fact that
decomposition products cannot always be detected in the latter case,
nor can the conductors be distinguished by their temperature coefficients.
If, however, a metal is in contact with a substance in which its ions can
exist, it will show a solution-tension which is recognizable by the de-
velopment of an electromotive force. Conversely, in such a case the
presence of an electromotive force indicates that the conductivity is
electrolytic, while no electromotive force is developed if the conductivity
is metallic. Prom this standpoint we must conclude that iodine, and
even sulphur are electrolytic conductors. Some substances, as for ex-
ample, silver sulphide, are electrolytic conductors at ordinary tempeia-
tures, and metallic conductors at lower temperatures, and since no abrupt
change from one form of conductor to the other can be observed, it foltows
that both kinds of conductivity may exist side by side. This transition
from one form of conductivity to the other is exhibited in carbon, where
the conductivity increases with the temperature up to a certain point,
and then on further heating decreases. Developing the subject further,
the author concludes that there is a practical similarity in the two kinds
of conductivity, the number of ions in metallic conduction being very
great, and their motion very rapid.

In order to determine the degree of dissociation of fused electrolytes,
Amdt (Ber., 40, 2937, 3612) has used fused boric oxide as a solvent.
The conductivity of this at 900° is very small (Jb= 0.000,021). When
sodium metaphosphate is dissolved in this solvent, the equivalent con-
ductivity is found to decrease with decreasing concentration. Taking
into account the increasing viscosity with decreasing concentration, it
appears that the equivalent conductivity is independent of the concen-
tration. This is most simply explained by assuming that the fused
sodium metaphosphate is completely dissociated, so that there is no

REVIEWS. 623

turther dissociation on further dilution with boric oxide. Amdt be-
lieves the same to be the case with all fused electrolytes which are made
up of univalent ions.

A clear and comprehensive review of the whole subject of non-aqueous
solutions is given by Carrara in the Gazzeiia (37, i, 525), which is fully
abstracted in the Chem. ZetUralbl. (1907, 2, 1576). It is the conclusion
of the author that between aqueous and non-aqueous solutions a great
similarity exists, but in non-aqueous solutions the phenomena are at-
tended by many and most diverse compHcations. In the meantime the
accumulation of data regarding these solutions must be carried on before
generalizations are possible.

Beckmann and his pupils continue (Z. anorg. Chem., 51, 236; 55, 371)
their work on the accumulation of these valuable data. Using quinoline
as a solvent the boiling point method gives normal molecular weights for
the halide compounds of zinc and cadmium, while the values for cuprous
chloride show increasing association toward CujCl, with increasing
concentration, but indicate the simple formula at infinite dilution.
Cobalt and nickel chloride and bromide show little tendency toward
association even in concentrated solutions, and their compounds with
quinoline give normal molecular weights. Phosgene, giving double
molecules for acetic and benzoic acids, belongs to the class of weakly
dissociating solvents. In it iodine (I,), iodine trichloride, arsenic and
antimony trichlorides, antimony pentachloride and SaCl, give normal
molecules. The S3CI3 molecule is also normal in a solution of ethyl chloride
and in liquid sulphur dioxide. In all three of these solvents sulphur
dichloride shows abnormality. In the first two, its molecular weight is
respectively 147 and 130, while that of SCI, would be 103. In liquid
sulphur dioxide the molecular weight of sulphur dichloride is 226, greater
than that required for the molecule S2CI4 (206). In this last solvent
potassium iodide was fotmd to have the doubled formula, as previously
determined by Walden.

Group L — A very full investigation of the oxides of the alkali metals
has been given by Rengade in Compt rend, (143, 592, 1152; 144, 753, 920;
I4S» 236), and summarized with additions in Ann, chim. phys. ((8), 11, 348).
When the metals of the alkalies are heated in the air they bum to higher
oxides, but if insufficient oxygen is furnished to bum the metal com-
pletely to the normal oxide, this seems to dissolve in the excess of metal.
By heating in a vacutun this excess may be distilled off, leaving the normal
oxide in a crystalline condition, often, as in the case of mbidium oxide,
in comparatively large octahedra. Sodium oxide, Na^O, is white and
hardly changes its color on heating; potassium oxide, white when cold,
becomes clear yellow at 200°; rubidimn oxide is pale yellow when cold
and golden yellow when heated, while caesium oxide is orange-yellow cold
and darkens on heating to carmine-red, purple-red, and at 150° black.
The oxides are somewhat volatile on heating and above 400° melt and
decompose into the metal and the dioxide. They are similarly decom-
posed in liquid ammonia, the soditun oxide least readily, and the com-
pound of the metal with ammonia reacts in turn with the dioxide form-
ing the amide and the hydroxide. By hydrogen the oxides are con-
verted into an equimolecular mixture of hydroxide and hydride, and on
heating to 300^ in a vacuum the latter is decomposed. The halogen

624 REVIEWS.

elements react with the oxides only when warmed. The reaction then
becomes violent. The same is true of many other reagents such as sul-
phur and sulphur dioxide, while hydrogen sulphide reacts violently in
the cold. Boron and carbon react only above 400°. Dry carbon dioxide
is absorbed at about 300°. Caesium and rubidium oxides absorb the
vapor of their metals at ordinary temperature in a vacuum, but the metal
distils off on heating to 60° or 80°. Potassium oxide absorbs potassium
vapor less readily and sodium is not absorbed by its oxide. On warming
in the air, higher oxides are formed, giving in the case of rubidium and
caesium the dioxides, trioxides and tetroxides, all of which were prepared
and are described. As regards the heat of formation, they vary from
82.4 cal. for Rb20 to 97.7 cal. for NajO. The heat of formation thus
does not increase regularly with the molecular weight. The work of
Rengade fills an important gap in the chemistry of the alkaU metals.
With this should be mentioned the work of de Forcrand on lithium oxide
(Compt. rend,, 144, 1321, 1402). This oxide was prepared by heating the
pure hydroxide in a current of hydrogen at 780°. It can also be prepared
by using lithium carbonate in the place of the hydroxide. At this tem-
perature lithium oxide has a very low vapor pressure, but at higher tem-
peratures it volatilizes appreciably.

There are numerous references in chemical literature to the formation
of a copper peroxide by different methods. These and other methods
have been investigated by Moser (Z. anorg. Chem., 54, 121) with the
result that the only instance in which the peroxide is formed is when a
30 per cent, hydrogen peroxide solution is added to a fairly concentrated
(2N) copper sulphate solution. There is an immediate precipitation of a
yellowish green copper peroxide, which seems to have the formula
CUO2.H2O. To avoid the presence of the acid formed in the reaction,
Moser found it best to use instead of copper sulphate a suspension of
finely divided copper hydroxide. In this case the peroxide is brown,
seems to be crystalline, and may be washed clean by ice water. The
product is decomposed by boiling water, and more violently by alkalies,
and dissolves easily in acids with decomposition. It also breaks up when
in a moist condition, but more slowly when dry% Its rapid decomposi-
tion by alkalies explains why Moser could not obtain it by oxidation of
copper in alkaline solution or suspension with chlorine, and also why it
cannot be prepared by the action of sodium peroxide. However, MuUer
in reviewing Moser's work {Ibtd.y 417) finds that if a copper solution is
treated with very strong (13N) sodium hydroxide solution and allowed
to stand for several months, a small amount of copper goes into solution,
and that if chlorine then be led into the solution the peroxide is formed,
but begins to decompose as soon as the current of chlorine is stopped,
the original blue color of the solution being restored. The peroxide is
also formed when a concentrated alkaline solution of chlorine acts on
metallic copper. The method of Moser, however, seems to be the only
way to prepare the peroxide, which must be considered as a very un-
stable compound. Mtiller also calls attention to the fact (Z. Elekiro-
chem.y 13, 25) that on the electrolysis of a very strongly alkaline (12-14.V)
solution of copper hydroxide there is formed at the anode a dirty yellow
copper peroxide which seems to have the formula CujOj. This may be

REVIEWS. 625

also formed on a copper anode as an orange-red or yellow coating by the
electrolysis of concentrated sodium hydroxide.

Group II . — According to Glassmann {Ber., 40, 3059) the only com-
pound in which the bivalence of glucinum is above question is the acetyl-
acetonate. Glassmann has now prepared glucinum picrate by the
neutralization of pi'cric acid in aqueous solution with glucinum carbonate.
The normal salt is formed which is soluble in numerous organic solvents.
The determination of its molecular weight by the cryoscopic method in
acetophenone confirms the bivalence of glucinum, which only among
French chemists seems to be seriously doubted. By the action of water,
glucinum picrate is converted into a basic salt.

Lohnstein (Z. Elektrochem. , 13, 613) has described a passive state of
metallic magnesium. The metal, which is rapidly soluble in dilute acetic
acid, as in most acid and even neutral solutions, if immersed in acetic
add to which a sufficient quantity of potassium bichromate has been
added, is not attacked at all, but seems to be in a 'passive' condition.
Solution and evolution of gas begin immediately if the metal is made the
anode of a cell in which the electrolyte is an acetic acid solution of the
bichromate. The stronger the acetic acid, the more potassium bichro-
mate is needed to induce the passive state, although small quantities
reduce the action of the acid on the metal. The passive state disappears
with the addition of chlorides and sulphates to the solution, and the metal
dissolves in proportion to the amount of these salts which has been added.
Lohnstein thinks that these phenomena are due to catalytic processes.

In a paper before the British Association (Chem. News, 96, 100) at
the Leicester meeting, the properties of calcium, especially in its rela-
tions to other metals, were discussed by Pratt. Owing to its large atomic
voliune, calcium has, even in small quantities, a marked influence upon
the physical properties of other metals. The chemical activity of those
metals which are easily attacked by reagents is much increased by alloy-
ing with calcium, and in many cases the alloy is more active than either
of its components. The action of calcium upon metals of large atomic
volume is greater than upon those of lesser volume. This may be a
general principle, and not merely applicable to calcium alloys. Calcium
alloys do not give promise of much industrial application, except as far as
possibly small quantities of the metal may be used for purposes of harden-
ing, but calcium bids fair to have an extended industrial value in the
metallurgical purification of other metals.

Two compounds containing three metals, NaKHgj and NaCdHg, have
been prepared by Janecke (Z. physik. Chem.j 57, 507), the first of the
kind. These compounds were discovered by a study of the melting and
solidification points of different mixtures of the metals. The compounds
have higher melting points than any of the binary alloys which come near
them in composition, the former melting at 188° and the latter at 325°.
Janecke has also (Ibid., 58, 245) confirmed the work of Kumakow on the
compounds of potassium with mercury, and finds the melting point of
KHg to be 178° and that of KHgj 279°. The three compounds whose
formulas were not definitely determined by Kumakow, Janecke finds to
be YijAg^, melting point 204°; KjHgg, 173°; and KHgg, 70°. Kumakow
(Z. anorg, Chem.y 52, 416) has extended his observations to the amalgams

626

imviEWS.

of rubidium and caesium, and finds here also as characteristic the mer-
curides corresponding to NaHg, and KHg,. In addition to CsHg, (melt-
ing point 208.2°) CsHg4 and CsHg, were found, with melting pomts
163.5® and 157.7°. As far as examined the rubidium curve resembled
that of caesium. Kurnakow calls attention to the fact that the formulas
M'Rj, M'R^ and M'R^ seem to be characteristic for m'ercurides and cad-
mides.

Although appearing in This Journal (29, 844) the work of G. McP.
Smith on ammonium amalgam should not be omitted from this review.
After examining the three views of the constitution of this much-studied
substance; that of Berzelius, supported by LeBlanc, that it is a com-
pound of mercury with the metalHc radical NH^; that of Moissan that
it is an ammonia compoimd of mercury and hydrogen; and that of Rich
and Travers that it consists of free ammonium NH^, dissolved in mercury,
Smith shows conclusively that it must be considered exactly anatogous
to the amalgams of the other alkali metals, the radical NH4 acting as an
alkali metal. The work of Coehn, indeed, where copper, cobalt and zinc
salts were precipitated by ammonium amalgam, would be sufficient to
establish this view, were there not a possibility that this precipitation
might be due to nascent hydrogen, formed in the decomposition of the
amalgam. Barium and potassium are not precipitated by nascent hy-
drogen, but Smith, on treating ammonium amalgam with barium and
potassium salts, effected an exchange between the ammonium of the
amalgam and the barium and potassium ions. The amalgam is thus a
solution of a very unstable compound, (NHJHg»„ in mercury and is
properly called ammonium amalgam. At about zero it begins to break
up into mercury, ammonia and hydrogen, and the entangling of these
gases in the mass causes the familiar frothing, which is not a property
of the compound but a phenomenon of its decomposition.

Early in the year there was published a posthumous paper {CompU rend.
144, 593) by Moissan regarding a property of platinum amalgam. When the
amalgam is shaken with water there is formed a semi-soUd, buttery mass,
which exceeds the original volume of the amalgam several times. This
emulsion, for such it seems to be, is stable at 100° and at — 80°. Plati-
num amalgam forms' a similar emulsion with sulphuric acid, ammonia,
salt solution, glycerol, acetone and many other organic liquids, but not
with benzene. The simplest method of preparing this emulsion is to
shake 2 cc. of mercury with 12 cc. of water to which a few drops of a
10 per cent, solution of chlorplatinic acid have been added. Lebeau
[Ibvi,, 843) gives further particulars received from Moissan regarding
the amalgam. The amount of platinum necessary to produce the emul-
sion phenomena is very small. It is quite noticeable when a 0.038 per
cent, amalgam is used, and the best results are obtained when about one-
half of I per cent, platinum is present. The amalgams of the other
metals of the platinum group show no tendency to form similar emulsions
The presence of amalgams of other metals, such as zinc, tin, lead or cal-
cium, destroys the emulsifying power. The emulsion with a 5 per cent
gelatin solution shows under the microscope a structure similar to that
of a soap foam, in which the air is replaced by the liquid. It was noticed
that in forming an emulsion with platinum amalgam and ether, if the
tube in which the amalgam was shaken was closed by the finger, no emul-

REVIEWS. 627

son was formed, but if a clean, dry cork stopper, or better a rubber stop-
per was used, the emulsion was readily formed.

Group III. — No inconsiderable amount of work has been done upon the
rare earths, but little of this work calls for notice here. Barbieri (Atti.
accad. Lined Roma (5), 16, i, 399) has studied the properties of several
of the rare earths as catalytic agents, using as tests the reaction between
nitric acid and oxalic acid, and that between potassium permanganate
and oxalic add. In the former reaction eerie salts act much like those of
manganese as catalyzers, lying in this respect between manganese and
iron. Incidentally it was found that cobalt salts lie between those of
iron and nickeL Salts of lanthanum, praseodymium, neodymium,
and yttrium have no influence upon the reaction. By measuring the
time of reduction of potassium permanganate in sulphuric acid solution
of oxalic acid, the order of catalytic action was found to be manganese,
cerium, cobalt, praseodjonium, neodymium, lanthanum and nickel.
Here again cerium lies close to manganese, which reminds the author of
the observation of Mendel^eff that in the rare earths one seems to see
analogues of the members of the iron group, especially as cerium in many
respects resembles manganese. This seems a rather remarkable predic-
tion in the light of the chemical knowledge at the time it was written.
By the fractional crystallization of ytterbium nitrate from nitric acid,
Urbain {CompU rend., 144, 759) has succeeded in decomposing it into neo-
ytterbium with atomic weight close to 170, and a small quantity of a new
earth, for which Urbain proposes the name lutecium, Lu, derived from the
old name of Paris. The atomic weight of lutecium is not much above 174.

Electric furnace products. Du Jassonneix has continued his researches
upon the borides, adding quite a number to those already prepared (Compt.
rend., 143,897, 1149; 145, 121, 240; -Ber., 40, 3193). In his work with chro-
mium he finds that chromium oxide can be reduced by boron only in the elec-
tric furnace, and that while two definite compoimds exist, CrgB, and CrB,
they can be isolated only when one starts out with an almost homogeneous
melt of nearly the desired composition. Boiling adds attack these com-
pounds with ease, and the second one is acted on in the cold. By using
the thermite process, Wedekind (Ber., 40, 297) has also prepared a boride
of chromium which has a composition near CrB, but has a lower specific
gravity and deddedly greater resistance to acids than the boride pre-
pared by du Jassonneix. The oxides of manganese are readily reduced
by boron in the electric furnace, the products being MnB, and MnB.
The latter only is magnetic, contrary to the opinion of Wedekind, and is
easily attacked by adds. It has been supposed that the green flame
with which the evolved hydrogen bums when borides are attacked by
adds, is due to the presence of an unisolated hydrogen boride, but du
Jassonneix considers that it is merely due to the presence of traces of boric
add, or in cases to boron chloride. In addition to the iron boride, FeB,
prepared by Moissan, du Jassonneix has prepared both FcjB and FeB,;
they are formed by the direct tmion of reduced iron and boron, the former
dther in the electric furnace or in a gas furnace, the latter in the electric
furnace only. FeB, is very resistant. The only borides of cobalt and
nickel that could h^ prepared are CojB, NijB, CoBj and NiBj. These
can be formed by heating a mixture of the elements in a current of hy-
drogen at 1 100-1200®. Wedekind's researches (Ibid., 40, 1259) were

628 REVIEWS.

chiefly with reference to the magnetic properties of the borides. He
found the pulverulent manganese boride, MnB, to be half as strongly
magnetic as powdered iron, while in compact form it was about one-
fourth as magnetic as iron. Manganese antimonide, MnSb, is more
magnetic than the boride, and the phosphide, Mn^Pj, is also magnetic.
The boride of manganese is recommended by Hoffmann (Z. angew, Chem,,
19, 2133) as superior to the boride of iron for the preparation of boron
sulphide. The manganese boride is heated in a current of hydrogen
sulphide to about the melting point of antimony. The boron sulphide,
when heated in hydrogen sulphide, fuses and on cooling gives a vitreous
modification, different from either the crystalline or amorphous. This
vitreous modification is more stable in the air than the others but like
them is decomposed very rapidly by water into boric acid and hydrogen
sulphide.

In the preparation of zirconium carbide, Moissan used the pure oxide
and sugar carbon, and found the reduction difficult with a current of
1000 amperes. Wedekind finds (Chem.-Ztg,, 31, 654) that by using the
natural oxide and pure coal, zirconium carbide is readily formed by pro-
longed heating at 600 amperes. The fused or sintered mass is very- re-
sistant to water, air and hydrochloric acid, but not to concentrated nitric
or sulphuric acid. It thus fully resembles Moissan's carbide. This
zirconium carbide, ZrC, is an excellent conductor of electricity and Wede-
Idnd suggests its use as electrodes. On heating in nitrogen, the nitride
is formed, but no cyanide.

In order to determine whether it were possible to prepare a silidde of
copper, richer in silicon than Cu4Si, Vigouroux {CompU rend, , 144, 917) heated
a mixture of copper and excess of silicon in the presence of lead, bismuth
and antimony, in a current of hydrogen for three hours at 1200°. The
copper silicide formed distributed itself differently in the different metals,
but in every case the limit was reached with 10 per. cent, of silicon, corre-
sponding to CUjSi. Another method of investigation of these siliddes
was that of Rudolfi (Z. anorg, Chem., 53, 216), who fused together copper
and commercial silicon and studied the product as an alloy. Accoimt was
taken of the iron present in the silicon as Fe^Si, and in preparing the
richer silicon alloys, an alloy with low silicon content was used, in the
place of pure copper. The quantities were so chosen as to give the same
volume (5 cc.) of alloy in each case, and the temperatures used were the
melting points of antimony, gold and nickel (630.6°, 1064°, 1451®).
Two distinct compounds were found, CujSi and CUiaSi4. Up to 5 per
cent, silicon the alloys are ductile, but the higher the content of silicon,
the more frequently is annealing necessary. Under 5 per cent., the
alloys are about as hard as copper, but from 5-10 per cent, silicon, they
increase in hardness very rapidly. Above this the hardness increases
very slowly until 60 per cent, is reached ; the alloys with above this amount
of silicon are of approximately the same hardness as pure silicon. The
hardness is not appreciably increased by chilling. The red color of
copper is very materially, lightened by even traces of silicon, the i per
cent, alloy being brass yellow, while the 6-10 per cent, alloys are silver
white. With more silicon the steel-gray color of silicon is gradually
approached.

MoSij and WSij have been prepared by Defacqz (CompU rend., 144, 84S,

REVIEWS. 629

1424) by fusing copper silicide with metallic molybdenum or tungsten in
the electric furnace. These silicides are very stable when heated in the air
and very resistant to all acids except to the hydrofluoric-nitric acid mix-
ture, and also to fused acid potassium sulphate, but they are easily at-
tacked by hot chlorine and fused alkalies. The WSij may also be pre-
pared by the reduction of a mixture of silica and tungstic acid with alumi-
num. Honigschmid has also prepared {Monatsh,, 28, 10 17) MoSij and
WSi, by this method, as well as TaSij, which resembles the others except
that it is somewhat soluble in hydrofluoric acid. Quite similar but less
resistant is manganese silicide, MUgSij, prepared by Gin by the reduction of
rhodonite in the electric furnace. Lebeau, however (Compt rend,, 143,
1229; 144, 85) thinks Gin's product is not pure MugSij but an impure MnjSi.
Wedekind calls attention to the fact that the silicide of manganese is
never magnetic (Ber. physik. Ges.y 4, 412) while manganese forms magnetic
compounds with most of the other moderately negative elements. Thus
manganese carbide when prepared in the electric furnace is magnetic
and the same is true of the nitride when it has been very highly heated.
At a high temperature the unmagnetic MnAs changes into the magnetic
Mn^s; some of the numerous manganese phosphides are magnetic,
others not; the bismuthide (MnBi?) is strongly magnetic although bis-
muth is typically diamagnetic. Wedekind considers that these facts,
together with the fact that numerous compounds of chromium, cobalt
and nickel, as well as those of iron are magnetic, prove that magnetism
is not merely an atomic property, but also a property of molecules. The
only other new silicides prepared during the year are PtSi, made inde-
pendently by Lebeau (Compt rend., 145, 241) and Vigouroux (Ibid., 376)
by fusion of the constituents in the electric furnace, and the double silicide,
CUjPtSi, prepared by the latter by fusing platinum with copper silicide.

The preparation of nitrides possesses an industrial importance from
the fact that they yield ammonia on hydrolysis, and hence numerous
patents have been taken out along this line. It has long been noticed
that nitrogen is much more rapidly absorbed by calcium carbide when
in the presence of calcium chloride, and this idea is covered by patents.
Going out from the fact that calcium chloride is hygroscopic and its
presence may give rise to the formation of acetylene and thus occasion
dangerous explosions, Carlson proposes (Chem.-Ztg.y 30, 1261) to re-
place the chloride by fluorspar, which he claims gives equally good re-
sults. Bredig has taken up the investigation of this catalytic action of
calcium chloride (Z. Elektrochem, y 13, 69, 605), and finds that at the
temperature of 800° the absorption of nitrogen is very much increased
by the presence of 10 per cent, of the chloride, and that the other chlorides
of alkaUes and alkaline earths also have an accelerating action, but much
less marked than that of calcium chloride. While the fluorides, oxides,
phosphates and sulphates have some accelerating action, it is also far
less than that of the chlorides. From the fact that the free metals,
calcium, magnesium and sodium, have little catalytic action, Bredig
concludes that the h3^othesis that the action is primarily an absorption
of nitrogen by the free metal, cannot be true. The fact that the absorp-
tion is so little increased by the presence of the easily fusible chlorides,
such as those of lithium and potassium, seems to show that the increased
action is not due merely to a lowering of the melting point. It is un-

630 REVIEWS.

questionably dependent upon the specific nature of the added substance.
The results of Bredig are in the main confirmed by those of Foerster and
Jacoby (Ibid., loi), except that they find that calcium fluoride has a
much more marked action than was found by Bredig, but this action does
not become manifest to an appreciable extent until the temperature of
900° is reached. They recommend the use of fluorspar, but a somewhat
higher temperature is necessary than when the chloride is used- Fischer
(Ber., 40, mo) suggests the use of the calcium carbide-chloride mixture
in the preparation of argon from the atmosphere, oxygen being absorbed
with the formation of oxide and carbon, the nitrogen with the formation
of calcium cyanamide and carbon. By circulating air over the mass
at 800° crude argon is rapidly obtained. Fichter (Z. anorg. Chem,, 54,
322) would make use of crude aluminum nitride for the preparation of
ammonia. The nitride is made by heating aluminum bronze and a
small quantity of carbon in the form of soot, in a current of air. To
obtain a pure aluminum nitride nitrogen and not air must be used. The
nitride is decomposed very slowly by the moisture of the air, but rapidly
by heating with an alkaline solution. Serpek has taken out German
patents (Kl. 12 .181991-2) for the preparation of aluminum for the manu-
facture of ammonia, by heating a mixture of aluminum carbide and a
small amotmt of carbon in the air. The nitride obtained readily gives
off almost all its nitrogen as ammonia with boiling water. In his second
patent Serpek recommends the addition of a very small quantity of
hydrochloric add gas or sulphur dioxide to the nitrogen, for the purpose
of accelerating its absorption by the calcium carbide.

Group IV, — Some preliminary experiments have been described by
Parsons {Proc. Roy. Soc. (A), 79, 532) on the effect of high temperature and
pressure on carbon. Coal was subjected to a pressure of 30 tons to the
square inch and a current varying from 6000 to 50,000 amperes at two
volts. In spite of cooling, the steel walls were somewhat fused. The
carbon was in every case changed to soft graphite and no trace of diamonds
was found. In the effort to prevent erosion of the walls of the cylinder
the carbon was packed in magnesia, but this was quickly converted into
magnesium carbide. Even at a pressure of 100 tons and a current of
12 volts and 100 kilowatts no diamonds could be found. When caibon
was heated with carbon dioxide, the monoxide was almost excluavely
formed. Incidentally it was found that at a pressure of 30 tons to the
square inch, liquid carbon dioxide is compressed to one-fifth of its volume.

The investigation of complex carbonates, which was begim several
years ago by Reynolds, has been extended by Wood and Jones (Proc.
Cambridge Phil. Soc, 14, 171) to many new metals. The precipitate
formed by potassium carbonate with solutions of most metals is soluble
in excess, but this solution is decomposed on boiling. If, however, potas-
sium bicarbonate is added, the solution is stable. In many cases, as with
cobalt, copper, nickel, ferrous manganese, uranium, zinc, cadmium,
bismuth, calcium, silver and magnesium salts, a crystallized double salt
is deposited on standing. The general formula of these salts of the biva-
lent metals is K2M"(C03)2.4H20, but no copper salt of this formula was
obtained, although it has been described by Reynolds. The copper
salts prepared were the anhydrous salt and the monohydrate, and the
equilibrium of both of these with their solutions was worked out. While

REVIEWS. 631

ammonium sulphide and potassium ferrocyanide precipitate copper from
the solutions of the double carbonate, and potassium cyanide decolorizes
the solution, potassium iodide has no effect upon it, and from its elec-
trolytic behavior it appears that it is dissociated into potassium cations
and Cu(C03)2 anions, and the latter are further somewhat dissociated
into copper cations and carbonate anions. From a cobalt solution treated
with potassium carbonate the potassium cobaltocarbonate crystallizes
out as the normal tetrahydrate. The cobalt solution is similar to that of
copper but more stable. On heating it is blue but on cooling becomes a
deep red-violet as at first. The CoCCOg), ion thus seems to be red, which
is perhaps noteworthy, since some of the blue cobalt solutions are attrib-
uted to the presence of the C0CI4 ions, which also contain quadrivalent
cobalt.

For the investigation of fused silicates there are decided disadvantages
connected with the use of platinum vessels, but Tammann has suggested
the use of test tubes of carbon, rendered non-absorbent by an especially
dense layer on the surface. These can be heated in the electric furnace
easily as high as 2100°, and permit the study of fusions of silica with
various oxides to be carried out with great facility. Using these tubes.
Stein (Z. anorg. Chem., 55, 159) has investigated the preparation of a
large number of silicates, formed by fusing the oxides together. He
also finds that pure silica, which is viscous at 1600°, becomes at 1750°
thin fluid, and sublimes. The sublimate, which forms several rings, is
tridymite above, and just below a broad ring of vitreous silica. Even
by slow cooling the fused silica could not be made to crystallize. Heat-
ing quartz revealed by discontinuous expansion the existence of a transi-
tion point at SS^^y while chalcedony showed one at 173° and flint none
between 100° and 600®.

An easy preparation of titanium tetrachloride has been described by
Ellis (Chem. News, 95, 122). Rutile is easily powdered after having
been heated to 1000° and chilled in water. The powder is then mixed
with half its weight of aluminum powder in a Hessian crucible at 500 **,
and the mixture ignited by burning magnesium. The product is broken
up when cold and heated to a red heat in a current of dry chlorine.
Titanium tetrachloride distils over and can be separated from the silicon
tetrachloride formed at the same time by fractional distillation. The
fact that some of the latter product is formed shows that free silicon
must have been formed in the reaction with the aluminum, but inas-
much as silica is with difficulty reduced by aluminum alone, the cause of
the reduction must be the great heat generated in the reduction of the
titanium oxide by the aluminum.

An important piece of work is being carried out by Rosenheim in going
over the chemistry of zirconium to see how much of the work of the past
can stand, in the advances of inorganic chemistry of recent years. It is
needless to say, that here, as practically everywhere else that this test is
applied to inorganic chemistry, nmch of the earlier work becomes null
and void. Rosenheim's second and third papers (Ber., 40, 803, 810)
deal with salts of some of the more common acids. Two strong ten-
dencies appear in zirconium salts, which have been long recognized, viz.,
the formation of basic (zirconyl) salts, and the formation of complex
add-zirconates, in which the zirconium is in the anion. The former

632 REVIEWS.

tendency is illustrated by the great difficulty of preparing normal m-
conium salts. Even from a concentrated hydrochloric acid solution the
zirconyl chloride and not the zirconium tetrachloride cr^'^stallizes out,
and the presence of hydrobromic or nitric acid does not overcome the
hydrolytic tendency. By evaporating a solution with excess of nitric
acid at 15° over phosphorus pentoxide the hydra ted normal nitrate is
obtained, but even here there appears to be considerable ground for
supposing that the compound is really a complex zirconyl-nitric acid
(nitrato-zirconic acid), H2ZrO(N03)4.4H30. The normal zirconium ace-
tate may be prepared by the action of anhydrous acetic acid on the anhy-
drous chloride. Quite a series of new compounds have been prepared by
Rosenheim by this reaction with the anhydrous zirconium chloride and
organic acids, aldehydes and esters. In general, two of the chbrine
atoms of the chloride are replaced, and if any moisture is present, these
two atoms are replaced by oxygen by hydrolysis. The compounds with
salicylic ester and aldehyde, ZrClj(O.C,H4.C02CH3)2 and ZrCyO.C^H,.
CH0)2, will serve as examples of these compounds. The zirconium
tetracetate undergoes partial hydrolysis in the presence of moisture,
forming zirconyl acetate, but in aqueous solution is rapidly and com-
pletely hydrolyzed ; indeed, this is suggested as an excellent method of
preparing a solution of colloidal zirconium hydroxide. The normal
sulphato-zirconic acid, H4Zr(S04)4, has not been prepared by Rosen-
heim, nor by Hauser, who has been studying the sulphates of this metal
chiefly from the standpoint of physical chemistry (Z. anorg. Chem., 54,
196) ; but the potassium salt of this acid has been long known (so-called
double sulphate of zirconium and potassium), and Rosenheim prepared
the analogous sodium salt. The ordinarv ' zirconyl sulphuric acid ' has the
formula HjZrOCSOJj-sHaO, or possibly H2Zr(OH)2(SOj2-2H20. In more
concentrated sulphuric acid solutions Hauser obtained an *acid sulphate'
of the formula Zr (804)2.112804. 3 11,0, which should more probably be
considered as one-fourth hydrolyzed normal sulphato-zirconic add,
H3Zr(0H) (804)3.21120. The corresponding oxalic acid compound,
HgZr^OH) (02(54)3. 7H2O, was prepared by Rosenheim, as well as the
potassium salt of the normal oxalo-zirconic acid, K4Zr(C2O4)4.5H20, and
the half hydrolyzed acid, HjZrO (0204)3.31120. The latter is the ordinan'
zirconyl oxalate. The only salt of a complex tartrate that was obtained
was the potassium salt of the half hydrolyzed tartrato-zirconate,
K2ZrO(C4H40e)2.3H20. Hauser (Ibid,, 53, 74), by heating the sulphate,
which had been previously dried at 400°, in a current of hydrogen sulphide
at a moderate red heat, obtained an oxysulphide, ZrOS. If this is ex-
posed to the air before it has become completely cold, it ignites spontane-
ously. The oxysulphide of thorium was obtained in a similar way and
was somewhat more stable. Matignon (Ann. chim. phys. (8), 10, 130)
finds that thorium dioxide is slowlv converted into thorium tetrachloride
by heating in a porcelain tube in a current of dry carbonyl chloride, but
the conversion is more satisfactorily brought about by heating in a current
of carbon tetrachloride at a temperature not quite high enough to sub-
lime the thorium chloride. If the operation is interrupted before the
oxide is converted completely into the chloride, and the product rubbed
up with absolute alcohol, the crystallized oxychloride, ThOClj, is ob-
tained, which is insoluble in alcohol but soluble in water. Pure metallic

REVIEWS. 633

thorium cannot be prepared by the action of sodium on the chloride,
as the oxide will always be present. The hydride of thorium dissociates
very easily, its dissociation pressure reaching 760 mm. below 400°. It
is, however, not immediately acted on by chlorine at ordinary tempera-
ture. Attention should also be called to a paper by Schenck and Rass-
bach (Ber.y 40, 2185) on the chemical equilibria between lead sulphide
and its oxidation products. It does not admit of brief abstraction but
is an example of the application of the study of conditions of equilibrium
to an important metallurgical problem.

Group V, — But three or four papers have appeared the past year on
the combustion of nitrogen in the electric flame, and these do not indi-
cate much progress in the study of the reaction, although it is claimed
that considerable advances have been made from a practical standpoint.
When one takes into account the enormous industrial importance and
possibilities connected with the manufacture of nitric acid and nitrates
from the air, and on the other hand the great differences in output with
the slightest variations in conditions, it will be realized that few pro-
cesses oflFer more opportimities for investigation.

An interesting addition to the compounds of nitrogen is that of mono-
chloramine, NHjCl, which has been prepared by Raschig (Chem.'Zig.,
31, 926). Starting with the well-known blue coloration produced when
aniline is oxidized in aqueous solution by hjrpochlorites and the fact that
no color is produced if the hypochlorite has been previously treated
with ammonia, Raschig found that when one molecule of ammonia is
added to one molecule of sodium hypochlorite, monochloramine and
sodium hydroxide are formed quantitatively. The solution can be dis-
tilled in a vacuum at low temperature and a pure solution of the new
compound obtained, indeed, by using concentrated solutions and a high
vacuum it was found possible to obtain it in pale yellow, oily drops,
floating on the distillate. Monochloramine is very volatile, with a very
irritating odor, resembling that of nitrogen chloride, and is very un-
stable. With potassium iodide it reacts giving a dark brown solution of
what is possibly moniodamine, but which soon decomposes into ordinary
iodide of nitrogen. The chloramine reacts with ammonia, with the
production of either nitrogen or hydrazine, according to circumstances.
Traces of iron, cobalt and other metals act as catalytic agents accelerating
the decomposition into nitrogen, especially in the presence of substances,
such as acetone, which decrease the viscosity of the solution. On the
other band, boiling the solution, especially in the presence of such sub-
stances as increase the viscosity, and of an excess of ammonia, tends to
promote the formation of hydrazine. By the use of albumen, casein and
gelatin it was found possible to obtain from 60-80 per cent, of the theoreti-
cal yield of hydrazine, and the possibility is suggested of using the method
in the commercial production of hydrazine.

By dissolving nitric anhydride in freshly distilled sulphuric anhydride,
Kctet and Karl (Compt rend., 145, 238) obtain a product which distils at
about 218*' and solidifies to a hard, white cr3rstalline mass which melts at
124°. The same compotmd is formed by mixing solutions of each anhydride
in carbon tetrachloride. The substance has the formula (S03)4N206, and
seems to be anhydride of nitric and tetrasulphuric adds. It is very
hygroscopic and dissolves in water with the regeneration of the two adds.

634 REvuews.

It has a powerful action on most organic substances, frequently both
nitrating and sulphonating them, while if warmed both oxidation and
carbonization may take place. No satisfactory solvent for the anhydride
has been found. Potassium pemitrate, KNO4, has been prepared by
Alvarez (Chem. News, 94, 269; from Ann. chim, anal. appL, 11, 401) by
the action of sodium peroxide upon an alcoholic solution of potassium
nitrate at a low temperature. On evaporating the solution the pemitrate
is obtained in good crystals, which are neutral in reaction and give charac-
teristic, but very unstable precipitates with solutions of metallic salts.
The pemitrates of the alkaline earths are white crjrstalline precipitates.
In a similar manner Alvarez has prepared perphosphates, perarsenates
and pertungstates, NaPO^, NaAs04 and NaWO^. The pemitrate is
recommended as a powerful oxidizing agent for combustions, and it is
suggested that the perphosphates can perhaps be used in mecUdnc. An
interesting series of double compounds with nitrogen sulphide has been
prepared by Davis (J. Chem. Soc, 89, 1575) by direct adcUtion in chloio-
form solution. SnCl4.2N4S4, SbCl5.N4S4 and M0CI5.N4S4 were thus pre-
pared. Tungsten hexachloride and titanium tetrachloride were both
first reduced and then gave the compounds WCI4.N4S4 and 2TiCl,.N4S4,
No compoimds could be obtained with ferric chloride or the trichlorides
of antimony and arsenic. All of the addition products were very un-
stable in moist air.

A useful lecture experiment to show the conversion of yellow into red
phosphoms and at the same time illustrate the action of a catal3rtic agent
is described by 2^cchini (Gaz. chim. ital., 37, i, 422). Into a glass tube
30 cm. long and 7 or 8 mm. diameter and closed at one end, is put suflS-
cient dry yellow phosphoms to fill the tube one-third full when melted
The tube is then heated to about 180° in a concentrated sulphuric add
bath. A fragment of iodine is then dropped onto the surface of the fused
phosphoms, and the conversion of the yellow phosphoms into the red
proceeds rapidly. According to Wolter (Cfcew.-Z/g., 31, 640) when
phosphoms ' sesquisulphide,' P4SJ, is shaken with a solution of iodine
in carbon bisulphide, the color of the iodine disappears, and becomes
golden yellow, and on cooling or on the addition of a little benzene or
petroleum ether, characteristic orange leaflets of the di-iodide, P4S,I,,
are formed. Since none of the other sulphides of phosphoms give charac-
teristic compounds with iodine, this reaction can serve to detect the pres-
ence of the * sesquisulphide * in matches and other easily inflammable
masses.

A number of salts of sulphato-arsenious adds have been prepared by
Kuhl (Arch, der Pharm., 245, 377), by heating together arsenious oxide
and a sulphate in concentrated sulphuric add solution, and evaporating
off the acid. These salts are all somewhat basic, if one may use that
term toward the arsenious add, but should perhaps be better called
oxy-salts. Such are K4AsjO(S04)4, CaAsjO(S04)5 and PbAsjOjCSOJr
Somewhat similar salts of antimony (Z. anorg. Chem., 54, 256) are derived
from a meta-sulphato-antimonious acid, as AgSb(S04)2 and Ca(Sb(S04),)r
6H3O. Corresponding salts of tin, prepared by Wdnland and KfiU
(Ibtd.f 244) are derived generally from the meta-sulphato-stannic add,
as K3Sn(S04)3 and CaSn(S04)3.3H20. Some are, however, derived fiom
the ortho-add, as ThSn(S04)4.2H20 and CeHSn(S04)4. The titanium

REVIEWS. 635

salts (Ibid,, 253) are derived generally from meta-sulphato-titanic acid, as
CaTiCSO^)^. With molybdic add also double sulphates are formed
{Ibid,, 259), one type being a pyro-sulphato-molybdate, K2Mo304(S04)8«
6H,0, and another having only one oxygen of a pyromolybdate replaced
by the sulphate group, as K3Mo30^(S04).6H20. All of these compounds
are looked upon by Weinland and Kiihl as being arsenites, antimonites,
stannates, titanates and molybdates, in which the oxygen atoms are
more or less completely replaced by sulphate groups.

In quite an extended study of vanadiiun, Rutter (Ibid., 52, 368) has
prepared salts of bivalent vanadium by the electrolytic reduction of
vanadic acid in the presence of sulphur dioxide, using a tile diaphragm
and mercury cathode with low temperature. While it was not possible
to prepare crystals of vanadous sulphate, crystals of the double ammonium
sulphate were easily obtained, (NHJ3V(S04)2.6H20. Vanadous sulphate
oxidizes so readily that in the absence of oxygen a dilute solution de-
composes with evolution of hydrogen, while if the solution is concentrated
hydrogen sulphide is given oflF. When solutions of bivalent vanadium
are mixed with vanadium solutions of other valences, equilibria are at
once attained. Bivalent and quadrivalent give trivalent vanadium
as a green solution, while vanadic acid with bivalent vanadium gives
first a quadrivalent and then a trivalent solution. A study of the accelera-
ting action of vanadium pentoxide on oxidation processes has been made
by Naumann and his pupils (Jour. pr. Chem. [2], 75, 146). Small traces
of the pentoxide accelemte very markedly the oxidation of sugar to oxalic
acid by nitric acid, and the reaction can be easily studied by precipitating
the oxalic add formed and titrating with permanganate solution. If
the temperature is allowed to reach 70°, the oxalic add is further oxidized
to carbon dioxide. When a mixture of air and alcohol vapor is led over a
layer of asbestos which has been saturated with vanadium pentoxide,
the alcohol is oxidized to aldehyde with some acetic add, while the vana-
dium asbestos becomes red-hot. The oxidation of stannous salts to
stannic by nitric acid or by potassium chlorate and hydrochloric acid is
greatly accelerated by the presence of vanadic acid, but on the other
hand, the oxidation of ferrous and manganous salts seems scarcely aflFected.
The accelerating action 'of vanadium pentoxide is attributed by Naumann
to the readiness with which it gives up one atom of oxygen, becoming the
tetroxide, which in turn is quickly re-oxidized by whatever oxidizing
agent happens to be present. Thus if to a slightly warmed, add solu-
tion of the pentoxide is added a sugar solution, the blue color of the
tetroxide at once appears, only to give place to the yellow color of the
pentoxide as soon as a few drops of nitric add are added.

The interest connected with the use of tantalum in the incandescent
Kght has naturally begun to stir up chemists to the further investigation
of this puzzling element and its neighbor, columbium. While no in-
considerable number of chemists have worked with these metals, it is
probably true that less is known of their chemistry than of any other
two elements, if we except those that show radioactivity. Werner von
Bolton (Z. Elektrochem., 13, 145) has, probably for the first time, pre-
pared pure columbium, previous workers apparently having had in their
hands only a lower oxide or a carbide. Bolton mixes columbium pentoxide
with paraffin and makes it into fibers that are reduced to the tetroxide

636 REVIEWS.

by heating in charcoal powder. The tetroxide as thus obtained is a
conductor and is reduced to the metal by heating highly with the alterna-
ting current in a vacuum. If the direct current is used, only a lower
oxide is obtained, which is, however, a good conductor of electricity.
Columbium was also prepared by the Goldschmidt reaction. It is a
bright light gray metal, imacted upon by adds. It is about as hard as
wrought iron, 'malleable and ductile, and can be welded at a red heat
It is so passive as an anode that it may find industrial appHcation. Al-
though it fuses only at 1950**, it 'dusts' so badly at high temperatures
that its use in incandescent lights is precluded. When heated in hy-
drogen the hydride, CbH, is formed, which is readily oxidized, while the
metal itself is only slowly acted on by oxygen at even a red heat. With
nitrogen it forms a nitride. At a red heat in chlorine the pentachloride
is produced, and it is also attacked by fused alkalies, by sulphur and
selenium. It forms no amalgam with mercury but alloys with iron in all
proportions. The chlor- and bromcolumbates and tantalates have
been studied by Weinland (Z. anorg. Chem., 54, 223) and for the most
part are formed according to the type M'jCbOClg (or CbOCl5.2M'Cl).
The chlorcolumbates of the alkalies crystalUze in octahedra, seemingty
belonging to the type of chlorplatinates in which one atom of chlorine
is replaced by one atom of oxygen. Weinland has also found {Ber.,
39, 4042) that the oxychloride of * quinquivalent chromium' prepared
by him last year, forms double salts with the alkali chlorides, having the
formula Mj'CrOClg (CrOCls.2M'Cl). These crystals also have the octa-
hedral habit, although only the caesium and rubidium salts belong to the
regular system. A similar case is that of the monoxychlorosmates dis-
covered by Wintrebert, and another the dioxychlorruthenates and dioxy-
chlorosmates. With organic bases other types of chlorcolumbates
appear, but all are compounds of the oxychloride, CbOClj, and the same
is true regarding the chlorotantalates. The bichloride of tantalum has been
prepared as a dihydrate, TaCl2.2H20, by Chabri6 {CompU rend., 144, 804)
by the reduction of the pentachloride with sodium amalgam at a red
heat, and crystallization from concentrated hydrochloric acid in a vacuum.
The green color of this chloride is seen when tantalum solutions are ^^
duced by various agents, but it very readily becomes oxidized. The
properties of both these metals have also been examined by Muthmann
{Ann., 355, 58), with essentially the same results as found by von Bolton.
A moderate red heat is required to oxidize columbium, the tetroxide,
Cb^O^, being formed, while tantalum bums at a lower temperature to the
pentoxide, TajOj.

Group VI. — In a recent paper (Z. anorg. Chem., 56, 233) Marino argues
in favor of a new class of dioxides. When manganese dioxide is treated
with sulphur dioxide, manganese dithionate is the principal product,
though there is some sulphite formed. With lead dioxide, sulphur dioxide
gives at first lead sulphite and oxygen, which latter immediately oxidizes
the sulphurous acid to sulphuric. The course of the reactions, as fol-
lowed by Marino, is:

I. PbO,+SO,=PbS03-fO

11. SO,-fO-fH,0=H,SO,

III. PbS0, + H,S04=PbS0,+H^y

ki^vi^ws. 637

With true peroxides, the reaction is:

I. BaO, + HjSOj = BaSO, + H3O3

III. BaS08+HA = BaSO,4-HaO.

In neither of these cases is any trace of dithionate formed. From the
reactions with hydrochloric acid, it may be assumed that in lead and
manganese dioxide the oxygen is united solely to the metal and that the
metal is quadrivalent, while in barium dioxide the barium is bivalent
and the atoms of oxygen are united together, as shown by the hydrogen
peroxide formation. From the dithionate reaction it is probable that

the formula of manganese dioxide is Mn^ , but from the fact that lead

dioxide forms only the sulphite, it must have some other formula. The

only two which conform to the conditions are Pb/^ | and Pb/^ II . The

former of these is the more probable since the sulphite formed from it
has the as)rmmetrical formula, as was experimentally shown by its re-
action with dimethyl sulphate. Its reaction with sulphur dioxide is

O /SO,

thenPbf^ | -f-SO,=Pb^ | +0. Lead dioxide may thus be considered

NO X)

to belong to a hitherto undescribed type of dioxides. Since the forma-

IV >j?0
tion of a dithionate is conditioned upon the group MT , and since a

dithionate is formed when sulphur dioxide reacts with the sesquioxides of iron,
cobalt and nickel, it follows that the constitution of these sesquioxides is

M = 0 M = 0 Q

not, as generally assumed, ^O or ^O , but 0 = M==m/^ . In the

M=0 M=0 ^

opinion of the reviewer the conclusions of Marino do not seem to be well
justified. Aside from the question as to whether there is any real differ-

ence between the two constitutions M/^ and M ^ | (which is extremely

^O \o

doubtful), the constitution proposed for the sesquioxides is improbable.
There is no other evidence to indicate that all of the metal in these com-
pounds is not imiformly trivalent (or [Ma]^^- Further, the fact that
MnO, gives dithionate, and PbOj primarily sulphite, does not necessitate
a diflFerence in constitution of the dioxides. The difference in solubility
of lead and manganese compounds is quite sufl5cient to account for the
breaking down of the primarily formed quadrivalent sulphites along

different Unes. The primary reaction would be Mf' -|- 2802= M^ — ^ wi^

^o \o/ '

In the case of manganese the reduction in valence of the metal is to

638 RHVIBWS.

Mn<f^ I and in the case of lead to the insoluble PI

X)— SOa
(S0,4- H20«)HjS04, as found by Marino.

The old problem of the constitution of the thiosulphates has been
attacked by Julius Meyer (Ber., 40, 135 1) from the standpoint of the
double thiosulphates, but no very definite results have been obtained.
The double thiosulphates are in general quite unstable, but a pretty
full series of the lead, silver and copper thiosulphates with the alkalies
was prepared, Cs,Sa08.PbS20j.2H30 is a salt of a type which frequently
occurs, though quite a number of other t)rpes exist, sometimes more than
one with the same two metals, as for example the rubidium-copper thio-
sulphates, of which three distinct salts were prepared. In the case of
some of the ammoniacal double thiosulphates of silver both white and
yellow salts were obtained,which Meyer thinks may possibly point to a
difference in constitution, the silver in the one case being attached to
oxygen and in the other to sulphur. The difference in color may, how-
ever, be due to other causes, as the salts have not the same formula. In the
case of the double rubidium salts, the white salt is Rb2SaOj.Ag,Sj03.NH„
and the yellow salt 3RbjS2O8.4AgaSaO8.NHg.

In reviewing the sixth group, the work of Norris, published in
This Journai, (28, 1675), on the elementary nature of tellurium shoidd not
be passed over. There seems to be little doubt but that tellurium is
rightly placed as a member of the sixth group of the periodic system,
but the most careful determinations of its atomic weight agree in placing
it above iodine. Mendel^eff and others have plausibly conjectured that
the discrepancies are due to the presence of some element, such as
Mendel6eff's dvitellurium, in our ordinary tellurium, and special effoits
have been made by several chemists to separate out such an element
In earlier work with Fay and Edgerly, Norris purified potassium bromo-
tellurate by fractional crystallization, Bratmer sublimed the tetrabromide
of tellurium, and Kothner distilled tellurium itself fractionally. Since
the boiling points of both the bromides of selenium and tellurium, and
of the elements themselves, are not far apart, it is probable that a Wgher
element of the same group would not differ greatly in its own boifing-
point or in that of its bromide. Consequently it might be difficult to
separate such an element if it were present in tellurium. On the other
hand, there is a great difference between the boiling points of the dioxides
of sulphur, selenium and tellurium, and it is probable that the diosde
of a heavier element of the sulphur group could be readily separated from
the other members by sublimation. Norris therefore purified tellurium
by the sublimation of the dioxide, but no difference in the atomic weights
of the different fractions could be detected. In order to purify tellurium
from any element of any other group, the reaction of sodiiun thiosulphate
with tellurium dioxide to form the tellurium analogue of the pentathionate,
NaaS4Te08, was utilized. This breaks up by alkalies into the tetra-
thionate and tellurium. It is hardly possible that any element of another
group could replace sulphur in the pentathionate, so the precipitated
tellurium from this compoimd may be considered free from any element
except sulphur (and possibly selenium). On further purification this
tellurium shows the same atomic weight as that purified in other waySi

REVIEWS. 639

This work of Norris's presents further strong confirmation of the ele-
mentary nature of what we now know as tellurium, and the solution of
the problem of its anomalous position in the periodic table must be sought
in some other direction. A paper has just appeared by Marckwald (Ber.,
40> 4730) on the atomic weight of telluriimi which seems to bring out a
new phase of the subject. Marckwald admits tliat the work of Norris and
others has conclusively demonstrated the absence of any other element in
their tellurium, which could affect its atomic weight, but he claims that
most of the atomic weight determinations thus far made are defective.
He has carried out a number of determinations based on heating HeTeOg
and weighing the TeO, formed, and gets fairly concordant results, witli an
average of 126.85 ± 0.02. As this is below tie accepted atomic weight of
iodine, it puts a new face on the whole matter, and the criticisms and re-
sults of the other workers in this field may prove interesting.

In the light of Werner's theory of the constitution of inorganic com-
pounds, the question of the constitution of the product formed when
chromium chloride is dissolved in pyridine and the solution precipitated
by water, is of interest. The product has the empiric formula, CrCl,
(Pyr)j. According to Werner's theory none of the chlorine atoms should
be ionizable, but the insolubility of the compound in water precludes the
possibility of determining this directly. Pfeiffer (Z. anorg. Chem., 55,
97) finds that the compound can be easily dissolved in concentrated
nitric add to a deep green solution, from which water precipitates the
original compound in crystalline form. From this he argues that the
chlorine cannot be ionizable, since if it were, some of the chlorine would
be replaced by the nitrate group. Furthermore, the compoimd, while
insoluble in water, is soluble in pyridine, chloroform, methyl alcohol
and other organic solvents, which renders it probable that it has nothing
of the character of a salt. The compound must hence be looked on as

/ CI3.
the coordinated group (Cr ). Several papers have appeared during

the year on the conditions which pertain in solutions of chromates and
bichromates, but the problem still seems far from a final solution, and
must be passed over here. Groger has continued to work upon the
double chromates {Ibid., 51, 348; 54, 185), and finds that two quite general
types of double chromates prevail. The tjrpe which is found in most of
the double alkali sulphates of the bivalent metals, M'jM'' (804)3.61130,
is also found among the chromates, but hardly with the same frequency.
It is found, for example, in the double alkaU chromates of nickel and
cobalt. On the other hand, a commoner type is M'2M"(Cr04)2.2H20.
This is foimd not only in the double chromates of cadmium, zinc and
magnesitun with potassium, but also in the potassium calcium chromate.
Other types are also found, the double chromates of lead and univalent
mercury with potassium being anhydrous. The large number of com-
plex molybdates prepared by Hall have already been described in This
Journal (29, 692). Rosenheim has confirmed the formula of the potas-
sium molybdi-octacyanide (K4Mo(CN)8.2H20) described by Chilesotti.
It appears from titration experiments with potassium permanganate
that this peculiar compound is a derivative of quinquivalent molyb-
denum. This only serves to increase the anomalous character of this

640 REVIEWS.

salt, which represents the unique case of a stable anion, solubk[ii^water,
which contains eight coordinated groups about one atom. (I^iite a
number of salts of this anion were prepared by Rosenheim, and they are
comparable in stability with the ferrocyanides, A quite complete in-
vestigation of the compounds of quadrivalent uranitun has been pub-
lished by Colani {Ann. chim, phys. [8] , is, 59). The anhydrous chloride
was prepared by the action of chlorine and chloride of sulphur on the
dioxide, the product containing other chlorides if any other oxides are
present. The tetrabromide was readily formed but the anhydrous tetra-
iodide could not be prepared. The sulphide, selenide, nitride, phosphide
and arsenide were all found to be normal in their formulas. A laige
series of phosphates and double phosphates was prepared, the latter
giving generally rather simple types.

Group VIL — During the past year considerable work has been pub-
lished from the University of Danzig by Ruff and his pupils on fluorine
and its compounds (Ber., 39, 4310; 40, 2926; Z. anorg. Chem,, 52, 256 ;Z.
angew. Chem., 20, 12 17). The pentafluoride of antimony was prepared
by the action anhydrous hydrogen fluoride upon antimony pentachloride
at zero, the temperature being gradually raised and the hydrogen chloride
and hydrogen fluoride being successively distilled off. It was not found
possible to prepare the pentafluoride by the action of hydrofluoric add
upon antimony pentoxide. With a small quantity of water the penta-
fluoride forms a dihydrate, but dissolves in more water to a stable solu-
tion, in which no precipitate is formed in the cold with either hydrogen
sulphide or potassium iodide. While dry metals have little action upon
the pentafluoride, several of the elements form addition products. Such
are SbFgl, (SbFj),! and SbFjS. At about 250® an equilibrium seems to
be attained with a formula (SbFs)!.^!, and this whether the mixture
heated contained an excess of the pentafluoride or of iodine. With
liquid ammonia a compound is formed with the formula (SbFj,(NHJ„
which according to Ruff is probably a difluohydrate of diaminodianti-
monotrifluoramide, which may be formulated HF.NH2.SbF,.NH.SbF,.
NHa.HF. With tungsten hexachloride, tungsten hexafluoride is formed,
but a better method of preparing the latter is the action of anhydrous
hydrogen fluoride upon tungsten hexachloride. Arsenic trifluoride can
also be used. Timgsten hexafluoride is the first fluoride of a metal which
is a gas at ordinary temperature. At 19.5° it is condensed to a liquid
which freezes at 2.5°. It fumes strongly in the air and is very sensitive
to moisture. The oxytetrafluoride, WOF4 was prepared, and a mixture
containing this and the dioxydifluoride, WO,Fj, but the latter could not
be obtained pure. While efforts to prepare the hexachloride of molyb-
denum have not proved successful, and the hexafluoride could not be
obtained by the action of hydrogen fluoride or any of the other fluorides,
in some cases fluorine compounds of hexavalent molybdenum were ob-
tained, but they were generally the oxytetrafluoride or the dioxydifluoride,
corresponding to the tungsten oxyfluorides mentioned above. The
molybdenum hexafluoride was finally prepared by the direct action of
fluorine upon metallic molybdenum, obtained by the Goldschmidt pro-
cess. MoFe fuses at 17® and boils at 35® and is a very reactive substance.
Another interesting compound prepared by Ruff is the add potassium
compound of lead tetrafluoride, PbF4.3KF.HF. This may also be looked

REVIEWS* 641

upon as the tripotassium salt of fluo-orthoplumbic acid, KjHPbFj. This
is readily prepared in quantity and when heated gives off a part of its
fluorine, hence it is proposed by Brauner to use it in the preparation of
fluorine. It is probably the most stable known halide compound of
quadrivalent lead. Bismuth pentafluoride was also prepared, hydro-
fluoric add being used to dissolve the product obtained by the action of
chlorine upon an alkaline suspension of bismuth. Lebeau has also been en-
gaged in work upon fluorine and has prepared (CompL rend., 144, 1042,
1347; 145, 190) selenium tetrafluoride by the direct action of fluorine upon
selenium. It boils somewhat above 100°. With an excess of fluorine
no higher fluoride could be obtained. This is, however, contrary to the
work of Prideaux, who obtained a hexafluoride, the vapor density of
which, as determined by Ramsay, corresponded to the hexafluoride.
In his later papers Lebeau seems to admit the formation of the hexa-
fluoride. The subject of the basicity of hydrofluoric add has been further
studied by Kremann (Monatsh., 28, 917) and Pellini and Pegoraro (Z.
EUklrachem., 13, 621). The former studied the conductivity of the
solutions, using as a vessel a hollow block of paraffin. On the basis of
Ostwald*s dilution law, the add has the formula HjFj. Pellini studied
the conductivity on neutralizing hydrofluoric add with alkalies, and
from this concludes that the add acts as if it were bibasic and consisted
of a combination ,of two monobasic acids, one of which is strong and the
other weak. The free add and the neutral salts thus act as binary elec-
trolytes, while the hydrogen fluorides and the neutralization phenomena
conespond to a bibasic add.

Recent work by Schwarz (Z. angew, Chem., 20, 138) on bleaching
powder confirms the hypothesis that the underlying reaction of its for-
mation is Cla-|-HOH:^HCl-|-HOCl. This explains why the presence of
water is necessary to the reaction. So long as the lime is present in ex-
cess the reaction is merely a neutralization of the adds according to the
above equation. As soon, however, as the lime has all been used up, the
excess of hydrochloric acid liberates h)rpochlorous acid from the bleaching
powder and this, with the hypochlorous acid formed in the above re-
action, oxidizes undecomposed hypochlorite to chlorate. The excess
of chloride which is always found in commercial bleaching powder may be
due, according to Schwarz, to the loss of oxygen from the hypochlorite,
or to hydrochloric acid present in the chlorine used. If the amount of
water present in making bleaching powder is too great, the same effect
is pnxluced as with an excess of chlorine, since the hjrpochlorite is very
easily hydrolyzed. The fact that the synthetic bleaching powder (prepared
by the action of chlorine monoxide on calcium oxide in the presence of mois-
ture) gives up much less chlorine when treated with carbon dioxide, is prob-
ably to be explained by considering it merely a mixture of caldum chloride
and caldum hypochlorite, while that prepared in the ordinary commercial
method is a true double salt, CaCl(OCl). Bleaching powder from
stiontia resembles very closely in preparation and properties, that from
lime. Two new salts of some little interest are hydrazine chlorate and
perchlorate, prepared by Salvadori (Gaz. chim. iial,, 37, ii, 32) by the
neutralization of a solution of hydrazine hydroxide with the free acids,
and evaporation in a vacuum. The chlorate is exceedingly hygroscopic
and has three times as great explosive force as fulminate of mercury.

642 RBVIBWS.

Alcohol is rapidly oxidized by its solution even in the cold to aldehyde
and acetic acid. Hydrazine perchlorate is, on the other hand, far more
stable and can be fused at 131° on platinum foil, and burned quietly at a
higher temperature. It is, however, exploded by a blow. It can be
boiled in solution, even in the presence of alcohol, without decomposition.
A specimen of the perchlorate was preserved for two years wichanged
over calcium chloride.

Several years ago the method of preparing anhydrous chlorides by heat-
ing the oxides in a current of chlorine and chloride of sulphur was intro-
duced. Bourion (Compt. rend., I45| 243) now finds that this reaction is
equally applicable to the preparation of bromides, and he has obtained
a number of new bromides of the metals by heating their oxides in a
current of bromine containing a small amotmt of the chloride of sulphur.
The amount of the latter reagent necessary for good results varies in
different cases.

A new suggestion regarding the place of manganese in the periodic
table has been made by Reynolds (Chem News, 96, 260). Recognizing
the slight resemblance between manganese and the other elements of the
seventh group and the fact that no other elements of higher atomic weight
have been foimd to take the vacant places in the second series of group
seven, he transfers manganese to the first period of the eighth group,
making this series, Mn, Ru, Os. The second series becomes Fe, Rh, Ir,
and the third series, Ni, Pd, Pt. One advantage of this arrangement is
that nickel precedes cobalt in position as it should with its undoubtedly
lower atomic weight. It may be questioned, however, whether thK
arrangement has actually come as near the natural arrangement as the
ordinary table. Cobalt would here stand alone as a single member of a
fourth series of group eight. It is tme that group seven with only one
series would correspond to the so-called group zero with the inert gases,
which has only one series, but as a matter of fact the halogen group seven
corresponds rather to the alkali group one, and the compounds of man-
ganese are no more unlike those of the halogens than are those of copper,
silver and gold unlike the alkalies. According to Re3molds's arrange-
ment, the iron-platinum metal group represents a transition from the
sixth group to the first, while the inert gases would be the transition
from group seven to group one. If the iron-platinum metals and the
inert gases represent the two divergent series of g^roup eight, then the
iron-platinum metals are a natural transition from the positive series of
group seven to the negative series of group one, while the inert gases form
the transition from the negative series of the seventh group to the positive
series of group one. Mention ought to be made here of the fact that
Holmes (This Journal, 29, 1277) has succeeded in preparing the long
sought for tetrachloride of manganese, by the action of dry hydrogen
chloride upon freshly precipitated manganese dioxide, suspended in
carbon tetrachloride. It is hydrolyzed in the presence of even a trace of
water and loses chlorine on heating. This work of Holmes renders
probable what has long been surmised, that the first product of the action
of hydrochloric acid on manganese dioxide is the tetrachloride, and the
reason that this has eluded so many investigators is not so much the
inherent instability of quadrivalent manganese compounds (a thing in

REVIEWS. 643

itself improbable) as the great tendency of the compounds of quad-
rivalent manganese to hydrolyze.

Group VIIL — ^That in the action of iron on water the temperature and
the surface are the two determining factors has of course long been rec-
ognized, but few have realized how extensive the influence of increasing
the surface by using finely divided iron might become. Bimie (Chem.
WeekbL, 4, 291) has studied this reaction using ferrum reductum and the
vapor of water at different temperatures. When 10-20 grams of this
iron are heated in a current of steam with an ordinary gas jet, 100 cc.
of hydrogen can be collected in a few moments. It is even possible to
obtain a rapid evolution of hydrogen by the use of 'florists' wire/ if a
suflBdently high heat is used. At a temperature of 15° with 500 grams
of ferrum reductum little hydrogen was obtained the first day, but after
that from 100-500 cc. every 24 hours. A decided evolution of hydrogen
is apparent when two kilos of florists' wire is boiled with distilled water.
The color of iron alum (and incidentally of other ferric salts) has long been
an unsolved puzzle. It has been described as colorless, as yellow, as
violet, as blue, but it has often been conjectured that the ordinary ame-
thyst color is due to manganese, since the salts of trivalent manganese
vary from amethyst to red according to concentration. Christensen
(Daftske Vid. Selsk. Fork., 19061 4, 173) has experimented with the frac-
tional crystallization of iron alum and of pure iron alum to which a trace
of manganic sulphate (manganic acetate in sulphuric add) has been added.
It was fotmd that crystals could be obtained varying from colorless to
garnet red, depending on the content of manganese. In the more deeply
colored crystals, it is safe to say mixed alum crystals are present, but
in those which are amethyst the color is possibly due to a ferric-manganic
sulphate, such as has been prepared by 6tard. While this salt is green,
its addition to a colorless iron ammonium alum gave violet crystals.
Pure iron alum is colorless but colorless iron alum is not necessarily pure,
for on fractioning such an alum a mother-liquor was obtained from which
violet crystals were obtained. Bellucd and his pupils have continued
the work (Atii accad. Lincei, Roma (5), 15, ii, 467 ; 16, i, 654) begun several
years ago on the nitrososulphides of iron (nitroprussides). By action
upon the sodium salt, NaFe4(NO)7Sg.2H30, with the hydrochloride of
hydrazine in slight excess, the hydrazine nitroprusside was obtained.
The hydroxylamine salt was similarly formed but required rather longer
heating to complete the reaction. The potassium, thallium, rubidium
and caesium salts were readily formed by treating the hydrazine salt with
the corresponding chloride, and the salts obtained in this way were all
anhydrous, while when otherwise prepared they are hydrated. Nitro-
prussides were also prepared of a number of organic bases, such as phenyl-
hydrazine, pyridine, tetra-alkyl ammoniums, phenylened^mine and semi-
carbazine. Cobalt hexamine (luteo salt) also forms a crystalline com-
pound.

An article has lately appeared by Gross (Elektrochem. Z., 14, 146)
claiming the decomposition of platinum by the alternating current.
Potassium carbonate was heated to a yellow heat in a platinum crudble
and subjected for several hours to an alternating current of 120 volts and
35 amperes, with 50 alternations per second. A small amount of potas-
sium nitrate was added to the melt. The platinum was strongly attacked

644 REVIEWS.

and in and over the melt giuphite-like crystals were formed. After
treatment with water a brown substance remained which was soluble in
water but after heating to a red heat could be dissolved only in aqua
regia. Its solution gave a dark brown precipitate with hydrogen sulphide.
On evaporation of the filtrate from the melt a red- powder was obtaiiied
which was free from platinum but on solution in hydrochloric add showed
the same properties as the brown substance. The above-mentioned
needles, on solution in aqua regia showed also similar properties. The
total amount of new substance was 15 per cent, less than the loss of the
platinum crucible and the electrodes. Gross considers that the red
powder is a hydrate and the brown material the new substance itself, ob-
tained by the decomposition of platinum. The same results were ob-
tained when potassium hydroxide or a mixture of nitric and sulphuric
acids was used as an electrolyte. Further information is to be looked for,
as the data given aipe insufficient to form a conjecture as to the meaning
of the published results, and especially as to whether the other metals
which are present in all commercial platinum may not be responsible
for the results obtained by Gross.

Gutbier has continued his work upon the halopalladites and palladates
of organic bases as well as the pallado-amines of the same bases (^er.,

39, 4134). Pyridine, picoline, quinoline and a number of primary alkyl-
amines were used and the results obtained are similar to those which have
been previously described. The halopalladites are formed by direct
union of the components in hydrochloric add solution and can generally
be recr)rstallized from hydrochloric add, but on treatment with water
they are decomposed with the formation of the palladamjnes. Gutbier
has also (Ibid., 40, 690) prepared a large number of hexachlor- and
bromruthenates of organic bases. His method is to saturate a solution
of ruthenium trichloride with the halogen and then add the solution of the
hydrochloride of the base, or to mix the trichloride with the base and then
saturate the solution with chlorine or bromine. Since earlier investiga-
tions have shown that the true trichloride or rather pentachlomithenic
add, HjRuClj, and its salts are not converted into the hexachlorides
directly by the addition of chlorine, but only after they have been con-
verted into the aquochlorides, it must have been the latter, M',Ru(H,0)Clt,
with which Gutbier worked. The add solution of the trichloride is readily
converted into the aquochloride by heating its solution with alcohol
and other organic substances, so that this change may well have occurred
in the presence of the organic substances with which Gutbier was work-
ing. The properties of Gutbier's hexahaloruthenates agree with those
of the corresponding hexahaloruthenates of the alkali metals. A very
useful piece of work has been carried on by Paal and Amberger (/Mi,

40, 1378) on the quantitative determination of osmium. When a com-
pound of osmium can be heated in a hydrogen current and reduced to the
metal, this can be very satisfactorily handled on a Gooch filter, as has
been well worked out by Wintrebert ; but where the osmium is in solution
the problem is difficult. Paal and Amberger have gone critically over
the whole ground of the quantitative precipitation of osmium, testing no
less than ten proposed methods, generally with very tmsatisfactory re-
sults. When the osmium is in the form of osmium tetroxide fosmic
add') it can be completely predpitated by alcohol on long standing,

REVIEWS. 645

though the precipitate is a hydrogel and hard to wash. Other com-
pounds of osmium, as the osmates, are also precipitated by alcohol but here
again the precipitate is difficult to free from alkalies, if they are present.
No method, which can be considered thoroughly satisfactory, is known
by which osmium can be quantitatively precipitated. The same is even
more true of ruthenium, for no method is yet known by which this metal
can be completely precipitated from solution.

WA8HIMOTON & LBB 17NIVBR8XTY,
I,BXIirOTON« Va.
Jannaryi, 1908.

ON THE NON-EQUIVALENCE OF THE FOUR VALENCES OF THE

CARBON ATOM.^

By J. U. Nbf.
Received January 8, 1908.

Three assumptions are made with reference to the carbon atom in
our present system of organic chemistry: first, that the valence of this
atom is invariably four; second, that the four valences are equivalent;
and third, that they are distributed in three dimensions and act in the
direction of the axes of a tetrahedron.

If we believe, as many at present are inclined to do, that the chemical
forces are analogous to or identical with the electrical forms of energy,
it at once becomes improbable that the four valences of the carbon atom
can be equivalent. Had Berzelius, for example, realized that the mole-
cules of hydrogen and oxygen were each composed of two atoms he would
have developed his electrochemical theory from a different standpoint;
he would have concluded, as have Clausius, Sch5nbein, and many others
in more recent times, that the forces holding the two atoms of oxygen
or hydrogen together in these molecules must be alternately positive and
negative.

Let us first analyze critically, however, the evidence which has led us,
for some fifty years past, to assume that the four valences of a carbon
atom are equivalent. In the last analysis the fact that up to the present
moment every one of the great number of monosubstitution products of
methane has been fotmd to exist in one modification only is all we have
to justify our assumption. There is bu£ one acetic acid, one nitromethane,
one anitine, one acetaldehyde, etc., etc. It is evident that we are here
drawing positive conclusions from negative evidence — always an un-
reliable and dangerous process of reasoning. How can we ever be certain
when we have a monosubstitution product before us that it is not always
one and the same hydrogen atom of marsh gas which has been replaced?

A Belgian chemist, Henry,* has spent a number of years in trying to
prove by experiment that any one of the four hydrogen atoms of methane,
a, 6, c or d, may be replaced by the carboxyl or nitro group and yet give
thus one and the same acetic add or nitromethane respectively. Such
experiments can not conceivably be decisive, for they necessitate the
unjustifiable assumption that the various atoms or radicals bound to the

* Sec This Jousnal, 26, 1 549-1 577. Read at the Chicago meeting of the American
Chemical Society.

* Bun. acad. royal de Belgigue (3), la, 644; 15, 333.

646 REVIEWS.

four different valences of a carbon atom hold their places. A study of
the optically active a-brompropionic adds, the brom-phenylacetic and
succinic acids (see below) shows for instance very decisively that such an
interchange of radicals is continually taking place at ordinary temperature.

The present situation with reference to the nature of the four valences
of the carbon atom may therefore very properly be stunmed up in the
following words; we assume their equivalence until there is proof to'the
contrary.

During the past ten years or more there has accumulated quite a mass
of evidence pointing tmquestionably to the conclusion that the four
affinity units of carbon are equivalent in pairs only, as shown by the
following expression, iZCZ7» ^ which the plus and minus signs do
not necessarily mean positively and negatively charged valences, but are
simply used to discriminate between two different kinds of affinity units.

Allow me to call your attention to certain reactions shown by the fol-
lowing optically active, space isomeric, a-substituted propionic adds.

COOH COOH

-h OH HO +

H— C + NH, or NH, + C — H

— X = [ClorBr] X —

CH, CH,

d-lactic add. Mactic add.

(i-alanine (aminopropionic acid). ^alanine,

c^-brompropionic acid, etc. /-brompropionic add.

When d- or /-brompropionic add, or its ester, is treated with ammonia
we always obtain the corresponding, optically active, or-amino add, d- or
/-alanine, or its ester. Similarly when these same brom add esters are
treated with various metallic hydroxides, the corresponding, optically
active, d- or /-lactic add derivatives are obtained.

During the past twelve years an enormous amount of evidence has ac-
cumulated which shows that the interaction of the primary and secondary
alkyl halides, RCH^X and RjCHX, with ammonia, metallic hydroxides, or
with silver salts of various adds proceeds through the following succes-
sive stages:

(a) R,CHX:2tR3C<+HX + 2NH,->R,c/ +NH«X.

(6) R,CHX:;tR,C<+H— X+M— OH-^

R,C< + MX + H— OH -> RjCHOH +MX.
(c)R,CHX -^ R,C< + H— X + Ag— A ->

R,C< + AgX + H— A-^R,CHA + AgX.
Now the two space isomeric a-brompropionic adds, named above, may

Br. XH,
be regarded as a-carboxylated ethyl bromides, /C<^ and

H/ N:OOH
Hv /CH,

yC(^ ; they must therefore be entirely analogous in their re-

Br/ N:OOH

actions to the secondary alkyl bromides. It is furthermore evident at a
glance that if the four valences of carbon were equivalent both i-

REVIEWS. 647

and /-brompropionic add must give by loss of hydrogen bromide,

>yCH,
^ , through dissociation, one and the same ethylidene car-

^COOH
boxylic acid ; this substance must theno bviously absorb ammonia, H — NH,,
water, H — OH, or the acid H — A, to give equal amounts of both dextro-
and Zaevo-alanine or lactic acid derivatives respectively.

This can, however, tio^bethe result provided the two valences of carbon
in ethylidene carboxylic add are unlike; the two methylene derivatives
obtained from d- and /-brompropionic add by loss of hydrogen bromide
must then in fact not be identical but represent two isomers of space as

+ V /CHg — V yCH,

shown by the formulae, ^C^ and yC<, . These sub-

— / .N:OOH + / ^COOH
stances then obviously absorb the dissociated ammonia, H — NH3, or water,
H — OH, present to give d- or /-alanine or the corresponding d- or /-lactic
add derivative — ^which agrees with the facts,* The moment, there-
fore, we admit the correctness of the interpretation given of the action of
alkyl halides with ammonia, water, etc., the facts presented above prove
with predsion that two of the four valences of carbon can not be equiva-
lent.

This conclusion also enables us to comprehend in a more simple man-
ner the phenomenon of autoracemation, and espedally to understand
why optically active substances can maintain their independent existence
notwithstanding the fact that the various groups or radicals bound to the
different valences of carbon are also continually dissociated from them;
thus d' or /-lactic add, d- or /-alanine, d- or /-brompropionic add although
partially dissociated at ordinary temperatures into d- or /-ethylidene
carboxylic add and water, ammonia or hydrogen bromide respectively,
maintain thdr identity for a long period of time. Nevertheless the opti-
cally active brompropionic acids, as well as thdr esters, and the corre-
sponding monobromsuccinic and phenylacetic acid derivatives go over-
as Walden has shown,* on long standing (two to five years) at ordinary
tempemtures into a mixture of equal amoimts of the d- and /-compounds.
Dexiro and laevo lactic add, on the other hand, can be Jcept indefinitely
without change at ordinary temperatures and yet each changes very
quickly, in two to three days, at 140° into racemic, t. e.,d-/-lactic add.

The analogous d- and /-iodopropionic adds are as yet unknown, but
it is certain that they must transform themselves with great speed at
ordinary temperatures into racemates so that their actual isolation is
problematical. Why are there these remarkable differences in the stability
of quite analogous optically active compounds? It is due simply to a
difference in the relative amount of existing dissodation. The alkyl
iodides are relatively far more dissodated into C« Haw and HI than the

^ I am purposely avoiding here a discussipn of the remarkable optical inversion
observed by Walden [(Ber., 32, 1833-55) and by Fischer (Ibid., 40, 489)] in a few
isolated instances in the malic and lactic add series — a reaction which all admit is
abnonnal and one which, in my opinion, when better imderstood, will also lead to the
conclusion thatfthe four valences of carbon are not equivalent.

* Ber., 31, 1420

648 REVIEWS.

corresponding bromides, chlorides or fluorides — hence the well-known
far greater activity of the alkyliodides, alkylsulphates, etc. Further-
more it is obvious in the case of the a-substituted propionic acids under
discussion, that as soon as the percentage of dissociation in these com-
pounds into d' and /-ethylidenecarboxylic acid reaches a certain definite
limit, then a very slow transformation of these two space isomeric methyl-

+ v /CH3 _^ — V yCHs

ene derivatives into each other, pC^ .^ y^C » ™^st

— / X:OOH -f ^ ^COOH

automatically take place — hence the phenomenon of autoracemation.

Finally, and this is important, since the above interpretation of the
reactions of the alkyl halides is not yet generally accepted by chemists,
we may reach the same conclusion, namely, that the four valences of
carbon are not equivalent, from an entirely different standpoint by con-
sidering another series of reactions shown also at ordinary temperatures
by optically active amino acids. When d- or /-alanine ester is treated
with nitrous acid, nitrosyl bromide or chloride respectively we always
obtain the corresponding, optically active d- or /-lactic, brom- or chlor-
propionic ester. Now the well-known conversion of primary amines into
alcohols or alkyl halides by means of nitrous acid or nitrosyl halide has
been shown to proceed, in the aliphatic series, through the following
stages, a or 6 :

/OH
(a) HO— N = 0 + H— NHC„H,«4.i -> HO— N<; -^

^NHC„H,«+x

C«Han< II +2H,0 -^ C«Ha«+,OH-hN,+H,a
^N

yOH
(6) X— N=0-fH— NHC„Ha«+i -> X— N< ->

\nhc.h^+,

C«Ha«<; II +H,0-hHX~^ C«H,n+iX-fN,+H,0.

i, e.y through an intermediate formation of a fatty diazo compound.

Consequently d- as well as /-alanine ester must go over, when treated with

the reagents named, into the two space isomeric d- and /-a-diazopropionic

Nv^ /CH3 N^ XH,

esters, II >C^ and II )>C<^ ; the first of these must give

NC ^COOR N<. ^COOR

with water or haloid acid d- lactic or d-halogenpropionic acid, whereas the
second one is necessarily converted, with evolution of nitrogen, into the
corresponding antipodes. It is of course at once apparent that if the two

CH,. /N
valences of carbon bound to nitrogen in diazopropionic ester, }QjC I ,

coor/ ^N

were equivalent, both d- and /-alanine must give on treatment with
nitrous acid or nitrosyl halide respectively equal amounts of the d- and
/-derivatives, t. e., racemates or at any rate the same products — ^whichis
however not the case ; consequently the two diazopropionic esters whidi

are the intermediate products^ in the various reactions are not identical
but isomers of space — ^in other words the two valences of carbon joined
to nitrogen in these compounds are not equivalent.

Have we any other evidence which throws doubt on the equivalence
of the four valences of carbon? Yes. I refer to asymmetric addition,
which shows us in a most simple manner why we have such a vast array
of optically active substances in the vegetable and animal kingdom.
You are all familiar with the properties of the unsaturated hydrocarbons
called the olefines; they absorb with great ease, by addition, such sub-
stances as hydrogen, halogens, sulphuric acid and haloid adds. Now
propylene absorbs the two last-named reagents to give almost exclusively
the isopropyl derivatives,

CH,CH— CH, CH3CH— CHj

I I and I I . Why is this ?

HOSOa— O H X H

The absorption reaction takes plax^e first because we have in propylene
a certain very definite but small number of molecules in an active molec-
ular condition, thus,

CH,CH«CHa :^ CHjCH— CH,

Inert propylene. Active propylene.

Furthermore the concentrated sulphuric acid or the dry haloid acid is
also partially dissociated into the active masses, H — and — OSO,OH or
— X respectively. Hence a simple union takes place between these active
substances to give the addition products named above.

Now if we assume the equivalence of the four valences of carbon it is
very difficult to understand why the addition of the dissociated H and X
particles does not take place equally on both active carbon atoms of
propylene giving like amounts of the propyl and isopropyl derivatives,
CHjCH— CH, and CH,CH— CH,, respectively. This is self-evident when

X H H X

we realize that the two active valences of active propylene are not equiva-
lent but must be represented thus:

CH.CH = CH3 Tt CHaCH — CH,(I) and CH,CH — CH,(II).

- - ^ I I I I

— + + —

We have now only to make the further assumption that active propylene

contains at ordinary temperatures relatively more active (II) than active

(I) molecules; the addition of the dissociated reagent, H — ^X, must then

+ —
obviously give CH,CH — CH, as the chief product.

* In view of the importance of ascertaining this with certainty, I am at present
working on the problem of isolating these two isomers of space. It is obvious, as
Dr. A. F. McLeod suggested to me in reading this paper, that the long known space
iaomerism existing among the oximes of ketones and aldehydes, as well as among the
diazo and azo compounds, may likewise be due to the non-equivalence of the valences
of the carbon and nitrogen atoms respectively.

6$o KEw Books.

Etnil Fischer's work in the sugar group shows that prussic acid when
added to various aldoses often gives a decided preponderance of one
of the two isomers theoretically possible. Similarly the benzilic add
rearrangement, which also depends upon an addition of active carbon
monoxide to various aldoses,^ often proceeds asymmetrically.

We are thus able to understand, in a strikingly simple way, why various
enzymes can convert the sugars into optically active destruction products
such as d- or /-lactic acid, etc., etc. Finally we may ask ourselves whether
the two unlike valences of carbon are positively and negatively charged.
It is impossible to answer this question with certainty at the present
time; it is my conviction, however, that the peculiar ease with which
carbon forms chains of great complexity and stability, as well as the fact
that the vast majority of the carbon compounds are non-ionizable sub-
stances whose reactions proceed mainly in the manner indicated above,'
must be attributed to the non-equivalence of the four valences of the
carbon atom ; furthermore the existence of various compoimds containing
bivalent carbon can also be readily* understood on this basis.

NEW BOOKS.

Immunochemistry. The Application of the Principles of Physical Chemistry to the
Study of the Biological Antibodies. By Svantb Arrhbnius. The Macmillan
Company. Price, $i.6o.

The preface of this book states that the contents contain a summan*
of six lectures on the immunity reactions, delivered at the University
of California during the summer of 1904, amplified by the addition of
new matter covering the subject to the date of publication. No informa-
tion is given as to whether this is a translation of the German edition,
or conversely; but as the German edition preceded the American by sev-
eral months the latter is presumably a translation, which assumption
is supported by the occasional occurrence of characteristic German
forms of construction.

Under the title of *'Immuno-chemistry'* (a useful term which this
book will probably cause to be adopted into the vocabulary of the "ira-
munologist") Arrhenius gathers much of the literature bearing upon
the studies that have so far been made of the chemical nature of the
reactions of immunity, but most of the space is devoted to discussion
and interpretation of the results so far obtained by the application of
the methods of physical chemistry to the problems of immimity. As
by far the greater part of this work has been done by, or under the direc-
tion of the author or his colleague, Dr. Madsen, the "Immuno-chera-
istry*' partakes largely of the nature of a monograph upon the physical
chemistry of immunity reactions; consequently a review of the book
almost necessarily resolves itself into a criticism of the value of the in-

> Cf. Ann., 357, 231-3.

■ See This Journal, 26, 1577.

NBW BOOKS. 651

vestigations upon the physical chemistry of the immunity reactions
that are therein recorded. Certainly when a physical chemist of the
pre-eminent standing of Prof. Arrhenius collaborates with so experienced
an investigator of the problems of immtmity as Prof. Madsen, the out-
come of their labors is sure to receive most respectful consideration,
and whether one does or does not agree with all their conclusions, none
will deny that they have opened an almost untouched field of research
to general investigation. The results and hypotheses produced by their
work and recorded in collected form in this book are certain to stimu-
late a great amount of research and controversy, which are botmd to
be fruitful whether or not they confirm the conclusions of Arrhenius;
therefore the actual value of this book is certain to be very great, and
it constitutes a welcome addition to the literature of immunity. As
to the actual significance of the results so far obtained through the study
of immunity reactions by the methods of physical chemistry, there is
room for difiFerence of opinion. Already many of the conclusions ex-
pressed in this book, and previously published in special periodicals,
have been sharply attacked; and in the opinion of the reviewer these
attacks have been justified, for there is no question that among the
many investigations recorded are to be found serious errors of expen-
ment and of interpretation. To discuss the specific points of Arrhenius's
work that have been found open to attack would require more space
than can well be allotted to a book review, but it may be stated that the
chief source of error would seem to lie in the enormous disparity that
exists between the very exact methods of physical chemistry and the
very uncertain materials of immunology, which makes the application
of one to the other a very hazardous procedure.

Only the investigations of numerous workers through a long period
of time can finally decide the actual value of the work done and the con-
clusions reached by Arrhenius and Madsen and their adherents, but
whether supported or not, its stimulating influence is certain to have
a profound eflfect upon the future development of our knowledge of this
all-important subject of immtmity. As the authorized exposition of
this work, Arrhenius's **Immuno-chemistry" will of necessity stand as
a classic in the literature of immunity, and to pick at the faults of style
and rhetoric would be trivial; to attempt to judge its actual present
worth or to predict future valuation would be presumptuous. Its use-
fulness is assured, and it will be read eagerly, and zery critically, by every
student and investigator of the processes of immunity. It is un-
fortunate that the exigencies of the subject are such that the book will
be read understandingly by but a very limited number of scientists;
the unavoidable use of the complicated terminology of immunity will
stand in the way of the chemist, except in those sections devoted to the

652 RECENT PUBLICATIONS. '

physical chemistry of enzyme reactions, where there is gathered much
valuable material; on the other hand, the constant and necessary use
of mathematical expression will be a serious obstacle, unfortunately,
to many immunologists. H. Gipeon Wells.

RECEirr PUBLICATIONS.

Andres, L. E.: Zellaloid and seine Verarbeitung. Wien: 1908. gr. 8. 37455.
M. 6.

Baker, T. Thorns: The Spectroscope: Its Uses in General Analytical
Chemistry; an intermediate text-book for practical chemists. New York: WilEam
Wood & Co. 1908. 8vo. $1.75.

Bbrthblot, M., et Jungpi^bisch, E.: Traits 6t6mentaire de Chimie Oiginqiue.
4^ d. Vol. I. Paris 1907. L'ouvrage complet. 2 vols. 1904-1907. 2252 pp.
M. 42.

Blochmann, R.; Luft, Wasser, Licht und Wilnne. 9 Vortr&ge aus dem Gebiet
der Experimentalchemie. 3 Auflage. Leipzig: 1907. gr. 8. 149 ss. M. i.

Cohen, E.: Das Lachgas. Eine chemischkultur-historische Studie. Leipzig:
1907. M. 3,60.

Conduchs, a.: Contributk>n a P^tude des Ozyui^es et Carbamidoxima. Paris:
1907. 141 pp. M. 4.

Faraday, M.: Chemical History of a Candle. New Edition. London: 1907-
I S.

Hammarsten, Olop.: Text-book of Physiological Chemistry. Authorized trans-
lation from the author's enlarged and revised 6th German edition, by J. A. MandeL
New York: John Wiley & Sons. 1908. $4.

Handbuch der angewandten physikalischen Chemie in EinzeLdarstdlungen. Herans*
gegeben von G. Bredig. Band VIII. Leipzig: 1907. I gr. 8. 204 ss. M. 9. Infaalt:
MtUler, A. AUgemeine Chemie der Kolldde.

Hawk, P. B.: Practical Physk>logical Chemistry. London: 1907. 16 s. 6 pp.

Hildebrandt, H.: Neuere Arznelmittel. Beziehungen zwischen dereo chem-
ischen Konstitution und pharmakologischen Wlrkung.^Ldpzig:^ 1907. gr. 8. 168
ss. M. 4,20.

Jt^PTNBR, Hanns von. : Lehrbttch der chemische Technologie den Energks.
Ill Bd. Die chemische Technologie d. strahlenden u. d. elektrischen Energie. Leip-
zig: 1908. M. 10.

Kraus, K. : Ezperimentierkunde. Anleitung zu physikalischen und chemischefl
Versuchen an Volks- und Bilrger-schulen. Wien: 1907. 360 ss. M. 4,20. -p4l

Lbwis, E. T.: Inorganic Chemistry. New York: G. P.SPutnam's Sons, f 1908.
8vo. 424 pp. $1.25.

MoYB, A.: Die Gewinnung und die Verwendung des Gipses. (Bibliothek def
gesamten Technik, 72 Bd.) Hanover, 1908. M. 2,40.

OsTWALD, W.: Prinzipien der Chemie. |Eine Einleitung|in*alle chenuscfaen Lefar-
bucher. Leipzig: 1907. M. 8,80.

Roi«OFF, M.: Grundriss der physikalischen Chemie. Leipzig: 1907. gr. 8.
M. 5.

ScHULz, F. N. : Praktikum der physiologischen Chemie. Kuizes Repetitonnm.
3 Aufl. Jena: 1908. M. 2,50.

Spektraltafel nach der Orighialzeichnung von G. Kirchhoff und R. BwMCQ* Nene
Auflage. Wien: 1907. M. 6.

Vou XXX. May, 1908. No. 5.

THE JOURNAL

OF THE

American Chemical Society

[Contributions proh thb Rbssarch Laboratory of Physicai. Chsustry op
THB Massachusetts Institutb of Tbchnouxsy No. 34.]

SOLUTIOir OF METALS IN NOlf-METALLIC SOLVEIITS; n.' ON THE
FORMATION OF COMPOUNDS BETWEEN METALS

AND AMMONIA.

Bt Chaklbs a. K&aus.
Received February 14, 1908.

Solutions of metals in liquid ammonia were first obtained by Weyl,*
who brought together sodium and potassium with gaseous ammonia
under pressure. However, he mistook the solutions which are formed
under these conditions for simple compounds and assigned to them the
formulae NaNH, and KNH,, respectively. These compounds he sup-
posed to be structurally analogous to the hypothetical free ammonium
group, being derived therefrom by substitution of an atom of hydrogen
by one of the metal in question. Taking account of this relationship
in his nomenclature, he introduced the terms sodammonium and potass-
ammonium, respectively, which nomenclature has been largely adopted
by subsequent investigators and commentators.

The study of solutions of metals in ammonia was materially advanced
by Seely,* who showed conclusively that solutions result in the action of
ammonia on the alkali metals. Prom a consideration of the optical
properties of these solutions, he concluded that a compound was not
formed between the two components. Neither Weyl nor Seely, how-
ever, was able to adduce quantitative data in support of his contention.

These investigations on the action of ammonia on the alkali metals

* For the first paper of this series, **I. General Properties of Sohitions of Metals in
Liquid Ammonia," see Tms Joxjrnal, 29, 1557-157 1 (1907).
" Ann. Physik, 121, 6oi (1864).
' Chem. News, 23, 169 {ifitqi).

654 CHARLQS A. ERA^S.

seem to have excited little active interest for, excepting an isolated ob-
servation by Gore/ we find no further investigations recorded until
1889, when Joannis' undertook an extended series of investigations in
this field. To him belongs the credit of bringing quantitative data
to bear on the problem of the compounds formed by soditun and potas-
sium with ammonia. He devised a means of isolating and analyzing
these compounds, to which he assigned the composition NaNH, and KNH,
respectively, and to which, like Weyl, he ascribed an ammonium stnic-
ture. Employing the method devised by Joannis, Moissan obtained
the compounds LiNH,, Ca(NH,)/ and LiCH,NH„* while Mentrel ob-
tained the compound Ba(NH,)e* and Roederer the compound Sr(NH,)|.*

In the preceding paper^ attention was called to the fact that the con-
centrated solutions of metals in ammonia exhibit metallic reflection
and are consequently opaque. It is plain, therefore, that the fonna-
tion of a compound cannot be ascertained by visual observations nor can
separation of the different phases in these concentrated solutions be car-
ried out by the simple means usually employed in the preparation of a
pure substance. The method adopted by Joannis in preparing and
identifying the compounds in question is therefore an indirect one, as wSl
be seen from the description given below. Objections have been raised
from time to time to the results obtained by Joannis as well as to those
obtained by other chemists emplo3ang the same method.

It is the purpose of the present paper to determine, if possible, whether
or not compounds are formed. To this end I shall first examine such
evidence as is already at hand. The solutions of sodium and potassium
in ammonia have been studied extensively and, as will be seen below,
the available data are sufficient to enable us to draw the conclusion that
solid compounds are not formed. In the case of other metals it has been
found necessary to adduce new experimental evidence. It will thus be
shown that lithium, like sodium and potassium, does not form a solid
compound, while calcium forms the compotmd Ca(NH|}e with ammonia.
Finally the questions of constitution and nomenclature will be consid-
ered and the physical properties of the compound Ca(NH,)e will be dis-
cussed briefly.

Criterion for the Appearance of New PhaaeB.

In determining whether compoimds are formed between a solvent
and a dissolved substance, it is of primary importance to possess a dear

» Phtl. Mag., 44, 315 (1873).
' Compt, rend,, zog, 900 (1889).
» Ibid., 127, 685 (1898).
^ Ibid,, 128, 26 (1899).

* Ibid,, 135, 790 (1902); Bull, soc, Chim., 29, 493 (1903).

* CompL rend., 140, 1252 (1905).

* This Journai«, 29, 1570.

SOLUTION OF Ml^ALS IN NON-METAI^UC SOLVENTS. 655

knowledge of the phase relations before attempting to identify any of
these phases as compounds by means of chemical analysis. Since the
solutions of metals in ammotiia constitute two component systems, it
follows that if anmionia is withdrawn from the system, the pressure must
become constant as soon as a third phase appears. If, on continuing
the withdrawal of ammonia, one phase disappears the pressure will again
vary when ammonia is withdrawn. If, however, one of the three phases
is substituted by a new third phase, the pressure changes abruptly to a
new constant value. One of the phases present is always gaseous am-
monia; the other two phases may either be both solid or liquid, or one
may be liqiud and the other solid. It is an easy matter to determine
how many new phases make their appearance in withdrawing ammonia
from a dilute solution of a metal until the free metal and gaseous ammo-
nia are left behind. The study of the vapor pressure of a system of
metal and ammonia is therefore a necessary preliminary in determining
whether compounds are formed. The nature of the phases present at
any time may, in geneml, at once be determined by visual-examination,
and by proper means it is always possible to transform the entire system
into any desired phase, when its composition may be determined by
anal3rsis.

Non-existence of the Compounds NalTH, and KITH,,

The pressure of the systems sodium-ammonia and potassium-am-
monia have been carefully investigated by Joannis.* He finds that in
the case of both these systems, if ammonia is withdrawn from a solution
of the metal, the vapor pressure falls until a solid phase makes its ap-
pearance, after which the pressure of the system remains constant until
only free metal and gaseous ammonia are left. This would seem to show
that the solid phase which initially separates from solution is free metal.
Joannis, however, believes that such is not the case. He observed that
the soHd, which initially precipitates, appears to possess the same color
of metallic reflection as does the solution itself, while, as is well known,
the free metal possesses white metallic reflection. He therefore con-
siders that this solid may be a compound of soditun with ammonia. We
shall simply call this substance "solid compound" in order to avoid
circumlocution. This substance is evidently a new phase; the question
only remains to show whether it is a compound, as Joannis believes it
to be, or, otherwise, free metal and saturated solution as the vapor pres-
sure relations indicate. According to Joannis, free metal does not make
its appearance until the saturated solution is completely converted into
the soHd compound, after which, on further withdrawal of ammonia,
the free metal appears, without being accompanied by any change in

' Loc. cii,: For details of the method described in this paragraph v. Ann, ckim,

t^y^f 7i 13-36 (1906).

656 CHARLES A. KRAUS.

the pressure. He assumes, therefore, that the dissociation pressure of
the *' solid compound " is exactly equal to the vapor pressure of its satura-
ted solution. Analysis of the "solid compound" gave him a composi-
tion corresponding to the simple formula MeNH,, where Me may be
either Na or K.

According to the phase rule, the system, vapor, saturated solution,
"solid compound," and free metal must be an invariant one and obtain-
able at a single temperature only, since two components and four phases
are present. Roozeboom* first called attention to these facts and sug-
gested that Joannis had carried out his experiments at the tempera-
ture of this invariant point. Joannis, however, showed that at a series
of temperatures no change occurs in the pressure from the moment that
a solid phase begins to separate until only free metal and gaseous am-
monia are left. This proves conclusively that the solid phase initially
separating out of the solution is identical with the solid phase which is
finally left behind, namely, free metal, and that a solid compound is not
formed between the metal and ammonia. The ** solid compound" can
consist only of free metal and saturated solution of the same in ammo-
nia. Ruff and Geisel' have recently expressed this view and in addi-
tion they have adduced some evidence to show that the solid compound
consists of a free metal covered with a film of solution, which adheres
through the action of strong surface forces. Joannis,* replying to this
paper of Ruff and Geisel, throws some doubt on the correctness of the
conclusion which they have drawn from their experimental results. In
the section dealing with lithium, whose behavior is in every way similar
to that of soditmi and potassium, I shall adduce independent evidence
showing conclusively that, in accordance with the view of Ruff and
Geisel, the so-called compound LiNH, contains solution and free metal
This evidence was not obtained primarily to show that compounds do
not exist, for as to this point the thermodynamic evidence would seem
to be sufficient in itself. Since, however, Joannis* believes that his hy-
pothesis may be reconciled with the phase rule, it may be as well to give
evidence which is quite independent of any theoretical considerations.

This reconciliation of his hypothesis with the phase rule Joannis* be-
lieves to have been effected by Moutier.* A careful examination of
Moutier's paper fails to show, however, that such a reconciliation is pos-
sible. A consideration of the free energy of a system can lead to no re-
sults other than those obtainable by other thermodynamic methods.

* Compt. rend., no, 134 (1890).
' Ber., 39, 831 (1906).

' Ann. chim. phys., 11, 101 (1907).

* Loc, cit.

* Loc. cit.y 103; Ann. chim. phys., 7, 34 (1906).

* CompL rend,f xzo, 518 (1890).

SOLUTION O^ MBTALS IN NON-METALLIC SOLVENTS.

657

The conclusions reached by Moutier are in fact applicable to a single
temperature only and not to a series of temperatures, as be believes them
to be.

Vapor Pressure of the System LilfH,.

According to Moissan/ the compound LiNH, is formed. Since he
employed the same method as did Joannis in the case of sodium and
potassium and since confirmatory vapor pressure measurements were
lacking, it seemed advisable to determine the pressure relations of the
system Li-NH,.

The apparatus employed need not be described in detail. It consisted
essentially of a tube permanently attached to a manometer and contain-
ing the solution. Provisions were made for connecting this tube with
a source of pure ammonia vapor or a vacuum pump as desired. After
introducing the metal into the containing tube, it was placed in a bath
at — 10° and ammonia vapor was introduced under a pressure slightly
greater than one atmosphere. The process of condensation and solu-
tion takes place with great facility in the case of lithium, even at ordinary
temperatures. When the process of solution was complete the container
was placed in a bath at 20*^ and the excess of ammonia was allowed to
escape under a pressure of about 100 cm. of mercury. On withdrawing
ammonia by means of the pump, the pressure fell until it reached a value
of about 10 centimeters, after which it remained constant. At this
point a solid phase began to separate out and on continuing the with-
drawal of ammonia the amount of solid increased. The solid phase ap-
peared to have the same color as the solution and the solution apparently
disappeared long before all the ammonia had been completely withdrawn.
The pressure, however, remained constant imtil only free metallic lithium
was left behind. When but very little ammonia was present the color
distributed appeared imiformly over the entire surface of metal, giving
it a slight tinge of color. On adding more ammonia the color became
more pronounced while, at the same time, the crystals of metal, which
previously were very sharply defined, lost their sharp outline, their edges
becoming indistinct. The phenomenon is such as we should expect if the
crystals were covered by a film of highly colored metallic solution.

The equilibrium pressure of the saturated solution of lithitun in am-
monia was measured at a niunber of temperatures with the following results :

Temperature.
19.3**

Pressure.

cm.

Mean.

9.53 cm.

J9.48
)9.58

/3-3I
» Compt. rend., 127, 685 (1898).

3.31

331

Temperature.
9 75**

20.3**
20.3**

Pressure.
5.68 cm

5.72

9.91

10.06

9.90
10.06

Mean.

5.70 cm.

9.98

658

CHARI^ES A. KRAUS.

The pressures first recorded were obtained after abstracting, the second
after adding ammonia.

From the pressure data the heat evolved when one gram-mokcuk
of ammonia vapor combines with metal under equilibrium conditions
to form a saturated solution may be calculated from the equation:

^ T,-T, ^ p^

Using the values of p at o® and 20.3*', respectively, we thus obtain
Q = 8698 calories. The value of P at 9.75° calculated from this value
of Q is 5.699 cm. while the value 5.70 was fotmd.
Independent Evidence Showing that the Solid Compound LlRH| is Hot

Formed.
The pressure relations recorded above show conclusively that a com-
pound is not formed between lithitun and ammonia. It seemed worth
while to examine the behavior of lithium solutions somewhat further
in order to determine, if possible, the sources of the error in the result
of Moissan. In common with other investigators who have adopted
the method of Joannis, he finds that so long as the amount of ammonia
present exceeds that corresponding to the formula LiNH,, that is, when
saturated solution is present, any free bit of metal at once absorbs am-
monia from the saturated solution to form the compoimd. On the other
hand, when the amotmt of ammonia is less than that which corresponds
to the formula LiNH,, a free surface of metal remains permanently free
from ammonia. This is, in fact, the criterion which these investigators

employ in determining when the saturated solution dis-
appears and the supposedly "solid compound" begins to
dissociate. Since no change occurs in the pressure when
the composition passes through the point corresponding
to the composition LiNH,, it isdifficult to understand why
there should be any difference in the behavior of the
system on one side and the other of this point. The fol-
lowing experiments were imdertaken for the purpose of
obtaining more light on this question.

A quantity of lithiiun was introduced into a tube of
the form outlined in Fig. i. Ammonia was condensed
at 0° until the metal had all dissolved, after whidi
ammonia was withdrawn by means of the pump until
a portion of the metal had been precipitated. A por-
tion of the solution spattered along the walls of the tube
EC, while a few isolated drops of solution could be seen
* em pi eyed in clinging to the walls of the smaller tube AC.
studying lithium On immersing the tube as far as E in a bath at 15°.
solutions. 10° below room temperature, the solvent at once evapo-

r A*

JOL

SOI^UTION OF METAtS in" NON-imTAI^UC SOLVENTS. 659

rated from the isolated drops in AC, leaving metal behind, while along the
walls of the tube EDC the metal retained ammonia. This portion of the
tube was then warmed with the hand, whereupon the ammonia evaporated.
As soon as this heating process was discontinued, the solution from the
bottom could be seen creeping up over the walls of the tube. The solution
passed from crystal to crystal, tracing out characteristic figures, such' as
would naturally result if a liquid were to creep over a surface covered
with irregularly spaced particles. Isolated particles of metal like those
in the tube AC underwent no change whatever. On plunging the entire
tube in the bath the particles in AC remained unaflfected.

After repeating the above operation a number of times, it was ob-
served that the amotmt of metal in the space DC -had greatly increased.
This is evidently due to the fact that the metal on the walls of the tube
acts as a wick by means of which solution is drawn from the colder por-
tion of the tube in the bottom to the warmer portions above. Here
the solvent evaporates to condense again in the bottom, while the metal
is left behind in the warmer portion of the tube. To test this further,
the tube was left immersed as far as £ in a bath at 12®. At the end of
45 minutes the metal had crept up the tube AC over a distance of more
than a centimeter, forming a very heavy deposit. The deposit of metal
in CD had now become so heavy that it could not be freed from anmionia
by warming with the hand. The tube was left immersed as far as D
in a bath at o^ for some time, after which all the ammonia was withdrawn.
On examination it was found that no metal was left in the bottom, it
having collected in the warmer portion of the tube DC.

A small quantity of ammonia was now introduced, suflScient to form
only a thin film on the surface of the metal. The tube was placed in
ice-water as far as C, while the metal deposit extended about 1.5 cm.
above C in AC. On warming the tube at B with the hand, the metal
deposit crept up the tube with visible speed, and at the end of about
ten minutes the entire surface of the tube AC was coated with metaL
The tube AC was kept warm for some time, after which the ammonia
was again withdrawn. Nearly all the metal now appeared collected
in the tube AC.

It is remarkable that this process should take place so rapidly against
a temperature difference of nearly 30°. No better illustration could
be given of the strength of the surface forces coming into play between
the metal and its solution. It is to be remembered that the amotmt
of ammonia was but a small fraction of that required to form a com-
pound of the composition LiNHj. This experiment, therefore, shows
conclusively that liquid is present in the system even when, according
to Moissan, only the compound LiNH, should be present. These experi-
ments show, moreover, that the behavior of the S3^tem is the same irre-

66o CHARLES A. KRAUS.

spective of the amount of ammonia present so long as this is less than
that necessary to form a saturated solution of all the metal present. Those
phenomena which Moissan employed in isolating the supposed compoand
LiNHj, I have been quite unable to reproduce. It is clear, then, that
in accordance with the vapor pressure relations, the properties of lith-
itmi in the presence of small quantities of ammonia, are such as to pre-
clude the possibility of a compound being formed at the temperatures
of the present experiments.

The behavior of other alkali metals is doubtless similar to that of
lithium. In the case of sodium I have observed this experimentally-
Caesium and rubidium have not been investigated as regards their vapor
pressure relations. Moissan^ obtained what he believed to be the com-
pounds CsNH, and RbNH^. It seems not improbable, however, that
the same errors underlie the results obtained with these metals as has
been shown to imderlie those obtained with the remaining alkali metals.

The Vapor Presgure of the System Ca-KH^

It having been shown that the alkali metals do not form compounds
with ammonia, it seemed important to examine at least one member of
the group of the alkaline earths. For this purpose calcium was selected
According to Moissan, the compound CaCNHs)^ is formed, the method
employed being that of Joannis, which has been shown to be untrust-
worthy. Pressure data were not obtained by Moissan, but he states
that above o^ calcium combines with ammonia without liquefaction,
while at lower temperatures liquefaction takes place. This indicates
that a solid compound is formed whose dissociation pressure lies below
one atmosphere, while the vapor pressure of its saturated solution lies
above one atmosphere at temperatures in the neighborhood of o®. That
this is correct, follows from the experiments about to be described.

Observations on the phase relations in the more dilute solutions of
calcium in ammonia were made in connection with some conductivity
experiments. Solutions of calcium, like those of sodium, separate into
two liquid phases. The point of complete miscibility of these two phases
lies at much higher temperatures in the case of calcium than it does in
that of sodium. Even at room temperatures, the concentration of the
two phases differs very widely. At — ^33** the concentration of the di-
lute phase does not exceed i/io gram-atom per liter, while that of the
concentrated phase is. such that pronounced metallic reflection results.
Owing to the tendency of the concentrated phase to cling to the walk
of the containing tube and also, perhaps, because of the small difference
in the specific gravity of the two phases, it was not possible to separate
them into two layers, one above the other. The fact that the concentra-
* Loc, cit.

SOLUTION OF HBTALS IN NON-MBTAI4JC SOLVBNTS.

66i

0
E

111"

Mta.

ted solution clings to the walls of the container made it very difficult to
determine the nature of the system. However, on using very large quan-
tities of solvent and small quantities of metal the concentrated phase
could be plainly seen adhering at intervals to the walls of the tube or
floating about in the dilute solution.

In studying the concentrated solutions of calcium, a tube of the form
outlined in Pig. 2 was employed. The tube G serves as a receptacle for
the metal or its solution. At F it is joined
to the manometer system C by means of a bit
of rubber tubing. The stop-cocks D and E
are provided so that at any time G may be
detached and weighed, the cocks being closed
beforehand. When G is again attached, the
air is exhausted from the tube ED before open-
ing E. The cocks A and B make connection
with a source of pure ammonia and a vacuum
pump, respectively.

A piece of metal, freshly cleaned, is intro-
duced into G through H, which is immedi-
ately sealed off, after which G is exhausted.
The tube G having been weighed before-
hand, the weight of metal is obtained by
detaching G and weighing. The amount of
ammonia present in G at any time may be
obtained in a similar maimer.

After having introduced a quantity of metal ^^f ^.-Apparatus employed
,. ^ -i^*.*** ^ Ml analyzinR cralcium com-

accordmg to the method descnbed above, G pounds.

was placed in a bath at 20® and ammonia

\'apor was introduced under a pressure slightly greater than one atmosphere.
At the end of 10 minutes appreciable absorption of ammonia had not taken
place. Even at 0°, absorption was inappreciable at the end of 5 min-
utes. The containing tube was then placed in a bath at — ^33° and am-
monia was condensed. It was noticeable that calcium dissolves at a
much slower rate than do the alkali metals. When the metal was all
in solution, the tube was allowed to warm up, the excess of ammonia be-
ing allowed to escape under a pressure of i 1/3 atmospheres. In the
neighborhood of 0°, the liquid disappeared, leaving behind what appeared
to be a solid metallic substance identical in color with the solution from
which it was precipitated and possessing the mechanical properties of
a solid.

The tube was now placed in a bath at 22° and ammonia was with-
drawn by means of the pump. The pressure fell rapidly and reached a

662 CHARLES A. KRAUS.

constant value at lo centimeters. Pressure readings were carried out
at different temperatures with the following results:

Tempenture. Pressure.
I0.8® 4.60

O® 2.28

21.7** 9.07

21.7® 9.07

43.7** 30.67

The last two determinations were made at different times. The equilib-
rium pressure could be reached from one side only. This was due to the
extreme slowness with which ammonia combines with calcium. Such
combination does take place, however, as will be seen below. In the
last experiment a slow increase in pressure was noted, due, without doubt,
to the formation of amide and hydrogen from the two constituents.

Calculating, as above, the value of Q from the pressures at o*' and 21.7°
we find Q = 10,230 calories for the heat evolved when one gram-mole-
cule of ammonia combines with calcium to form the compound in ques-
tion. This value of Q gives the pressures 4.66 cm. and 30.20 cm. at 10.8®
and 43.7°, respectively. The corresponding values fotmd are 4.60 cm.
and 30.67 cm.

The tube was now cooled to — ^33°, and ammonia was again condensed
until the metal was dissolved. Leaving the tube G in its bath of bofl-
ing ammonia, solvent was withdrawn with the pump. The pressure
soon reached a constant value at about 50 centimeters and a solid sub-
stance apparently crjrstallized from solution. Evidently a compound
precipitates out of solution at this pressure, which in turn loses ammonia
at a much lower pressure.

A fresh piece of calcium weighing 0.4189 gram was introduoed into
G, and ammonia was condensed until the process of solution was com-
plete. Leaving the tube in its bath, ammonia was withdrawn imtil the
pressure became constant and solid began to precipitate. The follow-
ing pressure observations were then made :

Temperature. Pressure, Approached from

— ^32 . 5 ® 47 . 42 cm. higher pressure

— 32.5® 46.98 " lower

—32.5® 47 28 " higher

—50^ 19.28 " higher

— 50° 19.28 ** lower

— 32.5** 47 18 " higher

— 32.5** 47 -08 " lower

—32.5** 47.42 " higher

As may be seen, the pressure reading differs slightly according as the
equilibrium is approached from higher or from lower pressures. Thib
is due to the slowness with which equilibrium establishes itself. Tak-

SOLUTION OP HSTALS IN NON-HSTAIXtC SOLVBnTS. 663

ing the mean of these observations, we obtain the following values for
tbe pressure of the saturated solution, namely: ( = 50°, p = 19.28 cm.,
' - —32-5°, P = 4718 cm.

For tbe heat of solution, we obtain from these data, by calculation,
the value Q = 5458 calories when one gram-molecule of gaseous am-
monia dissolves the compound under equilibrium conditions to form a
saturated solution.

The pressure-composition curve for calcium and ammcmia at — 32.5° is
represented in Fig, 3, where the ordinates represent pressures in cetiti-
meters and the abscissae composition m mols of calcium per too mols
of calcium and ammonia. Portions of the curve are exaggerated in order
to bring out certain points. The correct pressures, however, af^ear on
the margin. Along AB we have the change in pressure of a dilute solu-
tion with concentration. At B, a second liquid phase of concentration

Per cent, of meul accdTdlDK to ronnDU weight.

Fig- 3< — Vapor pressure — composition ci^ve for caldttm in ftmrnonia.

c appears, and the pressure remains constant until the dilute phase of
concentration b disappears. This pressure is within a millimeter of the
atmospheric pressure, since b is very small. Along CD the pressure
iaOs to 47.18 cm., when the solid compound of composition Ca(NH,),
(see below) appears. The pressure now remains constant imtil the satu-
rated solution of composition d disappears, when the pressure falb
abruptly to 1.8 mm. (calculated) and metallic calcium appears. This
pressure is maintained as long as the solid compound remains. That

664 CHARLES A. KRAU&

the solid compound does not lose its ammonia in two steps instead of one
was shown by measuring the pressure: first when the compound had
lost but a little ammonia, and second when only a small amount of am-
monia was present in the system. This result will be referred to below
in connection with the possible formation of the compound Ca(NH5)4.
Moissan^ states that calcium does not absorb anmionia above 20^
As already stated, calcium in the massive form absorbs ammonia very
slowly even at o®. This is to be expected, since a compound is formed
which does not liquefy in the presence of ammonia. By employing very
finely divided calcium, as it may be obtained by completely withdrawing
the ammonia from a solution of the same, it was found that at tempera-
tures above 20® ammonia combines with calcium, although the process
is a very slow one.

Composition of the Compound of Calcium and Ammonia*

Experiment i. — In the preceding experiment in which the vapor pres-
sure of a saturated solution of the compound of calcium in ammonia
was determined, 0.4198 gram of metal was employed. At the end of
the pressure experiments, which occupied in all about 8 hours, the solu-
tion was placed in a bath at o^ and ammonia was withdrawn until a
pressure of about 10 centimeters was reached. The tube was then weighed.
The contents of the tube weighed 0.9873 gram in excess of that of the
metal present. Assuming that this excess in weight is due to ammo-
nia combined with calcium, we may calculate n the number of molecules
of ammonia per atom of calcium. We thus have

09873X401 _ .

"* 0.4198 X 17.06 "" ^•^^^•

The correct value is probably either 5 or 6. In view of the fact that the
solution was prepared eight hours before analysis was made, it is not im-
probable that a portion of the metal reacted with the solvent according
to the equation: Ca -f 2NH3 = Ca(NH,)j + H^.

Experiment 2, — To avoid errors due to possible formation of amide
the following experiments were carried out as rapidly as possible. In
this experiment 0.2489 gram of metal was employed. Ammonia was
withdrawn at 0° until the pressure nearly reached the dissociation pres-
sure of the compound. There was found a gain of 0.6218 gram, corre-
sponding to n = 5.874.

Experiment 3. — The calcium introduced weighed 0.2075 gram. After
pumping off the excess solvent at 20°, there was found a gain of 0.5142
gram, giving n = 5.825. .

Experiment 4. — ^In this experiment 0.5142 gram of metal was em-
ployed. The excess solvent was removed at o^ as in Experiment 2.
^ Loc. cit.

SOI^UTION OF METALS IN NON-METALWC SOLVENTS. 665

There was found a gain of 1.2830 grams, from which n may be calculated
to be 5.864.

After weighing, the tube was again attached to the pump and the am-
monia was completely eliminated. The contents of the tube now weighed
0.01 16 gram in excess of the weight of metal initially present. This
indicates that amide is formed during the experiment. It was accord-
ingly decided to carry out several experiments in which the solvent
should be withdrawn at lower temperatures. It is to be mentioned, how-
ever, that at lower temperatures a longer time is required in removing
the excess of solvent.

Experiment 5. — Employing 0.3888 gram of calcium from which the
excess solvent was withdrawn at — ^33 ^'j a gain of 0.9772 gram was ob-
tained. This gives n = 5.909.

Experiment 6, — This experiment is a duplicate of No. 5. There was
employed 0.4491 gram of metal and found a gain of 1.1255 grams. Prom
these results n is found to be 5.891 molecules of ammonia per atom of
calcium.

Collecting the results of the last five experiments we have :

BxperimeBt No. n. Temperature.

2 5.874 o<>

3 5.825 20®

4 5.864 o®

5 5.909 —33**

6 5.891 —33**

The temperatures here given are those at which the excess of anunonia
was withdrawn. It is plain that at lower temperatures the value of n
is consistently larger than at higher ones, while at the same tempera-
ture the results are in good accord. That a portion of the metal reacts
with ammonia is indicated by the fact that a slow but steady increase in
pressure may be observed, particularly at higher temperatures. We
may conclude, therefore, that the composition of the compounds of cal-
cium is represented by the formula Ca(NH3),.

As already stated, Moissan describes a compound CaCNHj)^. It might
be though that the compound Ca(NHj)^ breaks down in two steps, in
which case CaCNHj)^ should appear as an intermediate product. The
experiments described above show conclusively, however, that the com-
pound dissociates according to the equation

Ca(NH.)e ^ Ca + 6NH3 .
Compounds of Barium and Strontium.

A compound of barium, whose composition is Ba(NH3)3, has been de-
scribed by Mentrel.* Systematic pressure determinations were not car-
ried out for the purpose of determining the phase relations, but it is stated

^ Loc. cit.

666 CH ARISES A. KRAUS.

that above oertain temperatures a solid compound is obtained which
does not liquefy in excess of ammonia. It can scarcely be doubted,
therefore, that a compound is formed. Employing the method of Joannis,
he finds the decomposition decreasing from 6.97 molecules of ammonia
per atom of metal at — 50° to 6, 10 at o®. He believes the compound
Ba(NH,)Q to be formed, the excess ammonia being present as a solid
solution in this compound. There seems to be need of further evidence
on this point.

Roederer^ has investigated the action of ammonia on strontium. His
results are in every way similar to those of Mentrel in the cate of barium.
He believes the compoimd Sr(NH,)e to be formed. Here also the com-
position is a function of the temperature and a solid solution is suggested
by way of explanation.

Nature of the Compound CadTH,)^,

It was stated at the beginning of this paper that Weyl, on discovering the
solutions of sodium and potassium in ammonia, concluded that com-
potmds resulted, to which he assigned an ammonium structure. Joannis,
beHeving that he had. isolated compotmds of sodium and ammonia, re-
tained these conceptions of Weyl as to the constitution, and other in-
vestigators have, for the most part, accepted this view. In the case of
compotmds containing a considerable number of ammonia molecules,
the structural formulae advanced are rather complex to say the least,
and they lack for support a single physical property or a single reaction
that would indicate a structure such as has been proposed. How little
this theory of constitution was based upon facts is well illustrated by
the fact that those compounds which first led to the ammonium theory
have now been shown to be non-existent. In the case of the metals
of the alkaline earths where six molecules of ammonia are present per
atom of metal and where all are given oflF at the same pressure, it is no
longer necessary to take the ammonitun theory into consideration. The
term metal-ammonium should therefore be dropped from the literature
as Ruflf and Geisel have suggested.*

Before discussing further the question of the constitution of the com-
potmd Ca(NH,)e it will be necessary to consider, briefly, the properties
of this substance. It has already been mentioned that the compound
is identical in appearance with the solution from which it is precipitated,
i. e., that it possessed the satne optical. properties. To whatever molec-
ular condition these opticar properties may be due, it is plain that they
obtain in both solid and solution aM' are therefore not dependent on
the physical stg.te.of the system* . Now the compound and its solution
not only possess the same optical properties: but they likewise possess
the ^me ejbectrical properties, for both;the solid and its solution exhibit
^ Loc, cU,

SOLUTION OF METAI3 IN NON-METAUIC SOLVENTS. 667

metallic conduction. It seems possible, therefore, that the same factors
which govern the optical properties of these two substances also govern
their electrical properties. This suggests that if the factors governing
the electrical and optical properties may be determined they will lead
to some knowledge as to the state of the compound (^(NHa)^ in its solid
state. A further discussion of this point cannot be undertaken, how-
ever, until the data relating to the properties of the solutions in ammo-
nia have been pesented.

That the compound Ca(NH,)8 is capable of existence is an important
fact. We have here for the first time a compound in which a metal ap-
pears combined with a solvent without at the same time being com-
bined with a strongly electronegative element or group of elements,
as is commonly the case in solvated salts. The nature of the forces com-
ing into play when calcium combines with ammonia must be quite differ-
ent from those involved in the combination of a metal with a negative
element to form a salt, for, while in the latter case all metallic proper-
ties are lost, in the former the metallic properties persist. The salts are
usually considered to be valence compounds, and their formation is sup-
posed to involve forces of an electric nature. The compotmd Ca(NHg)8 ^
belongs to the class of compounds which, like the solvates, are commonly
grouped under the head of molecular compotmds. That electrical forces
play only a minor part in the compound (CaNH,)^ is indicated by the per-
sistence of the metallic properties.

It seems probable that the ammonia present in the calcium compound
is combined with the metal in the same manner as in the case of ammo-
niated salts or solvated ions. Indeed the thought lies near that Ca(NHs)o
is simply a free positive ion which is present to some extent when a cal-
cium salt is dissolved in ammonia. It is interesting, also, to note that
this compotmd corresponds with both Abegg's theory of contravalences
and Werner's co-ordination number.

Since the compotmd Ca(NH,)j appears to be a solvate of the metal
calduniy the name calcium, hexammoniate may be suggested as a consis-
tent nomenclature.

Incidentally, attention may be called to the fact that the presence
of a large number of nitrogen and hydrogen atoms in the compotmd
does not interfere with its properties as a metal. The presence of non-
metallic elements in a compound does not necessarily preclude the ex-
istence of metallic properties. Further examples of compounds of this
character will be discussed later.

Summary,
The existing experimental data relating to the supposed formation of
the compounds NaNH, and KNH, are examined in the light of the phase
nile and the conclusion is reached that these compounds do not exist.

668 GitBERt NEWTON LEWIS.

In a simibr manner the non-existence of the compound LiNHg is es-
tablished. In this case independent evidence is given which shows con-
clusively that in a system containing lithium and a small molecular per
cent, of ammonia a saturated solution of the metal in ammonia is formed
This result is in agreement with the phase relationships existing in the
system and demonstrates the inapplicability of the method employed by
Moissan in obtaining and identif3dng the supposed compound of lithium
and ammonia.

It is shown that calcium forms a solid compound with ammonia whose
composition is represented by the formula CaCNH,),. The optical proper-
ties of this compound are apparently identical with those of its satura-
ted solution in ammonia, and like its solution, the compound exhibits
metallic conduction.

The vapor pressures of saturated solutions of lithium and of Ca(NH,^,
in ammonia have been determined, as well as the dissociation pressures
of the compound itself. The heats of formation of the corresponding
solutions and of the compoimd from metal and gaseous ammonia are
calculated to be 8700, 10230, and 5460 calories per gram-molecule of am-
monia, respectively.

The constitution of the compotmd CaCNHj)^ is discussed. It ap-
pears that this compound is of the nature of a solvate, corresponding,
perhaps, to an ammoniated calcium ion. It is suggested that the com-
potmd be called calcium hexammoniate in order to take account of these
relations in the nomenclature.

Boston, Pebrttary 6th, 1908.

[Contributions prom raA Research Laboratory op Physical CHBMisTitY of
THE Massachusetts Institute op Technology. No. 25.]

THE OSMOTIC PRESSURE OF CONCENTRATED SOLUTIONS, AlID

THE LAWS OF THE PERFECT SOLUTION.

Bt Gilbbrt Nbwtom Lbwis.
Received March 5, 1908.

The laws of the infinitely dilute solution have been thoroughly es-
tablished. There can be no reasonable doubt as to the accuracy of Henry's
law for the vapor pressure of the solute, Raoult's law for the vapor pres-
sure of the solvent, or van't Hoff's law for the osmotic pressure, in the
case of an infinitely dilute solution. In fact if any one of these laws is
shown to be correct, the other two must follow as a direct consequence
of the laws of thermodynamics.

Unfortimately, we never work with an infinitely dilute solution, and
too little attention has been given to the question of the validity or even
the mutual compatibility of the laws just mentioned in concentrated
solutions and even in the so-called dilute solutions.

OSMOtiC PRESSURE 01^ CONCEN^TRATBD SOLUTIONS. 669

There is a well known thermodynamic relation between the osmotic
pressure of a solution and the lowering of the vapor pressure of the sol-
vent, which enables us, in every case, regardless of the concentration
of the solution, to calculate the osmotic pressure when the vapor pres-
sure lowering is known, and vice xersa. It is therefore possible to cal-
culate the osmotic pressure of a given solution, first, on the assumption
that van't Hoff's law^ is correct, and second, on the assumption that
Raoult's law* is correct. It is commonly supposed that for fairly dilute
solutions these two methods of calculating the osmotic pressure give
identical results, but this is not the case. For example, the osmotic
pressures calculated in these two ways for a normal solution of cane sugar
differ by 20 per cent, and even at the dilution of 0.005 normal the differ-
ence is still o.i per cent. It is obvious, therefore, that even in that region
to which we are accustomed to apply the term, ** dilute solution," the
law of Raoult and the law of van't Hoff are not compatible. If one is
true, the other must be false. What then shall we regard as an ideal
or perfect solution, one that obeys the law of Raoult or one that obeys
the law of van't Hoff, or shall we choose another criterion which differs
from both of these ?

Morse and Frazer,* who have recently succeeded in measuring osmotic
pressures up to 25 atmospheres by a direct method, propose to replace
the law of van't Hoff by the following equation, which gives values for
the osmotic pressure more in accord with those obtained experimentally
in the case of sugar and glucose :

„ nRT

Here V is not the volume of solution but the volume of pure solvent
in which n mols of solute are dissolved. These authors propose, there-
fore, to substitute for the system in which concentrations are expressed
in mols of solute in one liter of solution (volume normal system) another
in which concentrations are expressed in mols of solute dissolved in one
liter of pure solvent (weight normal system). In most cases the differ-
ence between these two systems is much less than it is in the case of the
two substances of high molecular weight investigated by Morse and Frazer.
Thus a weight normal solution of sugar is only 0.82 volume normal, a
difference of about 20 per cent., but in the case of methyl alcohol, am-
monia and hydrochloric acid, substances of small molecular weight se-

1 n sanRT/V, where n is the osmotic pressure, R the gas constant, T the absolute
temperature, and n is the number of mols of solute dissolved in the voltmie V of the
solutioii.

■ (^ — p)/Po ■■ «i/(*+^)» where Po is the vapor presstire of the solvent in the
pure^state, p that of the solvent from the solution, and n, is the number of mols of
solute dissolved in n mols of solvent.

•Amer. Ckem. /., 34, i (1905); 37, 324, 425, 558; 38, 175 (1907).

670 GILBERT NKWTON LEWIS.

lected at random for this calculation, the difference in concentration of
weight normal and volume normal amotmts to only 2, 4, and 2 per cent.,
respectively. It is fortunate, however, that they did study those very
substances in which the difference between the two systems is most pro-
nounced, for we are thus forced to face certain questions concerning
moderately dilute solutions which have been too often evaded.

It will be the purpose of this paper not only to find what theoretical
justification there may be for the above modification of the van*t Hoff
equation, but also to determine in general which of the various laws of
solutions may be most suitably chosen to define the perfect solution.

Before beginning this inquiry it may be well to discuss briefly another
question raised by Morse and Frazer, who write with some disparage-
ment of the methods of determining osmotic pressure which rest upon
thermodynamic calculations. Without undervaluing in any degree the
importance of direct measurements of a quantity which has played so
important a part in the development of modem chemistry as osmotic
pressure, it must nevertheless be definitely affirmed that we have at
our disposal seveml means of determining the osmotic pressure which
are readily capable of furnishing results many times as accurate as any
yet obtained by direct measurement. These methods will, therefore, be
briefly considered in the following section :

Direct and Indirect Osmotic Pressure Measurements.

The exact definition of osmotic pressure, and some of the thermo
dynamic relations in which the osmotic pressure is involved will be dis-
cussed briefly in notes at the end of this paper. There it will be shown
that the osmotic pressure of an aqueous solution may be obtained at once
from the freezing point by means of the equation

n = 12.06A — 0.021 A^ (i)

where n is the osmotic pressure in atmospheres and A is the lowering
of the freezing-point in centigrade degrees.^ From this equation the
osmotic pressure of any solution up to ten or fifteen times normal may
be obtained with an accuracy which depends only upon the precision
of the freezing-point determinations and upon the accuracy of the vahie
used for the heat of fusion of ice. - Since the error in the latter quantity

^ If we £issume that at infinite dilution van't Hoff's law holds exactly, and take
R =0.08207 liter atmospheres per degree, from the work of D. Berthelot {Z,EUklro-
ckem.f 10, 621, 1904), then we find from equation (i) that the molecular lowenfl^
of the freezing-point of any aqueous solution at infinite dilution is 1.858**, which difien
materially from the value commonly used, namely, 1.85. The latter valoe istised
by Morse and Frazer, but they should use the value 1.843, for they do not employ
the international atomic weights but those based on hydrogen as unity. The md
is therefore reckoned in the latter system, and not in the customary one, in Tabks
I and II where Morse and Prazer's data are used.

OSMOTIC PRESSURE OF CONCENTRATED SOLUTIONS. 67 1

is probably not more than about o.i per cent., it is obvious that except
for the most dihite solutions osmotic pressures may be foimd in this way
with an accuracy which is more than ten times as great as Morse and
Fiazer claim for their direct measurements. It is interesting to com-
pare the osmotic pressures obtained by Morse and Frazer with those
calculated by equation (i) from the freezing-point measurement of the
same authors. This comparison is made in Tables I and II. The first

Tabls I. — Cane Sugar.

. _ _ Per cent.

M. A. Ilobs. ncalc. difference.

O.I 0.195 2.44 2.35 4

0.2 0.392 4.80 4.73 I

0.3 0.585 7.16 7.05 2

0.4 0.784 9.40 9.45 I

0.5 0.985 II. 8 II. 8 o

0.6 I. 19 14.2 14.3 I

0.7 1.39 16.8 16.7 o

0.8 1.62 19.3 19.5 I

0.9 1.83 22.1 22.0 o

z.o 2.07 24.8 24.9 o

0.867 1.77 21.3

1 . 25 2 . 68 32.2

1-54 342 41.0

1.63 3.63 43.5

2.10 4.88 58.4

Tablb II. — GtrUCOSE.

_ _ Per cent.

M. A. nobs. ncalc. difference.

0.1 0.192 2.40 2.32 3

0.2 0.386 4.65 4.65 O

0.3 0.576 7.01 6.94 I

0.4 0.762 9.30 9.18 I

0.5 0.952 II. 6 II. 5 I

0.6 I. 15 14.0 13.8 I

0.7 1.34 16.4 16. I 2

0.8 1.53 18.8 18.4 2

0.9 1.72 21.2 20.7 2

I.O 1.92 23.6 23.1 2

column gives M, the number of mols of solute in one liter of water, the
second the lowering of the freezing-point, the third the osmotic pres-
sure directly measured, the fourth that calculated from A, and the fifth
gives in round numbers the percentage difference between the observed
and calculated values. In Table I, I have also given (below the line)
the osmotic pressure of cane sugar solutions calculated from the freezing-
point measurements of Ewan.* These seem in perfect accord with the
values of Morse and Frazer and extend to higher concentrations. It
* Z. fhysih, Chem., 31, 22 (1899).

672 GIlrBERt NEWTON LEWIS.

is to be noted that each calculated value is obtained for the temperature
at which the solution in question freezes, while the observed \'ahies were
found at a few tenths of a degree above zero, but the correction for this
small temperature difference is too small in comparison with the experi-
mental errors to be considered.

It is apparent that the observed and calculated values for the osmotic
pressure agree vdthin the limits of error of the former. The tables in-
dicate, moreover, that the experiments with glucose were somewhat
less reliable than those with cane sugar.

Since, therefore, freezing-point measurements offer a simple and ex-
act means of determining the osmotic pressure at the freezing-point,
it is possible from them to determine the osmotic pressure at other tem-
peratures, if we know its temperature coefficient. Morse and Frazer
have considered it impossible to predict the value of this coefficient,
but they have overlooked the simple thermodynamic equation, which
may be derived immediately from the familiar energy equation of Hehn-
holtz, namely,

n-9=Tg. (2)

where n is the osmotic pressure, T the absolute temperature and q is
the heat of dilution, that is, the heat evolved when one cc. of solvent
is added to a large quantity of the solution. This quantity q is known
for a large number of solutions and in any case may be very easily de-
termined. For cane sugar we have very accurate knowledge of this
quantity for one temperature, 15®, from the independent but entiiclr
accordant work of von Stackelberg* and Ewan.* According to their meas-
urements in the case of a weight normal solution q is equal to 0.12 cal.
or 5 cc.-atmos. Substituting the latter value in equation (2) and caffing
n at 15° approximately 24 atmos., which according to the experiments
of Morse and Frazer cannot be far wrong, we find

dn ^ 24 — 5

^ "" 273 + 15 '
or about 0.07 atmos. per degree. In other words, while the osmotic
pressure of an ideal solution at 15° changes 0.35 per cent, per degree
the normal sugar solution changes only 0.27 per cent, per degree. Un-
fortunately we do not know the heat of dilution of sugar solutions at
lower temperatures, but since in other cases von Stackelberg has showi
that it increases with decreasing temperature, it is probable that it does
in this case also. The temperature coefficient of osmotic pressure
therefore probably become smaller at lower temperatures and may even
become negative (when ^ >n), which would explain the surprising fact

» Z. pkysik. Chem,, 26, 533 (1898).

OSMOTIC PRESSURE O]? CONCENTRATED SOLUTIONS. 673

discovered by Morse and Frazer that the osmotic pressure of cane sugar
is about the same at o® as it is at 25®.

Since the heat of dilution may be very readily measured at any tem-
perature, we have by its means a remarkably simple method of deter-
mining the osmotic pressure at any temperature, if it is known at one.

For obtaining the osmotic pressure of a solution at any temperature
there is another perfectly general indirect method which has been fre-
quently employed* and recently has been improved to such a point that
it rivals in accuracy the freezing-point method.^ It depends upon the
thermod3mamic relation between the vapor pressure from a solution
and the osmotic pressure, which may be expresssed in the equation*

n_!«n. = ^^^ (3)

Here n is again the osmotic pressure, a. is the coefficient of compressibility
of the solvent, V^ is its molecular volume, In stands for natural logarithm
and pQ and p are respectively the vapor pressure of the solvent in the pure*
state and in the solution. Several applications of this equation will be
made in the following section.

The Law of Ideal Solutions.

What we shall call a perfect or ideal solution is somewhat a matter
of choice. We might define as an ideal solution one which obeys the
law of van't Hoff, or the modified form of this law proposed by Morse
and Frazer, or the law of Raoult, or the law of Henry. These laws are
essentially identical for the infinitely dilute solution, but for a solution
of finite concentration we are at liberty to choose one but not all of these
laws to define the ideal solution. No one of them is true for every solu-
tion at every concentration, and we must therefore choose that one which
holds most nearly for the greatest number of substances over the widest
limits of concentration.

I shall attempt to show that the most fundamental law of solutions
and the one by which the perfect solution is best defined is the following
modification of the law of Raoult. At constant pressure and tempera-
ture the activity^ of the solvent in a perfect solution is proportional to its
mol fraction. That is,

i = io^, (4)

where ( is the activity of the solvent in the solution, (^ the activity of
the pure solvent, and N, the mol fraction, is the number of mols of sol-

* See for example, Noyes and Abbott, Z. physik, Chem., 33, 56 (1897).
■ See Smits, Z. physik, Chem., 51, 33 (1905).

* For the development of this equation, see note 3, at the end of this paper.

* For a definition of the term activity see Lewis, "Outlines of a New System of
Thermodynamic Chemistry," Proc. Amer. Acad., 43,239(1907); and Z. physik, Chem.,
^if 129 (1907); C. A.f 1908, 61;,

674 GILBERT NEWTON I^WIS.

vent in one mol altogether of solvent and solute. Since, however, the
conception of activity is new, and since if the vapor of the solvent obeys
the gas law the activity is proportional to the vapor pressure, we may
with sufficient exactness for our present purposes, substitute the vapor
pressure of the solvent for its activity and write

P = Po^'f (5)

that is, in a perfect solution the vapor pressure of the solvent is propor-
tional to its mol fraction.^ Thus in a solution containing o.i mol solute
to 0.9 mol solvent, N = o .9 and the vapor pressure of the solvent should
be nine-tenths of its vapor pressure in the pure state, or if the solution
contains 0.25 mol solute to 0.75 mol solvent p should be 0.75 p^. This
is simply a statement of Raoult's law in its simplest form.*

There are no cases in which the law of van't Hoff or the modified fonn
of this law proposed by Morse and Frazer have been shown to hold at
concentrations higher than normal. (In a normal solution in water
the mol fraction of the solute is about 0.02.)

Indeed, at very high concentrations van't Hoff's law cannot hoW, for
the osmotic pressure of a solution approaches infinity as the percentage
of solvent approaches zero, while the osmotic pressure calculated from
the van't Hoff equation never exceeds a few hundred atmospheres even
when we approach the condition of pure solute. On the other hand, it
will be shown presently that the law proposed by Morse and Frazer ordi-
narily gives, at higher concentrations, osmotic pressures far higher than
those which actually exist. But often the law of Raoult (and the modi-
fied law of Henry) has been shovsm to hold at aU concentratUms from
o per cent, to 100 per cent, of solute, and while in many other cases this

^ The point of view here adopted is practically identical with that which for
several years has been advocated by J. J. van Laar in numerous publications.

' It is important to note that equation (4) leads us immediately to a simple eqoi-
tion for the activity or the vapor pressure of the solute. In the paper previously
referred to I have proved the f ollovdng exact equation for the change in the acti^v
of each component of a binary mixtiure with change of composition, namely, .

where N, is the mol fraction and l| the activity of one constituent which we win cbS
the solute, and N and { are the corresponding terms for the other constituent wfaicfa
we will call the solvent. Now, when equation (4) is true, dln^ = dlnS. Siibstita-
ting in the above equation and noting that by definition N, = i — N we find

N,(tfn^i + <iN = o, or,
<fln5j = (flnN„or

where K is a constant. We see therefore that in a perfect solution it is also tmethat
the activity of the solute is proportional to its mol fraction. If we substitute ^the
vapor pressure of the solute, for ^„ ^, = KNj, which is, in a slightly modified fono,
the law of Henry. In other words, if both vapors obey the gas law, thelawofHeorT
may be derived thermod3mamically from the law of Raoult and must hold if tbatbv
does.

OSMOTIC PRESSURE OP CONCENTRATED SOLUTIONS. 675

law does not hold, the greatest deviations are always found in those cases
in which we have reason to believe that the solvent and the solute form
complex compounds either with themselves or with each other.

Many illustrations might be given to show the remarkable scope of
Raoult's law. I will choose a binary mixture which has been studied
more carefully over a wide range of concentration than any other, namely,
benzene and ethylene chloride. The vapor pressures are taken from
the excellent paper of Zawidski.* We will call benzene the solvent and
ethylene chloride the solute. In Table III, in which the data marked
by Zawidski as questionable are omitted, the first column gives the num-
ber of grams of solute to one gram of solvent, the second gives the par-
tial vapor pressure of the solvent at 50°, and the third gives the molecu-
lar weight of ethylene chloride calculated from the vapor pressures by
Raoult's law. The calculated molecular weights, are constant, even up to
the highest concentration, where the solute constitutes over 90 per cent,
of the solution. The average of these calculated molecular weights is
99.1 while the actual molecular weight of ethylene chloride is 99.0. (We
have therefore every ground for believing that also in the pure state ethyl-
ene chloride exists in the form of simple molecules.)

Grams CSH4CIB
to X sf. C«H«.

0.0

0.525
0.904
1.39
2-43
3.89
14.54

Tablk III.

M. W.

p. of CaHfl.

CtBUCl^

268.0

• • • ■

189.8

99-4

156.0

98.2

127.8

98.7

92.4

99-3

65.9

99.0

21.8

100. 0

Average, 99.1

Theoretical, 99.0

If then we define a perfect solution as one which obeys Raoult's law,'

it is interesting to find what the law is connecting osmotic pressure and

concentration in a perfect solution. This law, which is less simple than

either the law of van't Hoff or that of Morse and Frazer, may be derived

directly from equations (3) and (5), and is

T RT

n—tan* -_^if^N (6)

2 Vo

* Z. physik. Chem,, 35, 129 (1900).

' Strictly speaking, we define a perfect solution as one which obeys equation (4)
n4tber than equation (5), but the more precise method which employs the activity
instead of the vapor pressure leads to exactly the same equation for the osmostic pres-
sure as we shall derive here.

676 GILBERT NEWTON LEWIS.

I UT

n-i«n» ^fo»(,_N). (7)

where V^ is the molecular volume and a the compressibility of the sol-
vent.

In Table IV the osmotic pressures of cane sugar solutions are calcu-
lated from equation (7). The first column gives the weight nonnal
concentration; the second, the volume normal; the third, N^, the mol
fraction of solute ; the fourth, the osmotic pressures calculated by the
van't HofI equation ; the fifth, those calculated by the equation of Morse
and Frazer; the sixth, those calculated by equation (7); the seventh,
Morse and Prazer's observed values.

Tabus IV.

Cone

Cone.

Mol fraction

wt. nomi.

▼ol. norm.

ofaelute.

•

van't Hoff. Morae and Fraaer.

X<ewia.

ObKTved.

O.I

0.098

0.00180

2.34

2.41

2.40

0.2

0.192

0.00358

4- 58

4.81

4.80

4.74

0.3

0.282

0.00537

6.73

7.23

7.21

7.23

0.4

0.369

0.00715

8.81

9.64

9.60

9.67

0.5

0.452

0.00892

10.8

12.0

12. 1

0.6

0.532

0.0107

12.7

14.5

14.4

144

0.7

0.610

0.0x24

145

16.8

16.7

16.9

0.8

0.684

0.0142

16.3

19.3

19.2

19.4

0.9

0.756

0.0159

18. 1

21.7

21.5

21.8

I.O

0.825

0.0177

197

24.1

239

24.5

While the values given by the equation of van't HoflF differ from those
observed by nearly 25 per cent, at the higher concentrations, it will he
seen that the pressures given in the fifth and sixth columns agree through-
out with the observed values, within the limits of experimental error,
and differ from each other by only one per cent, even at normal concen-
tration.

This agreement between the osmotic pressures calculated from the
equation of Morse and Frazer and those calculated by equation (7) will
always be found at moderate concentrations, as the following considera-
tions show. The second term in equation (7), except at the very high-
est concentrations, is comparatively insignificant, amounting usually
to only a few per cent, of the value of n even when the osmotic pressure
is as high as a thousand atmospheres. At moderate concentratioDS
we may, therefore, neglect this term and write equation (7) in the fonn,

n ^fo,(i-N,). (8)

Now the equation of Morse and Frazer may be written in the form

„ RT/ Ni \ ,.

OSMOTIC PRBSSURB OF CONCENTRATED SOLUTIONS. 677

for * is the number of mols of solute to one mol of solvent and V^

I — Nj o

is the volume of one mol of pure solvent.

Equation (8) developed in series gives

°''W(^*"^2^*' + i^>*+ ••••)'

and similarly equation (9) gives,

n--^(N,+N,>+N,»+....).

These equations differ only in the higher powers of Nj and therefore
give identical results at such concentrations that the terms containing
these higher powers are negligible. When the mol fraction of the solute
is 0.02 the values of n calculated from these equations differ by one per
cent. For all more dilute solutions, therefore, the osmotic pressure of
a perfect solution may be calculated within one per cent, from the equa-
tion of Morse and Frazer.

At higher concentrations, however, the difference between these two
equations becomes very great, as is shown in Tables V and VI. Table
V deals with solutions of ethylene chloride in benzene, and simply re-
states in a new way the facts brought out in Table III. Table VI con-
tains data on solutions of propylene bromide in ethylene bromide. In
both tables the first column gives the mol fraction of solute; the second,
the partial yapor pressure of the solvent, taken from the work of Zawid-
ski;* the third, the osmotic pressure calculated by the van't Hoff equa-
tion; the fourth, that calculated by the equation of Morse and Frazer;
the fifth, that calculated by equation (7), while the last column gives
the actual osmotic pressure obtained thermodynamically from the vapor
pressures by means of equation (3). The molecular volumes of benzene
and ethylene chloride at 50® are taken, respectively, as 0.092 and 0.082
liter, and the coefficient of compressibility of benzene as o.oooi. The
molecular volumes of ethylene and propylene bromides at 85** are taken,
respectively, as 0.092 and 0.113 liter, and the coefficient of compressi-
bility of ethylene bromide as 0.000,06.

Table V.
CH^Cl, in CJHe at 50^.

Nj.

>t.

▼aii*t Hoff.

Morse and Praser.

Lewis.

Pound.

0.0

268.0

• •

• *

• ■

. .

0.293

189.8

120

lOI

100

0.416

156.0

128

205

157

158

0.522

127.8

160

315

215

217

0.657

92.4

196

549

313

310

0.754

65.9

223

880

413

406

0.920

21.8

268

3290

743

735

' Omitting the values which the author marks as questionable.

678 GILBERT NEWTON LEWIS.

Tablb VI.

CjH.Br, in CJH

:,Br,

at 85®.

Ni.

▼an't Hoff.

Morw and Praser.

I,ewls.

Pound,

0.0

172.6

• .

• •

■ •

• a

0.147

145- 1

0.222

132.2

0.298

121. 1

136

114

0.412

lOI.I

121

224

171

173

0.526

81.9

150

351

241

240

0.620

64.0

173

520

313

319

0.720

48.0

198

820

412

414

0.800

34-3

218

1280

522

525

0.860

23s

232

i960

640

649

0.915

13.8

241

3440

806

827

We see from these tables how closely in these two cases the actual
osmotic pressures agree with those calculated by equation (7), and how
far from the truth are the pressures calculated both by the van't HofF
equation and by that of Morse and Frazer. These two solutions are,
according to our definition, perfect solutions, within the limits of experi-
mental error, for all concentrations from o to over 90 per cent, of sohite.
Since, moreover, these cases are not unique but have been chosen out of
a large number of similar cases merely because of the greater experimen-
tal care with which they have been investigated, it is to be presumed
that even those solutions which are not perfect at all concentration
will, on the average, follow the law expressed in equation (7) to higher
concentrations than they will the law of van't Hoff or that of Morse and
Frazer.

In view of the experiments of Morse and Frazer, it has recently been
proposed* that in the ordinary equations of chemical equilibrium the con-
centrations expressed in the volume normal system should be replaced
by those expressed in the weight normal system. This is imdoubtedly
an improvement, but the equations thus obtained are not entirely cor-
rect, even when all the substances concerned are present as perfect solu-
tions.

In order to find an exact equation, let us consider a reaction occurring
as follows :

where x^ mols of Xj combine with X2 mols of Xj to form x^ mols of X^
etc. It is readily seen from the considerations advanced in this paper
and from the thermodjmamic laws of chemical equilibrium,' that the
general equation of chemical equilibrium, regardless of the concentra-

*Walden: Z. physik. Chem., 58, 500. This paper also contains a ktterfrom
van't Hoff on this subject.

' Lewis: Loc. cii., equation (XXIII).

OSMOTIC PRBSSURK OF CONCENTRATED SOLUTIONS. 679

dons of the reacting substances, provided that they are all present as
perfect solutions, is as follows:

^'sijjI^' ' ] -K (a constant), (lo)

where Np N,, etc., are the respective mol fractions of X^, X,, etc.

This, then, is the form which the mass law assumes when the substances
concerned form perfect but not infinitely dilute solutions, and for such
cases it is rigorously exact.

Note z.

If a solution and the pure solvent are separated by a semipermeable
membrane the solvent will flow through the membrane into the solution,
where its escaping tendency is less. The only way of preventing this
flow is to make the escaping tendency of the solvent the same on both
sides of the membrane. There are two simple ways of accomplishing
this, (i) to increase the pressure on the solution until the escaping ten-
dency of the solvent in the solution is raised to equal that of the solvent
in the pure state, (2) to diminish the pressure on the pure solvent until
its escaping tendency is lowered to equal that of the solvent in the solu-
tion.

The osmotic pressure may therefore be defined in two ways, (i) as
usually defined it is the increase in the pressure on the solution neces-
sary to bring the latter into equilibrium with the solvent; (2) Noyes,'
however, prefers to define the osmotic pressure as the diminution in the
pressure on the solvent necessary to bring it into equilibrium with the
solution. Neither of these definitions is entirely free from objections,
but since the second one permits a much simpler mathematical treat-
ment than the first, it has been adopted throughout this paper. The
osmotic 'pressures defioied in these two ways differ only when there is a
total change of volume when a small quantity of solvent is added to a
solution. There is no such volume change in the case of sugar and glu--
cose as shown by the experiments of Morse and Frazer and of Ewan.
We have been justified, therefore, in applying our equations, which in-
volve the osmotic pressure according to the second definition, to the
results of Morse and Frazer, who worked with the osmotic pressure of
the first definition.

Note 2.

The exact equation coimecting osmotic pressure and freezing-point
may be found as follows: Let us consider an aqueous solution in equilib-
rium with ice at the temperature T and the pressure p, and also in equilib-
rium with these, through a semipermeable membrane, pure water at
the same temperature and at the pressure p — n, n obviously being the
osmotic pressure. Now if the temperature changes by (TT and the pres-
^ Z. physik. Chem^ 35, 707 (1900).

68o GILBERT NEWTON LEWIS.

sure on the solution and ice remains equal to P, that on the water must
be changed in order to maintain equilibrium. The necessary change
in pressure we will call du. Since the water and ice are in equilibrium
at the beginning, the activity ^ of the water and the activity ^ of the
ice must be equal, and these, moreover, must remain constant as the
temperature changes. Hence,

and

or,

Now the change in the activity of the ice is due only to the change of
temperature, that is,

but the activity of the water is changed both by the change in tem-
perature and the change in pressure, that is,

Equating the last members of these two equations, we have

Now substituting for the partial differentials their values from the
fundamental thermodynamic equations^ and combining the first two
terms gives

dn —L

wher^ L is the heat absorbed in the fusion of one gram of ice and v is
the volume of one gram of water. For T we may substitute 273.1 — A,
where A is the lowering of the freezing-point below the centigrade zero,
whence

dA t;(273.i — A)*
In order to integrate this equation, L and v must be known as functions
of A. According to a well-known principle the change of L with A is
given by the equation

L = Lt— CA,
where L© i? the heat of fusion at 0° C. and C is the difference between
the specific heats of water and ice. According to the best available
data, C is about 0.5 if our imit of energy is the calorie, or 2 1 if our unit
of energy is the cc.-atmos. The value of ho obtained in the very ac-

^ Lewis: Loc. cU,, equations V and VIII. It is of course to be noted that by
definition n is a negative pressure.

OSMOTIC PRESSURE OF CONCENTRATED SOLUTIONS. 68 1

curate experiments of Smith* was 334.2 joules, assuming the electromo-
tive force of the Clark cell to be 1.434 V at 15°. Since this value entered
twice into the calculation of Smith, if we adopt for the Clark cell the
value now accepted of 1.433 V, the value of Lo must be lowered by 2 parts
in 1434 and becomes 333.7 joules, or 3294 cc.-atmos.*

We may therefore write,

L = 3294 — 21 A.
Strictly speaking, L is a ftmction of the pressure also, but the pressure
eflFect may easily be shown to be too small to be considered in the present
cakuhtion.

The volume of a gram of water also depends upon both temperature
and pressure. We shall see that one degree lowering of the freezing-
point corresponds to about 12 atmos. change in the osmotic pressure.
Hence from the known coefficients of thermal expansion and compressi-
bility we find that the value of v may be expressed very closely by the
Unear equation,

V = i.ooo H- 0.0008A.

Substituting now in the above equation the values of L and v, and per-
forming the multiplications and divisions indicated, we obtain 3- as a
series function of A, namely,

-^=■12.06 — 0.0414A — o.oooo9A'-|-

Except for very high values of A and A^ term and all the higher terms
are negligible. Dropping these terms, therefore, and integrating, we
have

n — 1 2.06A — o. 2 1 A*.

By this equation the osmotic pressure corresponding to any freezing-
point lowering may be calculated immediately and, if the experimental
data used are as accurate as they appear, the error of the calculation can
hardly exceed a few tenths of a per cent, even at osmotic pressures of
several hundred atmospheres.

Note 3,

The connection between the osmotic pressure and the vapor pressure
of the solvent from a given solution is obtained as follows : From the
fundamental thermodynamic equation for the change of the activity
of a substance with the pressure,' we have

din — y
dn "" RT '

* Phys. Rev., X7, 231 (1903).

'Guttmaim (J. Phys. Chem,, xi, 279 (1907)) has made a similar recalculation
of Smith's value but applied only one-half of the correction applied above.

* Lewis: Loc. cit, equation V.

682 GII^BERT NBWTON I.BWIS.

where V is the molecular volume and k is the activity of the ptue solvent.
When the vapor of the solvent behaves like a perfect gas whose pressure
is /> we may write

Hence, in such a cate,

dlnp — V
dn"RT*

V may be regarded as constant for small values of n but at high pressures
we must consider the compressibility of the solvent. If the coefficient of
compressibility of the solvent is denoted by. a, and the volume when the
osmotic pressure is zero by V,,,

V = V^(i-an).

Substituting this value of V in the above equation and integrating, we
have

where />^ is the vapor pressure of the pure solvent, p that of the solvent
in the solution of osmotic pressure n. This equation is derived for the
case that the vapor of the solvent obeys Boyle's law. In any other case
a more complicated formula must be used.

Summary*

The simple laws of the infinitely dilute . solution become mutually
incompatible in solutions of finite concentration. It is therefore neces-
sary to choose one law to serve as a criterion of the perfect solution. The
only law of dilute solutions which ever holds in concentrated solutions
is the law of Raoult. This law is stated in a slightly modified form and
a perfect solution is defined as one which obe3rs this law. A number of
solutions are mentioned which behave as perfect solutions over the whok
range of concentrations, from o per cent, to loo per cent, solute.

The indirect methods of determining osmotic pressure are discussed
and an exact relation between the osmotic pressure and the freezing-
point lowering of an aqueous solution is obtained. It is also pointed out
that the osmotic pressure at one temperature may be obtained from that
at any other when the heat of dilution is known.

Adopting Raoult's law in its modified form as the characteristic law
of the perfect solution, it is possible vrith the aid of thermodynamics
alone to obtain an equation connecting the osmotic pressure and con-
centration of a perfect solution. The equation thus obtained permits
the exact calculation of osmotic pressures in perfect solutions, up to looo
atmos. In comparatively dilute solutions the pressures thus obtained
are substantially identical with those given by the equation of van't

THE INDESTRUCTIBII<ITY OF MATTER, ETC. 683

HofF, as modified by Morse and Frazer, but at high concentrations the
divergence between the two equations is very great.

An exact form is obtained for the mass law. in concentrated perfect
sohitions.

Boston, March 3, 1908.

[CONTRIBUnON PROM ROGERS'S LABORATORY OP PHYSICS OP THS MaSSACHUSSTTS

Institute op Tbchnoi/x>y.]

THE mDESTRUCTIBILITY OF MATTER AND THE ABSENCE OF
EXACT RELATIONS AMONG THE ATOMIC WEIGHTS.

Bt Danibl p. Comstock.
Received March 9» 1908.

The two chief reasons briefly stated for believing in the evolution of
the elements one from another are, first, that some such process is un-
doubtedly taking place in the case of the radioactive substances, while
we are being forced toward the conclusion that all the elements are radio-
active to some degree; and second, that in the hottest stars only two
known elements occur, namely, hydrogen and helium, while as we pass
successively to cooler and cooler stars the other elements gradually make
their appearance in a more or less orderly manner. Apparently this
can only mean that at these transcendental temperatures the forces
due to molecular or atomic impact are comparable with the interatomic
forces involved in the breaking up of one element to form another, and
hence the combination necessary for the formation of the heavier ele-
ments can take place only after the temperature has sufficiently dropped.

There is one seemingly fatal objection, however, to any very simple
statement of the evolutionary theory and. this objection has not been
sufficiently emphasized. The difficulty is this, that so far as we know there
are no exact simple relations between the various atomic weights, whereas
if we are to assume, as the simplest form of the evolutionary theory does,
that the lighter elements come from an atomic disintegration of the
heavier ones, or vice versa, it is evident that simple additive relations
must exist.

As we know, many simple, additive relations do exist, but they are
approximate, not exact, and the deviations from exactness, though small,
are larger than we can explain from error in atomic weight determina-
tions.

On the basis of common conceptions, therefore, the evidence seems
contradictory, certain facts seeming to require the simple evolutionary
idea, while another fact, the inexactness referred to above seeming to
deny it.

I wish to show that on the basis of the electrical constitution of mat-
ter this inexactness is not only to be explained but it is to be expected.

684 DANIEL F. COMSTOCK.

By the electrical constitution of matter is meant merely that concep-
tion, which has grown in favor among physicists of late, which considers
the atom to consist wholly or in part of a group of electric charges. Some
of these make themselves evident in an ionized electrolyte or in a gas,
but in general the fact that the two electricities, positive and negative,
are present in equal amounts makes the atom neutral as regards action
on an outside point.

Now since moving charges always act like currents, and hence nmst
set up magnetic fields, it must follow that when an apparently uncharged
atom is set in motion there will be set up inside of it a magnetic field,
which depends for its strength and distribution on the number, position,
and magnitude of whatever charges the atom may be supposed to con-
tain. It will therefore require more energy to set the atom in motion
because of the necessity of building up this field, and hence the atom will
have an added inertia, i. e., an added mass because of these charges.

Now it has been shown mathematically by the author (Phil, Mag.,
Jan., 1908), without making any assumption as regards the structure
of the atom, that this added mass due to the electric charges is strictly pro-
portional to the total electrical energy contained in the atom.

By electrical energy is meant the energy which we know must exist
in space wherever there are electric lines of force. Two charges, one
positive and one negative, attract each other, and if these are separated
to a distance, work must be done against their mutual attraction. Since
the charges themselves have changed in no way, the energy put in must
exist in the surrounding space in a form generally known as ** strain in
the ether." Magnetic lines of force also correspond to energy stored in
space and the statement proved in the paper mentioned above is that
the added mass which an atom possesses because of the electric charges
which it is supposed to contain depends only on the total electric and
magnetic energy which the atom contains and is directly proportional
to this energy.

The expression found may be written

M = ^|
3 V*

where M is this "electric mass," V is the velocity of light, and E is the

total contained electromagnetic energy of the kind described above.

The amount of energy which corresponds to a gram mass is enormous.
It is readily calculated from the formula, and is found to be 7 X 10^
ergs, approximately, so we may make the statement that, on the
present bases, the inertia of a gram mass is dtie to the existence within H
of 7 X 10^^ ergs of confined energy.

Now, when the atom of an element breaks up the process is a violent
one and a large quantity of energy is lost, i. e., goes ultimately into heat.

THE INDBSTRUCTIBDUITY OF MATTER, BTC. 685

Hence, if this is electric energy lost — and on the basis of the electrical
conception it is electrical energy — ^it follows from the above that there
must have been a loss in mass accompanying the atomic disintegration be-
cause of the energy lost. Hence, taking the simple case of an atom of
A, splitting up violently and forming atoms of the elements B and C,
we would find that the mass of B and C, taken together, would be a lit-
tle less than the mass of A, from which they came, the difference corre-
sponding to the loss of mass accompanying the loss in energy.

Now the loss of energy when an atom of radium breaks up is known,
and calculation shows that this would give a loss in mass which would
be of about the same size as the deviations in atomic mass or weight so
common in the table.

In an important paper ^ Rydberg has shown that the atomic weights of
the first twenty-seven elements of the periodic system approximate to
whole numbers very much more closely than chance could bring about.
He has also shown that the atomic weights of these elements are best
considered as the sum of two parts (N + D), where N is an integer and
D is a fraction, in general positive and smaller than unity. If M is the
number of the element in the system (called by Rydberg the ** Ordnungs-
zahl"), then N is equal to 2M for the elements of even valence and 2M -f i
for the elements of odd valence. Below is given a table showing the
various quantities. I have used, however, the International Atomic
Weight values for 1907 instead of those Rydberg used.

N. N.

Sign.

aM.

aM + i.

Atomic
, weight.

7.03
91

0.03
I.I

II. 0

12.00

7
8

14.01 — 0.99
16.00 0

19.0

20.0

Na...^.
Mg

II
12

23 05
24.36

0.05
0.36

27.1
28.4

O.I
0.4

Ti..

v..

Cr.

Mn.

Fe.

2M.

2M4-1.

weight.

31.0

32.06

0.06

35.45

0.45

39.9

3.9

39.15

0.15

40.1

O.I

44.1

I.I

• • •

48.1

O.I

51.2

0.2

52.1

0. 1

55.0

0.0

55.9 -

—O.I

Many besides Rydberg have, of course, noticed and studied the curious
deviations in the table of atomic weights and Rydberg's work is men-
tioned merely because it seems unusually explicit.

The orderly arrangement of the series is striking. It will be noticed

» Z. anorg, Chem., 14, 66 (1897).

686 DANIISI^ ^. COMSTOCK.

that in three cases only are the D's greater than unity and only in two
cases are they negative.

Rydberg points out that although the heavier elements do not con-
form well to this scheme, i. e., do not in general give the small fractional,
values of (D) noticed above, yet this is in reality no valid objecticm,
for the numerical values of the weights of heavier elements depend much
more on the value of the arbitrary unit chosen than do those of the lighter
weight elements, and hence, they can have little influence one way or the
other in estimating the validity of the curious relations he sets forth.

The whole question is, of course, whether these differences represent
real physical deviations from something or whether they are merely
mathematical remainders. Rydberg certainly believes them to repre-
sent physical realities, and considering the before-mentioned overwhelm-
ing improbability that the approximation of the atomic weights to whole
numbers is due to chance, we can hardly doubt that he is right.

The question will doubtless be asked why is there no loss of mass found
when a violent chemical reaction takes place and energy is lost. The
expression for loss of mass in terms of loss of energy, when written in the
differential form (substituting V = 3.io**) is

AM«^ IO-* AE,
27

and if we substitute the heat of reaction of any known chemical reaction
for AE, we find that AM is too small to be detected even by the delicate
experiments of Landolt.
For the reaction

O, -f 2H, - 2H,0,

AM for one gram molecule is about lo""** of a gram.

In radioactive changes, however, the energy is enormously greater
and hence is to be detected as before mentioned.

If we consider the whole atom is electric instead of only part, as we
have considered above, a conception which is by no means artificial snoe
it has been proven that the mass of an electron is entirely due to its charge,
we reach the interesting conclusion that on this basis the "Indestructi-
bility of Matter" is only a corollary of the "Conservation of Energy,"
for if the atoms are to be considered as electrical systems it follows that
the law of the Conservation of Mass, which is essentially the same as the
Indestructibility of Matter, must be approximately true, though not
strictly true. It must be approximately true because no known chem-
ical or physical change, with the exception of radioactivity, involves any-
thing but a relatively minute liberation or absorption of energy, but the
law cannot be absolutely exact, for even this minute loss of energy must
involve, on the present basis, its corresponding loss of mass. Thus even

THB INDBSTRUCTIBBWTY OF MATTER, «TC. 687

the cooling of a hot body must involve a certain diminution of mass,
though this, of course, is extraordinarily small, much smaller even than
the loss due to most chemical reactions.

It would be interesting to search for other evidence of the existence
of this vast store of energy in matter. Such evidence would involve
effects, which would be approximately proportional to the density. Such
an effect is found in the case of the absorption of the ^-rays of radium
when passing through different kinds of matter. Here the absorption
is proportional to the density over a very wide range. Strutt has shown
that in the case of fourteen substances whose absorption he measured,
where the density varied from 0.007 in the case of sulphur dioxide, to
21.5 in the case of platinum, the ratio of the absorption to the density
has an avemge deviation of only 20 per cent.

From a ph3rsical point of view, also, it is interesting to note that, since
gravitation has always been found to be strictly proportional to mass,
it follows from the above that the electrical structure of matter requires
that gravitation should be proportional to the energy contained in
the gravitating bodies and to this energy alone. Thus gravitation must
be considered on the present basis as existing between quantities of
confined, electromagnetic energy and not between "masses'* in any other
sense. More knowledge bearing on the electrical theory of matter would
therefore throw considerable light on the outstanding mystery of gravi-
tation.

Summary,

We have seen that, assuming the electrical theory of matter — ^the
theory, that is, which considers the atoms as systems composed of elec-
tric charges — ^it follows that the mass of a piece of matter is determined
solely by the amount of electromagnetic energy which it contains and
is proportional to this amount. The energy in ergs which endows one
gram with its mass is equal to three-fourths of the square of the velocity
of light, or, in round numbers, twenty million horse-power-hours, or the
energy corresponding to the work of one thousand horses working for
two years. This enormous amount is by no means impossible since the
amount of energy which the radioactive substances are known to lose
in passing to lower forms is quite appreciable in comparison.

On the electrical theory of matter it therefore follows that the chief
property of matter, the property which gives us a quantitative defini-
tion, namely its inertia or mass, is really a property of the energy stored
up within the structure which defines the space relations of a piece of
matter and is not a property of the structure itself. Thus it follows
that the law of the '* Conservation of Mass," which we have here reason
to beHeve is only approximate, is in reality a corollary to the law of the

688 HERBERT N. MCCOY.

''Conservation of Energy" and thus this latter law and the "Indestnic-
tibility of Matter" are closely akin.

From the above it follows that on the present basis any loss of energy
must involve a decrease in mass. Thus when a chemical reaction which
liberates heat has taken place, when a body is cooled or when by the
process of radioactivity one substance loses energy and is transformed
into another substance, there must be a decrease in the mass of the whok
and hence also a decrease in weight. In the first two cases mentioned,
however, the change is too small to be detected but in the last case the
change should be appreciable, and we have a ready explanation of the
irregularities which occur in the table of atomic weights.

Finally, it is pointed out that since gravity is proportional to mass
it would appear that gravitation must be considered as acting between
quantities of confined energy and not between "masses" in any other
sense.

[Contribution prom tbb Kbnt Chemical Laboratory, Univbrsry op Chicago.]

TWO NEW METHODS FOR THE DETERMmATION OF THE
SECONDARY IONIZATION CONSTANTS OF DIBASIC ACIDS.

By Hbrbb&t N. McCot.
Received March 3, 1908.

The ionization of a dibasic acid, H^X, takes place in two stages, repre-
sented by the following equations :

H^HX ^ k^H^X; (i)

//•X = kJFIX; (2)

where k^ and &, are the ionizaton constants, and where the formulae
represent the molar or ionic concentrations of the corresponding sub-
stances. With the exception of a few moderately strbng acids, like ox-
alic, all organic dibasic acids are found by conductivity measurements
to dissociate essentially according to equation (i), in solutions more
concentrated than milli-normal ; they thus behave like monobasic adds.^
The second constant, ib,, is always much smaller than the first. Whik
the secondary ionization produces but a negligible effect in solutions
of the free acid, ip solutions of the acid salts its effect is of great impor-
tance.

When an acid salt, like NaHX, is dissolved in water it reacts, par-
tially, forming the free acid and the neutral salt, according to the equa-
tion, thus:

2NaHX :^ HjX + NajX.

The state of equilibrium reached is governed by equations (i) and (2),
which, by combination, give

^ Ostwald: Z. physik. Chem., 3, 281 (1889).

IONIZATION CONSTANTS OF DIBASIC ACIDS. 689

J!^^h (3)

Eqtiation (3) applies to solutions containing all proportions of add,
acid salt and neutral salt, as well as to solutions resulting from the pure
acid salt and water. I have found* that the equilibrium in solutions
of the carbonates of sodium is accurately represented by the relation-
ship

Where the formulae now represent the total concentrations of the re-
spective substances, Equation (3a) becomes identical with (3) for very
dilute solutions, in which the salts are practically completely ionized.
For a fixed concentration of total sodium, a good constant was found for
all proportions of carbonate, bicarbonate and carbonic acid. The con-
centration of the later component was proportional to the concentration
of the gaseous carbon dioxide, with which the solution was in equilib-
rium; the gaseous concentration, as determined by analysis, multiplied
by the coefficient of solubility, gave the concentration of the free add in
solution.

The same principle may be applied also to the study of solutions of
salts of all non- volatile acids, provided they are sufficiently soluble both
in water and in some inert solvent, which is immisdble with water, the
aqueous solution, containing the acid and neutral salts, is shaken, until
equilibrium is reached, with a solvent in which the add is soluble but
the salts insoluble. The concentration of the un-ionized free acid in the
aqueous layer is directly proportional to the concentration of the same
substance in the immiscible solvent, the proportionality factor being
the partition coefficient of the free acid alone, for the two solvents. A
simple analysis of the aqueous solution gives the remaining data for the
calculation of the concentrations HX and X.

The aqueous solution of the add and neutral salts of a dibasic acid,
HjX, will, in general, contain the following seven^ molecular and ionic
substances: NaHX, Na^X, HjX, Na, H, HX and X. If P is the parti-
tion coeffident, then //jX, the molecular concentration of the un-ionized
free add, is equal to 0.5 P times the equivalent concentration of the ether
solution. The relations between the concentrations of the six remain-
ing substances may be expressed by six equations, as follows:

H*HX = k.H^X, (i)

If «! and ttj are the degrees of ionization of the acid and neutral salts
respectively, then, as close approximations,

* Am. Chem. /., 29, 437 (1903).

' In moderately dilute solutions the amount of NaX ions is probably negligibly
small.

690 HERBERT N. MCCOY.

HX = a,(NaHX + HX) (4)

and

X = a,(Na^ + X). (5)

Since the solution is ele<jtrically neutral, we may write

Na + H ^ HX -h 2X. (6)

If th6 total concentration of the sodium is called m,

NaHX + 2Na^ ■}- Na ^^ m. (7)

Finally, if the equivalent add concentration of the aqueous solution
(as shown by the titration with standard alkali) minus P times the equiv-
alent concentration of the ether solution be called C, we may write

NaHX + HX + H ^ C. (8)

(4) and (5) give

HX = a^iC—H). (9)

(i) and (9) give

HX~^+ ^ {^y-k,a,H^. (10)

If H is very small compared with C, equation (9) becomes

HX = a^C^ nearly. (11)

Equation (lo) also reduces to (11) if h^Jrl^ is very small compared

^ (¥)■•

(6) and (7) give

NaHX + 2Na^ + HX + X—H = w. (12)

(4), (5) and (12) give

— + — - w. (13)

(9) and (13) give

2 Ofi J

If HX = ttjC, equation (11),

x-«,[

^-«»(^). ('5)

Equations (10) and (14) must be used for highly ionized adds, like
oxalic and dibromsuccinic, but the simplified forms (11) and (15) may
be used, without appreciable error for weaker adds like succinic.

The following pages give an account of a few preliminary experiments
made with salts of succinic add, at 20°, using ether as the auxiliary
solvent. The partition coefficient of succinic acid for water and ether
was found by shaking an aqueous solution of pure succinic add with
carefully purified ether at a temperature of 20°, for a period of five min-
utes, a separate experiment having shown that equilibrium was reached

IONIZATION CONSTANTS Oi^ DIBASIC ACIDS. 69 1

in that length of time. Forty cc. of the aqueous solution required 27.96
cc. of decinormal barium hydroxide, phenolphthalein being used as
indicator. Forty cc. of ethereal layer, after evaporation of the ether
and re-solution of the add in water, required 3.78 cc. of the same barium
hydroxide solution. The apparent partition coefficient is 27.96/3.78 =
7.40, but this must be corrected, since part of the acid in the aqueous
solution is ionized, while that in the ethereal solution is not. The true
partition coefficient is, as Nemst has shown ^ in all such cases, the ratio
of the unionized portions of the acid in each solution. The aqueous
solution was 4.28 per cent ionized, as calculated from its concentration,
0.0349 normal molecular, and the known ionization constant,' 66.5 X
lo"^. The true partition coefficient is, therefore, 7.40 (i — 0.0428) =
7.08. Another aqueous solution of 0.1362 normal and therelore 2.19
per cent, ionized, gave for the partition coefficient 7.16 (i — 0.0219) =
7.00. The mean of the two results, 7.04, is the true coefficient, which
is independent of the concentrations of the solutions.

The details of a determination of the concentrations of the components
of a solution containing both add and neutral sodium succinate are il-
lustrated by the following example: 20.00 cc. of exactly 0.2 molecular
succinic acid and 20.00 cc. of exactly 0.25 normal sodium hydroxide
were mixed with 50 cc. of purified ether and sufficient water so that after
being shaken the volume of the aqueous solution would be approximately
50 cc. The mixture was then shaken vigorously for 5 or 6 minutes,
at 20**, in a separatory funnel having three superimposed bulbs connected
by short tubes, which were graduated. The two lower bulbs were each
of 50 cc. capacity, while the upper bulb held about 150 cc. The two
smaller bulbs and the connecting tubes were carefully calibrated. After
being shaken, the solutions were allowed to settle. The volume of the
ether was 50 cc. ; that of the water solution was 50.42 cc. ; therefore the

sodium concentration, m, was = 0.09916. 20.00 cc. of the

' ' 50*42

aqueous solution required 11.66 cc. of N/io barium hydroxide; 40.00 cc.

of the ethereal solution required 6. 11 cc. of N/ioo barium hydroxide.

From these results it follows that the equivalent acid concentration of

the aqueous solution was 0.05830, and that of the ethereal solution

0.001528. Therefore C = 0.0583 — (7.04 X 0.001528) = 0.04754. If

we take a, « 0.80 and aj = 0.70,* then HX = aiC = 0.03803 and

X = 0.5 a, (m — C) = 0.01807. H^X = 0.5 X 7.04 X 0.001528 =

0.00538.

These values substituted in equation (3) give kjk^ = 14.8.

^ Z. physik. Chem., 8, no (189 1).

• Ostwald: Ibid,, 3,^282 (1889).

* Sec following paper.

692

HERBERT N. MCCOY.

The results of seven experiments, of which the above calculation rep-
resents the fifth, are given in the following table :

Molecular cone HtX.

Ionic cone. IfX.

Ionic cone. X.

k^.

0.01487

0.0520

0.0120

15.2

0.01258

0.0504

0.0129

15.7

0.01060

0.0480

0.0140

15.5

0.00862

0.0448

0.0152

15.3

0.00538

0.0380

0.0181

14.8

0.00276

0.0301

0.0218

15.1

0.00105

0.03o8

0.0258

16.0

(16)

Mean, 15.4

In all of the experiments represented by the above table the total
sodium concentration was approximately N/io. For constant con-
centration of total base, the ratio fej/fej is seen to be practically constant;
k^ = 0.000,0665, therefore ikj = 0.000,0043. Further extensive meas-
urements, using the method here delineated, have been carried out by
Mr. Chandler, as described in the article following.

The secondary ionization constant of a dibasic acid, ^2» ^°^y ^ calcu-
lated very readily by a second new method from the conductivities of
dilute solutions of the acid and neutral sodium salts. A. A. Noyes has
shown ^ that, for a dilute solution (say N/1024) of the acid salt,

(fe^ + m + H)H'
^" k,(m—H) •
Of the quantities on the right-hand side of equation (16), w, the total
concentration of the sodium is known, fcj is also known; //, the concen-
tration of free hydrogen ions, is the only unknown. This may also be
found from relationships based upon the following considerations: For
a very dilute solution of the pure acid-salt we may write

m^ + wAj -fi HXXffj^ + XX^ = mA^ (17)

where X^ is the ionic conductivity of the univalent basic ion (sodium)
and Xff, Xfjx and X^ the ionic conductivities indicated by the subscripts;
Jj is the observed molecular conductivity of the solution of the so-called
acid salt ; X^ is known and X^ may be obtained from a single determination
of the conductivity of a very dilute solution of the neutral salt.

It is obvious that Xfjx can not be determined in the ordinary way, on
account of the formation of H and X ions from the acid salt, but the
value of X^x ^ay be calculated, in most cases, from the observed con-
ductivity of the acid salt solution, when the ionic concentrations of the
various sorts of ions have been determined by the equilibrium method
already described; this would not, however, lead to an independent
determination of fe,. It has already been clearly shown by Ostwald'

• Z. phystk. Chem., 11, 495 (1893).

* Ifnd,, a, 840 (1888).

IONIZATION CONSTANTS OF DIBASIC ACIDS. 693

and by Bredig* that ionic conductivity depends tipon the composition of
an ion, and that the conductivity may be very closely estimated from
the composition. The values of Xfj^ found in the two ways agree well.
It has been observed, further, by Mr. Chandler,' that Xjjj^ is, in all cases,'
very nearly equal to o.^^Xx- ^^ ^^ accept this as a fixed relationship,
the result will be at least a very close approximation. Therefore

For a very dilute solution of an acid sodium salt of a dibasic acid, it
is easily seen' that

H^ = X—H, (19)

and

H^ + HX + X = w. (20)

(i), (19) and (20) give

{K + m + H)H^

(19) and (20) give

HX = m -^ H—2X; (22)

(17), (18), (21) and (22) give

jj, fn(2Ai—2\i—kx)—K(Xff-¥o.7^x)H ^ kim(Ai—\—o. sX^x)

2kH + ^X 2\if+\x

(23)

ni(2Ai—2\—\x)—ki(\H+o^7^x) _ ..

2(2\HTKd ""^ ^""^^

and

k{m(Ai — K — 0'3^;ir)
2\h+>^x

=6. (25)

// = a+ Va* + 6. (26)

Equations (24) and (25) may be simplified for sodium salts at the

definite temperature, 25°, as follows. Let the concentration, w, = 1

1024

and let A2 be the equivalent conductivity of the salt Na^X at this con-
centration, at which the equivalent conductivity due to the sodium ion
is about 49.5. Therefore, X^ = 49.5 and Xx = 2 (A2 — 49- 5). The value
of Xjf is 352, at 25**, according to Kohlrausch and Steinwehr.* Substi-
tution of these values in (24) and (25) gives

m(Ai—A2)—K(o.7A2+ 14O /^^x

2A2+60S ^ ^^

and

' Z. physik, chem., 13, 19T (1894).

■ Next article.

' Equations (19) and (20) have been used by Noyes, Loc. cU.

* Berl. Aktui. Siizber., 26, 570 (1902).

694 ^* ^ CHANDLER.

24, + 605

In equation (26) the value of // is expressed in terms of quantities
which are either known or easily determined experimentally, and, there-
fore, k^ may now be obtained by substitution of the value of H, so found,
in equation (16). The following application to the case of tartaric add
will serve as an illustration of the method, while numerous additional
acids are similarly treated in Mr. Chandler's article. For tartaric add,

K = 970 X io~^;* a m ^ , iii = 141 and J, = 108. Therefore,

1024

a = — 2.167 X IO"■^ c = +6.480 X io~®, H =117.6 X 10"^ and

^2 = 34-3 X lo"^. The ratio kjk^ « 28.3.

[Contribution prom Tim Kbnt Chismical Laboratory, UNTVBRSiry op Chicago.]

THE IONIZATION CONSTANTS OF THE SECOND HYDROGEN ION

OF DIBASIC ACIDS.

Bt E. E. Chandler.
Received March 3, 1908.

It is generally believed that dibasic a.cids ionize in two stages, thus:

H3X :^ H + HX and HX :^ H + X.

From a study of the conductivities of dibasic acids, Ostwald' concluded
that, excepting strong acids like oxalic, the second stage of the ioniza-
tion did not take place to an appreciable extent at concentrations greater
than milli-normal, since the primary ionization constant, k^, as calcu-
lated from the relation,

H^HX ^ k^H^, (i)

where the formulae H,HX, etc., represent molecular or ionic concentrations,
was really a constant for all concentrations greater than 1/1024 normal.
For smaller concentrations the apparent value of k^ as calculated on the
basis of equation (i). usually increased appreciably. This increase is
the result of the ionization of the second hydrogen ion.

If the secondary, ionization constant, i. e., the ionization constant of
the second hydrogen ion of a dibasic acid, is called fe,, then

H*HX == k^*HX. (2)

The magnitude of this constant has been determined previously for a
considerable number of acids by four entirely diflferent methods.

Trevor* determined the rate at which dilute solutions of add salts of
dibasic acids invert cane sugar, and from the results calculated the con-
centrations of the hydrogen ions in the solutions used. It was assumed

' Walden, Z. physik. Chem., 8, 445 (1891).

• Ibid., 3, 281 (1889V

• Ibid., 10, 321 (1892).

SECOND HYDROGEN ION OF DIBASIC ACIDS. 695

that the dissociation of a salt like NaHX into Na and HX ions is prac-
tically complete in dilute solutions and that further ionization of the HX
then yields H ions, the concentration of which governs the speed of in-
version. A little later A. A. Noyes^ developed a formula by means of
which he calculated from Trevor's data the secondary ionization con-
stants of the dibasic acids corresponding to the salts used by
Trevor. Tower* foimd values approximating those of Trevor and Noyes
by the use of oxidation cells. Smith* cairefuUy repeated and extended
Trevor's work, experiments bemg made to test the reliability of the
method. Wegscheider* obtained secondary ionization constants from
the conductivity of the free acids. A fourth, entirely distinct method
was used by McCoy to find the secondary ionization constant of carbonic
add.' This method was later extended and applied to the study of
succinic acid.* The method of McCoy as used for the latter acid, is based
on the following considerations. It was shown that when an acid salt
as NaHX, of a dibasic acid is dissolved in water it reacts thus:

2NaHX :^ HjX + Na^X.

The state of equilibrium in such a solution, as well as in one containing
any arbitrary ratio of total acid and base, is governed by the relations
expressed by equations (i) and (2), which by combination give

In order to determine the state of equilibrium one must know the con-
centrations of the components. To find the concentration H2X the solu-
tion may be shaken, until equilibrium is reached, with an immiscible
or partially misdble solvent, such as ether, in which the free add is solu-
ble, but the salt insoluble. The concentration of the acid in the ethereal
layer, multiplied by a factor, which is a constant for a given add at a
fixed temperature, gives the concentration of the free add, in molecular
form, in the aqueous solution. This factor is the partition coeflSdent
of the free add for water and ether, which in this case is the ratio of the
concentration of the molecular H^ in an aqueous solution of the add
alone, to that of the total add in an ethereal solution which is in equilib-
rium with the former. If the total concentration of the base, w, is known
for the aqueous solution containing H^X, NaHX and NaX, a determina-
tion of the total addity, as shown by a titration, gives the remaining
factor for the calculation of the concentrations of HX and X. The for-

* Z. physik. Chem., 11, 495 (1893).
«/Wd., 18, 17 (1895).

nbid., 25, 144, 193 (1898).

* MoruUsk., 23, 599 (1902); 26, 1235 (1905).

• Am, Chem. /., 29, 437 (1903). ...

• Preceding paper.

696 K. B. CHANDLER.

mulae by which these calculations are made are deduced in the preced-
ing paper. It is there shown that in general

HX-'f+yj ("-f ) - A.«.H^, (4)

X-«.('

and

m + C HX\. ,,

C is the equivalent acid concentration of the water solution minus P
times the equivalent acid concentration of the ether solution ; P is the
partition coefficient; a^ and a, are the degrees of ionization of the add
and neutral salts respectively, and k^ is the primary ionization constant
of the free dibasic acid; m is the total concentration of the base (say
sodium) in the solution. In most cases these formulas may be greatly

/aJC\ '
simplified; where k^a^H^ is very small compared with y^) »

HX - a^C nearly. (6)

In such cases

X - „.(!S=C). w

The method just outlined was applied to the sodium salts of succinic add
by Professor McCoy, at whose suggestion I have studied the conditions of
equilibrium in solutions of the salts of a number of other dibasic adds and
from the results have calculated their secondary ionization constants. In
the preceding paper it is shown how the secondary constants may also be
calculated from the conductivities of the solutions of the acid and normal
salts. I have also determined the conductivities of solutions of the salts
of those acids studied by the partition method and from the results have
obtained a set of independent values of the secondary ionization con-
stants.

The experimental work described in the paper consists of three prin-
cipal parts: the determination of

(i) The Partition Coefficients;

(2) The Equilibrium Constants;

(3) The Cofltiductivities of the Salt Solutions.

I. The Partition Coefficients.

It is well known that a quantity of a substance shaken with a mixture
of two immiscible or slightly miscible solvents at a fixed temperature is
divided between the two in a constant ratio,* if the solute has the same
molecular composition in the two solvents. If there is in either solvent
partial dissociation or association of the solute, then it is found that a

* Berthelot and Jungfieisch, Ann. chim, phys. (4), 26, 396 (1874).

SBCOND HYDROGEN ION OF DIBASIC ACIDS. 697

constant ratio of concentrations obtains only for those portions of the
solute having the same molecular composition in the two solvents.^

This ratio is known as the partition coefficient. I have found ether
to be the best solvent in general for use in the determination of the state
of equilibrium in solutions of dibasic acids and their salts. The experi-
ments of McCoy showed that the salts of succinic acid are not taken up
by the ether. In most cases ether dissolves the acids satisfactorily and
gives for aqueous solutions of the free acid a constant partition coeffi-
cient when allowance is made for ionization, thereby indicating that in
the ethereal solution the acid exists in the simple molecular form. In
other solvents, i. e., chloroform, many acids are partially associated.^
Moreover, an ethereal solution will readily separate from an aqueous
solution, with which it has been shaken, while solutions of some of the
other solvents tried do not readily do so. To purify the ether it was
shaken with dilute sodium hydroxide solution, washed repeatedly with
water, and finally distilled. The middle portion only was used. All
of the partition coefficients were determined at 25® ±0.01, except one
series at 0°. The substances were shaken together in a plain cork-stop-
pered bottle in a thermostat, first by a mechanical shaker, until they
had come to the temperature of the bath, and then violently for a few
minutes by hand and, finally, allowed to settle for about half an hour.
This procedure was fotmd sufficient to insure the attainment of equilib-
rium and to effect a complete separation of the ethereal and aqueous
layers. In order to remove a sample of the aqueous solution un-
mixed with ether, the top stem of the pipette was closed by the moist-
ened forefinger of one hand as the pipette was passed through the ether
layer; the body of the pipette was then grasped in the other hand. Suf-
ficient heat was thus communicated to expand the enclosed air and ex-
pel the drop or two of ether that had entered the stem of the pipette.
The solutions were titrated with standard barium hydroxide, phenol-
phthalein being used as indicator. Before the titration of the ethereal
solution, water was added and the ether distilled off. Most of the acids
used were products of the firm of C. A. F. Kahlbaum.

The results of the partition experiments with the free acids are given
in the tables of Series I. The first column gives the molar concentra-
tion, Af of the aqueous solutions after treatment with ether. The de-
gree of ionization, a, of an acid was calculated from its primary ioniza-
tion constant, k^, by means of the equation

(i—aW

= *,

^ Nemst, Z. physik Chem., 8, no (1891). Hendrixson, Z. anorg. Cktm,, 13, 73

(1897).

' HendrixsoD, Loc, cU.

698

E. E. CHANDI^ER.

For oxalic and dibromsucdnic adds and for dilute solutions of makic
add k^ is not constant; for oxalic and maldc adds, a was obtained by
interpolation from the results of Ostwald^ and for dibromsucdnic add
from those of Walden.' The third column contains the values of the
ratio, p, of the concentration of an aqueous solution of the add to that
of the corresponding ethereal solution. This ratio may be called the un-
corrected partition coeflSdent. The true partition coefl5dent, P, is the
ratio of the concentration of the unionized or molecular add in the two
solvents. It is easily seen that P = p(i — a). The values contained
in the last column are calculated from this relationship. All of the tables
except one refer to partition between water and ether at 25® ±ox)i.
Table IV refers to succinic add and these solvents at 0°.

I. — ^Partition Coefficients.

Tablb I. — OxALicAcm.
io*Jk, » 38000.

100 a, p.

0.3815

36 13.86

8.87

0.2767

42 15.44

8.84

0.2251

44 15-74

8.8x

0.1911

46 16.35

8.83

0.1339

51 17.77

8.61

0.0887

60 21.66

8.64

0.0x98

79 42.54

8.85

0.0103

86 61.38

8.35

0.0054

91.5 100.00

Tabids II. — ^Maix>nic Acn>.
xo'ik, ■— 1580.

8.50
8.64

100 a. p.

0.1478

9.8 10.94

9.86

0.1121

II. 5 11.07

9-79

0.0862

12.6 11.28

9.86

0.0331

19.6 12.22

Tablb III.— Succnac Aero.
lo'Jfe, *— 66.5.

9.82
9.83

100 a. p.

0.1708

1-9 7.73

7-59

0.0582

3-2 7.79

7.54

0.0287

4.7 7.73

7.36

0.0217

5.4 7.81

7.39

0.0120

7.2 7.95

7.37

0.0059

10. 0 8.39

•

7.55

0.0039

12.2 8.42

7.39

0.0023

15-6 8.79

742

^ Z, physik. Chem., 3, 281 (i88q).
» Ibid., 8, 479 (189O.

7 45

SSCOND HYDROGEN ION OF DIBASIC ACIDS.

699

Tasub IV. — Succinic Acid at o®.

0.0705
0.0702

0.0374
0.0200
0.0126
0.0116
0.0063
0.0039

100 a.

3.0

4.43

4.45

4.1

4.43

5.6

4.59

7.0

4.64

7.3

4.68

9.8

4.67

12. 2

4.85

4.30
4.31
4.25
4.33
432

4.33
4.21

4.26

4.30

Tablb v.—

-Pdcbuc Acid.

IO«*i

= 32-3.

100 a.

P-

0.00998

5.54

0.7095

0.670

0.00702

6.56

0.7170

0.670

0.00480

7-87

0.7195

0.663

0.00284

11.30

0.7480

0.663

0.00179

12.60

0.7075

0.653

0.0280
0.0085
0.0072
0.0063
0.0056

Tabls VI. — Gi^uTASic Acid.
io«*i - 47.4,

100 a.

4.0

7-2
7.8

8.3
8.9

A
372
3.84
391
392

3-93

0.664

3
3
3
3

57
56
60

59
58

3.58

TabitS VII. — Suberic Acid.

io«ib,

=329.

100 a.

0.00986

5.3

0.215

0.204

0.00544

7.2

0.228

0.211

0.00175

12.2

0.246

0.216

0.00084

17.2

0.258

0.214

0.00049

21.9

0.274

0.2x4

0.212

700

E. B. CHANDLER.

Tablb VIII.-

— AzBLAic Acid.

lo'ik,

= 25.

100 a.

0.00310

8.6

0.0679

0.0621

0.00178

II. 2

0.0702

0.0623

0.00x23

13-4

0.0718

0.0622

0.00096

15.0

0.0747

0.0635

0.00077

16.6

0.0753

0.0627

0.00064

18.0

0.0782

0.0641

0.00058

x8.8

0.0800

0.0648

0.00051

19.9

0.0810

0.0649

0.00046

20.9

0.0823

0.0651

0.00033

24.1

0.0868

0.0659

Tablb IX. — Sbbacic Acid.
lo'ik, = 23.8.

0.0637

100 a.

0.00062

17.8

0.0213

0.0175

0.00058

18.3

0.0213

0.0x74

0.00047

20.1

0.0221

0.0x76

0.00036

22.6

0.0232

0.0179

100 a, p.

0.0261

19.4 0.809

0.637

0.0197

21.9 0.822

0.642

0.0131

26.1 0.873

0.645

0.0119

27.2 0.894

0.651

0.0085

28.4 0.932

0.667

0.0057

36.7 0.996

0.63X

0.0056

36.9 1.006

Tablb XI. — ^Mbtaphthauc Acid.
lo'ilPi = 287.

0.635
0.644

100 a. p.

0.000398

56.2 0.0821

0.0359

0.000272

62.7 0.0943

0.0352

0.000263

62.9 0.0944

0.0350

0.000250

64.1 0.0949

0.0341

0.00229
0.00163
0.00148

Tablb XII. — Camphoric Acid.

io**i = 229.
100 a. p.

9.5 0.0387

I I. 2 0.0398

1 1. 7 0.0403

0.0350

0.0353
0.0357

0.0353

SECOND HYDROGEN ION OF DIBASIC ACIDS.

701

Tabls XIII. — ^iTACONic Acid.
lo'ikj s=i 151.

100 a.

P-

0.0615

4.8

2.99

0.0306

6.8

3.06

0.0161

9.2

3.18

0.0103

II. 4

3- 20

0.0091

13.8

3.22

0.0039

21.5

3.48

Table XIV.

-— Malbic Acid.

IO«*,:

= 11700.

100 a.

0.0993

29.0

9.60

0.0486

385

XI. 19

0.0337

44.1

12.20

0.0253

48.7

13.27

0.0196

53.0

14.40

0.0143

58.4

16.06

O.OIOO

63.7

17.78

0.0054

73.1

23.70

100 a.

0.0271

16.9

0.0114

24.8

0.0096

26.6

0.0092

27.1

0.0041

37.6

0.0041

37.6

Table XVI.-

100 a.

0.0879

16.3

0.0283

26.8

0.0253

28.2

0.0131

36.7

0.0056

49-9

0.0327
0.0307
0.0302

Table XV. — ^Puharic Acid.
io*ik, = 930.

0.782
0.871
0.889
0.893
1.053
1.040

TaBLS XVII. — DiBROMSUCCINIC AciD.
lo'ikj a= 34000.

100 a. p.

67.6 0.0578

68.6 0.0595

69.1 0.0603

2.83
2.87
2.88
2.81
2.83
2.85

2.86

6.82
6.88
6.82
6.82
6.78
6.69

6.45
6.38

6.71

0.650
0.655
0.652
0.648
0.658
0.650

0.652

-MONOBROMSUCCINIC AciD.

io*ikj = 2780.

P-

0.413

0.344

0.479

0.352

0.482

0.348

0.599

0.380

0.678

0.340

0.345
p.

0.0187
0.0187
0.0186

0.0187

702 E. B. chandi.br.

The results found for oxalic add are remarkable ; while the vahe of
p increases more than sevenfold within the range of concentrations used
the value of P is so nearly constant that the variations may reasonably
be considered as due to experimental errors. Now the value of a tised
in calculating P by means of the equation P = /)(i — a) is that deter-
mined from the conductivity of the acid. In its change of ionization
with change of concentration, oxalic acid does not follow exactly the
Ostwald dilution law, in which respect it resembles salts and the stronger
adds and bases. One of the most important problems connected with
the ionic theory has been to decide whether the degree of ionization
as determined in the ordinary way from the results of conductivity meas-
urements of a good electrolyte is correct. Assuming the validity of
the partition law, the constancy of the value of P seems clearly to indi-
cate the accuracy of the; values of a used. The view that the correct
value of a is given by conductivity measurements is in accord with the
conclusion of A. A. Noyes, who has subjected all the other evidence on
this point to a critical review.* The constancy of the partition coeffi-
dent, P, is equally good for the balance of the acids, for most of which
the dilution law holds true.

n. Equilibrium in Solutions of Salts of Dibasic Adds; Determination

of Equilibrium Constants.

It has been shown that the state of equilibrium in a solution containing
neutral salt, add salt and free dibasic add is represented by the equation

H^X • X *,• ^^^

The methods of determining the concentration of each of the constituents
of such a solution was outlined in the introduction. In order to calcu-
late the total concentration of the basic ion it is necessary to know the
final volume of the aqueous solution. When water is shaken with ether,
previously saturated with water, the volume of the aqueous layer in-
creases; this increase in volume was in some cases measured directly in
the manner indicated by McCoy for succinic solutions. In most ex-
periments, however, it was found more convenient to determine the in-
crease in volume in another way. It was fotmd by experiment that
when water or a dilute solution, such as was used in these experiments,
was shaken with ether previously saturated with water, that the in-
crease in volume of the aqueous solution always amotmted, at 25®, to
7.8 per cent., and at o®, to 15 per cent. The details of the experimental
procedure for the determination of the equilibrium constant may be
illustrated by an example taken from the work of glutaric add. A quan-
tity of pure glutaric acid suffident to give a mixture of the acid and neu-

^ International Congress, St. I/ottis, 1904, Vol. IV» p. 311.

SECOND HYDROGEN ION OI^ DIBASIC ACIDS. 703

tral salts was added to 25.00 cc. of exactly normal sodium hydroxide
and the solution diluted to 250 cc. Exactly 20 cc. of the resulting solu-
tion were shaken in a plain 250 cc. bottle with about 75 cc. of ether at a
temperature of 25® ±0.01 tmtil equilibrium was reached. This occurred
within ten minutes. After the solution had stood in the thermostat
for about thirty minutes longer and the ether and the aqueous layers
had completely separated, a portion of each was removed, in the man-
ner described for the determination of the partition coefiicients, and titra-
ted with standard barium hydroxide, the ether solution being first evapo-
rated after the addition of water. In all cases phenolphthalein was
used as indicator. 20 cc. of the aqueous solution required 10.70 cc. of
0.0972 normal barium hydroxide; 50 cc. of the ether solution required
18.60 cc. of 0.00992 normal barium hydroxide. Therefore the (equiva-

10.70 X 0.0972
lent) acid concentration of the aqueous solution = — =*

0.0520, and the (equivalent) acid concentration of the ethereal solution

18.60 X 0.00992
= — — = o .00369.

C = 0.0520 — 3.58 X 0.00369 = 0.0388.

H2X = 0.5 X 3.58 X 0.00369 = 0.0066.

Since the aqueous volume is increased 7.8 per cent, by the absorption

of ether, the total concentration of the sodium = w = 0.1/1.078 =

o .0928. HX and X may be calculated by the simplified formulas : HX =

ajC, and X = aal ). «! = 0.79 and a, = 0.70. Therefore,

HX = 0.0306 and X =* 0.0189.

io*ibi = 47. Therefore, lo^fe, = 6.3.

Since the sodium salts of monobasic acids are, in general, very nearly
equally ionized at equal concentrations, it seems probable that the ten-
dency of an acid salt, NaHX, to ionize into Na and HX must be about
the same as that of the simpler salts, NaX. Consequently it has been as-
sumed that ttj, the degree of ionization of NaHX, is equal to that of sodium
acetate at the same concentration. The degree or ionization, a2, of a
normal salt NajX has been taken the same as that of normal sodium suc-
cinate at the same concentration. The equilibrium concentrations
have been determined for solutions of the sodium salts of all the adds
m Series I with the exception of oxalic, in which case the potassium salts
were used on account of their greater solubility in water. Some diffi-
culties were encountered in certain cases. The aqueous solution of potas-
sium acid oxalate contains but a very small proportion of free acid and

704 E. E. CHANDLER.

neutral salt. This condition is unfavorable for an accurate detennina-
tion of the equilibrium concentrations and therefore of the ratio kjk^.
The same kind of difficulty in still greater degree appeared in the study
of maleic acid solutions. It was necessary in each case to use a propor-
tionately large quantity of the ether solution to obtain an accurate titra-
tion. Mono- and dibromsuccinic acids also present difficulties in that
each is acted upon slowly by water, the first to give hydrobromic and
malic acids, ^ and the second to give hydrobromic and brommaleic adds.'
However, the action of water at 25° is too slow to cause very apprecia-
ble errors.

Tabls XVIII. — Glutaric Aero

«l-

a,.

//fX.

NX,

kjkf.

kjk,.

0.75

0.64

0.1855

0.0808

0.0167

0.0606

0.0335

6.6

0.75

0.64

0.1855

0.0807

0.0170

0.0605

00335

6.4

6.5

0.78

0.695

0.1022

0.0405

0 0068

0.0316

0.0214

6.8

0.78

0.695

0. 1022

0.0437

0.0089

0.0341

0.0203

6.4

0.78

0.695

0. 1022

0.0448

0.0090

0.0349

0.0200

6.8

0.79

0.70

0.0936

0.0464

0.0II4

0.0366

0.0165

7.1

0.79

0.70

0.0936

0.0528

0.0165

0.0417

0.0143

7.4

0 79

0.70

0.0936

0.0529

0.0165

0.0418

0.0142

7.4

0.79

0.70

0.0936

0.0480

O.OII9

0.0379

0.0160

7.6

7-4

0.79

0.70

0.0928

0.0388

0.0066

0.0306

0.0189

7-5

0 79

0.70

0.0928

0.0464

O.OII3

0.0366

0.0162

7.3

7.4

0.81

0.73

0.0618

1.0084

0.0025

0.0068

0.0195

9.5

0.81

0.73

0.0618

0.0164

O.OI5I

0.0133

0.0166

7-1

0.81

0.73

0 0618

0.0269

0.0489

0.0218

0.0127

7.6

0.81

0.73

0.0618

0.0316

0.0763

0.0256

O.OIIO

7.6

o.8t

0.73

0.0618

0.0344

0.0445

0.0279

O.OIOO

8.2

0.81

0.73

0.0618

0.0406

0.0150

0.0329

0.0077

9-4

8.1

0.83

0.76

0 . 0464

0.0254

0.0064

0.0211

0.0080

8.7

0.83

0.76

0.0464

0.0261

0.0068

0.0217

0.0077

8.9

0.83

0.76

0.0464

0.0248

0.0070

0.0206

0.0082

7.4

0.83

0.76

0.0464

0.0248

0.0067

0.0206

0.0082

7.7

0.83

0.76

0.0464

0.0252

0.0068

0.0209

0.0081

8.0

8.2

0.84

0.77

0.0366

0.0199

0.00526

0.0167

0.00643

8.3

0.84

0.77

0.0362

0.0197

0.00514

0.0165

0.00635

8.4

0.85

0.87

0.0316

0.0172

0.00490

0.0146

0.0562

7.8

0.85

0.87

0.0309

0.0172

0.00490

0.0146

0.00534

8.2

0.90

0.83

0.0155

0.0083

0.00204

0.0075

0.00299

0.90

0.83

0.0155

0.0086

0.00236

0.0077

0.00286

8.9

9.0

Table XVIII gives the details of all experiments with glutaric acid.
The symbols m, C, HX, and X represent the respective concentiatioos
in terms of gram equivalents per liter, while H^X refers to gram mok-
cules per liter. The simpler formulae, (6) and (7), have been used for
calculating HX and X in all cases, except for oxalic and dibromsuccinic

* Tantor, /. Russ. Chem. Soc, 23, 339 (1892). Beilstein, Org, Chem., i, 658.
' Vaii't Hoff, Etudes de Dyn. Chem. Amsterdam, page 14.

SECOND HYDROGEN ION OF DIBASIC ACIDS. 705

adds. Table XIX gives, in condensed form, the results for other acids;
in each case several determinations were made at each concentration,
as in the case of glutaric acid. The details are omitted in order to save
space.

Tabls XIX.
Ratio of Ionization Constants, i^/JE^ at Various Concentrations, m.
Oxalic m 0.2691 0.1346 0.0769 0.0598 0.0448 0.0245

^i/^s S71 981 919 S40 1087 S^
Malonic m 0.3x34 0.1880 0.1180 0.0940 0.0784 0.0587 0.0470 0.0393

V*-» 357 390 444 459 474 47© 488 493

Sucdnic m 0.1870 0.0834 0.0312 0.0170 0.0046

ki/^M 15-5 i6-9 19-2 21.0 26.6
Glutaric m 0.1855 0.1022 0.0936 0.0618 0.0464 0.0362 0.0155

k^k^ 6.5 6.7 7.4 8.1 8.2 8.4 9.0

Suberic m 0.1855 0.08240.03x1 0.0x69

kjk^ 4.2 4.7 5.4 6.1

Pixnelic m 0.X859 0.0833 0.0309 0.0169

V*s 3-9 4.4 5.2 5-6

Azelaic m 0.X855 0.0824 0.031X 0.0x69

*,/*, 4 2 4.8 5.2 5.8

Sebadc m 0.1855 0.0824 0.0337 0.0x69 0.0046 0.00x8 0.00092 0.00046

^lAi 4-0 4.2 4.8 5.7 7.4 9.x 9.2 X0.4

(^-Phthaiic.... m 0.1858 0.0825 0.0309 0.0x70 0.0089

k^k^ 165 200 240 265 290

m-PhthaHc. . . m 0.09260.04630.03x1

V*i 5.3 6.6 6.5

Camphoric... m 0.0923 0.046X 0.0154 0.0084

V*t 8.x 9.3 XX. 3 X3.2

Itacooic m 0.09260.04680.0x560.0084

*i/^ 33-7 38.9 46.1 57.0

Maleic m 0.0926

*,/*, 30000

Pumaric m 0.0926 0.0468 0.0x56 0.0084

*,/*, X8.3 20.3 25.8 30.0
Monobrom-

sucdnic... m 0.0926 0.0463 0.0x56 0.0084

ki/kj 46.0 54.0 69.0 91.0
Dibromsuc-

dxiic m 0.0927 0.0463 0.0x57

*i/*a IO-9 13-5 12.6

In discussing the results we may consider the influence of three fac-
tors upon the value of the equilibrium constant: (i) The effect of the ratio
of the total base to the total acid in solutions of the salts where the con-
centration of the base is^ constant; (2) The effect of the total concentra-
tion of the base; (3) The effect of temperature. The experiments of Mc-
Coy on carbonates and succinates showed that, for constant concentra-

total base, m, to total acid in any experiment, is ^ , . In

706 E. E. CHANDLER.

tions of the total base, the value of kjk^ is independent of the ratio of
base to acid. My experiments lead to the same conclusion. If an aque-
ous solution of the pure add salt (that is, one having equal molecular
concentrations of base and acid) is shaken with ether the ratio of base to
add becomes greater than tmity. In the experiments here described
no attempt was made to maintain equal concentrations of base and add
in the aqueous solution ; indeed in the case of glutaric add the ratio was
varied greatly in order to test the independence of kjk^- The ratio of

the seventh experiment with glutaric add this ratio is 1.044; ^ ^^ ^^
it is 1.282 ; and in the twelfth it is 1.644. Sudi a high value of the ratio
as the last is unfavorable to accuracy because of the exceedingly small
titer of the ether solution ; this doubtless accounts for the abnormal value
found for fej/A?,.

On the other hand, in the seventeenth experiment the ratio, is 0.735
and it is interesting to note that in spite of the large excess of acid above
that required to form the dry add salt the solution contains an apprecia-
ble amount of the ion of the neutral salt. H^re the great excess of add
is also unfavorable for an accurate determination of the constant; never-
theless, the value found is not far from the mean. In the first experi-
ment vrith a sebacic add the ratio of base to add is 1.84. . Owing to the
limited solubility of the add in water, it is impossible to decrease this
ratio very much at large concentrations of the base. Sebadc add, how-
ever, is so much more soluble in ether than in water that there is always
suffident add in the ether solution to make an accurate titration possi-
ble and no difficulty was found in getting concordant values of kjk^.

I have found that fej/Afj increases with decreasing concentration of the
total base. A similar result was found by McCoy for carbonates. Smith
also observed a diminishing value of fe, vrith dilution. This is equiva-
lent to an increase of kjk2 if k^ is constant.

The values of fej/fej have been found by inter- and extrapolation for
the concentrations o.i, o.oi and o.ooi of total base vrith results shown
in the following table. The last two columns contain the ionization
constants of the first and second hydrogen ions. The values of k^ were
calculated from those of kjkz for o.ooi normal concentration:

Table XX.

Cone. 0.1. Cone. 0.01. Cone, o.oox. xo*Ai. io%

Oxalic 930.0 930 930 38000 40.9

Malonic 460 610 780 1580 3.03

Succinic 16.5 23.5 30.5 66.5 2.18

Glutaric 7.0 9.8 14.0 47.4 S-fi

SECOND HYDROGBN ION OF DIBASIC ACIDS. ^0^

TablA XX {Continued),

Cone. 0.x. Cone. o.oi. Conco.ooi. icfik\. io*l^.

Suberic 4.6 6.4 8.1 29.9 3.67

PnncHc 4.4 5.9 7.4 32.3 4.37

Aickdc 4.6 6.1 7.6 25.3 3.33

Scbadc 4.1 6.5 9.2 23.8 2.59

(h-Phthalic 190 290 390 1210 3.10

fn-Phthalic 5.2 8.1 10.8 287 26.6

Camphoric 8.0 12.4 16.4 229 14.0

Itaconic 32.0 51 70 151 2.16

Maleic 3cxxx> (45000) (60000) 1 1700 o. 20

Ftunaric 17.0 30 43 930 21.6

Monobromsuccinic 44-o 78 112 2780 24.8

Dibrcmsucdnic 10 (16) (22) 34000 1540

The values in parentheses, for maleic and dibromsucdnic acids, are
uncertain.

The effect of temperature was determined by means of a series of ex-
periments at o® with succinates. The following results were obtained:

Table XXI.

m 0.1739 0.0869 0.0497 0.0290 0.0158

*i/*« •••• 15-5 16.7 18.0 19.7 23.1

It will be seen that although the partition coeflScient of succinic add
at o® (fourth table, Series I) is 4.30, instead of 7.45 at 25°, yet the value
of kjk2 is practically the same at the two temperatures. This fact is
in accord with many well-known results, which show that degrees of
ionization and ionization constants are, for most substances, nearly the
same at b® and at 25®.

It is interesting to note that kjk^ for the dibromsucdnic acid is not
greatly different from kjk^ for succinic add, although both k^ and fe,
are enormously greater. Halogen substitution has therefore affected
the dissociation of the two hydrogens about equally.

A series of experiments was carried out with sebadc acid, using chloro-
form instead of ether. The results of the partition experiments, con-
ducted at 25*^ ±0.01 are given in the following table. A^ and A^ repre-
sent the molecular concentrations of the add in the water and chloro-
form solutions respectively. It is well known that there is association
of the molecules of many substances when dissolved in chloroform. Al-
lowance for a only, therefore, fails to make the values of P constant. The
smaller values of P for the more concentrated solutions indicate, of course,
a greater degree of association in the chloroform.

Tablb XXII. — ^Partition Coefficisntb of SEBxac Acid.

Aw. Ac a, p. P,

o.ooioo 0.000963 14.3 1.04 0.89

0.00059 0.000438 18.2 1.36 1.09

0.00030 0.000176 24.5 1.70 1.29

9.90018 0.000087 30.3 2.08 i.i^5

7o8 E. E. CHANDLER.

The following table contains the results obtained in the determination
of kjk2 for sebacic acid by the chloroform method. The column head-
ings have the same significance as in the tables where ether was used,
but the column headed A^ gives in addition the concentration of the add
in the chloroform. The experimental treatment is the same as that
described for ether extraction except in two particulars, (i) The solu-
bility of chloroform in water is so slight that the volume of the water
solution is not changed when it is mixed with chloroform, so that the
concentration of the total base is not altered thereby. (2) Inasmuch as
there is no constant partition coefficient, its value for each experiment
with a salt solution usually must be determined by interpolation. No
partition experiment was made with the free sebacic acid in which the
chloroform concentration was so great as that of the first experiment of
the following table. The corresponding value of P, 0.84, was obtained
by graphical extrapolation :

Table XXIII. — Sebacic Acid.

«l-

««.

Ac.

H^.

HX.

kjkf.

0.73

0.63

0.20080

0.02104

0.00148

0.84

0.00125

0.0154

0.0566

3-33

0.79

0.70

0.09245

0.01182

0.00085

0.91

0.00077

0.0093

0.0282

4.01

0.79

0.70

0.09245

0.01195

0.00094

0.89

0.00083

0.0094

0.0282

3 79

0.85

0.78

0.03346

0.00454

0.00026

1. 17

0.00031

0.0039

0.0113

4-34

0.85

0.78

0.03346

0 . 00484

0.00033

1.14

0.00038

0.0041

0.0112

395

0.88

0.82

0.01826

0.00269

0.00016

1.30

0.00021

0.0024

0.0064

4.29

0.88

0.82

0.01826

0.00253

0.00014

1-33

0.00019

0.0022

0.0065

392

The first value of kjk2 is uncertain on account of the uncertainty of
the factor P = o. 84. The remaining values of i^i/fej ^^e almost constant.
However, too much stress should not be laid on the constancy, as the
method is involved by the phenomenon of association.

For an acid which is difficulty soluble in water, kjk2 may be deter-
mined in a manner still different from that described above. The solu-
bility of the acid in pure water is first determined and allowance made
for dissociation in the usual way. When the solid acid is present in ex-
cess the concentration of the undissociated portion of the dissolved add
in aqueous solution of the free acid is equal to its concentration when the
salts also arc present. It is well known that it is not easy to determine
with accuracy the solubility of a difficultly soluble substance. About
two months were spent in an endeavor to find the solubility of sebadc
acid in water. Equilibrium between the dissolved and undissolved add
was sought by approaching 25^ from both a higher and a lower tempera-
ture. Rapid shaking was found to increase the solubility considerably.
The result is probably to be explained by the comminution of the crys-
tals by the greater agitation, it being apparently well established that
smaller particles are more soluble than larger ones.* Forty-four ex-

* Ostwald, Z. phys. Chem., 34, 495 (1900); Hulett, Ilnd., 37, 385 (1901).

SECOND HYDROGEN ION OF DIBASIC ACIDS. 709

periments gave an average molar concentration of 0.00118; individual
values ranged from 0.00125 to 0.00105. At a molar concentration of
o.oon8, the degree of dissociation is 13.3 per cent. Therefore the con-
centration of the undissociated portion is 0.00118 X 0.867 = o.ooio.
To filter the solution, it was drawn into a pipette by means of a filter
pump through a closely packed cotton plug. It was then titrated. That
error did not arise from imperfect filtration was shown by the fact that
separately filtered portions of the same solution had the same concen-
trations.

To determine kjk^ sodium hydroxide of known concentration was
shaken with excess of acid, filtered as described, and titrated. The fac-
tors of equation (3) are determined in the same manner as for the parti-
tion method, except that the concentration, i/jX, remains 0.0010 for all
experiments. This method of determining fej/A, for sebacic acid was not
satisfactory. It was even more difficult to get a constant equilibrium
between the acid salt, the neutral salt, the dissolved and undissolved
acid, than between the last two alone.

The values found for hjk^, averaged about 5, but some values more
than double this amount were found without my being able to assign
any reason therefor.

The same method was also tried with suberic acid, but only one ex-
periment was made. The molar concentration of its saturated solution
was found to be 0.0144, at which concentration the degree of dissocia-
tion is 4.4 per cent, and the molar concentration of the undissociated
add is 0.0138. In a single experiment in which the total concentration
of sodium was 0.201 the value 5.5 was found for hjk^ for suberic acid.
It is possible that an acid of about this solubility would give constant
results and regret is expressed that more experiments were not made
with suberic acid by this method.

m. The Determination of the Secondary Ionization Constant of a Dibasic

Acid by the Conductiyity Method*

The conductivity of the negative ion, HX, of an acid salt, NaHX,
cannot be found in the usual way on account of the ionic interactions
of the solution of the acid salt. The values used by Ostwald* and Bredig'
as the conductivities of HX ions, for the calculation of the degrees of
ionization, from the conductivities of the free dibasic acids, were merely
estimated from the composition of the ion.

I have found the approximate values of such ionic conductivities
in the following manner: The concentrations of acid and neutral salts

* Z. phys. Chem., 2, 840 (1888); 3, 281 (1889).

* llnd., 13, 191 (1894).

7IO S. E. CHANDI<BR.

and of free H ions were calculated for a dilute solution, say N/iooo, of
the pure acid salt, by means of the value kjk^ found by partition experi-
ments. The observed equivalent conductivity of the solution of the
acid salt diminished by the conductivity due to the neutral salt and H
ions represents the conductivity of the ions of the acid salt. This differ-
ence, by division of a number representing the fraction of the sodium
actually in the form of acid salt gave the true (hypothetical) equivalent
conductivity of the acid salt. By subtracting from the latter value the
known ionic conductivity of sodium the ionic conductivity of HX re-
mained.

The results so obtained clearly revealed a very simple relationship.
The conductivities of the HX ions were all approximately proportional to
those of the corresponding X urns. The detailed results of these calcula-
tions are omitted, but the relationship discovered has been utilized to
calculate the secondary ionization constants by a method entirely inde-
pendent of the data of the partition experiments. The good agreement
between the values of k^ as found by the partition and conductivity meth-
ods serves, of course, as an equally satisfactory test of the accuracy of
the law just announced.

The same relationship between the conductivities of HX and X was
observed when the conductivities of the HX ions were estimated in an-
other way. The work of Ostwald^ and Bredig^ has served to show that
the conductivity of an organic ion is dependent upon its composition.
Univalent, isomeric ions are equally mobile, and univalent ions, com-
posed of the same number of atoms, have practically the same conduc-
tivity. One may therefore safely assume that the conductivity of any
ion, HX, of a dibasic acid, H^X, is equal to that of an ion of the most
nearly related monobasic acid. Thus the conductivity of the add suc-
cinic ion, HCO2.CH3.CH2.CO2, may be considered equal to that of the
ion of butyric add, CHg.CHj.CHj.CO,.

In Table XXIV, Jx A is the equivalent conductivity of the X ion of the
corresponding dibasic add ; X is the equivalent conductivity of the nega-
tive ion of that monobasic acid most closely resembling the dibasic add
of the same line. The ratio,>l/i^Xi is nearly a constant, the mean value
of which is 0.595; or practically 0.6. We may therefore consider that

It is readily seen that the value of X^^t l^ist colunm, Table XXIV,
calculated by this equation, does not differ greatly, in any case, from
that of X,

* Loc, cU,

SECOND HYDROGEN ION OI^ DIBASIC AClDS.

711

OzaJic 72.4

Malonic 64. i

Sucdnic , 56.0

Glutaiic 51 .0

Suberic 44. 3

Azdaic 43. 3

Sebadc 40. i

Camphoric 40. 8

o-Phthalic 49. 8

ff»-PhthaUc 50.0

Itaconic 55.7

Makic 60.0

Fumaric 60. 6

Monobromsucdnic 56 . 4

Dibromsucdnic 55 . 2

Tablb XXIV.

X. X/JXa-. o.3Xa'=X^a'.

Acetic 41 . 6

Propionic 36. 8

Butyric 33.0

Valerianic 31.0

Phthaluric 26. 5

PhthalaniHc 26. 2

Toluic 32.1

" 32.1

Angelic 31.5

Crotonic 34.5

" 34.5

Butyric 33.0

330

0.575

43-4

0.574

38.5

0.589

33-6

0.608

30.6

0.598

26,5

0.605

26.0

24.1

24- 5

0.651

29.9

0.642

30.0

0.566

33.4

0.575

36.0

0.569

36.4

0.585

33.8

0.598

33.1

Mean, 0.595

By means of the relationship just discussed, McCoy has shown in the
preceding paper, how the concentration of hydrogen ions in the solution
of an add salt of a dibasic acid may be calculated from conductivity data.
For the acid sodium salt of N/1024 concentration,

// = a 4- Va' + b (9),

where

and

2il, + 605
kim{Ai — 0.6 At — 20)

(10)

(I I)

ft.

(12)

ft . 2 J, + 605

and J| and A^ represent the observed equivalent conductivities of the add
and normal salts respectively, at 25® and N/1024 concentration. The
value of H calculated by means of the preceding equations may be used
to calculate ft, by substitution in Noyes's equation :*

(ftt + m + if)H»
fti(fii— H)

I have measured the conductivities of those add and normal salts
studied by the partition method. The results are given in Tables XXV
and XXVI. The conductivities of the add salts of oxalic and malonic
adds, as well as of many of the neutral salts, have been measured pre-
viously by Walden' and Bredig.' My results agree well in general with
those of Walden and Bredig, in so far as comparison is possible.

* Z. phys. Chem,, xi, 495 (1893).

•/Wrf., 8, 433(1891).
» Ibid., 13, 191 (1894).

712 E. E. CHANDLER.

Tablb XXV. — CoNDUCTiviTiBS AT 25® IN RECIPROCAL Ohms OF AciD Sai*t&— NaHX

Dilution. 32. 64. X38. 256. 51a. 1024. 2048. 4096.

Oxalic 97 109 123 139 158 182 211 250

Maionic 80.5 84 87 90.6 97.3 104.8 113 126

Succinic 76 80 83.3 86.5 89.5 94 loi iii

Glutaric 73.4 78.7 82.4 85 88 91.6 98 106.4

Suberic 68.9 72.5 75.7 78.3 81.5 85 91 99

Azelaic 68.9 72.5 75.7 79.3 81.5 85 89.8 97

Sebadc 75 78 80 82.9 88.3 95

o-Phthalic 70.6 75 80 85 92 100 112 125

m-Phthalic 91.3 100.6 115 136 171

Itaconic 74 77.8 81.4 85.2 89 94.8 102.6 116

Maleic 79 84 88 91 . 5 95 99. 5 104 no

Fumaric 80.5 86.2 93 104 118. 7 140 170 205

Monobromsuccinic 78.6 86 96.5 109 127 151 180.6 217

Dibromsuccinic 150 187 231 280 331 381 422 456

Tablb XXVI. — Normal Salts — Na^X.

Dilution. 32. 64. 128. 256. 512. 1024. 2048. 4096.

Oxalic 99 105 no 115 118 120 121 122

Maionic 91 -7 98.2 102.4 105.5 109 112 114 115. 5

Succinic 87.3 92.3 96.7 99.7 102 104 105.5 106

Glutaric 82 . 9 88 . 2 92 . 5 95 97 99 loi 102

Suberic 76 81.4 85.3 88.1 90.6 92.3 94.5 96

Azelaic 75 80 84 87 89 91 92 93

Sebacic 81 84 86.4 88 90 91

o-Phthalic 81.5 86. 5 91 94 96 98 99 100

m-Phthalic 92 95 98 loi 105

Itaconic 86 90.6 95 98.4 102 104 105 106

Maleic 90 95 99 102 . 5 105 . 5 108 109 109.5

Fumaric 89.5 94.6 98.5 102 165 108 no 112

Monobromsuccinic 86 91 95 99 102 105 108 112

Dibromsuccinic 84 89 94 98 10 1 104 107 no

Table XXVII gives the results calculated from these conductivities
by equations (9) and (12):

Table XXVII.

A^. A,. xo^ifc]. 10*17. 10^^

Oxalic 182 120 38000 193 49

Maionic 104.8 112 1580 35 2.1

Succinic 94 104 66.5 12.9 2.7

Glutaric 91.6 99 47.4 1 1. 4 2.9

Suberic 85 92.3 29.9 7.4 1.9

Azelaic 85 91 25.3 7.7 24.

Sebacic 82.9 88 23.8 7.6 2.5

o-Phthalic 100 98 1210 44-4 3-9

m-Phthalic 115 98 287 67.9 24.

Itaconic 94.8 104 151 18.7 2.8

Maleic 99.5 io8 11700 15. t 0.26

Fiunaric 140 108 930 113 32

Monobromsuccinic. . 151 105 2780 151 39

Dibromsuccinic 381 104 34000 675 1600

SECOND HYDROGEN ION OF DIBASIC ACIDS.

713

Table XXVIII contains all of the values which I have found for the
ionization constants of the second hydrogen ion of the dibasic acids
studied, together with those found by Smith, Wegscheider and Trevor.

Table XXVIII. — Secondary Ionization Constants, lo'ifej.

Partition. Conductivity. Smith. Wegscheider. Trevor.

Oxalic 41

Malonic

Sucdnic

Glutaric

Suberic

Pimelic

Azelaic

Sebadc

o-Phthalic

ffi-Phthalic

Camphoric

Itaconic

Maleic

Fumaric

Monobromsuccinic

• •

•

2.0

2. 1

I.O

2.2

2.7

2.3

3-4

2.9

2.7

3.7

2.5

4-4

• •

3-3

2.4

2.7

2.6

2.5

2.6

3.9

■ •

• •

0.7

2.2

2.8

2.3

0.20

0.26

0.4

• •

1600

• •

•

16.0

10. o

Dibromsuccinic 1540

The methods of Trevor and Smith are similar, except that a different
constant is used for the inversion effect of a completely dissociated acid.
Their solutions were 1/128 normal with respect to the acid salts and the
experiments were carried out at 100°, as the speed of inversion was
too slow at 25°. This should not materially effect the value of ^2» un-
less there was some decomposition of the acid at high temperature. Smith
found that malonic acid suffered decided decomposition at 100° and the
value given is an interpolation. He also found that a neutral salt or
even water caused quite as rapid an inversion as an acid salt, although
in the former cases the rate was irregular. From these considerations,
it is remarkable that values obtained by Smith's method agree so well
with others.

The method of Wegscheider, using the free acid, is probably less ac-
curate than the method using the acid salt. Even though the latter
is complicated by the presence of the neutral salt. The former method
is evidently more accurate for acids having a large value of ifej than for
those for which ^2 is small. The values found by Wegscheider for suberic,
fumaric and monobromsuccinic acids agree very well with those I have
found. But his values of Ajj for maleic and malonic acids, which other
experimenters find to be small, are certainly too large.

In conclusion, I wish to express my obligation to Dr. McCoy, under
whose supervision the foregoing work has been carried out.

714 ^' 1>EKAY tHOMPdON AND M. W. SAG^.

ON THE FREE EITERGT OF mCKEL CHLORIDE.

Bt M. dbKat Thompson and M. W. Saob.
Received Pebmaxy 14, 1908.

z. Introduction*

In a previous paper^ by one of the authors, calculations of the free energy
have been made, from potential measurements, of all compounds for
which the necessary data already existed, but so far no direct free energy
determinations of salts have been made on the principle underlying
these computations. The object of the following investigation is to de-
termine the free energy of nickel chloride by this method, for a fuU de-
scription of which reference may be made to the above paper. Briefly,
for the salt here investigated it consists in measuring the electromotive
force of the cell:

Nickel I Saturated sol. of NiCl, | Pt + CI,.

If the salt contained no water of crystallization and the chlorine weie
at a pressure of 76.0 cm. of mercury the free energy increase of one mol
of the salt would be

iXF - —2EF

where E is the measured electromotive force and F is one faraday.' On
account of the fact, however, that nickel chloride takes on water of crys-
tallization, a slight modification of the ideal process by which the nickel
and chlorine are allowed to imite reversibly is necessary.

Nickel chloride, in contact with water, forms the hydrate NiCl4.6H,0.*
A hydrate with four molecules has also been detected,* but this existed
in contact with a saturated solution at a temperature about 80** and,
consequently, would not enter into this process. According to Lescoeur,
the anhydrous salt is produced directly on dehydrating that with tw)
molecules of water. This conclusion is based on vapor pressure meas-
urements at 100^, starting with NiCls*^^)^ ^^^ making measurements
after successive evacuations of the apparatus. The final analyas of
the hydrate gave NiCl3.i.45H20, which might be a mixture of NiClr
2H,0— NiClj-iKjO or NiCLj.2H,0— NiCl,. Although this point is doubt-
ful, it will be assumed that there is no hydrate with one molecule of
water.

The salt in contact with the saturated solution of nickel chloride is
then NiCl2.6H20, and in the ideal process, as nickel and chlorine unite
and precipitate from the solution, water would be removed from the
solution, and in order to keep the amotmt of water in the solution con-

* This Journai*, 28, 731 (1906).

' To bring the conoeption of free energy increase into conformity with the pres-
ent more general usage, the terms involving RT are included in AF. See Haber,
"Thermodynamik technischer Gaserak," p. 9.

■ Lescoeur, Ann. chim, phys, [6], 19, 533 (1890).

* Etard, Ibid. [7], 3, 545 (1894).

FRBB ENERGY 01^ NICKEL CHLORIDE. 715

stant, this has to be supplied leversibly and isothennally at the same
rate at which it is removed. This may be accomplished by means of
isothermal distillation, as follows: Consider water at the temperature
in question enclosed in a cylinder with a frictionless piston. Allow six
moles to evaporate reversibly and isothermally. The free energy in-
crease of the system is — 6RT. Now separate the six mols and allow them
to expand reversibly and isothermally until the pressure is equal to the
vapor pressure of a saturated solution of NiCl2.6H20. The free energy
increase of the system in expansion is

—eRT log ^,

Psoi

if pff^ is the vapor pressure of water at T° absolute and p^^^ is the vapor
pressure of the solution. The water vapor may now be condensed re-
versibly into the solution, at the same rate at which it is removed by the
salt crystallizing out. The free energy increase of the system in this
last step is +6/?T, so that the net result is the free energy increase,

— dRTlog

psol

£iF.=—2EF — 6RT log^ (i)

psoi

is therefore the free energy of NiCl2.6H20, referred to gaseous chlorine
at atmospheric pressure and liquid water, both at T° absolute.

The free energy of NiCl2.2HjO may be obtained from this by allowing
NiClj.6H30 to change reversibly into NiCl3.2HjO and four molecules of
liquid water. To do this allow four mols of water to evaporate at the
vapor pressure p^ of the system NiCl2.6H20 — ^NiCl3.2H20, whereby AF
is — 4RT. The water vapor must now be compressed to the vapor pres-
sure pff^ of water, in which process AF is

+ 4RT log ^.

The water vapor is then condensed to liquid water, in which AF is +4RT.
The free energy increase then of NiCl2.2H20 is

AF2= — 2EF — 6RT log ^^ +^T log ^ . (2)

Psol Pi

By an exactly similar process the free energy increase caused by
NiCl2.2H20 going over to NiCl, and 2H2O is

H-22eTlog^,

where />, is the vapor pressure of the system NiCl2.2H20 — ^NiClj, and
hence the free energy increase of formation of NiClj is

AF« — 2J5F — 6/?T log?^^ + 4/?riog^ +

psol Px

2/^Tlog^- — 2EF + /?riog^^* (3)
Pa rir

7l6 M. DHKAY THOMPSON AND M. W. SAGS.

To all the above expressions a small correction must be made for the
gaseous chlorine, whose pressure will in general not be 76 cm. of mer-
cury. Assuming the vapor pressure of the solution to be unaffected by
the presence of the chlorine in the enclosed space above the chlorine
couple* would be b — p^^ where b = barometric pressure. The increase
in free energy of the system when the chlorine expands from 76 cm. to

^—Psoi is

AF = -/?riog^-7^,

which must be added to the right-hand side of equation (3).

2, Potential Measurements,

The apparatus in which the potential measurements were carried out
consisted of an H tube 15 cm. high with a glass stop-cock in the hori-
zontal arm. The tubes of the vertical arms were 2 to 3 cm. in diameter.
The stop-cock was not lubricated and the potential of the cell was usually
measured with the stop-cock turned off.

The chlorine electrode was a smooth sheet of platinum with 10 per
cent, iridium, 10 cm. long by 3 cm. wide. This was half way immersed
in the solution, and chlorine gas, generated by the electrolysis of strong
hydrochloric acid, bubbled over it from below. The chlorine was led
from the cell by a rubber tube to a bottle containing potassium hydrox-
ide, where it was absorbed. The pressure of the gas over the chlorine
electrode was determined by a small manometer in the escape tube and
by the barometer reading. The nickel electrode was formed by electro-
lytic deposition of nickel on platinum or copper. It is known that nickel
exists both in the active and passive states and that active nickel rapidly
changes over to the passive state if exposed to the oxygen of the air.
Pure nickel immersed in molal nickel sulphate gives about — 0.2 volt-
while according to Muthmann and Frauenberger' active nickel has a
potential of -t-0.32 volt. The nickel having this potential was prepared
by electrolytic deposition on platinum *' according to the directions for
electro-analysis.*' This statement, and that which follows later, ac-
cording to which the solution used was neutral nickel sulphate and the
impressed voltage eight volts, are all the directions given for obtaining
a nickel deposit giving the voltage quoted above. The first measure-
ments of the nickel couple were attempts to reproduce this poten-
tial in a molal solution of neutral nickel sulphate. The measurements
were made by the Poggendorf method with a Weston standard cell and
Lippmann electrometer. The potential between the nickel sulphate and

* See Tms Journal, 28, 733.

" All single potentials in this paper refer to the " normal electrode" as —0.560
volt at 20°.

* Sitzungsber. kgl. Bayr. Akad. Wiss., 34, 201 (1904).

FREE ENERGY OF NICKEL CHLORIDE. 717

normal potassium chloride was eliminated by a saturated solution of
potassium chloride or ammonium nitrate placed between the two.^ The
potential was measured within thirty seconds after plating. It was
found that on electrolyzing a neutral sulphate solution at room tempera-
ture with eight volts a green deposit appeared on the cathode whose
potential was about that given by Muthmann and Frauenberger, and
which decreased rapidly. On heating a solution containing 60 grams
NiCl2.6H20 per liter to 90° and electrolyzing with 12.5 volts the nickel
came down in a black spongy form, and no green compound was visible.
The highest potential obtained was -I-0.112 volt. Many other solutions
were used, but none gave such high results as the neutral sulphate. For
example, a solution containing 60 grams NiS04.6H30H-2o grams i,NH4)2SO^
+ 40 cc. NH^OH per liter, electrolyzed at 20° with eight volts, gave a
deposit of nickel whose potential was +0.017 volt. This was a gray, firm
deposit, and could be easily duplicated. Our only explanation of Muth-
mann and Frauenberger's high value for nickel is that some of this green
precipitate came down with the nickel. If this nickel was obtained in a
cold solution with eight volts, this is certain to have happened. The
nickel deposits obtained in the hot solution did not maintain a constant
voltage, but fell off as rapidly as in the case of the green precipitate ob-
tained in the cold. Attempts to increase the voltage by polarizing with
hydrogen were not successful. In preparing the nickel electrodes for the
final measurements the anode and cathode were separated by the wall of
a porcelain cup and in all cases pure electrolytic nickel anodes were used.
This cup was to protect the cathode from any sulphuric acid set free
at the anode, as it was found, as stated by Muthmann and Frauenberger,
that any free acid reduced the potential.
The actual measurement of the cell

Nickel I Sat. sol. NiCl2.6H20 | Pt + CI,

at 20® was now taken up. A saturated solution of the chloride was pre-
pared by rotating bottles for several days in a thermostat maintained
at 20° by means of a mercury regulator and electric lamps. The dis-
solved nickel was determined by electro-analysis and three determina-
tions gave 60.5, 60.6 and 60.9 grams of anhydrous salt per hundred grams
of water. This is somewhat less than 64 g. chloride to 100 g. water as
found by Etard.^ The density of this solution referred to water at 4°
was 1.474.

Before making measurements, chlorine was allowed to bubble over
the platinum-iridium plate for an hour or so, so as to give the solution
in this arm of the H tube time to get saturated with chlorine and to drive
out all air above the solution. After leaving the cell it was tested by

* Bjernim: Z. phystk. Chem., 53, 428 (1905).

' Landolt-BSmstein-Meyerhoffer Tables, 3rd Edition, p. 562.

7l8 M. DBEAY THOMPSON AND M. W. SAGE.

absorbing in potassium iodide in an inverted test tube. The absence of
bubbles showed the gas to be pure chlorine. The electromotive force of
this couple was measured against the ** normal electrode" and was found
to be — 1.59 volt. The nickel electrode was now prepared as described
above and changed from the plating solution to the cell as quickly as pos-
sible and the reading on the bridge taken immediately. This could be
done within thirty seconds from the time the plating was finished.
Cleaning ofif the sulphate adhering to the elctrode by dipping in water
did not affect the result. The electromotive force of the cell thus meas-
ured varied between 1.65 and 1.707 volts, the latter being the highest
value obtained in a large number of measurements. This, therefoie,
corresponds to the most active nickel obtained above. The pressure of the
chlorine gas at the time of the latter measurement was 752 mm. mercury.
As there is approximately 0.2 volt difference between the highest value
here obtained and that which would have resulted if the nickel couples
had as high a value as that assigned them by Muthmann and Freuen-
berger, some decomposition point determinations of the saturated sohi-
tion between small platinum plates were made for the purpose of deter-
mining whether the above measured value is reversible. The values
obtained were not used for computing the free energy, but were simply
to decide whether an error of 0.2 volt in the above value were possibte.
The values found were 1.61, 1.63, 1.60 and 1.60 volts. These are some-
what less than the highest value found above, but the agreement indi-
cates that 1.7 volts and not 1.9 volts is the voltage of the cell.

$• Vapor Pressures.

The vapor pressure of the saturated solution of nickel chloride has
been found by Lescoeur* to be 8.0 mm. Hg and that of the system
NiCla-eHaO— NiCl3.2H30 to be 4.6 mg, Hg, both at 20°. The system
NiCl,.2H,0 — ^NiCl, was found to have a vapor pressure at 100** of 123
mm. No measurement is given of this quantity at 20**, but the state-
ment is made that it is zero, which is, of course, impossible.

By means of the integrated van't Hoff formula this pressure could be
calculated, however, from the pressure at 100°, if the heat of hydration
of NiCl2.2H20 were accurately known. This is equal to the difference
between the heat of hydration of NiCl,.6H20 starting from NiCl, and
from NiClj.2H20. The former quantity is 20,330 calories.* The follow-
ing table shows the latter value, q, obtained from Lescoeur's vapor pres-
sure of the system NiCl2.6H20 — ^NiC1.2H20 by the formula:

4.S7iog»g = g.(^-^).

* Ann, chim. phys. [6], 19, 533 (1890).

* Landolt-BOrastein-Meyerhoffer Tabellen, p. 461.

.FREE ENERGY OF NICKEL CHLORIDE. 719

All except the last column are taken from Lescoeur.

Temp. Prca. in mm. Hg. qi.

'5 3-Tv 10540

250 6

14760

20 *»

4.6?
0.5J

30° 10

^l '♦-) 14000

15° 3

3°: '°-> .5470

40® 24.

Subtracting the average value of q^ from 20,330, the heat of hydration
of NiCl2.2H20 is obtained. Computing this from the value of the vapor
pressure of NiCl2.2H30 at 20°, 11.5 mm. results, which is greater than
that of NiCl3.6H20, and is therefore impossible. It is evident, there-
fore, that this value of the heat of hydration is not accurate enough for
this purpose.

The value of this pressure was obtained directly by the tensimeter
method. For this purpose one tensimeter was filled in June and
three in July.* One bulb contained phosphorus pentoxide and the
other the salt. The liquid in the manometer was cottonseed oil of spe-
cific gravity 0.9185. The tensimeters were placed on their side, care-
fully pumped out, and at the same time warmed. They were then left
several hours in connection with the pump and again pumped out and
warmed before sealing oflF. In October they were placed in a thermo-
stat at 20° and allowed to remain until January. The following table
gives the readings in mm. of oil:

Date. No. x. No. 2. No. 3. No. 4.

Dec. 30 II. 4 7.0 3.6 5.4

Jan. 7 6.3 broken 6.4

Jan. 25 II. 6 6.8 6.1

Mol water to I mol salt 2.06 2.05 .... i.99

Dec. 30, after the measurements, the bulbs of Nos. 2 and 5 containing
the NiClj were placed in a beaker of hot water for a few minutes to see
if the evolution of water vapor could not be accelerated. The pressure
increased to 10 to 15 cm. of oil, due evidently to the expansion of water
vapor already present, but went back to nearly the same values after-
wards. The water in the salt in the tensimeters was determined by ana-
lyzing for the amount of nickel present subsequent to these measurements,
and the results were those shown in the table above. It is evident from
the above that No. i was not properly filled, and is therefore omitted in
taking the mean. The average of Nos. 2 and 4 is 0.43 mm. of mercury.

In order to check this value by a different method it was decided to

> See Findlay, "The Phase Rule", p. 88.

720 M. DEKAY THOMPSON AND M. W. SAGE.

measure it at iio°and compute by the van't Hoff equation from this
and from Lescoeur's value at ioo°, the value at 20°.

For this purpose a tensimeter was filled with some nickel chloride that
had been in the drying closet for more than a week at 100®. Mercury
was used in the manometer and phosphorus pentoxide in the drying
bulb. It was first placed in a steam-bath to see that it checked with
Lescoeur's value. The tensimeter was left at four in the afternoon, and
next morning was found at 123 mm. Hg, reduced to zero degrees, at
which pressure it remained constant for the following twenty-four hours.
This agrees identically with Lescoeur's value. It was then placed in a
bath of vapor from boiling toluene and in a few hours reached a constant
pressure of 218 mm. Hg., at a temperature of 110.7° as obtained froma
thermometer corrected at this point by a standard from the Reichsanstak.
The value at 20° computed from this is 0.46 mm.

The tensimeter was then opened and heated while exhausted and con-
nected to the air pump for an hour and a half at a temperature varying
from between 90° and 120°, thereby driving off some water. It was
sealed off while hot and again placed in the bath of toluene vapor. The
constant pressure reached as before in a few hours was 216 mm. Hg., at
110.6°. The pressure computed from this measurement for 20° is 0.47
mm.

These values agree well enough for the present purpose with the aver-
age of those from direct measurements. The best value to adopt is the
average of the four which is 0.45 of mercury. It is interesting to note
that on cooling, the mercury in the tensimeter immediately goes back to
nearly a zero pressure, showing that water is taken up by the chloride
more rapidly than it is evolved.

Substituting in formula (3) E = 1.71 volts, p^i = 8.0, p^ = 4.6 and
p2 = 0.45, AF = — 74,400 cal. The correction due to the pressure of
the chlorine not being 760 mm. is negligible.

The heat of the reaction, or the increase in total energy, is — 74,500021.,
assuming the compound to be formed from gaseous chlorine. If the
chlorine were solid and superheated to this temperature, the total energy
increase is obtained by increasing the above, value by the sum of the heat
of fusion and the heat of vaporization of chlorine which has been esti-
mated at 7000 cal.* This gives — 67,500 cal. as the total energy change,
starting with solid chlorine. Likewise, if the chlorine were solid the
free energy increase AF would be diminished by the quantity RT, or
approximately 600 cal., giving the value F = — 75,000. The ratio of
free to total energy, for chlorine in the solid state, consequently is i.ii,
while the ratio of the same quantities with chlorine in the gaseous state
is 1. 00. The free energy and the total energy of this salt, therefore, are
* This Journai*, 28, 741 (1906).

NEGATIVE VISCOSITY. 721

approximately equal, as was found to be true in all other cases for which
this calculation has been made.

4* Precision Discussion,

The term 2EF in formula (3) amounts to 79,040 calories, the second
term to only 4600 cal., from which it is evident that even a large error,
in the vapor pressures would have only a relatively small effect on the
final result. Assuming an error of o.i mm. in />,^„ p^ and o.i mm.
in p, which seems an outside limit, the resultant effect is 600 calories,
or about 0.8 per cent, error in AF. Assuming a possible error of 0.02
volt in E the combined error in AF would be 990 cal., or 1.3 per cent.

5. Summary.

The free energy AF of NiClj was determined by measuring the poten-
tial of the cell.

Ni I Sat. sol. of NiClj.eHjO | Cl^ + Pt.

A formula for AF was deduced, involving this potential, the vapor pres-
sure of the saturated solution, of the system NiCl2.6H20 — ^NiCl2.2H30
and of NiCl2.2H20 — NiCl2. The pressure of the last system was obtained
both by direct and by indirect measurement. The free energy and total
energy of NiCl2 were found to be approximately equal.

Blbctrochbmical I^aboratoky,
Mass. Inst, op Tbch., Boston.

A STUDY OF THE SOLUTIONS OF SOME SALTS EXHIBITING NEGA-
TIVE VISCOSITY.

Prbdbrick H. Gbtman.
Received March 10, 1908.

In 1873 Hiibner* determined the viscosities of a series of solutions
of alkaline halides of equal densities and observed that some of the salts
diminished the viscosity of water.

Subsequently Sprung^ made an extensive study of the viscosities of
saline solutions between the temperatures 0° and 60° C.

He divided the salts examined into two groups as follows :

(i) KCl, KBr, KI, KNO3, KClOs, NH.Cl, NH.Br, NH.NOg.

(2) K2SO,, NaCl, NaBr, Nal, NaNOg, NaClOs, "^^^^O^, (NH,)2SO„
BaClj, SrClj, CaCl2, LiCl, MgSO^.

He pointed. out that at low temperatures the salts of the first group
lower the viscosity of water and at higher temperatures they increase
it. The salts of the second group always increase the viscosity of the
solvent, the viscosity of the solution becoming less as the temperature
is raised.

' Pogg. Ann., 157, 130.

^ Ibid., 159, I. . _.

722 FREDERICK H. GETMAN.

The experimental work of Sprung was later confirmed by the investi
gations of Slotte^ and Wagner,* the latter having studied the viscosities
of the solutions of forty different salts at several concentrations.

Arrhenius,* ICanitz^ and Miitzel* have measured the viscosities of saline
solutions which show the phenomenon of negative viscosity and still
more recently Ranken and Taylor* have made some extremely accurate
determinations of the viscosities of solutions of potassium chloride and
ammonium iodide at different temperatures.

One of the first to offer an explanation of negative viscosity was Euler,'
who made use of the theory of electrostriction proposed by Nemst and
Drude." In terms of this theory the ions are enveloped in a strong elec-
tric field, owing to their charges, and the intervening liquid is subjected
to great stress, so that the tendency of the ions to increase the viscosity
in inverse proportion to their speeds of migration is offset by the elec-
trostriction.

Wagner* has pointed out that Euler*s theory is untenable, siuce the
viscosity of the solvent may be diminished by solutes which are non-
electrolytes.

As a possible explanation of the phenomenon, Wagner suggests that
the amount of solvent in a given space is diminished by the solute and
that this leads to a lowering of the viscosity.

If the solute have a high viscosity, then the viscosity of the solution
will be greater than that of the solvent.

Ranken and Taylor*® have pointed out that at 8® urea diminishes the
viscosity of water, but as the temperature is raised the viscosity of the
solution becomes greater than that of the solvent.

Recently the author" determined the viscosities of several solutions
of potassium salts having lower viscosities than that of the slovent. The
determinations were extended from dilute to concentrated solutions
and it was found that in every case the viscosity-concentration curves
passed through a minimum.

The suggestion was put forward that the abnormal behavior of the potas-
sium salts resulted from the combined action of the ions and the undis-

• IVied. Ann., 30, 257.

' Z. physik. Chem.f 5, 31.

• Ibid., I, 285.

• Ibid,, 22, 336.

• Wied. Ann., 43, 15.

• Trans. Roy. Soc, Edinburgh, 45, 397.
' Z. physik. Chem., 25, 536.

• Ibid., IS, 79.

• Ibid., 46, 867.
»® Loc. cU.

" /. chim. phys., 5, 344.

NEGATIVE VISCOSITY. 723

sociated molecule. The potassium ion appearing to lower the viscosity
of the solvent, while the different anions and the undissodated molecules
tended to increase it.

Shortly after this paper was written Jones and Veazey* suggested a
possible explanation of negative viscosity.

In the course of their investigations the anomalous behavior of potas-
sium thiocyanate directed their attention to this phenomenon and they
found that the solutes which diminish the viscosity of water are those
whose cathions have large atomic volumes. If the ions of the solute are
large, relative to the molecules of the solvent, then the effect of the dis-
solved salt will be to reduce the viscosity of the solvent.

In view of what has been done in this field it has seemed of sufficient
importance to extend my work and include other saline, aqueous solu-
tions exhibiting negative viscosity.

The salts chosen for more careful study were ammonium chloride,
ammonium bromide, ammonium iodide, ammonium nitrate, and rubid-
ium iodide.

Apparatus and Method.

The method of measuring the viscosities was the well-known transpira-
tion method of Poiseuille-Ostwald. The viscometers were placed in^ a
bath of water maintained at constant temperature and the times of trans-
piration were measured with a stop-watch which had been carefully com-
pared with an accurate chronometer.

The densities were determined with a pycnometer of the Sprengel-
Ostwald type.

The viscometers were frequently cleaned with chromic acid and special
precautions were taken to protect them from dust.

The constants of the tubes were frequently checked by measuring
the times of transpiration of pure distilled water and solving for k in the
formula

where 17 is the absolute viscosity of water at a definite temperature, d
the density of water at the same temperature, and / the time of trans-
piration as measured by the stop-watch.

The values of 17 and d were obtained from the tables of Landolt and
Bomstein.

The concentration of the solutions was determined either by direct
weighing or by analysis.

The salts used were obtained from Kahlbaum and were sufficiently
pure to warrant using them without recrystallization.

* Am, Chem, J,, 37, 405.

724

FREDERICK H. GETMAN.

Results,

In the tabulation of results the symbols have the following signifi-
cance:

m = concentration of solution in gram-molecules per liter.

d = density of solution referred to water at 4° C.

77 = viscosity in C. G. S. units (dynes per square centimeter).

Tablb I.
NH4CI— 25°.

0.4437

I. 0071

0.00889

0.8874

I. 0138

0.00885

133"

1.0204

0.00882

I . 7748

1.0268

0.00880

2.2185

' I 0331

0.00878

2.6622

1.0394

0.00887

3 1059

1.0458

0.00897

3 • 5496

I. 0516

0.00904

4.437

1.0630

Tablb II. »
NH,C1.

0.00925

VicP.

i?30**.

VS^.

0.68

0.0128

0.0080

0.0057

1.62

0.0123

0.0081

0.0058

2.93

0.0120

0.0082

0.0061

4.34

0.0120

Table III.

0.0085

0.0064

NH,Br— 25°

■

fit.

0.216

1.0121

0.008867

0.432

1.0247

0.008796

0.647

1.0352

0.008756

0.863

I . 0468

0.008680

1.079

1.0583

0,008644

1.323

I. 0715

0.008575

1.588

I . 0858

0.008480

2.646

1.1414

0.008254

3.486

1.1758

0.008356

4-357

•

1.2273

0.008470

4.920

1.2605

Tablb IV.
NH^Br.

0.008560

i?io«>.

V3<P.

Vyfi.

1.63

0.0116

0.0077

0.0056

2.58

0.0112

0.0077

0.0057

3- 76

O.OIII

00079

00061

» Sprung, Pogg. Ann., 159, i,
* Sprung, Loc. cit,

NEGATIVE VISCOSITY.

725

m.
0.500
0.751

I.OOl

1. 501
2.002
2.502
3.002
4003
5 004

0.125
0.25

0.5
i.o

4.0

6.0

0.125
0.25

0.5
1.0

15
20
10

15
20

Table V.

NH4l--25^

I .0447

0.008577

1.0675

0.008440

I. 0913

0.008237

I. 1377

0.008086

I. 1839

0.007827

I . 2304

0.007779

1.2765

0.007842

1.3692

0.007997

I 4591

0.008321

Table VI.

NHJ— 30"*.

17.

1 .0071

0.00792

1.0181

0.00786

1 .0401

0.00774

I .0944

0.00753

1-3513

0.00729

1.5285

0.00823

Table VII. »

NHJ-45^

17.

1.0026

0.005950

I. 0145

0.005925

1.0380

0.005855

1.0853

0.005780

Table VIII.

NHJ.

1 .001

0944

O.OII46

0937

O.OI03I

0929

0.009277

2.002

1889

0.01044

1874

0 . 009568

1859

0.008752

3.002

2826

0.009947

2806

0.009297

.2786

0 . 008648

4.003

3763

O.OOOOI

3738

0.00931

3717

0 . 008692

5.004

.4661

I .01051

.4641

0.009761

.4611

0.009017

Ranken and Taylor, Loc. cit.

726

l^REDERICK H. GETMAN.

Tablb IX.

NH,NO,— 25^

•

i».

0.256

1.0059

0.008853

0.512

I. 0137

0.008797

0.767

I. 0215

0.008724

1.023

1.0300

0.008655

1.279

1.0373

0.008602

1-454

1.0432

0.008579

2.012

1.0602

0.008504

2.245

1.0676

0.008500

2.587

1.0777

0.008565

2.909

1.0873

0.008627

3.636

I . 1078

0.008850

4.664

I. 1396

Table X.*
NH,NO,.

0.009308

mcp.

V310P.

VsbP.

0.745

0.0124

0.0079

0.00567

1. 521

0.0119

0.0079

0.0057

3.367

0.0Z20

0.0085

0.0062

4.646

0.0128

0.0091

0.0070

6.22

0.0145

Tablb XI.
Rbl— 25^

0.0113

0.0088

0.264

I. 0371

0.008714

0.528

1 .0592

0.008489

0.792

I. 1225

0.008303

1.056

1.1738

0.008147

1.583

1.2475

0.007887

1.758

1.2662

0.0079x8

2.639

I . 4149

0.007796

2.931

I. 4612

0.007870

In Fig. I the tabulated values of viscosity X 10* are plotted as ordinates
against the corresponding concentrations as abscissae. For conveniens
in comparison, Fig. 2 gives similar data for the corresponding potassium
salts and for sodium chloride and sodium bromide. The curves in Fig.
I are for 25° while those in Fig. 2 are for 18°.

The viscosity-concentration curves preserve the same order for the
three bases, sodium, potassium and ammonium and apparently rubidium.
In all cases the chlorides have the greatest and the iodides the least
viscosity.

The ammonium salts depress the viscosity of water to a greater extent
than the corresponding potassium salts, while rubidium iodide produces
the maximum effect of all.
* Sprung, Loc, cU,

NBGATIVB viscositv.

In tenns of the theory put forward by the author the minima in the
viscosity-concentration curves are explained as due to equilibrium be-
tween the tendency of the cathions to diminish the viscosity of the sol-
vent and the tendency of the anions and the undissociated molecuks to
increase it.

At the minimum we find the solutions to be approximately 70 per
cent, ionized.

Viscosity and Tempcratore.

Early in the study of the viscosity of pure liquids and solutions it was
discovered that a slight variation in temperature produced a decided
change in the viscosity, generally the viscosity being diminished by rise
of temperature.

The attempts which have been made to express viscosity as a function
of the temperature have been but partially succes^l.

Slotte' found that the viscosities of a series of chromate solutions at
temperatures between 10° and 40" could be quite accurately determined
by means of the formula

' Witd. Ann., 14, a

'^JTl-^' ">

728 FREDERICK H. GBTHAN.

where a, /9, and ;■ are constants and ( is the temperature.
Subsequently the same investigator' proposed the formula

a and c being constants and the exponent n varying in value from u
to 1.9.

Fig.*.
From purely theoretical reasoning Graetz' suggested the formula

in which A is a constant, t^ and t, the critical and melting temperaluiw.
respectively, and t the temperature at which the viscosity of the liqu"
is sought. This formula has been found to give good results for puK
liquids within a limited range of temperature.
De Heen' has proposed the formula

' BeiM., 16, 182.

' Wied. Ann., 14, 15.

' BuU. Ac. Bets. (3), I

^[hM

NEGATIVE VISCOSITY. 729

where ij^j is the viscosity at 0° and A and n aft constants.

Of the several formulae enumerated those of Slotte and Graetz appear
to be most widely applicable.

The effect of change in temperature on the viscosities of the solutions
studied is shown in Figs. 3 to 6, in which are given a series of viscosity-
concentration isothermals.

Fig. 3.

The solutions of ammonium chloride, ammonium bromide and ammo-
nium nitrate become less and less negatively viscous as the temperature
is raised, until at 50°, the temperature of the highest isothermals, they
are all positive.

Ammonium iodide exhibits the same tendency but because of its marked
negative viscosity it is evident that a higher temperature must be reached
before the viscosity changes sign. The minimum in the viscosity -con-
centration isothermals shifts toward the more dilute region as the tem-
perature is raised.

Since the isothermals are not parallel, it is evident that the rate of
change of viscosity is not uniform at different temperatures.

The viscosities of tbennal solutions of the salts studied in this and a

rRBDBRICK H. GBTHAN.

Hg.4.

previous investigation' can be calculated with a fair degree o( accuracy
by formula (i) of Slotte.'

Tables XII to XXII give the observed and calculated values of ij (or
different temperatures together with their diflferences, A. In ewry case
the greatest difference occurs in the values for lo", the calcubtal
value being, with but two exceptions, too large. It is evident that the
formula ceases to be applicable below 15°.

Neglecting the values for 10° and eUminating several values which ait
known to be slightly in error, the average departure from the observed
value is 8 units in the fifth place of decimals.

The results for three salts, ammonium chloride, ammonium iodide
and potassium chloride, are shown graphically in Fig. 7, the observed
values being designated by a cross and the calculated by a small elide.
Tablb XII.
NaQ.

NSGAtlVB VlSCOSItY.

731

Tablb XIII.

NaBr.

a = 27.

^saO.OOII.

7- = 0.5523.

ir(ObMrved).

i|(C«lcuUted).

0.01358

0.01383

— 0.00025

0.00948

0.00952

— 0.00004

0.00853

0.00859

— 0.00006

0.00610

0.00607

TablB XIV.
NaNO,.

— 0.00003

a «= 27.

= 0.00103

r-=- 0.5523.

i7(Observed).

i7(CalcuUted).

0.0138

0.0139

O.OOOIO

17.

O.01118

0.01135

— 0.00017

0. 01 107

0.01066

+0.00041

0.00949

0.00959

— O.OOOIO

0.00866

0.00000

0.00617

0.00614

Tabt,k XV.
KCl.

+0.00003

0 = 30.

= 0.00101.

r = 0.05520.

^(ObMnred).

i7(Calculated).

0.01270

0.01279

— 0.00009

0.01055

0.01049

+0.00006

0.01004

0.00000

0.00902

0.00000

0.00814

0.00819

—0.00005

0.00580

0.00589

TabI.K XVI.
KBr.

—0.00009

a = 30.

j9 = 0.00124.

7- = 0.5520.

i7(ObMrved).

i7(CalcQlated).

0.01222

0.01256

—0.00033

0.01030

0.01026

+0.00004

0.00796

0.00000

0.00574

0.00566

Table XVII.
KI.

+0.00008

a «— 30.

^ SB 0.0015.

7"=- 0-5520.

^(ObMnrcd).

^(Calculated).

O.OI180

0.01230

— 0.00050

0.01007

O.OIOOO

+0.00007

0.00760

0.00770

— O.OOOIO

0.00553

0.00540

+0.00013

732

l^REDERICK H. GETMAN.

Table XVIII.
KNO,.
a = 30. /? = 0.0010.

t. i7(Ob8erved). ^(Calculated).

10 0.01242 0.01280

18 0.01044 0.01050

20 0.00995 0.01004

30 0.00820 0.00820

40 o . 00690 o . 00689

50 0.00585 0.00590

Table XIX.
NH.Cl.
a = 30. /? = 0.0011.

/. i7(Obsenred). i7(Calcalated).

10 0.01275 0.01270

17.6 0.01040 0.01049

25 0.00808 0.00893

30 0.00808 0.00810

50 0.00578 0.00580

r =

0.5520.

+0.00038
— 0.00006
— 0.00009
0.00000
+0.00001
— 0.00005

r =

0.5520.

+0.00005
— 0.00009
— 0.00005
00002
00002

Table XX.

NH^Br.

a — 30.

P — 0.0015.

r — 0.5520.

i7( Observed).

i7(Calculated).

0.01220

0.01230

— O.OOOIO

0 . 00866

0.00953

— 0.00087

0.00780

0.00770

+ 0.00010

0.00570

0.00540

Table XXI.
NHJ.

+0.00030

a — 30.

fi — 0.0018.

r — 05520.

i;(Ob8erved).

i7(Calcalated).

0.01146

0.01200

—0.00054

0.01031

0.01047

— 0.00016

0.00928

0.00924

+0.00004

0.00824

0.00823

+0.00001

0.00753

0.00740

+0.00013

0.00578

0.00556

Table XXII.
NH.NO,.

+0.00022

a = 30.

P = 0.0013.

r— 0.5520.

i7(Ob8erved).

^(Calculated).

0.01223

0.01250

— 0.00027

0 . 00866

0.00873

—0.00007

0.00790

0.00000

0.00568

0.00560

+0.00008

NEGATIVE VISCOSITY. 733

Visco^ty and Coaductivi^.
As early as 1856, G. Wiedemann' called attention to the fact that vis-
cosity and conductivity are related. From a study of solutions of cop-
per sulphate he was led to formulate the relation

*>; . .

— = constant,

where 7 is the viscosity and k is the conductivity of the solution whose
concentration is m.

Gotrian^ measured the viscosity and conductivity of solutions at differ-
ent temperatures, but failed to establish any clear relation between thtra.

FiR. 5-

lAter, Grossman* recalculated Gotrian's results and found that the
product of the viscosity by the conductivity was a constant, indepen-
dent of the temperature.

Several empirical relations have been pointed out by various workers,

' Pogg- Ann., 99, 339.

'Ibid., 157, 130.

* Wied. Ann., 18, 119.

734 PRBDBRICE H. GGTMAN.

among whom may be mentioned Arrhenius/ Euler,' Strindbeig* and
HoUand.*

In 1896 Moore,* in an investigation of the viscosities of sotne salt solu-
tions, attempted to find some relation between viscosity and comiuc-
tivity, but without much success. He makes the significant mnaik

Kg. 6.

that "more extended obsen^tions must be made upon the relation of

viscosity and conductivity, perhaps even some new method of compan-

son arrived at, before the two subjects are placed in their right relatkm."

The recent work of Bousfield and Lowry* has led to the formula

connecting conductivity and temperature, the parenthesis being Slottc's
expression for the change in viscosity with temperature.

» Z. ^ynk. Ch^m., 9, 4^7.

* Ibid.. 35, 536.

'/Wrf., 14,331.

*WM. Ann., 50, 261.

*Phfs. Rn., 3, 321.

' Ptoc Roy. Soe., 74, a8o.

NEGATIVE VISCOSITY.

It seemed of interest to compare the viscosities of the normal solutions
with their conductivities as given by Kohlrausch,' making use of Wiede-
mami's relation.

The results are given in Table XXIII, X denoting the equivalent c
ductivity.

Naa....
NaBr....
NaNO,. .

KBr

KI..

KNO,

NH,a

NH^r

NH^

NH,NO,

' Dot LtiivemBgen d*r EUktrotj/ti.

Yablb

XXIII.

Temp.

— 18".

0116

01 IS

0114

0105

0103

103
103

0104

0103

OIOI

1 03

00977

104

OIOI

736 FREDERICK H. GETMAN.

The values of i^X for the halogen salts of potassium and ammonium
are nearly constant, the average value being 1.032.

It is evident, however, that some other factor besides those given is
required to establish the relation between viscosity and conductivity.

Discussion of Results.

A survey of the results obtained with the potassium, ammonium and
rubidium salts examined seems to confirm the theory suggested by the
author, that the phenomenon of negative viscosity is due to a tendency
of the cathions to lower the viscosity of the solvent while the anions and
the undissociated molecule tend to increase it. Jones and Veazey* have
called attention to the fact that those salts whose cathions have the
greatest atomic volumes exhibit negative viscosity when dissolved, pro-
vided the atomic volumes of the anions are not so small as to counter-
act the effect of the cathions.

In terms of their theory we should expect, for a series of salts having
the same anion, that the lowering of the viscosity would vary directly
with the atomic volume of the cathion. These authors have shown
this to be approximately true for the normal solutions of the chlorides
of potassium, rubidium and caesium. From Figs, i and 2 we find that
rubidium iodide lowers the viscosity to a greater extent than potassium
iodide while the lowering produced by ammonium iodide is slightly less
than that produced by rubidium iodide. Owing to scarcity of material
it was impossible to extend the measurements on rubidium iodide to
more concentrated solutions.

Differences in the degree of dissociation at different temperatures
cannot be employed to explain the greater negative viscosity observed
in each case for the lower temperaures.

It is probable that the lower temperatures favor the formation of molec-
ular complexes which, owing to greater volume and smaller surface,
tend to diminish the viscosity.

Dunstan' has furnished experimental evidence for the formation of
these complexes, which are stable only at relatively low temperatures.

By means of this supposition it is possible to explain the case of nega-
tive viscosity presented by urea at 8° and other instances of negative
viscosity observed with solutions of non-electrolytes.

The empirical formula of Slotte for calculating viscosities at various
temperatures is found to apply to the solutions studied between 15° and
20°.

Assuming with Wiedemann that the migration velocity varies in-
versely with the viscosity for a fixed potential gradient, the product of

* Jones and Veazey, Loc. cit.

* J. Chem. Soc.f 85, 817; Z. physik. Chem., 49, 590.

ANHYDROUS CHLORIDES ON TELLURIUM. 737

the viscosity and conductivity should give a constant. It has been found
that this relation is only approximately true for the solutions investi-
gated.

Columbia Univbrsxtt, March, 1908.

THE ACTION OF VARIOUS ANHYDROUS CHLORIDES ON TELLURIUM

AND ON TELLURIUM DIOXIDE.

Bn Victor Lbnhrr.
Received February 25, 1908.

In an earlier paper ^ the action of sulphur monochloride on elementary
tellurium has been shown to result in the production of tellurium tetra-
chloride and sulphur, when the sulphur monochloride is in excess, while
Kraflft and Steiner,^ in studying this reaction, observed that when an ex-
cess of tellurium is heated with sulphur monochloride, tellurium dichlor-
ide results.

Further study on the action of tellurium and the dioxide with active
reagents has shown that with many of the anhydrous chlorides, especially
with those which are liquid at the ordinary temperature, tellurium tetra-
chloride is produced. In certain cases the tetrachloride immediately
separates from the solution in pure form while with a number of reagents
of this character actual union takes place and a crystalline condensation
product separates.

Tellurium Dioxide and Sulphur Monochloride, — Tellurium dioxide,
when treated with an excess of sulphur monochloride, is transformed
into the tetrachloride, sulphur dioxide being formed at the same time,
according to the equation:

TeOj + 2S3CI2 = TeCl, + SOj + 3S.

Analysis, TeCl4: Calculated, CI, 52.79; Te, 47.21.

Found, CI, 52.07; Te, 46.65.

When, on the other hand, an excess of tellurium dioxide is heated

with sulphur monochloride, the reducing action of this reagent steps in

and the result is that the dichloride of tellurium is formed, thus:

TeOj + S2CI2 = TeClj -f S + SO2.

The formation of tellurium tetrachloride by the action of excess of
sulphur monochloride on tellurium dioxide takes place readily; the re-
action can be materially hastened by warming, and under these condi-
tions, preparation of a large amount of the tetrachloride can be accom-
plished in a very short time. Extraction of the salt with carbon bisul-
phide is advisable in order to remove an excess of sulphur.

Behavior of the Oxy chlorides of Sulphur toward Tellurium and Tellurium

' This Journai^, 34, 188,
• Ber., 34, 560.

738 VICTOR LBNHER.

Dioxide. — ^With thionyl chloride, either the oxide of tellurium or the
metal yields tetrachloride when the anhydrous chloride is in excess and
dichloride when the element or oxide is in excess. These reactions are
quite similar to the action of sulphur monochloride on the metal or oxide.
Sulphuryl chloride reacts with the element forming tellurium tetra-
chloride, but when the metal is in excess the dichloride is the resulting
product. Tellurium dioxide in contact with sulphuryl chloride acts only-
very slowly, if at all, in the cold, but when the two are heated together
in a sealed tube, crystalline products are formed which, by anal3rsis, ap-
pear to be due to the simple solution of the dioxide in the sulphuryl
chloride and subsequent recrystallization of condensation products.
The end products of the reaction are dependent on the reacting masses
and on the pressure at which the reaction is carried out.

TeUurium and Thionyl Chloride^ with C. W. Hill. — ^Tellurium and
thionyl chloride, when heated together in a sealed tube for twenty-four
hours, yield crystals of the tetrachloride. The supernatant liquid was
dark colored and was found to contain tellurium.

Analysis, TeCl*: Calculated, Cl, 52.79; Te, 47.21.

Found, Cl, 52.00; Tc, 47.47» 46.60.

In studying the action of thionyl chloride on tellurium it was found
most satisfactory to heat the metal in a porcelain boat in a current of
the vapor of thionyl chloride, which was produced by boiling thionyl
chloride in a distilling bulb. The boat in which the tellurium was con-
tained was placed in a combustion tube connected directly yrith the dis-
tillation bulb, and the combustion tube was drawn out into a series of
bulbs. This apparatus was chosen inasmuch as the armngement allowed
the products of the reaction, by carefully regulating the temperature,
to be carried from bulb to bulb, and in the first bulbs practically all of
the sulphur formed in the reaction was retained, allowing the tellurium
tetrachloride to be collected in the bulbs farthest from the boat.

Anal3rsis, TeCl4: Calculated, 01,52.79; Te, 47.21.

Found, Cl, 52.80, 52.20; Te, 47 01, 46. 95-

The reaction of thionyl chloride on tellurium may be expressed,

2SOCI3 + Te = TeCl, + SO, + S.

Thionyl Chloride and Tellurium Dioxide, with C. W. Hill.— By heating
thionyl chloride with tellurium dioxide for twenty-four hours in a sealed
tube, long slender crystals of tellurium tetrachloride were obtained.
When the tubes in which the reaction was carried out were opened sul-
phur dioxide was liberated.

Analysis, TeCl«: Calculated, 01,52.79; Te, 47.21.

Found, I, Cl, 52 . 3ii 52 . 71 ; Tc, 47 • i^t 47 64.
" II, Cl, 52.55, 52. 13; Tc, 47.55, 47.30,

ANHYDROUS CHLORIDES ON TELLURIUM. 739

On heating the dioxide in the vapor of thionyl chloride, tellurium
tetrachloride is also readily obtained. Analyses II, above.

StUphuryl Chloride and Tellurium, with C. W. Hill. — Tellurium and sul-
phuryl chloride, when heated together in a sealed tube, )deld tellurium tetra-
chloride. On heating tellurium in the vapor of sulphuryl chloride, the metal
is first transformed into black dichloride, after which it is further chlorin-
ated to the tetrachloride by the action of excess of the sulphuryl chloride.

Analysis, TeCl^ ; Calculated, CI, 52 . 79 ; Te, 47 . 2 1.

Found, CI, 52.51, 52.36; Te, 47.14, 47 30- •

StUphuryl Chloride and Tellurium Dioxide, with C. W. Hill. — Tellurium
dioxide acts quite differently toward sulphuryl chloride than it does with
thionyl chloride or sulphur monochloride . When heated in the vapor of sul-
phuryl chloride to as high a temperature as it is possible to conduct an ex-
periment in hard glass combustion tubing, there is apparently no action.
No volatilization of the tetrachloride occurs, and in our experiments,
when a weighed amount of tellurium dioxide was heated in the vapor
of sulphuryl chloride to as high a temperature as the glass would per-
mit, the boat and contents weighed precisely the same after this treat-
ment as before. This would indicate that under these conditions the
vapor of sulphuryl chloride is without any appreciable action on tellu-
rium dioxide.

When, however, tellurium dioxide and sulphuryl chloride are brought
together in a sealed tube and heated, the system undergoes complex
changes. When the dioxide was heated in a sealed tube with sulphuryl
chloride for twenty-four hours at 165°, a small amount of a crystalline sub-
stance was formed, but the larger part of the dioxide remained unchanged.
The heating was continued for seventy-two hours when the dioxide com-
pletely disappeared, the liquid having assumed a dark yellow color. On
allowing the tube to cool, crystals deposited which were entirely unlike
tellurium tetrachloride. When the tube was opened so much sulphur
dioxide had been formed that the major part of the contents were lost
when the pressure was released on opening the tube. The portion saved
showed a content of 40.38 per cent, of chlorine.

The experiment was repeated a number of times, the tubes being heated
for forty -eight hours to 175°. In each case the tube was strongly cooled
before opening and in each tube a crystalline product was obtained.

Analysis showed the crystalline substances formed to vary in composi-
tion. Their composition by the analytical data obtained may be indi-
cated as follows :

3TeO„ 4S0aCla; 5Te08, qSOjCI,; TeO,, 2S0,C1,; 2TeO„ 5SO,Cl,.

Sulphuryl chloride appears to have no action on tellurium dioxide
at the ordinary temperatures or when the dioxide is heated in the vapor

740 VICTOR LHNHBR.

of sulphuryl chloride, but when the two are heated together to high tem-
peratures in a sealed tube, condensation products of varying composi-
tion are obtained.

In each experiment it was found that the excess of sulphuryl chbride
contained large quantities of tellurium.

Genend Remarks.

In studying the action of the chloride of sulphur or the oxychlorides
of sulphur it is observed that not infrequently one of the products of the
reaction .along with tellurium chloride is sulphur dioxide. Frequently
in our experiments, in sealed tubes, large volumes of sulphur dioxide
escaped on opening the tubes, while in the tube itself pure white tellurium
chloride remained. This is an interesting illustration of mass action,
as well as of conditions in the system, since sulphur dioxide readily pre-
cipitates tellurium from an aqueous or add solution of the chloride, and
the dioxide when heated in a current of sulphur dioxide gas is readily
reduced to metal.

While the action of the various chlorides of sulphur on tellurium or on
tellurium dioxide goes on readily, a number of other anhydrous chlorides
act similarly but not infrequentl}'^ by-products are formed which con-
taminate the chloride of tellurium which is produced. For example,
arsenic trichloride or antimony trichloride react with tellurium dioxide,
yielding tellurium tetrachloride and arsenic trioxide or antimony tri-
oxide. In these cases it is very difficult to obtain the tellurium chlor-
ide in a high degree of purity. Lead tetrachloride will convert either the
element or the oxide to tellurium tetrachloride. Here again an impure
salt is formed and a mixture is obtained from which it is difficult to ob-
tain pure tellurium salt.

Phosphorus trichloride, in contact with the dioxide of tellurium, quickly
reduces it to the elementary condition, and when brought in contact
with the element in pure condition is without action.

Carbon tetrachloride is without action on tellurium or on tellurium
dioxide even when allowed to be in contact with either of them for a
great length of time.

Tellurium Dioxide and Antimony Pentachloride. — Antimony penta-
chloride reacts with tellurium dioxide in the cold, the action of heat ma-
terially hastening the reaction, with the formation of the compound
TeCl^SbClj.

Analysis, TeCVShCl*: Calculated, Te, 22.49; Sb, 21.16; CI, 56.35.

Found, Te, 22 .02; Sb, 21 . 74; CI,

The action of antimony pentachloride on tellurium dioxide can be con-
sidered as first resulting in tellurium tetrachloride, which immediately
unites with an additional molecule of antimony pentachloride, forming
the addition product TeCl4SbCl6:

THE HOMOGENEITY OF TElrLURIUM. 74 1

(a) 5TeO,+4SbCU=5TeCl4 + 2SbA;

(b) TeCl^ + SbCl^ = TeCl4,SbCl5 ;

or sTeOj + 9SbCl5 = 5 [TeCl^.SbCy -h 2Sb A-

The compotmd TeCl4.SbCl5 appears in white tabular crystals which are
leadily decomposed by water.

Phosphorus Oxychloride and Tellurium Dioxide, — ^When phosphorus
oxychloride is brought in contact with tellurium dioxide and the two are
allowed to remain together in a warm place, a crystalline mass of large
flat monoclinic plates begins to form. The excess of phosphorus oxy-
chloride can be readily removed by means of carbon bisulphide. These
crystals are deliquescent and are readily decomposed by water.

Analysis, TeCl4.POCl,: Calculated, Te, 31.32; P, 7.61; D, 61.06.

Found, Tc, 30.22; P, 790; Cl, 59- 13-
While ^iphosphbrus oxycMloride doubtless first reacts with tellurium
dioxide to form tellurium tetrachloride, the tetmchloride, as soon as pro-
duced, unites with one molecule of the excess of phosphorus oxychloride
forming the addition product. The first reaction could be indicated:

sTeO, + 4POCI8 = 3TeCl4 + 2P A»

and the entire reaction may be expressed:

sTeOj + 7POCI3 = 3[TeCl, . POCIJ + 2P A-

When the above reaction is carried out with an excess of phosphorus
oxychloride, a considerable portion of the reaction product remains in
solution in the excess of the reagent.

It thus appears that by the action of the various anhydrous chlorides
which have been studied on tellurium or tellurium dioxide, three series
of products can form; either tellurium tetrachloride or the dichloride can
be produced, or by the n^e of such compounds as phosphorus oxychloride
or antimony pentachloride, double chlorides can be obtained.

UNIVBR8ITT OF Wisconsin,
Madison, Wis.

THE HOMOGENEITY OF TELLXTKIUM.

By Victor I^bnhbr.
Received Pebruaiy 34, 1908.

When tellurium or its dioxide reacts with a number of the liquid anhy-
drous chlorides,* crystals of tellurium tetrachloride or of a double chloride
are formed along with a mother liquor which contains the excess of the
reacting anhydrous chloride. This mother liquor contains such by-
products of the reaction as may be soluble in it, and has also been found
invariably to contain greater or less quantities of tellurium.

When sulphur monochloride is the reacting liquid, the amount of tel-

* See preceding paper.

742 VICTOR LENHER.

luriutn appealing in the excess of the reagent is quite small, so small in-
deed that a fairly good separation of tellurium from selenium can be
effected by use of this reagent in the cold, inasmuch as selenium dissolves
readily in sulphur monochloride. When, however, the sulphur chloride
is warm, its solvent power on tellurium chloride is materially increased.

Other chlorides studied, such as thionyl chloride, sulphuryl chloride,
phosphorus oxychloride and antimony pentachloride, when in excess,
dissolve considerable quantities of tellurium chloride or of the double
chloride as the case may be.

Inasmuch as the question of the complexity of tellurium has been
repeatedly raised, especially so after the prediction of Mendel&ff in his
London address,^ when he predicted that tellurium contained another
element, whose atomic weight would be 212 and which he called dvi-tel-
lurium; and after the work of Brauner,^ when he thought from some of
his experiments that the so-called element was actually a composite,
it has been deemed important to examine these mother liquors, arismg
from the reactions above indicated, to ascertain whether any of the re-
agents studied would possibly allow of a splitting apart of tellurium.

Norris, Fay and Edgerley' studied the elementary character of tel-
lurium by carrying out an elaborate fractional crystallization of the doubk
bromide of tellurium and potassium. Later Norris^ made fractional
distillations of tellurium oxide, following the distillations by determina-
tions of the atomic weight. The conclusions drawn by Norris are that
his results have furnished positive evidence in favor of the elementan'
character of tellurium.

Baker and Bennet* have recently published an account of their investi-
gations on the atomic weight and elementary character of tellurium.
Their work was carried on for a period of thirteen years. They studied
the fractional crystallization of telluric acid, the progressive solubility
of barium telliirate, the fractional distillation of the metal, chloride and
dioxide, the fractional decomposition of the hydride, the fractional
electrolysis of the. bromide and chloride and the fractional precipitation
of the chloride by water. They have been unable to distinguish any
difference in the atomic weight and consider 127.6 to be the true atomic
weight.

Study of the Products of the Reaction of Various Anhydrous Chlorides
on Tellurium and lis Oxide: — ^The mother liquor arising from each of
the reactions studied was decomposed by water, and the tellurium pre-

* /. Chem. Soc, 55, 649.
' /Wd., 55, 411.

' Am. Chem. /., 23, 105.

* This Journal, 28, 1675.

* J. Chem. Soc.f 91, 1849.

THE HOMOGENEITY OF TELLURIUM. 743

cipitated from hydrochloric acid solution by sulphur dioxide, after which
it was carefully purified from such materials as might be introduced by
the reagents used.

In each case the tellurium obtained was found to answer to all of the
tests to which the original material responded and behaved in all respects
similarly to what we commonly know to be tellurium.

The method of approaching this problem can be illustrated by the ex-
periments in which the action of phosphorus oxychloride on tellurium
dioxide was studied. In this case a crystalline compotmd is found which
careful analysis has shown to have the composition TeCl4.P0Cls. The ex-
cess of phosphorus oxychloride used was observed to carry considerable
quantities of tellurium in solution. This mother liquor was carefully
decanted from the crystallized salt and treated with water. By this
procedure, the phosphorus oxychloride is converted into phosphoric and
hydrochloric acids and the tellurium tetrachloride decomposed into hydrated
dioxide.

After addition of sufficient hydrochloric acid to completely dissolve
the tellurium dioxide, treatment with sulphur dioxide yielded elemen-
tary tellurium. It has been observed in working with this particular
liquid that when an attempt is made to precipitate tellurium out of a
solution containing a large quantity of phosphoric acid the reaction is
very much retarded and the tellurium precipitates very slowly. This
can be overcome, in a large measure, by the addition of hydrochloric acid
and considerable dilution with water. The retardation of precipitation
of tellurium in presence of phosphoric acid is doubtless due to lack of
dissociation and the precipitation can better be effected after addition
of considerable quantities of hydrochloric acid.

The tellurium, after having been converted into chloride by means of
aqua regia, was again precipitated by sulphur dioxide, converted into
the basic nitrate and crystallized from nitric acid solution, as has been
suggested by Norris, Fay and Edgerley^ for the purification of tellurium.
The basic nitrate, when ignited, gave the dioxide.

Determination of the amount of tellurium in the dioxide by precipita-
tion as element gave the ratio

0.85635 TeO, 0.6845 Te 127.52 at. wt. Te.

The tellurium thus obtained from the excess of phosphorus oxychlor-
ide showed by the ratio of tellurium dioxide to element that no apparent
change in atomic weight had taken place, and that no splitting apart
of the element had occurred was evidenced by the fact that the salt
reCl^-POClj which did not dissolve in the excess of phosphorus oxy-
chloride, and which contained all of the tellurium not appearing in the

* Am. .Chem. /., 23, 105,

744 VICTOR I^ENHER.

mother liquor, showed by analysis a rational formula with tellurium
taken at the commonly accepted atomic weight.

The mother liquors appearing in all the reactions of the anhydrous
chlorides which were brought in contact with tellurium or the dioxide
were examined in a manner similar to that with phosphorus oxychloride
and the same character of results were obtained. A rational formula
for the insoluble crystalline product always appeared when a careful
analysis was made, while the tellurium appearing in the mother liquor,
after decomposing the excess of reagent and purifying the element from
such products of the reaction as might contaminate it, always showed
the ratio of tellurium dioxide to element corresponding to the atomic
weight of 127.5 for tellurium.

Tellurium and Ferric Chloride. — By heating together in the dry con-
dition tellurium and anhydrous ferric chloride, reaction ensues with the
production of ferrous chloride, and tellurium tetrachloride. This reac-
tion can also be accomplished and much more conveniently so by the
continued action of a hot solution of ferric chloride on tellurium,

4FeCl3 + Te = TeCl, + 4FeCl2.

Pethybridge* has studied the action of a solution of ferric chloride having
a density of 1.18 on telluride gold ores and claims to be able to ejctract
the tellurium as tetrachloride with the simultaneous formation of ferrous
chloride, leaving behind gold and silver as follows:

SFeClg + ( AuAg)Te, = 2TeCl4 + 8FeCl, + (AuAg) .

This is in line with the results previously obtained by the author,' whidi
indicate that the tellurium in the natural tellurides of gold is not strongly
united to the gold.

While tellurium reduces ferric chloride to ferrous with the simultaneous
production of the tetrachloride, ferrous salts will precipitate a part of
the telluriuq;! when added in excess to a solution of the tetrachloride in
hydrochloric acid.

Crane' and Rose* have attributed this precipitation of small quanti-
ties of tellurium by means of ferrous sulphate to the presence of tellurium
dichloride in the tetrachloride. Tellurium dichloride is, however, de-
composed by water or aqueous hydrochloric acid into the element and
oxide or chloride, and with the knowledge that tellurium dissolves in
hydrochloric acid in the presence of the air to tetrachloride directly,
it would seem that the precipitation of small quantities of tellurium
by means of ferrous salts is largely controlled by mass action. The rc-

* U. S. Patent No. 709,037, Sept. 16, 1902.

* Tms Journal, 24, 355.

' Am, Chem. J., 23, 408. ^

* Pogg, Ann., 21, 443. _ _ : .. - .

THB HOMOGENEITY 01^ TELLURIUM. 745

actions between ferric chloride and tellurium and ferrous salts and hy-
drochloric acid solutions of the tetrachloride are indeed reversible.

In view of this partial precipitation of tellurium by ferrous salts, a
series of experiments was conducted, in which this partial precipitation
was made the basis of a fractionation. To this end a quantity of tellur-
ium originally obtained from the Baltimore Smelting and Rolling Co.,
was precipitated from a caustic soda solution by means of grape sugar.
The carefully washed material was dissolved in aqua regia and was pre-
cipitated by sulphur dioxide from a hydrochloric acid solution. The
tellurium thus obtained was fused with pure potassium cyanide, the re-
sulting fusion extracted with water and a current of air passed through
the solution to precipitate the tellurium out of the solution of the alkaline
telluride, leaving sulphur or selenium present in solution as alkaline
thiocyanate or selenocyanate. The tellurium purified in this manner
was introduced into a porcelain boat which was placed in a porcelain
tube. This tube was connected with a supply of hydrogen, which was
generated from pure zinc and hydrochloric acid and which was washed
with water. The tellurium was distilled from the boat (the tube being
heated by a blast lamp) into the cooler portion of the tube where it was
condensed. The distilled metal was collected and introduced again into
a clean porcelain boat and tube and the distillation repeated.

Twenty-five grams of this metal were dissolved in a mixture of nitric
and hydrochloric adds and a hydrochloric acid solution of the tetra-
chloride formed by boiling off the excess of nitric add. To this solution
was added a solution containing about 25 grams of ferrous sulphate,
which contained a small amount of ferric salt, when a precipitate of about
0.1 gram of elementary tellurium appeared. The quantity of precipi-
tate obtained did not appear to be materially increased by the addition
of a considerably larger quantity of the ferrous sulphate solution. The
solution containing the tellurium and iron salts was treated with add
sodium sulphite and the tellurium predpitated. This tellurium was
then reconverted into chloride by means of hydrochloric and nitric acids
and ferrous sulphate again added to the hydrochloric acid solution. A
second predpitate of about the same size as before, appeared. This opera-
tion was repeated twenty-four times, fractions being obtained each time
of approximately the same size. These twenty-four precipitates were
combined, dissolved in hydrochloric and nitric acids and a hydrochloric
add solution obtained as before. Treatment of this solution with fer-
rous sulphate again yielded a small precipitate of tellurium, the size of
the predpitate being about the same as in the previous series of
fractions. A second predpitation was made and was combined with
the previous one for an atomic weight determination. The metal was
treated with nitric acid in a small Jena flask, and the resulting basic ni-

746 VICTOR I^ENHER.

trate gradually heated to 440^, when the weight was found to be con-
stant.

0.1694 Te gave 0.21 19 TeO, and if O = 16, Te = 127.55.

Inasmuch as the original material showed the same atomic weight,
it is evident that by this method of fractionation no portion can be found
which shows a figure radically different from the commonly accepted
figure.

Action of Hydrochloric Acid on Tellurium in Presence 0} Air. — Although
tellurium is ordinarily considered as insoluble in hydrochloric acid, yet
it is actually attacked slightly by the acid when exposed to the air, if
sufficient time is allowed for contact. This action can be demonstrated
by bubbling a current of air, for several weeks, through concentrated
hydrochloric acid in which is suspended metallic tellurium. In a com-
paratively short time the acid becomes yellow, indicating the presence
of tetrachloride.

This action of hydrochloric acid on tellurium in the presence of air has
been made the basis of an experiment in which it would be possible to
have differential solution take place should tellurium contain a higher
member of the series of elements in which it is commonly placed.

One hundred grams of carefully purified tellurium was precipitated
from hydrochloric acid solution by means of sulphur dioxide and the
precipitate washed until the wash water would not react for chloride
with silver nitrate. This freshly precipitated and finely divided metal
was introduced with pure concentrated hydrochloric acid into a filter-
ing flask and a current of air which had been previously washed with
water and then passed through pure hydrochloric acid, was drawn through
the solution for three months. At the end of this time a considerable
part of the metal had dissolved. The solution was filtered and on treat-
ment of the solution with acid sodium sulphite, tellurium was precipi-
tated. Two portions of tellurium were here obtained, one of which
was a smaller portion of a few grams obtained from a tellurium solutiwi
that had been formed by the slow action of hydrochloric acid and air
on tellurium, and this solution had remained in contact with the main
part of the finely divided metal for three months. The portion which
dissolved and that still remaining undissolved were found to be identical
in chemical character, obeyed the same reactions, and when converted
into oxide, and this oxide analyzed, gave an atomic weight of 127.5.

In the hands of the author, tellurium has shown no signs of breaking
apart in any of the reactions which had for their object a study of the
elementary character of the metal, nor in the various reactions or de-
rivatives which have been studied in our laboratory has the element
shown any indications of other than simple character.

The statements of Mendelfeff as to the possibility of tellurium contain-

POTASSIUM CHLOROPLATINATE. 747

ing another element and his views as to its incorrect atomic weight have
led to an extended series of critical studies on this subject. Brauner
studied the atomic weight for six years, Norris for nine, Baker and Ben-
net for thirteen, and the author for ten. We see, therefore, thirty-eight
years of experimental work directed to the study of the atomic weight
of an element which to-day remains as an element whose elementary
character is well established, and which stands with a higher atomic
weight than the next element in the horizontal series. While it still
remains an exception to the periodic arrangement of Mendel^eff , and so
far as its comparison with iodine is concerned, is abnormal, yet in the
main points of its chemical behavior and those of its compounds, it prop-
erly belongs associated with sulphur and selenium in the sixth group.

The element tellurium still remains, however, an exception to the
periodic arrangement of Mendel6eflf, its atomic weight being higher than
that of iodine. We must hence conclude that either tellurium is abnor-
mal in a direction which has not yet received careful study, or that we
do not yet appreciate all of the principles of the periodicity of the ele-
ments.

Univbksitt of Wisconsin,
Madison, Wis.

A STUDY OF THE SOLUBILITY OF POTASSIUM CHLOROPLATINATE.

Bt B. H. A&CBIBAI.D, W. G. Wilcox and B. G. Bucklbt.

Receiyed March 9, 1908.

The importance of knowing accurately the solubility of a precipitate,
in the liquor from which it has been thrown down, if the precipitate is
to be used for a quantitative estimation, is apparent to every one. This
is the case, in particular, with such substances as potassium chloroplatinate,
when this salt is to be used for the determination of the amount of potas-
sium occurring in feldspars and such igneous minerals, where the amount
of chloroplatinate which is to be weighed is, generally speaking, compara-
tively small. The difficulty here encoimtered is, in a measure, overcome
by the methods worked out by Hille brand ^ and his pupils, who evaporate
the solution containing the salt in question, together with the correspond-
ing sodium salt, to dryness and then dissolve out the sodium salt with
absolute or eighty per cent, alcohol. The trouble, however, although
lessened, is still of the same nature as before, and even imder the best
conditions some of the potassium salt must be carried into solution,
grsing a high value to the sodium. With these features of the case in
mind, it was thought that it would be of some interest to study the solu-
bility of the potassium chloroplatinate somewhat more carefully than had

»HiUcbrand, "Analysis of Silicate and Carbonate Rocks," Bull. 305, U.S. Geol.
Survey. Washington : Analysis of Rocks.

748 ^- H. ARCHtBAtD, W. C. WitCdX AND 6. G. BtTCEt^Y.

hitherto been done, under somewhat similar conditions to those encoun-
tered in precipitating this salt from solution in ordinary analytical work.
The experiments which have been carried out with this end in view, are de-
scribed in the following pages.

Bunsen and Kirchhoff ^ were the first to work on the solubility of the
chloroplatinate of potassium. They made measurements of its solubility
in water, at various temperatures between 0° and 100°. Crookes,' when
studying the solubility of thallium salts, determined the solubility of
potassium chloroplatinate in water at 15^ and 100®. Fresenius' deter-
mined the solubility of this salt in solution of ethyl alcohol and water,
which contained 55, 76 and 97.5 per cent, of the alcohol. Gibbs,* while
working out the separation of the platinum metals, found that "potas-
sium chloroplatinate is insoluble in strong, cold, aqueous solutions of
potassium chloride." Precht* determined the solubility of this chloro-
platinate in absolute alcohol and in 96 and 80 per cent, alcohol-water
solutions. He also made some very interesting observations on the
preparation of the pure salt. Peligot* studied the solubility of the salt
in question in solutions of ethyl alcohol and water, and he also made a
determination of the solubility in absolute methyl alcohol. No S3rsten]atic
study has been made of the solubility of the chloroplatinate in different
concentrations of potassium chloride, nor has the influence of other alco-
hols been ascertained. It was thought that it would be worth while
to obtain some data with regard to both of these points, to somewhat
extend the observations mentioned above, and also to ascertain the in-
fluence on the solubility of the chloroplatinate of the presence of sodium
chloride.

Apparatus. — ^A constant temperature bath was necessary, in whidi
the solutions could be maintained at a given temperature until they
should have attained complete equilibrium. The bath used had a capac-
ity of about 40 liters. It was insulated with a thick coating of felt
on the sides, and a thick board of asbestos on the top. It carried the
usual arrangement for keeping the solutions constantly agitated, a small
hot-air engine of the Hendrici form supplying the motive power for run-
ning the stirrer. The bath was supplied with an Ostwald gas regulator,
by means of which the temperature could be kept constant, if necessary,
within 0.1°.

All burettes and pipettes used were carefully standardized by weigh-
ing the water which they delivered.

* Pogg. Ann., 113, 372 (1861).
■ Chem, News, 9, 37 (1864).

• Anal. Chem, und Pharm., 59, 117.

* Am. J. Sci. (2), 31, 70.

» Zeii. anal. Chem., 18, 509 (1879).

• Monit. scient. (4), 6, 872 (1892). . . _^

POTASSIUM CHWROPI^ATINATE. 749

The tubes, which contained the solutions while they were in the bath,
were of hard glass and had a capacity of about 55 cc. The rubber stop-
pers used to seal them were boiled in dilute caustic soda for several hours
before using, to remove surface impurities.

Two thermometers, A and B, were used in the course of the work,
to indicate the temperature of the bath. They were standardized at
0® in a bath of melting ice ; again at the transition point of sodium sul-
phate,' 32.383°; and again at 100°.

Preparation of Materials. — The chloroplatinic acid used in preparing
the potassium chloroplatinate was prepared as follows: Platinum scrap
was boiled for several hours with c. p. concentrated hydrochloric acid,
washed with distilled water, and the boiling repeated for about the same
length of time with c. p. concentrated nitric acid, to remove surface im-
purities. After another washing the scrap was dissolved in aqua regia,
using only as much of the nitric acid as was necessary to dissolve the
platinum. The color of the precipitate formed with the platinum solu-
tion and potassium chloride showed that there must have been an appre-
ciable amount of iridium in the platinum scrap. To remove the iridium,
advantage was taken of the fact that ammonium chloroiridate is much
more soluble in water than ammonium chloroplatinate. 100 parts of
water at 20° dissolve only 0.666 part of ammonium chloroplatinate,
while 5.00 parts of ammonium chloroiridate will dissolve in 100 parts
of water. Accordingly, a somewhat dilute solution of ammonium chlor-
ide, that had been once recrystallized, was added slowly, with constant
stirring, to the chloroplatinic acid solution, care being taken not to pre-
cipitate quite all of the platinum. The ammonium salt thrown down
was of a very brilliant yellow color, while the chocolate color of the resi-
due left upon evaporation of the mother liquor, showed that a large part
of the iridium had remained behind.

After thoroughly washing the precipitate of ammonium chloroplat-
inate it was reduced in hydrogen to spongy platinum, and the platinum
again dissolved in aqua regia. At this point a slight black residue re-
mained, which would apparently not dissolve in the aqua regia, and which
we supposed to be iridium. The solution of the spongy platinum was
now precipitated as before with ammonium chloride, the precipitate
reduced, the spongy platinum dissolved in aqua regia, and again pre-
cipitated with the ammonium chloride. The mother liquor left from
this precipitation showed no sign whatever of the presence of iridium,
while the residue which was noticed when the first portion of spongy
platinum was dissolved did not appear after the second precipitation.

As a further proof of the absence of iridium from the product thus
^ Richards and Wells, Proc. Am, Acad. Arts and Sciences, 38, 431 (1902).

750 E. H. ARCHIBALD, W. G. WILCOX AND B. G. BUCKLEY.

prepared, one-half of the sample was taken and reduced, the spongy
platinum dissolved in aqua regia, and again precipitated with ammo-
nium chloride. The precipitate thus obtained was successively recrys-
tallized three times from water. The mother liquors from these recrys-
tallizations were combined with that from the precipitation, and the
whole evaporated. However, no indication of iridium appeared in the
solution or in the residue.

The final precipitate of ammonium chloroplatinate, which was now
assumed to be free from iridium, was reduced, the platinum dissolved
in aqua regia, and this was followed by two evaporations, carried almost
to dryness, in the presence of an excess of concentrated hydrochloric
acid, to remove all traces of nitric acid. Precht,' and later Dittmar and
McArthur,^ and more recently Noyes and Weber,' have shown how neces-
sary it is to get rid of all the nitric acid, before precipitating with potas-
sium chloride, if a precipitate of known composition is to be obtained. The
residue from the final evaporation was taken up with enough water to give
a solution containing about 5 per cent, of platinum, and this was added
slowly, with constant stirring, to a dilute solution of potassium chloride.

The potassium chloroplatinate prepared in this way was thoroughly
washed and carefully dried at a temperature of about 80°. It was kept
over anhydrous calcium chloride until used.

From the analysis of several portions of chloroplatinate prepared in
different ways, and from solutions of varying dilution, it appears that
there is considerable tendency for the precipitate to carry down with it
other substances from the solution. This point is receiving some atten-
tion in this laboratory. It is enough to say here that the precipitate
obtained as above did not contain an appreciable excess of the platinum
or of the potassium chloride.

The potassium chloride used in the above preparation, as well as that
used directly in making the solubility determinations, was prepared by
dissolving Kahlbaum's chemically pure product in distilled water and
precipitating this with hydrogen chloride gas, generated by boiling chem-
ically pure hydrochloric acid. The potassium chloride precipitated in
this way wafe twice recrystallized from water, washed, and dried at a tem-
perature of 1 10°.

The sodium chloride used was prepared in the same manner as the
potassium salt.

Methyl alcohol, free from acetone, was digested for nine hours with a
large amount pf fresh calcium oxide. It was then distilled. The dis-
tillate was treated with a fresh quantity of calcium oxide, and again dis-

* Loc. cii.

' Trans. Roy. Soc. Edin., 33^ 561 (1887).

* This Journal, 30, 13 (1908),

POTASSIUM CHLOROPLATINATE. 751

tilled. That portion coming over at 65.5° was collected and used in
these experiments.

The other alcohols, ethyl and isobutyl, were freed from water in a simi-
lar manner. Those portions distilling over at 78.4*^ and io6°, respec-
tively, were used.

These alcohols, after being dehydrated, were carefully protected from
the moist air of the laboratory.

The water used in this work was prepared by distilling the ordinary
distilled water of the laboratory, after adding a few crystals of potas-
sium permanganate and a few drops of sulphuric acid. As a further
indication of its purity, its electrical conductivity was found to be as
low as could be expected.

Measurements. — ^We will first consider the solubility of the salt in
water at different temperatures. The method of procedure was as fol-
lows: To the tubes were added about 50 cc. of water, and somewhat
more chloroplatinate than would dissolve at the particular temperature
at which the determination was to be made. It should be stated here
that two different determinations were made for every pqint investiga-
ted. The temperature of the bath was now raised to the required point
and the thermostat regulated so as to maintain the water at this tem-
perature. The tubes were now attached to the paddles of the stirring
apparatus and constantly rotated for from twelve to twenty-four hours.
At the lower temperatures, equilibrium is attained very slowly, and in
these cases the stirring was continued over night. Several experiments
had shown that the solutions would become completely saturated within
this time.

In order to determine how much of the chloroplatinate had been dis-
solved, after the required time had elapsed the stirrer was brought to
rest and a test tube raised in the bath, until the top stood just «ibove
the water. When the excess of salt had completely settled, which, with
the heavy precipitate we are here concerned with, takes place in a very
short time, the stopper was removed, and as quickly as possible 25 cc.
of the sohition were withdrawn with the calibrated pipette, and first
transferred to a tared specific gravity bottle and weighed. This pro-
cedure was necessary in order to obtain the data required to express the
results in terms of grams of salt dissolved in 100 grams of solvent. This
solution was then washed into a weighed platinum dish. After evapo-
rating carefully and drying at a low temperature, the dish and residue
were weighed. Further drying of the dish and contents was followed
by another weighing, and this treatment was repeated until the weight
became constant. This process was repeated with 25 cc. of the contents
of a second tube, and we thus had two measurements of the solubility

Weiffht of
KsPtCledUsoWed.

Temperature.

2«

0.48x2

i6«

0.6718

25**

0.8641

35^

1. 132

48 «•

1-745

752 E. H. ARCHIBALD, W. G. WILCOX AND B. G. BUCKLBY.

at this particular temperature. The mean of these two determinations
gives the values tabulated below.

The results of the determinations of the solubility of the chloropkti-
nate in water at different temperatures are shown below in Table I.^ These
values represent the amounts of salt, in grams, soluble in 100 grams of
water, all weights being reduced to the vacuum standard. For the pur-
pose of this reduction the specific gravity of the chloroplatinate was
taken to be 3.54:'

Tabls I. — Results Showing thb Solubility of Potassium Chloroplatinaib ux

Water at Different Temperatures.

Weight of
Temperature. KsPtCU diMolTcd.

59 *• 2.396

68® 2.913

ts*' 3.589

92® 4.484

These results are shown graphically in Plate I, where the temperatures
are plotted as ordinates and the weights of chloroplatinate dissolved as
abscissae. In order that the results might be compared with those of
other observers the amount of salt dissolved for the temperatures io°,
20 '^j 30°, etc., taken from the curve in Plate I, are given below, together
with those of Bunsen :'

Table II. — Results Comparing the Solubility Values for Potassium Chlobd-

PLATINATE in WaTER, OBTAINED BY PREVIOUS INVESTIGATORS,

WITH THOSE Obtained by the Authors.

Weight of KfPtCle diMoWed. Weight of KsPtCU disaolttd.

Temperature. Authors. Bunsen. Temperature. Authors. Bunsen.

o** 0.4784 0.74 60** 2-444 2.64

ID** 0.5992 0.90 70** 3055 3-^9

20® 0.7742 1. 12 80® 3-7II 3-79

30® 1. 000 1. 41 90° 4-360 4.45

40® 1-355 1-76 100° 5030 5.18

50® 1.865 2- 17

• It has been pointed out to us by Prof. W. A. Noyes that Dr. Weber found thai
a solution of potassium chlorplatinate became add upon boiling. We find that this
hydrolysis goes on even at ordinary laboratory temperatures although very slowly, the
reaction not being completed within five days. The effect of this reaction upon the
solubility appears to be very slight, as a solution of the salt saturated at 95^ ^
maintained at this temperature in contact with the solid phase for 36 hours, upon
cooling to 75® gave practically the same value for the solubility as a solution saturated
at 75® and not allowed to rise above this temperature. Several of the determinations
given in the above table were obtained by approaching the saturation point from both
directions. This hydrolytic reaction is being studied further at different temperatures
and we hope to have something further to commtmicate in the near future..

• Landolt-Bomstein-Meyerhoffer, Tabellen.

• Loc. cU.

POTASSIUM CHLOROPLATINATH.

753

Crookes* states that one part of this chloroplatinate will dissolve in
io8 parts of water at 15®, and in 19 parts at the boiling-point of water.

- 2 - —

^ JL

w y

^ y

U J.

W f-

V I

a 7-

J-

a A J-

\ ^

^ -/

|l 7 «^„ X

u \ / fV^t \

■ 3 /

s. /

"8 y

3# 2 '

al -. -, . . ..J

/^ n ^rT Krr> IT

rfit |l./499t^tS9l3if4W«fH52

It is at once apparent that the present results show an appreciably
lower solubility of the platinum salt in water, particularly at the lower
temperatures, than other investigators have found. It seems possible
that the salt with which they worked may not have consisted entirely
of the chloroplatinate, as Precht,* and quite recently Noyes and Weber*
have shown that it is very difficult to prepare pure chloroplatinic acid.
If the nitric acid is to be completely removed by evaporation, this pro-

* hoc, cit.

754 H. H. ARCHIBALD, W. G. WILCOX AND B. G. BUCKLEY.

cess must be carried out repeatedly, and in the presence of a large ex-
cess of hydrochloric acid. Any contamination which resulted from an
incomplete removal of the nitric acid would doubtless give high results
for the solubility of the chloroplatinate. It is possible, too, that the
more complete removal of the iridium in the present instance may also
have lowered the solubility slightly.

Before dismissing this point, it should be stated that ever\' care was
taken to have the solutions saturated with the chloroplatinate, at the
particular temperature at which the experiment was being made, be-
fore they were removed from the bath. In some cases they were rotated
in the bath for 50 hrs. before being analyzed.

It should be noted that the curve in Plate I is a continuous one over
the whole range of temperature investigated.

Solutions of Methyl Alcohol and Water. — Solutions of methyl alcohol
and water, as solvents, were next tried, the procedure being much the
same as in the previous case, except, that here, four parallel determina-
tions were usually made. As the experiments were being carried on
very near the temperature of the laboratory, it was thought better, al-
though perhaps not necessary, to filter the 25 cc. of the solution whidi
had been measured out with the pipette, before evaporating. It was
found that when the amotmt of alcohol in the mixture was as high as
thirty per cent., reduction of the chloroplatinate to platinum black took
place during the evaporation. This method of determination had there-
fore to be abandoned. In its place we used a colorimetric method as
follows: A standard solution of potassium chloroplatinate in pure water
was prepared, containing seven grams of the salt per liter of sohition.
The unknown solution to be determined was poured into a Nesskr tube
of 100 cc. capacity and distilled to the mark. Using the standard solu-
tion as a basis, another solution was prepared in a second tube, whidi
had the same strength of color as the unknown. The volume of the
standard solution required to produce this color, multiplied by the amount
of chloroplatinate in unit volume, gave the amount of salt in the original
unknown* Care was always taken to have the conditions the same in
the unknown as in the known solution. For instance, both solutions
always contained the same amounts of alcohol or potassium chloride as
the case might be ; they were also kept at the same temperature, which
was a little higher than the temperature at which the determination was
being made . It was found that by this method we could detect the pres-
ence of 0.0005 gram of potassium chloroplatinate dissolved in 100 cc.
of water, or one part of the salt in two hundred thousand parts of water.
It is believed that this method gives a degree of accuracy in the deter-
mination of a number of points, in particular those dealing with the
dilute solutions, that has not hitherto been reached.

POTASSIUM CHLOROPI^ATINATE. 755

With regard to the mixture of ethyl alcohol and water, as well as those
of isobutyl alcohol and water, the determinations were carried out in all
respects like those described above.

The results for the solubility of potassium chloroplatinate, in various
concentrations of methyl, ethyl and isobutyl alcohol and water, are
given in Table III. The measurements were all carried out at a tempera-
ture of 20°. In the case of isobutyl alcohol and water only two measure-
ments were made, as this alcohol only dissolves to a small extent in water.

In order that these values might be comparable with those obtained
by other investigators, the composition of the solutions is expressed in
grams of alcohol per one hundred grams of solution, while the amounts of
salt dissolved are given in grams per one hundred grams of solvent. As
before, all weights are referred to a vacuum.

Tablb III. — Results Showing thb Solubility of Potassium Chloroplatinatb in

Solutions of Different Alcohols in Water.

Per cent. Weight of Per cent. Weight of Per cent. Weight of

ofmethvl KtPtCleln of ethyl KsPtClein of isobutyl KtPtClein

alcohol. 100 gms. solution. alcohol. 100 gma. solution. alcohoL 100 grams solution.

0.00 0.7742 0.00 0.7742 0.00 0.7742

8.10 0.4434 3-996 0.5258 8.20 0.625

16.55 0.3050 8.08 0.4122 (saturated) 0.318

25.38 0.2188 12.25 0.3325

34.58 0.1495 16.51 0.2565

44.25 0.0877 25-30 0.1698

54.34 0.0478 3452 0.1088

70.42 0.0185 44- 16 0.0742

87.7 0.0061 54.3 0.0356

100. o 0.0027 649 0.0199

76.0 0.0131

87.7 0.0052

100. o 0.0009

Table IV.

Values showing the solubility of potassium chloroplatinate in mixtures, made up
of even percentage values by weight of water and alcohol. For the sake of compari-
son, Peligot's results for ethyl alcohol mixtures are given in the last column.

* Weight of KftPtCle in 100 gms. of solution.

Per cent, of
alcohol by weight.

MethTl alco-
hol mixtures.

0.00

0.774^

5 00

0.535

10.00

0.412

20.00

0.2642

30.00

0.1831

40.00

0. 1165

50.0

0.0625

60.0

0.0325

70.0

0.0182

80.0

0.0124

90.0

0.0038

100. 0

0.0027

Ethyl alco-
hoi mixtures.

Peligot*s results,
ethyl alcohol mixtures.

0.7742

0.75

0.491

■ • • •

0.372

0.5

0.218

0.35

0.134

0.28

0.076

0.14

0.0491

0.12

0.0265

0.08

0.0128

0.06

0.0085

0.05

0.0025

0.02

0.0009

756 E. H. ARCHIBALD, W. G. WILCOX AND B. G. BUCKLEY.

These results are shown graphically in Plate 11, the per cent, of alco-
hol being plotted as ordioates and the amounts of salt dissolved in grams,
as abscissae. From this curve the above results (Table IV) are taken tor
the round concentrations.

A glance at the curves in Plate II shows, at once, that the solubility of
the salt in the alcohol water solutions varies in a regular manner with
the amount of alcohol present, decreasing gradually as the per cent.
of alcohol increases. It should be noted, too, that it is appreciably more
soluble in the methyl alcohol solutions than in those of ethyl alcohol.
This is true also of the pure alcohols, for the absolute methyl alcohol

POTASSIUM CHLOROPLATINATB. 757

dissolves about six times as much salt as the absolute ethyl alcohol. This
shows that a more complete precipitation of the chloroplatiriate is ob-
tained by using ethyl alcohol than methyl.

Attention should also be drawn to the considerable difference between
the solubility of the chloroplatinate in 90 per cent, alcohol and in abso-
lute alcohol. The removal of the last 10 per cent, of water lowers the solu-
bility seven times. This shows the importance of using absolute alcohol
in the precipitation of this salt. When sodium chloroplatinate is also
present, however, it appears from Morozewicz's* work, that solutions con-
taining over 80 per cent, alcohol decompose the sodium salt, precipitating
sodium chloride.

Potassium and Sodium Chloride Solutions, — The solubility of the chloro-
platinate in solutions of sodium and potassium chloride was next exam-
ined. These solutions were prepared by weighing out a certain amount
of dry salt, making up to a definite volume, and subsequently diluting
this standard by means of carefully calibrated pipettes. The deter-
mination of the amount of salt (chloroplatinate) which had passed into
solution, was made by the colorimetric method. These determinations
were, however, in the case of the sodium chloride solutions, always
checked in the following manner. When the solutions were first made
up, the amount of platinum salt added, as well as the volume of the solu-
tion, was carefully determined. When the solutions were ready to analyze,
they were taken from the bath, the stopper removed, and the whole
quickly filtered through a previously weighed Gooch crucible, the resi-
due of the chloroplatinate being washed into the crucible with absolute
alcohol, this, serving at the same time to wash the mat of the crucible.
After carefully drying the crucible and contents, they were weighed.
This method gave results which were in close agreement with those ob-
tained by the colorimetric method.

Tabi^b V. — Results Showing thb SoLUBnriTY of Potassium Cmx)ROPLATiNATB in
Solutions op Potassium and Sodium Cmx>iaDB.

Concentration of
KCl •olntions.

Weight of KaPtCl«
in 100 gms. of solution.

Concentration of
NaCl solutions.

Weight of KtPtCU in
100 gms. of solution.

0.000

0.7742

0.000

0.672

0.200

0.0236

0.050

0.700

0.250

0.0207

O.IOO

0.729

0.500

0.0109

0.250

0.758

1. 000

0.0046

0.500

0.775

2.000

0.0045

0.750

0.791

3.000

0.0043

1. 000

0.805

4.000

0.0042

2.000

0.834

Saturated

0.0034

The results obtained with the sodium and potassium chloride solutions,
are shown in Table V. The concentrations are expressed in terms of
' Chem, Absi. Am, Chem, Soc, x, 972.

758 E. H. ARCHIBALD, W. G. WILCOX AND B. G. BUCKLEY.

gram molecules per liter. The measurements were carried out at jo"
in the case of the potassium chloride sohitions, and in the case o( the
sodium chloride ^lutions at 16".

These results are shown graphically in Plate III, the concentrations
of the solutions, as regards potassium and sodium chloride, being pbtted
as ordinates and the amounts of chloroplatinate dissolved as abscissae.

If we consider the potassium chloride solutions, we see that the amount
of salt dissolved diminishes rapidly as the concentration of the potas
slum chloride increases, until we reach a concentration of about one gram
molecule per liter. From this point On, the increase in the concentra
tion of the potassium chloride has very little effect upon the amount

POTASSIUM CHLOROPLATINATE. 759

of chloroplatinate dissolved. This is surely what we would expect.
Potassium chloride being, in its aqueous solution, a strongly ionized salt,
we have here a high concentration of the potassium ion ; in this case the
one common to both salts. This will prevent the dissociation of the
chloroplatinate, and therefore very little will go into solution. Apparently,
by the time we have reached a concentration of about one gram mole-
cule per liter, with regard to potassium chloride, equilibrium is practically
established between the potassium chloride undissociated and the potas-
sium and chlorine ions, as the addition of more chloride has little effect
on the solubility of the chloroplatinate.

In the case of the sodium chloride solutions, we see that the amount
of chloroplatinate dissolved increases as the amount of sodium chloride
present increases. This increase is rapid at first, but after a concentra-
tion of 0.05 gram molecules of sodium chloride is reached, we see from
Plate III that this increase is much smaller and very nearly proportional
to the increase in concentration of the sodium chloride. The equilibrium
is here between the ionized potassium chloride formed and the dissocia-
ted potassium chlorplatinate, thus:

+ —

2NaCl :^ 2Na -I- 2CI

•KjPtcu :^^ + Ptci..

The more dissociated sodium chloride there is present, the more sodium
ions there are to unite with the PtCU ions to form undissociated sodium
chloroplatinate, and therefore the greater will be the amount of potas-
sium chloroplatinate which will dissolve. This being the case, that por-
tion of the sodium chloride which is dissociated will be largely instru-
mental in causing the potassium chloroplatinate to dissolve. And this
is what we find to be the case. The first portion of sodium chloride
added to pure water has a greater effect than later portions, as it almost
entirely dissociates, while the same amount added to a concentrated
solution has almost no effect.

If we compare the solubility of the chloroplatinate in potassium chlor-
ide solutions, with its solubility in the water-alcohol solutions, we see that
about the same amount of the platinum salt is dissolved by a solution
of potassium chloride containing two gram molecules per liter, as by a
water-alcohol 'solution containing ninety per cent, of alcohol. If then,
platinum was being precipitated from a solution, and no^ potassium, a
more complete precipitation of the platinum would be 6hlained by add-
ing a saturated solution of potassium chloride than l^y adding absolute
alcohol.

The work described in the foregoing pages shows that *•

76o G. S. JAMIESON, I«. H. LEVY AND H. L. WBI<I^

(i) Small amounts of potassium chloroplatinate in solution can be
estimated colorimetrically with considerable accuracy. This will be
true of any salt which gives the PtClo anion, provided the color of the
cathion is not such as to interfere.

(2) The chloroplatinate is less soluble in solutions of ethyl alcohol
and water than in water solutions of either methyl or isobutyl alcohol.
Only 0.0007 gram of the salt dissolves in 100 cc. of ethyl alcohol at 20°.

(3) The solubiUty of the chloroplatinate in potassium chloride solu-
tions decreases with the increase in concentration of the potassium chlor-
ide, imtil a concentration of one gram molecule per liter is reached. Be-
yond this point, increasing the concentration of the potassium chbride
has practically no effect.

(4) The solubility of the chloroplatinate in solutions of sodium chloride,
increases rapidly until a concentration of 0.05 gram molecules per liter
is reached. For more concentrated solutions the increase in solubility
is small and almost proportional to the increase in concentration of the
sodium chloride.

Chemical Laboratory of Syracusb Univbrsity,

Syracuse, N. Y.

OS A VOLUMETRIC METHOD FOR COPPER.

By G. S. JaMIBSON, I,. H. I«BVY AND H. I«. WBLL8.

Received Pebnxaiy 96, 1908.

The process to be described is based upon the titration of cuprous
thiocyanate with potassium iodate solution in the presence of a large
excess of hydrochloric acid. This method of titrating a number of re-
ducing substances, such as free iodine, iodides, arsenites and antimonites,
in a very satisfactory manner, is due to L. W. Andrews.^ The reaction
depends upon the formation of iodine monochloride and the disappear-
ance of the iodine color imparted to an immiscible solvent, such as chloro-
form or carbon tetrachloride.

We find that cuprous thiocyanate is oxidized by iodine chloride sharply
and quantitatively with the formation of cupric salts, sulphuric and hydro-
cyanic acids, according to the equation

4CuSCN + 7KIO8 + 14HCI =

4CUSO, + 7KCI + 7ICI + 4HCN + 5H,0.

This oxidation is similar to that obtained in Parr's method* for the titra-
tion of CuSCN by means of potassium permanganate solution, but we
consider the iodate titration far preferable to the latter in simplicity and
accuracy.
To show the applicability of the Andrews method to the titration of

^ Tms Journal, 25, 756 (1903).
' Ibid,f 22, 685 (1900).

thiocyanates, the following experiments (by L. H. L.) were carried out,
using the pure dry compounds and a solution containing 10.706 grams
of potassium iodate per liter (1/20 of formula-weight), and titrating
in glass-stoppered bottles in the presence of about 5 cc. of chloroform
and about one-half of the final volume of concentrated hydrochloric
acid.

Substance taken. KlOfUsed. Substance found. Error.

Gram. cc. Gram. Gram.

NH4SCN o.iocx> 39.4 o.iooo 0.0000

NH4SCN O.IOOO 39.5 0.1003 +0.0003

AgSCN 0.2000 36.05 0.1995 — 0.0005

AgSCN O.IOOO 18. 1 0.1002 -f 0.0002

CuSCN 0.2000 57.5 0.1999 — o.oooi

CuSCN O.IOOO 28.8 o.iooi +0.0001

In some of the above experiments the substance was dissolved before
titrating, in a hydrochloric acid solution of iodine monochloride which
had been carefully adjusted over chloroform to the colorless point of the
latter, but the use of this reagent was found to be unnecessary.

We find that filter paper has no effect upon this titration. This fact
permits the use of paper for filtering cuprous thiocyanate in this process.
It appears that organic substances are generally inactive towards iodine
monochloride, for Andrews titrated tartar emetic directly with accurate
results, and we have found by direct experiments that ethyl alcohol,
acetic acid, formic acid, and formaldehyde do not interfere with the
iodate titration.

The following results were obtained (by G. S. J.) by precipitating
cuprous thiocyanate with sulphurous add and ammonium thiocyanate
from a measured solution of copper sulphate of known strength, fil-
tering, sometimes on asbestos, sometimes on paper, and titrating in the
manner previously mentioned. Another 1/20 formula-weight potas-
sium iodate solution was used here: i cc. « 0.001817 g. Cu.

Copper taken. KIOs used. Copper found. Error.

Gram. cc. Gram. Gram.

0.0486 26.7 0.0485 — O.OOOI

0.0486 26.8 0.0486 0.0000

0.0388 21.3 0.0387 — 0.0001

0.0351 1945 0.0353 +0.0002

0.0486 26.7 0.0485 — 0.0001

0.0486 26.9 0.0488 +0.0002

In applying the method to ores, we find that lead and antimony, if
not removed, will produce high results, but both these metals are re-
moved (to such an extent at least that they do not interfere in the slight-
est degree) by the evaporation with sulphuric add in the process which
will be described. Silver also must be removed.

For the sake of convenience, we have employed the normal potassium

762 G. S. JAMIESON, L. H. LEVY AND H. L. WELLS.

iodate in the place of the acid iodate used by Andrews. The normal
iodate can be purchased in a pure condition (it should be perfectly neu-
tral to test-paper), or it may be easily prepared by the method of Groger^
from potassium iodide and potassium permanganate. The salt should
be dried at 100° before weighing. We have used, besides the previously
mentioned 1/20 formula -weight solutions, one about twice as strong,
but for convenience in calculation in determining copper we recommend
the use of some multiple of 5.892 grams per liter which gives exact milli-
grams of copper per cubic centimeter, according to the multiple taken.
For instance, with 11.784 grams KlOg per liter, i cc. = 0.002000 gram
copper.

Method of Analysis. — ^To 0.5 gra'm of the ore in a 6 oz. flask, add 6 to
10 cc. of strong nitric acid, and boil gently, best over a free flame, keep-
ing the flask in constant motion and inclined at an angle of about 45*^, un-
til the larger part of the acid has been removed. If this does not com-
pletely decompose the ore, add 5 cc. of strong hydrochloric acid and con-
tinue the boiling until the volume of liquid is about 2 cc. Now add grad-
ually and carefully, best after cooling somewhat, 6 cc. of strong sulphuric
acid, and continue the boiling until sulphuric acid fumes are evolved
copiously. Allow to cool, add 25 cc. of cold water, heat to boiling, and
keep hot until the soluble sulphates have dissolved.^ Filter into a beaker,
and wash the flask and filter thoroughly with cold water.^ Nearly neu-
tralize the filtrate with ammonia and add 10 to 15 cc. of strong sulphur
dioxide water. Heat just to boiling and add 5 to 10 cc. of a 10 per cent,
solution of ammonium thiocyanate, according to the amount of copper
present. Stir thoroughly, allow the precipitate to settle for 5 or 10 min-
utes, filter on paper, and wash with hot water until the ammonium thio-
cyanate is completely removed.

Place the filter with its contents in a glass-stoppered bottle of about
250 cc. capacity, and by means of a piece of moist filter paper transfer
into the bottle also any precipitate adhering to the stirring rod and beaker.
Add to the bottle about 5 cc. of chloroform, 20 cc. of water and 30 cc.
of concentrated hydrochloric acid (the two latter liquids may be pre-
viously mixed). Now run in standard potassium iodate solution, in-
serting the stopper and shaking vigorously between additions. A vio-
let color appears in the chloroform, at first increasing and then dimin-

* Z. angew. Chem., 1894, 13.

' The decomposition and conversion into sulphates here described closely follows
the directions of Low, "Technical Methods of Ore Analysis," p. 79, in connection
with the iodide method.

*With substances containing appreciable amounts of silver a few drops of hy-
drochloric acid should be added before making this filtration, but not enough to dis-
solve any considerable amounts of the lead sulphate or antimonic oxide that may be
present.

VOLUMETRIC METHOD FOR COPPER.

763

ishing, until it disappears with great sharpness. The rapidity with which
the iodate solution may be added can be judged from the color changes
of the chloroform.

In order to make another titration it is not necessary to wash the
bottle or throw away the chloroform. Pour ofiF two-thirds or three-
fourths of the liquid in order to remove most of the pulped paper, too
much of which interferes with the settling of the chloroform globules after
agitation, add enough properly diluted add to make about 50 cc. and
proceed as before. In this case, where iodine monochloride is present at
the outset, the chloroform becomes strongly colored with iodine as soon
as the cuprous thiocyanate is added, but this makes no difference with the
results of the titration.

In the following experiments (by G. S. J.) weighed quantities of pure
copper were put through the above course of analysis in the presence
of antimony, and in some cases lead also:

(i cc. KIO3 = 0.003610 gmm Cu.)

Copper taken.

Antimony.

Iodate used.

Copper found.

Error.

Gimm.

Lead.

cc.

Gram.

0. 1136

0.06

31 -35

O.II31

— 0.0005

0.0691

0.06

1905

0.0688

— 0.0003

0.0733

0.06

present

20.30

0.0733

0.0000

0.0673

0.06

present

18.75

0.0677

-Ho. 0004

0.0650

0.06

18.08

0.0651

4-0.0001

0.0486

0.03

13.50

0.0487

+0.0001

0.0486

0.03

13.48

0 . 0486

0.0000

Several ores, sulphides, some of which contained lead or antimony,
were analyzed by the process (by G. S. J.) in order to compare the re-
sults with other methods: (KIO3 10.706 g. in 1000 cc; icc =0.001817
g. Cu.)

Ore taken.
Gram.

, 0.5000

. 0.5000

. 0.3584

. 0.2000

Ilia

III6

IIIc 0.5000

IVo 0.2000

1V6 0.2000

V 0.2000

Iodate used,
cc.

32.2

39.8

41 .2

22.9

28.9

21.08

21. 10

20.8

Copper found.
Per cent.

11.70
14.46
20.88
20.80
20.86

19.15
19.16

18.89

Copper by other method.
1 1. 7 1 Electrolytic

14.50 "

20 . 70 Iodide

20.70

19.02 Electrolytic

19.02 "

18.80 Iodide

So far as ease and rapidity are concerned, one of us (G. S. J.) has made
in just one hour an analysis of a copper ore, including weighing and
calculation, by the method given above. Following Low^s* modification
of the much-used iodide method, the time was one hour and twenty min-
utes. There is no doubt that the iodate method is the easier and quicker
of the two.

* Loc. cit.

764 S. W. PARR.

In view of the large excess of potassium iodide employed in the iodide
method, it is probable that the iodate method is cheaper.

The potassium iodate solution is perfectly stable and can be preserved
without change for years, if protected from evaporation. Ordinarily
it is unnecessary to standardize the solution, except by weighing out a
known amount of the salt and dissolving it in a known volume. However,
should there be any uncertainty in regard to the purity of the salt, or in
connection with the relations of the volumetric apparatus, it would be
advisable to standardize with pure copper, putting it through all the opera-
tions of the process, and thus eliminating also any slight constant enors.

With the precaution just mentioned, the process is capable of reaching a
very high degree of refinement, for the method of titration is one of the
sharpest and most unifonn in its results. Since most of the iodine goes
into the small volume of chloroform, the accuracy of the end-reaction
is extraordinary.

SHBPFIBLD IfABORATOKT,

Nbw Havbn, Conn.

SODIUM PEROXIDE m CERTAIN QUAinTTATIVE PROCESSES.

Bt S. W. Parr.
Received February 21, 1908.

Sodium peroxide as a reagent in qualitative analysis, described by
the writer,* has been found, after a number of years of actual service, to
have advantages which entitle it to a far wider recognition and a more
detailed study for that particular purpose. The same article states
**that other properties have developed, mainly of interest in quantita-
tive methods, which it is hoped will be of sufficient value to warrant
further notice." It is in connection with this latter phase of the sub-
ject that the following processes are offered.

The adaptations of sodium peroxide here referred to are largely the
outgrowth of the writer's experience in the use of that substance as a
combustion medium for calorimetric determinations. This material
has been recognized for some time as a good fusion reagent where both
solution and oxidation are to be effected, but the usual methods of fusion
in an open vessel are characterized by too great violence and danger of
loss as well as by serious decomposition of the containing vessel. By
carrying on the fusion in a closed vessel, it is possible to so adjust reagents
as to bring about a quiet fusion without spurting. Owing to the con-
centration of the very great heat within the mass, the walls of the con-
taining vessel must be kept relatively cool by submerging the same in
water, thus preventing the corrosion of the container while at the same
time there is no interference with the chemical reaction. An idea of

» Tms jouKNAL, 19, 341.

SODIUM PEROXIDE. 765

the intensity of the heat in the interior of the mass may be gained from
the experience occasionally met with of having soft iron wire melted
into a rotuid shot. Experiments have been carried on to determine the
best conditions for securing a quiet fusion upon substances of widely
varying character. In this work a closed cartridge only has been em-
ployed. Experiments have been made with bombs of various sizes and
of various compositions, A bomb was made of 30 per cent, nickel steel
but no special advantage was observed in its use. A bomb of about five
times the usual capacity, was also made of the same material but the in-
crease in the charge which this permitted resulted in the development
of too great a quantity of heat to warrant its general adoption.
In case the walls are melted through, the contact of the molten
mass with the surrounding water produces a disturbance
with more or less of the characteristics of an explosion.
Pringsbeim* has adapted the method to the determinarion
of the halogens in organic compounds and described a special
form of closed crucible for carrying out the process. It is
the purpose of this article to make record of a still wider
use of the method and to call attention to the adaptability
of the bomb which accompanies the Parr calorimeter. It
may not have any great advantage over the closed crucible
of Pringsheim, but the fact that the bomb is not infre-
quently already at hand may be a reason for calling attention
to it.

Fig. I ilhistrates the bomb in its present form. The igni-
tion is brought about by dropping into the stem a short slug
of soft iron or pure nickel wire heated to redness. It will
lodge at the lalve M, when, by pressing down quickly with
the forceps at O, the hot wire is released and drops into
the mixture. From the construction of the cartridge it may
be seen that, when submerged, the holes in the screw caps
provide tor the circulation of water about the two ends of
the cylinder, thereby permitting the use of rubber gaskets.
An air space about the lower part of the cylinder as at E
provides for a less rapid cooling of the walls of the
chamber than would be the ease if the cylinder at that part
were in direct contact with the water. The ends of the
cylinder are removable and the fused mass is driven ut
with a short rod provided for the purpose. A jet of hot water ~
readily i^shes out the interior surfaces and the entire fusion
is thus brought into solution with a minimum amount of water, 50 or
75 cc. being ample.

' Am. Chewt. J., 31, 3S6.

766 S. W. PARR.

In applying the process to inorganic substances it is necessary to have
a mixture which will carry on the combustion independently of the ma-
terial tmder examination. It is also true that many organic substances
require, for complete combustion, conditions not provided by sodium
peroxide alone.* Thus, with carbon and hydrogen, the conditions for
the reactions are indicated separately by the following equations:

2Na202 + C = NajCOs + NajO;
NaA -h Na,0 + O -h 4H = 4NaOH.

A series of tests for determining the most suitable reagents has resulted
in the following mixture, applicable also to inorganic as well as organic
substances. For a detailed statement of the action of these reagents,
reference should be made to a recent article on ** Calorimeter Constants.'"

Standard Fusion Mixturb No. i.

ID grams (i measure) of soditim peroxide,
0.5 '* potassium chlorate,
0.5 " benzoic add.

A thorough mixing is effected by enclosing the charge in the bomb
and shaking; the bomb is then submerged to prevent the melting through
of the metal and the red hot slug is dropped through the stem as above
indicated. The process is complete in two or three minutes, when the
cartridge may be opened and the fusion transferred to a beaker, where
the dissolving and boiling out of the oxygen is accomplished in about five
minutes more. These are the fundamental conditions for a large class
of substances, some of which are detailed as follows:

Sulphur and Arsemc in Pyritic Ores of Iron and Copper, — ^If to the
charge as just described we add 0.25 gram of a pyritic ore, we shall have
a complete decomposition of the pyrites along with the combustion,
so that within a very few minutes after the weighing and mixing in the
cartridge (fifteen minutes at the most, and usually within ten minutes
if desired), we may have our sulphur as sulphate in solution ready for pre-
cipitation with barium chloride. Nor is accuracy sacrificed to speed,
for the total absence of contamination from the sulphur of gas flames is
avoided. Moreover, as Hillebrand has shown,* with the bulk of the sohi-
tion measuring 200 cc. to 300 cc. dehydration for the removal of silica
is unnecessary. This method, therefore, is especially well adapted to
the determination of sulphur in mineral substances, and is greatly to be
preferred to the usual Fresenius method. For pyrites, moreover, it has
decided advantage over the Lunge procedure. This is especially true in
the case of roasts containing small amounts of sulphur and with pyrites
containing copper. In all these cases a perfect fusion is secured. Upon

' This Journal, 29, 16 16.

* Ibid., 29, 1618.

» BuUetin 305, U. S. Gepl, Surv., p. 160.

SODIUM PEROXIDE. 767

boiling, the ferrate is decomposed, precipitating the iron, but a clear solu-
tion results upon acidifying. If it is desired to precipitate the copper,
it is necessary to make slightly acid and then faintly alkaline with sodium
carbonate, since the copper hydroxide is soluble in an excess of alkali.
Arsenic, if present, is in the fusion as sodium arsenate and may be deter-
mined by any of the accepted methods. In the following tests a pyritic ore
with approximately 10 per cent, of copper was used. The arsenic was deter-
mined for comparison by the distillation method.^ A parallel fusion
was also made, using a known quantity of pure arsenic trioxide and cop-
per oxide.

TaBLB I. — DETERMINATION OP SULPHUR IN ORBS BY FuSION WITH SODIUM PBROXIDB.
Material. Pusion with Na^Os. I«ttBge*s method.

Galena i4-59 1383

Zinc blende 29 . 28 28.73

Arsenical pyrites 29 . 32 28 . 52

Table II. — Determination of Arsenic by Fusion with Sodium Pbroxidb

Arsenic found

Arsenic found gravimetrically Arsenic
▼olumetrically cal- as AsjOs and cal> present

Material. culated to AS9O1. culated to ASfOt. as AssO*.

As,0, 4- CuO 100. 3 .... 100. o

AsjO, + CuO .... 99-97 100. o

Arsenical ore with iron and r (a) 11.42 .... ii-37

copper pyrites i (6) 1 1 . 68 ' .... 1 1 • 43

Sulphur in Coal, Coke, Ashes, Etc, — ^In the case of coal and coke, the
method has very decided advantages over the usual Eschka process.
With coal and coke, one-half gram of these substances should be taken
and the benzoic acid omitted from the fusion mixture. In ashes, how-
ever, the fusion is affected by the use of the benzoic acid, as above de-
scribed. If desired, the fusion will easily allow the use of from one-half
to one gram of ash material. For sulphur in coal and coke, Sundstrom^
has described a special crucible based on the reaction with sodium per-^
oxide, as suggested by the use of that material for calorimetric purposes.
Von Konek' makes direct use of the bomb as provided for calorimetric
determinations. Both of these writers, with still others, give compara-
tive data showing the satisfactory nature of the results. By use of the
fusion mixture, as above given, a much wider range of substances is cov-
ered, including those with little or no ability to carry on the combus-
tion by themselves.

By reference to Table I, it will be seen that the results for sulphur
in mineral matter are slightly lower by the Lunge method. This is a
natural result of the methods employed, since the fusion process gives

^ Z. anal. Chem., 21, 266.
» This Journai,, 25, 184.
* Z. angew. Chem., 1903, 516.

'«
*

768 S. W. PARR. S

the total sulphur, including any sulphates present. In case of high iron,
it is, of course, removed by filtration before precipitating the sulphur*

Sulphur in Rubber. — Rubber may be completely oxidized if brought
to a reasonably fine state of division. This may readily be accomplished
by grating the rubber on a new file, though the purjer grades may be
better cut into small pieces with a sharp knife. One- or two-tenths of a
gram with the standard charge will give a perfect combustion. The
precipitation of sulphur as a sulphate is carried on in the usual manner.

Table III. — Sulphur in Rubber. ^

Sulphur from
Amount peroxide

taken. \ method. Bschki**

Material. Gram. Method. 'Percent. method.

"Ebonite" rubber packing o.i 10. o g. peroxide (a) 4. 59 4.32

o. 5 g. chlorate \_ . ^

0.3 g. benzoic ^

add (&) 4-30. 4-41

White soft rubber tubing o. i same < ,..' - : -

* 1(6) 0.92"^ '

Halogens f Sulphur, Etc., in OrgarUc Compounds. — As a substitute for
the Carius method for decomposing organic compoimds of the halogens,
the method is in every way to be preferred. It is applicable also to organic
compounds of sulphur, phosphorus, arsenic, etc. Pringsheim* also refers
to this use of sodium peroxide as a qualitative reagent and gives details
for carrying out the combustion process. Our own experience proves that,
with organic compounds a simple modification of the standard charge
is sufficient in that the sum of the material under examinatioiji phis the
benzoic acid should equal the usual amount of combustible, viz., 6.5
gram. That is, if 0.2 gram of an organic substance such as aniU^e hydro-
chloride is taken, the amount of benzoic acid should be 0.3 gram. However,
these quantities are flexible within certain limits, though the total amount
of organic combustible should not greatly exceed 0.5 or 0.6 gram.

Since, with the halogens, it is necessary to avoid the use of chlorine
compounds, the fusion reagents, potassium chlorate and benzoic add,
above given, may be replaced by a mixture made up as follows;

Boko Magnesium Mixturs.

Parts.

Boric add in fine powder 5

Potassium nitrate, powdered 4

Magnesium, powdered i

The amount of the above boronitrate-magnesium mixture may be in-
creased from 0.5 to 2 grams or until a satisfactory combustion is secoied
by means of the increased quantity of metallic magnesium present. The
charge, therefore, for this class of compounds would be:

* Am. Chem. J., 31, 386; Ber., 37, 2155.

SODIUM PEROXIDE. 769

Fusion Mdcturb No. 2.

10 grams (i measure), sodium peroxide,
I to 2 grams boro-magnesium mixture,
o. 3 to o. 5 gram carbon compound.

In the case of organic substances that are liquid at ordinary tempera-
tures, the conditions are not altered except with very volatile substances
where an exact weight is diflScult to obtain. For such material a very
Kght bulb of glass is blown from thin-walled tubing of about two or three
mm. caliber. This is fused ojff with a capillary stem and weighed. With
care, it is not difficult to make bulbs of one-half inch or more in diameter,
weighing less than 0.2 gram. When so made, they may be easily broken
after the charge is made up and the cartridge closed. About 0.2 or 0.3
gram of liqtiid is drawn into the bulb and the capillary placed in the
cartridge directly upon the bottom and the standard charge with No.
I or No. 2 placed above it. After closing the cartridge, a sharp rap on
a solid desk will break the glass, when, by shaking, a thorough mixture
is secured. This procedure has given good results with benzene' and
similarly volatile materials.

Tabids IV. — Cmx)RiNS by Fusion with Sodium Pbroxidb.

Material. Fation mixture. Pound at AgCl. Theoretical.

Sodium chloride Ftision mixture ^o. 2 60. 48 60. 60

Aniline chloride Fusion mixture No. 2 27 . 63 27 . 36

Carborundum. — Finely divided carborundum bums readily by use of
fusion charge No. 2. The carbon in this fusion may be determined
volumetrically. In a separate sample, however, the metallic iron should
be determined by extracting the same with a magnet and the free silica
may be determined by volatilizing with hydrochloric acid, since the sili-
con of the carbide is not attacked by that reagent. The total silicon
from the fusion should then be corrected to ascertain the amount in com-
bination as SiC.

Table V. — ^Analvsis of Carborundum.

Constituents. Method. Per cent.

SiO, VolatiUzed with HF 8.27

Iron as Fe Removed by magnet 4. 37

Silicon as Si Fusion mixture No. 2 63. 58

Carbon as C Volumetric from fusion mixture No. 2 33 . 67

While various other substances have been tested with satisfactory
results such as shales, fire clays, titanium, ores, etc., it is thought that
the above descriptions cover a sufficiently wide range of material to in-
dicate the uses to which the method may be put.

I am indebted for analytical results to six or eight senior and graduate
students of this department who have from time to time worked with

770 D. F. CALHANE.

the methods as indicated. Special acknowledgment should be made
to Mr. F. W. Gill for results on arsenic, chlorine and carborundum.

UNIVBRSITT of iLLIZfOIS,

Urban A, III.

THE COMPARATIVE OXIDIZING POWER OF SODIUM PEROXIDE

AND ITS USE IN QUALITATIVE ANALYSIS.

BT D. P. Calhakb.
Received Pebmary xa, 1908.

The detection of chromium in qualitative analysis rests on its oxida-
tion to the chromate and the precipitation of chromate of lead in acetic
acid solution. The oxidation is usually effected by chlorate of potash
on the solution of chromium hydroxide in strong nitric acid. If by any
chance the nitric acid chosen is not strong enough or has become diluted,
the oxidation will not occur and the test fails. It is necessary for suc-
cess that the hydroxides possibly containing iron, aluminium and chro-
mium be freed of water as much as it is possible by drying them quickly
by heating. In this part of the procedure the student usually fails to
work properly, not enough water is removed, and the subsequent oxi-
dation does not occur. As I have found, in teaching classes in this sub-
ject, so much trouble in getting the student to properly observe precau-
tions, it seemed of advantage to apply' a test which would be effective
and not so much dependent on a certain set of conditions.

Some exp)eriments were accordingly made on the action of certain other
oxidizing agents. Among these were chosen sodium peroxide and bro-
mine water. The preference was for the former, as it is to be obtained
in a convenient solid form, and possesses few of the disagreeable features
of bromine water. A o.i N solution of chrome alum was prepared, eadi
cubic centimeter holding 0.0332 g. of chrome alum, answering to
0.005 g- o^ CrjOg. Next, the oxidation of portions of this solution
with peroxide of sodium and bromine water was carried out to see
what the comparative efficacy of these two reagents is on solutions con-
taining known amounts of chromium. In this connection an interesting
feature in the action of sodium peroxide on chromium solutions was met
with, that at first led to the belief that bromine water was more eflSdent.
The dilutions were carried to the point where the amount of chromic
oxide present was only 0.000125 g., corresponding to 0.000085 ?• ^'
chromium. At this extreme dilution, a safe test was secured with
bromine water, but the sodium peroxide apparently failed at a con-
centration answering to 0.00025 g. chromic oxide. The bromine water
appeared to act equally well both hot and cold at all the different con-
centrations from 0.005 g. chromic oxide down to 0.000125 g. With
the sodium peroxide there was no test with the presence of 0.005 g.

OXIDIZING POWER OF SODIUM PBROXIDE. 77 1

chromic oxide, if the solutions were treated cold. When the sodium
peroxide was added to the hot solution a small amount of yellow pre-
cipitate was obtained. An explanation of this surprising fact was
arrived at later.

The next procedure was to find out what the limit of delicacy for the
test is in actual analysis. Normal solutions of aluminium nitrate and
ferric chloride were prepared, one cubic centimeter of which answered
to 0.071 g. aluminium nitrate and 0.054 g- ferric chloride, giving a pre-
cipitate of 0.026 g. aluminium hydroxide, 0.0356 g. ferric hydroxide,
and 0.0033 K' chromic hydroxide from i cc. of the chrome alum. The
analysis in the first instance was carried through with a mixture of i cc.
of each of the three solutions. The aluminium and iron here exceeded
the chromium in the mtios of 8 and 12 times, respectively. The solu-
tion of the mixed salts was treated with 10 cc. of barium carbonate emul-
sion and allowed to stand for 2 or 3 minutes. Next a filtration was made,
and the filtrate tested for iron, chromium and aluminium. The tests
proved complete precipitation of the metals. The barium carbonate
residue was dissolved in hot hydrochloric acid. The hydroxides of the
three metals reprecipitated with ammonia. These were dissolved in
hydrochloric acid and the diluted solution divided into two equal parts.
There could be present in each part the equivalent of 0.0025 g. chromic
oxide, answering to 0.0016 g. chromium hydroxide. One part was added
to a large excess of bromine water, after previously having added excess
of sodium hydroxide. After filtration from the iron, no test for chro-
mium appeared on adding acetic acid and lead acetate. In the other por-
tion similarly treated, using sodium peroxide as the oxidizing agent,
a good test was obtained. Next, the above procedure was repeated,
using 2 cc. of the iron and aluminium solutions and i cc. of the chro-
mium. The mtios here were 24 iron, 16 aluminium to i of chromium
hydroxide. Here again the bromine water failed to act sufficiently for
a test, while the sodium peroxide gave a good test. In the third experi-
ment 5 cc. each of the iron and aluminium solutions were taken and o . 5
cc. of the chrome alum solution. After proceeding as before, the bro-
mine water failed to give a test, while 0.5 g. sodium peroxide gave a re-
liable indication. The ratios here were 81 aluminium and 11 1 iron to
I of chromium. The mixed precipitate contained 0.130 g. aluminium
hydroxide, 0.178 g. ferric hydroxide, and 0.0016 g. chromium hydroxide.
In each of the last two tests the maximum amount of chromic oxide
present for oxidation was 0.0008 g., corresponding to 0.0005 S- chromium.

In the next experiment it was desired to contrast the delicacy of the
usual method with potassium chlorate and nitric acid with the peroxide
procedure. Amounts taken were 5 cc. aluminium nitrate, 5 cc. ferric
chloride, and 0.5 cc. chrome alum. The analysis was carried through

772 D. F. CALHANE.

in the usual way and the precipitate of iron, chromium and aluminium
hydroxides given by ammonia was dissolved in strong nitric acid and
boiled with potassium chlorate. The solution was divided into two equal
portions and one tested with sodium hydroxide acetic acid and acetate of
lead. The result was negative for chromium. Here, as in the previous
cases, 0.0008 g. of chromium hydroxide was present. The result shows
that the limit of successful oxidation by this method has been exceeded.
The time required to carry this out is longer and the test is much less
delicate than with the peroxide method.

Another test was made to compare the fusion method of oxidation
with that using bromine water, potassium chlorate and nitric acid and
sodium peroxide. The usual method of fusing the hydroxides of iron,
chromium and aluminium mixed with dry potassium nitrate and sodium
carbonate on platinum was followed, having present the same amount
of chromium, iron and aluminium as in the previous cases. The result
here for chromium by the usual test as lead chromate after oxidation,
was surprisingly convincing. Five cubic centimeters of the ferric chlo-
ride aluminium nitrate and 0.5 cc. of the o.i N chrome alum solution were
precipitated by ammonia. The precipitate was filtered, washed and
fused in a platinum crucible with potassium nitrate and sodium carbon-
ate mixture for a few minutes. The cooled mass was lixiviated with hot
water. The iron was filtered ojff and one-half of the clear filtrate, which
was decidedly yellow, was made acid with acetic acid and lead acetate
added. A good yellow color was produced due to the formation of lead
chromate. This test appears to be equally delicate with the sodium per-
oxide method. There was present in this fusion test 0.0008 g. chromk
oxide, or 0.0005 g. chromium.

In the experiments having to do with the oxidation of the pure chrome
alum solution with bromine water and peroxide, it was surprising that the
peroxide apparently failed to oxidize the chromium at a concentration
where bromine water gave excellent results. It was found, however,
that if the solution was hot during the test a very slight precipitate of
lead chromate was obtained with the sodium peroxide procedure. Bro-
mine water acted equally well in hot or cold solution. It has been pre-
viously shown that in actual analysis sodium peroxide is by far the more
powerful oxidizing agent.

It was found on further experimentation that whereas a solution of
bichromate was readily oxidized to perchromate by barium peroxide in
acid solution, a chromic salt gave no indication of oxidation to the per-
chromate by the above agent. The chromic salt was only slightly oxid-
ized to the chromate.

Sodium peroxide acting on a solution of the chromic salt gave a yel-
low color. On acidification a violet color appeared, which concentrated

DETERMINATION OF CARBON IN STEElr, ETC. 773

in ether indicated perchromic acid. In add solution sodi]um peroxide
gave no such oxidation. If a chromic salt were oxidized by sodium per-
oxide as above, the solution acidified with acetic acid and lead acetate
added, in dilute solutions such as were used in this work, no precipitate
appeared. On allowing the solutio;a to stand, or more quickly, on warm-
ing, the customary yellow precipitate of lead chromate appeared and
oxygeii continued to be evolved for some time.

These results show the reason of the failure to obtain the tests for chro-
mate of lead in the earlier part of the work where dilute solutions of
chrome alum were severally treated with sodium peroxide and bromine
water. The sodium peroxide forms an alkaline solution with the evolu-
tion of oxygen. The chromic salt is oxidized to the perchromate, the
sodium salt being formed in the alkaline solution. This substance is
stable and gives a yellow color to the solution. On acidification with
acetic acid, lead acetate produces no precipitate, as the lead perchromate
is soluble in this medium. On standing or warming, oxygen is given off
and the lead perchromate breaks down to the chromate, giving the custom-
ary yellow precipitate.

Oxone, the fused form of sodium peroxide, acts the same as the unfused
variety, as would be expected. For analytical work it is less desirable,
owing to the impurities it contains. In addition to the silica found in
the unfused variety, it carries about 1.5 per cent, of copper as the result
of several analyses showed. In addition, small amounts of iron were
found to be present. The results obtained in this investigation show
that sodium peroxide is the best oxidizing agent for chromium in solu-
tion. Sodium perchromate is formed by this agent in alkaline solution.
This solution is stable, and, on acidification, oxygen is rapidly and con-
tinuously evolved. The oxygen can be liberated as fast as desired, thus
giving a powerful oxidizing source simply controlled.

The properties of this alkaline perchromate solution will be further
investigated.

WORCBSTBR POLTTBCHNIC IZfBTITUTB,

WoRCBSTBR, Mass.

THE DETERMINATION OF CARBON IN STEEL, FERRO-ALLOYS,
AND PLUMBAGO BY MEANS OF AN ELECTRIC

COMBUSTION FURNACE.

Bt C M. Johnson.
Received March 3, 1908.

Several months ago it occurred to the writer that the Hoskins resis-
tance wire could be applied to the heating of combustion tubes. A
drawing was prepared for a furnace of a muffle type to heat four tubes
lying in the same plane and parallel.

774 C. M. JOHNSON.

After some correspondence it was agreed, at first, to try a single tube
furnace. It consists of a steel tube 295 mm. x 76.3 mm. containing a
non-conducting paddng of magnesia oxide. In the center is a qmrti
tube wound with the Hoskins wire.' Inside of this tube is placed aD-
other of the same material of 19 mm. inside diameter and 600 mm. kmg,
in which the combustions are made.

A. Mercury pressure gauge for detection of leaks and stoppages,

B. Jar for stick potassiuin hydroxide or for any solid dryer or absorb-

C. Safety jar for potassium hydroxide solution, preventing solution

from backing over into rubber tubing.

D. Caldum chloride jar.

E. Soda lime jar.

F. Mercury valve, to prevent reverse action.
Ci. Ekctric combustion furnace.

H. Jar for granular anc to remove
Acid fumes,
Litharge fumes.
Sulphur fumes,
Chlorine fumes.
1. Jar for phosphoric anhydride to remove water.
J. Absorbent and weighing apparatus for carbon dioxide.
This furnace was to be durable, if not heated above 1 100°, but it buined
out in three days. It was then rewired with greater resistance and was
guaranteed, if not heated above 1000°. Fearing the furnace might again
desert the cause, the writer put in a small 32-ohm rheostat that happened
to be at hand. With about one-fourth of this resistance the furnace,
on a 220-volt direct current, has been maintaining a constant tempera-
ture. To secure complete combustion of steel, it is very essential thai
the heat be maintained as close to 950" as possible, t, e., as little under
that temperature as practicable. If the temperature drops to about
900°, or under, the results obtained are liable to be from 0.01 to o.io per
cent, too low, unless red lead is mixed with the drillings. Hence, if one
• The apparatus can be supplied by the Scientific Materials Co., of Pittsbuis-

DETERMINATION OF CARBON IN STEEL, ETC. 775

desires to opetate with oxygen alone, the necessity of keeping the tem-
perature from 940° to 960® centigrade cannot be made too emphatic.

The oxygen is purified by passage through jars of stick caustic pot-
ash, potassium hydroxide solution, calcium chloride, and soda lime in
the order named. The oxygen then passes through a mercury valve
into the porcelain or quartz (fused silica) tube, half of which is filled
loosely with ignited asbestos. The products of the combustion are puri-
fied from acid, sulphur, litharge, or chlorine fumes by passing through
a jar of granulated 30-mesh zinc. The water is removed by a jar of phos-
phoric anhydride.

For steels containing 0.30 to 1.50 per cent, carbon two^ grams of fine
drillings, not over one-fourth mm. thick, are taken. For still lower per-
centages of carbon 3.0 to 5.0 grams of drillings of not over 20-mesh size
are selected.

The sample is weighed into a clay boat. (The boat is molded and
burned in the laboratory by a boy at a trifling cost.) The steel begins
to bum by the time the stopper of the combustion tube is in place. Two
grams of steel are decarbonized in three minutes and five grams in six
minutes. The burning is continued for ten minutes more with oxygen
passing through the combustion tube at a rapid rate. The weighing
apparatus is detached, wiped and weighed. Twenty-five minutes aflford
ample time for a single combustion, counting all operations.

The weighing apparatus and the jars for the purifying train are the
writer's design and were first published, in part, with illustrations in
the January Journal of the Engineers' Society of Western Pennsylvaniay
1906, and more fully in This Jqurnai,, 28, 862 (1906). This weighing
apparatus (J) is used forty times before it is refilled. As it is always
weighed against a duplicate for a tare, after the fortieth combustion its
tare is used as an absorber for forty more combustions, so that when a
pair has been freshly filled the operator knows he can complete eighty
combustions before he needs to refill his weighing outfit.

While no red lead is necessary for steel combustions, some of the al-
loys such as ferro-chrome, carbonless chrome, and ferro-boron, require
that red lead be mixed with the drillings or powder to break the metallic
bond and permit of decarbonization. Ferro chrome is the most refractory
as from a carbon content of more than 4 per cent, only 0.2 per cent,
was obtained by burning as in steels with oxygen alone, at a tempera-
ture of 940*^. Pig-iron also requires some red lead. In general, about
one-half the amount of lead required for decarbonization in a gas furnace
is suflBcient for the same work in the electric furnace, by reason of the
higher heat attainable within the range of durability. A few of the
many comparisons made in this laboratory between the combustions in

776 C. M. JOHNSON.

a gas furnace with red lead and oxygen and combustion in oxygen alone
are given in Table I:

Tabus I.

Weight of Amount of Per cent

drillings red lemd carbon

Sample. Method. taken. used. Soond.

No. I Steel Electric 4 grams none 0.09

" I '* Red lead 4 " 7 grams 0.09

" 288 " Electric 2 " none 1.176

" 288 " Red lead 2 " 4 grams i . 175

No. 2 Steel Electric 5 grams none o. 121

" 2 " Red lead 4 " 7 grams o.xii

" 3 " Electric lyi " none 0.976

" 3 " Red lead iK " 4 grams 0.967

" 4 " Electric 3 " none o. 109

" 4 " Red lead 5 " 7 grams 0.118

" 5 " Electric 2 " none 0.469

" 5 " Red lead 2 " 4 grams 0.474

" 6 " Electric 2 " none 0.736

" 6 " Red lead 2 " 4 grams 0.737

" 7 " Electric 3 " none 0.118

" 7 " Red lead 4 " 7 grams 0.117

" 8 " Electric 2 " none 1.17

" 8 " Red lead 2 " 4 grams 1.168

" 9 " Electric 2 " none 1.15

" 9 " Red lead 2 " 4 grams i . 16

" 10 " Electric 5 " none 0.046

" 10 " Red lead 4 " 7 grams 0.040

Tablb II. — Fbrro-alloys and Plumbago.

Weight of Amount of Percent,

drillings red lead carbon

Sample. Method. taken. used. fonnd.

Ttmgsten powder Electric 2 grams none 0.003

Red lead 2 " 4 grams o.oio

Plimibago, No. 153 Electric 0.3 " none 50 -700

Red lead 0.2 " 4 grams 50.800

Plimibago, No. 356 Electric 0.3 " none 51650

Red lead 0.2 " 4 grams 51 -300

Plumbago, No. i Electric o. 2 " none 94-90o

Red lead 0.3 " 4 grams 94-3a>

68.5 percent. Ferro-chrome Electric i.ogram i gram 4.21

Red lead i.o " 4 grams 415

Ferro-vanadimn, No. 134. Electric i.o " none 3-^2

Red lead 1.0 " 4 grams 3.09

Ferro-titanium, No. I Electric 2.0 grams none 0.22

Red lead 2.0 " 4 grams 0.24

Ferro-boron, No. i. ..... . Electric i .0 gram i gram i .73

Red lead 1.0 " 4 grams 172

Carbonless chrome No. 9 — Electric 1.0 " i gram 0.08

96.0 percent, chromium — Red lead 1.0 " 4 grams 0.09

Pig iron Electric 1.0 " none 3- 20

"B" Electric 1.0 *' o.sgnim 3 5«

Red lead 1.0 " 4 grams 3-5^

DETERMINATION OF CARBON IN STEEt, ETC. 777

The advantages of the electric heating apparatus are obvious. Very
little heat is radiated; economy of space is attained; tubes are heated
gradually and cooled gradually; time required is the minimum; labor cost
is plainly the lowest because of simplicity and rapidity and no expensive
platinum tubes or boats or crucibles are used.

Some may say, *' Why not bum the steel in air?" The answer is that
the cost of oxygen is small, 1/3 cent per combustion, and the steel bums
twice as fast. Oxygen can now be had at 5 c. per ai\ ft. in 100 cu. ft. cylin-
ders. The method is accurate for all steels. As pointed out in the
writer's article in This Journai^^ and in his preliminary paper read be-
fore the Pittsburg Section in Dec, 1905, one may lose as much as 50 per
cent, of the carbon in certain alloy steels by attempting to dissolve the
borings in either neutral or acid double chloride of copper and potas-
sium.

The best protection for the bottoms of clay or porcelain boats is a
liberal la3rer of ignited silica $and, such as is used for acid open-hearth
furnace bottoms. The silica rock is crushed to about 20-mesh and igni-
ted in a muffle furnace at a bright red heat, cooled, and kept in glass-
stoppered bottles.

To secure complete decarbonization it is necessary either that thin
drillings be used or, if the sample contains much coarse or bulky material,
it should be selected. This can easily be accomplished by pouring the
borings on a 20-mesh sieve and shaking all of the steel of 20-mesh
size and the still more finely divided dust on to a 60-mesh sieve, which
retains only the 20 to 60-mesh material. This always represents a good
average sample.

Further, the drillings should be placed in as compact a mass as possi-
ble. If curly drillings are scattered along the entire length of the boat
instead of being put in a deep, compact body, borings that are a little
thick will frequently be found to still contain unbumed metal. This
detail is a very important one. Of course, the reason is that drillings
lying in close contact heat each other to incandescence during the bum-
iag with oxygen.

Also, during the period when the oxygen is being absorbed in large
quantity by the burning metal, the flow of the gas shottld be regulated
so that there is a bare excess and no more. That is, the oxygen should
be turned on in sufl&cient quantity that the gas is bubbling through the
weighing apparatus slowly. If the gas is mshed through during this
period the steel becomes violently heated and slags with the sand and
the sides of the boat, destroying the latter. Worse yet, low results are
obtained frequently in this way, probably due to the formation of carbon

' Loc. cit.

778 C. M. JOHNSON.

monoxide, which is driven out of the hot portion of the tube before it is
oxidized to the dioxide.

If the oxygen is turned into the tubes in sufficient quantity to main-
tain a slow stream during the period of the burning, the end point of the
combustion is distinctly shown by a ^udden increase of the speed of the
bubbling through (J). The rush of oxygen is then checked but the rate
of flow is still mther rapid for the final ten minutes.

The weighing apparatus (J) is filled not quite to the bend of the inkt
tube with a solution of potassium hydroxide made by dissolving 500
grams of the latter in 500 cc. of water. The drying tube at the outlet of
J is closely filled with pieces of stick caustic potash cracked to about
half the size of a grain of wheat. To prevent the caustic potash from
coming in contact with the small rubber stopper in the drying tube a
loose plug of asbestos is placed at that point. The little bulb of this
drying tube is filled about half full of glass wool. If dry sticks of caustic
potash are cracked quickly, the small pieces can be conveyed to the dry-
ing tube in dry condition and constitute not only a splendid guard against
loss of moisture from J but are also equally effective as an absorbent
of carbon dioxide.

If a porcelain boat is used, the 15 x 75 mm. Royal Meissen boat is the
best shape and most endurable of any porcelain boats that the writer
'has tried. When putting in the sand bottom, fill the front half of the
boat about two-thirds full and then with the butt end of the forceps make
a trough in the sand, working it well up the sides of the boat. Pour
the drillings from the weighing bottle into this depression. By so doing
the drillings are kept in a compact mass and when the combustion is
completed the burned steel can be lifted out in a small cake. In this way
a boat can be used ten to fifteen times.

When a great many combustions are made daily, the fused silica, or
electro quartz, tube is the most durable. The continuous spraying of
oxides against the walls of a porcelain tube weaken it and when the cur-
rent is turned off and the tube is permitted to get cold the contractiOT
causes a rupture.

To prevent the contents of D, B, E from clogging the inlets and out-
lets, large plugs of cotton are used at these points. Glass wool ptags
should be used in H and loose plugs of ignited asbestos in I. Bnough
mercury is placed in the bottom of F and A to form a seal. The inlet
end of the quartz tube heats somewhat and it is better to wrap it several
times around with a strip of cheese-cloth, the end of which dips into a
150 cc. beaker of water suspended directly imdemeath by means of cop-
per wire. During the absorption of carbon dioxide the outlet of J is
protected from ingress of moisture or carbon dioxide or fumes from the

EXTRACTION OF POTASH FROM FELDSPATHIC ROCK. 779

room by a drying tube not shown in the^figure. It is filled with pieces of
stick caustic potash broken to the size of a pea.

I«ABORATORY OP THB PARK STBBL WORKS,

Crucible Stbbl Co. op Ambrzca,
Pittsburg, Pa.

(CONTMBUTIGN PROM THE OFPICB OP PUBUC ROADS, U. S. DBPT. AGRICULTURE.)

THE EXTRACTION OF POTASH FROM FELDSPATHIC ROCK.»

By Allbrton S. Cushman and Prbvost Hubbard.
Received March 4, 1908.

The extraction of potash from native rocks has long been considered
one of the most important as well as one of the most difficult problems
of industrial chemistry. In spite of the enormous resources of the North
American continent, there has not yet been found anywhere on it an
available source of potash, thus necessitating the importation from abroad
of many himdreds of millions of pounds per annum of the salts and com-
potmds of this important substance. ^ One result of this lack of a native
source of supply has been to stimulate the use in agriculture of hard wood
ashes, which are even at the present time brought from Canada in con-
siderable quantities to the added devastation of the fast disappearing
forests. In addition to this, cotton hull ashes from the South are shipped
to the North, which merely robs the soil of one portion of the country
to supply the deficit in another. The pegmatitic granites and feldspathic
dykes of the eastern and central western United States offer an unlimited
source of supply which only awaits an economical method for making
it available. Many of these feldspar deposits run as high as 10 per cent,
in potash (KjO) and it follows, therefore, that a quarry only fifty feet
square and fifty in depth contains about 2,000,000 potmds of this alkali.
Under the stimulus of the rapidly growing cement industry, great ad-
vance has been made in the last few years in the art and economics of
fine grinding, which must of necessity be the first step in any process
which attempts the extraction of potash from feldspar or other minerals.
At the present time, in the manufacture of Portland cements, at least
two extremely fine grindings as well as a burning at a high temperature
are accomplished so economically that the finished product can be packed
in bags or barrels and sold in some places for a price equal to about three-
tenths of a cent per pound or about six dollars per ton. In view of the
fact that a short ton of 10 per cent, feldspar contains about ten dollars'
worth of potash at prevailing prices the problem of extraction is not on
first thought an unpromising one from an economical standpoint.

There are about twenty well defined rock-forming minerals as com-

* Paper read before the N. Y. Section, Am. Chem. Soc., Feb. 7, 1907. Published
with the permission of the Secretary of Agriculttu*e.

780 AI^LERTON S. CUSHMAN AND PREVOST HUBBARD.

prised in the following tabular list, only a few of which, however, may
be considered as possible sources of raw material. Of these orthodase
and microcline are probably the most important. The potash feldspars
occur in large dykes or deposits in various parts of the country, and have
been particularly developed in Maine, Connecticut, New York, Pennsyl-
vania and Maryland, where they are mined and ground for use almost
exclusively in the ceramic industries. Many of these pegmatite de-
posits, however, because of insufficient coarseness, too large a percentage
of quartz, or too great an abundance of iron-bearing minerals, are un-
fitted for finer uses and therefore are especially available as a raw ma-
terial for potash extraction.^

Potash-bearing Sii<icates Arranged According to Dana.

orthosilicates.

KaUophilite: Silicate of alumina and potash. EI^O, 27.20-29.30
per cent.

Microsommite : Sulpho-silicate of alumina, lime, soda, and potash.
K|0 , 6.25-7.82 per cent.

Nephelite: Silicate of alumina, soda, and potash. K,0, 4.55-7.14
per cent.

Hauymte: Sulpho-silicate of alumina, lime, and soda (potash). K,0,
0.33-4.96 per cent.

Algerite: Silicate of alumina, magnesia, and potash, and water with
some CaCO,. K3O, 9.97 per cent.

METASIUCATBS.

LeuciU: Silicate of alumina and potash. K^O, 18.90-21.48 per
cent.

POLYSILICATES.

Orihoclase: Silicate of alumina and potash. K,0, 5.40-15.99 per
cent. Constitutes at least 15 per cent, of the earth's crust and occurs
in nearly all varieties of acid igneous and metamorphic rocks, as well as
in many sandstones, conglomerates, etc. ^^

Microcline: Same as orthoclase, except as to crystal habit.

Hyalophane: Silicate of alumina, baryta, and potash. K,0, 7.82-
II. 7 1 per cent.

Anarthoclase: Silicate of alumina, lime, soda, and potash. K|0,
2.50-1 1.90 per cent.

HYDROUS SIUCATES.

Muscovite: (White or potash mica) silicate of alumina and potash.
KjO, 6.83-1 1. 10 per cent.

Lepidolite: (Lithia mica) fluoro-silicate of alumina, lithia, and potash.
KjO, 10.78-12.34 per cent.

^ See Mineral Resources of the U. S., 1906, E. S. Bastin.

EXTRACTION OF POTASH FROM FELDSPATHIC ROCK. 7^1

Zinnwaldite: (Lithia mica) composition like lepidolite but contains
iron. KjO, 10.46-10.58 per cent.

Biotite: (Iron or black mica) silicate of alumina, magnesia, iron and
potash. KjO, 6. 18-10.08 per cent.

Phlogopite: (Magnesia mica) silicate of alumina, magnesia, and pot-
ash. KjO, 7.06-10.32 per cent.

Lepidomelane: (Black mica) silicate of alumina, iron and potash.
K^O, 6.06-9.45 P^r cent.

Roscodite: (Vanadium mica), silicate of vanadium, iron, magnesia,
alumina and potash. KjO, 7.59-8.87 per cent.

ZeolUes: Hydrous silicates of alumina, lime, soda, run from traces
to 11.09 per cent. KJO. The principal potash-bearing varieties are Pkil-
lipsite (0-11.09 per cent. K2O), Chabasite (0-4.39 K^O), Analcite (0-2.83
per cent. K,0), and Nairolite (0-1.17 K^O).

The question of the use in agriculture of very finely grotmd feldspar
without any further treatment has previously been discussed by one of
us,* and systematic experiments are being carried on year by year by
the Department of Agriculture in order to test the value of potters' feld-
spar ground to 200 mesh as a tobacco fertilizer. It has been pointed
out that by grinding the spar, either dry or in the form of a slurry with
lime, ammonium salts or gypsum, the potash can be made more quickly
available to plants.' It is an interesting question whether these simple
methods could not be developed in order to prepare feldspathic rock
for use in agriculture.

The principal representatives of the potash soda feldspar group, ortho-
dase and microcline, have the composition generally approximating
to R,0, AI2O3, 6Si02, in which R may be either potash or soda. It is
probable that very few of the large deposits will run higher than 10 per
cent, potash and in many the percentage of soda is considerable and
presents a decided complication in any scheme of potash extraction.

When in a finely grotmd condition, orthoclase is not a difiicult silicate
to decompose and a number of methods have been proposed and suc-
cessfully carried out on a small scale. None of these have, however,
up to the present time developed into successful commercial operations.
Water alone, to a slight extent, decomposes feldspar and will extract a
small amount of potash from fine-grotmd orthoclase, but the action is
not continuous and soon practically ceases, owing to the accumulation
of the resulting decomposition products, which may form new combina-
tions of a more or less insoluble nature and which also protect the active
surface area of the particles from continued decomposition. If the in-
soluble products are removed by abrasion, solution, electrolysis, or other

» BuU. 104, Bur. Plant Ind., U. S. Dept. Agr.

• Bull. 28, Oifice Public Roads, U. S. Dept. Agr., p. 16.

782 Al^I^ERTON S. CUSHMAN AND PREVOSt HUBBAIO).

means, the action goes further, but in any case the amount of decompo-
sition depends upon the available surface offered by the powder and there-
fore upon its degree of fineness. This is true of any reaction which is
purely a surface one and is a fstctor which, in many similar cases, deserves
more consideration than is ordinarily given it. Some figures have been
presented in a previous publication which show the enormous increase
in surface area which results from reducing one pound of feldspar in the
shape of a solid cube to the condition of the finest possible powder. It
has been shown that whereas the original cube would have a surface
area of 29.3 square inches, the same material reduced to particles just
capable of passing a screen containing 200 meshes to the linear inch would
have an area approximately equal to 24,900 square inches. If further re-
duced entirely to the finest possible powder, the surface area mounts into
millions of square inches.^

In a previous publication on the decomposition of the feldspars the
authors have already discussed to some extent the effect of fineness of
grain on the rate of decomposition and have presented certain data in
regard to an electrolytic method for extracting the soluble alkalies from
ground rock.' Some results of a study of very fine powders have also
been presented by the authors in a paper on the air elutriation of fine
powders in which a laboratory method for separating the very finest powders
from coarser particles was described and the relation of fineness to sur-
face area discussed.' By means of an apparatus which may be briefly
described as consisting of a series of settling chambers through which
the powdered spar is forced by air pressure, a number of samples of the
material were obtained differing considerably in the relative size of their
particles. From microscopic measurements of the average diameter
of the largest, smallest and medium-sized particles in each sample, and
taking into account the relative proportion of each of these different
sizes present, a close approximation was made of the actual surface area
presented by a unit weight of powder, by means of the following formula
where M|, M,, M, equal the per cent, of large, medium and small sized
particles and Z^, /, and /, equal their respective diameters:

.-./•Ml

«(x'+f+^)-

The object of this paper is to present the results of the continuation
of these researches, together with a brief review of our own previous
work and that of other investigators.

In our former work the decomposition of an impure orthoclase, as

* BuU, 104, loc. cit,

■ BuU. 28, Office of Pvblic Roads, U. S. Dept. Agr,

• This Journai<, 29, 4.

EXTRACTION OF POTASH FROM FHI^DSPATHIC ROCK. 783

procured in bulk from a commercial firm, was studied, no attempt being
made, however, to obtain exact measurements of the fineness of the prod-
uct, although it was fotmd that nearly 98 per cent, would pass through
a standard 200-mesh sieve. This material was found, by analysis, to
have the following composition and was used in all of the experiments
described in this paper:

SiUca (SiOJ 68. 29

Alumina (A1,0J 18. 27

Potash (K,0) 9-32

Soda (Na,0) 3.60

Phosphoric anhydride (PaO^). 0.53

Total 100.01

When leached with water, the powder gave a yield of 0.025 P^r cent.
alkalies in solution, although it was demonstrated that a greater amount
had actually been produced by the action of the water and that the
difference between this amount and that passing into solution had been
combined with or absorbed by the insoluble substances also formed.
Wet grinding in a ball mill, by which the particles were further broken
down and the insoluble products removed by abrasion, produced a yield
of 0.32 per cent, alkalies. It was fotmd that the absorbed alkalies could
be removed by electrolysis, and by this method the dry ground powder
gave a yield of 0.14 per cent, alkalies while as high as 0.45 per cent, were
extracted from the wet re-ground material. From fourteen re-grindings
and electrolytic runs on the same sample it was found possible to ex-
tract 3.54 per cent, alkalies with water alone, which represents a decom-
position of about 27 per cent, and indicates that total decomposition
could be brought about by repeating the operation of regrinding and
electrolysis a sufficient number of times. Because of the time and energy
involved in this method, the results obtained are, of course, interesting
from»a theoretical point of view only.

The action of water can, of course, take place only upon the free sur-
faces of the particles, and as this surface is a function of fineness of grain
it was decided to attempt the determination of the relative amount of
alkalies which could be extracted by water from samples of different
degrees of fineness. For this purpose some of the feldspar was obtained
in massive form, crushed in an iron crusher and then passed through a
set of standard mesh sieves. The powder retained between each consecu-
tive pair of sieves was given a number, and its particles assigned the
average diameter of the two meshes. For the coarse material this method
was considered sufficiently accurate. Material which would pass through
the 200-mesh sieve was separated by the air elutriation into three sam-
ples and microscopic measurements were made as previously described*

784

ALLERTON S. CUSHMAN AND PRHVOST HUBBARD.

The area of a unit weight, or in this case, a unit volume (i cc), of solid
material broken down to the sizes represented by the different samples
was then calculated, from the formula previously given, for use in com-
paring the relative amount of the alkalies liberated. The results of
these measurements are given in the following tables:

Table I. — Coarse Powders.

No.

Mesh.

Average diameter,
mm.

Area per oc.
sq. cm.

10- 20

I . 3970

20- 30

0.7239

30- 40

0.4695

128

40- 50

0 • 3465

173

50- 80

0.2493

241

80-100

0. 1605

374

100-200

0 . 0994

604

Table II.-

-Fine Powders.

Diameters.

Area

sq.

n^r cr

No.

mm.

Pcr cent.

mm.

Per cent. mm.

Per cent.

pel Wk.

cm.

.OI22

8 0

.0061

15 0 . 00090

, 202

.0063

3 0

.0025

10 0.00018

292

,685

.0037

3 0

.0007

14 O.OOOIO

510

,486

In order that a number of electrolyses could be carried on at the same
time and under the same conditions, a battery of six cells was constructed
as shown in Fig. i.

-©-

>0 ° oQ ^ ^O ° ^O ^ °0 ® ^O

Fig. I. — Arrangement of electrol3rtic cells.

Each of these cells consisted of a 400 cc. Jena glass beaker which served
as cathode compartment and a cylindrical wooden pine cup 5 inches
high — 1.5 inches external diameter and 1/8 inch thick, for the anode or
slime chamber, which also served as a porous diaphragm. An adjusta-
ble rubber collar placed around the tops of these cups allowed them to
be hung in the rack above the beakers so that the bottom of each cup
was about 0.5 inch from the bottom of the corresponding beaker. The
anodes were made of heavy platinum wires which passed through stoppers
fitted in the mouths of the wooden cups, and the cathode in each ceD

EXTRACTION OF POTASH FROM FBLDSPATHIC ROCK. 785

consisted of a layer of mercury covering the bottom of the beaker and
connected to the binding post by means of a platinum wire running
through a sealed glass tube. The binding posts were arranged in such a
manner that the cells could be connected either in parallel or series.
When run in parallel the anode and cathode wires were connected to the
binding posts joined to the main line while a series connection could he
made by means of the auxiliary binding posts placed between the cells
in such a manner that the cathode of one could be connected to the anode
of the next. The cups were first treated with alcohol to remove as much
resin as possible and then well soaked in water. Blank runs were made
upon each cup until no indication of alkalinity could be observed by the
use of phenolphthalein in the cathode compartment.

Twenty grams each of the coarse samples were slimed in the anode
chamber with 50 cc. distilled water and 200 cc. distilled water was placed
in the cathode compartment. The cells were run in series on a 220-
volt line. A number of 48-hour runs were carried on until practically
no more alkalies were liberated, and titrations of the cathode liquor with
N/io HNO, made at the end of each run to determine the amount of
K,0 + Na,0 extracted. Each run was begun with fresh distilled water
in the cathode compartment. Electrolyses of the fine powders were

Relative arck (sq. cm. percc.).

Fig. 3. — Electrolyas of orthoclase with water (coarse powders).

786

allbrTon s. cuskman and prevost bubbard.

made in the same manner with the exception that, the supply being some-
what limited, 3 grams were used instead of 20. The results for both
the coarse and fine powders are given below and are folloited by dk-
grams (Figs, 2 and 3). It was found necessary to construct diagrams

Fig. 3. — Electrolysis of orlhocla.se with water f fine powders),
of different scale for the coarse and fine powders, as the relative areas
exposed make it impossible to compare the different order of magnitudes

Table III. — Coarsb Powders.

No. I. No. I. No. 3. No. 4- No. 5- No. 6. Me.;.

Sq. cm. per cc 43 83 118 173 241 374 604

Per cent. KjO + Na,n 0.013 0.017 0.021 0.016 0.027 0029 0 os<

Table IV.— Fine Powders.

No. 8, No. 9. No. ».

Sq. cm. per cc 53. 'o' ^92.685 5104**

Per cent. K,0 + Na.0 0,423 0,683 "S;;

Although these results plainly show an increased yield, the solutini
effect is not in exact ratio to the increase of surface area. It is prote*
bly true that as the particles decrease in size the tendency to clump and
coagulate increases, and that, therefore, the theoretical surface area ob-
tained does not entirely come into play. Although our investigatimis
do not include a sufficient number of different sized powders to demon-
strate this point, it is apparent, from theoretical constdemtions, tbat

EXTRACTION OF POTASH FROM FELDSPATHIC ROCK. 787

some reverse action must take place with increasing fineness. If
this were not tnie, powdered orthoclase made up of ultimately fine par-
ticles should be rapidly and completely decomposed by water, which
is not the case. This point will be discussed more fully later on, but it
may be stated that this case is not essentially dissimilar to that of the
gas law, which is subject to an increasing correction with increasing
pressure until finally it no longer holds. It is not to be supposed that
the decomposition of the feldspar is actually effected by the action of
electrolysis, as the electric current merely transfers the soluble basic
products of the hydrolysis. The simplest reaction which can be written
to express the action of water on orthoclase is as follows:

KAlSijOg + HOH = HAlSigOg + KOH.

By a splitting off of silica from the hydrated aluminum silicate we
can account for the formation of kaolin.

iHAlSijOg + H,0 « (H,0)a.AlA.(SiO,), + 4Si03.

It is certain, however, that the actual reactions which take place dur-
ing the process of kaolinization in nature are more complex than these,
and that intermediate zeolitic compounds of varying composition are
formed in which water, if not potash itself, is held in the form pf a solid
solution.^ It is difficult to determine whether similar intermediate prod-
ucts are formed when water acts on orthoclase powder under the* con-
ditions maintained in these experiments. It has been shown, however,
that in whatever condition the potash is held in the decomposition prod-
ucts of orthoclase it is possible to extract it by electrolysis.

As feldspar is known to be partially decomposed by concentrated sul-
phuric acid, it was thought that a comparison of the decomposition of
different sized powders could be readily obtained by this means. A series
of digestions and determinations of alumina was accordingly made, and
while the time exposed and relative proportions of powder and acid varied
somewhat in the different experiments, as described under the respec-
tive tables, the method in general was as follows:

Weighed amounts of the powder were mixed with known volumes
of HjS04 (sp. gr. 1.84) and allowed to digest on the steam-bath for definite
lengths of time. They were then diluted and filtered, and either the whole
or an aliquot part of the clear filtrate was analyzed for alumina in the usual
manner, by precipitation with NH^OH. If the filtrate was cloudy, a suflS-
cient amount of ammonia was added to cause the suspended particles
to settle, but not enough to neutralize the acid and precipitate the alumina.
An aliquot portion of the clear supernatant liquid was then analyzed.

The coarse powders were obtained in the same manner as described
for the electrolysis, with HjO, with the addition of thoroughly washing

' Z. anorg. Chew., 15} 318.

788

ALLBRTON S. CUSHMAN AND PRBVOST HUBBARD.

them on the sieves with a. strong stream of water after they bad been

dry-sifted. They were then dried and all traces of iron from the cnisher

removed by means of an ekctro-magnet. &x grams of each sample

were digested with lo cc. of acid for eighteen hours, as described. Fig.

4 and its accompanying table give the results of these experiments.

Tablb v. — Coarse Powdbhs.

Sq. eto, per cc . ^j. Sj. 141 *«■

Per cent. A1,0, 0.041 0.078 0.098 0.160

RelatiTc area (w). cm. per cc).
Fig. 4. — Solubility of orthoclase in Sulphuric add {coarse powders).

Four different experiments were made upon the air elutriated powden,
the conditions and the results being given in the following table,
made to accompany Fig. 5. Microscopic measurpments and determina-
tions of relative volume proportions were made in the same manner as
described for the fine material in the electrolytic experiments. As the
surface areas were also calculated in the same manner, it seems unneces-
sary to give the actual measurements here.

Table VI.

Sym

Chamber.

Chamber

boi

Cooditloni.

3 gms., 34 hrs.

Percei

I. A1,0,

0.333

0.747

10 cc. H^.

sq. on

percc.

5,958

11.184

6 gms.. 18 hrs.

Percer

t. A1.0.

o.a67

0,480

10 cc- H^,

sq.cm

percc.

5.958

II. .84

□

6 gms., 96 hrs.

Percer

t. A1,0.

0.313

0,643

1,090

rocc.H^,

sq. cm

percc.

5,958

11,184

34. =84

6 gms., 19 hrs.

Percet

t. A1,0.

0..87

0,357

0,587

10 cc H^,

sq. cm

percc.

5.<i8

9.546

J., 366

EXTRACTION OF POTASH FROM FELDSPATHIC ROCK. 789

Fig- 5- — Solubilitj' of orthoclase in sulphuric acid (fine ponders).

In reviewing the diagrams for solubility with respect to area exposed,
it will be seen that there is unquestionably a tendency toward greater
solubility as the powder becomes finer, but here again in no case is the solu-
bility shown to be directly proportional to the surface area. As
these differences are in every case too great to be laid to the charge of
experimental error in analysis, it was decided that it would be of inter-
est to determine the solubility of some substance about which there could
be no question in regard to surface area exposed to the action of a sol-
vent. Sheet copper was selected as being the most homogeneous ma-
terial readily obtainable, and pieces of known area, after being thor-
oughly cleaned, were subjected to the action of normal nitric acid for
the same length of time in the same vessel, especial precautions being
taken to keep the temperature as constant as possible and thus to avoid
convection currents. The loss in weight of the samples was then de-
termined and the results plotted as shown in Fig. 6.

A number of tests were made, in which the size of the test pieces and
the concentrations of acid were varied, but in every case the curve repre-
senting any one series was more or less erratic in its direction.

It is certain that varying conditions play an important part in the rela-
tive solubility of different sized particles of the same material, but within
certain limits the solubility or decomposition effect rises rapidly with
increaang fineness.

790 AIX8RT0N 5. CUSHMAN AND PREVOST HUBBARD.

In previous work it was demonstrated that when powdered feldspar
was slimed with water and a small amount of hydrofluoric add and elec-
trolyzed in the manner described for the water electrolyses, a far greater
amount of decomposition took place than could be theoretically deduced
from the action of the quantity of hydrofluoric acid present.' Thisiesutt
was due to the regeneration of the acid and in the run described it vas
found that 87 per cent, decomposition had been produced by the action
of 20 cc. of 35 per cent, hydrofluoric acid upon 200 grams of thegro'indoi-
thoclase. All of the bases, inchiding alumina, were earned to the catlnde
compartment, the alumina being held mainly in solution as an atkaliiit
alumioate, while silica al(^e is the final product left at the anode.

Square iuche*.
Pig. 6. — Solubility of copper in nitric acid.
The mechanism of these reactions would seem to be as follows: Taking,
tor the sake of convenience, an abbreviated form of the orthocJase mok-
cule and conddering the primary reacticfti when attacked by hydroflu""'
acid, we may write the following equations: ^

I. 4KAlSi,0, + 64HF = 4KF + 4AIF, + i2SiF, + 32H,0.
If potassium 'fluoride is added to a solution of aluminum fluondc, a
difficultly soluble double salt is formed, and so we should exp^ tbc
following reaction :*

' BuU. 38, Ol^ce PubUe Road , .".'. S. W«^. A^r.
' Danuner'B Anorg. Cbetn., Ill, 97.

KXtRACtlON OF POTASH FrOM FBI*DSPATH1C ROCK. 79 1

2. AlFg + 4KF - A1F,.4KF.

This slightly soluble double salt is slowly hydrolyzed, the potash
and alumina being carried to the cathode chamber while the hydrofluoric
add is set free at the anode to immediately attack fresh orthoclase par-
ticles.

The. silicon tetrafluoride is changed to hydrofluosilidc acid and silicic
add according to :

3. i2SiF4 + i6H,0 = 4Si(OH)4 + 8H^iF«.

Finally it appears that the hydrofluosilidc acid is again broken up by
hydrolysis according to :

4. H,SiF, + 4H,0 = 6HF + Si(OH),.

It has been shown that only about one-tenth of the amotmt of hydro-
fluoric acid necessary to complete the reaction, according to the stoichio-
raetrical relations involved, suffices to bring about complete decomposi-
tion of the orthoclase. This can only be explained by supposing that
some such reactions as are given above take place.

The next problem was to study the effect of fineness of grinding on
the rate of decomposition. No relation between the amount of decom-
position with respect to area exposed can be determined imless the re-
action proceeds to a finish and as, in the electrolysis of orthoclase with
hydrofluoric acid, this point is reached only when decomposition is com-
plete, it is evident that a comparison between these two factors is impos-
sible by this means. The effect of fiineness upon the time necessary for
complete decomposition imder the same conditions can, however, be de-
termined, and with this object in view electrolyses with hydrofluoric
add were made on the three fine powders described as Nos. 8, 9 and 10
in the water electrolyses. One cubic centimeter of hydrofluoric add
was added to each of the anode liquors and the nms continued as be-
fore, with the exception that the cells were connected in parallel instead
of in series. The length of these runs was the same for all of the cells
in each particular case, but the amount of current flowing varied some-
what, of course, with the individual cells. In the sixth run another
cubic centimeter of add was added in order to hasten the operation.
The cathode liquors were carefully titrated with standard add after
each run, and the results obtained are given in the following table:

Tablb VII.

Time. No. 8. No. 9. No. 10.

Run. Hours. cc. cc. cc.

With water alone 2 . 70 4. 35 5 . 55

1 with H F 48 6.45 8.55 13-65

2 " " 24 8.55 9.65 11.20

3 " " 24 7.70 9.00 II. 15

4 " " 24 5.00 5-35 7.35

5 " " 24 0.30 2.20 2.80

792 AI*I*ERTON S, CUSHMAN AND PREVOST HUBBARD.

TablS VII {(CofUinued).

Time. No. 8. No. 9. Na 10.

Run. Honrs. oc. cc. oc.

6 " " 24 5.40 10.80 18.20

7 " " 24 2.00 6.40 6.15

8 " " ,. 24 2.20 8.95 6.10

9 " " 24 5.45 11.25 1.95

10 " " 48 9.65 5.20 0.15

11 " " 24 8.85 0.50 0.00

12 " " 24 8.70 0.30

13 " " 24 5.15 0.00

14 " " 24 2.70

15 " " 24 2.40

16 " " 24 1.50

17 " " 24 o.io

18 " " 24 0.00

Total cc 84.80 82.50 84.25

No. of hours 456 336 288

It will be noticed that the reaction was completed with the finest
powder first, the next finest second, and the coarsest third.

The actual amounts of alkalies set free as shown by the total titra-
tions are given in the following table. In order to calculate the per cent.
decomposition which this represents, it was found necessary to deter-
mine the quantity of alkalies present in the three classes of material
obtained from the air elutriator, as the relative amount of quartz pres-
ent in the samples varied with the diflFerent sizes, owing to its greater
hardness and consequent lack of fineness. All of these results are given
below:

Tablb VIII.

No. 8. No. 9. No. 10.

Alkalies liberated from 3 gms o . 3998 gni. o. 3890 gm. o. 3972 gm.

Per cent, alkalies liberated 13 . 33% 12 . 97% 13 . 24%

Per cent, alkalies contained 13 . 29% 13 . 63% 13 . 74%

Calculated per cent, decomposition. 100.00% 95- 16% 96.36%

To check this work very roughly, an analysis was made of the residue
obtained from the electrolysis of sample No. 3. It was found to con-
tain silica and alumina in the following proportions :

Per cent.

SiOj 98. 80

AlaO,+ 1.13

Total 99-93

It would seem, therefore, that all of the alkalies had been liberated
instead of 96.36 per cent., as represented in the preceding table. The
low results obtained on samples 9 and 10 by titration are undoubtedly
due to a slight diffusion of the hydrofluoric acid through the wooden

EXTRACTION OF POTASH FROM PBlDSpATHIC ROCK. ?93

cups to the cathode compartmetit. The reason for the fact that some
alumina was present in the residue is explained by the low solution pres-
sure of aluminum fluoride, which allowed the more soluble alkaline fluo-
rides to be transported flrst. It was noticed that as the yield of alkalies
decreased alumina came over in greater amounts, and this transfer would
have undoubtedly been complete had electrolysis been carried further.

Assuming, therefore, that decomposition in each of the three cases
was complete, the approximate effect of surface area upon time neces-
sary to bring about this decomposition can be shown as in Pig. 7, when
the ordinates represent the time covered by the total number of runs
and the abscissas the surface area presented.

RcIaUn am (looo iq. cm. per cc).

Fig. 7. — Electrolytis of ortboclase with hydrofluoric acid (fine powders).
In reviewing the diagrams showing the effect of fineness upon the ex-
tent of decomposition with water, as well as upon the time required to
bring about complete decomposition with hydrofluoric acid, it will be
noted that while the results do not strictly conform to theory, important
differences in the reacting power of the different sized powders are shown
to exist. The same remarks which were made upon the experiments
with sulphuric acid are applicable here, and while it is true that a num-
ber of factors which it is impossible to take into account have a very
important bearing upon work of this nature, it is neverthekss evident
that in the extraction of potash from feldspathic rocks or, in fact, in any
process of a similar nature, the effect of surface area of fineness of the
material is a factor which deserves serious consideration. The slight
solubility of the double fluorides of aluminum and potassium is one cause
of the slowness of the separation of the bases in the electrolytic pnw^ss.

794 ALLERTON S. CUSHMAN AND PREVOST HUBBARD.

but it is possible that this difficulty might be overcome if the work were
being conducted upon a large scale. Even if these difficulties did not
exist, however, the prevailing price of potash and high cost of electrical
energy would probably be a bar to the electrolytic process.

A study of the decomposition of the feldspars could not be considered
complete imless it included all the methods of attack which have been pro-
posed by others, or which have suggested themselves to the writers.
The analytical method of J. Lawrence Smith, for decomposing silicates,
has very naturally been made the basis of a number of processes, on some
of which patents have been granted in various countries. Patents for
fusion methods with salt or lime, or both, in combination, were granted
in England to Tilghman 1847, Newton 1856, and Ward 1857.* In
1882 Spiller proposed a method for making potash alum from feldspars,
which depended upon treating a mixture of the ground minerals with
sulphuric acid. Pemberton showed later that this method could not
possibly produce alum for a sufficiently low cost to justify its use.' More
recently, Rhodin's process of fritting at 900°, a mixture of 100 parts of
ground feldspar with 53 parts of quicklime and 40 parts of salt, has been
to some extent exploited in England and in Sweden.' By this process
it is claimed that about 90 per cent, of the potash present is converted
into chloride and can be easily leached from the frit. It is possible that
this mixture could be burned in continuous rotary kilns operated in a simi-
lar manner to those in use in cement manufacture. The fact, however,
that the salt and lime combine with the silica and alumina to make a bv-
product of little or no value has undoubtedly prevented the development
of this method into a commercial process. Lake has lately obtained
an English patent* for treating leucite with sodium hydroxide or carbon-
ate and quicklime. The inventor writes the following reactions in de-
scribing his process:

1. Alj08K30.4SiOa + 4NaOH + sCaO =

Na20.3Ca0.4SiOa -f AI^O^.Nsl^O -h K,0 + 2H,0:

2. Al203.K30.4Si03 + 2NajC08 + sCaCOH), =

Na^O . 3CaO . 4Si03 -f AljOj-Na^O -f K3O -b 2CO, + 3^,0.

Although complete decomposition imdoubtedly takes place by this
method, the above reactions are open to criticism. There is no appar-
ent reason why the soda should all enter into combination, leaving
potash in a free state. Blackmore^ has been granted a patent for
the separation of alkali salts from insoluble combinations by treat-

• /. Soc. Chem. Ind., 20, 5, 440.

• Chem. News, 47, 1 206, 5.

• /. Soc, Chem. Ind., 20, 5, 438.
^ English Pai,, 17,985.

• • U. S. Patent, 772,206 (1904).

EXTRACTION OF POTASH PROM F^I^DSPATHIC ROCK. 795

ment with carbonic acid under high pressure. The writers have had no
experience with this method, but it is doubtful whether, even if it is
successful chemically, it could be developed into a commercial process.

The writers have found that fusion or fritting with calcium chloride
will decompose feldspar but here again, even under the most favorable
circumstances, it is unlikely that such a method of attack could be made
economical.

Sodium nitrate and mixtures of sodium carbonate and nitrate decom-
pose feldspars at comparatively low temperature. As sodium nitrate
is extensively used as a fertilizer, and as it may possibly be eventually
manufactured in part from atmospheric nitrogen, it is at least interest-
ing to note in passing that it is possible to enrich it with soluble potash
by fusion with ground feldspar and regrinding.

It will be seen, however, that the weak point in all the methods of at-
tack so far discussed is the unavoidable formation of large quantities
of by-products which, though made from more or less costly raw ma-
terials, are of no value in the end. This difficulty can apparently only
be overcome by the use of potash compounds to attack the feldspar,
since the potash used is at least as valuable after the process is completed
as it was before. The writers have tried a number of methods of attack
with various compounds of potash. Some of these experiments have
led to interesting results which may possibly be found to have a bearing
upon the practical problems of potash extraction. If fine-ground ortho-
clase is mixed with potassium carbonate or hydroxide and heated at a
dull red heat for a short time, the decomposition of the feldspar takes
place rapidly. The decomposition becomes complete when the ortho-
clase and potassium carbonate are present in the ratio of 1:1.6 parts.
Ou taking up with hot water a white flocculent precipitate of definite
composition is formed which is readily soluble in dilute acid. On care-
ful analysis of this white precipitate, made from different samples of ortho-
clase, it appears to be the potassium analogue of natrolite, the well-
known sodium aluminum orthosilicate, the formula of which is

(H,0),K30Al303(Si03)3.
Fusions were made of two different orthoclases in the following manner:

One-half gram of the powdered orthoclase was ground with 0.8 gram
of potassium carbonate, the mixture transferred to a platinum crucible
and heated to a dull red for 15 minutes. When cool, the fusion was taken
up with hot water, digested on a steam-bath for five minutes in order
to obtain a flocculent precipitate, and filtered, the precipitate being
washed with boiling water until the last washings were neutral to litmus.
The precipitate or synthetic silicate was then dissolved in hydrochloric
add and analyzed in the ordinary manner. The following table gives
the results obtained, calculated in per cent, of the original powder taken:

796 ALLERTON S. CUSHMAN AND PREVOST HUBBARD.

Table IX.

Synthetic sili- Synthetic sili-

Theo- Ortho- cate from A. Ortho- cate from B.

retlcal clase. , ^ » clase. • »

orthoclase. A. i. 2. B. i. 2.

SiO, 64.84 68.29 33.68 33.80 66.52 32.78 32.78

A1,0, 18.29 18.27 18.36 18.24 18.32 18.32 18.26

P2O5 0.53 0.15 0.16 ...

K,0 16.87 9.32 16.28 16.39 13-53 1754 17-36

Na,0 3.60 1. 19 1. 16 1.63

icx).oo 100.01 69.66 69.75 100.00 68.64 68.40

Using the data given above, the percentage composition of these pre-
cipitates can be calculated, showing the probable synthesis of a potash
analogue of natrolite. The percentage composition of an anhydrous
potash form of natrolite is also given for the sake of comparison.

Composition of Composition of Composition of

anhydrous potash anhydrous residue anhydrous residoe
form of natrolite. from orthoclase A. from orthoclase B.

SiOa 47.97 48.51 47.84

A1,0, 27.06 26.32 26.69

KiO 24.97 25.17 25.47

100.00 100.00 100.00

The above results c^n best be interpreted by the following reactions:

1. Reaction during fusion.

K^O . AlACSiOj)^ 4- 6K2CO3 = eKjSiOg + K^Al^ + 6C0^

2. Reaction of fusion with water:
6KaSiOs + K2AI2O4 + sHjO =

(H20)2.K20.Al203(SiOj)3 -f 6KOH -f 3K^0y
Leaving aside the question of whether or not such a method of attack
could be considered from a commercial standpoint, it is none the less
apparent that complete decomposition of the feldspar takes place, with a
synthesis of 'a silicate rich in potash which would be easily made availa-
ble in the soil or for further extraction treatment in the factorv.

The latest process for the extraction of potash from feldspar which
has been granted a patent in the United States is that of Swayze,^ which
takes advantage of the principle of the potash attack upon the silicate.
According to the specifications for this process, coarsely ground feldspar
is first heated to partially destroy its crystalline structure, and it is then
heated under pressure with a strong solution of caustic potash. By sub-
sequent chemical treatment it is proposed to manufacture potash, alum
and fine silica as the end products of the various operations. Although
it is doubtless true that orthoclase can be largely decomposed by this
method, the reactions involved are necessarily similar to the fusion meth-
ods just described with subsequent treatment with water, and it is doubt-

* U, S. PatetU, 862,676 (1907).

FLASK FOR FAT DBTHRMINATION. 797

fill whether the necessary separations of the products could be profita-
bly made.

Another method which has suggested itself, as a result of the writers'
investigations, is treatment of the ground orthoclase with a certain pro-
portion of hydrofluoric acid, taking advantage of the reactions given on
pages 790-1, but without the subsequent electrolysis. By filtering cold
on a cloth filter, nearly all of the potash present is held in the residue in
the form of a double fluoride with alumina. By subsequent heating
with ground limestone or lime, the pdtash is readily made soluble, and,
if it is desired, can be leached out. Unfortunately, the excess of lime
present makes it impossible to recover the hydrofluoric acid.

It would seem probable that some one of these methods which have been
described or suggested, could be developed under favorable circum-
stances into a successful commercial process. The above data are pre-
sented in the hope that it will stimulate experimentation on a larger scale
than is possible in a chemical laboratory and result in a successful solu-
tion of an important industrial problem.

The restdts given in this paper may be summarized as follows :
(i) Fine grinding of feldspars renders the potash partly available
under the action of water. The addition of certain substances, such as
ammonium salts, lime and gypsum, increases this effect.

(2) It is possible to completely extract potash by an electrolytic method
either with or without the addition of hydrofluoric acid, but it is prob-
able that this method could not be used commercially on account of its
cost.

(3) The effect of fineness of grinding has been studied and data given
showing the relation of surface area to rate of decomposition.

(4) It is shown that there are numerous fusion methods which could
be used successfully if the cost were not too high. The attack on the
silicates by means of potash or its compounds yields some interesting
reaction products which might possibly be made use of.

(5) The attack with hydrofluoric acid is suggested as a possible method
that deserves further study.

[Contribution prom the Bureau op Chemistry, U. S. Department op Agricul-
ture.]

FLASK FOR FAT DETERMINATION.

W. I,, Dubois.
Received Pebruaiy 21, 1908.

The Knorr apparatus for the extraction of fat employs a flask which
is fragile, very difficult to clean and expensive to replace. A number of
attempts have been made to supplant these flasks with simpler ones

798

WILLIAM A. JOHNSON.

Fig. I.

more easily deaned, and the breakage of which would not be such an
important matter.

The first flask designed to meet this requirement is that described by
Wheeler and Hartwell.^ In this apparatus the designers have used a

straight-necked flask holding about loo
cc, fitted with a rubber cup channeled
so as to receive the condenser. Some
workers, however, have found this de-
vice somewliat unsatisfactory, owing to
the short life of the rubber cups and
some danger attending their use. The
modification of this idea, shown in Fig. i,
was designed by Mr. F. W. Robison, of
the Michigan Dairy and Food Depart-
ment. As is indicated in Pig. 2, the seal
consists of a maple cup made to fit over an ordinary
rubber stopper through which the neck of the flask is
passed. The seal is made by mercury in a mamier
similar to the device employed by Wheeler and Hart-
well. This form of apparatus is now used in one of
the laboratories of the Department of Agriculture and
is considered a great improvement over the Knorr
flask.

The flask designed by the writer and shown with
connections, in Fig. 2, is a modification of the above.
Being of the Erlenraeyer type, cleaning is more easily
accomplished, while at the same time all the good fea-
tures of the above-described flasks are retained. The
one used in this laboratory holds about 100 cc. and
weighs approximately 30 grams, this tare affording
considerable strength while not affecting the accuracy
in weighing. In practical use the flask is proving about all that could be
desired in regard to safety and ease of manipulation, facility in cleaning
and small expense for replacement.

Buffalo Laboratory.

^•«»c^'

.i»|fc«f

r»W»i

Fig. 2.

A PROPOSED METHOD FOR THE ROUTINE VALUATION OF

DIASTASE PREPARATIONS,

By William A. Johnson.
Received February 3i, 1908.

In view of the rapidly increasing number of starch-digesting products
on the market, and the exaggerated claims which are made for some of
^ This Journal, 23, 338.

ROUTINE VALUATION OP DIASTASE PREPARATIONS. 799

them in the literature, through which they are advertised to physicians
and others interested, the necessity for a simple and approximately ac-
curate method of valuation becomes every day more apparent. This
necessity would not be so urgent if the manufacturers of these amylo-
lytic ferments had agreed among themselves on a uniform method by
which the enzymic activity may be determined, but this they have not
done and apparently are not likely to do.

In the last two years committees of the Council on Pharmacy and
Chemistry of the American Medical Association have undertaken to pass
on the validity of the claims of the manufacturers of medicinal substances
for the strength and purity of their products, among which products
the diastatic mixtures occupy an important place. In the course of
some investigations on the subject, which I have carried out under the
direction of one of these committees, I have made a number of observa-
tions which have a bearing on the question of the value of the methods
and these I now wish to put on record.

It may be said at the outset that nothing fundamentally new will be
offered here. Indeed this is not necessary in view of the classic re-
searches of Brown and Morris, Brown and Heron, O'Sullivaii, Roberts, Kjel-
dahl, Lintner, Effront and others, to say nothing of the older studies of
Dubrunfaut, Payen and Musculus.^

But most of the methods of measuring the ferment activity as brought
out in these long studies had for their main object the valuation of malt
employed in the brewing industries, and because of this fact they are not
usually available for the work we have in hand. In general, such valua-
tion may be made by noting the amount of the diastase preparation re-
quired to completely discharge the color of the iodine reaction in a given
weight of starch paste of definite strength, or by noting the amount
of sugar formed from an excess of starch, by the enzymes in some definite
time at a proper temperature. From a theoretical standpoint, methods
based on the latter determination would seem to have the advantage,
as sugar formation rather than starch disappearance is the end practically
required. But in some classes of preparations the enzyme is mixed
with so large a quantity of glucose or maltose that the determination of
more sugar formed would be found difficult in practice, especially where
. the ferment activity is low. This objection does not obtain in the ob-
servation on the disappearance of the starch-iodine reaction.
It is true that this starch-iodine method has been frequently condemned

* The general literature on the subject of the digestion of starch is of course vol-
uminous, and no attempt will be made to quote all of it. But reference may be made
to the convenient r^sum^s in the following works: v. Lippmann, "Chemie der Zucker-
anen," III Edition. Oppenheim, "Die Femiente imd ihre Wirkungen." Eflfront,
"Di^ Diastas^n,"

8oO WILLIAM A. JOHNSON.

because the end point indicated by the disappearance of the iodine reac-
tion is a point measuring the formation of a mixture, possibly, of some-
what complex dextrins with sugar, rather than of maltose itself. But
if it may be shown that the disappearance of the iodine reaction follows
always when a rather constant amount of maltose is formed, the method
may be made of value for practical purposes. The correspondence be-
tween sugar formation and starch disappearance obtains, however, only
under definite conditions, the redetermination of which was the first
object of my experiments.

Roberts was the first, apparently,* to work out a method for the com-
parison of diastatic activities through the aid of the iodine-starch reac-
tion* This general method was first used by F. C. Junck' and later by
J. M. Francis,* who supplied many working details, which contribute
much to the general accuracy and convenience of the process. In the
accurate comparison of diastases in this way the essential points to be
observed are these : i . A pure standard starch must be made and this
must be used in the form of a thin paste of constant value. 2. The ex-
periments must be so conducted as to show a sharp end reaction bet¥?een
the iodine and vanishing starch. 3. A standard time limit must be
adopted and rigorously adhered to. 4. The reaction must be carried
out at a constant temperature.

In my experiments, in agreement with many others who have invesr
tigated the subject, I have found potato starch the best material to use
as a standard. Most of the com starch on the market seems to be wholly
unfit for the purpose; in fact, different samples tested have given often
final results varying by 50 per cent, or more from each other. But potato
starch comes from the market in nearly pure form and by a simple treat-
ment may be made suitable for use. For my experiments I washed
500 grams repeatedly with distilled water, by decantation, then sudced
as dry as possible on a Buchner funnel. The mass was then spread on
glass plates and dried 3 hours in an air current at a temperature of 50®.
This made it dry enough to rub in a mortar, after which it was dried at
80® through four hours, which brought the moisture content down to
9.5 per cent., when the product was rubbed up again and bottled. It
is not advisable to try to dry beyond 90 per cent, of pure starch, as the
anhydrous starch absorbs moisture so quickly as to introduce inaccura-
cies in weighing. Prepared in this way, the microscopic examination
shows clean granules, free from fracture and free from foreign substanas.

It is hardly necessary to insist that a colorless end point is much more
accurately and easily observed than is the point where the blue starch-

* Proc. Roy, Soc, 32, 145. Cit. Maly's "Jahresber." for 1881, p. 290.
' Am, Jour. Pharm., June, 1883.

• Bulletin of Pharmacy, February, 1898.

ROUTINE VALUATION OF DIASTASE PREPARATIONS. 8oi

iodine reaction gives place to the reddish starch-dextrin reaction. I
have worked then, in every case, to the colorless end point. With a little
practice this may be uniformly noted by different observers working
with the same materials, yet many persons have failed to realize the
importance of this, as will be shown below.

On the question of the time limit in the digestions, there is greater
room for difference of opinion, and various intervals from five minutes
to one hour have been suggested by different workers. As the rapidity
of starch conversion is very accurately proportional to the amount of
enzyme present, it is in any case possible to reduce the time required
to secure the desired end reaction by starting with a larger weight of the
ferment, and the time finally selected as the standard or limit must de-
pend, therefore, on a practical acquaintance with a wide range of sub-
stances in which such tests are made. In carrying out a number of tests
in parallel, as is always the custom in such work, five minutes is an in-
conveniently short time to complete the various manipulations neces-
sary; on the other hand, an hour, or even half an hour, is a relatively
long time, which must be considered in part wasted if the same degree
of accuracy can be secured in a shorter period. Now, it has been found,
as a matter of fact, that no one of the digestive ferments on the market
is so strong as to convert over 300 times its weight of starch into reducing
sugar in ten minutes, while many of them have between one-tenth and
one one-hundredth of this activity. Suppose, therefore, that we start
with I g. of starch, the weights of enzyme preparation necessary to effect
conversion in this time would run from 3.33 up to 33.3 or 333 mgs.,
quantities which are reached by convenient dilutions. Ten minutes
seem to afford ample time to make the final color tests and I have there-
fore adopted this period for all this work.

In making such digestions, a temperature of 60° is often used, but as
we are dealing with products which in practice are to be used at the tem-
perature of the body, generally, it is preferable to take a temperature
of 40° as the standard, and this I have done.

Practical Details,

In working this method we need first a standard starch paste. This
is made by weighing out enough of the starch, prepared as above, to cor-
respond to 20 grams of pure anhydrous starch. Of 90 per cent, starch
we take, therefore, 22.22 grams. This is stirred up to make a uniform
cream with 100 cc. of water, and the mixture is then poured into 800 cc.
of boiling distilled water, in a flask. The boiling is continued ten min-
utes and then more water is added to make 1000 grams by weight. The
mixture is heated and shaken to distribute the starch imiformly. The
contents of the flask should be practically clear and free from all lumps.

802 WILUAM A. JOHNSON.

For each test quantities of 50 grams each are weighed into a series of 250
CO. flasks, clamped in a large water-bath kept at 40^.

The iodine test solution is made by dissolving 2 grams of iodine and
4 grams of potassium iodide in 250 cc. of water. Two cc. of this sohitioii
are then diluted with distilled water to make 1000 cc.

In making up the diastase solutions, the operator must be guided
by the results of a few preliminary experiments in each case. For
liquid malt extracts, for example, 10 cc. diluted to 100 cc. will be gen-
erally a proper strength, while in the examination of the dry prepara-
tions on the market, 200 to 500 milligrams dissolved or suspended in 100
cc. of distilled water will usually answer. These solutions are used in
this way. Small definite volumes of the dilution are added to the flasks
containing the starch paste in the thermostat, and with the least possi-
ble loss of time. The mixture is well shaken. The volumes added nmv
be as follows, but all diluted to that of the largest volume before mix-
ing: I cc., 2 cc., 3 cc., 4 cc., 5 cc., 6 cc. In about eight minutes tests
are begun by reniovmg volumes of 5 drops of each digesting mixture bj
a pipette and adding this to 5 cc. of the diluted iodine solution in a dear
white test tube standing over white paper. It is best to have a row of
these tubes mounted to receive the liquids to be tested. If at the end
of ten minutes drops from one of the flasks fail to give the iodine reac-
tion, we are ready for a second and more accumte test. Weigh out now
100 gmms of the paste into each of 6 bottles and, assuming that the end
point was found in the first test to be between 4 and 5 cc., add accurately
to the different flasks these volumes of the diastase solution: 8 cc., 84
cc., 8.8 cc, 9.2 cc, 9.6 cc, 10 cc. These volumes should stand ready
and all diluted to 10 cc. so that they may be poured into and shaken up
with the starch without delay. The tests with the iodine solution are
repeated as in the first trial and new limits are found between whicb
the real value must lie. For example, at the expiration of ten minutes
the paste to which 8.8 cc. of the diastase solution are added may show a
faint yellowish dextrin color, while that with the 9.2 cc. is colorless.
For all of our practical purposes it is not necessary to go beyond this.
In fact, we cannot carry our readings to a much greater degree of accuracy
because of the difficulty in distinguishing between the final shade from
the disappearing erythrodextrin and the achroodextrin, using these
terms in a geneml sense, rather than in the sense of assuming the actual
existence of these forms.

Much of the uncertainty in the determination of diastatic values, as
fotmd in the literature, doubtless comes from the failure to recognise
the importance of working to a colorless end point whenever this is pos-
sible. Roberts^ pointed this out clearly in his work, but his suggiesdons

^ Loc, cit.

ROUTINE VALUATION OF DIASTASE PREPARATIONS. 803

have been generally overlooked, because perhaps they were made in the
course of physiological, rather than in the usual technical, investiga-
tions. Most results for starch-converting power which we find adver-
tised are evidently obtained by working to a rose-red end point, as the
Pharmacopoeia allows in the case of the pancreatin test. For this reason
many of the strong products which I have examined appear to be some-
what weaker than claimed. This is shown by the results of the follow-
ing table, in which the statements of digesting power are given in both
ways. The diastase products tested are among the best known in the
market and are widely advertised. In giving the starch-converting power
of such preparations anhydrous starch does not appear to be taken as
the standard in any case. Average commercial starch contains about
15 per cent, of water, which should be allowed for in making fair com-
parisons.

Table of STARCH-CoNvERxmo Powers.

Value in anhy- Value in aohj- Value in commer-

drous starch, to drous starch, to cial starch, to

colorless end loss of blue loss of blue Value as

No. point. color. color. claimed.

1 100. o 1430 168.0 150.0

2 16.0 22.5 26.0 150.0
3(liq.) 0.0 0.0 0.0 7 +
4 113. 170.0 203.0 200.0
5(liq.) 3.6 5.2 6.1 8.0

6(liq.). . 6.0 8.6 10. 1

7(liq.) 6.0 8.6 10. 1

The values given in the last column are those claimed by the respec-
tive manufacturers, and are based apparently on digestion to the loss
of the blue color only. From the advertising literature it is seldom pos-
sible to discover the exact basis of the valuation. The values in the
second column, of the table, expressing the digestion carried to the achro-
mic point, are much more accurate than those in the following column,
where the disappearance of the blue color is recorded. It is not possi-
ble to judge of this as closely as desirable for a quantitative test.

Sugar Formation.

It is interesting now to note the amounts of sugar formed by these
preparations in equal times, and such determinations were made in this
manner. Having found the relative values of the preparations, amounts
of each sufficient to convert i gram of starch to the achromic point in
ten minutes were taken and mixed with 50 grams of starch paste. For
each substance five tests were made, the fiasks being kept in the thermo-
stat through periods of 10, 30, 60, 120 and i8o minutes, respectively.
At the end of the proper time a flask was removed, quickly brought to
the boiling-point to check further action, and the sugar then determined
as maltose by the FehUng titration. The results of these determina-

8o4 WILLIAM A. JOHNSON.

tions are shown in the following table, for some of the products described
above:

Milligrams op

Rbducing Sugar Calculated as Maltose

, from

I Gram of Anhy-

DROUS

Starch.

10 minutes.

30 minutes.

60 minutes.

120

minutes. 180 minutes.

613

788

866

611

822

933

1042

1094

adiq.)

Inert.

Too weak to measure.

622

783

850

855

630

788

845

858

633

777

863

872

635

777

866

In this table, as in the other, numbers 2 and 3 are preparations from
a fungus. Numbers i> 4 and 6 are pancreas preparations from three
different firms, 4a and 46 are products of one firm purchased some
months apart, while 6a and 66 are products of another firm bought at
different times. It is interesting to note that the different products
resemble each other very closely in their behavior, and that in all cases
the amount of actual sugar formed in ten minutes, that is when the achio-
mic point is reached in the iodine test, is about 60 per cent, of the theo-
retically possible complete amount, assuming that i gram of starch should
yield 1.055 grams of maltose. At the end of one hour the amounts
have gone up to over 80 per cent, of the possible, and beyond this there
is practically no change.

The behavior of the preparation from the ftmgus is interesting. While
its converting power for a short interval is like that of the others, the
conversion becomes relatively stronger with time, and evidently proceeds
beyond the production of maltose. The results of the calculations from
the titrations must be interpreted in this way. It must be remembered,
however, that the weight of the preparation required to convert starch
with rapidity is much greater than with the other ferments used.

The above results are entirely in accord with those which have been
found from time to time for the diastase from malt. A complete con-
version is not practically possible, although by precipitating out the
dextrins and adding fresh ferment it may be carried somewhat farther
than is here done. The values here obtained are easily comparable
with those of the iodine method.

The diastase preparations used in the above tests were all practically
free from sugar to begin with. But we have as commercial articles a
class of products made from malt by various extraction processes in
which there is always a large amount of sugar, from the malt, and fre-
quently added glucose. The examination of these mixtures, which are
usually in the form of thick sirups, presents greater difficulties. Some

ESTIMATION OF HUMUS. 805

of these articles I have now in hand and expect to report on them at a
hter time.

NOX.TBWBSTBRM UNXVBKSITY MBDICAL SCHOOL,

Chxcaoo, III.

THE SEPARATIOir OF CLAY Df THE ESTDCATION OF HUMUS.

Bt C. a. Moobrs AMD H. H. Hampton.
Received March 6, 1908.

That a serious error may be introduced into the estimation of humus
by the official method has been pointed out by a number of investiga-
tors. The chief cause of this error has been the weighing of clay with
the humus extract and the consequent reckoning of the combined water
in the clay as humus. To avoid this difficulty, Cameron and Breazeale*
filtered the extract through a Pasteur-Chamberland filter and deter-
mined the humus in the clear filtrate ; Peter and Averitt* have suggested
the use of a factor with which to make a correction for the loss in the
clayey residue; and a third, or evaporation method has been used by
the authors.' This paper presents a comparison of the results obtained
by the three methods.

In the filtration method the ammoniacal humus extract is filtered
through a Pasteur-Chamberland filter, which retains all the clay, and
the humus is estimated as usual, by evaporation, etc., of the clear fil-
trate. In getting the results here reported, a silver-plated, containing
tube was used on account of the ready solubility in ammonia of the cop-
per of the usual brass tube.

By the factor method Peter and Averitt make a deduction from the
total loss in weight of 10 per cent, of the residue left after the humus
has been burnt oft. In Table I are given the results obtained by making
both a 10 per cent, and a 14 per cent, deduction.

In the evaporation method the ammoniacal humus extract contain-
ing clay in suspension is evaporated to dryness over a steam bath, by
which means the clay is flocculated so that after extraction with 4 per
cent, ammonia it can be retained on a common filter paperi Two evapo-
rations and.filtrations are necessary as a rule in order to get a clear fil-
trate, in which the humus is determined as usual.

The percentages in the first column, obtained with clay present, are
not only liable to be irregular, as is shown under 636, but are undoubt-
edly too high even when the clay is allowed to settle out for several
weeks before taking out the aliquot portion for evaporation, as was done
for Nos. 602 and 666.

* This Journal, 26, 29-45.

* Ky. Sta. Bull., No. 126, pp. 63-126.

■ Term, Sta, Bull,, Vol. 19, No. 4, p. 50. ,^ ,

8o6

C. A. MOOERS AND H. H. HAMPTON.

Table I. — Per Cent, of Humus Estimated by the Different Methods in Repre-
sentative Tennessee Soils.

Soil
No.

Character of soil.

636 Fertile clay loam

3
4

602

666

584

Poor clay loam |

Silt loam | ^

Rich loam

«
.c

2.04

2-34
2.64

2.55

1.28

1.26

1. 19

1. 13

2.54

CI
TS hi)

5.23

6.75

7.97
8.78

1-95
2.06

2.01
1.68

5-68

Correction
method.

1.52
1.66
1.84
1.67

1.08
1.05

0.99
0.96

1-97

1.31
1.40

1.32
1 .01
0.97

0.91
0.89

1.74

8
O

%
0.64
0.65
0.77
0.59
0.76

u
9
JO
u

V
V

'O be

« B

fi(
0.22
0.18
0.21
0.24

0.23

S
u
9

u
V

e
3 .

• B

1.36 0.2S

1-33 0.19

1.38 0.23

1.26 0.33

0.29
o.iS

0.53 0.12

1.08
I. II

0.82 0.09
0.88 0.25

577 Rich loam.

3.07 7.91 2.28 1.96

1.67
1.77

0.49
0.65

1.73 0-50
1.72 0.69

The correction method undoubtedly gives better results than the pie-
ceding, but the duplications of Nos. 636 are not concordant, and there
is a question as to the proper factor. For these soils a 14 per cent, ratbei
than a 10 per cent, correction gives the most harmonious results.

The filtration method proved highly unsatisfactory, chiefly on account
of the humus absorbed by the filter. No simple way to avoid this de-
fect was found. The first three results on No. 636 were obtained from
successive filtrations of the same solution. These filtrations were made
after the first runnings had been discarded. Each of the after parts
that were analyzed represented about one-third of the remaining liquid.
The fact that by the evaporation method, in which perfectly clear solu-
tions were obtained, the results were, in round numbers, from 40 to 100
per cent, greater than those gotten by the use of the filter, is conduave
evidence of the inadequacy of the filtration method.

The evaporation method yielded at least fairly concordant results,
but as they were obtained in the early part of our work they are proba-
bly not as uniform as the method will permit under close attention to
details.

In Table II are given the percentages of humus by the evaporation
method, and of total nitrogen by the Kjeldahl method, foimd in adja-
cent plots of an experimental field at the Tennessee Experiment Statkn
farm. Lime had been applied to one-half of each of the plots at the
rate of 1800 pounds per acre three years before the samples were taken
for analysis. The results indicate that by the aid of this simple modi-
fication small changes in the humus-content of a soil can be measured.

RESEARCHES ON QUINAZOLONES. 807

Tablb II. — ^Epprct of Liming on the Humus and Nitrogen Contents op Soils '

PROM Adjacent Plots.

Limed half. Unlimed half.

Plot Humua. NitroD^en. Humus. I^itrogen.

No. Fertilized with Per cent. Per cent. Per cent. Per cent.

F4 Mineral Fertilizer 1.28 0.119 1.32 0.119

F5 '* '* 1.38 0.121 1.42 0.129

F6 Farmjrard manure 1.47 0.131 1.47 0.136

F8 Mineral fertilizer i . 37 o. 126 i . 41 o. 126

G4 " " 1.32 0.122 1.37 0.119

G5 " ** 1.33 0.118 1.34 0.125

06 Farmyard manure and mineral fer-
tilizer 1.39 0.126 1.42 0.131

G8 Mineral fertilizer 1.04 o.ioo i.io 0.108

Average 1.32 0.120 1.36 0,124

TbNNBSSBB AORICULtURAL EXPBIIIMBNT STATION,

Knoxvillb, Tbnn., March 4, 1908.

(Contributions from thb HAVEiiBYBR I/Aboratories of Colubabia University*

No. 151.)

RESEARCHES ON QXJINAZOLONES (TWENTIETH PAPER) ON CER-
TAIN 7-NITRO-2-METHYL-4-QUINAZOLONES FROM

4-NITR0ACETANTHRANIL.>

By Marston Taylor Boobrt and William Klabbr.
Received Pcbuary 38, 1906.

Bogert and Steiner* and Bogert and Seil' have already reported on the
synthesis of 7-nitro-2-melhyl-4-quinazolones from 4-nitroacetanthranil
and various primary amines by the Anschutz, Schmidt and Greiffenberg*
reaction. The present paper records the continuation and extension of
this work.

The reacion involved is a simple one, and may be conveniently ex-
pressed as follows:

/N.COCH, /NH.COCH,

0,N.C,H3< I + RNH, = 0,N .C,H,<( -

^CO ^CO.NHR

/N : C.CH,
0,N.C,H,/ I +H,0.

^CO.NR

In one or two cases we isolated the intermediate amide.

The primary monamines used were ammonia, methyl-, n-propyl-,
benzyl- and j9-naphthylamines, aniline and /^-anisidine. All of these

* Read at the General Meeting of the American Chemical Society, December
28, 1906.

* This Journal, 27, 1327 (1905)-
»/W(i., 29, 517 (1907).

* Ber., 35, 3480 (1902).

8o8 MARSTON TAYU)R BOGERT AND WILLIAM KLABER.

condensed smoothly with the acetanthranil, giving good yields and dean
products.

It was also found possible to condense the anthranil with amino ni-
triles and amino acid esters, but not with the free amino add or its salts.
Glycine ester and nitrile and anthranilic ester and nitrile were used in
the experiments, and the corresponding quinazolones obtained. By
the hydrolysis of the quinazolone nitriles, or by the action of ammonia
upon the quinazolone esters, the corresponding amides were prepared.
And from the latter, in turn, the nitriles were regenerated by the action
of acetic anhydride.

With hydrazine, both the 3-aminoquinazolone and the corresponding
diquinazolonyl were obtained.* From the amino compound, acetyl and
di-acetyl derivatives were prepared, as well as the phenylhydrazone of
the monacetyl compound. By the Biilow condensation' with ethyl
diaceto succinate, the pyrrole compound was produced.

7-Nitro-4-quinazolone (7-nitro-4-hydroxyquinazoline) was prepared
from 4-nitro-2-aminobenzoic acid and formamide,'and 7-amino-2-methyl-
4-quinazolone by reducing the corresponding nitro compound.

Experimental,

4-Nitroacetanthranil was prepared, as described by Bogert and Steiner/
by nitrating a-toluidine in' presence of excess of concentrated sulphuric
acid, acetylating the 4-nitro-2-toluidine, oxidizing the nitroacettohii-
dide to the nitro acetaminobenzoic acid, and converting the latter into
the corresponding nitroacetanthranil by the action of acetic anhydride.

7'NitrO'4-quinazolone (j-Niiro-4.-hydroxyquinazoline) ,

.N = CH _^ N==CH

(7) 0,N.C,H,< I TI 0,N.C,H,< | .— 4.Nitro-2-ami-

\CO— NH x:(OH)=N

nobenzoic acid was gently fused with excess of formamide, giving a dark
red solution. After removal of the excess of formamide, the residue was
purified by treating with boneblack and crystallizing from alcohol.
Long, slender, glistening yellow needles m. p. 276° (cor.).

Nitrogen found, 22.21. Calculated for CgHgOgN,: N, 22.0.

j-NitrO'2-meihyl-4'quinazolone{j-Niiro-2-me{kyl-4rhydroxyquif^

.N = CCH, ^ N=— C— CH,

0,N.C,H,< I irO,N.CeH3< I , has been re-

\CO— NH \C(OH)=N

ported previously by Bogert and Steiner* and Bogert and Seil.* By care-

» Compare Bogert and Seil, This Journal 28, 884(1906); Bogert and Cook, Ihii.,
28 1453 (1906).

= Ber., 35, 4312 (1902); 39, 2621, 3372 (1906).

' Niementowski, J. prakt, Chem. [2], 51, 564 (1895).

* Loc. cit.

RKSEARCHES ON QUINAZOLONES. 809

ful purification, we have succeeded in raising the melting-point from
275-7° (as given by them) to 287-90® (cor.). As thus purified, the sub-
stance forms silky needles of a pale greenish cast. It was prepared both from
the anthranil and ammonia, and by the action of heat upon the ammonitmi
salt of 4-nitro-2-acetaminobenzoic add. It dissolves in neutral sodium
carbonate solutions with evolution of carbon dioxide, but can be reprecipi-
tated from this solution by saturating with carbon dioxide, as it is insolu-
ble in the acid carbonate. It is also soluble in concentrated aqueous
ammonia, but reprecipitates when the solution is boiled. It crystal-
lizes imchanged from fused ammonium acetate. Stannous chloride re-
duces it to the 7-amino compound (described beyond).

Oxidation of y-NitrO'2-meihyl-4rquinazolone, — Seven grams of the quin-
azolone were dissolved in 300 cc. dilute sulphuric add (one of concen-
trated add to five of water) containing 14 grams chromic anhydride,
and the solution boiled for about 60 hours. On cooling, small, color-
less crystals separated. Dissolved in sodium carbonate solution, re-
precipitated by acetic add, and recrystallized from alcohol, colorless
scales were obtained, melting at about 327® (imcor.), and dissolving
with a red color in caustic or carbonated alkali solutions. Not enough
of the substance was obtained for an analysis.

Potassium Salt of y-Nitro-2'methyl-4rhydroxyquinazoline, — The quinazo-
line was dissolved in dilute potassium hydroxide solution and this solu-
tion then saturated with solid potassium carbonate. The yellow pre-
dpitate was filtered out, dissolved in absolute alcohol, and the filtered
solution evaporated to dryness, leaving a pale yellow crystalline mass
easily soluble in water or alcohol. It crystallizes from alcohol with alco-
hol of crystallization as a pale yellow solid, the color deepening to an orange
on driving out the alcohol. It was analyzed by titration with fifth-nor-
mal sulphuric acid, using phenolphthalein as indicator.

Potassium found, 16.21. Calculated for CgH^OsNgK: K, 16. 11.

Silver Salt. — ^The quinazoline was dissolved in concentrated aqueous
ammonia, the solution heated to boiling, and ammoniacal silver nitrate
solution added gradually as long as it gave a predpitate. The precipi-
tate was voluminous and curdy, and of a faint yellow color. It was
washed thoroughly with dilute ammonia water and dried. It is decom-
posed quantitatively when heated with a salt solution. Analyzed in
this way, it gave the following result :

Silver found, 34. 35. Calculated for CgHoOgNjAg: Ag, 34.60.

/N : C.CH,
7'NiirO'2^'dimelhyl-4-qutnazolone, OjN.CjHj^ | . — Bogertand

^CO.NCH,

Steiner* prepared this substance by the action of methylamine upon the

* Loc. cit.

8lO MARSTON TAYLOR BOGBRT AND WILUAM KLABER.

4-nitroacetanthranil. We have also obtained it by heating the methyl-
amine salt of 4-nitroacetanthranilic add at 190-200°, but the yield by
this latter method is poor and the product quite impure. It is unchanged
by fusion with ammonium acetate or by heating with ammonium for-
mate for three hours in a sealed tube at 230°. Bogert and Steiner de-
scribe the compoimd as forming light yellowish green crystals, m. p.
144-5° (cor.). We have succeeded in getting it nearly colorless and of
a m. p. 151-2° (cor.).

7-Nitro-2-inethyl'3-fhpropyl-4-qutnazolone, from 4-nitroacetanthranil and
n-propylamine, crystallizes from dilute alcohol in colorless needles, m,
p. 140° (cor.). It is slightly soluble in water and soluble in alcohol.

Nitrogen fotmd, 16.9. Calculated for C^K^fi^N^: N, 17.0.

7-NitYo-2'meihyl'3'phenyl'4-quinazolone, from the nitroacetanthranil
and aniline, crystallizes from alcohol in nearly colorless, diamond-shaped
plates, m. p. 209° (cor.).

Nitrogen found, 14.82. Calculated for CigHuOaNj: N, 14.94.

7'Nitro-2-methyl-j-benzyl-4-qui7iazolone, from the nitroacetanthranil and
benzylamine, crystallizes from alcohol in coarse, yellowish, cubical forms,
m. p. 131-2° (cor.).

Nitrogen found, 14. 27. Calculated for CieH^OaNj: N, 14.23.

Its hydrochloride forms pale yellowish crystals, m. p. 229-30° (cor.),
which lose their HCl when boiled with alcohol.

Nitrogen found,* 1 2 . 92. Calculated for Ci8Hi40sN8Cl : N, 12. 67.

y'Nttro-2-methyl-3-p-anisyl-4-quinazolonef from the nitroacetanthranil
and />-anisidine, crystallizes from alcohol in glistening scales, of a faintly
yellowish cast, m. p. 228° (cor.).

Nitrogen foimd, 13.49. Calculated for Cif^H^iO^N^: N, 13.50.

y-NitrO'2'meihyl-3'^-naphihyl'4-quinazolone, frbm the nitroacetanthranil
and j9-naphthylamine. Fine, colorless needles (from alcohol), m. p.
218-9° (cor.).

Nitrogen found, 12.64. Calculated for Cj^jjOgN,: N, 12.68.

J- A mino-2-meihyl-4-quinazolone (j-amino-2-nieihyl-4-hydroxyquinazO'
line), — ^The 7-nitro compound was reduced with stannous chloride and
hydrochloric acid. The double chloride of tin and the amino quinazoline
which separated was boiled with dilute sodiimi carbonate solution and
filtered hot. On cooling, the 7-aminoquinazoline separated from the
filtrate in long, silky, colorless needles, m. p. 311° (cor.), identical with
the quinazoline obtained by the action of potassitun hydroxide solution
upon 7-acetamino-2-methyl-4-quinazolone (which will be described in
another paper). It is soluble in hot water, alcohol or caustic alkali solu-
tions, but not in cold sodium carbonate solutions. Crystals obtained
from water or alcohol carry approxim^-tely half a molecule of the soh'ent.

Nitrogen found, 24.22. Calculated for C^HgON,: N, 24.0.

RESEARCHES ON QUINAZOIX)NES. 8ll

7 -Nitro -2 - methyl -3 - amino - 4 - quinazolone. — 4-Nit roacetanthranil was
added gradually to an excess of 50 per cent, hydrazine hydrate solution
(aqueous). Considerable heat was developed and the reaction was com-
pleted by further final warming. When cold, the insoluble product
was extracted with a small amount of alcohol, to remove nitroanthranilic
add and small amounts of impurities, and the residue was then purified
by crystallization from alcohol. Shining, pale yellow needles, m. p.
223® (cor.). With acetic anhydride it gives both mono- and diacetyl
derivatives.

Nitrogen foimd, 25 . 73 and 25 . 74. Calculated for CbH^OjN^: N, 25 .45.

The yield of pure aminoquinazolone was generally about 50 per cent,
of the theory.

y-NitrO'2-methyl-j-amino-4-quinazolone and phenylhydrazine, — The amino
quinazolone was mixed with excess of phenylhydrazine and the
mixture heated just to boiling. As soon as ebullition began, the fiame
was removed, as considerable heat is developed by the reaction. Am-
monia was evolved. When the action moderated, the heating was re-
newed for a short time, the mass allowed to cool somewhat, excess of
acetic acid added and the mixture warmed up again. This second phase
of the reaction also proceeds with considerable rise of temperature, due
to the combination of the acetic acid with the excess of phenylhydrazine
and the formation of acetylphenylhydrazide, and the flame should be
removed as soon as the reaction threatens to become violent. The dark
colored product is heated for a short time further, concentrated to small
volume and allowed to crystallize. The mixture is diluted largely with
cold water, the insoluble material crystallized from alcohol and washed
with ether. From very dilute potassiiun hydroxide solution it separa-
ted in fine needles, which were recrystallized from alcohol, giving clus-
ters of colorless, feathery needles, m. p. 230° (uncor.), insoluble in ether,
soluble in hot water, easily soluble in alcohol.

Found : C, 48 . 62 and 48 . 82 ; H, 5.9 and 5.52; N, 25 . 60 and 25 . 54.

These figures are far removed from those calculated for the phenylhy-
drazinoquinazolone, the phenylhydrazinophenylhydrazone, or the oso-
tetrazole which might be formed from the latter by loss of a molecule
of aniline. The compound has not been identified.

A similar product was obtained by heating the amino quinazolone
with two molecules of phenylhydrazine in cumene solution, water and
ammonia being evolved. When the heating was conducted in nitroben-
zene or alcoholic acetic acid solution, but little action took place.

7-Niiro-2-m€thyl-3-acetamino-4-quinazolone, — ^The above quinazolone,
when heated with acetic anhydride, gave a clear solution. This solu-
tion was concentrated to crystals, which were washed with carbon tetra-
chloride and recrystallized from benzene. Short, colorless prisms, m.

8l2 MARSTON TAYLOR BOGERT AND WILLIAM KLABER.

P« 233° (cor.), soluble in alcohol, insoluble in carbon tetrachloride, and
but sparingly soluble in cold benzene.

Nitrogen found, 21 .49. Calculated for C11H10O4N4: N, 21 .37.

Phenylhydrazone. — The acetaminoquinazolone and phenylhydrazine
were heated just to boiling and the flame removed. If large amounts are
used, the reaction proceeds with considerable violence and frothing.
On cooling, the mass solidified. It was washed with ether and crystal-
lized from alcohol. Beautiful, glassy prims, m. p. 315° (cor.).

Found: C, 57.80; H, 4.90; N, 23.4. Calculated for CiyH^OjN,: C,
57.95; H, 4.54; N, 23.86.

7-Nitro-2'methyl-S'diacetainino-4'quin<izolone,

Q.N.C^H,^ I . — This was obtained from either the amino

\CO.N(COCH,),

or acetaminoquinazolone by the further action of acetic anhydride. The
diacetyl compound is much more soluble in benzene than the monacetyl
and the two can be separated in this way. The diacetyl derivative was
separated from the benzene mother liquors by careful addition of gaso-
line. It crystallizes in pale yellow, glassy plates, m. p. 132° (cor.),
soluble in alcohol, benzene, or acetic anhydride. Oxidation with neu-
tral permanganate yielded only 4-nitroanthranilic acid and its acetyl
derivative.

Nitrogen found, 18.79 and 18.89. Calculated for CuH^OjN/. N,

18.42.

When the diacetyl derivative was boiled with phenylhydrazine in alco-
holic acetic acid solution, the only change noted was the formation of
the monacetyl compoimd. No hydrazone was observed. A boiling
alcoholic solution of aniline had the same effect in changing the diacetyl
to the monacetyl derivative.

'^^'/''DtnitrO'2,2''dtmethyl'4,4'-diketoUtrahydro-3y3'-diquinazolylf

.N : CCH,CH,C : Nv
O.N.CeH,< I I >CeH,.NO,.— One gram of sopercent.

^CO.N N.CCK

aqueous hydrazine hydrate solution was added to 6.4 grams 4-nitro-
acetanthranil and the mixture heated for a short time on the sand-
bath. When cold, the solid product was extracted with dilute acetic
acid, to remove aminoquinazolone, and the residue crystallized from a
mixture of ethyl and isoamyl acetates. Small, yellow, granular crys-
tals, m. p. 337 . 5° (cor.). The yield was poor.

Nitrogen found, 20.63. Calculated for CjsHuOeNe: N, 20.58.

The substance is soluble in ethyl or isoamyl acetates, but is insoluble
in water, alcohol, and in the majority of the ordinary organic solvents.
It dissolves in acetic anhydride and, on cooling, white crystals separate,

RESEARCHES ON QUINAZOLONES. 813

m. p. 227** (iincor.). probably similar to the acetic anhydride addition-
product described by Bogert and SeiP for the corresponding 5,5'-dinitro-
diquinazolyl.

4-Nitroaceianthranil and Guanidine. — ^When these substances were
heated together in aqueous solution, in equal molecules, the guanidine
salt of 4-nitroacetanthranilic acid was obtained. It fonns coarse, lentil-
shaped crystals, m. p. 247° (cor.).

In one case, where two molecules of the anthranil were used to one of
the guanidine, a small amount of a yellow substance was isolated, which
melted sharply at 253®, resolidified and did not re-melt below 300°. It
was not obtained in sufficient amount to identify.

Ethyl 7-NitrO'2-meihyl-4-quinazolonyl'3-aceiate,

.N : CCH,
OjN.CeHj^ I .—Attempts to condense 4-nitroacet-

^CO.N.CH^COAHfi

anthranil with free glycine or with its sodium salt failed. Heating with
aqueous solutions caused only hydrolysis of the anthmnil, while direct
dry heating resulted in decomposition.

Three and one-half grams glycine ester hydrochloride were dissolved
in the smallest possible amount of water and the hydrochloric acid re-
moved by the addition of the moist silver oxide from 4 . 66 grams of silver
nitrate. Five grams of the anthranil were then added to the mixture,
and the temperature raised rapidly to the boiling-point. The close of
the reaction was indicated by the clearing of the solution, followed by
the appearance of turbidity in the supernatant liquid. The quinazo-
lone ester was extracted with ether and purified by crystallization from
the same solvent. Asbestos-like sheaves of colorless crystals, m. p.
139-40'' (cor.).

Nitrogen found, 14.38. Calculated for CuHiaOgNg: N, 14.43.

After the ether extraction, the residual solution was found to contain
some nitroacetanthranilic acid.

7'NiirO'2'in€thyl'4-quinazolonyl-j-acetamid€y

.N : CCH,
OjN.C.H,^ I . — The above ethyl ester was boiled with

\CO.N.CH,CONH,

excess of concentrated aqueous ammonia until a clear solution was ob-
tained. On cooling, colorless, very fine, silky needles separated. Re-
crystallized from alcohol, they showed a m. p. of 275° (cor.).

Nitrogen found, 21 .45. Calculated for CiiHio04N4: N, 21 .37.

When boiled with acetic anhydride, it gave the nitrile described below.

'^-Niiro-2'ineihyl-4-quinazolonyl-j-aceionitrile,
* Tms Journal, 28, 892 (1906).

8 14 MARSTON TAYI<OR BOGERT AND WII.UAM BXABER.

.N : CCH,
OjN.CjH,^ I . — 4-Nitroacetanthranil and anhydrous aceto-

^CO.N.CH.CN

nitrile were found to be without action upon each other at the boiling-
point of the nitrik (8 1**) or in a sealed tube at i6o^

4.17 grams potassium hydroxide were dissolved in a small amount of
water, the solution cooled in an ice pack, and 5 . 78 grams aminoacetoni-
trile sulphate (50 per cent, excess) added slowly with stirring, care being
taken to prevent any considerable rise of temperature. Five grams
nitroacetanthmnil were then quickly stirred in and the mixture heated
rapidly to boiling. After boiling for about five minutes, the dark red
solution was allowed to cool, cold water &dded, the mixture thoroughly
stirred, and the precipitate filtered out. This was treated with bone-
black in acetone solution, and repeatedly crystallized from the same
solvent. Colorless, shining scales, m. p. 207-8° (cor.), insoluble in water
or petroleum distillates, soluble in alcohol, acetone or dilute acetic add.

Nitrogen found, 23 . 10. Calculated for CjjHgOjN^: N, 22 . 90.

4rNitro-2'aceiaminohippuronitriley
(4)0,N(2)CH3CONH.C,H3.CONHCH3CN.— This intermediate amide was
found in the acetone mother liquors from the above quinazolone. It is
very much more soluble in acetone than the quinazolone and can be sepa-
rated by this property. It cr^^stallizes in colorless prisms, melting with
effervescence at 194° (cor.) (probably changing to the quinazolone),
and is readily changed to the quinazolone by heating with very dilute
potassium hydroxide solution.

Nitrogen foimd, 21 . 53. Calculated for CiiH,o04N4: N, 21 . 37.

When ethyl /?-aminocrotonate and 4-nitroacetanthranil were heated
together, only 7-nitro-2-methyl-4-quinazolone (7-nitro-2-methyl-4-hy-
droxyquinazoline) was obtained. This is, of course, due to the fact that
the aminocrotonic ester loses ammonia very readily, and this ammonia
then condenses with the anthranil to the simple quinazolone.

Methyl 7'Nitro-2'm€thyl'4-quinazolonyl-3-o-benzoate,
.N : CCH,
0,N.C-Hs<; I . — Neither anthranilic acid nor its

\C0. N. CeH,. COOCH.Co)

sodium salt could be made to condense with 4-nitroacetanthranil.

The anthranil was heated for a short time with excess of methyl anthra-
nilate and the mixture then allowed to cool. The resultant glassy solid
was extracted with alcohol, the alcoholic solution treated with bone-
black, and the precipitate which separated on cooling treated with cold
dilute potassium hydroxide solution, to remove nitroacetanthranilic
acid. The residue, crystallized from alcohol, gave canary-yellow, granu-
lar crystals, m. p. 175° (cor.), quite soluble in alcohol, but insoluble in
water.

RESEARCHES ON QUINAZOLONES. 815

Nitrogen found, 12.52. Calculated for CiyHijOgN,: N, 12.38.
7-NitrO'2'meihyl-4'quinazolonyl-3'0'henzamide^

/N : CCH,
OjN.CjH,^ I . — Anthranilic nitrile was just cov-

^CO.N.CeH,.CONH,(o)

ered with water and the mixture heated to boiling. The nitrile remained
largely undissolved as a yellow oil. 4-Nitroacetanthranil was then added.
The oily nitrile turned dark and finally solidified to a yellow mass, a pre-
cipitate also appearing in the supernatant liquid. Potassitun carbonate
was added to the cold solution, to remove nitroacetanthranilic acid.
The residue was then extracted with ethyl acetate, to eliminate certain
easily soluble impurities, and the undissolved portion crystallized from
nitrobenzene, the excess of solvent being washed out with alcohol.
Straw-colored needles, m. p. 320-1° (cor.), soluble in nitrobenzene or
acetic anhydride but only moderately soluble in alcohol.

Nitrogen found, 17.34, 17 28 and 17.38. Calculated for CieHi,04N4:

N, 17.3-

The product might have either of the following structures:

/NH.COCH, .N : CCH,

0,N.C,H,<; or 0,N.CeH3< |

^CONHCeH.CN ^CO.N.CeH^.CONH,

(I) (11)

Boiled with acetic anhydride, it gave the benzonitrile quinazolone de-
scribed below, from which it can also be produced by five to six hours*
boiling with alcohol, the conversion being a quantitative one. This de-
hydration could occur quite readily with either structure, but the easy
re-hydration of the benzonitrile quinazolone to the same product would
seem to us to exclude formula (I). Further, the compound does not
melt and re-solidify below the m. p. of the quinazolone, as is usually the
case with such intermediate amides as (I), nor does it change to the quin-
azolone on long boiling in nitrobenzene as (I) might be expected to do
in line with the synthesis of the benzonitrile quinazolone described be-
low.

Attempts to bring about a further condensation, thus:

yN : CCH, xN : CCH,

0,N.C.H,<^ I — >- 0,N.CeH,< I were unsuccessful.

^C.N.CeH,.CO \C.N.C,H,

O H;^ N N CO

7-Niiro-2-methyl-4-qutnazolonyl-3-o-benzonitrile,

/N : CCH,
0,N.C,H^ I . — Pure benzonitrile was found to be with-

X;O.N.C,H«CN(o)

out action upon 4-nitroacetanthraniL

8l6 WILUAM MCPHERSON AND WILBUR L. DUBOIS.

By fusing anthranilic nitrile with the nitroacetanthranil, and extracting
the product with ether, heavy red needles were obtained, m. p. 225°
(uncor.).

In a second experiment, the condensation was carried out in boiHng
nitrobenzene solution. The reddish crystals were treated with bone-
black in benzene solution, and recrystallized from alcohol. Slender,
glistening, yellowish needles, m. p. 234° (cor.) ; soluble in alcohol, ben-
zene, ether, nitrobenzene or acetic anhydride.

Nitrogen found, 18.23 ^^^ 18.45. Calculated for CieHjoOjN^: N.
18.30.

Long boiling with 95 per cent, alcohol changed this compound com-
pletely to the amide mentioned above, from which it could, as stated,
be regenerated by dehydration with acetic anhydride. As anthranilic
nitrile itself is rather resistant to hydrolysis, the ease with which this
quinazolone nitrile changes to the amide is rather surprising.

7'Nitro-2'methyl-4-quina2olonyl-3'(^2,ydimeihyl'3y4'dicar^^

/N : CCH, yC(CH,) : C.COOCjHj
0,N.C,H,^ I yN^ I .—Bttlow has shown* that

CO.N— ^ ^C(CH,) : C.COOC,H,

hydrazines of the type RNHNHj or RjNNHj condense with diacetosuc-
cinic ester to derivatives of pyrrole.

Three grams of the 3-aminoquinazolone and four of ethyl diacttosuc-
cinate were dissolved in 40 cc. of giacial acetic acid, the solution boiled
for three hours, and then concentrated. A small amount of precipitate
separated on cooling. This was removed, and alcohol added to the mother
liquor. A white precipitate resulted, which was purified by repeated
crystallization from alcohol, until it appeared in colorless, shining, minute
scales, of a constant m. p. of 171° (cor.).

Found: C, 56.45; H, 4.8; N, 12.86. Calculated for CjiHaOyN^.C,
57.0; H, 4.97; N, 12.67.

HAVBMBTBR I«AB0RAT0RIB8, COLUICBIA UNIVBR8XTY,

Nbw York, Pebnuii^ 35, 1908.

[Contribution prom the Chemical Laboratory ok the Ohio State UNivERsmr]

ON THE ACTION OF a-BENZOYLPHENYLHYDRAZINE ON THE

HALOGEN DERIVATIVES OF QUINONES,

Bt William McPhbrson ^nd Wilbur L. Dubois.
Received March 36, 1908.

The action of phenylhydrazine on quinones was first investigated by
Zincke.^ He showed that phenylhydrazine, as well as its salts, acted

* Loc. cit.

» Ber., 16, 1563.

ACTION Ol' a-BENZOYl,PHENYI,HYDRAZINE. 817

energetically on the quinones of the benzene series with evolution of
gas, but did not isolate the compounds formed. More recently it was
shown by one of us, working under the guidance of Dr. J. U. Nef , that if
the a-hydrogen in phenylhydrazine is displaced by an acyl group the re-
sulting compounds condense smoothly with benzo- and naphthoquinones
in a perfectly normal manner, forming hydrazones.' Thus, with benzo-
quinone, the action is expressed by the following equation:

C,HX + CeH,NAcNH, - C^HX + H,0.

^O ^N— NAcCeHj

The action of naphthylhydrazine on both benzo- and naphthoqui-
nones has also been studied by McPherson and Gore,^ and the results of
these investigations have had an important bearing in determining the
constitution of the hydroxyazo compounds.

Zincke also studied the action of phenylhydrazine on the halogen de-
rivatives of the benzoquinones' and foimd that decomposition occurred,
accompanied by evolution of nitrogen. An investigation was begun
in this laboratory to determine the action of acylated phenylhydrazines
on the halogen derivatives of the quinones, and a preliminary paper
was published.* Utifortvmately the notes on all the work done were de-
stroyed by the burning of the laboratory, and it has been possible only
recently to resume again the investigation. Since a number of other in-
vestigators are working in this field, the following report is published
with the hope that I may be allowed to reserve this phase of the subject
for investigation. The action of the a-acylated phenylhydrazines on the
more recently discovered orthobenzoquinone, as well as its halogen de-
rivatives, is also being studied with the hope that some light may be
thrown on the constitution of the hydroxyazo compounds-
It has been fotmd that the a-acylated phenylhydrazines act on the
halogen derivatives of the benzoquinones in three different ways, as
follows:

I. A regular condensation may take place with the formation of hydra-
zones. One would naturally expect this reaction with the monohalo-
genated quinones, since it is with thse derivatives that hydroxylamine
most readily condenses to form oximes. So far as the experiments have
been carried out, it is only with these derivatives that hydrazones are
formed. With monochlorbenzoquinone the reaction is expressed as
follows:

* B€r., 28, 2414; Am, Chem. /., 32, 364.

• Am, Chem. /., 25, 485.
» Ber., 16, 1563.

^ This Journal, 22, 141.

8i8

WILLIAM MCPHERSON AND WILBUR L. DUBOIS.

C=0

Hc/Ncci

ICH

+ C^HgNAcNH,

C=0

Hc/N:ci

+ H,0.

C = N — N AcCeHj
2. One of the hydrogen atoms in the NH, group of the hydrazine,
together with one of the atoms of the quinone hydrogen, may be re-
moved by the oxidizing action of a second molecule of the quinone. Thus
with trichlorquinone the action is as follows:
C==0 C=0

cic/\:ci ClC^^CCl

CIC

c=o

+ QHsNAcNH, + O =
CH ClC

c=o

C— NH— NAcC^Hj

+ HA

cicL ^cci

c=o

C— NH— NAcC,H.

This reaction is in accord with the fact that hydroxylamine, which
readily forms oximes with the monohalogenated quinones, does not do
so with the trihalogen derivatives.

3. One of the hydrogen atoms in the NH3 group of the hydrazine may
combine with a chlorine atom of the quinone, splitting off hydrochloric
acid. Thus, with tetrachlorquinone, the reaction is expressed by the
following equation :

C=0 C=0

cic/Vci ClC^^CCl

+ C.H5NACNH, = + HCl.

CIC

C--0

Experimental Part,

Action of a-Benzoylphenylhydrazine on Monochlorbenzoquinone, Mono-
chlorquinonebenzoylpkenylhydrazone,

/CCl = CHv
O = C<f y>C = N— NC^HgOCeHg.— The chlorquinone was pre-

^CH = CH^

pared from chlorhydroquinone by Clark's method.* Twenty -five grams of
the chlorhydroquinone gave 20 grams of the chlorquinone when treated
with the reagents in the following proportion : Chlorhydroquinone 25
g., manganese dioxide 36 g., water 125 cc, sulphuric acid 25 cc

Five and four-tenths grams of the chlorquinone and 10 grams of a-ben-
zoylphenylhydrazine sulphate were dissolved by heating in separate
portions of a mixture of 200 cc. of alcohol and 50 cc. of water. The solu-
tions were cooled to room temperature and then mixed and heated slowly
on a water-bath to from 50® to 60®, the mixture being stirred occasion-

* Am. Chem. J., 14, 555.

ACTION OF a-BENZOYLPHENYLHYDRAZINE. 819

ally. As the temperature rose a yellow solid separated in the form of
short, lath-shaped crystals. The mixture was set aside until it had ac-
quired the room temperature and the solid was filtered off. A yield of
6 grams was obtained. The product was nearly pure, and after two
crystallizations from benzene-ligroin melted sharply at 172.5° and gave
the following results on analysis :

Calculated for C,^H,jNaOaCl: C, 67.72; H, 3.89; N, 8.34; CI, 10.53.
Found: C, 67.70; H, 4.15; N, 8.28; Cl, 10.02.

Monochlorquinonebenzoylphenylhydrazone is a yellow, crystalline com-
pound very difficultly soluble in alcohol and ether. It readily dissolves in
chloroform and benzene. From benzene it separates in rhomboids with
faces almost equal in length, while from chloroform it separates in rhom-
boids having one axis greatly elongated. It is almost insoluble in ligroin
and can best be purified by adding ligroin to a concentrated benzene
solution. Phenylhydrazine acts upon it with evolution of nitrogen,
although the action is not so energetic as with the corresponding non-
halogenated benzoquinonehydrazone.

Action of Saponifying Agents on MonoMorquinonebenzoylphehylhydra-
zone, — ^The hydrazone was saponified by adding sulphuric add and stir-
ring until complete solution took place. The solution was then added
slowly to water (not the reverse) when a yellow compound was formed, which
collected on stirring. This was filtered, dried and purified by dissolving
in ligroin. It separated from the ligroin in yellow, lath-shaped crystals,
melting at 86*^. The compound was soluble in sodium hydroxide and
corresponded in properties to an hydroxyazo compound.

Constitution of Monochlorhenzoquinonephenylhydrazone, — In accordance
with the method of preparation, the hydrazone must have one of the
following formulas, depending upon whether the hydrazine condenses
with the oxygen atom in the ortho or in the meta position to the chlorine.

C=0 C=0

HC<^CC1 H(

HciJcH HC^Cl

C = N — NC^HgOCeHj C = N — NC7H,0C.H,

I. II.

It has been shown that when the corresponding non-halogenated
compounds are saponified, hydroxyazo compounds are formed, the hydro-
gen which replaces the benzoyl group probably migrating from the nitro-
gen to the oxygen.

Now if formula I is correct, then on saponification one would expect to
obtain benzeneazoorthochlorphenol, while if formula II is correct, then
the corresponding benzeneazometachlorphenol would be formed. One

^ I

820 WILLIAM MCPHERSON AND WILBUR L. DUBOIS.

would naturally expect formula I to be the correct one, since it has been
shown that hydroxylamine condenses by preference with the oxygen
atom in the meta rather than in the ortho position to the chlorine. Ac-
cordingly, benzeneazoorthochlorphenol was prepared. The following
method gave a good yield and a fairly pure product:

Three grams of orthochlorphenol were dissolved in a solution of 7
grams of sodium hydroxide in 100 cc. of water, and the resulting solu-
tion diluted to looo cc. To the cold solution was added a solution of
benzene diazonium chloride made by diazotizing 3.36 grams of aniline
dissolved in 15 grams of concentrated hydrochloric acid and 30 cc. of
water with 2.61 grams of sodium nitrite dissolved in 20 cc. of water.
The resulting solution remained clear. On neutralizing with dilute hy-
drochloric acid, the hydroxyazo compound separated. It was purified by re-
peated crystallizations from ligroin, from which it separated in from brown
to yellow lath-shaped crystals.

Calculated for C^H^NjOCl: C, 61 .47; H, 3.90; N, 12.07.
Found: C, 61.20; H, 4.19; N, 12.27.

The benzeneazoorthochlorphenol so prepared is identical with the
product obtained by the saponification of chlorbenzoquinonebenzoyl-
phenylhydrazone. Each of the compounds, as well as a mixture of the
two, melt at 86°. Each, when benzoylated by the Batunann reaction,
give identical benzoyl derivatives, melting at 109°. Hence, formula I,
given above for the hydrazone, must be the correct .one.

Benzoyl Derivative of Benzeneazoorthochlor phenol ,

QC\\ CH

C^HjO.O — c/ Nc — N = NC^H,.— This was obtained by Bau-

mann's reaction. After adding the benzoyl chloride, the mixture was
thoroughly shaken and allowed to stand until the oily precipitate solidi-
fied. The product was easily purified by crystallization from alcohol

Calculated for Ci,Hj,N,0,Cl: C, 67.72; H, 3.89; N, 8.34.
Found: C, 67.34; H, 3.99; N, 8.62.

Benzoylbenzeneazoorthochlorphenol is a reddish yellow compound melt-
ing at 109*^. It is very soluble in hot alcohol and the solution, on cooling,
becomes almost solid with a mass of curved, hair-like crystals. It is
very soluble in benzene and moderately so in ether, separating from the
former in clusters of feather-like crystals and from the latter in clusters
of needles. It is not identical but isomeric with chlorbenzoquinoneben-
zoylphenylhydrazone. The hydroxyazo compound contains the benzoyl
group joined to oxygen, while in the hydrazone the benzoyl group is
joined to nitrogen.

Action of Reducing Agents on the Two Isomeric Benzoyl Derivatives,—
The hydroxyazo compound was dissolved in acetic acid, the solution cooled

ACTION Olf a-BENZOYLPHENYLHYDRAZINE. 821

and zinc dust added ; the solution at once became clear. The zinc dust was
removed by filtration. On adding water to the filtrate a white precipi-
tate was obtained, which, after purification from benzene-ligroin, melted
at from 157-158°. This was dissolved in alcohol and a few drops of a
solution of ferric chloride added. On standing, the original hydroxy-
azo compound was regenerated and separated from the alcohol in the
characteristic curved, hair-like crystals. No analysis was made of the
white compound obtained by the reduction, since its reactions show it
to be the dihydro derivative of the original hydroxyazo compound of the
following composition :

C^H^O— O— c/ V— NH— NHC,H,.

x:h=ch/

The isomeric hydrazone, when reduced in a similar way, gave an al-
most black solution from which, on the addition of water, a black solid
separated. While this probably contained benzanilide, it was found
impossible to get it sufficiently pure to determine its identity beyond
any doubt.

Conversion of Monochlorquinonebenzoylphenylhydrazone into Benzoyl-
oxyazochlorphenol. — ^Willstatter and Veraguth recently have described* a
most interesting and important rearrangement of the acyl derivatives of
quinonephenylhydrazones into the corresponding oxyazo compounds.
By means of this reaction the chlorquinonebenzoylhydrazone described
above is readily converted into the isomeric hydroxyazochlorphenol. The
hydrazone was dissolved in absolute ether and heated for several hours
with a small amount of anhydrous powdered potassium hydroxide, as
described by Willstatter and Veraguth. The hydrazone was converted
almost quantitatively into the isomeric oxyazochlorphenol, the benzoyl
group migrating from the nitrogen to the oxygen.

The Action of a-Benzoylphenylhydrazine StUphate on Monochlorquinone
in a Solution of Glacial Acetic Acid. Benzeneazoorthochlor phenol Sul-
phate.— Two grams of a-benzoylphenylhydrazine sulphate and 5 . 5 grams
of monochlorquinone were dissolved in separate portions of 40 cc. of
glacial acetic acid. The solutions were cooled to room temperature,
then mixed and slowly heated on the water-bath with occasional stirring.
At about 60° a mass of dark red, silky needles separated. The mixture
was set aside until cold and the crystals filtered off and dried. The re-
sulting substance was insoluble in all the common organic solvents ex-
cept hot glacial acetic acid and acetone, and only sparingly soluble in
these. It was purified by washing repeatedly with hot benzene. The
resulting compoimd melted with decomposition at 1 88°- 190®. The
properties of the compound were similar to the hydrochlorides of the hy-
* Ber., 40, 1432.

822 J. BISHOP TINGLE AND H. P. ROELKER-

droxyazo compounds described by Hewitt and Pope.* This led to the
belief that it might be the sulphate of benzeneazochlorphenoL Water was
poured over it, the mixture heated slightly and filtered. The resulting
compoimd was found to be benzeneazoorthochlorphenol. The sulphuric
acid in the water vTas determined.

Calculated for CnHjNjClO.HgSO^: HgS04, 30.09.
Found: HgS04, 29.65.

The formation of this compound is undoubtedly due to the saponifica-
tion of the hydrazone at first formed, the resulting oxyazo compound then
combining with the sulphuric acid originally present in the hydiame
sulphate to form the corresponding sulphate. While the hydrochterides
of a number of the hydroxyazo compoimds have been prepared, so far
as I know, this is the first sulphate to be described. Other quinones
gave similar results, so that this method of preparation would seem to
be a general one for the preparation of the sulphates of the hydroxyazo
compounds.

Columbus, Ohio.

STUDIES m ITITRATION. H.*— MELTING POINT CURVES OF BDIART
MIXTURES OF ORTHO- META- AND PARANTTRANILINES:
A NEW METHOD FOR DETERMINING THE
COMPOSITION OF SUCH MIXTURES.

Bt J. Bishop Txnolb akd H. P. Roblkbr.
Received March 3, 1908.

During the last academic year the senior author, in conjunction with
Dr. F. C. Blanck, carried out a somewhat extensive investigation of the
nitration of iV-substituted aniline derivatives. In the course of this work,
the results of which have been awaiting publication, in the AmericaM
Chemical Journal since the middle of August, 1907, the need was frequently
felt for a simple, expeditious method for the determination of the com-
position of mixtures of the isomeric nitranilines. Moreover, for our
purpose, it was necessary that the method should be applicable to rela-
tively small quantities of material. The only processes of which we
have been able to find descriptions in the literature consist of recrystalli-
zations, accompanied, in some cases, by conversion of one or more of the
nitranilines into some simple derivative. Apart from the question of
their accumcy, such methods did not appeal to us because they certainly
involve a considerable amount of labor, and probably demand for their
successful operation a relatively large quantity of material

Besides their diflFerences in solubility, the nitranilines vary somewhat
in color, in their strength as bases, and, quite widely, in their meltingr

* Ber., 30, 1624.

■ Bishop Tingle and Blanck, Am. Chem, J., 36, 605 (1906).

STUDIES IN NITRATION. 823

points. The color of the compounds, especially in dilute solutions, ap-
peared to be too nearly alike for our purpose, and a determination of the
partition of acid between the isomers involves too much trouble, conse-
quently we decided to investigate the curves produced by plotting the
melting points of binary mixtures of the nitranilines against the com-
position. In this connection we should like to mention that A. P. Holle-
man has been for some time engaged with the investigation of the influ-
ence of water on the course of nitration of benzene, acetanilide, etc.,
and he has also studied the influence of substitution in the benzene nucleus. ^
In the course of a paper with C. H. Sluiter* he gives a brief tabular state-
ment of the melting points of a few mixtures of />-nitracetanilide and
dinitracetaniUde. In another communication, with which we were not
acquainted until after our own work was in progress, he speaks of "per-
haps being able to determine the composition of binary mixtures of aro-
matic nitro compounds by reading oflf from a curve the depression in the
melting point of A caused by the addition of varying quantities of B."
As will be seen from the contents of our paper, such a mode of procedure
is inapplicable to mixtures of the nitranilines and is essentially different
from the plan which we have adopted. HoUeman also describes briefly
and without working details, a "method of extraction," which is appli-
cable to the determination of the composition of ternary mixtures in
general.*

Experimental.

Materials, — ^The nitranilines were obtained from Kahlbaum, they
melted at 71®, 114° and 147*^, the temperatures given in the literature
for the 0-, m- and />-compounds, respectively. Before use they were re-
crystallized from 95 per cent, alcohol, in such a way that half of the sub-
stance employed remained in the mother liquor. This treatment produced
no change in the melting point.

The alcohol^ employed in the experiments described below was the
ordinary con:imercial 95 per cent. It was filtered before use.

Preparation of Mixtures, — ^A weighed quantity of each nitraniline
was dissolved separately in alcohol, at the ordinary temperature ; the solu-
tions contained about 2 per cent, of ortho- and metanitraniline, respec-
tively, but in the case of the para derivative the concentration was only
about I per cent. The quantities of each solution required to produce
a mixture of the composition desired were run out from a burette and
the alcohol evaporated on the water bath. In some cases the evapora-
tion was allowed to take place at the ordinary temperature. The resi-
due was then placed in a desiccator on filter paper, allowed to remain

* Ber., 39, 1715 (1906).

' Rec, trav. ckim., 26, 208 (1906).

• Rep. Congres. chim. Pharm. Liege,, p. 283, July, 1905.

824 J. BISHOP TINGI^E AND H. F. ROELKER.

about 24 hours and the melting-point determined. Fresh solutions were
made up each day as required.

Orthonitraniline is very slightly volatile at the ordinary temperature
and, like the other isomers, vaporizes somewhat with boiling alcohol
The magnitude of the error introduced in this manner cannot be very
great, because mixtures of similar composition, prepared by evaporating
the alcohol at the two temperatures mentioned, always showed the same
melting point. Moreover, mixtures which had been wrapped in filter
paper and exposed to the air showed no change in their melting point
after several months. As will be seen from the tables, the composition
of the mixtures which we employed varied by intervals of 2 per cent.

During the course of this work we had reason to suspect the accuracy
of the statements made in the literature regarding the solubility of the
nitranilines and we therefore made the following determinations. Sepa-
rate portions of the three isomers were mixed with quantities of alcohol
insufficient for complete solution, and the liquids were maintained for
about 30 minutes, at 40*^, in a thermostat. The clear solutions were
then decanted and allowed to cool slowly to 15°. After remaining for
a time at this temperature, in contact with the crystals which had sepa-
rated, 10 cc. of each clear liquid was withdrawn by means of a pipette,
the solutions allowed to evaporate at the ordinary temperature, and the
residue dried in a desiccator. At 15°, 10 cc. of the respective solutions,
in 95 per cent, alcohol, contain 1.5845 grams of ortho-, 0.4960 gram of
meta- and 0.4030 gram of paranitraniline, respectively.

Table I. — Mblting Point op Mixtures of Ortho- and Metanitraniunbs.

Percent. Percent. Percent. Percent. Percent.

3rthc

>. M. p. c

>forth<

>. M. p. 0

f orth<

>. M. p. 0

forth!

J. M. p. 0

r ortho.

M. p.

68.7«>

55.3**

56. 6«

57-5°

107.0*^

68.o«»

550«»

56.7°

60. 2«

108.0^

67.o*»

54.80

56. 8«

58.2<»

109.0°*

63.0**

54. 60

56.8°

57.8°

IIO.O®*

61.0°

54 0*>

570*'

57-2*>

10 '

IH.O®*

60.6**

55-4^

57.3**

64. qO

108.0°*

58.0*'

55. 60

57. 60

70.4**

109.0°*

57.3°

55. 7**

58. oo

So.qO

III.5'**

55.5^

55.8<>

58.6«»

87.8<>

112. o°*

55.4^

56.5*'

59.2<»

104.0°

The melting points were determined in the ordinary manner; the ther-
mometer employed was a ** standard" one, capable of being read to 0.1°.
The temperature of the bath was usually raised rather rapidly until it
was within 25° of the melting point of the substance under examination,
after which the increase was about 8° per minute. No difl&culty was
experienced in obtaining constant results, even when the conditions
were somewhat varied. In certain cases, which are duly noted, it was

STUDIES IM NITHATION. 835

found to be desirable to determine the temperature at which the sub-
stance gave clear globvtes on the sides of the melting point tubes, t>ecause
for the mixtures mentioned this was much more definite than the point at
which the material melted completely and collected at the bottom of the
capillary. This procedure does not afiFect the use of the curves for quan-
titative purposes.

The temperatures marked * are those at which the substances melted
completely (see above). These mixtures, containing from 20 to 2 per
cent, of orthonitraniline, became clear and globular at temperatures
higher than those required to produce the same effect in the case of speci-
mens containing a higher proportion of ortho compound; consequently
this change of standard does not invalidate the use, for analytical pur-
poses, of the curve given below. In Fig. i, the data contained in the above
table are plotted in the form of a curve.

In Table II are given the melting points of mixtures of meta- and para-
nit raniline.

The mixtures marked * became clear and globular {vide Table I) at the
temperatures given, but did not melt completely. In a few cases the
melting-point was not sharp, therefore we determined the limits of tem-
perature within which the substance liquefied and give the mean value

J. BISHOP TINGLE i

) H. F. ROELKBR.

Tablb II — Mblting Points op Mxxtores op Meta- j

> Parakithanilinb.

8i 107

■5°

105

0°

108

o"t

■5'

104

5°

It)

5°t

.0°

lOt

0°

5°

107

0°

.0"

101

0°

o*^

113

s-f

■5°

100

0°

118

5°

0°

5°

113

o°t

■ 5°

0°

3°t

• S"

8=t

122

-o"

0°

5"'

128

o=t

5°

104

a=t

127

mm
18

130

•35

136

'38

140

142

'43

145
146

in the table; these figui

larked i". So far as we have been able to
test the question, we have found these temperature limits to be constant.
The 30 per cent, mixture was almost, but not completely, melted at 1 18.5°,
whereas that containing 4.6 per cent, of meta derivative behaved exactly
like a pure compound. The data in Table II are reproduced in the curve
Fig. 2.

Mixtures of ortho- and paranitraniline melt in a highly irregular man-
ner and the curve prepared from our data is quite unsuited for analytical

STUDIES IN NITRATION. 827

purposes. At first we were inclined to refer this irregularity to the pres-
ence of moisture, but this is hardly probable because the pure constit-
uents, after exposure to air, exhibit the correct melting points quite sharply ;
moreover, all our mixtures were treated in exactly the same maimer,
so that the disturbing factor should have affected each one to an equal
extent. In order to make certain of this point, however, some of the
mixtures were dried with very special care, but no alteration was apparent
in their melting points. It may be added that of the mixtures showing
the more pronounced irregularities, various specimens were prepared
at different times, but the results were unchanged. The mixtures which
exhibit the most marked breaks in the curve (containing 58-68 per cent,
of ortho compound) do not contain the constituents in any simple molecu-
lar ratio.

At present we have no explanation to offer of the reason for the irregu-
larities in question. It may be that they have a casual connection with
the well-known simultaneous production of ortho and para disubstituted
benzene derivatives and they are decidedly reminiscent of the alternate
rise and fall shown in the melting point of the aliphatic monobasic acids.
We are at present investigating this question more fully.

Application of the Melting Points for Analytical Purposes.

An inspection of the curves for the ortho- meta- and meta-para-mix-
tures shows that the direct determination of the melting point would be
sufficient, within some regions of temperature, to demonstrate the com-
position of a substance containing either of these constituents in unknown
proportion. Within a greater range of temperature, however, there are
at least two mixtures of widely differing composition which have iden-
tical melting points. In such cases the determination is made as follows :
Assuming that the material tmder examination consists of ortho- and
metanitraniline, a portion of it is mixed with an approximately equal
quantity of, say, pure metanitraniline, and the melting points of this
mixture and of the original material are determined simultaneously, on
the same thermometer. If the original material melts at a higher tem-
perature than it does after the addition of metanitraniline, its composi-
tion is indicated by a point on the left side of the curve, otherwise by the
corresponding position on the right-hand branch. If pure orthonitrani'
line is used for mixing, these relationships will be, of course, reversed.
In the case of the meta-para-curve the addition of pure paranitraniline
will lower the melting points of a mixture whose composition is situated
to the left of the eutectic point, whereas the remaining possible mix-
tures, to the right of it, will have their melting points raised by the addi-
tion of the pure para compound. Admixture with pure metanitraniline
will, necessarily, reverse these relations.

It will be obvious that the quantity of pure isomer which is added is

8a8 W. J. KARSLAKE AND W. J. MORGAN.

of no particular importance; we have found it convenient to take an
amount approximately equal to that of the material under examination.

The method has been tested by the junior author, who received a
number of mixtures, the composition of each of which was imknown to him.
His results were accurate to 2 per cent, and in some cases a^ed even
more closely.

In dealing with ternary mixtures it is advisable to crystallize them
fractionally from alcohol once or twice. In this manner we have found
that we could obtain two portions, one containing substantially all the
orthonitraniline together with some meta-, while the other portion con-
sisted of the m.eta- and paraisomers. We have made a few experiments
on the separation of ortho- and pamnitraniline by crystallization. So
far as we can judge, the separation can be made quantitatively. The
mixture is weighed and treated with just sufficient boiling alcohol (95
per cent.) to dissolve it. When cool, the volume is measured. The paia-
nitraniline which deposits is collected and to its weight is added that of
the quantity dissolved in the filtrate; the sum of these, subtracted from
the weight of the original material, represents the orthonitraniline.

Summary,

1. Curves have been constructed showing the relationship between the
melting points and the composition of the three binary mixtures of the
isomeric nitranilines.

2. In the case of the mixtures of ortho- and meta- and of meta- and
paranitmniline, these curves are comparatively regular; they fall to the
eutectic point and then rise to the melting point of the pure compound.

3. The curve representing the relationship between the melting point
and composition of mixtures of ortho- and paranitraniline is highly ir-
regular. At present it is impossible to give a satisfactory explanation
of this phenomenon.

4. The curves mentioned above can be used to determine the composi-
tion of mixtures of otho- and meta- and of meta- and paranitraniline,
respectively, by a simple and expeditious method which requires only a
minimal quantity of material. The results are usually accurate within
two per cent.

McMastbr University.

Toronto, Canada. •

December, 1907.

SOME DERIVATIVES OF i,3-DIMETHYL-2,6-DI]nTROBENZEITE-4-

SULPHONIC ACID,

Bt W. J. Karslakb and W. J. Morgan.
Received March 10, 1908.

The initial material used was the potassium salt of i ,3-dimethyl-2,6-
dinitrobenzene-4-sulphonic acid, which was prepared as follows:* 800
i Qaas and Schmidt, B. 29, 1424.

1 ,3-DIMETHYI,- 2 ,6-DINlTROBENZENK-4-SULPHONIC ACID. 82 9

cc commercial w-xylene were shaken in a separatory ftmnel with 800
cc. sulphuric add, sp. gr. i . 84, for about an hour, no attempt being
made to regulate the rise in temperature. After standing for an hour
or so the m-xylenesulphonic acid solution was drawn off from the undis-
solved portion of the hydrocarbons and slowly added, with constant
shaking, to a mixture of 1000 cc. of ordinary fuming nitric add and 500
cc. of ordinary fuming sulphuric add, the temperature being kept below
100°. After standing twenty-four hours, the mixture was heated
to 125-130^ for six or seven hours, allowed to cool, and poured into about
eight liters of cold water. After filtering off the insoluble predpitate
formed, which consists mostly of 2,4,6-trinitro-w-xylene, the filtrate was
heated, neutralized with caldum carbonate, again filtered, and a hot solu-
tion of potassium carbonate added in excess. The predpitated
caldum carbonate was filtered off, the filtrate evaporated to a
vohime of three or four liters and the potassium salt of ■1,3-dimethyl-
2,6-dinitrobenzene-4-sulphonic add allowed to crystallize out. A yield
of 640 grams was obtained and from the melting-point of the sulphon-
chloride, and sulphonamide, its identity established. It was later learned
that a better yield and a purer product could be obtained by isolating
the i,3-dimethyl-6-nitrobenzene-4-sulphonic add first formed and then
nitrating this with about twice its weight of fuming nitric add diluted
with double the wdght of ordinary sulphuric add, sp. gr. i . 84.

The sulphonchloride was made in the usual manner by treating the
potassium salt with an excess of phosphorus pentachloride, pouring the
product into ice-water, filtering and washing, and crystallizing the resi-
due from carbon tetrachloride. It had a melting-point of 123°.

i,j-Dimetiiyl-2,6'difUtrobenzene-4-sulph(mamUde, — Twenty grams of
aniline dissolved in 100 cc. carbon tetrachloride were added to 20 grams
of the sulphonchloride previously dissolved in 200 cc. carbon tetrachlo-
ride. After heating on the water-bath for two hours in a flask connected
with a return condenser the carbon tetrachloride was distilled off, the
residue washed several times with dilute hydrochloric add and after-
wards with water. It was then dissolved in a ten per cent, solution of
caustic potash, filtered, the filtrate addified, with dilute hydrochloric add,
and the predpitated anilide crystallized from dilute alcohol. It separated
in short, yellow needles which melted at 154°. It is easily soluble in
ether or chloroform, less so in carbon tetrachloride, and insoluble in
water. Upon analysis it gave 9.02 per cent, sulphur (Liebig's method)
against 9.12 per cent, calculated for CiJUjfi^^S.

i,3'Diin£thyl'2,6'dimirobenzene-4'Sulphon^O'toluidt^ — Five grams of
the sulphonchloride dissolved in 30 cc. of carbon tetrachloride were added
to fotu: times the equivalent weight of o-toluidine previously dissolved
in 100 cc. of carbon tetrachloride and the mixture heated in a flask with

830 W. J. KARSLAKE AND W. J. MORGAN,

a return condenser for four hours on the water bath. After distillfflg
off the carbon tetrachloride the residue was treated as in the preparation
of the anilide. From ninety-five per cent, alcohol the o-toluidide sepa-
rated in white radiating nodules, while from a mixture of benzene and
alcohol it crystallized in white needles. It had a melting-point of 135^.
Upon analysis it gave 8.68 per cent, sulphur against 8.76 per cent, re-
quired for CjjHjjOaNjS.

i,S'Dimethyl'2,6-dinitrobenzene'4'Sulphonr-p'toluidide. — Three times the
equivalent weight of /)-toluidine was heated with 5 grams of the sul-
phonchloride in the presence of dilute caustic potash for about half an hour
on the water-bath. The mixture was then filtered and the filtrate acidi-
fied with dilute hydrochloric add until no further precipitation took
place. The /^-toluidide so obtained was crystallized from alcohol and
separated in yellow plates or in silky needles having a melting-pomt of
162^. It is soluble in chloroform, less so in carbon tetrachloride, and
insoluble in water. Upon analysis it gave 8.96 per cent, sulphur against
8 . 76 required by theory.

i,S'Diinethyl^2,6'dimtroben2ene'Sidphone-benzene. — ^Five gmms of the
sulphonchloride were dissolved in an excess of benzene and one gram of
anhydrous aluminum chloride added. The mixture was then heated
for three hours in a flask with a reflux condenser on the water-bath. The
excess of benzene was distilled off and the residue successively treated
with dilute caustic soda solution, dilute hydrochloric acid and water,
and then dried. It was finally digested with carbon tetrachloride and a
small quantity of boneblack, filtered, and allowed to cool. The sulphone
separated in large glistening plates, which melted at 178^. Upon analysis
it gave 9.70 per cent, sulphur against 9.52 percent, required for
Ci,H„OeN,S.

i'Carboxy-3-ni€thyl-2f6-diivUrobenzene'4-sulphomc Acidif). — Eight hun-
dred grams of the potassium salt of i,3-dimethyl-2,6-dinitrobenzene-4-
sulphonic add were dissolved at room temperature in five liters of water
and to it was added a solution of 1700 grams of potassium permanganate,
50 gmms of caustic potash, and enough water to dilute the whole to 120
liters. After standing three weeks at room temperatme the solution
was warmed on the water-bath, decolorized by the addition of a little
alcohol, filtered from the manganese dioxide, and evaporated to crystal-
lization. One hundred grams of imchanged original potassium salt sepa-
rated out. This was filtered off, the filtrate acidified with hydrochloric
acid, evaporated to dryness, and extracted with alcohol. The alcoholic solu-
tion upon evaporation to dryness gave a sirupy liquid which was treated
with phosphorus pentachloride, poured into ice-water, the insoluble
residue thoroughly washed with water and, after drying, crystallized
from carbon tetrachloride. The crystals thus obtained were brown in

a-DINAPHTHYL SEl,ENIDE AND TELLURIDE. 831

color, granular, and irregular in shape. They melted at 127-128** and
were probably an impure acid chloride. Upon digestion with concen-
trated aqueous ammonia they formed a product which, upon crystalliza-
tion from alcohol, gave yellow plates possessing no definite melting-
point and charring at about 290°. They were not analyzed. The resi-
due, which was insoluble in alcohol, was recrystallized several times
from water and then analvzed . The results obtained indicate that one of the
two methyl groups, CH3, present in the original salt, was oxidized to the
corresponding carboxyl group, CO. OH, but it yet remains to be shown
which one of them it was. We are hoping to determine this point in
the near future. Tentatively it is assumed to be that methyl group
which is in the first position (i). Upon this assumption the salt obtained
was the acid potassium salt of i-carboxy-3-methyl-2,6-dinitrobenzene-
4-sulphonic acid and contains one molecule of water of crystallization.
The analysis gave 11.04 P^^ cent, potassium, 5.00 per cent, water, 9.00
per cent, sulphur, against 1 1 . 08 per cent, potassium, 4 . 97 per cent,
water, 8 . 83 per cent, sulphur required by theory.

The acid barium salt was obtained by adding barium carbonate to the
hot solution of the acid potassium salt, filtering, and adding hydrochloric
acid to the filtrate. Upon coooling ther6 separated yellow plates con-
taining three molecules of water of crystallization. The analysis gave
15 -59 per cent, barium, 12.68 per cent, water, compared to 16.06 per
cent, barium, 12.62 per cent, water calculated according to theory.

The neutral barium salt was prepared by adding barium carbonate to
the hot solution of the acid potassium salt and filtering. It separated
in light yellow plates containing no water of crystallization. The analy-
sis gave 30.98 per cent, barium against 30.90 per cent, barium required
by theory.

The neutral strontium salt was formed by adding strontium carbonate
to the hot solution of the acid potassium salt and filtering. It separated
in almost white, square crystals containing no water of crystallization.
The analysis gave 21.81 per cent, strontium against 22.08 per cent,
required by theory.

University of Iowa,
Iowa City, La.

CONCERinNG a-DINAPHTHYL SELENIDE ASD TELLURIDE.

By R. B. I«yon8 and G. C. Bush.
Received March 12, 1908.

Some years ago, while collecting and tabulating material for the com-
parative study of the periodic relationship in the oxygen family, espe-
cially in the organic combinations of sulphur, selenium and tellurium,
vc were led to undertake the synthesis of certain aromatic compounds of

832 R. E. LYONS AND G. C. BUSH.

selenium and tellurium necessary to fill in the blank spaces in the chart.
The work was not completed because of our separation. The compounds
prepared in this connection are as yet unreported and may be described
as follows :

1. a-Dinaphthyl Selenide, (CioH7)2Se. — KiafFt and Lyons* prepared
phenyl telluride by heating together mercury diphenyl and metallic
tellurium. We undertook the preparation of a-dinaphthyl selenide in
an analogous manner from a-dinaphthyl mercury and selenium:

(CioH7)3Hg + 2Se = HgSe + {Cy,YL^)^.
A mixture of 4.5 grams a-dinaphthyl mercury, m. p. 242®, and 1.58 grams
finely powdered selenium, contained in a small Anschutz flask, was kept
in a Wood's metal bath at 190° under a pressure of 16 mm. for about 12
hours. A temperature above 200 ^^ was found unfavorable, causing the
dinaphthyl mercury to split into mercury and naphthalene. When cool,
the mixture was distilled with steam to remove the naphthalene and the
residue extracted with ether. Evaporation of the extracts gave 1.7
grams, or 54 per cent, of the theoretical amount of the dinaphthyl selenide.
The product crystallized from absolute alcohol, in which it is moderately
soluble, in fine, faintly yellow, glittering leaflets, which melted at 114®.
Neither treatment with charcoal nor crystallization from other solvents,
as ether or amyl alcohol, gave a colorless product. Exposure to air and
light effects rapid decomposition of the selenide.*

(C,oH7)iSe, Calculated: C, 72.28; H, 4. 18.
Found: C, 72.20; H, 4. 11.

The selenium in the compound was determined by the method of

Lyons and Shinn.' 0.3816 gram required 45.7 cc iV/io sodium thio-

sulphate. Calculated, 23.73; found, 23.65 per cent. Se.

2. a-Dinaphthyl Selenide Dibromide, (CioH7)2Se.Br2. — ^When the theo-
retical quantity of bromine was slowly added to a warm alcoholic solu-
tion of the selenide and the mixture allowed to stand for several hours
the dinaphthyl selenide dibromide separated out in the form of dirty
white needles. The yield was practically quantitative. The bromide
is soluble in amyl alcohol, but was more readily crystallized from carbon
disulphide as white, delicate needles which melted at 183° with decom-
position.

Calculated for (C,oH7)jSe.Br,: Br, 32.45.

Fotind: Br, 32.31.

Diphenyl selenium oxide* and diphenyl tellurium oxide* have been ob-

' Ber., 27, 1769.

* All selenium and telluriiun compounds described in this paper are so affected
in varying degrees.

» Tms Journal, 24, 1087*

* Ber,, 26, 2819.

* IHd., 27, 1770.

a-DINAPHTHYIv SHLENIDE AND TELLURIDE. 833

tained by treating the respective bromides with dilute solutions of sodium
hydroxide. The treatment of a-dinaphthyl selenide dibromide with solu-
tions of sodiuiti hydroxide of varying concentration at different tempera-
tures failed to produce the desired dinaphthyl selenium oxide, (CioH7)3SeO.
Boiling with 15 per cent, sodium hydroxide solution produced no change
in the substance. A 25 per cent, solution of the alkali was without ac-
tion at room temperature, but at about 90*^ the bromine was removed.
That oxygen did not take its place was shown by the recovery of di-
naphthyl selenide from the reaction mixture.

3. a-Dinaphihyl Selenide Dichloride, (CioH7)3Se.Cl2. — ^A current of dry
chlorine gas passed into an ether solution of the dinaphthyl selenide
produced immediately a heavy, white, amorphous precipitate of dinaphthyl
selenide dichloride. The chloride is insoluble in alcohol, ether, chloro-
form, benzene, ligroin, carbon disulphide and amyl alcohol; easily solu-
ble in xylene, from which it crystallizes in colorless prisms melting at

130^

Calculated for (CioH7)iSe.Cl,: CI, 17.53-

Found: CI, 17.40.

4. a-Dinaphthyl Telluridey (CiQHj)2Te. — ^This compound was obtained
by heating together molecular quantities of a-dinaphthyl mercury and
tellurium according to the method successfully used in the preparation
of phenyl telluride and a-dinaphthyl selenide,

(CioH7),Hg + 2Te = HgTe + (C«,H,),Te.
A mixture of 4 . 5 grams dinaphthyl mercury and 2 . 5 grams powdered
tellurium, contained in a small Anschiitz flask, attached to a vacuum
pump, was kept at 190-198° and 16.5 mm. pressure for about 8 hours.
After cooling, the solidified mass was distilled with steam to effect the
removal of the naphthalene which had collected in the upper portion of the
flask. The residue was then extracted with ether, the ether solution
quickly filtered and evaporated, to avoid excessive separation of tellur-
ium. The solid residue was purified by crystallization from much alco-
hol. Three crystallizations of the telluride from alcohol gave glittering
brownish yellow leaflets which melted at 126.5°. The yield of the crude
product was 53 per cent, of the theoretical. Excessive or prolonged heat-
ing of the reaction mixture brings about a secondary change resulting
in the formation of a yellow product which melts at about 190° and im-
parts to alcohol, ether, carbon disulphide and other solvents a very pro-
nounced fluorescence. The product was tellurium-free and was not
further examined.

(C,oH7)2Te, Calculated: C, 63.33; H, 3.69.
Found: C, 63.23; H, 3.49.

The tellurium was determined by decomposition of the substance with
red fuming nitric acid in a Carius tube, reducing the tellurous acid in

834 R- H. LYONS AND G. C. BUSH.

hydrochloric acid solution with sodium bisulphite, collecting the precipi-
tate on a weighed Gooch filter, drying at 105°, and weighing.

Calculated for (C,oH7),Te: Tc, 32.98.

Fotrnd: Te, 32.79.

5. a-Dinaphihyl TeUuride Dibromide, (C,oH7)2Te.Br,. — ^The addition of
bromine to an ether or alcoholic solution of a-dinaphthyl telluride gave
immediately a heavy yellow precipitate of dinaphthyl telluride dibromide.
The precipitate was washed with ether and crystallized from carbon di-
sulphide, in which it is sparingly soluble, as sparkling, lemon-yeUow
granules, which melted with decomposition at 244°. If the bromine is
added to an exceedingly dilute, warm, alcoholic solution of the telluride
and the mixture be allowed to stand for several hours, fine yellow crys-
tals of the bromide separate, which, after washing and drying, melt at
244°.

Calculated for (C,oH,)«Te.Br,: Br, 29.57.

Pound: Br, 29.67.

6. a-Dinaphihyl Tellv/ride Dichloride, (CioH7)2Te.Cl2. — ^When a current
of dry chlorine is passed into an ether solution of a-dinaphthyl telluride a
heavy white precipitate of the chloride is formed. The chloride is in-
soluble in all of the ordinary solvents, but is sparingly soluble in xylene,
from which it crystallizes in glittering, colorless granules, melting at 265'*.

Calculated for (C,oHOtTe.Cl,: CI, 15.74.
Found: CI, 15-83.

7. Diphenyl Telluride Dichloride, (CaH5)2Te.Cl2, was prepared by passing
dry chlorine gas into an ether solution of diphenyl telluride* and puri-
fied by recrystallization from xylene, in which it is very readily soluble.
Long, white prisms, melting at 160®.

Calculated for (CJHJaTe.Cl,: CI, 20.24.

Found: CI, 20.36.

8. P-Dinaphihyl Selenide Bichloride, (CioH7)jSe.Clj, was prepared by the
action of dry chlorine gas upon the selenide in ether solution. By recrys-
tallization from carbon disulphide it was obtained in almost coloriess
leaflets, melting at 146°.

Calculated for (C,oH7),Sc.C1,: CI, 17.53.

Found: CI, 17.60.

From the time of Doebereiner the similarity of sulphur, selenium and

tellurium compounds has been a subject of comment. There is, in general,

a pronounced similarity in methods of formation, in deportment and in

the effect of substituents in the compounds of these elements with organic

radicals. The changes in the physical constants, e. g., boiling-point,

melting-point, to be observed in a comparison of sulphides, selenidcs

* Ber.y 27, 1769.

* Ibid., 27, 1767.

a-DINAPHTHYL 3ELSNIDE AND TELLURIDE.

835

and tellurides of like radicals, are usually gradual and progressive with
the increase in magnitude of the atomic weights. Data concerning the
compounds described in this paper and some related compotmds, pre-
viously reported/ are placed in the following tables:

Sulphides. B. p,

, (16 mm.).

Selenidet. B.

p. (16 mm.).

Tellvridct. B. p. ( x6 mm. ) .

(CJIJ^

157**

(C^JiPe

167®

(C^O.Te

182®

C,H,.SH

172®

C^il^.Scn

183®

. . • •

M. p.

M.p.

M. p.

(C^J^.S

6o<>

(CeHJiSe.Sc

63.5^

• • V •

(CA)^

70.5*'

(C.N*)i3cO

114^

(C.H^,TeO

187®

«-(C,.H,)^

no®

a-(C,oHOiSe

114**

a-(C.H,).Te

126.5®

HC..H,)^

151^

P<C,^7)^

138.5^

• • • ■

(*^H,.CA).S

64®

((^<:H,.CJaJ^

62®

(a^H,.CHJ,Te

38»

(^CH^CJHJ^

57**

(^CH,.CeHJ^

69®

(^CH.C^4).Te

64" •

(CAa)iS

88®

(C.H,Cl)|ge

96®

• • • •

(C^«Br);S

109.5®

(C,H,Br)^

"55°

• • • •

Ths Hai^ogbn Addition Products,

CUoridei

M.p.

Bromides.

M.p.

(C,H0AC1,

(C,HO,Se.a, i83«>

(C^^.Te-Cl, i6o»

^-(C,oH,)ACl.

y»-(C.H,)iSe.a, 1460

/»-(C,.H,),Te.Cl,

or-(C|Q£i7}fS.d2 ....

a.(CioHT)«Sc.Cl, 130'*

a-(C,oH,).Tc.Cl, 265®

(/»-CH,.C.HJ,3.a,

(/^H,.C,^4),Sc.Cl, 177°

/>-CH,.CeHJ,Tc.a,

(o-CH,.CJHJ^.a, 152°

(o-CH^C,HJ,Te.Cl,

(C.HJ^.Br,

(C.H3)^.Br, 148®

(CHJjTe.Br, 203®

/?-(C,oH,)ABr,

i5-(C,oH,),Sc.Br, 161®

/?-(C,oH,).Tc.Br,

a-(C,oH,)^.Br.

a-(C,oHT),Sc.Br, 183°

a-(QoHT),Tc.Br, 244°

(A-CH,.C,H4)ABr,

(/»-CH,.C.HJ^.Br, 162®

(/»-CH,.CeHJ,Te.Br, 201 ®

{o-CH,.C.HJgSeBr, 84®

(o-CH,.CeHJ,Te.Br, 182®

Consideration of the rather limited number of these compounds now
available shows that the change in boiling and melting point generally,
but not always, varies directly with the magnitude of the atomic weight.
The first known and simpler compounds of these elements, e. g,, the phenyl
compounds, conform to this and exhibit differences which are strikingly
constant. However, examination of all of the data given in the above
tables shows that the change does not follow a rigid or fixed rule. In
some instances the changes are very abrupt and out of proportion to the
difference between the atomic weights; in other instances the change is
the reverse of the expectation and is without plausible explanation,
e. g., o-tolyl sulphide, selenide and telluride show the melting-points
64**, 62® and 38°, respectively. The melting-point decreases as the
atomic weight increases and the drop from 62^ to 38° is abrupt.

^ B#r., 261 3818; 27| 1764; aSy 167a

836 HBNRY A. TORREY AND H. B. KIPPER.

The changes are even more erratic and irregular in the halogen addi-
tion products. An extreme case is found in the o-tolyl selenide dichlor-
ide and dibromide melting at 152° and 84°, respectively. The faabgen
substitution products, thus far reported, exhibit diflferences which are more
nearly constant than is observed among the halogen addition products

Chbmical Laboratories,
Indiana Univbrsitt, Bloominoton, Ind.,

March, 1908.

(Contribution prom thb Chemical Laboratory op Harvard Collbgs*)

HYDRAZONES OF AROMATIC HYDROXTKETOITES. ALEAU-

mSOLUBLE PHENOLS.

SECOND PAPER.

Bt Hbnrt a. Torrey and H. B. Kippbr.

Received March 13, 1908.

Although it is a very general rule, so general indeed that it is almost
universal, that phenols are soluble in aqueous alkalies, there are certain
substances of this class that are marked exceptions. The work described
in this paper consists of an extension of the list of such compounds, aod
some investigation of the conditions to which this alkali-insolubility
is due. While studying the phenylhydrazones of certain hydroxy aceto-
and benzophenones, it was found, as mentioned in an earlier paper,*
that when the free hydroxyl was in the ortho position with reference to
the substituted ketone group, the substance was insoluble in strong
aqueous alkalies. Substances showing this characteristic property have
been obtained and studied by others. O. Anselmino' studied the phenyi-
hydrazones of homosalicylaldehydes and obtained alkali-insoluble com-
pounds similar to ours. Liebermann* first discussed in detail the alka-
line insolubility of benzene azonaphthol, which was later studied by Gold-
Schmidt and R. Brubacher,* McPherson' and Hantzsch and Farmer.*
St. V. Kostanecki has obtained certain iiitrogen-free phenols which are
insoluble in alkalies; thus A. Comelson and St. v. Kostanecki^ report
that 2-hydroxy-benzaldiacetophenone ,

HOQjH^CHCCHjCOCeHs),,
is insoluble in warm dilute sodium hydroxide, although it does dissolve
in hot 15 per cent, potassium hydroxide, and St. v. Kostanecki and R-
V. Salis* state that 2-ethoxybenzalresacetophenonmonoethyl ether,

* This Journai^, 29, 77.

* Ber., 35, 4099.
« Ibid,, 16, 2858.

* Ihid.f 24, 2306.

• Ibid,, 28, 2418.

• Ibid., 32, 3100.
' Ibid., 29, 242.

• Ibid., 32, 1030.

HYDRAZONES OF AROMATIC HYDROXYKETONES. 837

/OH (2)
(4) C^.OC,H,<;

X;OCH = CHC,H,OCyi, (2)

is insoluble in aqueous alkalies. Rogow^ and R. Fosse and A. Robyn,*
by the condensation of hydroxjraldehydes with ^-naphthol, obtained di-
naphthoxanthenes, containing hydroxyl groups, which were insoluble
in aqueous sodium hydroxide. Rogow^ mentions also condensation prod-
ucts of hydroxy-aldehydes with ^-naphthylamine which are almost in-
soluble in dilute alkalies. The phenylhydrazone of />-homosalicylalde-
hyde, as obtained by Anselmino, is a compound entirely insoluble in
cold, highly concentrated alkalies. The phenylhydrazone of salicyl-

aldehyde, on the other hand, although its hydroxyl group is in the ortho
position to the side chain carrying the hydrazone radical, is, neverthe-
less, soluble in alkalies. The solubility of the phenol is therefore influ-
enced by the presence of the methyl group, and this effect is even more
marked when more than one methyl group is present. Anselmino pre-
pared a number of derivatives of this nature, including in addition to
the phenylhydrazones, azines and semicarbazones. Those phenoxyalde-
hydes, however, which gave phenylhydrazones, insoluble in aqueous
sodium hydroxide, gave azines and semicarbazones which were soluble
in this reagent. He found also that phenylhydrazones of all /)af a-hydroxy-
aldehydes were normally reacting phenols, dissolving easily in alkalies
of all concentrations, and being reprecipitated with adds. Our inves-
tigation of the phenylhydrazones of the o-hydroxy-, aceto- and benzo-
phenones was begun with the idea of obtaining a product of a secondary
condensation between the hydroxyl group and the imide group, and the
alkali-insolubility of many of the phenylhydrazones which we obtained
lent color to the belief that such a condensation had actually occurred.
The analytical results and the subsequent work, however, made such a
conclusion untenable. We have already mentioned the influence of the
methyl groups on the alkali-insolubility of the derivatives of the alde-
hydes, as studied by Anselmino ; in none of the compounds which we have
investigated is there an alkyl group directly attached to the ring, but we
have found that the presence of other groups in the ring has the same

* Ber., 33, 3535; J, pr, Ch. [2], 72, 320.

^Compi, rend., 132, 789; 137, 858; 138, 282a; 140, 1538.

838

HENRY A. TORREY AND H. B. KIPPBR.

influence. Our work has been directed chiefly toward the study of phenyl-
hydmzones and other similar derivatives of resacetophenone and reso-
diacetophenone. In resacetophenone,

— CH,

II
O

one hydroxyl group is ortho and the other para to the ketone side chain
and its phenylhydrazone is, as would be expected, easily soluble in alkalies.
The phenylhydrazone of paeonpl,

^"^ C— CH,

however, in which the only free hydroxyl is ortho to the large side chain,
is insoluble in aqueous alkalies, although paeonol itself is readily soluble.
Since the phenylhydrazone of o-hydroxyacetophenone,

/\ C— CHg

C,H,

is soluble in alkalies, it is evident that the methoxy group in paeonol has a
real influence upon the solubility. We have found also that the CbH^COO-,
the CHjC - NNHCeHj, and the CeH^C = N . NHCeH^ groups are contributing

factors in this phenomenon of alkaline insolubility. To make clearer the in-
fluence of these last two groups upon the solubility, it is only necessary
to call to notice the fact that although the phenylhydrazone of o-hydioxy-
acetophenone, whose formula has just been given, is soluble, the bis-
phenylhydrazone of resodiacetophenone,

CH,

II
N

HNH
C.H,

-C— CH,

II
N

— OHNH
i^H,

HYDRA20^mS OP AROMATIC HYDROXYKETONES.

839

is exceedingly insoluble, even in hot aqueous alkalies. Thus, if we con-
sider the influence of the substituting groups upon one hydroxyl at a
time, we may say that the insolubility is brought about by the joint in-

I
fluence of the CHgC = N — NHC^Hj adjacent to the given hydroxyl and of

I
the other CH,C=«N — NHCeHj, this second group playing the part of

the methoxy group in paeonol or the methyl groups in the aldehyde de-
rivatives. While the benzidine derivative of resodiacetophenone, in
which two molecules of benzidine have condensed with two molecules
of the ketone, shows the same insolubility toward alkalies as the phenyl-
hydrazone, the azine and semicarbazones are soluble. If we tabulate
the phenols that we have studied according to their solubility in aqueous
alkalies, they fall into the following groups:

Soluble in Aqueous Alkalies.

Phenylhydrazone of resacetophe- Monophenylhydmzone of resodi-
none, acetophenone,

/\ C— CH, CH,— C— /\f C— CH,

— OHNH

in.

HO— vy— OHNH

II
N

Semicarbazone of resacetophe-
none,

■C— CH,

C,H,

Phenylhydrazone of nitioresaceto-
phenone,

HO— vy— OH NH

HO—

■C— CH,

—OH NH

1
C,H, '

NH,

Monoazine derivative of resodi-
acetophenone.

CH,-C

C— CH, CI^— C

Bisazine derivative of resodiaceto-
phenone.

HO—

■C— CH,

Nho-/\-

CHj— C

' Portion of nitro group not proved.

C— CH, CH,— C

C— CH,

840

HENRY A. TORRBY AND H. B. KIPPER«

Insoluble in Aqueous Alkalies.

Phenylhydrazone of paeonol,

-C— CH,

II
N

CH,0— V .— OHNH

C,H.

Phenylhydrazone of resacetoph^
none-4-monobenzoate,

■C-CH,

C,H^— COO— . y—OH NH «

Phenylhydrazone of resaceto-
phenone-4-monoacetate,

CHjCOO—

■C— CH,

— OHNH *

C,H.

Phenylhydtazone of lesodiaceto-
phenone monacetate,

CH,CO— /^

CH,COO—

-C-CH,

II
N

—OHNH

I
C^H,

Phenylhydtazone of resodiaceto-
phenone monobenzoate,

CHjCO— /^

C,H.— COO—

CCH, CH,— C

Bisphenylhydrazone of resodi-
acetophenone,

—OH NH » H— N HO—

I I

Q,Hg CeHj

■C-CH,

—OHNH

Bisphenylhydrazone of mono-
methyl ether of resodiacetophenone,

CH,— C-

HNCHjO—

I
C,H.

C— CH,

—OH NH

C,H,

* Position of Br not proved.
' Saponified by alkalies.

Bisphenylhydrazone of dibenxo
resorcinol,

CH,-C-

II
-N

C,H,

-C-CgH,

II
N

HN HO^ ^— OH NH

C,H.

HYDRAZONBS OP AROMATIC HYDROXYKBTONBS.

841

Bisphenylhydrazone of dibenzo-
hydroquinonol,

HO— /^

C^.-C-

II
N

—OH NH

I
C.H,

Bis />-brotnphenylhydrazone of
resodiacetophenone,

CH,— C-

HNHO—

■C— CH,

H^Br

— OHNH
I

C,H«Br

Condensation product of benzi-
dine and resodiacetophenone,

CH,— C

■C— CH,

—OH N

CH,— C

Dianilido monazine derivative
of resodiacetophenone,

CH,— C-

HO—

II
CH,— C-

xy^

Bisphenylhydrazone of monobromresodiacetophenone,

CH.— C /\ C— CH,

II
N

HNHO— \ /'—OHNH
C,H, ^'

C,H,

-C— CH,

II
N

HC,H,

H C,H.

II
— C— CH,

It will be noticed that in the compounds that we have studied, the insolu-
bility in aqueous alkalies is determined by the fwo following conditions:
(i) The free hydroxyl group is ortho to a large side chain, as

' Tms Journal, 29, 81.
* Position of Br not proved.

842 HENRY A. TORREY AND IL B. KIPPER.

CH,

- N — NHCeH.;

(2) other substituting groups are present, as OCH, or OOCCeH^, etc
If the basic group substituting the ketone oxygen is small, as in the bis-
azine or resodiacetophenone, the compoimd is soluble. Although An-
selmino found that the phenylhydrazone of homosalicylaldehyde ^vas
insoluble in alkalies, K. Auwers and R. Bondy^ have found that the
phenylhydrazone of 5-nitro-2-hydroxy-i-methyl-3-benzaldehyde is soluble,
while the same derivative of 5-nitro-4-hydroxy-i-methyl-3-benzaldehyde
dissolves with difficulty, although we prepared phenylhydiazones from
nitrated paeonol and from methylated nitroresoacetophenone. We have
omitted them from the experimental part of this paper, as they have not as
yet been sufficiently studied. The introduction of a less negative group,
as bromine, into the ring, does not affect the solubility, for we found
the bisphenylhydrazone of bromresodiacetophenone to be insoluble in
alkalies.

It is, in our opinion, impossible to give at the present time an adequate
explanation of the alkali-insolubility of these compounds.

Anselmino, in discussing the insoluble aldehyde derivatives, shows
that no secondary condensation between the imide and hydroxyl group
has taken place and calls attention to the fact that E. Fischer' has shown
to be erroneous the statement of Causse' that such a condensation oc-
curs between salicylaldehyde and phenylhydrazine in the presence of
acetic anhydride. Not only do our analytical results show that in the
phenylhydrazones recorded by us no such secondary condensation has
occurred, but it is evident that no condensation of the kind could take
place in the alkali-insoluble aniline and benzidine derivatives. This
hypothesis is thus made untenable. A second hypothesis is that an inner
salt has been formed, which, with the bisphenylhydrazone of resodiaceto-
phenone, would be formulated as follows:

CUf—Cr

Hv.

>N— O—

<^CH,

N
H

-K

The comparatively weak add nature of resodiacetophenone and the
ortho hydroxy ketones studied makes such a h3rpothesis seem extremely

* Ber,, 37, 3915.

* Ibid., 30, 1240.

' Comfi, rend,, 124, 505.

HYDRAZONBS OF AROMATIC HYDROXYKETONES. 843

tinlikely, and further, it would offer no explanation for the difference
in solubility in alkalies between the phenylhydrazone of o-hyroxyaceto-
phenone and the phenylhydrazone of paeonol, or especially between the
phenylhydrazone of salicylaldehyde and that of homosalicylaldehyde.
In order to give some test to this theory, we have introduced a bromine
atom into each of the phenylhydrazone groups in the bisphenylhydra-
zone of resodiacetophenone, thereby increasing the negativity of these
^oups and consequently decreasing the tendency toward salt forma-
tion, but without effect upon the solubility. It might be suggested
that the insolubility is due simply to the fact that the add nature of the
phenol has been highly depressed by the introduction of basic groups,
such as the phenylhydrazone or benzidine groups. If this were the true
explanation, we should expect the phenylhydrazones of paeonol, or of
the acetates or benzoates of resacetophenone and resodiacetophenone
to be more soluble in alkalies than the phenylhydrazones of o-hydroxy-
acetophenone or of salicylaldehyde instead of less soluble. The alkali-
insolubility of the compounds, then, cannot be explained by the hydrolytic
action of water, especially when one considers that the bisphenylhydra-
zone of resodiacetophenone is not dissolved by a solution of sodium hy-
clroxide of as great strength as i : i. It was suggested in a former paper ^
that these compotmds may have a quinoid structure and that the insolu-
bility is due to this; thus for paeonol phenylhydrazone we have the for-
mula,

CH,0—

_A==

We have been unable, however, to get any indication of such quinoid
oxygen by the action of hydroxylamine, although it should be said that
even if such a quinoid oxygen were present, it would be unlikely to re-
act with hydroxylamine, and Anselmino' has shown that in the phenyl-
hydrazones of the homosalicylaldehydes which are insoluble in aqueous
alkalies, the phenylhydrazone group reacts with pyruvic acid to give
its hydrazone, and that the phenol oxygen can be benzoylated by benzoyl
chloride in pyridine solution. We are not, however, yet ready to entirely
abandon the quinoid formulas as a possibility. The alkali-insoluble
phenols that we have studied bear a certain analogy to 2-naphthol-i-

* This Joxtricai^, 29, 77.

• Ber,, 35, 4101.

844 HBNRY A. TORRBY AND H. B. KIFPBR.

•N = N— CgH,

azobenzene, | I | , which is insoluble in aqueous aOca-

lies, and in which the four carbon chain | has the same influence

HC.

on the solubility as the carbon-containing groups in our compounds.
A. Goldschmidt and R. Brubacher,* by a study of the reduction products
of its acetyl and benzoyl derivatives, have shown that in all probability

=N— NHCA

this substance actually has this quinoid structure | I {

\/\y

Fosse and Robyn* attribute the alkali-insolubility of the rather differ-
ent compotmds studied by them to the presence of a quadrivalent oicygen
atom. If we assume that in our compounds, the oxygen of the hydroxyl
is quadrivalent, the formula for the phenyl hydrazone of paeonol, for in-
stance, might be written thus:

,_A

H—f > ^C-CH,

or .

CH,0— i i=0— N

Y I I

Jj H NH

Since, however, quadrivalent oxygen ordinarily shows basic properties,
such constitutions as these seem tmlikely.

A very striking fact regarding the alkali-insoluble phenols is that al-
though they are insoluble in aqueous alkahes they all dissolve with gieat
ease in alcoholic alkalies and may be precipitated by mineral adds from
these solutions unchanged.

In general, one may say regarding the alkali-insolubility of the phenols,
that it is due to the combined influence of a large group as

H H H

— C = N — NCeHg,— C = N — NC^Hj, etc.,

CH,
in the ortho position to the hydroxyl and a carbon-containing group else-

* Ber., 24y 2306.
" Loc, cit.

HYDRAZONBS OF AROMATIC HYDROXYKETONBS. 845

where in the ring. Whether the insolubility is due to an actual change
in structure cannot at present be stated.

In the course of this work we have noticed some interesting cases of
so-called steric hindrance, particularly in connection with the action
of phenylhydrazine on resodiacetophenone diacetate and dibenzoate.
When two molecules of phenylhydrazine were allowed to act on one
molecule of resodiacetophenone diacetate in a hot alcoholic solution,
containing acetic add, the product obtained was the monophenylhy-
drazone of resodiacetophenone monoacetate, instead of the bisphenyl-
hydrazone of the diacetate. The reaction proceeded, then, according
to the following equation :
H

CH,C — /\ — CCH, CI^C /^ CCH,

II 11

II
O

CHjCOO

O + 2C.H,NHNH, = O
OOCCH, CH,COO—

I
— OH'NH

CH,CONHNHCeHg +11,0.

The acetylphenylhydrazine formed ia the reaction was isolated from the
mother liquor. If a large excess of phenylhydrazine, not less than four
molecules, for example, was allowed to act on the diacetate under the
same conditions, both acetyl groups were eliminated with the forma-
tion of resodiacetophenone bisphenylhydrazone. The action of phenyl-
hydrazine on the dibenzoate of resodiacetophenone is exactly analogous
to its action on the diacetate. In an earlier paper^ we stated that phenyl-
hydrazine, acting upon resacetophenone dibenzoate gave the phenyl-
hydrazone of the dibenzoate, melting at 183°. We have found, how-
ever, that this is erroneous, and that here also the benzoyl group adja-
cent to the keto side chain is eliminated, giving the phenylhydrazone of
resacetophenonemonobenzoate. We experienced great difficulty in the
determination of carbon in this compound, the results being generally
about two per cent, high, until the method described in the experimental
part of the paper was adopted. A more convenient way to prepare the
same compoimd is to introduce the benzoyl group into resacetophenone
phenylhydrazone, by means of the Schotten-Baumann reaction. Even
though more than two molecules of benzoyl chloride were used, only one
benzoyl group was introduced, since the hydroxyl group adjacent to the
large side chain was protected from the action of the reagent.

Somewhat simikir interference phenomena were noticed in attempting
to prepare the bisphenylhydrazone of resodiacetophenonedimethyl ether.

* Tms Journal, 29, 80.

846 HENRY A. TORRBY AND H. B. KIPPER.

When the proper amount of phenylhydrazine was allowed to stand with
the dimethyl ether in hot or cold alcohol or alcohol and acetic add, only
the imchanged resodiacetophenonedimethyl ether was obtained after
crystallization from dilute alcohoL Other examples of hindrance of
this kind are not wanting; for instance, Baum^ and v. Meyer* have found

/ \"CH8

that ketonic compounds of the type <r yCOCH, do not react with

the phenylhydrazine, and Klinger and W. Kolvenbach^ were unable to
obtain a phenylhydrazone from acetohydroquinone, although its diben-
zoate gave a phenylhydrazone. On the other hand, we obtained the
bisphenylhydrazone from the monomethyl ether of resodiacetophenone
and from dibenzohydroquinone and dibenzoresorcinol.* In attempting
to methykte the hydroxyl groups in the bisphenylhydrazone of reso-
diacetophenone, by acting upon its solution in alcoholic potash with
methyl iodide, merely the unchanged hydrazone or its saponification
products were obtained. Anselmino also fotmd it impossible to methylate
the phenylhydrazones of the homosalicylaldehydes.

In the latter part of this paper we describe the action of metanitroben-
zoyl chloride upon hydroquinone diacetate and resordnol diacetate in
the presence of condensing agents such as anhydrous aluminium chlo-
ride, but we were entirely tmsuccessful in our attempt to introduce the
nitrobenzoyl group into the ring itself, in every case either one or both
of the acetyl groups being replaced by the nitrobenzoyl group. Thus,
from resorcinol diacetate, CeH^COOCCHg),, resordnol m-nitrodibenzoate,
CjH4(OOCCeH4NOj)„ was obtained. This replacement may be due, in
part at least, to the greater volatility of acetyl chloride as compared
with nitrobenzoyl chloride, and might be compared to the action of sul-
phuric acid on sodium chloride. Other analogous organic metathetical
reactions have been observed, as for instance, the displacement of the
acetyl group in acetaniline by the action of benzoyl chloride,* the forma-
tion of ethyl acetate by heating amyl acetate with ethyl alcohol at 240°,
and of amyl benzoate by heating ethylbenzoate with amyl alcohol,*
similar replacements by means of alcoholates,^ the displacement of the
methyl group* by the action of acetyl chloride and altuninium chloride

* Ber., 289 3207.

* /Wd., 29, 835.

^ Ibid,, 31, 1216.

* This Journal, 29, 81.

* Paal and Otten, Ber,, 23, 2587; and Ame Pictet, Ibid,, 23, 301 1.

* Friedel and Crafts, Ann., 133, 208.

' J. Purdie, Ber,, 20, 1554. Gattermann and Ritscfake, Ibid., 23, 1738. Jack-
son and Torrey, Am. Chem. J., 20, 404.

* Claus and Huth, /. pr. Ch. [2], 53, 59.

HYDRAZONBS OF AROMATIC HYDROXYKETONBS. 847

and the replacement of the isobutyryP by the acetyl group, and espe-
cially the formation of phenyl benzoate from phenyl acetate by the action
of benzoyl chloride in the presence of zinc chloride.*

Experimental Part.

4'Monomeihyl Ether of Resacetophenone Phenylhydrazone, — On shaking
a molecule of resacetophenone phenylhydrazone* with slightly more
than one molecule of dimethylsulphate in alkaline solution for eight
hours, a compound was obtained which melted at 108° and was insolu-
ble in ammonia and aqueous alkalies. Analysis showed that it was the
monomethyl ether of resacetophenone phenylhydrazone ; and since it is
identical with the phenylhydrazone of paeonol, the methoxy group must

be in the para position to the side chain:

CH,0—

-C CH,

N . It

I
— OHNH

in.

is soluble in alcohol, benzene, ether, chloroform, acetic add, and alcoholic
sodium or potassium hydroxide. It was purified by crystallization
from dilute alcohoL

Calculated for CiftH|,OaN,: C, 70.31; H, 6.25.

Found: C, 70.03; H, 6.34.

Paeonol,^ an aromatic ether, occurring in the root-bark of Paeonia
Mouian, has been used for medicinal purposes in Japan and China since
the earliest times. W. Nagai* showed that it was the 4-monomethyl
ether of resacetophenone and Tahara* synthesized it from resacetophe-
none and found that the hydroxyl group ortho to the aceto group was
methylated with some diflSculty. In order to compare the product of
the methylation of the phenylhydrazone of resacetophenone with the
phenylhydrazine obtained from paeonol itself, the latter compound was pre-
pared by allowing one molecule of resacetophenone, dissolved in dilute
alkali, to stand for seveml days with one molecule of dimethyl sulphate.
By using a shaking machine, the results were much more rapid and satis-
factory. After acidifying, the paeonol was extracted with benzene, in
which it is much more soluble than is resacetophenone. After evapora-
tion of the solvent, the paeonol was purified by distillation in vacuo;

^ Brauchbar and Kohn, MofuUsh., 19, 27.

* Ddbner, Ann., 2x0, 255.

■ Bull. Soc. Chem, [3], 6, 154.

* Ber., 35, 1292.

* Ibid,, 249 2847.
' Ibid,, 24, 2460.

848 HENRY A. TORREY AND H. B. KIPPBR.

under 30 mm. the paeonol came over at about 210^ and under 5 mm.
at 180°. The crystals which formed on cooling melted at 50-51°. The
phenylhydrazone of paeonol has already been described by F. Tiemann/
but was prepared by us for the sake of comparison and, as has been said,
was identical with the monomethylether of resacetophenone phenyl-
hydrazone described above.

The Phenylhydrazone of Resacetophenonemonobemoate,

CI (2)
.—By allowing the alkaline solu-
H,N,HC,H, (i)

tion of the phenylhydiazone of resacetophenone to stand for several days
at room temperature with more than two molecules of benzoyl chkride,
a compound was obtained which was insoluble in cold aqueous alkaties,
although solution with accompan3dng saponification takes place skwiy.
The insoluble precipitate was collected upon a filter, dried with suction
and upon a clay plate. The benzoate was crystallized from hot alcohol,
there being a considerable difference between the solvent powers of the
hot and the cold solvent. M. p. 181-2°. It is also soluble in benzene and
glacial acetic add ; in aqueous sodium hydroxide it is insoluble, but dis-
solves slowly on boiling. This compotmd was described in an earlier
paper* as the phenylhydrazone of resacetophenone dibenzoate, since the
analytical results pointed to this formula, but our subsequent work upon
it has shown it to be the monobenzoate. Great diifficulty was experienced
in the combustion of this substance, the results for carbon usually being
too high for the monobenzoate. Satisfactory results were finally ob-
tained by proceeding as follows: The substance was mixed with fine
copper oxide, and two reduced copper spirals were placed in the front
part of the combustion tube. Before beginning the burning, a tube con-
taining a reduced copper spiral was attached to the forward end of the
combustion tube and dry air, free from carbon dioxide, was passed over
this hot spiral, until the combustion tube proper was filled with an at-
mosphere rich in nitrogen and containing little oxygen. The special
tube containing the copper spiral was then replaced by the absorption
apparatus and the combustion was begun. The substance was heated
very gradually and, as the burning progressed, oxygen was introduced
into the tube. In two determinations a little oxygen was introduced
into the rear end of the tube early in the combustion.

Calculated for C^HjgOjN,: C, 72.83; H, 5.20; N, 8.09.
Found: C, 74-77, 73-96, 72. 53» 72.921 7238, 72.75;!!, 4.91, 6.08,3.85.6.15,
4.43, 5.41; N, 8.56, 7.84, 8.25, 7-99-

This same compound was also obtained from the resacetophenone diben-

* Ber,, 24, 2854.

* This Journal, 29, 80.

HYDRAZONES OP AROMATIC HYDROXYKBTONBS. 849

zoate^ by the action of phenylhydrazine in hot alcoholic solution, which.
shows that the benzoyl group has replaced the hydrogen of one of the<
hydroxyl groups and not that of the imido group. The removal of one
benzoyl group is in accord with our observations on the action of phenyl-
hydrazine on resodiacetophenone diacetate and dibenzoate, described
below, and it is reasonably certain that the benzoyl group removed is
that nearest the ketone side chain.

The Phenylhydrazone of Resacetopkenonemonoacetate, — One molecule of
resacetophenonemonoacetate,' treated with slightly more than one mole-
cule of phenylhydrazine in hot dilute alcohol containing some acetic
acid, gave a compound melting at 127-8®. It is slowly dissolved in aque-
ous alkalies with the probable saponification of the acetyl radical. It
is soluble in alcohol and less so in ether and chloroform ; insoluble in ben-
zene. It was crystallized for analysis from alcohol.

Calculated for C,eH,eO,N,: N, 9.86.
Found: N, 9.38.

The action of phenylhydrazine on resacetophenonediacetate* as well
as the action of hot acetic anhydride on the phenylhydrazone of resaceto-
phenone, gave a thick, viscous product, which could not be obtained in
a crystalline form.

MonanUroresacetophenone Phenylhydrazone,

(4) HO.

XeH3NOaC.CHaNjHCeH6(i).— The phenylhydrazone of mono-
(2) HCK

nitroresacetophenone was obtained by dissolving the mononitroresaceto-

phenone^ in hot alcohol and adding slightly more than one molecule of

phenylhydrazine. On cooling, beautiful dark red crystals separated out,

which, after crystallization from alcohol and acetic acid, melted at 232-4®

^th decomposition. It is somewhat soluble in hot benzene or toluene

with a brownish color, and in hot alcohol or ichloroform with a reddish

color. It dissolves in glacial acetic add with a reddish color and may be

precipitated by water. As would be expected, it is soluble in alkalies;

the precipitate obtained by neutralizing this solution with hydrochloric

acid at first appears yellowish brown, but with an excess of add rapidly

becomes red in color.

Calculated for C,4Hi,04Na: N, 14.63.

Found: N, 14. 11.

(4) HOv
Resacetophenone Semicarhazone, /CeH,C : CHjN^HCONHjC i ) . —

(2) HO/

* Torrey and Kipper: This Journal, 29, 80.

* J. pr, Chem. [2], 23, 149; Am. Chem., J., 7, 276.
■ Ber., 30, 297.

* 7. pr. Chem, [2], 23, 151.

850 HENRY A. TORRBY AND H. B. KIPPBR.

An aqueous solution containing lather more than one molecule of semi-
carbazide hydrochloride, together with an equivalent quantity of sodium
acetate, was added to one molecule of resacetophenone dissolved in alco-
hoL The mixture was shaken for some time. After filtration the reac-
tion product was dried on a porous plate and crystallized from benzene.
Yellow needles, melting with gradual decomposition at 214-220^1 were
obtained. The semicarbazone is soluble in alcohol, ether, and hot ben-
zene, and is soluble in chloroform. It is soluble in ammonia and aqueous
alkalies.

Calculated for C,H,|0,Ns: N, 20.09.
Found: N, 19.34.

The Preparation of Resodiacetophenone, — ^The method used for making
resodiacetophenone was essentially that employed by Crespieux.^ One
and one-half molecules of anhydrous zinc chloride were dissolved in one
and one-half molecules of glacial acetic add, and to this hot mixture one
molecule of resacetophenone was added, after which the whole was heated
to 140^ in an oil-bath. From a dropping ftmnel one molecule of phos-
phorus oxychloride was run slowly into the mixture, while the heating
at 140° was continued for one-half hour. The hot, viscous product was
then poured into water; the resodiacetophenone separating out was fil-
tered from the soluble zinc salts and crystallized from alcohol It was
foimd that a shorter heating (3-5 minutes) at 140-150° after the addition
of the phosphorus oxychloride was advantageous. While adding the
phosphorus oxychloride care must be taken not to allow the temperature
to rise above 150°, as carbonization is likely to take pkice. A mixture of
alcohol and benzene was found more advantageous than pure alcohol
for the crystallization of the resodiacetophenone, as it was found easier to
separate it from a reddish substance, probably resacet&i, formed at the
same time. When the latter had been formed in considerable amounts,
benzene was used with advantage as a crystallizing medium, as the red-
colored impurity is practically insoluble even in hot benzene. It is de-
sirable, however, to carry out the first crystallization in alcohol By
fusing resordnol diacetate with anhydrous zinc chloride, J. F. Bijkmann'
obtained a diacetodihydroxyphenone, which proves to be identical with
the resodiacetophenone made according to the method given above.
Eijkmann* showed that the same methoxyethoxydiacetophenone was
obtained in each case, whether the methoxyhydroxydiacetophenone
was ethylated or the ethoxyhydroxydiacetophenone was methylated;
further, by oxidation of dimethoxydiacetophenone with potassium per-
manganate, 4,6-dimethoxyisophthalic add, CeH.(COOH),'•3(OCH,)/^

' BuU, Soc. Chim, [3], 6, 152.
' Chem. Centr.f 1904, 1, 1597
' Ibid,, 1905, 1, 814.

HYDRAZONES OF AROMATIC HYDROXYKETONES. 851

was obtained, showing conclusively the constitution of the dihydroxy-
diacetophenone to be C«H,(COCHs),'-5(OH)3"^

The identity of resodiacetophenone, made as given above, with this
dihydroxydiacetophenone, was shown by making the mono- and dimethyl
ethers by the action of methyl iodide on an alcoholic solution of the potas-
sium salt of resodiacetophenone. These melting-points were found to be
120° and 170**, respectively, the same as given by Bijkmann. Further,
the phenylhydrazones of both compounds were fotmd to be identical in
properties.

Salts of Resodiacetophenone, — Resodiacetophenone, when dissolved in
that quantity of sodiiun hydroxide which gave two molecules of the lat-
ter to one of the former, and evaporated to dr>Tiess on the steam-bath,
gave the white sodium salt. If lead acetate was added to some of the
above solution, a pinkish precipitate of the lead salt was obtained. The
addition of silver nitrate gave a white precipitate which blackened rapidly
in the air.

Resodiacetophenone Bis phenylhydrazones 2y4-Dihydroxy-i,^-Diacetophe-
none Bisphenylhydrazone, CeH2(OH)2(C . CHjNjHCeHg),. — ^This substance
was discovered by Crespieux* and has been described by us also in
a recent paper. ^ It is best prepared by allowing considerably more than
two molecules of phenylhydrazine to stand for several hours with an alco-
holic solution of resodiacetophenone. The substance crystallized from
aniline, diethyloxalate, acetone and alcohol, all gave the same melting-
point, namely 291®. This is 60® higher than that given by Crespieux.
Work which is now being carried on in the laboratory by Mr. R. D. Bell
and one of us has shown that varying quantities of resodiacetophenone
monophenylhydrazone are formed together with the bisphenylhydra-
zone, even when a large excess of phenylhydrazine is used and the
mixture is allowed to stand for some time. Such mixtures melt at
about 230**, but after treatment with a soditun hydroxide solu-
tion, which dissolves out the m.onophenylhydrazone without having
the slightest effect upon the bisphenylhydrazone, the melting-point is
immediately raised. That it is the resodiacetophenone monophenyl-
hydrazone which is dissolved out by the alkali was shown by precipita-
ting the filtrate with acid and analyzing the product after crystallization
from diethyl oxalate. The insolubility of resodiacetophenone bisphenyl-
hydrazone in even the most concentrated aqueous alkalies is very marked.
It is, however, easily soluble in alcoholic sodium or potassium hydroxide,
giving a yellow solution, from which the original compound, together with
small quantities of a decomposition product, nam.ely, the monophenyl-
hydrazone, m.ay be precipitated by mineral adds. If acetic acid was

" Bull. Soc. Chitn. [3], 6, 152.
* This Journal, 29, 80.

852 HENRY A. TORRE Y AND H. B. KIPPER.

used, instead of hydrochloric, the precipitate, as it stood in the liquid,
appeared ahrost white. The nrodification obtained by crystallization
from aniline had a rather deep orange color.

The Monophenylhydrazone of Resodiaceiophenone, 2,4'Dihydroxy-i,,^'
Diacetophenone Monophenylhydrazone, C;,H2(OH)jCOCH8.CCH,NNHC,H,.
— ^This compound was obtained exactly as the bisphenylhydrazone, ex-
cept that only a little more than one molecule of phenylhydrazine to one
of resodiacetophenone was employed. The light yellow crystals which
separated out were collected on a filter, washed, and then dissolved in
10 per cent, sodium hydroxide, to free them from the small amount of
bisphenylhydrazone formed, which is insoluble in aqueous alkalies. After
filtration the hydrazone was precipitated with hydrochloric acid. The
precipitate was washed with hot water and finally with alcohol, dried on a
porous plate and crystallized from brombenzene, in which the mono-
hydrazone is very soluble when hot and but slightly so in the cold. Light
yellow needles, melting with decomposition at 233**, were obtained.
By crystallization from hot acetone, in which the solubility is also high,
yellow, tetragonal plates were formed, likewise melting at 233®.

Calculated for C,eH,eO,N,: N, 9.86.
Found: N, 9.58.

This monohydrazone is soluble in ammonia and aqueous alkalies with
a yellow color. It is soluble in acetic acid, hot and cold, and precipita-
ted from this solution by water. It is slightly soluble in hot or cold alco-
hol, benzene, or chloroform, and readily soluble in hot acetone and ethyl
acetate.

Bisphenylhydrazone of the Monomethylether of Resodiacetophenone,

( I ) CeH,HN,CH, :C. .C : CHaN^HCeHs (.5)

/C^Hj^^^ .—One molecule of the mono-

(2) HO^ X)CH3 (4)

methyl ether of resodiacetophenone, dissolved in hot alcohol, was heated

on a steam-bath for a short time with slightly more than two molecxiles

of phenylhydrazine. A light yellow precipitate began to appear very

soon, and after cooling and allowing to stand for a few hours, an almost

quantitative yield of the hydrazone was obtained. It was crystallized

from benzene, in which it is quite soluble when hot. Long, transparent

needles Were obtained, melting at 245-6®, with decomposition, which were

insoluble in aqueous alkalies.

Calculated for CaH^OaN^: N, 14.43.
Found: N, 14.18.

When the dimethyl ether of resodiacetophenone was treated with

phenylhydrazine in hot or cold alcohol or in alcohol and acetic add, no

action took place, as the product, after crystallization from dilute alcohol,

proved to be merely tmchanged resodiacetophenone dimethyl ether.

HYDRAZONES OF AROMATIC HYDROXYKETONKS. 853

Resadiaceiopkenane Diaceiaie (i,3'Diacetophenone Dtacetat€'2,4),

(1) CH3COV .OCCH, (5)

y>C,H,<^ * — Five grams of resodiacetophenone

(2) CH,COO/ \OOCCH3 (4)

were dissolved in about 30 grams of acetic anhydride and boiled for two
hours under a reflux condenser. The solution was allowed to cool and fil-
tered from the precipitate which separated. After drying and crystalliza-
tion from alcohol, colorless needles, melting at 120^, were obtained, while
from benzene the compound crystallized in transparent hexagonal plates.
It is readily soluble in chloroform and in hot alcohol or benzene, and is
far more soluble in cold benzene than is resodiacetophenone.

Calculated for C,4Hj40«: €,60.43; H, 5.04.
Found: C, 60.15; H, 5.56.

Resodiacetophenone Dibenzoate {i ^fy-Diacetophenone Dibenzoaie),

(1) CHjCO-^^^ .OCCH3 (5)

^C^H,^^ . — It was found possible to prepare the

(2) CeH,COCK X)OCC,Hj (4)

dibenzoate by heating resodiacetophenone with benzoyl chloride at 170-
180®, but much better results were obtained by the following method,
which gave an almost quantitative yield: Five grams of resodiaceto-
phenone were dissolved in thirty grams of warm pyridine, the solution
was then cooled in ice-water, at which temperature some resodiaceto-
phenone separated, and a little more than two molecules of benzoyl chlo-
ride were added in small portions with constant shaking. The mixture
was then allowed to come to room temperature and finally heated
just enough to make the solution complete. After allowing to cool
again, first water and then a solution of sodium carbonate were added
and the dibenzoate was washed with ammonia. Drying and crystalliza-
tion from alcohol gave a constant melting-point of 118-9°. It is soluble
in benzene, alcohol and chloroform, but more so in the hot than in the
cold solvent ; it is also soluble in hot and cold glacial acetic acid and very
soluble in hot toluene. It is readily saponified by boiling with a ten per
cent, sodium hydroxide solution.

Calculated for CjjHjgOj: C, 7 1 . 64; H, 4 . 48.
Found: C, 71.48; H, 5.25.

Monophenylhydrazone of Resodiacetophenone Monoacetate,

(5) CH,CO. .C : CHaN.HCeH^ ( i )

^C8H2<^ . — Two molecules of phenylhy-

(4) CH3COO/ X)H (2)

drazine were allowed to act on one of resodiacetophenone diacetate dissolved
in a mixture of hot acetic acid and alcohol. On cooling, a h^vy white
precipitate fell, which after filtration, drying on a porous plate, and crys-
tallization from benzene, was obtained in the form of transparent plates.

854 HENRY A. TORREY AND H. B. KIPPER.

melting, when pure, at 191-2°; from alcohol it crystallized in white
needles, giving the same melting-point. In the mother liquor, from the
preparation of this compound, acetylphenylhydrazine,* C^HgNHNHCOCH,,
was isolated, proving that one at least of the acetyl groups must have
split out. The interference of these groups is shown in an even more
marked degree by the action of four molecules of phenylhydrazine, when
both acetyl groups are removed and the bisphenylhydrazine of resodiaceto-
phenone is formed. The monophenylhydrazone of resodiacetophenone
monoacetate is rather soluble in hot alcohol and benzene, but only slightly
so in these solvents when cold ; very soluble in hot toluene ; soluble in gla-
cial acetic acid, and precipitated with water. It dissolves slowly in cold
sodium hydroxide, rapidly on boiling, with decomposition. By acidi-
fication of the solution from the action of either hot or cold hydroxide,
resodiacetophenone monophenylhydrazone, m. p. 233**, was obtained.
Since the solution has the odor of phenylhydrazine, it is evident that a
small amount of this group is also split out.

Calculated for C,,H„04N,: N, 9. 59.
Found: N, 8.70, 9.15.

Monophenylhydrazone of Resodiacetophenone Monobenzoaie,

(5) CH,COv .C:CH,N,HC.H, (i)

)>CeH,<^ . — ^The action of phenylhydra-

(4) CeH.COCK NdH (2)

zine on the dibenzoate of resodiacetophenone is exactly analogous to its
action on the diacetate described above. Two molecules of phenylhydra-
zine react with one molecule of the dibenzoate in hot alcoholic solution,
and on cooling a white compound separates, which after crystallization
from alcohol finally gives colorless, transparent prisms, vrith constant
melting-point of 2 14-5 **. The compound is soluble in hot alcohol, benzene and
chloroform, but less so in the cold. It may be precipitated from its acetic
acid solution by water. It is insoluble in ammonia and in ten per cent,
sodium hydroxide, the solvent action is very slow in the cold, but rather
rapid on boiling, when doubtless saponification takes place. It will be
seen that in the formation of this hydrazone, one benzoyl group has been
removed by the action of the phenylhydrazine.

Calculated for C„H,o04N,: N, 7. 22.
Found: N, 7 . 23, 7 . 49.

Resodiacetophenone Bisparahromphenylhydrazone^

( 1 ) C,H,BrHN, : CHCv /CCH3 : N^HC^H^Br (5)

^CeH,<; . — An alcoholic solu-

(2) HQ/ X)H (4)

tion of two molecules of />-bromphenylhydrazine and one of resodiaceto-
phenone was boiled on the steam-bath under a reflux condenser for one

* Ann.f 190, 129.

HYDRAZONieS OF AROMATIC HYDROXYKETONES. 855

hour. The light yellow needles thus obtained were purified by crystal-
lization from acetone, in which the compound is fairly soluble in the hot
and less so in the cold. It melted at 270-1 ^ with decomposition. It is
readily soluble in hot nitrobenzene, aniline and monobrombenzene, but
slightly so in these solvents when cold; insoluble in alcohol, benzene,
chloroform, toluene, ligroin, and amyl alcohol; slightly soluble in hot
acetic acid and ethyl acetate. It is insoluble in ammonia and aqueous
alkalies, but soluble in alcoholic sodium hydroxide; boiling with very
strong potassium hydroxide (i : i) causes decomposition.

Calculated for CaH^OjN^Brgi Br, 30.07.
Found: Br, 29.88.

Addition and Substitution Products Obtained from the Interaction of
Resodiacetophenone and Benzidine, — One molecule of resodiacetophenone
was dissolved in hot alcohol and to this a hot alcoholic solution of two
molecules of benzidine was added and the mixture was heated for about
one hour on the steam-bath under a reflux condenser. A precipitate of
yellow crystals was obtained which was filtered while the liquid was
still hot and was washed with hot alcohol. This product, which melted
above 300 ^^ was insoluble in the ordinary solvents, but soluble in hot ani-
line, monobrombenzene or nitrobenzene, the last two solvents causing a
slight browning; it is not decomposed by ammonitun hydroxide, but hot
acetic acid causes gradual decomposition and boiling with a ten per cent,
sodium hydroxide solution decomposes it slowly and concentrated hy-
drochloric acid causes a fairly rapid hydrolysis. From aniline, light
yellow needles, melting above 300®, were obtained, which proved, on
analysis, to be a condensation product of benzidine and resodiacetophe-
none in the proportion of one molecule of each. Because of the marked
insolubility in all the ordinary solvents, no attempt was made to deter-
mine the molecular weight. Since in condensation of benzidine with alde-
hydes and ketones both amine groups commonly enter into the reaction
and from analogy to the azine derivative described below, it appears most
likely that two molecules of benzidine have reacted with two molecules
of resodiacetophenone, giving a compound of the following formula:

CH,-C-CeH,(OH),-C— CH,

N N

C.H«

N N

CH,— C-C,H,(OH),-C— CH,

856 HENRY A. TORREY AND H. B. KIPPER.

Calculated for C^Hzfi^^^: C, 77. 19; H, 5. 26; N, 8. 19.
Foimd: C, 77-3i; H, 5.82; N, 8.22.

From the alcoholic filtrate from the preparation of the above benzi-
dine derivative, yellow needles separated on cooling, which were washed
with cold alcohol These melted slowly between 182** and 185° with de-
composition; a second and a third rapid crystallization from alcohol gave
the same melting-point, as well as crystallization from benzene. Since
the two reacting substances, benzidine and resodiacetophenone, are both
colorless, and the product obtained consisted of beautiful, well-shaped,
yellow crystals, it is not possible that it is simply a mixture of these two
substances. It is readily decomposed on standing with aqueous ammonia
or acetic add, with the former benzidine being precipitated and reso-
diacetophenone passing into solution, while with the latter resodiaceto-
phenone is precipitated and benzidine dissolved. On boiling the alco-
holic or benzidine solution of the low-melting derivative for several hours
under a reflux condenser, a yellow precipitate is formed, having the proper-
ties of the high-melting condensation product. The properties and
the analysis of the soluble low-melting compoimd indicate that it is an
addition product of one molecule of benzidine with four molecules of
resodiacetophenone.

Calculated for Ci,Hi^^4CioH,o04: N, 2.92.
Found: N, 2.36, 2.64.

The Mono- and bis-Azine Derivatives of Resodiacetophenone and an
Aniline Condensation ProdtLct Obtained from the Former. — Considerably
in excess of two molecules of hydrazine hydrochloride were allowed to
act on one molecule of resodiacetophenone on the steam-bath tmdera
reflux condenser in a i : i aqueous alcoholic solution to which sodium
acetate in an amount equivalent to the hydrochloride had been added. The
heavy, yellow precipitate which formed was collected on a filter and washed
first with hot water and then with alcohol and dried by suction and plating.
This derivative, although soluble in alkalies, is highly insoluble in the
common low-boiling organic solvents, but is soluble in hot nitrobenzene
or aniline ; with the latter, however, a further condensation takes place,
as described below. Purification by dissolving in ten per cent, soditmi
hydroxide and precipitation by hydrochloric acid gave a compound
which does not melt imder 300 ^^ but which changed from yellow to Kght
red between 240*^ and 260*^, and back again to the yellow modification on
cooling. Analysis shows that it is monazine of resodiacetophenone, in
which one molecule of hydrazine has condensed with two molecules of
the ketone :

HYDRAZONHS OP AROMATIC HYDROXYKBTONES.

857

CH,— C— CeH,(OH),— C--CH3

II II

N O

II
CH,— O-C,H,(0H),-C— CH3

Calculated for CtoH^OflN,: N, 7. 29.
Found: N, 7.07.

By carrying out the reaction between hydrazine hydrochloride and
resodiacetophenone with a solution of sodium hydroxide instead of so-
dium acetate, complete condensation between two molecules of hydrazine
and two molecules of resodiacetophenone takes place. When sodium
acetate is used, the marked insolubility of the first condensation product
probably prevents any further reaction. In the reaction in alkaline
solution a ten per cent, aqueous sodium hydroxide was used at the tem-
perature of the steam-bath, the alkali-soluble azine compound was
precipitated out with hydrochloric add and dried by plating and suction.
The solubilities were found to be about the same as for the monoazine
derivative. At 150® the red modification begins to appear, and is much
deeper in hue, A small amoimt kept between this temperature and
300® for about twenty minutes, and then allowed to cool, returns again
to the yellow form. On heating in a mixture of acetic and hydrochloric
adds, both azine derivatives are decomposed; resodiacetophenone was
recovered from the solutions by diluting with water. Analyses of the
compound gave results which correspond with the following formula, a
bisazine of resodiacetophenone:

CH,— C-

NHO—

<:^CH,

NHO—

CH,

H
H

-OHN

OHN

C— CH,

Calculated for C^^^O^^^: N, 14. 73.
Found: N, 14.48, 14.43.

On heating the monazine with aniline, further condensation takes

place, viz., between two molecules of aniline and the two free carbonyl

groups, giving a dianilido monazine of resodiacetophenone, which is

insoluble in aqueous alkalies.

858

HBNKY A. TORRBY AND H. B. KIPPBR.

CH,— C-

NHO—

-OH N

■C— CH,

CH,

NHO—

OH N

II
— C— CH,

Calculated for C„H,o04N4: N, 10.48; C, 71.91; H, 5.62.

Found: N, 10.97, 10.65, 10.66; C,- 71.74: £[,4.54.

Action of Bromine on Resodiacetophenone. — ^The direct action of bro-
mine on acetophenone* gives the a>-bromaceto derivatives, while in the
action of bromine on paeonol acetate in carbon disulphide, some of the
bromine goes into the side chain, giving exo-brompaeonol acetate,
CeHa(OCH,)aOCOCH8COCHjBr, and at the same time esobrompaeonol,
OHCaHjBrOCHjCOCHa, is formed by the action of bromine on the paeonol
formed by saponification of the acetate by hydrobromic add.' In study-
ing the action of bromine on resodiacetophenone, we fomid di£Sciilty
in obtaining satisfactory results, since by no variation of the method of
preparation or purification could a compound of sharp melting-point be
obtained. We have, however, obtained a monobrom derivative in a fair
degree of purity and this seems pretty certain to be a ring and not a
side-chain substitution product. One molecule of bromine was added
to one of resodiacetophenone and the mixture was allowed to stand for
six or eight hours with occasional stirring until hydrobromic add ceased
to be evolved. A white, crystalline product was obtained which, after
numerous crystallizations from chloroform and alcohol, failed to give a
sharp melting-point, gmdual liquefaction taking place between 197-
202 *'. Analysis, however, gave a satisfactory result for a monobrom
derivative.

Calculated for CioH,04Br : Br, 29.31.
Found: Br, 29.38.

Some of the product was allowed to stand with a ten per cent, sodium
hydroxide solution for several hours and was finally throvra out of solu-
tion with hydrochloric add. The predpitate had a light yellow color,
which crystallization from chloroform did not change and the melting-
point was still not sharp, 200-3**. Analysis showed that practically

* Ber.f 10, 2006.
» Ibid., 30, 299.

HYDRAZONES OF AROMATIC HYDROXYKETONES. 859

-no bromine had been removed by this treatment, which points to sub-
stitution in the ring rather than in the side chain. Found, 28.80 per
cent, bromine. When the bromination was carried on in chloroform or
when several molecules of bromine were used, the products contained
large quantities of bromine, much of which could be removed by treat-
ment with alkalies. The products obtained by bromination had a very
strong action on the mucous membrane. On treatment with potassium
hydroxide, this action was entirely destroyed.

Bromresodiacetophenone Bisphenylhydrazone,
•CeHBr(OH)j(C:CHsN2HC8H5)j,.— About two molecules of phenylhydra-
zine were added to one molecule of the monobrom resodiacetophenone
•dissolved in hot alcohol. On standing for a few hours, the bisphenyl-
hydrazone separated out in beautiful light yellow crystals, slightly solu-
ble in hot benzene and toluene and almost insoluble in alcohol and ether.
Crystallized from chloroform, the compound decomposes at 215-20°. It is
insoluble in ten per cent, sodium hydroxide in the cold; on heating it dis-
solves with saponification of a phenylhydrazine group, made evident by the
-odor of free phenylhydrazine. The phenylhydrazone of the bromresodi-
acetophenone which had been treated with alkali was also made and the
same light yellow product was obtained.

Calculated for CjjHaOjN^Br: Br, 17.66.
Found: Br, 17.86.

In connection with our work on the dihydroxybenzophenones, we at-
tempted to obtain nitro derivatives by the action of acid chlorides of the
nitrobenzoic acids on the acetate, benzoate, nitrobenzoate, and methyl
ether of hydroquinone in the presence of various catalyzing agents, as the
chlorides of aluminum, zinc or copper. In no case, however, did we ob-
tain benzophenones, but with the acetates found that one or more of the
acetyl groups had been replaced by nitrobenzoyl groups. Thus the ac-
tion of two molecules of nitrobenzoyl chloride on one molecule of hydro-
quinone diacetate in the presence of anhydrous aluminium chloride at
130-140° gave, besides a small amount of unchanged nitrobenzoyl chlo-
ride, two compounds which were readily separated from each other by
their different solubilities. After allowing the reaction to proceed for
about two hours with occasional additions of aluminium chloride, the re-
action product was treated with very dilute hydrochloric acid and finally
-with a little ammonia to free it from nitrobenzoic add. After drying on
a porous plate, the portion soluble in alcohol was extracted and crystal-
lized from this solvent, in which it is quite soluble when hot and somewhat
less so in the cold; its solubility in benzene is similar. The melting-point
"was 113®. It is insoluble in cold aqueous alkalies, but dissolves rather
rapidly on heating. Analysis shows that the compound is hydroquinone

86o HENRY A. TORRBY AND H. B. KIPPER.

monoacetatemononitrobenzoate) one acetyl derivative having been re-
placed by the nitrobenzoyl group,

/OOCH3 (i)

C.H,<(^

00CCeH4N0a (4)

Calculated for C,4H„OeN: C, 59 80; H, 3.65; N, 4.65.
Found: C, 59.18; H, 4.02; N, 4.63.

That portion of the product insoluble in alcohol was foimd to be in-
soluble also in most of the common organic solvents. It was found pos-
sible, however, to crystallize it from hot acetic anhydride, from which
it is deposited in small white needles. It is very soluble in hot nitro-
benzene and but slightly so in the cold solvent; it is somewhat less solu-
ble in hot diethyl oxalate than in hot nitrobenzene. The melting-point
was 268*^. Condensation of two molecules of w-nitrobenzoyl chtoride
or the free acid and one of hydroquinone in the presence of phosphonis
oxychloride gave the same compound. It is slowly saponified by alco-
holic potash, giving some m-nitrobenzoic acid. The method of prepara-
tion, the saponification product and the analyses show the compoimd
to be hydroquinone-m-nitrodibenzoate,

X)OCC,H,NO, (i)

c,h/

X)OCC,H,NO, (4)

Calculated for CmH„OsNs: C, 58.82; H, 2.94; N»6.86.
Fotmd: C, 58.63; H, 4.19; N, 6.62.

Resorcinol m-nitrodibenzoate was formed in a similar attempt to pre-
pare the benzophenone, and likewise by the direct action of ff^•nit^)ben-
zoyl chloride on resorcinol. In this case mere heating on the steam-
bath for a few hours gave a completed reaction, while with hydroquinonol
a temperature of 130° was required. The solubilities are similar to those
of the analogous hydroquinonol derivative. It was crystallized from
nitrobenzene, m. p. 172°. Analysis gave results for resorcinol m-nitro-
dibenzoate,

.OOCC^H^NO, (i)

X)OCC,H,NO, (3)

Calculated for C,oHi,OgN,: C, 58.82; H, 2.94; N, 6.86.
Found: C, 58.47; H, 3.27; N, 6.69.

The action of w-nitrobenzoyl chloride on resorcinol dibenzoate in a
number of different solvents such as carbon disulphide, phosphorus tri-
chloride and phosphorus oxychloride was tried without result. Ortho-
nitrobenzoyl chloride on resorcinol dibenzoate in the presence of anhy-
drous zinc chloride and infusorial earth gave entirely unsatisfactory re-
sults, while 0' and w-nitrobenzoyl chlorides on resorcinol or hydroqui-

HYDRAZONES OF AROMATIC HYDROXYKETONES. 86 1

nonol fn-nitrobenzoate in the presence of various catalyzers, as anhydrous
zinc and aluminum chlorides and metallic copper gave no reaction.
The dimethyl ether of resorcinol is a derivative in which the substituting
groups are comparatively stable and should offer little interference;
with o-nitrobenzoyl chloride and zinc or aluminum chlorides, however,
the add chloride was decomposed and with w-nitrobenzoyl chloride,
although hydrochloric acid was given off, it was found that the reaction
consisted principally in the replacement of the methyl by the nitroben-
zoyl groups. This reaction is in line with the observation of A. Claus^
that acetyl chloride in the presence of zinc chloride tends to remove
the methyl groups in the methyl ether of resorcinol. Since an excellent
method for preparing resodiacetophenone is the isomerization of resor-
cinoldiacetate by means of zinc chloride, an attempt was made to iso-
merize the resorcinol- w-nitrobenzoate ; with anhydrous zinc chloride
at 130-140° there was no change, and with mixtures of zinc chloride
and aluminum chloride or with alumintun chloride alone, carbonization
occurred before the desired reaction took place.

Since acetaldehyde' and benzaldehyde,* acting on quinone, give, be-
side the acetate and benzoate and quinhydrone, considerable quantities
of the dihydroxy aceto- or benzophenone, it was thought possible that
the nitrobenzaldehydes might react in a similar way upon quinone, yield-
ing dibydrox3aiitrobenzophenones:

O OH

H-^^H H^ ^^C— CeH,NO,

+ 0,NC.H,CHO =
Hv/H Hv yHO

II
O

Accordingly, quinone was dissolved in melted o-nitrobenzaldehyde
and allowed to stand in the stmlight for several weeks, but no condensa-
tion took place, the nitrobenzaldehyde having been isomerized simply
to nitrosobenzoic acid, a reaction which G. Ciamidan and P. Silber*
have shown takes place with the former substance under the influence
of sunlight. Salicylaldehyde, in the sunlight, acted as a reducing agent
on quinone, giving quinhydrone and salicylic add. The study of alkali-
insoluble phenols is being continued in this laboratory.

* /. pr, Chem. [2], 53, 39.
' Ber,, 31, 12 14.

^ Ibid., 24, 1 341.

* Ibid., 34, 2040.

CAMBRIDOB, MA88.

362 HBNKY A. TORREY AND C. M. BRBWSTBR.

THE ACnOir OF PHTHAUC AITHYDSIDE OB RESACETOPHEROHE.

By Hbnrt a. Torrbt and C. M. Brbwstbr.
Received March 13, 1908.

In the preceding paper by H. B. Kipper and one of us we have called
attention to the replacement of the acetyl group by the nitrobenzoyl group
in resorcinol diacetate and hydroquinonol diacetate. In the present ar-
ticle we wish to record a more remarkable reaction in which the acetyl group

in resacetophenone, HO — <f y — COCH3, a compound in which that

OH
radical is directly attached to the ring, has been replaced by a group of
higher molecular weight. By heating two molecules of resacetophenone and
one molecule of phthalic anhydride with a dehydrating agent, such as an-
hydrous zinc chloride, sulphuric acid, or phosphoric acid, then acetic acid is
given off and fluorescein is formed in large amotmt. It has been known
for a long time that acetophenone, heated with concentrated sulphuric
add, gives benzoic acid and benzenesulphonic acid^ and Hoogewerf and
van Dorp' have shown that in substituted aromatic methyl ketones
carrying a methyl group in the ortho position to the acetyl group, sul-
phuric acid has the power of replacing the ketone mdical by hydrogen,
while Klages' had found somewhat earlier that although the aromatic

ketones of the type <^ y — COR were unaffected by heating with phos-
phoric add, those ketones in which an alkyl group is in the ortho posi-
tion may be changed into the corresponding hydrocarbon; the decom-
position takes place most easily with diortho alkyl derivatives. The
formation of fluorescein from resacetophenone and phthalic anhydride
does not seem to be due to the formation of resorcinol by the splitting
out of the acetyl group which then reacts with the phthaUc anhydride,
for when resacetophenone was heated with concentrated sulphuric
acid no odor of acetic acid was noticed.

The procedure in this method for preparing fluorescein was as follows:
The mixture of 20 grams of resacetophenone and 10 grams of phthalic
anhydride, after slightly moistening with water and adding 5 grams
of concentrated sulphuric acid, was heated with constant stirring until
melted. The rate of condensation was watched by dissolving small por-
tions in soditun hydroxide and observing the color and quality of the
predpitate obtained on acidification. When a large yield of a bright

' Krekeler, Ber,, 19, 678.

' Koninklijke Akad. van Wetenschappen te Amsterdam, 1901, 173. CA«m. Cemir^
7a [2], 1117.

• Ber., 32, 1549.

OPTICAL ROTATION OF SPIRITS OF TURPENTINE. 863

yellow precipitate was secured, the reaction was stopped and the mix-
ture poured into a large quantity of cold water, when the fluorescein
separates out in a bright yellow, flocculent mass; if a dark precipitate is
obtained, the mixture has been heated too long. The fluorescein may
then be purified by fractional precipitation from an alkaline solution
with add, which gradually removes a tarry impurity probably contain-
ing resaceteln. Another method of purification consists in dissolving
the crude fluorescein in concentrated sulphuric acid diluted with an equal
volume of water and precipitating after filtration by neutralization
with sodium hydroxide. That acetic acid is actually given off in the re-
action was shown by the odor, by the ferric chloride and the ethyl ace-
tate tests. That the compound formed was actually fluorescein was
shown by conversion into the diacetate, the dichloride and eosine. The
diacetate made either by heating with acetic anhydride or by acetyla-
tion with acetyl chloride and pyridine in glacial acetic acid, after puri-
fication by crystallization from acetone, melted at 201®, while Baeyer*
gives 200®.

Calculated for C,|H,,Or: C, 69. 23; H, 3. 88.

Pound: C, 69.04, 69.24; H, 4.03, 4.09.

From this diacetate fluorescein was obtained by saponification. The
dichloride was made by heating with phosphorus pentachloride at 180^.
After crystallization from alcohol it gave a melting-point of 249® (Baeyer,
252®). Calculated for C^fi^PJ^h' CI, 19.22; Found, 19.20.

Eosine was made by adding bromine to an alcoholic solution of the
fluorescein. Calculated for C^gOgBr: Br, 49.36; Found, 48.76. Suc-
cinyl fluorescein was formed with elimination of acetic acid when the
condensation was carried on with succinic acid instead of phthalic anhy-
dride.

THE OPTICAL ROTATION OF SPIRITS OF TURPEHTINE.

BT CHA8. H. Hbrtt.
Received February xo, 1908.

Among the physical properties of spirits of turpentine, none has proved
of more interest than its optical rotation. In most specimens this prop-
erty is very marked, and as the liquid is colorless and the determination
readily made, many data are fotmd on this subject in chemical literature.
SHght variations in the rotation of different samples are to be expected,
as spirits of turpentine is not a chemical compound but a mixture of
substances, chiefly terpenes. From the results of ntunerous observa-
tions upon commercial samples, the view commonly held previous to
1 89 1 was that French spirits of turpentine, distilled from the oleoresin
of Pinus marUima, is levo-rotatory and that American spirits of turpen-
* Ann., 183, I.

^64 CHAS. H. HBRTY.

tine, distilled in years past, almost wholly from Pinus palristris, is dextro-
rotatory. The difference in the character of the rotation was ascribed,
therefore, to the different species from which the spirits of turpentine
was produced.

Recognizing the fact that American spirits of turpentine is distilled
from more than one species of pine, J. H. Long,^ in 1891, undertook a
study of the volatile oils distilled from oleoresins of well identified indi-
vidual trees, these trees embracing the several species of pines subjected
to turpentining in our southern states. He found that spedmeos from
Pintis pcdustris (Long Leaf Pine )gave dextro-rotatory oils, while those
from Finns heterophylla (Cuban or Slash Pine) gave levo-rotatory oils.
Since the oleoresins from these two species are indiscriminately mixed,
at the time of collection in the woods, the rotation of the oil distilled from
such a mixture would naturally vary. Pinus palusiris is the predomi-
nating species and Long attributed to this fact the dextro-rotatory character
of American spirits of turpentine. This view has been generally accepted.

The fact that spirits of turpentine is frequently adulterated with opti-
•cally inactive mineral oil, led A. McGilP to make a large number of de-
terminations of the rotation of commercial samples of spirits of turpen-
tine, in the hope of utilizing this property for the detection of adultera-
tion. From the widely varying results obtained he was compelled to de-
■clare the method useless.

New evidence upon this point has been obtained from investigations
■carried on in this laboratory in collaboration with the U. S. Forest Ser-
vice, the experimental work having been carried out by Messrs. George
A. Johnston and W. S. Dickson under the direction of the writer. In
•order to gain a better knowledge of the oleoresins from the two principal
species of pine utilized in the turpentine industry at the present time,
fourteen trees were selected on a Florida turpentine farm. One-half
of these were Pinus palustris, the other half Pinus heterophylla. Three
trees of each species were tapped for the first* time at the beginning of
the experiments. In each case a small, young pine, a medium pine, and
a large, old pine were selected. In another set four trees were selected,
two each PinvrS palusiris and Pinv^ heterophylla. These trees had been
subjected to turpentining during the previous year, the chipping, or
weekly scarification, on all of them having been unusually shallow, only
about one-half as deep as is commonly practiced. In a third set four
trees were selected, two each of Pinus palusiris and Pinus heterophylla,
which had been turpentined during the previous year, and on each of
these the depth of the chipping was the normal cut. The trees in each
5et were chipped at intervals of seven days.

* /. Anal. Appl. Chem., 6, i.

' Bulletin No, 79, Inland Revenue Dept., Canada,

OPTICAL ROTATION OF SPIRITS OF TURPENTINE. 865

Special precautions were taken in the collection of the oleoresins. The
cup and gutter system described in BtUleHn No. 40, [/. S. Bureau of For-
estry, was used. Instead of the clay cup commonly used, oyster pails
were substituted. The entire apparatus was covered with black oil-
cloth fastened securely into the bark of the tree above the chipping sur-
face, thereby protecting the resin from light and avoiding the filling of
the pails with rain water. Every four weeks these pails were removed
from the tree, tightly stoppered and immediately shipped to this labora-
tory for examination. The specimens so obtained were extremely pure
and free from chips. After removal of the pails, the metal gutters were
raised to a point near the chipping surface in order to minimize the amount
of oleoresin which might stick to the exposed portion of the trunk above
the gutters.

The distillation of the oleoresin was carried out in a 500 cc. Kjeldahl
flask, surrounded by a bath of cottonseed oil. Steam from a small
boiler was first passed through a small iron pipe in which it could be
superheated, then into the distillation flask through a glass tube having
on its end a bulb containing a number of openings. By this means
strong agitation of the molten oleoresin was obtained. Thermometers
were placed both inside the flask and in the oil-bath. The mixed vapors
of steam and spirits of turpentine were passed through a Hopkins con-
densing bulb to prevent the carrying over of solid particles of resin, con-
densed in an ordinary Liebig condenser and collected in a separatory
ftuinel. After drawing off the lower layer of water, the spirits of turpen-
tine was transferred to a dry flask and allowed to stand over night with
calcium chloride. The determinations of the optical rotation of the vola-
tile oils were made with a Schmidt and Haensch half -shadow polariscope,
sodium flame, at 20^.

In the following table are given the results from the first collection
of the oleoresin in early spring:

Tabls I.

Optical rotation
Tree Diameter 100 mm. tube,

designation. Species. (inches). Character of chipping. 3o<>C.

Ai P. heterophyUa 7.0 ist year, normal depth — 2o**5o'

A2 " 14.5 '* " + 0^15'

A3 " 24.5 " " — i5**o'

A4 P.palustris 7.3 " *' +15*^40'

A5 " 15.0 " " -h 8«» 9'

A6 " 21.0 " " +i8®i8'

Ci P. helerophylla 12.3 2nd year, shallow — 27^11'

C2 " • 8.2 " " — 26^28'

C3 P.palustris 13.0 ** " — 7<»26'

C4 " 8.7 " " + 7'*3i'

Di " 9.0 2nd year, normal depth +io*>5o'

D2 " ' 13.5 *' " + I**23'

D3 P, heterophyUa 13.0 " " — 18®35'

D4 " 90 *' '* — 29^26'

866 CHAS. H. HBRTY.

These results show a wide variation in the optical rotation of the vola-
tile oils from the individual trees, even among trees of the same specks.
In a general way the figures give support to Long's view, namely that
the volatile oils from the Pinus palustris are dextro-rotatory and those
from Pinus heterophylla levo-rotatory. That this is not strictly tme,
however, is evidenced by the dextro-rotation of A2 (P. heiero-
phylla) and more especially by the levo-rotation of C3 (P. palustris).

With these variations in the first collection from the several trees,
the question naturally arose, would the variations change as the season
advanced or would the figures prove constant for the individual trees?
The rotations for the successive collections follow in Table II:

Tablb II. — Opticai* Rotation xn 100 mm. Tube, 20® C.

Collection. Ai. Aa. A3. A4. A5. A6. Ci.

I.. — 20**5o' +o*'i5' — 15** o' +i5°4o' +8^9' -hi8«»i8' — 27*'ii'

2.. —22° 5' — o*»3o' — 14*»26' +i5**22' +8<>5o' +I7*'43' — 26*»48'

3.. —21^45' +o«»i5' — 15°55' +I4*'i5' -f8°2/ +19^30' — 26*»25'

4.. — 2i«> 7' — i°i5' ~I5°50' +14*^20' +8°34' +i8*>46' — 23<»32'

5.. —20*30' —2® 5' — 15*15' +I4°2l' +8*32' +I9°24' — 2I*»I2'

6.. — 20°i5' — 3°3o' — 15*'27' +I4**35' +8*4' +i8°i6' — 2i**46'
7.. — 22O15' — 5*^45' — 17**52' +I2«49' +7*^6' +i4**4/ — 21*35'

Collection. Ca. C3. C4- Di. Da. D3. D4.

I.. —26*28' —7*26' +7*31' +10*50' +1*23' —18*35' —29*26'

2.. — 25''37' —6*^42' +7''20' +11*23' +2*40' — 17** o' — 27**45'

3.. —26*20' — 4°45' +13° 7' +2*25' —15*20' —28*19'

4.. —26*30' —4*29' +12*46' +2*25' —15*0' —27*38'

5.. —26*7' — 3''55' +13° o' +1*13' — 14*38' —27*48'

6.. —26*0' —4*5' +13*0' +1*15' —14*7' —26*11'

7.. —26*28' —6*6' +10*48' —0*55' —14*19' —26*12'

Note. — ^The yield of oleoresin from C4 was so small, after the first and second
collections, that not enough volatile oil could be obtained on distillation to fill the
100 mm. tube.

From this table it is seen that the rotation in most cases is quite con-
stant throughout the year. The most marked exception is A2 (P. heUro-
phylla). It is evident that some distinct change in the biological ac-
tivity of this tree has taken place, for while the rotation is reasonably
constant during the first half of the year, a steady increase in the levo-
character of the oil is apparent during the last half. In the case of Ci
(likewise P. heterophylla) somewhat the reverse has taken place. A
rather marked decrease in the levo-rotation is shown just at the middle
of the year, then the rotation remains practically constant during the
last half. In the case of C3, another t)rpe of change is represented, the
levo-rotation decreasing up to the middle of the season and again increas-
ing during the latter half.

With the limited facts at hand, it is impossible to interpret the signifi-

METHOD OF ANALYZING SHELLAC. 867

cance of these changes. That tree which shows the most marked varia-
tion, A2, is a healthy, vigorous tree, from which variations would be
least expected. Nor can an explanation be oflfered for the wide varia-
tions in the optical rotation of oils from the same species. All of the
trees in Series A are located within 20 yards of each other and have, there-
fore, the same general conditions of climate, light and soil. Fractiona-
tion of the volatile oils from these show practically the same rise in boil-
ing-point for the same volume of distillate. It would seem, therefore,
that these volatile oils, consisting so largely of pinene, are mixtures
principally of dextro- and levo-pinene, the preponderance of the one or
the other determining the optical rotation.

0NIVBR8ITT OP NORTH CAROLINA,

Chapel Hill, N. C, February 3, 1908.

A METHOD OF ANALYZING SHELLAC.'

Bt Parker C. McIlhxnet.
Received March 13, 1908.

The method of analysis which is in most common use at the present
time, both in England and in the U. S., for the determination of the
amotmt of rosin in shellac, is an indirect method depending upon the
different powers of a shellac and of rosin to absorb iodine from a suitable
solution. Different operators prefer different methods of making this
test, some preferring to use the old Hubl method, and others the more
modem -Wijs method, as modified by Langmuir. Either of these meth-
ods is capable of giving reasonably satisfactory results, although the
Langmuir method is certainly much to be preferred, both on the score of
accuracy and of speed. Another method which is in use is that proposed
by Parry, depending upon the solubility of the resinate of silver made
from common rosin, in ether, while the corresponding resinates from shel-
lac are insoluble. This method labors under several disadvantages
and sources of error, of which the two principal ones are the solubility
of the unsaponified portion of the shellac in ether, and the danger of a
decomposition of the resinate of silver before it can be separated and
determined.

A direct method of separating shellac and rosin and recovering the
rosin, at least, in a substantially unchanged form, is greatly to be de-
sired, and several experimenters have attempted to make such a sepa-
ration by taking advantage of the solubility of rosin in petroleum ether,
a solvent in which shellac is insoluble. No metjiod of extracting from
even a finely pulverized sample of shellac the portion soluble in petro-
leum ether seems to be capable of removing more than a small part of
the rosin contained in the sample.

' Read before the New York Sdction on March 6, 1908.

868 PARKBR C MCILHINBY.

Shellac dissolves in alcohol, except for the ^ax contained in it; petro-
leum ether dissolves to some extent in alcohol and it was though that
by first dissolving the shellac to be analyzed in alcohol, then adding
to the solution all the petroletmi ether which it would dissolve, and then
adding water, so as to so dilute the alcohol that it would no longer have
any material solvent power upon either rosin or shellac, there should
result a separation of the dissolved petrolic ether containing in sohition
the rosin and the wax, but free from the resinous constituents of the
shellac. Upon trying this process, it appeared that it was correct in
principle, but that on account of the limited solubility of petrolemn
ether in alcohol, the separation was not quite complete. Upon substi-
tuting for ordinary alcohol, glacial acetic add, or absolute alcohol, in
which the sample of shellac to be examined is dissolved and then adding
to it petroleum ether, which is misdble in all proportions with acetic
acid, or absolute alcohol, it was fotmd practicable to combine with the
rosin and the wax so large an amount of petroleum ether, that upon add-
ing water, an almost complete separation of the rosin and wax from the
resinous part of the shellac could be effected.

Based upon these facts, the following process was devised: Dissolve
in 20 cc. of glacial acetic acid (about 99 per cent.), or the same volume
of absolute alcohol, 2 gram$ of the sample to be analyzed. This requires
a gentle heat. Add to the solution, after cooling, 100 to 300 cc. of petro-
leum ether, boiling under 80**. This addition of petroleum ether should
be made slowly, because the addition of so large an amount of petroleum
ether precipitates from its solution a part of the shellac, coinbined, ap-
parently, in case acetic add was the original solvent, with acetic add
to form a liquid predpitate. It is manifestly desirable that this pre-
dpitation of part of the shellac should not be effected by too sudden an
addition of petroleum ether, as it might then contain some rosin carried
down mechanically with it.

We now have a solution containing both petroleum ether and glacial acetic
add, or absolute alcohol, and containing in it, in solution, all the rosin, all
the wax and most of the resinous part of the shellac. Add now to this solu-
tion, drop by drop, 100 cc. of water, agitating the liquid during the addition.
The water ''unites with" the alcohol or acetic add, and separates from the
liquid the petroleum ether with whatever is soluble in petrolic ether.
This indudes the rosin and the wax. The shellac is also predpitated,
but as it is insoluble in petroleum ether it remains as a predpitate sas-
pended in the diluted alcohol or acetic add. The separation of the t^vo
liquids takes place very rapidly, and it is an easy matter to effect a com-
plete separation of the two layers in a separating funnel. The petro-
leum ether layer is washed once or twice with water and then filtered
through a dry paper into a weighed flask, from which the petroleum

METHOD OF ANALYZING SHELLAC. 869

■ether is then distilled off, leaving the residue of rosin mixed with wax,
which is weighed. To the weighed residue neutral alcohol is added,
and the flask heated to dissolve the rosin in the alcohol. The liquid is
then titrated with N/5 or N/io caustic potash. This gives a measure
of the amount of rosin present, as the average combining equivalent of
rosin does not vary greatly from 346. This is the figure used for this
titration in the Twitchell process. It is much better, however, not to
depend upon his determination of the rosin but to proceed to an actual
separation by adding to the neutralized alcoholic solution, a distinct
excess of alkali, and a sufficient quantity of petroleum ether; the mix-
ture is then transferred to a separating ftmnel, thoroughly agitated,
and some water added. The liquids are then separated, the petroleum
ether layer being washed with water, and the alcoholic solution of the
losin extracted once more with petroleum ether. The petroleum ether
sohitions are then united, the petroleum ether distilled off, and the resi-
due of wax weighed. The alcoholic solution containing the rosin is then
boiled until the alcohol has been expelled, and, if necessary, a further
addition of water made ; then dilute hydrochloric add is added, so as to
precipitate the rosin. This should be done in a weighed flask; the acidi-
fied liquid can now be decanted off through a wet filter paper so as to
leave behind in the flask the major part af the rosin, which is washed
with water. The remainder of the rosin upon the filter paper after
sufficient washing, is extracted with petroleum ether back into the weighed
flask from which the water has been decanted as perfectly as possible.
Upon distilling off the petroleum ether, the rosin is left behind in a con-
dition to weigh. If the whole of the last traces of water have not been
removed by the distillation with petroleum ether, it is well to add a
further quantity of dry petroleum ether and distil again.

Attempts were made to use other solvents, particularly coal tar ben-
zene instead of petroleum ether. It was found, however, that benzene
dissolved considerably more out of the shellac than the wax.

If it is desired to determine only the rosin in the shellac, it is unneces*
sary to distil off all the petroleum ether from the solution containing
the wax and the rosin. This solution may, instead, be treated directly
with an alkali capable of combining with the rosin, while leaving the wax
unattached. The most convenient method which I have found for ac-
complishing this is to add to the petroleum ether solution, after filtering
it to remove from it anything which it may contain in suspension, a solu-
tion of sodium hydroxide in absolute alcohol. Such a solution I have
found it convenient to make by dissolving in ordinary 95 per cent, alcohol
the appropriate amount of metallic sodium. Such a solution, when added
to the petroleum ether solution, mixes perfectly with it, and the alkali
combines with the rosin, and the mixed solution allows the saponified

870 PARKBR C. MCILHINEY.

rosin to be extracted from it by agitation with water, or better, witk
slightly diluted alcohol. Instead of using a solution of sodium hydrox-
ide in absolute alcohol, it should be possible to use the method of sepa-
rating rosin from neutral substances in petroleum ether solution, that is used
in the Twitchell process of determining rosin in admixture with fatty
acid. This consists in agitating the petroleum ether solution with an
alkaline aqueous solution, containing some alcohol, made by dissolving
one gram of potassium hydroxide with 10 cc. of alcohol in water, and
diluting to 100 cc. I have found the other method of procedure advan-
tageous, however, and beUeve it to give a more exact separation. The
solution drawn off from the petroleum ether, after extraction by alkali
and containing in solution the rosin, combined with potassium or sodium,,
together with whatever alcohol has been used to promote the combina-
tion and to facilitate the mechanical separation of the two liquids, is
heated for some time to remove the alcohol, and is then acidified with
hydrochloric or with sulphuric acid to precipitate the rosin. The pre-
cipitated rosin is then weighed as before.

Without enterihg into a discussion of the exact chemical composi-
tion of shellac wax, it may be proper to state that the wax may be sepa-
rated commercially from shellac by two essentially different methods,
which will probably give waxes having somewhat different compositions*
If the shellac is dissolved in an alkaline solution and the wax which re-
mains in suspension is filtered out, it is probable that different results
will be obtained from what would be obtained by the other process, con-
sisting of a solution of the shellac in alcohol, and filtration of the solu-
tion to remove the wax which will then remain in suspension. It is to
be expected that these two processes of solution would leave, in the un-
dissolved wax, small amounts of ingredients of different characters, and,
furthermore, the alcohol used in the second process would dissolve small
amoxmts of the wax which would probably be quite insoluble in an alkar
line aqueous solution. Again, it is quite probable that a caustic solu*
tion would behave towards shellac wax somewhat differently from a di-
lute carbonate solution. For the sake of uniformity and simplicity, I
am assuming that shellac wax is quite free, when pure, from all acid
substances capable of tmiting with free alkali. It is, nevertheless, true
that the user of shellac who wishes to have a determination made of
the amotmt of wax which it contains, usually desires this information
in order that he may know how much of the shellac will remain undis-
solved in alcohol. This amount of insoluble wax will, of course, vary
to some extent, depending upon the strength of the varnish that he pre-
pares with it, that is to say, the number of pounds of shellac which he uses
to a gallon of alcohol. As the analyst can hardly be expected to take
account of this in making his test, it appears more rational to regard as

MOTHOD OF ANALYZING SHELLAC. 87 1

^vax the whole of those matters contained in the shellac that remain in-
soluble in an alkaline solution.

In making a determination of the amount of wax in a sample of shelkc,
by this method, it is to be observed that some constituent of shellac wax
is evidently only difficultly soluble in petroleum ether. Upon submitting
^veral portions of the same lot of shellac to analysis in this way, the
only difference made in the different determinations being that varying
amounts of petroleum ether were used, the series of analyses gave, with
increasing amounts of petroleum ether, increasing percentages of wax,
until the proportion of about 125 to 150 cc. of the solvent to i gram of
a shellac containing about 5 per cent, of wax was reached. It may be
that by using some other solvent such as benzene for the determination
of wax in this way, a smaller amount will suffice, but as it seldom happens
that only wax without rosin is to be determined, petroleum ether is the
most generally applicable solvent. The solubility of rosin in petroleum
•ether is so easy and complete that no difficulty is experienced in extracting
from 2 grams of shellac 50 per cent, of rosin, using 100 cc. of petroleum
ether.

Whenever in the course of an analysis by this method a quantity of
material is separated by acidifying the aqueous solution which should
contain the rosin, its identity may be established with some certainty
by determining its iodine figure by the Langmuir method, and its acidity
by titrating with alkali in alcoholic solution.

The petroleum ether referred to here is a solvent made by redistilling
71® B6. benzine, separating for use that part which distils below 80®.
This fraction constitutes a large proportion of 71^ benzine, and such a
ledistillation gives, at a comparatively small cost, a satisfactory solvent
for the purpose. If a determination of only the rosin is desired, it is of
course tmnecessary to be very particular as to the volatility of the solvent
used, as it is unnecessary in such case to distil it off before extracting
the rosin by alkali.

By treating pure shellac according to this process, it is possible to sepa-
rate from it a small amount, sometimes as much as i per cent., of ma-
terials soluble in petroleum ether. This small amount of resinous mat-
ter, when examined, proves to be something essentially different from
common rosin. Its odor and its low iodine figure indicate that it is
some resinous constituent of shellac, perhaps a small amount of the major
constituent, which is slightly soluble in petroleum ether. As it is diffi-
cult to imagine that by this process rosin if a normal constituent of pure
shellac would escape detection, and as the small amount of resinous
matter here obtained is essentially different from rosin, it is reasonable
to conclude that, contrary to the idea held by many, common rosin or
a material similar to it, is not a natural constituent of pure shellac, but

872 CHAS. H. HERTY AND W. S. DICKSON.

that any rosin or colophony, which can be separated in a state of reasona^
ble purity from the sample of shellac, was originally added to the shel-
lac, as an adulterant.

The process here described allows the analyst to separate in a form
convenient for exhibition either as evidence in court or as an ocular
' demonstration for his client, any rosin which may have been added as
an adulterant to shellac.

Shellac varnishes may contain beside true shellac not only rosin, but
other gums and resins soluble in alcohol. It becomes, therefore, a mat-
ter of interest to ascertain how some of these other resins behave when
treated by this process. Two samples of maniUa, when treated, using
absolute alcohol as the first solvent, gave respectively, 41.2 and 43.3
per cent, of matter soluble in petroleum ether. The acidity of these
two lots of matter soluble in petroleum ether was in the case of the first
sample such that i cc. of normal alkali neutralized 41 1.7 milligrams
and in the case of the second 470.7 milligrams. Two samples of Kauri
gave, respectively, 37.9 and 27.0 per cent. Upon titrating with stand-
ard alkali these portions soluble in petroleum ether, it appeared that i
cc. of normal alkali was capable of neutralizing 903.6 mg. and 742.5 mg.,
respectively. Of Sandarac, two samples, when similarly analyzed, gave
34.96 and 36.19 per cent., having such an acidity that of the first 541.2
mg. would neutralize i cc. normal alkali, and of the second, 552.5 mg.
would neutralize i cc. Of Dammar, 89.9 per cent, proved to be sol-
uble, while the resin of Shorea roburta, a sample of which was kindly
sent by Mr. W. Risdon Criper, of Calcutta, gave 69.5 per cent, of soluble
matter.

A number of attempts were made to effect a satis&ctory separation
of the wax before separating the rosin from the shellac. It was found,
however, that on account of the solubility of wax in alcohol and in gla-
cial acetic add, this separation could not well be made by filtering out
the wax before the addition of petroleum ether. Neither were attempts
which were made to separate the wax by a preliminary solution of the
shellac in aqueous alkali successful in furnishing a method that at all
approached in feasibility to the method already described.

145 E. 33RD ST.« New York Citt.

THE VOLATILE OIL OF PINUS SEROTINA.

By Cha8. H. Hbrtt and W. S. Dzckion.
Received Pebmary lo, 1908.

Scattered among the forests of Long Leaf pine along the Atlantic sea-
board, there are found, usually in mixed stands, patches of Pond pine
(Pinus serotina) and Loblolly pine {Pinus taeda). These pines are
seldom subjected to turpentining, as the yield of oleoresin is not so plenti-

VOLATILE OIL OF PINUS SEROTINA. 873

ful as from the predommating types Pinus palusUris and Pinus hetero-
phylla. ' Nor are the two species usually distinguished locally, the name
"black pine" being applied to each. The striking odor of the wood of
Pinus serotina when freshly cut made desirable an investigation of its
volatile oil, and in collaboration with the TJ. S. Forest Service, the oil
has been studied in this laboratory during the past year. Well identi-
fied trees were selected in Florida. The trees were regularly chipped
throughout one season of eight months. The product from each tree
was collected every eight weeks. The oleoresin closely resembles that
from Cuban pine (P. heierophylia) being quite liquid and containing
relatively about the same proportion of crystalline acids. To this low
percentage of crystalline matter is to be assigned doubtless, as in the
case of P. heierophylia, the absence of "scrape" formation on the scari-
fied surface of the tree, a formation so typical of P. pcUustris.

The volatile oil was distilled from the oleoresin by steam in the appa-
ratus described on page 865 above. The oleoresin evidently contains
a greater proportion of mucilaginous substances than that from the more
common pines, for it was much more difficult to distil. On heating to
140^, the usual temperature of distillation, and introducing steam, the
easily molten mass froths badly. This could be avoided only by raising
the temperature at the outset to 160*^. At this temperature, the vis-
cosity is diminished sufficiently to enable a complete distillation to be
carried out without frothing. During the latter part of the summer,
however, and during the autumn, the amount of this mucilaginous sub-
stance evidently increased, and to such ^ an extent that it became prac-
tically impossible to distil off the volatile oil. Partial success was secured
by the addition of concentrated sodium hydroxide solution to the dis-
tUling flask.

The resin left after distillation is pale yellow, similar to the best grades
of commercial resin. Acid number 167.

The volatile oil, freed from water by standing in contact with cal-
cium chloride, was a limpid liquid with a fragrant odor suggesting at once
the presence of limonene. The physical constants of the oil follow:

Sp. gr.: 20®, 0.8478.

Sp. rotation: 20**, — 105*36'.

Index of refraction: 20®, i . 4734.

Add number: o.

Saponification number: i . 54.

Iodine number: 378.

Solubility in ethyl alcohol at 22 . 5**:

95 per cent, alcohol i . 35 parts required to dissolve i part of volatile oil.

90 per cent, alcohol 4 . 80 parts required to dissolve i part of volatile oil.

85 per cent, alcohol 8 . 10 parts required to dissolve i part of volatile oil.

80 per cent, alcohol 16. 20 parts required to dissolve i part of volatile oil.

70 per cent, alcohol 56.00 parts required to dissolve i part of volatile cnl.

874 AUGUSTUS H. GIlrL.

Comparative evaporation with the volatile oil of P. palusUris, at room
temperature, in shallow watch glasses, 0.2 gram of each used.

P, paluslris. P. sevoiimm.

Time. Percent. Percent

Loss after }4 hour 35. 7 20.30

Loss after i hour 62 . 5 37.y>

Loss after i}4 hours 91.7 53-40

Loss after 2 hours 96.0 68.47

Loss after 5 hours 97 . 8 98.8

On fractionation the following results were obtained :

Index of refraction , KotAtion in loo mm.

Temperatures. Per cent, distillate. xP. tube vP.

172-175*' 27.4 1. 4716 —87*53'

175-180** 57.0 1.4724 — 92**2l'

180-185** 8.4 1.4744 — 92*14'

185— f- 7.2 1*5045

Repeated fractionation at atmospheric pressure showed some poly*
merization. From a fraction, 175-176°, a large yield of limonene tetra-
bromide was obtained. Melting-point 103°-! 04®. The solution of the
tetrabromide in chloroform was levo-rotatory, — 70.0®.

A study of the oxygen absorbing power of this volatile oil in compari-
son with that of the ordinary spirits of turpentine obtained from P.
palusiris showed a much larger absorption by the oil of P. seroivna dur-
ing the early days of the experiment, but the total absorption after three
months' exposure to northern light was practically the same in each.

Univbrsxty op North Carolina,
Cbapbl Hill, N. C, February 4, 1908.

ON THE OZmATION OF OLIVE OIL.

By Augustus H. Gill.
Received Pebmary 14, 1908.

Some years ago it became a question of the determination of the kind
of *'wool oil" that had been employed in the noianufactute of certain
"tops." Tops may be defined as wool roving or wool which has been
partially spun. In their manufacture the wool is scoured and oiled,
usually with an olive oil emulsified with either ammonia or sal soda, then
it is carded and spun. As the tops are stored "in the grease," as the ex-
pression is, two months may elapse before they are used, so that the oil
spread over these fibers has ample opportunity for oxidation.

The oils extracted from the tops had the characteristics shown in the
table below:

Top. No. 1. 2. 3. 4.

Date left mill Aug. Nov. Nov. Nov.

Date tested Jan. Dec. Jan. Jan. Dec. Jan.

Iodine No 39 53.6 42.1 45.5 54.0 40.5

Saponif. No 213 225 221.5 207.5 216.0 221.5

OXIDATION OF OUVE OIL.

875

The January samples were different from those tested in December,
although from the same lot.

Ballantyne^ gives the following iodine figures for olive oil which had
been "kept in direct sunlight, uncorked and agitated every morning."

Original oil. i month. 2 mos. 3 mos. 5 mos. 6 mos.

83.2

82.5

82.3

8i.6

79.1

Sherman and Falk* give the following figures:

Fresh.

Olive oil 83.8

Lard oil 73 . 3

69.3
73.3

•I

78.2

Bzposed.

77-4
66.7
54-6
56.2

Prom the results obtained from the oils extracted from the tops it
would seem that the oil could not have been by any possibility olive oil.

As an aid in settling this question, olive oil was exposed to atmospheric
agencies under varying conditions, as follows:

I. Olive oil A was emulsified with ammonia and soda in the usual
way, the emulsion sprinkled upon absorbent cotton and allowed to lie
in the laboratory for seven weeks, covered with paper to keep off dust.
The time (seven weeks) was approximately equal to that which the oil
had been upon the top when received. The oil was then extracted from
the cotton (sample B).

II. Another portion of olive oil A was emulsified as in I and extracted
from the emulsion by ether (sample C). This was to see what effect,
if any, the emulsification process had.

III. A portion of olive oil A was oxidized by drawing a current of air
through it eight hours per day for seven weeks (sample D).

IV. A fourth portion of the original oUve oil was allowed to stand for
this same length of time in an open beaker in the laboratory (sample E).

V. Another portion of the original oil was emulsified and sprinkled
upon the tops themselves, which had been extracted with naphtha and
allowed to remain for five weeks as in Experiment I. The oil was then
extracted (sample F).

VI. Lastly a sixth portion was heated for two hours in an open dish
to 120^ to see the effect of heat (sample G).

Tablb op thb Oxxdation op Ouvn Oil undsr Various Conditions as Shown by

THB IoDINB NUMBBR.
A.

Otirinal
<3l.

H.
Original oil

Emulsified

Heated

after standing

oil from

oil by
itself.

Blown

Open

"Top"

Ditto

7 weeks.

cotton.

oil.

beaker oil.

oil.

ixP.

closed.

83.2

62.0

84.0

84.0 83.9 78.2

82.0

83.5

' /. Soc, Chem. Ind., xo, 29.
• This Jotjrnai,, ^7, 606.

876 FRED W. MORSE.

The iodine number of the oil supposed to have been used on the tops
at the mill was 84.3; assuming that the change was the same as in the
laboratory, that upon the tops should not have been below 63 or cer-
tainly should not have reached 53-54, as both my determinations and
that of the mills chemist showed.

The work corroborates that of Ballantyne and shows that except when
spread out in a finely divided condition as upon cotton, olive oil changes
but little on exposure to the air or heat.

The high saponification numbers, 207-221, indicate that the oil has
undergone oxidation and also the entire absence of any unsaponifiable
or mineral oil. It would seem from these results that the oil used
upon the tops was most likely lard oil; had it not been for the possibility
of cholesterol in the wool this could have been shown by a test for choles-
terol in the extracted oil.

In conclusion, the writer wishes to express his indebtedness to Mr.
H. S. Bailey, by whom the analytical work was performed.

Massachusetts Institute op Tbchnolooy,
Boston, Mass.

THE EFFECT OF TEMPERATURE ON THE RESPIRATIOn OF

APPLES.

By Fred W. Moksb.
Received February 17, 1908.

While engaged in an investigation of the effects of different methods of
storage on the chemical composition of apples, it was foimd impractica-
ble with the methods of analysis in common use to determine the varia-
tions produced by comparatively small changes in temperature, or in other
words, whether 32° F or 45° F affected the composition of the fruit in a
different ratio.

This difficulty was due to the fact of the destruction of some of the
apple constituents by respiration, which could be easily deduced from the
exhalation of carbon dioxide and water, and the practically constant
proportions of water and dry matter which existed in spite of a steady
loss of water, and decrease in weight.

It seemed possible that the rate of chemical change might be measured
by determining the rate of exhalation of ^carbon dioxide. Some simple
experiments with fruit under bell jars, over mercury, and over water,
soon showed that temperature had a very marked effect on the exhalation
of the respiratory products.

A respiration apparatus was plaimed and constructed as follows: The
chamber in which the fruit was to be placed was a cylindrical vessel of
copper supported by a tripod in an upright position. The bottom of
the cylinder was formed like a funnel ending in an outlet tube of brass

RESPIRATION OF APPt,ES. 877

to connect with absorption apparatus. The top of the cylinder was
closed by a disc of glass, which rested on a narrow shelf soldered to the
inner wall of the cylinder. An inlet tube entered the cylinder just be-
neath the shelf.

The dimensions of the cylinder were, diameter 20 cm., height 25 cm.,
and conical bottom 5 cm. in depth. The shelf for the glass plate was
2 cm. below the upper edge. The total volume was a little more than 6
liters. The inlet and outlet tubes were 6 mm. in diameter.

The respiration chamber was placed inside a larger galvanized iron
tank, which could be filled with ice or water in order to control the tem-
perature. The outlet tube was passed through a tubulure near the bot-
tom of the outer vessel; The inlet tube was passed through a hole in the
cover of this vessel, led to a point near its bottom, then coiled around the
tiipod and, finally, soldered to the inlet orifice near the top of the respira-
tion chamber.

Air, before entering the chamber, would thus be brought to the tem-
perature of the surrounding tank, and the carbon dioxide would fall
and accumulate in the funnel-shaped bottom. A slow current of air
would maintain a constant temperature within the respiration chamber
and the carbon dioxide would be readily removed because of its density.

The air was freed from carbon dioxide before entering the chamber
by passing it through a solution of potassium hydroxide. The exhaled
carbon dioxide was collected in two absorption tubes connected with the
outlet. The first tube contained a 20 per cent, solution of potassium
hydroxide and the second tube a standard solution of barium hydroxide.
The determinations were made by titrating the alkaline solutions with
half-normal hydrochloric acid.

In titrating the potash solution, phenolphthalein was added as an indi-
cator and the acid added until neutral, then methyl orange was added
and an exact measurement made of the amount of standard acid now re-
quired to neutralize the potassium hydrogen carbonate. The barium
solution was titrated with the standard acid and phenolphthalein.

In the majority of experiments the potassium hydroxide absorbed all
the carbon dioxide.

The procedure in an experiment was as follows:

About 2 kilograms of sound Baldwin apples were placed in the respira-
tion chamber, and the glass disc was firmly sealed in place with putty.
A current of air was drawn through the apparatus by means of an aspira-
tor holding about 16 liters of water. The rate of flow was adjusted so
that the air would be renewed three or four times during the experiment,
and yet not pass too rapidly to permit complete absorption of the gas
by the alkaUne solution.

Blank experiments were repeatedly made to include the carbon dioxide,

878 PRED W. MORSE.

which might leak in, together with that in the atmosphere of the empty
chamber and the amount present in and absorbed by exposing the solu-
tions while filling and empt3ring the absorption tubes, and titrating.
Corrections were then made by deducting results of blank tests from the
amounts obtained in experiments with the fruit.

The leakage gave but 2 to 3 mg. of carbon dioxide; but the contamina-
tion of the solutions was of considerable consequence, requiring about 2.5
cc. of the half-normal add to correct it.

The earlier experiments were conducted only during the day, because
the aspirator was fotmd unreliable in maintaining a continuous current
of air. Therefore, each morning, air was drawn rapidly through the
chamber until its atmosphere had been renewed repeatedly, before the
absorption tubes were attached for the collection of the exhalations.

Later in the season, a larger aspirator was employed by which a con-
tinuous run could be maintained for two days or more at a time. Some
of the experiments were made at room temperatures, some with the outer
tank filled with ice -water, and some with it filled with closely packed ice.
No attempt was made to regulate the temperatutes closely, during these
trials, but readings of a thermometer in the room or in the outer tank
were recorded.

Since the weights of fruit and length of time varied with different ex-
periments, the amounts of carbon dioxide were calculated on a common
basis of one kilogram of fruit and one hour of time, and the quantities
expressed in milligrams.

The data of the different runs follow, with the results arranged in groups,
according to the temperatures. The room temperatures and the ice-
water temperatures ranged through several degrees, but all results are
included.

Carbon Dioxidb Exhaled by i Kiu> op Applbs Pbr Hour.

Room temperature, x8^ to 25® C Medium temperature, 5^ to xo** C. Low temperature, o^ C.

Hrs.

Mgs.

Hrs.

Mgs.

Hra.

Mg^

Oct. 16

S}4

16.4

Oct. 22

8.0

Oct. 29

2.3

Oct. 17

18.7

Nov. 7

6ye

7.3

Oct. 30

3.6

Nov. 5

12.6

Dec. 2

8.7

Nov. 8

3-8

Nov. 6

12.6

Dec. 3

9-5

Mar. 2-4

46K

2.4

Nov. 12

18.2

Dec. 4

6V^

8.8

Mar. 4-6

55K

2.8

Dec. 9

18.0

Dec. 5

7.9

Mar. 9-1 1

2. J

Dec. 10

17.9

Dec. 6

9.6

Dec. 11

6}^

23.0

Mar. i6>i8

5.5

Dec. 12

26.7

Feb. 26-28

17.9

It was invariably noticed that a number of hours was required to hrin^
the apples into equilibrium with the surrounding temperature, hencx,
when changing from room temperature to that of melting ioe, the lespiia-

RESPIRATION OF APPLES. 879

tion would slowly decrease until it reached the normal rate for the latter
environment.

The average rate of exhalation of carbon dioxide was 18 mg. per kilo
an hour at room temperatures, 8.1 mg. at medium temperatures, and
2 . 7 mg. at zero.

A comparison of the results obtained at the different temperatures
showed that they were not directly proportional to the variations in tem-
perature, but that there was an acceleration of the rate of respimtion as
the temperature rose, which corresponded to the law of acceleration of
chemical action by rise of temperature.

At room temperature, the amount of carbon dioxide per kilo an hour
was about 6 times what it was at o®. The intermediate temperatures
gave an average result which was 3 times that of o**, and not quite one-
half of that at the higher temperatures.

The investigation was now dropped for a time, tmtil an opportunity
offered for better control of temperature. Although there was a strong
probability that the acceleration of respiration was in accord with the
acceleration of chemical action with rise in temperature, there were un-
certainties about the actual temperatures involved.

A year elapsed before the investigation was resumed. Particular at-
tention was given to the temperature of the apples between the runs in
the respiration apparatus, in order to avoid the lag in temperature change
in the apples themselves.

The apples were brought in from a cool storage cellar where the tem-
perature was running between 45° F and 50° F, or 8° to 10° C. It was
therefore decided to try the temperature of 10° first, and save the time
required to bring the fruit to some other temperature. The tempera-
ture of the chamber was maintained by means of ice-water, just enough
ice being added from time to time to keep the thermometer in the outer
chamber at 10^. The warm room, of course, tended constantly to raise
the temperature.

The apparatus was the same used^in previous ^trials. The aspirator
was run as rapidly as was consistent with thorough absorption of the
carbon dioxide. The same methods of determining carbon dioxide
were employed and everything done to make the results comparable
with the earlier ones.

The experiments were conducted for several successive days. After
the day's run was made, the apples were removed to a large jar and placed
in a pail of cold water which sat in the coolest part of the building where
the temperature had been noted to remain at about 10° during the night.
In the morning of each day, they were replaced in the respiration appara-
tus and the experiment thus resumed. On one morning the water was
foimd to have fallen to o^ and no run was made on that day.

88o FRED W. MORSE.

For the temperature of o°, the outer tank was kept full of closely packed
snow throughout the day. When each day's run was completed the ap-
ples were tightly closed in a large jar and buried in a snow-bank out of
doors. They were also put out there for about i8 hours before the first
run.

Room temperature proved difficult to control. The rooms were hot in
the latter part of the day, cool at night and gradually warming during
the forenoon. Runs were omitted on days when the morning tempem-
ture of the room was below 15°.

The results of the three series of experiments are given below:

Experiment 11. Weight of apples, 1428 grams.

Temperature 10** C.
Mar. 15. 6 hours, carbon dioxide 116. 6 mg.

Experiment 12. Weight of apples, 1147 grams.

Temperature 10^ C.
Mar. 16. 6 hours, carbon dioxide 93.5 mg.
Mar. 18. 6 hours, carbon dioxide 85.8 mg.

Experiment 13. Weight of apples, X145 grams.

Temperattire o** C.
Mar. 21. 6 hours, carbon dioxide 38.9 mg. j

Mar. 22. 6 hours, carbon dioxide 39.6 mg.
Mar. 23. 6 hours, carbon dioxide 27.5 mg.
Mar. 24 6 hours, carbon dioxide 36. 3 mg.

Experiment 14. Weight of apples, 1142 grams.

Temperature 20**.
Mar. 25. 6 hours, carbon dioxide 150. 2 mg.
Mar. 28. 6 hours, carbon dioxide 145 . 2 mg.
Mar. 31. 6 hours, carbon dioxide 154.0 mg.

The results for the different temperatures calculated for an hour and a
kilogram of fruit are given below:

Carbon dioxide.
Dmte. Temperature. Mg.

Mar. 15 10** 13.6

Mar. 16 13.6

Mar. 18 12.5

Average, 13.2

Mar. 21 o** 5.7

Mar. 22 5.8

Mar. 23 4.0

Mar. 24 5.3

Average, 5.2

Mar. 25 20® 21.9

Mar. 28 21.2

Mar. 31 22.5

Average, 21.9

STABIUTY OF LKCITHIN. 88 1

The acceleration from o^ to lo^ is more marked than that from io° to
20°, but the ratio of 2 to i holds practically true and is 4 to i for the
rise of 20®.

While this work was in progress there was published by Bigelow, Gore
and Howard* a description of respiration experiments with apples where
there were two lots, one at 0° and the other at 15°. These results were
given in percentages of the original weights of apples. From their tables
it was found that but four dates could be compared, which are given
herewith, together with the percentages of carbon dioxide and the ratio
between o® and 15°.

Date. 0°. • 150. Ratio o*>: 150.

Jan. 5 0.234 per cent. CO, 0.794 per cent. CO, 1:3.3

Jan. 27 0.308 0.899 1:2.9

Mar. 2 0.436 1. 122 1:2.5

Mar. 30 0.484 1-375 1:2.8

The avemge ratio is 1:2.9, which is approximately that of 1:2 for a
rise of 10®.

All these results show concordance and prove^that apples undergo
chemical changes fully twice as fast and in some instances three times
as fast with a rise of temperature of 10° between o^ and 20°, or in other
words, at summer temperatures apples will undergo respiratory metab-
olism from 4 to 6 times as mpidly as in modem cold storage. The low
temperatures alsojprovej^that therej^must be a limit to the keeping quality
even there, since respiration and^consequent destruction of cell tissues
still goes on.

NBW HAMPSBI&B AOSICni.TUIlAL BXPBRXMBKT STATION,

Durham, N. H.

OBSERVATIOirS OIT THE STABIUTY OF LECITHm.

BT J. H. I«ONO.

Received February 19, 1908.

Numerous investigations published in the last four or five years on the
subject of the preparation of the lecithin compounds from eggs or from animal
and vegetable tissues have discussed more or less vaguely the stability
of these products under the influence of light, heat, and atmospheric
oxidation. It seems to be assumed that the lecithins in general suffer
very ready decomposition, but in the literature I am unable to find much
that is definite as to the extent of their decompositions which take place
under the influences referred to. In the course of certain experiments
in other directions I found the need of this information and felt obliged
to carry out some experiments to supply the desired data.

At the outset it may be said that the conception of the term ** lecithin "

' U, S. Dept, Agr., Bur. of Chem., BuU. No. 94, "Studies on Apples."

S6l J. H. LONC.

is still very vague, in spite of the extended studies of Thudichum/ Koch,-
Erlandsen/ Stern and Thierfelder,* E. Schulze* and colleagues, and othen,
in addition to the well-known investigations of the older literature, and
what may be affirmed of egg lecithin, as we know it to-day, does not neces-
sarily apply in full to an analogous product from the vegetable kingdom
or from brains. The extended experiments of Thudichum and Erlandsen
have contributed greatly to modify our views as to the brain and musck
extracts of a fatty nature containing nitrogen and phosphorus, whik
the recent paper of Stem and Thierfelder, referred to, shows in ckar
light the fact that egg lecithin, generally supposed to be comparatively
simple and containing probably two main constituents, must in reality
be much more complex; but the relatively small yields secured in the
fractionations described in these papers are probably not wholly due to
imperfect insolubility, but are much more probably due to partial decom-
position of some of the individual substances during treatment. That
this is the case is shown further by the rather marked acidity of some of
the fractions, which is probably due to separated glycero-phosphoric
acid, or other phosphoric acid derivative. The modifying influence of
the presence of some of these decomposition products on the reactions
of ** lecithin " is generally overlooked and will be referred to later.

However, it is not my intention to take up the preparation of varioos
lecithin products at the present time, but rather to present data bearing
on the stability of some of the best known representatives of the group,
as secured through generally recognized methods. Egg lecithin was
taken as the first of these products, as representing the simplest and mo^
readily prepared.

Experiments with Egg Lecithin.

Pteparation. — The largest quantity of this used was made from yolks
without previous drying. The yolks of 72 eggs were treated in tots of
12 eggs each. In each case 300 cc. of ether were added and shakenwith
the dozen yellows through several days ; this was followed by the addi-
tion of 500 cc. of alcohol, after which the mixture was well shaken repeat-
edly and allowed to settle. The alcohol-ether solution was filtered and
evaporated to a pasty condition at a low temperature, and finally by aid
of vacuum. The residue was taken up in pure ether, the solution fil-
tered, concentrated and precipitated with pure neutral acetone in ex-
cess. This operation of dissolving in ether and precipitating by acetooc
was repeated three times, the last product being carefully dried in a

* **Die chemische KonstUution des Gehirns des Menschen und der Thiere" 1901.
' Z, physioL Chem,, 36, 134; 37, 181; and elsewhere.

'Ibid., 51, 71-

* Ibid., S3, 370.

* Ibid., 40, 10 1, where other literature is dted.

STABILITY OF LECITHIN. 883

vacuum at a low temperature, but not so as to remove all the iroisture,
as iray be done by dr>4ng over sulphuric acid through a long period.
By drying in the latter way a very hard, homy product is secured, which
is not easily worked with later. In several lots prepared in my experi-
ments, the average water content left was 6 per cent. In all, about 90
grams of the final product were secured in the form of a light yellow
mass, which, on analysis, showed these results for phosphorus and nitro-
gen:

® Per cent.

P 3-59

N 1.82

These values calculated to the anhydrous condition give:

Per cent.

P 382

N 1.94

which correspond to an atomic ratio of

P:N: : i: 1.12,
which suggests a mixture containing some diaminomonophosphatide.
With this product the following tests were made, an aqueous emulsion
containing in 100 cc. 4.476 grams of the anhydrous substance being em-
ployed in most cases.

Effect of Heat. — ^Two portions of 25 cc. each were evaporated to dry-
ness in a current of carbon dioxide, the temperature being kept at about
60°. The residues found weighed 1133 and 1.136 grams in place of
1. 12 grams, about, as fotmd by long drying over sulphuric acid. This
experiment was repeated with three new portions of 25 cc. each.
These were evaporated at a temperature of 100° in a vacuum -drying
oven, through which a rapid current of carbon dioxide was passed. At
the end of two days constant weights were reached as follows:

A. B. c.

I. 115 I. 127 I. 122

The dry residues were washed in the small dishes with portions of 25
cc. and then with 10 cc. each of pure dry acetone, the acetone remaining
half an hour in contact with the residues. After pouring off the last ace-
tone the dishes were returned to the oven and again dried in carbon di-
oxide. The new weights found were then :

A. B. c.

1.048 1.070 1.063

Nitrogen determinations were made on these residues, which gave the
following results :

A. B. c.

Per cent. Per cent. Per cent.

1-97 1-93 I 95

It is evident that these final residues have about the same composition

884 J- H. tONC.

as the original, although some loss has occurred in the drying and wash-
ing with acetone. This loss is apparently through solubility of the sub-
stances as a whole and not through decomposition.

In another experiment 25 cc. of the emulsion were evaporated to dr}'-
ness in the open air. The product found was very brown and weighed
I . loi grams, that is, less rather than more than the normal. The nitro-
gen found in it was 0.0214 gram, which corresponds to 1.94 per cent,
of the dry weight. In spite of the dark color no important volatile de-
composition product had been formed.

Essentially the same result was foimd by long boiling. Twenty-five cc. of
the emulsion were diluted with 50 cc. of water. The mixture was placed in
a flask with a stopper furnished with a fine opening and boiled long enough
to bring the volume back to 25 cc, which required about an hour and a
half. The nitrogen found was i . 94 per cent, again. It is evident that
no volatile nitrogen products were formed, and the emulsion remained
perfect.

In another experiment 25 cc. of the emulsion were heated in a platinum
dish in an autoclave to a temperature of 175®, through two hours. After
cooling, the contents of the dish consisted of a clear liquid and a dark
fat-like ring on the dish at the surface of the liquid. The total nitrogen
in the dish was found to be 0.020 gram, or i .79 per cent, of the weight
of the original dry substance. There was evidently some loss, there-
fore, of this element.

From these several experiments it is evident that the effect of heat,
alone, is not very pronounced, if the conditions for oxidation are absent.
It has been noticed in several other cases that even after long heating
to 100° in an atmosphere of carbon dioxide, perfect, light-colored emul-
sions could still be secured, with no change of properties.

Acidity. — ^The various lecithin preparations which I have examined
show a decided acidity to phenolphthalein, and this property in their
product is referred to by Stem and Thierfelder.* This acidity may be
observed directly in the emulsions by titration with 0.1 N sodium hydrox-
ide, and more clearly after addition of neutral alcohol. Twenty-five cc.
of the above emulsion, containing 1.12 grams of anhydrous lecithin, re-
quired directly i . 3 cc. of this weak alkali, and after the addition of alco-
hol nearly 4 cc. In a second test i gram of the lecithin, or o. 94 gram of
dry product, was made into an emulsion with 15 cc. of water and 25 cc
of alcohol were added. On titration, I used now 3.6 cc. of the o. i nor-
mal alkali.

This acidity does not appear to be due to acid liberated on formation
of the emulsion with water, as it is also observed on titration of a solu-
tion in strong neutral alcohol. A solution made by dissolving 1.157
* Loc. cii.

STABIUTY OF LECITHIN. 885

grams of lecithin (anhydrous) in neutral alcohol, to which a little neu-
tral ether was added, required 4 . 2 cc. of the o . i normal alkali with phenol-
phthalein. The acid substance is therefore present in the original leci-
thin as separated by the process outlined, and the fact that in absence of
alcohol a weak aciditv is shown, while after addition of the latter a much
stronger acidity is developed, is evidence of the presence of two kinds of '
acid substances. One of them is soluble in water and is indicated in the first
part of the titration, while solution in alcohol is necessary to bring out the
second substance. It is likely that the glycero-phosphoric acid complex
is responsible for the first reaction, while separated fatty acids, on solu-
tion in alcohol, bring out the second. Pure glycero-phosphoric acid
behaves as a dibasic acid with phenolphthalein and is monobasic with
methyl orange, while the acid complex, as separated from lecithin, would
doubtless act as monobasic with the first indicator and neutral with the
second. The lecithin emulsions I have made are not acid to methyl
orange. From the above it appears possible, if not probable, that the
acidity observed in the lecithin is due, in part, at least, to small amounts
of dissociation or hydrolysis products, rather than to the substance itself.
If we may assume that the free add hydrogen of the phosphoric group is
fully combined in the titration in aqueous solution, each cc. of the o . i
normal alkali used would measure 8.07 mg. of lecithin decomposed, or
in the above case, would correspond to about 10 per cent, of the whole.
But this asstunption cannot be correct, as a part of the phosphoric acid
is apparently already combined with calcium. The ash of the lecithin
contains this metal, as shown by Thudichum for the brain lecithin, and
by Stem and Thierfelder for the ^gg product. Further light on the ques-
tion of acidity will be given in some experiments to be referred to below.

Electrical Conductivity. — Since a determination of electrical conduc-
tivity seems to furnish very interesting evidence as to the progress of
hydrolysis or other change with liberation of acid in the lecithin emul-
sion, I have made a large number of tests on this and other samples of
lecithin from various sources. At best, the conductivity is low, and its
range is indicated in the following table. The measurements were made
by the usual Kohlrausch telephone method, and always at a tempera-
ture of 20°, accurately maintained. The lecithin was made into an
emulsion with water of high purity, the conductivity of which may al-
ways be neglected for these tests. The emulsion employed contained,
like the one referred to above, 4 . 476 grams of anhydrous lecithin in 100
cc The variations in the conductivity with the dilution are shown be-
low.

The emulsion once formed, is comparatively stable, as shown by this
experiment. Twenty-five cc. of the original were mixed with 75 cc. of water
and heated on an actively boiling water-bath two hours, in a flask. After

886 J. H. LONG.

cooling, the remaining liquid was made up to loo cc, accurately, and
the conductivity found. It was «2o — 0.000300, that is essentially the
same as in the second dilution below.

Cone, in loo cc. '^>

4.476 0.000798

2.238 0.000508

I. 119 0.000299

0.559 0.000172

0.279 .'. Hi 0.000096

0.140 0.000054

To what is this conductivity due ? In the usually accepted fonnula
for lecithins there is a free hydrogen in the phosphoric group, but as inti-
mated above, the acid value of this must be low. In the preparation of
these lecithins, alcohol, ether and acetone of a high degree of purity
were used throughout. These liquids were tested for conductivity and
residues from evaporation of 50 cc. of each one, taken up with water,
were also tested. In no case was a conductivity found which was at all
appreciable. The triple solution in ether and precipitation by acetone
insured the freedom from electrolytes originally present. As a corre-
sponding degree of conductivity has been found in many other samples
of lecithin from different sources it would seem to be inherent in the mole-
cule, but that this is probably not the case the next experiment will show.
Another emulsion of the same strength as the last was made up and
examined in the same cell, which contained the usual platinum black-
covered electrodes, with capacity, C = 0.306. The conductivity was
found to be «2o = 0.000718. Thinking the nature of the electrodes might
have some effect, a new test was made in a cell with bright electrodes
and C = 0.383. I found now /C20 = 0.000715, and the value was not
changed after washing the electrodes with ether and alcohol.

Twenty-five cubic centimeters of the last emulsion were measured out
and precipitated with 45 cc. of acetone in a small separatory funnel,
and the precipitate washed with 10 cc. and finally with 5 cc. more of ace-
tone. The residue was dried in a current of washed carbon dioxide
and emulsified with water to again make 25 cc. In the same cell, with
the bright electrodes, I found now a resistance over 10 times as great, or
/Cjo = 0.000066. After standing 18 hours, the result was unchanged.
The acetone was evaporated and the residue made up to 25 cc with water.
During the evaporation a small amount of insoluble matter separated,
which appeared to consist of some of the dissolved lecithin, but which
was very^ light in color, while the lecithin residues proper are usually
quite dark. The aqueous solution contained a soluble substance, or sub-
stances, since a conductivity, /fjQ == 0.001576, was found and on titration
0.8 cc. of 0.1 normal sodium hydroxide was required with phenol-
phthalein.

STABILITY OF LECITHIN. 887

The last emulsion was treated anew with acetone, when the surprising
observation was made that a very large quantity of the latter must be
added. About 100 cc. were required to do now what was accomplished
^th 45 cc. in the first case, and more was needed to complete and to
ivash the precipitate. The latter was made up to a 25 cc. emulsion with
ivater, as before, and tested for conductivity, giving % = 0.000041.
This very considerable decrease may be here due in part to the loss of
the portion soluble in the excess of acetone, but the change in the first
case, after precipitation, cannot be so explained. Several experiments
have shown that the recovered lecithin, after the first acetone precipita-
tion and washing, is about 88 to 90 per cent, of the original weight. The
loss after the second precipitation is apparently much greater, while in an
attempt to precipitate a third time, nearly the whole of the substance
^vent into solution with the acetone. The emulsions foimd after pre-
cipitation and taking up with water are practically neutral to phenol-
phthalein.

It is evident from the above that the observed conductivity of the first
emulsion is due to something not true lecithin, and that when this is
separated precipitation by acetone is very difficult. To test this point,
some of the supernatant acetone from a first precipitation was evapora-
ted and the residue taken up with water. A few drops of this solution
added to the second emulsion caused it to precipitate with acetone im-
mediately. The same result was secured by adding a few drops of a very
dilute glycero-phosphoric acid solution; in this case a sharp result fol-
lo'wed at once, which suggests that this acid may be the fraction split
oflf from the original lecithin and is responsible for the observed behavior.
In precipitating lecithin from ether solution by acetone, in the process
of preparation, glycero-phosphoric add and other possible decomposi-
tion products seem to be carried down and remain with the finished mass,
but in precipitating from an aqueous emulsion this is not the case appar-
ently, and in this way a purer final product seems to be secured. On
this point, however, futher work is necessary, as experiments carried out
show, in some cases, a slightly lower nitrogen and higher phosphorus
content in the so-purified lecithin than in the other, which suggests that
the acetone, in presence of water, may have a splitting or hydrolyzing
action and leave a residue poorer in the fatty acid groups, and relatively
richer in the phosphoric acid group. It is possible, also, that a part of
the nitrogen may be split off from the latter as the following figures sug-
gest : Two emulsions were made, having 5 and 6 grams to 100 cc. These
i^ere precipitated with acetone, and the residues washed and dried in
carbon dioxide without loss. On weighing, it was found that 88.9 per
cent, of the original anhydrous lecithin was recovered. The two recov-
ered products were made up into emulsions again and portions taken

888 J. H. LONG.

for phosphorus and nitrogen determinations, with the following results,
considering the recovered lecithin as anhydrous:

A. B.

Per cent. Per cent.

P 4-27 3-98

N 1.65 1.68

It will be recalled that in the original dry material the nitrogen and
phosphorus were 1.94 per cent, and 3.82 per cent., respectively. It is
evident, therefore, that some splitting has followed, but the nature of
the reaction is not clear, especially in view of the facts of lower conduc-
tivity and lower acidity in the last emulsions.

Salt Precipitation. — It is stated above that an emulsion which will not
precipitate by addition of acetone, or at best imperfectly, may be caused
to yield a good precipitate by the addition of a trace of glycero-phosphoric
acid. It was found that weak salt solutions have the same action, and
this, with the original weak emulsions, as well as those made up after
acetone treatment. In this respect experiments have been made with
solution of sodium chloride, barium chloride, silver nitrate and aluminum
sulphate in various dilutions and with many other salts in certain n-olecu-
lar proportions. The precipitates are n-arkedly colloidal and do not
settle quickly. These findings do not seem to agree with the results of
Koch* for brain lecithin. According to this author, dilute emulsions of
brain lecithin yield precipitates with dilute solutions of divalent metals,
but not with mono- or trivalent metals. My findings are quite sharp
and conclusive, and, as will be shown below, hold for brain lecithin also.

In this coimection another interesting observation was made. It
was found very difficult to extract the lecithin from aqueous emulsions
by means of ether; in fact traces only seem to go into solution, and this
has been observed not only for the simple emulsions used in these experi-
ments, but also in some of the commercial lecithin emulsions on the mar-
ket. The addition of sodium chloride brings about an immediate solu-
tion, which is first shown by the color of the upper layer when the emul-
sion is shaken with ether in a tube, and which can be proven by decant-
ing the ether layer and evaporating. This behavior seems to depend
on the power of precipitating colloidal substances, possessed by many elec-
trolytes, and was shown by several other salts as well as by sodium chlo-
ride, but not by urea and sugar, which are crystalline but not electrolytes.
Further work is in progress on this interesting reaction.

Digestion Experiments. — ^The behavior of the fat-splitting ferment of
the pancreas was first pointed out, apparently, by Bokay,^ but his ex-
periments were not extensive enough to show the rapidity of the action.
I undertook some investigations in this direction but did not carry them

* Z. physiol. Chem., 37, 181.
■ Ibid., I, 157.

STABILITY OF L,ECITHIN. 889

far, as meanwhile the work of SchumoflF-Simanowski and Sieber* came
to my notice. In their extended experiments the rate of digestion by
several ferments in addition to that by the steapsin is satisfactorily
demonstrated. My method of work was in principle very different and
was intended to show the rapidity of acid liberation rather than the
amount liberated.

It is shown above that the pure lecithin in the form of emulsion has
little or no conducting power, while that mixed with small am.ounts of
decomposition products shows electrical conductivity in rather marked
degree. It was further shown that this was not increased by standing
or by warming on the water-bath. I found, in preliminary experiments,
that the emulsions, after being mixed with pancreas extracts prepared
in the laboratory, and incubated at 40° through a number of hours,
showed a greatly increased conductivity and increased acidity. But
in such cases part of the increased acidity is often due to the acids formed
by changes in the ferment mixture itself by enzymes or bacteria, and in
testing several laboratory extracts and commercial pancreas prepara-
tions in incubated solutions, I have observed this increased acidity and
increased conductivity. In working with the lecithin emulsions it was
necessary to guard against this source of error as far as possible by the
use of toluene or thymol in making up the ferment solutions, and even
the emulsions themselves.

In carrying out the tests the following solutions were used: First,
an egg lecithin emmlsion of approximately i per cent, strength in
thymolyzed water. This was found to have a conductivity, io^k^ =
2 . 66. At the same time a moderated active pancreas extract in thy-
molyzed water was made and this had a conductivity, io^k^q = 5.27.
A mixture of equal volumes of the two liquids gave io^k^q = 4. 03. This
mixture was kept in the thermostat at 40®, and portions were withdrawn
for tests from time to time with the following results:

Time. io^k«>.

o hours 4.03

3 " 412

21 '* » 6.77

45 " 10.16

69 " 12.50

96 " 13.02

In the 96 hours through which the experiment was carried the conduc-
tivity of the pancreas alone increased from io^icjq =5.27 to 6.34, or 20.3
per cent., while the increase in the mixture was 223 per cent.

At the same time there was a very marked increase in the total acidity
of the mixture determined by titration with o. i normal sodium hydro'xide
after the addition of alcohol. At the beginning 25 cc of the mixture re-

' Z. physiol. Chem., 49, 50.

890 J. H. LONG.

quired 1.6 cc. of the alkali, while at the end of the experiments over 6 cc.
were required. The increased acidity was about 4.5 cc. of o . i N alkali
which measures a marked degree of acid liberation, including, apparently,
part of the phosphoric acid. From the results given above for the in-
creased conductivity of the pancreas solution alone, it is probable that a
part of the developed acidity must be due to the ferment itself. This is
a point which is usually overlooked in fat-splitting experimicnts, and it
appears to have been overlooked in one of the results of Schumoff-Siman-
owski and Sieber,* in which the acid formed is evidently more than could
have been liberated, under the conditions of the experiment, from the
lecithin molecule.

Additional Experiments with Egg Lecithin, — Many of the tests made
above were repeated with egg lecithin obtained by somewhat different
processes, but it will not be necessary to go into the details of the results.
Some data from two cases only need be referred to.

In the first of them a product was obtained from boiled eggs by ex-
tracting with ether only. The eggs were boiled until thoroughly hardened,
and then the yellows were separated and ground up with clear quartz
sand without any preliminary drying. The mass so obtained was thor-
oughly extracted in the Soxhlet apparatus. The crude ether extract
was concentrated and the residue dried at a low temperature. It was
taken up with dry ether and precipitated with acetone in the usual way.
The precipitate was washed with acetone and dried in carbon dioxide.
A nitrogen determination gave 2.08 per cent. The electrical conduc-
tivity of an emulsion made up as before was found to be much lower
than with the former product, pointing, possibly, to the presence of smaller
amounts of dissociation products. The purified lecithin obtained by
treatment of the emulsion with acetone, when made up into a new emul-
sion, gave, likewise, a very low conductivity.

In the preparation of this last lecithin from hard-boiled eggs, the mass
left in the Soxhlet apparatus after ether extraction, was extracted through
several days with redistilled alcohol. The alcoholic solution was e\^po-
rated at a low temperature, leaving a considerable residue, in fact nearly
as large as that from the ether extraction. Most of this residue was
foimd to be soluble in absolute ether, which was somewhat remarkable,
in view of the preliminary treatment. On concentration of the ether solu-
tion and precipitation with acetone a light mass was secured closely re-
sembling that from the ether extraction. After repeated washings with
acetone it was dried and used as in the other case. A determination
of nitrogen gave a high result, viz.^ 2.34 per cent. This would suggest
the presence of a considerable amount of diaminophosphatide, and is
not at variance with the results of some of the experiments of Stem and

STABII.ITY OF LECITHIN. 89 1

Thierfelder.^ A phosphorus determination was not made because of
lack of material.

An emulsion was made with the portion not used in the nitrogen test,
and this had a concentration of 3.1 per cent. It was characterized
by a relatively high conductivity, and for equivalent concentrations
almost four times as great as for the portion extracted with ether alone.
The emulsions in both cases precipitate metallic solutions readily and
both have an acid reaction when tested directly. That the two extracts
are irarkedly different is shown not only by the different nitrogen con-
tents but by the great variation in conducting power. It is apparent
that the alcohol has brought a larger quantity of decomposition products
into solution than was the case with the ether.

Experiments with Brain Lecithin.

It is well known that the product termed lecithin, as obtained from
the brain, is usually a mixture of considerable complexity, and that some,
at least, of the constituents of this mixture are very unstable compounds.
This is well illustrated by a few commercial substances which are now
sold under the name "lecithin," and obtained from brain extracts. In
beginning some experiments with brain lecithin, I attempted to use a
crude product made by a local manufacturing firm, and which was ob-
tainable in quantity. This material dissolved readily in ether, and the
solution gave a good precipitate with acetone, but on attempting to re-
dissolve and purify it, it became dark. Even after repeated solution in
ether and precipitation by acetone the products secured remained dark
and failed, to yield a characteristic emulsion with water. The crude
extract had been prepared by extraction with the light hydrocarbon
sold as "hexane," and in some stage of the work had probably been
exposed to a high temperature, which brought about a marked decompo-
sition. The products of decomposition were evidently carried down
with the acetone precipitate in the first and following attempts at purifi-
cation. I mention these facts because they throw light on some of the
commercial "lecithin" preparations on the market, which in recent years
have been highly advertised as curative agents.

Not being able to utilize the commercial product, I dried down about a
kilogram of minced cilves' brains in a current of warm air, as recommended
by Erlandsen,* and extracted with ether thoroughly. This ether ex-
tract was concentrated and the solid pcMtion finally dried at a low tem-
perature in a rapid current of carbon dioxide. The pasty mass was taken
up with absolute ether, which left a residue amounting to about 60 per
cent, of the crude solid extract, and this new ether extract was concen-

* Loc. cit.

• Z. physiol. Chem., 51, 71.

892 J. H. LONG.

trated to a small bulk. This was precipitated with acetone, which fur-
nished a nearly white mass. The latter vms dried and after expelling
the acetone completely was redissolved in absolute ether and again pre-
cipitated with acetone in a large flask. The greater part of the super-
natant liquid was easily removed by pouring and the remainder was drawn
out by a current of carbon dioxide passed rapidly through the flask,
which meanwhile was immersed in a vessel of warm water. In this way
a good yield of a light yellow mass was recovered, with which the follow-
ing experiments were made. On analysis the phosphorus content was
foimd to be 3 . 76 per cent, and the nitrogen content 2.15 per cent., both
calculated on the anhydrous basis. This gives an atomic ratio,

P: N:: I : i .27.
The product is naturally a mixture of different phosphatides, and for the
purposes of the present examination is sufficient. The experience of
Stem and Thierf elder* shows the extreme difficulty of securing products
of constant composition, in quantity, from the analogous egg lecithin,
as well as the great losses which accompany re -solution and precipitation.

Effect of Heat — ^Tests were made here as with the egg product, and the
general results were essentially the same. However, this difference
was noted: in all the evaporations, whether of the aqueous emulsion or
of the ether solution, the brain product remained much lighter in cobr
than was the case with the other. This suggests a greater degree of sta-
bility, although from the apparently greater complexity the reverse
might be assumed. After evaporating the emulsions to dryness and making
up to the original volume with water, no perceptible change in color
followed, and no essential change in conductivity. From my various
experiments in this direction I must conclude that the lecithin compound,
such as is secured in the method of extraction outlined, is much more stable
than would be inferred from many statements in the literature. In the
process of preparation, that is, while the lecithin is mixed with other sub-
stances, it evidently changes readily, but when isolated is apparently
much more stable. In view of many observations I have made, I must
consider the statement of Bang' on this point as too strong.

Acidity. — The brain lecithin shows a greater acidity toward phenol-
phthalein than was noted with the egg product. As before, this is best
observed in the aqueous emulsion. To test the point quantitatively, an
emulsion was made containing in 100 cc. 4.52 grams. When directly
tested, 25 cc. of this, containing i . 13 grams, required 2.1 cc. of o. i nor-
mal alkali for neutralization. For a second 25 cc. mixed with an excess
of neutral alcohol, 7.4 cc. of the dilute alkali were required. This is a
strong degree of acidity.

* Loc. cit.

> **Erg^niss^ dfr Phynolo^ie/* VI Jahr^ii|^, p. 163,

STABlUTY Oif L«CITH1K. 893

A second 50 cc. of this emulsion was precipitated and washed with
pure neutral acetone, using 120 cc. in alL This acetone solution was
divided into equal portions, one of which was titrated for add and then
used for a phosphorus test, while the second portion was tested for nitro-
gen. In the titration, 6.5 cc. of o. i normal alkali wera required, and this
for a voltune corresponding to 25 cc. of the original emulsion. A good
test for phosphoric add was also secured. The portion reserved for
nitrogen was concentrated and decomposed in the usual manner for the
Kjeldahl determination. The ammonia obtained was 12.6 mg., corre-
sponding to nitrogen equivalent to 0.92 per cent, of the original leci-
thin in the volume taken.

The purified lecithin residue left after the acetone treatment was dried
in carbon dioxide, then freed from this gas by a current of air with gentle
warming, and made up with water to a volume of 50 cc. A portion
titrated was found to be perfectly neutral with phenolphthalein and
alkali, even after addition of alcohol.

Twenty cc. of the new emulsion furnished 0.0106 gram nitrogen,
which amounts to 1.17 per cent, of the anhydrous lecithin originally
present in the equivalent voliune.

Ten cc. of the emulsion gave 0.0105 gram phosphorus, corresponding
to 2 . 34 of the ledthin originally present in the equivalent volume.

It is evident that the treatment of the emulsion with excess of ace-
tone has resulted in precipitating ledthin, apparently, with considerable
loss, and also in changing the ratio of the nitrogen to the phosphorus.
In the original substance we had 3 . 76 per cent. P and 2.15 per cent. N,
with an atomic ratio of P:N:: 1:1.27. Here, in the "purified" emul-
sion, we have 2 . 34 per cent, of phosphorus and i . 17 per cent, of nitrogen
for the same recovered weight, or a ratio of P:N::i: i .17. In other
words, there is a relatively greater loss of nitrogen than of phosphorus,
as was found to be the case with the egg product, and the acetone treat-
ment may have then the effect suggested for the egg lecithin.

No quantitative determination of the weight lost on treating the brain
lecithin emulsion with acetone was made, but superficial observation
showed it was much higher than with the egg lecithin, and besides this
the supernatant liquid was not perfectly clear as in the other case.
We have then a rather marked degree of solubility in the acetone, and
this seems to be accompanied by the formation of some decomposition
products which are likewise soluble. It must be recalled, however,
that the total acidity of the acetone solution is not greater than the orig-
inal, but, in fact, a little less, which complicates any attempt at explana-
tion of what actually takes place in the treatment.

Electrical Condtictivity. — For these tests some of the same emulsion.

894 J- H. U)NG.

with 4.52 grams to 100 cc, was employed, and the observations were made
as before at 20°. The following results were obtained:

Cone, in loo cc. f^ao.

4.52 0.000721

2.26 0.000529

1. 13 0.000337

0.565 0.000195

0.283 O.OOOIII

In general, the values foupd are not greatly different from those for
the egg emulsions. It appears in this case also that precipitation with
acetone furnishes a product with much lower conductivity, as was shown
by precipitating 25 cc. with acetone, in a bottle, pouring off the acetone
and washing several times with fresh portions. The residue was dried
in an atmosphere of carbon dioxide and after expulsion of the gas was
emulsified with water and made up to 25 cc. again. The conductivity
was now found to be Kjjq = 0.000127, or about one-sixth of what it was
originally. The new emulsion was neutral in reaction. The great change
must be due to the loss of decomposition product, rather than to the
loss of lecithin itself through solubility.

Some of this last emulsion was allowed to stand about two weeks and
examined again to detect a possible increase in conductivity by hydroly-
sis through long contact with water, but no such increase was found,
and this again speaks for the comparative stability of the substance.

Precipitation by Salts. — It was found that emulsions of the egg lecithin
are readily precipitated by solutions of several salts and in a manner
quite distinct from that described by Koch.* Similar experiments were
made with the brain lecithin emulsions, and with the same general result,
which will not be given in detail here, as the observed relations are made
the subject of fuller investigations. Since the completion of the experi-
mental part of this paper, an article by Hoeber* has come to hand, in
which the author shows that carefully purified egg lecithin made up into
emulsion yields precipitates with many neutral salts without regard to
valence of the metallic ions. This is in full accord with the results of my
experiments.

Action of Light. — In various methods of preparation of lecithin given
in the recent journal literature, much is said about keeping the product,
as far as possible, in the dark. In some of my experiments I have done
this, while in others no such precaution was taken. To test the behavior
of light, I have made emulsions of both egg and brain lecithin and allowed
them to stand in stoppered flasks through periods of two weeks or more
in a well-lighted room with south and west exposure, and part of the

* Loc. cit.

' BeUrAge tur. chem. Physiol, und Path., zi, 35.

EMULSIONS OP LECItHlN. 89S

time in direct sunlight. I have hot observed in any of the flasks a change
of color, change in acidity, change in conductivity, or change in behavior
toward weak salt solutions, from which I am forced to conclude that
the light effect, if present at all, is very slight.

Summary of Results.

In^this work it has been shown that :

J . Emulsions of egg and bmin lecithin are comparatively stable with
respect to temperature. Increase of temperature, or long-continued
heating of the emulsions does not appear to increase the dissociation as
measured by acidity or conducting power. The action of light on the
emulsions appears to be very slight.

2. Lecithin emulsions have an acid reaction which is marked. On
precipitating the emulsions with an excess of pure acetone the residues
left, on being again brought into emulsion form with water, are neutral.
Precipitation of lecithin from ether solution by means of acetone seems
to furnish a product which becomes acid when treated with water. The
acetone precipitation from water effects also some decomposition, shown
by change in the P: N ratio.

3. The electrical conductivity foimd in the emulsions suggests the
presence of acid or basic groups, but after purification by acetone the
conductivity is so much reduced as to indicate that this phenomenon as
observed is not due to the lecithin itself, but to decomposition products.
It is likely that many of the reactions assumed to be characteristic of
lecithin are due to hydrolysis or other products.

4. Emulsions of both brain and egg lecithin are readily precipitated
by weak salt solutions. No relation between the precipitating power
and the valence of the metallic or acid ions of the salts is apparent. The
extraction of lecithin from emulsions is aided by the addition of salts.

My thanks are due to my assistant, Mr. Frank Gephart, who has made
the above lecithin preparations.

N0RTHWB8TBRN University Medical School,
Chicago, February, 1908.

ON THE BEHAVIOR OF EMULSIOirS OF LECITHIN WITH
METALLIC SALTS AND CERTAIN NON-ELECTROLYTES.

By J. H. I«ONO AND Frank Gephart.
Received March 13, 1908.

Although lecithin may be obtained, like many other fats, in a crystal-
line condition, its behavior is ordinarily colloidal, and when mixed with
a large quantity of water its relation, physically at least, to the colloids
is very marked.

Among the properties of the colloids which must be regarded as of the

896 J. H. IX>NG AND PRANK GEPHART.

highest interest, the behavior of their solutions or suspensions toward
salt solutions has attracted xecently much attention. Hofmeister^ was
among the first to call specific attention to the precipitating action of
many solutions on certain proteins, and these suggestions were followed
up by extended investigations of Pauli,' Spiro,' Hardy,* and others. The
work of Pauli was especially valuable in showing the order of the precipi-
tating power followed by the difierent cathions, and the modif)dng influ-
ences of the anions, while in the later article of Spiro, data are presented
to bring these phenomena in comparison with others and so lead to a
theory of the processes.

Considering the lecithins as colloids, Koch* has attempted to show
the relations between valence of cathions and the precipitation of weak
emulsions of this substance by solutions of numerous salts. According
to this author the behavior of lecithin emulsions is in many respects
analogous to that of the true colloids just referred to, but the precipita-
ting power of the salt solutions on the emulsions seems to be confined
to certain groups only. Mono- and trivalent metals are said to be ¥ritl]
out action, while the solutions of the common bivalent metals, Mg, Ca,
Sr, Ba, Co, Ni, Fe, Zn, Cd, Cu, etc., are active precipitants. Add solu-
tions (H ions) were also fotmd to act as precipitants, while a number of
anions, investigated in their combinations with metals of the first group,
were found to be without specific action. In this paper Koch notes
further the behavior of mixtures of salts in the precipitation of lecithin
and finds that certain amounts of the mono- and trivalent metals neu-
tralize the precipitating action of the bivalent metals. This is especially
interesting in the case of ferric chloride, which was found to prevent
precipitation by calcium nitrate.

While experimenting on the extraction of lecithin from certain solu-
tions and emulsions on the market as remedies, or "tissue builders,"
we made the observation that the ease or completeness of extraction is
very much influenced by the character of mineral n^itters or salts pres-
ent at the same time. Emulsions in water or glycerol which gave up
no lecithin directly to ether or chloroform were found to extract perfectly
by a shaking-out process after the addition of various salts. In follow-
ing up the question it was found that these salts all produced a more or
less perfect precipitation in water emulsions, and that this separation
by precipitation, or salting out, must evidently precede the actual ether
extraction. As some of this experience seems to be the reverse of that

* Archiv. exper. Path. u. Pharm., 25, i; 27, 295; 28, 210.

* Beiir, chem. Phys. u. Path,, 2, i; 3, 225; 5, 27; 6, 233.
■ Ihid.f 4, 300.

* Z. physik. Chem.y 33, 385.
' Z. physiol. Chem.f 37, 181.

EMUWIONS OF LECITHIN. 897

reported by Koch, we were led to give it a fuller study with a larger
number of substances. Before the results were completely worked out,
however, an interesting paper by Hoeber^ came to hand in which he re-
ports observations leading to the same conclusions which we had reached,
and which fail to confirm the findings of Koch. As our experiments
cover a somewhat wider range than those of Hoeber on this particular
point we give them in full, although as far as the simple question of the
precipitation of lecithin emulsions by monovalent metals is concerned
it might not be considered necessary.

In a previous paper by one of us,' on certain properties of lecithin, at-
tention was called to the fact that acetone precipitates from emulsions
of this substance a product which is different from the original and in
some respects purer; at any rate it is free from the add reaction usually
found in the lecithin from other processes. In our experiments, given
below, we have used both kinds of lecithin, with practically the same re-
sults, and have tried emulsions of various strengths from 0.005 ^ to
0.05 N. For all the reported tests, however, we have used the weakest
emulsion, which contaned 4 grams to the liter, as the molecular (here
normal) weight of the egg lecithin employed is about 800. The prepara-
tion of this lecithin is described in the last paper referred to. It may be
added that essentially the same results were secured with some brain
lecithin, described in the same paper. The weakest emulsion used is
so dilute that it may be filtered, yielding an opalescent filtrate in which a
precipitate is readily visible. In the first series of tests we used in each
trial 5 cc. of the 0.005 lecithin and i and 5 cc. of the salt solutions in nor-
mal strength, where the solubility permitted. With salts of low solu-
bility satumted solutions were employed. In the table below "t" indi-
cates increased turbidity, "p" actual precipitation, while by '^op" a slight in-
crease in opalescence is indicated. The action of a few acids is included
in the table given below.

The table shows the wide range of salts which possess the power of pre-
cipitating the lecithin emulsions, and the most marked difference which
may be noted is in the time required to cause actual subsidence of the
several precipitates. We have tried to distinguish between opalescence
and the appearance of turbidity, but the distinction in many cases is
far from sharp. It will be noticed that the action with acetic acid is
weak while with boric acid no effect whatever was observed. The pre-
cipitates formed by calcium, strontium and barium salts are at the out-
set heavier, apparently, than those formed by salts of the alkali metals,
but after 24 hours the differences disappear and all the precipitates
settle out. In most cases these precipitates do not appear to be true

* Beitr. chem, Phys. u. Path., ii| 35 (Dec., 1907).

* Long, Tins Jousnai^, preceding artide.

898

J. H. LONG AND i^RANK GEPHART.

chemical unions, as the lecithin may be separated by an extiactkni pro-
cess with ether, as referred to below, and as is shown also by the foUow-
ing behavior. After subsidence of the precipitates the greater part of
the supernatant liquid may be poured off. This liquid carries, of course,
the larger part of the added salt. If the same volume of distilled water
is then poured over the precipitate and the mixture shaken a new emul-
sion is formed which, however, is denser than the original emukkm,
and, besides, is not stable. A separation soon follows, due apparently
to the presence of small amounts of salt not removed in the decantatkn.
On repeating these operations once or twice, and thus removing all the
salt, stable emulsions like the original are secured.

Table I. — PREcn>iTATioN op o.cxjs N LBcriHiN.

At onc«.

In 2 hours. In 24 hoan.

SaUfl us«d.

I cc.

NaCl op

Ka op

NH,C1 op

NaNO, op

KNO, op

NH,NO, op

Na,S04 op

K^O^ op

(NHJ^O, op

CaCl,

SrCl,

BaCla

Ca(NO,),

Sr(NO,).

Ba(NO,),

Fe(NHJ,(SOJ.

FcCl,

Fe^'(NH.)(SOJ,

Tl^O* op

Pb(NOJ, op

Cda, op

CUSO4 t

Ha p

HNO, p

H,SO, p

HAH4O. p

HC^,0, t

H,BO, o

5CC.
op
op
op
op
■op
op
op
op
op

I cc.

op
op
op

t
p
p
p
p

p
t

5CC.

p
p
p
p
p

p
t

I cc

p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p

50c

p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p

The behavior with ether is indicative of the same thing, that is, the
physical rather tlian the chemical nature of the precipitations. To show
this a 0.025 N emulsion from the same lecithin as before was used, and
to it were added certain volumes of ether and salt solutions. After shak-
ing and allowing the mixtures to stand the ether layer which gradually

UMUWIONS O^ LECITHIN. 899

sepaxated showed always some color, from the lecithin, even with the
weakest dilutions of the salts, as shown in the table below for sodium
chloride solutions. Essentially the same results have been found for
other salts, and need not be repeated. The color of the ether layer meas-
ures roughly the amount of lecithin separated, within certain limits,
but it is evident that the amount of salt required to secure the maximum
color is soon reached in the successive trials. This amount of salt is
small. As the ether becomes colored the water layer below clears up
perfectly.

Tabls II, — Epfbct op Mixing Equal Volumbs op Lecithin Emulsion Salt Solu-
tion AND Ether.

atnngth of salt solution. Color of ether layer.

0.00500 N very slight color.

0.00625 N slight, but increased.

0.00830 N more distinctly colored.

0.01250 N decidedly yellowish brown.

0.02500 N marked brown color.

0.05000 N, . . ,

0.07500 N

o.iooooiV

o. 12500 N

It it t t

1 1 1 1 It

1 1 1 1 it

No increase of color seemed to follow after using the 0.025 N salt solu-
tion, and this amount appears to be sufficient to separate the whole of
the ledthin, as was found in some experiments in which larger volumes
were taken for the trials. In the above table the amount taken was 5
cc. of each liquid, but with the larger volumes it was possible to recover
most of the lecithin in the ether layers. It was found also that the leci-
thin was separated as such, and not as a salt or combination. This was
shown clearly in two experiments in which sodiuin chloride and barium
chloride were employed in relative excess. The supernatant ethereal
layers contaiaing the lecithin were removed, evaporated, and the resi-
dues ignited. Only a very minute trace of chlorine was found in either
case, and no barium in the second case by the sulphate test. These two
tests represent typical cases, as the precipitation by the sodium salt is
relatively slow, while that by the barium salt is rapid.

The behavior of salts of cadmium and lead in this respect is interesting.
Thudichum* has shown that these and several other salts give true chem-
ical precipitates with lecithin in alcoholic solution, and it is important
to note their action with emulsions. In making the actual tests it was
soon recognized that as far as the simple precipitation is concerned,
lead and cadmium salts do not differ essentially from the others of the
bivalent group, as Koch' points out. In making the precipitates in pres-

^ Di4 chemische KonstihUion'des Gehims des Menschen und der Thiere.
' Lac. cii.

900 J. H. U3NG AND I^RANK G^PHART.

ence of ether, and after a time decanting the layer of this substance
which separated, we found that not more than the minutest traces of the
heavy metals had gone into solution. These metals, as all the otheis,
remain in the aqueous IsLytr while the lecithin dissolves in the ether.
In making a test for lead in the residue from the evaporated ether, nothing
more than a faint coloration with hydrogen sulphide was secured at any
time, and this might well come from the sUght solvent action of the water
held by the ether. The results with some other metals were the same,
indicating that the lecithin must exist wholly in the colloidal form in the
water solution, in which condition it does yield a true metallic combina-
tion.

A few experiments have been made with some salts in which one or
both ions are relatively weak. In this list are included mercuric chloride,
cyanide and acetate, aluminum acetate, chromium acetate, and the sulphates
of both metals. With the three mercuric salts the precipitation reactions
were extremely faint, if at all present, and on adding ether practically no
coloration appeared. This behavior is especially plain in the case of
mercuric cyanide, from which it may be inferred that the degree of dis-
sociation of the salts may have some bearing on the problem. With
the acetates of chromium and aluminum, very weak reactions were ob-
served. These salts were made by double decomposition between the
pure sulphates and lead acetate, and held a trace of lead in the water
solutions. Solutions obtained by dissolving the washed hydroxides in
acetic acid were even more inert with the lecithin, but these solutions
were weaker. With the two sulphates good reactions were noticed.

In carrying out the last experiments it was observed that the age of
the lecithin emulsions has some influence on the results. Old emulsions
appear to be much less reactive than fresh ones, as we noticed in a num-
ber of the later tests. With an emulsion which had been prepared several
days, no reactions were obtained with the acetates of aluminum and
chromium, and rather weak tests with the sulphates. The results
reported above were obtained with fresh emulsions in general, and we
have carried out no tests to explain this exceptional behavior.

In this connection the action of non-electrolytes is interesting. It
was stated above that ether fails to extract lecithin from glycerol emul-
sions, and in following up this point glycerol, glucose, saccharose, urea
and egg albumen were added in molecular proportions to the same kind
of emulsion used in the other tests. With the glycerol no increase of
turbidity was observed, while with the others it is possible that there
was a little increase in the opalescence. According to Pauli, as quoted
above, the sugars, urea and other non-electrolytes have no action on
the true colloids. With some specimens of urea a slight precipitate
may be obtained, but this is due, doubtless, to the traces of sulphates

EMULSIONS OF LECITHIN. 90I

present in much of the product as obtained from chemical dealers. With
pure urea the reaction is extremely weak. On adding ether to the mix-
tures, and shaking, as before, no color was found in this when it came
to the surface on standing. These compounds cannot have, therefore, the
same action on the emulsion which was noted with the salts. When,
however, to each one of the mixtures a few drops of salt solution were
added, and the tubes shaken, an immediate coloration of the ether layer
followed as with the regular electrolyte solutions, described above. The
extraction with ether is again seen to depend on preliminary salt action,
and the presence of the non-electrolytes does not appear to inhibit this.
It is possible that some of these substances form loose compotmds with the
lecithin which are not soluble in the ether, but which are readily decom-
posed by the addition of salts. A number of such compounds have been
described,^ but in most cases tangible proofs of their existence are lack-
ing. From various physiological reactions the existence of combina-
tions between lecithins and certain toxins has been much better estab-
lished.

An important combination, however, has apparently been completely
overlooked, and that is, the complex formed by the union of lecithin and
bile salts. Lecithin in the form of emulsion is dissolved perfectly in an
aqueous solution of bile salts, and in considerable proportion. The com-
bination formed seems to be remarkably stable, and not readily decom-
posed in such a manner as to give up the lecithin to the usual solvents.
We are at present engaged in a fuller study of this complex, which is in-
teresting from several standpoints, especially in the separation of leci-
thin from bile.

Results. — ^It has been shown in these investigations that weak emul-
sions of lecithin are precipitated by a large number of solutions of salts
and adds and that the completeness of precipitation does not appear
to bear any relation to the valence of the cathions concerned. Within
each group, the alkali group for example, we have not been able to dis-
tinguish any characteristic diflFerences. It seems to be true, however,
that the precipitation is in some way related to the degree of dissocia-
tion of the various compounds. While, for example, the ordinary adds
and tartaric adds are very active, acetic add is weak and boric acid quite
inert. The mercuric salts tested were all weak, and the activity decreased
from the chloride through the acetate to the cyanide, which is practically
inert.

It has been found also that ether and similar solvents have a very
slight extracting power for the pure aqueous emulsions, which in most
cases is scarcely appreciable, and that after the addition of salts to the

* See convenient literature references in paper by Ivar Bang in Ergebnisse d^
Physiologic, 6, 163,^

902 J. T. WILLARD. ^

emulsions the lecithin is immediately taken up by the solvents. Thii
action, which is best shown with ether, is related to the precipitating
power of the salts, and it appears necessary to break up the emulsions
before extraction is possible. Mercuric cyanide atid boric add do not
precipitate the emulsions and after treatment with ether no lecithin is
dissolved by the latter.

Certain non-electrolytes tested do not appreciably precipitate the emul-
sions, and their presence does not aid the solution by ether, but the
addition of traces of salt solutions to the mixtures brings about an inmie-
diate solution of the lecithin, even when great excesses of the non-elec-
trolytes are used. This behavior has certain practical applications.

Finally, attention is called to the peculiar behavior of bile salts.

NORTHWBSTBRN UNIVBR81TT MBDXCAL

School, Chicago.

ON THE OCCURRENCE OF COPPER IN OYSTERS.

By J. T. WiLLA&D.
Received March 19, 1908.

Last spring the attention of Dr. S. J. Crumbine, secretary of the Kansas
State Board of Health, was called to some cases of illness following the
use of fresh oysters in which these were suspected to be the cause. Two
samples were sent to the writer for examination. They had a distinctly
greenish blue color, and qualitative tests showed the presence of copper.
The oysters also possessed a noticeable coppery taste. Quantitative
determinations of the amoimt of copper were made. One of the sam-
ples was found to contain 0.0437 per cent, of copper, or 0.302 per cent,
calculated on the dry substance; the other contained 0.0324 per cent.,
or 0.211 per cent, in the dry substance. As copper has been recognized
as a constituent of many species of moUusks it seemed desimble to test
other samples. Two other samples of fresh oysters and six samples of
canned oysters, sold imder the name of Cove 0)rsters, were examined
and in every case copper was found to be present. As the oyster season
was practically at an end at that time, further investigation was post-
poned until October, when analyses were made of a considerable number
of samples secured chiefly with reference to determination of water con-
tent. In all 34 distinct samples were analyzed. The results are shown
in the following table.

With the exception of those marked as bulk samples these oysters
were placed in glass jars as they were taken from the shells. In most
cases the amount of liquor present was too small to determine the copper,
but in other cases where the liquor was examined copper was found pres-
ent, and in no instance was copper absent from the oysters although in
sample No. 12,117 the amount was very small. The uniformity of the

COPPER IN OYSTERS,

903

Tabls Showing Coppbr in Oystsrs.

Per cent, of copper in

Serial
number.

2.094
2,095

2,096
2.097
2,098
2,099
2, 100
2, lOI
2, 102
2, 103
2, 104
2, 105
2, 106
2, 107
2, 108
2, 109
2, 1 10
2, III

2, 112

2,113
2,114
2, 115
2, 116

2, 117
2, 118

2,119
2, 1 20

2, 121
2, 122

2, 123
2,124

2, 125
2, 126

2. 127

Place of
porchaae.

Manhattan

< f
Washington, D.C.

Philadelphia

1 1

< <

( I

< (

New York

f I

1 1

I (

I <

Baltimore

Name.
Bulk, '* Booth's"

1 1

t <

1 1

< (

< <

t i

Liquid.

■ • .
0.0056
0.0016

Bulk, No. I
Bulk, No. 2
Curryoman
Rockaway.
York River
Hampton Bars
Cape Cod
Blue Points
Coam River
Blue Points
Lynnhaven Bay
Rockaways
Clarke River
Tucker Salts, Bame-

gat Bay
Rockaway, Partial

Salts
Chituque Salts, Jer-
sey Coast
Maurice River Cove,

Baltimore Bay
Cedar Rock Salts,

Jersey Coast
Maurice River Cove,

transplanted
Cape Cod
Lynnhaven, Va.
Rockaways, Long

Island
Blue Points, Long

Island
Sea Puits
Easton Bay, Md.
Lynnhaven
Chester River
Horn Harbor
West River, Md.
Swamp Point

0.0048
0.0048
0.0024
0.0064
0.0048
0.0016

• • • • •

• • • • •
« • • • •

• • • • •

• • ■ ■ •

• • ■ • •

• • ■ ■ ■
■ • • • •

• • • • •

Meat.
0.0072
0.0083
0.0084

• •••••

0.0056
0.0060
0.0056
0.0056
0.0068
0.0064

• •••••

• ••*•■

• •■•••

• ■••••

• ••••*

• •««■•

• •■•••

• ■••••

• ••••■

• •••■•

• ■••••

Sample.

Dry
sample.

0.0072

0.087

0.0072

0.079

0.0071

0.079

0.0062

0.059

0.0084

0.085

0.0022

0.023

0.0052

0.055

0.0055

0.043

0.0042

0.034

0.0061

0.058

0.0058

0.040

0.0041

0.029

0.0058

0 044

0.0044

0.034

0.0104

0.064

0.0080

0.060

0.0008

0.006

0.0076 0.052

0.0032 0.018
0.0068 0.048

o . 0008 o . 006

0.0016 0.012

0.0048
0.0032
0.0008

0.032
0.021
0.005

0.0052 0.047

0.0092

0.071

0.0072

0.051

0.0076

0.087

0.0164

0. 170

0.0084

O.IIO

0.0028

0.025

0.0048

0.052

0.0028

0.038

presence of copper warrants the conclusion that that metal is a normal
constituent of oysters. The much larger amotmts in the first ones ex-
amined should perhaps be regarded as abnormal and may have been
due to special conditions, the nature of which is tmknown. It is not

904 NOTES.

improbable that, especially with susceptible individuals, those oysters
containing the larger quantities of copper might be a cause of illness.

In respect to the mode of analysis, it n^ay be of interest to state tliat
in most cases the oysters were digested with a minimum of sulphuric
acid, as in the Kjeldahl method for the determination of nitrogen, the
clear solution was diluted and the copper deposited electrolytically.
Check tests of the reagents proved them to be free from copper.

Kansas Statb Agricultural Collbob,
Manhattan, Kans.

NOTES.

Notes on Mr. Keen's Paper ^ on the Volumetric Determinaiion of Zinc,

Mr. Keen disarms criticism by disclaiming any great originality for the
method he describes. Unfortunately he has not selected the best of
the old methods, and some of the things he advises are likely to cause
trouble.

The .methods given for preparing the ferrocyanide solution and for the
titration are the very excellent ones described by Dr. Low.

The method given for standardizing is complicated and unreliable.
A much simpler one is to partly dissolve a single large piece of high-grade
spelter in dilute hydrochloric acid, dilute the solution so that it will con-
tain about 5 grams of zinc per liter and determine the zinc by any relia-
ble gravimetric method. I prefer to determine the zinc as pyrophos-
phate, as the method is simpler than most, and I have found it extremely
accurate. If not more than three-quarters of the piece of spelter is
dissolved the solution will contain nothing but zinc, and consequently
needs no purification. Two or three liters can be made up at once and
used as a standard for many months. If extreme accuracy is required,
the zinc should be determined in weighed portions of the solution and
weighed amounts be used for standardizing.

In the standardizing and actual analysis the volume of the solution,
temperature, amount of free acid and of ammonium chloride, the indi-
cator and the method of using it should be kept within very narrow limits
or the results will be unreliable. The effect of ammonium chloride on
the amount of ferrocyanide necessary is usually neglected but it is quite
important.

Sampling Spelter. — The method proposed is very unreliable and likely
to cause errors. It has been condemned by the International Commit-
tee at the Congress of Applied Chemistry at Rome. The method they
recommend is by far the best, i. e., to saw the slabs and use the sawdust
for a sample. It is best to saw each slab entirely in two, it must at least
be cut to the middle each time. An ordinary band saw, such as is used
» This Journal, 30, 225.

NOTES. 905

for wood, answers perfectly; the saw should have rather fine teeth and be
run at a high speed. The feed, of course, must be slow.

Afuzlysis.— Common western spelter containing one or two per cent.
of lead will not all dissolve in hydrochloric acid, and the residue is likely
to contain both zinc and iron. It should be filtered out, dissolved in
nitric acid, evaporated with sulphuric, the lead sulphate filtered out
and the filtrate added to the main solution.

The methods of separation proposed are nearly all slow and the accu-
racy of some, at least, is very doubtful. The method* recommended
by the Committee of this Society on Unifonrity of Zinc Analysis is easier,
quicker, simpler and far more accurate than the one proposed, and it is
applicable to all zinciferous materials.

Aluminum Alloys, — In many cases the presence of aluminum does
very seriously afifect the ferrocyanide precipitation.
- nbw jBRSBT ziifc Co., Geo. C. Stone.

71 Broadway, New York.

The Detection and Identification of Manganese and Chromium in the
Presence of Each Other, — *'To the cold, dilute nitric or sulphuric add
solution of the substance or mixture to be tested is added one or two
cubic centimeters of a silver nitrate solution of the ordinary concentra-
tion, then a relatively large amount (two to five grams) of solid potas-
sium persulphate, and the whole carefully heated until the evolution of
oxygen due to the decomposition of the persulphate is practically over.
By this means the manganese is converted into permanganic acid and
the chromium into chromic acid. The permanganate color shows itself
first and is usually best seen during the first few moments of heating.
In order to detect the chromium present (the chromate or dichromate
color being usually obscured by the permanganate color), the cooled solu-
tion is shaken with one-fourth to one-third its volume of ether, hydrogen
peroxide added in excess, and the mixture well shaken. This decom-
poses the permanganate with evolution of oxygen and converts the chro-
mic add into perchromic acid, which dissolves to a blue color in the ex-
cess, of. etl^er. Acetic ester may sometimes be used to advantage in place of
ordinary ether." The method is delicate, easily performed in an ordinary
test-tube, and convenient in having no filtrations or fusions. In a course
of Qualitative Analysis it may be tried either upon the original material
or upon the proper group precipitate as one wishes. Halides should be
absent and in case much manganese is present only small amounts of the
substance analyzed should be taken, otherwise the manganese tends to
be predpitated as manganese dioxide instead of bdng converted into
permanganate. W. J. KarslakE-

UifivBRSiTT OP Iowa, Iowa City, Ia.

* This Journal, a8, 262 (1907).

9o6 NEW BOOKS.

ITEW BOOKS.

OrgAnic Chemistry for Advanced Students. By Juuus B. Cohbn, Ph.D., B.Sc..
Professor of Organic Chemistry in the University of Leeds. New York: Long-
mans, Green & Co. 1907. pp. viii + 632. Price, $7.00.

The purpose of this book is to supplement the ordinary text-books
of organic chemistry by giving more extended surveys of selected topics
of special interest. This is, of course, not a new or tmoccupied field,
for in this country we have Lachman's 'Spirit of Organic Chemistry," and
in Germany Ahrens's **Vortrage." The new book covers much more ground
than the former, but is not so comprehensive in scope as the latter. It
is a publication of the subject-matter of lectures which the author faas,
been delivering to his senior students at the University of Leeds, and its
field will 'appear from a glance at the Table of Contents. The chapters
are as follows: Historical Introduction, Isomerism and Stereoisomer-
ism, Stereochemistry of Unsaturated and Cyclic Compounds, Stereochem-
istry of Nitrogen, Isomeric Change, Steric Hindrance, Condensation,
Carbohydrates, Fermentation and Enzyme Action, Purine Group, Pro-
teins, Benzene Theory, Terpenes and Camphors, and Alkaloids. A se-
lect bibliography at the close of each chapter lists the more important
works on the subject, in addition to which there are full references through-
out the text.

The material has been chosen with care and discrimination and is pre-
sented clearly and concisely. The publishers' work is well done, t)T)e
and paper being very satisfactory.

The book is a very useful contribution to the literature of the subject,
and should be warmly welcomed by all advanced students and teachers
of organic chemistry, for it gives in compact form a conspectus of recent
progress along lines of particular interest. Thus, the chemist who has
had but little time to keep i;p with recent investigations in such mat-
ters as the stereochemistry of the sugars, the synthesis of terpenes, of
alkaloid^, of polypeptides, enzyme action, tautomerism, and the Hke,
will find here the desired information.

The reviewer mo§t heartily commends the work to the attention of all
interested in organic chemistry. Marston Taylor BogbrT.

Kurzes Lehrbuch der Organischen Chemie. Von • William A. Noy^, Professor

; der Chemie an der Universitat Illinois. Mit Genehmigung des Vcrfasscrs ins

., Deutsche Uebertragen von Walter Ostwald, und mit einer Vorrede von Pw-

PSSSOR WiLHELM OsTWALD. Leipzig: Akademische Veriagsgesellschaft, m. b. H

1907. 8^, xxiv + 722. Price, bound, lo.So Marks.

"The* justification for adding another volume to the long list of Gcr-
ifiari text-books on organic chemistry, and that a translation, must -be^
found in the independent treatment and originality, with which tbc
author conceived and carried out his work. In view of the enonnous
mass of facts of organic chemistry, it is of decisive importance for every

NEW BOOKS. 907

beginner that he should n:aster as soon and as thoroughly as possible
the fundamental points of view connecting these facts systematically.
By the original arrangement of its material, the m.ass of facts has been
treated systematically in so logical a way, that the student of this small
volume will be able to continue his studies without any fear of losing his
way in the thickets of organic chemistry. The author has also solved
the difficult problem of suggesting to the student the more advanced
fields to be conquered without discouraging him from attempting them.
This result has been attained by the great clearness of the style of pre-
senting the subject "

This free translation of a part of Professor Ostwald's introductory
statement to the translation of Noyes's "Organic Chemistry*' explains
sufficiently the signal honor that this book has received of being ren-
dered into German with the approval of the master mind, in Germany,
of the science of the teaching of chemistry.

The writer of this review would recall the fact that the most conspic-
uous feature in the presentation of material in Noyes's book lies in the
fact that the aromatic series of compounds is treated systematically
with the aliphatic series. In the first descriptive chapters, all the differ-
ent classes of hydrocarbons, including benzene and related compounds,
are first considered. This arrangement makes it possible to present
logically and without unnecessary duplication the reactions of closely
related groups of compounds of the aliphatic and the aromatic series,
such as the alcohols and the phenols, the amines and the anilines, and so
forth. It also makes it possible to discuss at an early stage some of the
numerous reactions leading from compounds of one series to those of the
other series.

The more critical study of the compounds of the two series is obliter-
ating more and more the lines of any fundamental differences between
the two and is recognizing, instead, differentiation in reactivity of groups
of compounds of analogous structure common to both series; for instance,
it may be recalled that acetacetic ester unquestionably shows the be-
havior of a phenol, both in its tendency to form salts and in the reactivity
of its methine group ( : CH — , in the enol form) towards halogens, nitrous
acid, diazobenzene, etc. A striking similarity in constitution is obvious
if we accept Kekul6's structure for benzene. Vice versa, this parallel
suggests that even monophenols form tautomeric compounds,

O OH O

/Nh h/Nh h/\

H^H H^^ H^yH

H, H H

9o8 NEW BOOKS.

the ktter derivatives of a dihydrobenzene, which would account in the
simplest way for the comparative ease with which phenol rings are oxid-
ized and opened. It is well-known that the analogy in the behavior
of 1,3-dihydroxy-and 1,3,5-trihydroxybenzenesandthatof 1,3-dicarbonyl
derivatives is even more striking.

The puzzling and central fact that benzene and its derivatives appear
to be more stable in the unsaturated condition (Kekul^'s formula) and re-
act in most cases as saturated compounds (in the ring) is also not without
parallel in the aliphatic series; for instance, the organic adds certainly
resist reduction in the tmsaturated carbonyl group almost as vigorously
as do certain benzene compounds; even when their carbonyl group ab-
sorbs certain reagents, yielding temporarily ortho derivatives, there is a
rapid reversion to the more stable tmsaturated carbonyl group — (in the
acid esters, amides, etc.). At the same time this same unsaturated
carbonyl group has an immistakable effect on the activity of the hydro-
gen of the immediately neighboring groups, e. g., such groups as CH and
OH — ^much as the activity of the hydrogen atoms adjacent to the tmsat-
urated groups of the benzene nucleus is enhanced, and when we have two
such neighboring unsaturated groups in aliphatic compotmds, ^. g., in
the 1,3-dicarbonyl series, the analogy is even more marked. Again, itt
have all degrees of gradation in such relations — the carbonyl group in
aldehydes is readily reduced by hydrogen, but towards very many other
reagents shows again the same tendenc)'^ to reversion to the unsattuated
condition as a stable form. Benzene derivatives are likewise not all
equally resistant to reduction and saturation, as shown by Baeyer, Bam-
berger and others for the phthalic acids, the naphthalenes and similar
compounds. Stability in the so-called unsaturated condition may there-
fore well be simply a question of peculiarities of structure and energy
content, common to all fields of chemistry. It maty not be amiss to recall
parallel cases of the resistance to saturation of tmsaturated compounds
in inorganic chemistry, as shown by the phosphines and arsines at ordi-
nary temperatures, and by ammonia above 400*^.

The arrangement used by Noyes, treating the aromatic compounds
with the aliphatic ones, appears to the writer therefore logical, both
pedagogically and scientifically; exhaustive studies of the relation be-
tween the two series and especially of the question of stability and reac-
tivity of unsaturated molecules, as made in the investigations of Bae)xr,
Nef and Thiele, may solve that perplexing problem of the structure of ben-
zene in the simplest of all ways, by demonstrating that there is no real
benzene problem, but a broader, greater problem of equilibrium condi-
tions of tmsaturated valences. J. STiSGUTZ.

BxerciBes in Slementary Quantitative Analysis for Students of Agrkuttore.

By AzARiAH Thomas LmcouN, Ph.D., and James Henri Walton. Jk.,

NEW BOOKS. 909

Ph.D. New York: The MacmiUan Co. 1907. 8vo. pp. xv + 218. Price,
$1.50 net.

The work includes introductory exercises in gravimetric analysis,
acidimetry and alkalimetry, permanganate and dlchromate titrations,
iodimetry, stoichiometryi and a section on agricultural analysis cover-
ing the examination of milk, butter, cereals and feeding materials, fer-
tilizers and soils.

The book is well written and contains a number of good illustrations.
It will be welcomed by those beginners in agricultural analysis who have
been obliged to use the methods of the Association of Official Agricultural
Chemists in bulletin form in lieu of a text-book. The procedures are
clearly and explicitly described and the explanatory notes are generally
good. The numerical data selected to illustrate normal composition
could in some cases be improved, but the only figures likely to be seriously
misleading are those for starch in grain products on page 121.

The failure of the authors to make use of the conceptions of ionization,
mass action and solubility product in the discussion of inorganic reactions
and the entire omission of electrolytic methods are unfortunate in a text-
book which is likely to represent the sole training in quantitative analy-
sis of many of the students who use it. These, however, are omissions
which may be supplied by the teacher and which the authors will proba-
bly correct in a subsequent edition.

The book will fill a real need in the case of the agricultural student
for whom it is especially intended and will be found useful and sugges-
tive to many others. It is commendably free from typographical errors
and its general make-up is excellent. H. C. Sherman.

Testing Milk and Its Products. By Faiuungton and Wall. Madison, Wis.:
Mendota Book Co. 1908. pp. 292. Price, $1.00.

The authors have revised their useful book. The present constitutes
the eighteenth edition, the first edition having been issued over ten years
ago. Considerable matter has been added, which includes new methods
that have come into recent use. L. L. v. a

The Chemistry of Commerce. By Robbrt Kbnnbdy Duncan. Harper Brothers.
Price, $1 . 50.

It is perhaps questionable whether ''Chemistry of Commerce" should
be reviewed in a scientific journal like that of the Chemical Society, inas-
much as the book can only be regarded as a report on certain spectacular
topics, some of which barely lie within the broad domains of chemistry.

At the present time, anything which tends to stimulate industrial
and applied chemistry in the United States, will be hailed with delight
by every chemist of the land. That "Chemistry of Commerce" is in-
tended to do this, is evident from the author's preface and introduction.
Whether he has succeeded in stimulating the masses in this highly tech-

910 NEW BOOKS.

nical branch of the science is a question which might best be left to the
layman himself. To the chemist, however, who is familiar with the
industries of Germany, the value of the book as a stimulus to industrial
chemistry lies litt4e above the zero mark. Germany leads the world
in industrial chemistry, not because of any attempt to popularize science
by means of educating the masses in these extremely technical branches,
but because the nation has pursued a diametrically opposite policy.
The highly trained few instead of the superficially trained many is the
secret of Germany's industrial success.

The book is made up of twelve chapters, some of which have alieady
appeared as magazine articles or '* researches, " as the publishers choose
to call them. The whole is cemented together by both a preface and an
introduction with numerous little prefaces thrown in, in order to bring
about catal>i:ic action in the mind of the reader.

The author was sent abroad for one year to "write up" the industries
of Europe. Evidently the time was too short, for some of the great
industries have been left out, or perhaps crowded out by the more pyro-
technical ones like the New Microbe Inoculation. That '*la\Tnen sub-
sist on a pabulum of illogical and, for the most part, sensational misin-
formation," is a stinging blow to scores of popular writers who are mould-
ing public thought and who never appear under the yellow flag. If
some one of these writers should consider it worth while, he might, using
"illogical" and "sensational," the same standards used by the author,
find in "Chemistry of Commerce" hues differing only by a very few wax'e
lengths from the sodium spectrum.

The chapter on alcohol is interesting and reminds one of some of the
popular newspaper articles which have appeared from time to time since
the new Food and Drug Act. The Ethyl and Maude pun, however,
seems a little out of place in any book or article which lays any claims to
the science.

Some of the other chapters as, for instance, Catalysis, Fixation of
Nitrogen, The Rare Earths, Modern Chemistry and Glass-Making, and
Cellulose are too familiar to the reader to need more than mention. Lime
nitrogen would probably have had a little more significant meaning to
the layman than Kalkstickstoff.

The last chapter on Industrial Fellowship is unique. The scheme is
not entirely new. It does, however, seem a little out of place.

In conclusion, let it be hoped that the author may not be disappointed
in his method of bringing about a great industrial awakening by his ap-
peal to the public. G^ORGB B. Frankforter.

Modern Pigments and their Vehicles. By Fred Mairs. New York: J. Wiky
& Sons. pp. 265. Price, $2.00.

This book is evidently written by a man who has had a great deal of

NEW BOOKS. 911

experience, and contains some very valuable hints, and some excellent
descriptions of the composition of pigments. In its chemistry it is a
trifle weak. Its style of composition is colloquial, and a publishing house
like J. Wiley & Sons should employ a scientific censor whose duty it is
to edit a book thoroughly. For instance, the statement that the formula
for white lead is ^PbCOj should not be published, but probably this is a
printer's mistake. The statement that red lead is a bi-oxide, and orange
mineral a ter-oxide is also incorrect.

Under the History and Chemistry of Red Lead, the author states that
red lead is the best priming paint for steel and other metals, and that
engineers and architects are unanimous in recommending it, and that
it is becoming more important every year now that so much structural
iron and steel are being used in the construction of buildings in all our
large cities. This is only one example of some of the haphazard state-
ments made in the book, because the direct opposite is the case. The
Singer Tower, The City Investing Building, the Metropolitan Life Tower
and the new Pennsylvania Terminal are four of the largest buildings with
steel construction that have ever been built, and not one of them has had
red lead applied as a priming or finishing coat, and I do not know of a
sky scraper of any importance excepting the Times Building, on which
red lead has been used. The author quotes the Norfolk Navy Yard,
but inasmuch as the Navy Department. in the United States is not pro-
gressive, and all their painting is done in situ, which is totally different
from the shop and field coating of building cdhstruction, we cannot at-
tach much importance to naval usage. This would tend to indicate
that engineers and architects are anything but unanimous in recommend-
ing red lead as a priming coat, and many of the railroads in the United
States who do use red lead use a special kind of ready-prepared or ready-
mixed red lead which contains a large percentage of reinforcing pigment
like silica.

On the other hand, Mr. Maire's book contains some excellent general
information for the painter. He has, however, omitted any reference
to wood turpentine and China Wood Oil' and speaks of naphtha and ben-
zine as materials having a horrible smell.

The chapter on the mixing of tints is excellent, and the general de-
scription of the dry colors is very good. The table of synonyms is per-
haps the best table of its kind ever published. Maximilian Toch.

Technologie der Fette und Oele, Bd. 11, Gewinnung der Fette und Ocle, Spezieller
Tett. By Gustav Heptbr, with the collaboration of G. LuTZ, O. Hbl-
L8R, Fbux KasslBR, and others. Berlin: JuUus Springer. 1908. pp. x+974,
with 19 plates. Price, 28 Marks.

The first volume of this valuable work appeared in 1907; Volumes

III and IV are promised during 1908. Hefter is director of the Aktien-

912 NEW BOOKS.

gesellschaft zur Pabiikation Vegetablischer Oele in Triest. We have a con-
siderable list of books on industrial and technological subjects by college
and university professors, by commercial analysts and consulting chem-
ists, but all too few from the pens of those who have attained high rank
in the industries of which they write. It is then, with real delight, that
we welcome this work on the technology of fats, written by industrial
men.

The subject-matter of the present volume is arranged in six general
divisions; viz., The Vegetable Oils, The Vegetable Fats, The Animal
Oils, The Animal Fats, The Vegetable Waxes, and The Animal Waxes.
Under the head of each individual fat, oil, or wax is detailed its history,
source, raw material, production, properties, trade relations and economic
significance. Methods of analysis are not given, inasmuch as these are
to be found well presented and in great detail in such authoritative works
as those of Lewkowitsch and Benedikt-lJlzer. In this way the written
page keeps faith with the title (a virtue none too common in techno-
logical works and worthy of commendation) and the work remains a
technology throughout.

The authors have gathered together from a great number of sounres
and by no means from chemical and technological sources alone, an im-
mense amount of valuable data bearing on the main subject and in point
of accuracy few works can boast a superiority to this one. A well-seasoned
acquaintance is shown with the special and general literature of the sub-
ject and with the patents and processes of various countries. It is pleas-
ing to find the historical side of the subject so capably handled, and at
the same time it is a source of satisfaction that the latest mechanical
devices and arrangements used in the fat industries are so accurately
and fully described and illustrated. Obsolete methods and apparatus,
if mentioned at all, are given but the briefest consideration in those por-
tions of the volume treating of modem industrial practice.

Of considerable interest and usefulness are the lists of synonymous
terms in various languages as applied to the various oils and fats and the
raw material from which they are derived, in the headings of the sub-
divisions and also in the body of the descriptive text.

The book is a mine of information for the chemist and technologist
and it can be heartily recommended to anybody interested in the oils
and fats industries.

The paper, typography, general and marginal indexing and the general
make-up of the voltune are of the usual excellence which characterizes
Springer's productions. W. D. Richardson.

Traits Complet D'Anaiyse Chimique A^qu^ Aiiz Essais Industrids. Pak
J. Post, B. Nsumann. DsundMB Edition Fran^aiss bntd^bsmbnt sbfon-

DUB, TRADUITB D'apRJ^S IS TROISI^MB EDITION AUJ3MANDB BT AUGMBNTto DB

RECENT PUBUCATIONS. 913

NOMBRSUSES additions: ' ParL. GAUTmiL' Tome Second-Premier Fasc. Chaux-
Mortiers et Ciments-Platre-Produits C^ramiques-Verre et Gla^uits. Avec 99
figures dans le teste. Paiis^ libraiiie Scientifique A. Hermann, 6 Rue de la
Sorbonne. 1908.

It is unnecessary for the reviewer to comment at length on this transla-
tion from the German of the well Iqiown and. generally excellent work
of Post and Nemnann, but it is regifettable that in revising, the trans-
lator has not substituted other analytical methods for certain of those
described, or at least added to them. For instance, on p. 94, for titanium
only the old method of separatioii by boiling in a nearly neutral solution
is mentioned, and no reference is made to the Lawrence Smith method
for allcaiies. The typography and general appearance of the work are
all that could be desired. W. F. Hili^ebrand.

RECENT PUBLICATIONS.

Baii^by, R D : The Brewer's Analyst. Systematic Handbook of Analysis
Relating to Brewing and Malting. London: 1907. 8vo. ' 434 pp. 13s. 4d.

Basrm,, E.: Pr€c]8 d Analyse Chimique Biologique g^n^rale. Paris: 1908.
420 pp. M. 5.

Betts, Anson Gardner: Lead Refining by Electrolysb. New York: John
Wiley & Sons. 1908. 394 pp. Svo. $4.

BntCHMORB, W. H.: Interpretation of Gas Analysis. New York: 1907. z2mo.

91 PP- $150-

BoutLANGER» E.: Industries agricoles de Fermentation: Brasseries. Hydro-
mels. Paris: 1907. 8vo. 549 pp. M. 5.

Brisker, C: Einfiihrung in das Stadium der Eisenhttttenkunde. Zusammen-
fassende Darstelltmg der Grundlagen des Eisenhflttenwesens. Leipzig: 1907. gr.8.
172 ss, M. 3.60.

Bruce, W. J.: System of Radiography. London: 1907. i6s. 3d.

Cadot, A.: Lemons de Chlmie. Fascicule, 4: Azote, Phosphore, Arsenic. Paris:

1907. 575 pp.

Chancrin, E.: Chimie g6n6rale appliqu^e a 1' Agriculture. Paris: 1907 261pp.
M. 2.

Da\, Lewis F.: Enamelling. New York: Scribner. 1908. 222 pp. Svo. $3.

Dsnigbs: Chimie analytique. 3 Edition. Paris: 1908. M. 8.50.

DuPARc, L. ET MoNNiER, A.: Traits de Chimie analytique qualitative. Suivi
de tables systematiques pour Analyse min^rale. 2 Edition. Paris: 1907 M. 7.50.

Erdmann, E.: Die Chimie der Braunkohle. Erweiterte Sonderdruck aus der
Festschrift des X. allgemeinen deutschen Bergmannstages. Halle: 1907.

FiNCK, E.: Precis d' Analyse chunique. 2 Edition. Partie II: Analyse quanti-
tative. Paris; 1907. 384 pp. M. 4. L'ouvrage complet, 2 parties, 1 906-1 907. M. 7.

PouRCROv, A. F.: Phflosophie Chimique. Paris: 1907. M. 2.40.

GoTTscHAtL, M.: Leitfaden der Chemie nach dem Arbeitsprinzip. Teil II;
Metalle. Mflnchen: 1908. gr.8. 74 ss. Das jetzt voUst&ndige Werk, 1907-1908.
M. 2.

Grotewold, C: Die Zuckerindustrie. Ihr Rohmaterial, ihre Technik, und
volkswirtschaftliche Bedeutung. Stuttgart: 1907. 176 ss. M. 2.

914 RECfiNt PUBLICATIONS.

Hbsrmann, p.: Farbereichemlsche UnterBttchungen. 2, erwdterte u. umge-
arbeitete Auflage. Berlin: 1907. gr. 8. 344 ss. M. 9.

HsFTSR, G. : Technologie der Fette and Oele. Handbuch der Gewinnung imd
Verarbeitung der Fette, Oele und Wachsarten des Pflanzen- und Tierrdchs (4 B&ode).
Band II. Gewinnung der Fette und Oele: Spezieller Teil. Berlin: 1908. gr. 8. 974
ss. M. 28. Band I. 1906. 759 ss. M. 20.

JahiMbericht ttber die Leistungen def Chemischen Technologie, von IL Wagnsb,
fortgesetzt von F. Fischsr. Generalregister fiber Band 41-50. Leipzig: 1908. gr.
8. 260 ss. M. 9.

Joi«Y, A. ST Lbspicbau, R.: Cours A^mentaire de Chimie Mtftaoz. Chimie
organique. 5 Edition. Paris: 1907. 558 pp. M. 4.20.

K0B8IX, F. VON.: Tafeln zur Bestimmung der Mineralian mittels einiadMr
chemiflcher Venuche auf trockenem und naasem Weg. 15 » neubearbeitete und ver-
mehrte Auflage, von K. Oebbeke. Munchen: 1907. gr. 8. 125 ss. M. 2.50.

DB La Coux, H. : L'Eau dans llndustrie. Composition, ^puration, analyse,
etc". 2 Edition. Paris: 1907. 540 pp. M. 13.50.

Landausr, J.: Die Ltttrohranalyse. Anleitung zu qualitativen chemischen
Untersuchungen auf trockenem Wege. 3, verbesserte und vermehrte Auflage. Ber-
lin: 1908. gr. 8. 186 ss. M. 6.

Lassar-Cohn: Einftihrung in die Chemie in leichtfasslicher Form. 3, ver-
mehrte Auflage. Hamburg: 1907. gr. 8. 301 ss. M. 3.

Lbwbs, Vivian B. : Liquid and Gaseous Fuels, and the part they play in modern
power production. New York: D. Van Nostrand Co. 1907. 334 pp. $2.

Marchis, L.: Production et utili8atk>n des Gaz pauvres. Paris: 1907. 322 pp.
M. 16.20.

Prbscott, S. C. and Winslow, C. £. A.: Elements of Water Bacterblogy, with
special reference to sanitary water analysis. 3nd ed., rewritten. New York: John
Wiley & Sons. 1908. 258 pp. $1.50.

RoscoB, Sir H. Enfibld and SchorlBmmbr, C: A Treatise on Chemistry. In
2 vol. Vol. 2. The Metals. New Edition, completely revised by Sir H. £. Roscoe
and Dr. A. Harden. New York: The Macmillan Co. 1908. 1436 pp. 8vo. $7.50.

RoYLB, H. M.: Chemistry of Gas Manufacture. London: 1907. 8vo. 344 pp.

13s. 4d.

Silbbrmann, H.: Fortschritte auf dem Gebiete der photo- und chemigraphischcii
Reproduktionsryerfahren 1877-1906. 2 B&nde. Leipzig: 1907. 307 u. 480 ss.
M. 50.

Stoughton, Bradlby: The Metallurgy of Iron and Steal New York: Hill Pub-
lishing Co. 1908. 500 pp. 8vo. $3.

Tbrry, H. L.: India-Rubber and its Manufacture. With chapters on Gutta-
percha and Balata. London; 1907. 8vo. 304 pp. 6s. 9d.

THoiiSBN Juuus: Thermochemistry translated from the Danish by Katharine
A. Burke. New York: Longmans, Green & Co. 1908. 495 pp. $2.50.

TONGB J.: Coal. New York: D. Van Nostrand Co. 1907. 275 pp. $2.

Wbnzbl, F.: Die periodische Gesetzmiissigkeit der Elemente nach Mendd6eff,
durchgesehen und erganzt. Wien: 1907. M. 4.

Young, Sydnby: Stoichiometry, together with an introduction to the study
of physical chemistry, by Sir W. Ramsay. New York: Longmans, Green & Co.
1908. 381 pp. 88 fig. $2.

Zbllnbr, Julius: Chemie der htfhem Pflze. Leipzigt 1907. 241 ss. M. 9.

Vou XXX. JuNB, 1908. No. 6.

THE JOURNAL

OF THE

American Chemical Society

PAPERS ON SMELTER SMOKE.

[SECOND PAPER.]
ARSENIC IN VEGETATION EXPOSED TO SMELTER SMOKE. >

Bt R. E. Swaik and W. D. Harkins.
Received January a8, 1908.

In a previous paper^ there were presented the results of a study of the
smoke emanating from a copper smelter now in operation near Anaconda,
Montana. It was estimated that at the time the examination was made
this plant was discharging daily from its main chinmey an average of
59,270 pounds (26,880 kilograms) of arsenic trioxide, along with notable
quantities of copper, antimony, lead, line, and other substances. The
smelter is situated at the base of a ridge which descends steeply from Mt.
Haggin, a perennially snow-capped peak rising near the main crest of the
Rocky Mountains. On one side of this ridge is the valley of Warm Springs
Creek in which the city of Anaconda is situated, and on the other side is
Mill Valley, each typical of a number of narrow valleys and ravines lead-
ing down from the main range into a broad basin thirty-five miles long and
four to six miles wide, the Deer Lodge Valley. The present paper will
deal with the distribution of certain of the more notable constituents
of the smoke, particularly arsenic, over the district in the vicinity of the
smelter. This is a region of variable winds which often blow with great
velocity, so the section affected by the solid or gaseous components of the
smoke is not confined to a narrow tract nor to any closely circumscribed
area. There are prevailing wind directions for each season of the year,
but these are not so constant as to restrict the action in any considerable

' The work on this paper was begun in 1902 by W. D. Harkins, and was presented
in abstract at the New York meeting of the American Chemical Society, December,
1906.

■ This Journal, 29, 970.

9l6 R. E. SWAIN AND W. D. HARKINS.

degree. Pouring out of the stack a thousand feet above the valley floor,
the smoke stream can be traced as far as the eye can reach in that normally
clear atmosphere, trailing down the valley for thirty miles toward Gar-
rison, or often eastward in the direction of Butte, or sweeping over into
Mill Valley and filling the narrow ravines which lead down from the
Continental Divide, fourteen miles to the south. In spite of the height
at which it is discharged the smoke may be seen very frequently to strike
the valley within less than a mile from the base of the stack, while on
quiet days it may rise to a considerable height and hang as a haze over
the valley.

The basis of the work herein described has been the vegetation of the
Deer Lodge Valley and adjacent territory, and especially the hay and
wild grasses, for the primary incentive to the investigation, part of which
preceded the examination of the smoke itself, was the claim on the part of
the farmers residing in the vicinity of the smelter that they were suffering
abnormal losses of live-stock, which they attributed to smelter fumes.
The first appearance of the peculiar symptoms of disease in the stock were
noticed only a few months after the smelter began to operate ; the quick
recovery of the milder cases upon removing them to distant pastures, and
the recurrence of the disorder on their return; and the fact that the great-
est disturbance was observed in those sections over which the smoke
drifted most frequently, suggested that the trouble was a local one and in
some way connected with the smelter.

In all cases the samples analyzed were collected by one of us or by both
of us together, either in stoppered bottles or in heavy glazed jute sampling
bags. With very few exceptions the results obtained are from air -dried
samples, this being deemed a more reasonable and practical basis to
which to refer the analyses than the more complete drying at ioo°.

From each of these samples a weighed quantity (30-200 grams) was
taken for analysis, placed in a large casserole covered with a watch glass
and concentrated nitric acid allowed to stream over it from a pipette.
Destruction of the tissue begins at once without the application of heat,
and in the case of samples with many leaves which present extensive
smfacesfor action, the mass may ignite* and bum, if not carefully watched
during the first stage of the decomposition. Excessive action must be
checked promptly by floating the vessel in cold water or, if that is in-
sufficient, by diluting the add with water. By stirring it occasionally
with a glass rod the sample is soon reduced to a thick, yellow, semi-liquid
mass after which heat may be applied to the casserole until the most of the
acid is expelled. Five to 8 cc. of concentrated sulphuric acid are then
added, drop by drop. A rapid decomposition of nitro-compounds fol-

^ If a large amount of acid is added there is no danger of ignition even if con'
centrated nitric acid is used.

PAPERS ON SMELTER SMOKE. 917

lows and thereafter the mass is reduced to a charred and crisp condition
by heating in an air bath to a temperature not exceeding i8o°.* If the
destruction of the tissue is not perfect enough after the addition of the
sulphuric acid, successive small portions of nitric acid may be added
during the subsequent heating, taking the precaution to cool the casserole
to loo® or less before each addition. This method, which is essentially
an application to plant tissue of the Chittenden-Donaldson method for
the destruction of animal tissue, yields a final residue capable of easy
extraction with acidified water. The extract thus obtained was filtered
into a graduated flask and an aliquot part taken for analysis.

In all but a few cases the arsenic was estimated by applying the Marsh-
Berzelius method and weighing the mirrors which resulted. In no case
in which results are given was an estimate based upon comparative
mirrors.

Copper was determined by evaporating another portion of the solution
to dryness in a porcelain dish, igniting to destroy all organic matter,
and after dissolving the residue in dilute nitric acid, precipitating the
copper electrolytically. The following table gives the results of the
analysis of vegetable tissue collected in the Anaconda region :

Table I. — ^Arsenic and Copper in Grass and Hay.

Parts per million.*

mber.

Sample.

Month.

Distance and
direction.

A8,0,.

Copper.

1902

G»

Nov.

0.25

1551

1800

Nov.

3.0

166

871

Nov.

4.0

708

Nov.

1-5

283

• • .

Nov.
1903

216

Oct.

4.0

128

Oct.

2.0

• • •

Oct.

• • •

Oct.

• a •

Nov.

15.0

NNE

164

Nov.
1905

150

NNE

405

237

Jan.

122

• • .

Jan.

5.0

100

563

Jan.

5.0

ESE

Jan.

2.0

SSE

176

^ Another method used by one of the writers will be described in a subsequent
paper.

' Parts per million are equivalent to ten-thousandths of i per cent., and multiplied
by 0.7 give the number of grains in 100 poimds of substance.

' H denotes a sample of hay taken from the stack ; G, a sample of grass cut from
the field ; L, leaves of trees ; B, bark of trees ; C, leaves of the cedar ; and V, leaves of the
lily-of-thc-valley.

gii R. n. SWAIN AND W. D. HARKINS.

TablS I {CofUinuid).

Number.

Sample.

Montli

Jan.

Jsui.

Jan.

Feb.

May

Jime

June

Sept.

Sept.
Z906

Feb.

June

July

Aug.

ParU

per million.

TMaAan«*A mmH

direction.

aho..

Copper

4.0

2.0

SSB

2.5

4.0

ESE

150

6.0

170

226

4.2

3.0

326

S-o

112

2.0

220

119

4.5

217

407

4.0

• ■ «

4.0

• • .

5.0

BSB

6.0

112

5.0

263

470

35.0

a • •

4.5

161

6.0

■ • .

12.0

NNB

190

2.0

■ • •

14. 0

NNB

221

6.0

. . •

0.5

427

...

5.0

140

5.5

180

3.0

4.0

3.0

107

4.0

4.2

8.0

NNB

III

5.0

3.0

1.5

157

2.0

359

1.5

460

1.7

293

1.7

583

1.5

350

1.7

376

6.0

•

2.0

508

0.7

431

6.0

1.5

300

1.7

683

PAPBRS ON SBIKLTBR SMOKB. 9 19

TablH I (Continued).

Distance and

Parts per n

Number.

Sample.

Month.

direction.

ASflOj

Aug.

1.7

482

Sept.

2.5

Sept.

2.5

100

Sept.

6.0

Sept.

4.2

Sept.

I.O

lOI

Sept.

I.O

236

Oct.

10. 0

Oct.

130

Oct.

350

Oct.

34.0

Nov.

4.2

NNE

121

Nov.

6.0

NNE

Nov.
1907

1.5

705

Jan.

I.O

265

Jan.

2.0

Jan.

vSvSE

Jan.

4.0

Jan.

6.0

SSE

Jan.

6.5

Jan.

8.0

Jan.

lOO.O

Jan.

750

Oct.

100. 0

A few of the above samples were analy7ed by each of us, usmg the same
balance sensitive to 0.005 nig-> but otherwise independently, and the
mean of the results obtained taken for the table above. The complete
data for each joint determination are given as follows :

Tabls II.

Per cent. ASfO*.

No.

Sample.

Difference.
Per cent.

Grass

0.0103

0.0097

0.0006

Grass

0.0089

0.0085

0.0004

Grass

0.0169

0.0170

O.OOOI

20-31

Hay

0.0089

0.0085

0.0004

Grass

0.0079

0.0075

0.0004

Grass

0.0219

0.0220

O.OOOI

Grass

0.0264

0.0261

0.0003

33-35

Hay

0.0034

0.0035

O.OOOI

In all but two cases of those cited above the results given are merely
duplicate determinations made on a single sample, but samples 20 and 31
and samples 33 and 35 were instances where each of us took a sample
from a stack independently of the other and conducted a separate deter-
mination.

920 R. E. SWAIN AND W. D. HARKINS.

The first samples taken from this district were collected in November,

1902, but the record of all except nine of the results of the analyses made
on this material and that of the following year was accidentally destroyed.
The smelter started early in 1902 and continued without interruption
throughout that year, so the results for grass represent a deposition of
arsenic extending over a period of six months, while the hay, cut in July,
was exposed less than three months. Samples 2 and 5 were taken from
the same farm, but one was left exposed in the field during the whole
period while the other was removed and stacked as hay. The difFerenct
here shown in the two values, 166 and 36, respectively, is fairly repre-
sentative of a condition which will be found to prevail with notable
uniformity throughout the analyses. An interesting exception is found
in the data for 1903, where samples 6 and 7 show quantities which are not
only relatively very low, but nearly equal. During 1902 and a part of the
following year the smoke from the smelter was delivered from four smaller
stacks which were erected near the center of the plant. But in June,

1903, the smelter was shut down until late in September in order to pennit
certain modifications to be made, chief among them being the final work
of construction upon the present high stack and extensive settling flue.
At the time these samples were collected, one from a haystack and the
other of grass from the open field, the smelter had barely begun operations
after the summer's inactivity, so that the two samples had been exposed
to smoke only during the period of growth until the time of the closing
down of the smelter in June, and both were exposed equally long. Sample
8 shows the result of an added exposure of over a month during the dry
season of the year. The moss sample from the same locality gave a
remarkable result which is to be explained only on the assumption that
owing to its peculiar matted growth the moss may have collected the
arsenic which fell upon it like a natural filter, and held on to it from the
previous year. During the year 1904 no collections were made, though
the hay samples taken in the early part of 1905 were from the 1904 crop.
The average amount of arsenic in these hay samples collected up to May,
1905, is 41.9 parts per million, while the average of the results for grass
is 120.4.

For the year 1905 a noteworthy sample is that of grass (No. 30) from
Garrison, thirty-five miles north of the smelter, which carried 35 parts of
arsenic trioxide per million. Two samples (Nos. 71 and 72) collected
in October of the following year from the same locality carried 29 and 21
parts, respectively, per million. The former sample was old grass collected
in May, at which time the grass from the preceding year usually attains
its maximum content of arsenic.

It must be admitted that, whatever the amount of arsenic discharged
daily \\ith the smoke, we have, without further evidence, no satisfactorj'

PAPERS ON SMELTER SMOKE. 921

measure of its distribution. If arsenic and copper and other toxic con-
stituents of smelter smoke were not to be found in the soil, their appearance
in the vegetaton of the smoke zone could be attributed at once to at-
mospheric deposition. But it may be claimed that the arsenic and
copper found in these samples are not deposited from the smoke stream,
but absorbed from the soil itself. It is a fact that growing plants may
absorb small amounts of copper from the soil, and a review of the litera-
ture would seem to leave no doubt that the same is true, in a very slight
degree, for arsenic.

Voelcker* experimented with grass grown on soils to which arsenical
superphosphate had been added, and Hehner, who made the analyses
for him, found on an average 0.0045 grain of arsenic per pound of straw
(0.7 part per million). Swedes under the same conditions gave no arsenic
in the bulbs, and in one case only 0.02 grain per pound of dried tissue
(2.8 parts per million) in the leaves and stems. Barley straw grown on
soil manured heavily with superphosphate containing 0.5 per cent, arsenic
trioxide carried 0.007 grain per pound (i part per million) while the leaves
of swedes and mangels grown on ordinary soils containing about 0.0002
per cent, of arsenic trioxide were found to contain 0.002-0.004 grain of
arsenic per pound (0.3 to 0.6 part per million). AngelP analyzed a large
variety of cereals and table vegetables which were grown on soil to which
superphosphate carrying 0.5 per cent, of AsjO, was added. Of thirty-two
samples examined twenty-two showed not a trace of arsenic while the
rest are reported as containing slight traces. Since, however, only an
acid extract of the plant in dilute hydrochloric acid was made, with no
effort to destroy organic matter before adding it to the Marsh generator,
and since only fifteen minutes were allowed for the arsenic mirror to
appear, the results are open to very serious question. Gautier' found
arsenic in many vegetable foodstuffs; in wheat (0.000,007 per cent.),
bread (0.0000071 per cent.), and in potatoes (0.000,0112 per cent.),^ but
not a trace in cabbage and beans. He also claims to have found it in
marine and fresh water algae.

Copper in traces has been shown repeatedly to be present in plants
grown on coppery soils. In fact the occasional presence of copper in
plants has been admitted for nearly a century.* Verdrodi* found notable
amounts of copper in buckwheat (0.87 per cent. CuO), maize (0.06-0.39
per cent.) and other cereals. These results, however, have been criticized

* Report Royal Arsenical Commission, Vol. II, p. 174.
■ Ibid., p. 10.

*Gautier, Compt. rend., 139, loi (1904).

*Sec Meissner-Schweigg, Journal, 17, 340 (1816); Phillips, Ann. chim. phys. [2],
19, 76 (1821).

•Verdr6di, Chem. Z., 17, 1932 (1894); and 20, 399 (1896).

922 R. B. SWAIN AND W. D. HARKINS.

by Lehmann, ^ who points out that they are ten times those obtained by
Tscherch from soils very rich in copper, and justly criticizes the method
adopted by Verdrodi for the estimation of copper, viz,, ashing the plant,
extracting the residue with nitric and hydrochloric acids, precipitating
with hydrogen sulphide, and without further treatment weighing the
precipitate as copper sulphide. Lehmann's own results, however, show
that minute quantities of copper may be present in certain plants which
subsist on a soil containing copper salts.

There are good reasons for believing that the values given for arsenic
and copper in the previous table are not due to absorption from the soiL
Plants do not absorb so much arsenic or copper as is there shown. The
writers have grown cereals on several soil samples which were collected
in the Deer Lodge Valley. These were taken outside the range of the smelter
smoke and planted to barley and timothy. In no case have they found
more than 0.0002 per cent, of arsenic trioxide in the matured plants, where
values as high as 0.0263 per cent, had been obtained from grass grown
on the same soil five miles from the smelter. Moreover, above every-
thing else the length of time the plant is exposed to the free atmosphere
in the smelting district is the determining factor in coimection with its
arsenic content. A few cases illustrating this have already been cited in
connection with some of the earlier results, and many others could be
selected. Nos. 53 and 61, samples of wild lily-of-the-valley, were collected
by one of us from the same spot about one and three-quarter miles from
the smokestack and in a section over which the smoke blows much of the
time in the summer months. The first one was cut July 3rd and the
other August 14th, or six weeks later. During this time the arsenic
content of the plants increased from 583 to 682 parts per million. At
the same time these were taken a kind of wild grass {agrifnrons dtvergens)
growing at the same place was sampled. These are designated as Nos.
49 and 62. When the first of these was collected the grass was already
dead, so the increase in arsenic trioxide from 293 to 482 parts per mllion
cannot be ascribed to processes of absorption from the soil. Nos. 44,
66 and 73 are meadow grass ^mples taken from the same field in the
months of July, September and November, respectively. The July
sample, covering the period of most rapid growth and most frequent
rainfall, carried 12 parts of arsenic trioxide per million. The September
sample more than covered the rest of the growing period and carried 34
parts, while the November sample carried 121 parts per million, thus
collecting an added 87 parts after all growth had ceased.

A striking proof that the arsenic is deposited from the smoke was found
by a study of the wind currents. The Mill Valley district southwest of the
smelter is the one toward which the smoke blows most during the early

* Arch, Hyg., 24, 3; and 27, i (1896).

PAPERS ON SMELTER SMOKE. 923

summer, while late in August the air currents begin to go northward down
the Deer Lodge Valley, and from this time until the snow covers the ground
the greater part of the smoke blows in this direction. The analyses show
that the grass of Mill Valley contains more arsenic than any other district
during the early summer. Thus samples 48, 50, 51, 52, 53 and 58, which
were gathered in July, 1906, contained respectively 157, 359, 460, 293,
583 and 431 parts of arsenic trioxide per million. North of the smelter,
samiple 12, taken in July 1906, from the Bliss ranch, contained 18 parts,
one taken in September 33 parts, and one taken in November 73 parts
of the trioxide per million. In 1905 the Bliss grass in June contained
67 parts, in September 61 parts, and in February, 1906 (grass of the
season of 1905) it contained 180 parts per million. Hay from the same
ranch cut in August, 1904, contained only 30 parts to the million, while
by the next April the grass in the same field had increased its arsenic
content to 263 parts. During 1906 the Para ranch gave 12 parts in July,
34 parts in September, and 121 parts in November. There is, of course,
some objection to a comparison of the grass with the hay, since a certain
portion of the arsenic of the latter is shaken off by the cutting and stack-
ing. Even with the grass it is difficult to obtain the original arsenic
content, since the sample must be cut, put into containers, and trans-
ported to the laboratory.

These results show very plainly that the greatest accumulation of
arsenic generally occurs after the period of growth is ended and the plant
is dead. There is little doubt that the rains during the early summer
wash much of the arsenical deposit from the vegetation, while in winter
the grass is protected by snow. The most favorable time then for the
arsenic to accumulate on the tissue is during the late summer and fall;
and during that period, in fact when absorption through the roots is out
of the question, since when the grass is dead the accumulation is most
rapid.

That this is deposited arsenic rather than absorbed arsenic is evidenced
further by the fact that by shaking dry hay or grass grown in the vicinity
of the smelter a finely divided dark-gray powder, running notably higher
in arsenic than the tissue from which it came, is obtained. Various samples
of this dust were collected by shaking the hay or grass on a fork over a
glazed cloth and separating leaves and fragments of the tissue by means
of an 80-mesh sieve. This dust accumulates in quantity on the floors
and rafters of hay lofts and in the mangers of feeding barns where it can be
swept up in large amounts. Several samples of dust were also obtained
from parts of threshing machines and analyzed for arsenic. Complaints
are often made by men working on these machines that the thresher dust
from grain in the smoke zone irritates the skin and the mucous membrane
of the eyes and nasal passages.

924

R. E. SWAIN AND W. D. HARKINS.

Table III. — Arsenic in Dust from Hay and Grain.

No.

Year.

Month.

Sample.

Direction

from
smelter.

Distance

from
smelter.

AstOg.
Per

cent.

AssOs.

Parts per

million.

CuO.
Per

cent.

Farm.

1905

June

H.» Dust

N. E.

4-5

0.0987

987

0.5320

Para

1905

June

< <

( (

4-5

0.0969

969

0.5600

< (

1905

Apr.

1 1

( (

0.4380

4380

0.7430

Staffanso

1905

Apr.

< 1

1 1

0.5140

5140

0.7970

( *

1905

Oct.

T.« Dust

0.0887

887

Bowman

1905

Oct.

< t

0.0594

594

Perkins

1905

Oct.

i i

0.0448

448

Watts

1905

Oct.

< 4

0.0410

410

Day

1905

Oct.

1 (

0,0941

941

Jones

1906

July

H. Dust

1-5

0.3526

3526

Lappin

1906

Nov,

« (

S. E.

0.9190

9190

Allen

These values are uniformly high, the richest of the hay dust samples
exceeding any of those for the plant tissues many times over. The
thresher dust samples run lower, due in part at least to the fact that the
conditions under which that dust was deposited are more apt to cause a
loss in arsenic and a concentration of the heavier silicious dusts. The
samples of hay dust from the Para farm were collected from the stack
from which hay samples 20 and 31 were taken. The hay averaged 87
parts of arsenic trioxide per million, while the dust shaken from it averaged
978 parts. The dust from the Staffanson hay contained 7700 parts of
arsenic trioxide per million, while the hay itself carried 261 parts. In
the case of Lappin 's hay the dust carried 9190 parts and the hay 50 parts
per million. Such results indicate clearly that the arsenic is deposited
on the surface of the plant and is not distributed throughout the tissue.

A contention which might be raised is that this dust is soil which has
been blown upon the hay, or which adhered to it during the process of
curing in the field. Some representative analyses of soil taken to a
depth of three inches are given in the following table:

Table IV. — ^Arsbnic in Soils.

Per cent.

Farm.

Distance.

AssOs.

Staton

2 miles S.

0.0019

Callan

3 "

s.w.

0.0029

Para

4.5 "

N.N.

0.0043

BUss

5.5 "

0.0061

Staff anson

3 "

0.0107

The Para hay dust then carried 22.7 times as much arsenious oxide as
did the soil. The Lappin hay dust carried 483 times as much as the soil
sample from the adjoining Staton farm; and the Staff anson hay dust
contained 72 times as much arsenic as the soil upon which it grew. It is

» H - Hay.
» T - Thresher.

PAPERS ON SMELTER SMOKE. 925

"extremely unlikely that dust from the soil is any large contributor to the
quantities of arsenic found in the hay dust. Rather would it appear
more reasonable to assume that in so far as the soil is present it is a diluent
which diminishes the percentage of arsenic. Originally the soil of this
district does not seem to have been arsenical. The top three inches of
uncultivated soil will usually afford small quantities of arsenic which
rapidly vanish at greater depths. A sample of top soil taken from under-
neath a well built log house erected long before smelting operations
began in that valley carried only a trace of arsenic, while soil from an
adjoining field was found to contain 0.0107 per cent, in the top three
inches. Soil which had never been plowed contained no arsenic what-
ever at depths of from twelve to fifteen inches, so the arsenic present in
the top soil would seem to be due wholly to the smelter emanations.

Samples of snow were scraped from the surface at a time when the
ground for miles was covered with a sheet which had not yet been dis-
turbed by the wind. Considerable amounts of soluble arsenic were
found, the arsenic being dissolved in the water formed when the snow
melted. The appended table gives the analytical results obtained:

Tablb V. — Arssnic in Snow.

Date,

Z906. Farm.

Distance
and direction.

Weight

of snow

grams.

Sq.ft.
surface area.

Time
exposed.

ASjO,^
per sq. ft.

Per cent,
of dust.

2-5 Para

5 miles N.

925

6.228

17 days

0.00306

0 I130

2—5 Williams

4 " N.

955

12.6

17 days

0.00026

0.1690

2-5 Bliss

5.5 " N.

• • •

17 days

0.00040

0.2810

3-18 Callen

3 " S.W.

808

1345

3 days

0.00057

3-18 Bliss

5.5 " N.

604

6.72

3 days

0.00058

The figures in the column ** Per cent, of dust " include the percentage of
soluble arsenic calculated as if it were a component of the water-insoluble
dust.

Early in the fall of 1905 vaselined cloths, each presenting a surface of
three to eight square feet, were exposed at several points near the smelter.
One was fastened to each side of a board tablet, somewhat larger than the
cloth, which was set up on poles about twelve feet above the ground and
broadside to the direction of the smoke stream at that place. Thus
one vaselined cloth faced the smelter and the other faced in the opposite
direction. In one case (No. 2) a cloth was stretched horizontally above
the other two with its vaselined surface upward. Samples i and 2 were
set up September 5th and taken down November 21, 1905. No. 3 was
set up October ist and taken down March i8th, but the vaselined surface
was frozen hard and smooth for most of the winter and its efficiency as a
collecting medium thus greatly decreased. The following data were
obtained from this experiment :

926 R. B. SWAIN AND W. D. HARKINS.

Table VI. — ^Arsenic in VivsBLiNKD Cloths.

AstOs per aq. ft. of cloth.

*-~^^""^"""^^— ^— ^^■~^~^"~^~^"^"^~~"^^^~"~~%
Direction Distance Pacing Pacing Pacing

from from No. days upwara, smelter, away.

No. Parm. smelter. smelter, exposed. gram. gram. gram.

X Bliss N. 5.5 77 ... 0.00052 0.00024

2 Para N. 5.0 77 0.01194 0.00700 0.000S3

3 Staton S. E. 5.0 169 ... 0.00327 0.00051

The results admit of only one interpretation, which is that the smelter
smoke is the source of the arsenic found in such excessive amounts in the
vegetatipp of the region about Anaconda. The question of whether this
is a condition for which there is no remedy is difiScult to answer. It is
doubtful whether the present limited water supply would permit of the
installation of wash towers, and the large amotmt of sulphur trioxide
in the smoke would be destructive to a bag-house s)rstem. With its
present high temperature and velocity the smoke must carry most of the
arsenic out of the stack. In spite of the use of dry filters and wash towers
and a consequent low velocity of the smoke, in an efifort to meet the re-
quirements of the amended Alkali Acts of 1892, the average escape of
arsenic trioxide from the chimneys of arsenic plants in England for the
past five years has been 0.0028 gram (0.041 grain) per cubic foot.*

In the thirtieth annual report' the Chief Inspector under the Alkali
Acts wrote: *'The inspection of these works (i. e., arsenic works) had
been much called for by residents in the district where they are found,
and the necessity for the adoption of remedial measures was shown by the
frequent litigation brought about by the alleged destruction of cattle
through eating grass said to be poisoned by the arsenic too freely dis-
tributed through the air from the chimneys up which it had been carried
by the draught. The arsenic driven off by the heat of the roasting furnace
was caught and retained as far as possible, in long flues, culverts, and
settling chambers, but although these were often of great extent, reaching
in one place, a length of 2895 feet, or more than half a mile, and having a
capacity of 60, 795 cubic feet, yet arsenic was always liable to pass away.
In one case a test of the chimney gases showed the presence of as much
as 7.40 grains of arsenic trioxide in a cubic foot of gases passing into the
air, and small flakes of it were seen falling continually in a mild snow-
shower." The average escape of arsenic trioxide per cubic foot of chimney
gases from plants provided with wash-towers, or with dry filters made by
packing large chambers with brushwood, for each year since 1896 was as
follows :

1896 0.080 grain

1897 0.086 *'

1898 0.098 *'

i * Annual Reports Chief Inspector under Alkali Acts 1902-6.

■ 30th Annual Report under Alkali Acts (1893), p. 91.

PAPERS ON SMELTER SMOKE.

927

1899 0.074 gi^m

1900 0.094

1901 0.083

1902 0.049

1903 0.039

1904 0.053

1905 0.036

1906 0.030

These results attest the difficulty of removing all the arsenic from
chimney gases even where, as in many of the plants referred to above,
the ftunace gases are cooled by wash-towers and the velocity of the
smoke is reduced by extensive settling chambers. The smoke proceeding
from the smelter at Anaconda on the basis of results presented in the first
paper carries about 0.200 gmin per cubic foot, a considerable velocity
even in the settling chambers and a chimney temperature of 180° both
interfering strongly with the deposition of the arsenic. The quantity
which is given off under the present smelting conditions is not likely to be
reduced except by the use of a less highly arsenical ore. As the ore
supply for the Anaconda plant in drawn entirely from the mines of the
Butte district and the arsenic content of these ores seems to be con-
stantly increasing as greater depths are reached, the prospect of finding
a ready solution of the problem is not encouraging. The great value of the
products lost in the smoke, as shown by the first paper of this series,
should encourage smelter companies to conduct extensive experiments
with a view to the mitigation of this evil.

Col

July
July
July
July
July
July
Aug.

8 Aug.

9 Aug.

10 Aug.

11 Aug.

12 Aug.

3
4
5
6

Table VI.»

lected. Sample. Source.

12,1907 Grass Everett, 300 ft. S.E. from smelter. .

12, 1907 Grass Everett, i mile S. E. from smelter.

12, 1907 Poplar leaves Everett, 5o ft. W. from smelter

12, 1907 Grass Everett, 50 ft. W. from smelter

12, 1907 Grass Everett, 100 ft. N. from smelter. . .

12, 1907 Grass Everett, i mile N. from smelter. . .

1 1 , 1 906 Sunflower leaves, i mile from Murray smelter, Utah

11, 1906 Grass, i mile from Highland Boy smelter, Utah. . .

12, 1906 Alfalfa, I mile from Highland Boy smelter, Utah.

12, 1906 Grass, 2 mile from Bingham smelters, Utah

12, 1906 Grass, i mile from Bingham smelters, Utah

12, 1906 MiUcweed, 3 miles from Highland Boy smelter, Utah

ASfOt.

Per
cent.

Parts per
million.

1.3000

13,000

0.9400

9,400

4.4000

44,000

0.2300

2,300

0.2300

2,300

0.0420

420

0.1072

1072

0.0062

0.0038

0.0021

0.0045

0.0027

Incidental to this work in cotmection with the smelter at Anaconda
samples of plant tissue have been collected from two other prominent

' A large number of grass samples from other smelter districts have been anal-
yzed since the completion of this paper. Many of these contain arsenic in large
, quantities.

928 W. D. HARKINS AND R. E. SWAIN.

smelting centers, from Everett, Washington, about thirty miks north
of Seattle, and from the region a few miles south of Salt Lake City,
Utah, where several large smelters are in operation.* At both of these
places complaints of injury to live stock have arisen from time to time.
Only a few of these samples have been analyzed with the above results.

The writers wish to express their indebtedness to Dr. John Maxson
Stillman for suggestions in regard to this work.

THB UNIVBR8ITT OF MONTANA, MlSSOULA, MONTANA, AND STANFORD UNIVKRSITT, CALXFOKNIA,

December 28, 1907.

THE CHRONIC ARSENICAL POISONING OF HERBIVOROUS

ANIMALS.'

(papers on smbltbr smoke, third paper.)

By W. D. Harkins and R. B. Swain.
Received April x, 1908.

The two outbreaks of supposed arsenical poisoning which have led to the
most investigation from a scientific standpoint, are the one in Manchester,
England; in the year 1900, and that in the district surroimding Anaconda,
Montana, during the year 1902- 1903. The former was confined to human
beings, the latter almost entirely to cows, horses, and sheep. As yet very
few data have been obtained to show whether or not the effects of the
arsenic in the latter case extended to the human beings who resided in the
district.

During the latter part of November, 1902, it was the fortune of one of us
to travel over about one hundred square miles of the territory surrounding
the new Washoe smelter at Anaconda. At that time the carcasses of
several hundred animals that had recently died lay scattered over various
ranches of the valley, and one ranch was visited where approximately sixty
carcasses, mostly horses, were seen in a group in one comer of the field- A
very large number of the animals were dissected, and practically all of them
gave evidence of arsenical poisoning, either acute or chronic.

As has been explained more in detail in a former paper, what was called
the "Old Works*' had been in operation for many years on the north side
of the valley of Warm Springs Creek, in which the city of Anaconda is
located. In January, 1902, smelting operations were transferred to the
**New Works," which are located on the south side of the same valley, on a
ridge extending down from the foot-hills. This ridge projects into the Deer

* See Ebaugh, Gases vs. Solids, This Journal, ig, 951, 970 (1907).

■ At the New York meeting of the American Chemical Society, December, 1906,
W. D. Harkins presented a similar paper including only his own work. The first
paper of this series deals with the amomit and character of the smoke given ofT by the
smelter, and the second treats of the arsenic content of the vegetation.

ARSENICAl, POISONING OF HERBIVOROUS AlJiMALS. 929

Lodge Valley in such a way that much more of the smoke from the smelter
was blown across or into the valley than during former years.

Although the smelter was in operation during the whole year, few cases
of serious sickness among live stock were noticed until September, from
which time the number of deaths increased rapidly until November, when
a maximum was reached. This is easily understood wheti the facts of the
case are considered as presented in the second paper of this series,* The
fresh grass of the spring has little time to accummulate arsenic from the
air, and the summer is a period of rapid growth and frequent rains ; in the
autumn comes a dry period during which there is no plant growth, so this
is the most favorable period for a rapid increase in the amount of solid
substances adhering to the leaves and stems of the plant. During this dry
period, the farmers were forced to drive or ship a great number of animals
from the district, and those remaining were stabled and fed upon hay.
The owners had found from their experience with the symptoms exhibited
by the animals, that the disease was greatly moderated by changing the
diet from grass to hay. This was explained by the much smaller amounts
of arsenic found in the latter.^

In 1903 the smelter company had built the great flue and stack described
in the first paper* in order to prevent the escape of the arsenic. In accord-
ance with an agreement entered into with the farmers, the smelter was shut
down from July ist to September 30th of this year, and when the plant was
started on the latter date the smoke was supposed to no longer scatter
poisonous substances over the valley, since it was passed through the flue
intended for its ptuification. During the siunmer and autumn of this year
there were very few cases of death among the animals of the valley and it
was believed by the residents of the valley that there would be no more
cases of sickness due to arsenic.

Two interesting exceptions were investigated in cases where the
farmers thought that arsenical poisoning had occmred. A drayman in the
city of Anaconda purchased a load of hay, supposed by him to have been
grown during the year 1903. After feeding the hay to his horses for about
a week one of them sickened and died, the most marked symptons before
death being recurrent convulsions. . A post-mortem examination showed
the usual inflamed condition of the stomach, and the presence of a number
of small ulcers, resembling those frequently observed in cases of arsenical
poisoning. An analysis of the hay gave 0.0285 per cent, (285 parts per
million) of arsenic trioxide. Various tissues of the animal were also
analyzed and the liver found to contain 1.30 milligrams of arsenic trioxide
to one hundred grams of tissue (13 parts per million). It seemed improb-

• This Journal, preceding paper.

• Table 3; or Table I, preceding paper.

• This Journal, 29, 971-3 (1907).

93 O W. D. HARKINS AND R. E. SWAIN.

able that hay which contained so much arsenic could have been grown in
the year when the smelter was closed for the three summer months, and
upon investigation it was found that the hay had been grown about three
miles north of the smelter during the preceding year.

In October, a sheep owner who lived about twenty-eight miles from the
smelter, and in a small valley somewhat protected from the smoke by an
intervening range of hills, foimd that his grass was becoming exhausted.
He rented a low field lying approximately fifteen Yniles northeast
of the smelter, and drove his three thousand five hundred sheep from his
home to this field, upon which there was a large amount of grass. After
pasturing there for a week, a number of the sheep became sick, and the
neighbors advised the owner to drive the flock to a feeding place farther
from the smelter. On the way home five hundred sheep died, but during
the next four weeks the mortality diminished. The total loss was six hundred
and twenty-five animals. The case was investigated by the state veteri-
narian. Dr. M. E. Knowles, who decided that the sheep had died from acute
arsenical poisoning. The four samples submitted by him for analysis gaw
the following results:

Tabls I.

Number. Organ. Parts ASfOs per milUon.

I Stomach 3 . i

2 Stomach Trace

3 Stomach and liver 3.0

4 Stomach and liver 4.0

Claims were made that the sheep had died from poisoning caused by the
alkali of the soil. A complete analysis of the soil, and a partial analysis of
the stomachs of the sheep, showed that this was not true. It is a fact that
the soil was high in the salts of the alkalis, containing 0.70 per cent, soda
(NajO), 1.38 per cent, potash (K^O), 0.40 per cent, sulphuric acid (SOJ,
0.00306 per cent, arsenic trioxide, 0.0 118 per cent, copper, together with lead,
a trace of antimony, etc.

A visit was made to the field to see if the cause of the death of the sheep
could be determined. The grass was cropped very close to the ground, and
over the lower part of the field there was a large amount of moss which had
been greatly disturbed, presumably by the sheep during their feeding.
On analysis it was found that the grass contained fifty-two, and the moss
four hundred and five parts of arsenic trioxide in a million. It was there-
fore a reasonable assumption that the sheep had been poisoned at first by
eating the grass, and in a greater degree by eating the moss at a time when
the grass was nearly exhausted. It is probable that the sheep were wry
hungry when they first reached the field as they had lived for some time
upon a meagre food supply.

There were no more complaints of serious damage from the fanners, so

ARSENICAL POISONING OI^ HEllBlVOkOUS ANlMAtS- 93 1

far as the effects of arsenic were concerned, until late in the autumn of
1904. It is true, that, while the effects of sulphur dioxide upon the plants
had been moderated on the average, it had become more severe at many
points some distance from the smelter. The charge began to be made by
the farmers that the same was true of arsenic, and that symptons of arseni-
cal poisoning had again manifested themselves in spite of the use of the
big flue and stack. Chemical investigations were actively renewed in
January, 1905, and have continued up to the present time. The results of
the analyses of smoke and forage have been given in the previous papers
of this series. A reference to these will show that enormous quantities of
arsenic were thrown out by the high stack, and that large quantities were
present on the vegetation.

Autopsies of a large number of animals were made, and many of the
samples analyzed. Great care was taken with all of the analyses so that
quantitative results of considerable accuracy might be obtained, and some
of the precautions used will be described in a subsequent paper. The
methods were first tested by analyses upon known amounts of arsenic.
Various substances were used in the decomposition of the tissue, but the
methods of Fresenius, von Babo, and Chittenden-Donaldson were adopted
almost exclusively. All reagents were scrupulously purified from the
most minute traces of arsenic, and were tested very frequently during the
course of the analyses. All of the glassware, glass tubes, rubber tubes, por-
celain-ware, etc., were tested to see if they would give traces of arsenic
when used in the manner necessary for the tests. Berlin porcelain was
used to the exchision of glassware, except for the Marsh apparatus, which
was made of Jena glass.

A modified Chittenden method which was much used in cases where
little fat was present, consisted in putting 100 grams of the sample into a
large casserole, and adding 100 cc. of nitric acid. The mixture was stirred
frequently until it had become liquid, and was then heated on an asbestos
board which rested on an electric stove. After heating for a half hour
longer the solution was cooled, and thirty cubic centimeters of concentrated
sulphuric acid added. The solution was heated until it began to turn
brown, and then the steam of a dropping funnel was introduced through
a hole in the watch glass cover, and nitric acid allowed to drop into the hot
solution just rapidly enough to keep it from turning dark. After this
addition has been kept up for some time, the solution could be evaporated
(while adding nitric add drop by drop) until the fumes of sulphuric acid
appeared, without a blackening of the solution. A part of the sulphuric
acid was then evaporated, nitric acid being occasionally added. The
cooled solution was diluted and re-evaporated in order to remove oxides of
nitrogen which might be present, and the cooled and diluted solution used
for the determination of the arsenic by a modified Marsh method. The

932 W. D. HARKINS AND R. E. SWAIN.

generator was kept active for six hotirS; stannous chloride being used to
increase the activity of the zinc. It was found that this length of time was
essential where weighable quantities of arsenic were determined. The
best results were obtained by the use of a special fire-brick furnace with
four very large burners. A second hard glass tube lying in a second
furnace was connected with the first in order to test the completeness of
the decomposition of the arsine, and arsenic was always found in this
tube in cases where action in the generator xbecame at all rapid.

The results of the analyses of a part of the animal samples are given in
Table 2.

Tabi^b 2.-

—Amounts op Arsbnic in

THE Tissues op Animals.

No. and organ.

Date.

Animal.

Distance.

Parts A%Ot per milliom

190a

I h

Sept.

Horse

4.7

2 L

Sept.

Horse

SSE

I.I

Sept.

Horse

2.8

Sept.

Horse

0.8

5 L

Sept. •

Cow

6 L

Sept.

CaU

6.2

7 L

Sept.

Calf

1.5

0.4

8 L

Sept.

Calf

1-5

0.3

Sept.
1903

Steer

8.6

10 S

Oct.

Simp

NNE

3.x

II S

Oct.

Sheep

NNE

Tnct

12 L&S

Oct.

Sheep

NNE

0.7

13 h&S

Oct.

Sheep

NNE

4.0

14 L

Oct.
1905

Horse

1.5

130

15 L

Jan.

FiUy

O.OI

16 Lu

Jan.

Geld

17 s

Jan.

Geld

O.OI

18 L

Jan.

Cow

3-3

19 L

Jan.

Geld

4.5

NNE

35.0

20 K

Jan.

Mare

NNE

13.3

»2I L

Jan.

Colt

2.6

22 L

Jan.

Cow

1.6

'23 h

Jan.

Cow

2.1

24 L

Jan.

Cow

II. 9

25 L

Jan.

Cow

10.3

26 L

May

Steer

NNE

O.OI

27 h

May

Cow

O.OI

28 L

Nov.

Cow

NNE

10.00

29 L

July

Cow

7-4

»30 L

Jan.

Sheep

O.OI

*■ The shoulder of colt 21 was covered with a green fat which contained 288 paits
of copper to the million. Sample 30 L contained 592 parts ; and sample 23 L, 88 put
of copper to the million.

ARSENICAL POISONING OF HERBIVOROUS ANIMALS. 933

TablB 2 (ConHnued).

No. and organ.

Date.

Animal.

Distance.

Parts AssOs per 1

31 L

Jan.

Sheep

6.8

32 L

Jan.

Calf

1-5

33 L

Jan.

Calf

6.5

34 h

Feb.
X906

Calf

1-3

35 L

Nov.

Geld

6.0

36 L

Oct.

Geld

3-4

37 L

Aug.

Mare

4-5

4.4

38 L

Aug.

Geld

1-7

ssw

3-9

39 L

Aug.

Horse

NNE

16. 1

40 L

Aug.

Horse

NNE

9-5

41 L

Jan.

Mare

NNE

42 L

Sept.

Geld

4.2

NNE

14.8

43 K

Jan.

Mare

2.1

44 L

Jan.

Mare

8.7

45 L

Jan.

Geld

NNE

7.6

46 L

Jan.

Geld

5.3

47 h

Feb.

Filly

4.5

NNE

3.1

48 L

Feb.

FiUy

17.8

49 I<

Feb.

Mare

4-5

NNE

8.7

50 Br

Feb.

Mare

4.5

NNE

4.5

51 H

Jime

Filly

4-2

NNE

460.0

52 L

Mar.

Horse

1.5

20.7

53 I-

Aug.

Mare

NNE

4.99

54 B

Aug.

Mare

NNE

20.67

55 L

Sept.

Horse

4-5

NNE

Trace

56 h

Sept.

Mare

6.5

52.5

57 L

Sept.

Geld

4-5

NNE

3.3

*58 K

Feb.

Colt

4-5

NNE

19.8

59 I^

Feb.

Colt

NNE

Trace

60 L

Feb.

Colt

4.5

NNE

255

61 L

July

Colt

NNE

31.7

62 L

July

Colt

4.4

63 L

Nov.

Colt

NNE

2.6

64 L

Oct.

Colt

4-7

65 L

Aug.

Colt

2.2

66 L

July

Colt

NNE

I.I

67 h

Feb.

Cow

II. 8

68 L

Feb.

Cow

SSE

33.8

69 L

Nov.

Cow

NNE

10. 0

70 L

July

Cow

14.2

71 K

July

Cow

6.2

72 L

July

Cow

63.12

73 L

Nov.

Cow

SSE

9.2

74 U

Nov.

Cow

SSE

16.0

75 Iv

Nov.

Cow

NNE

1.2

76 L

Nov.

Cow

4-2

NNE

10.4

77 L

Jan.

Steer

5-4

78 L

Oct

Calf

1.7

16.2

No. and orffan.

Date.

79 1/

Feb.

80 L

Jan.

81 L

Aug.

82 L

Sept.

934 W. t>. HAkKtNS AND R. B. SWAlN.

TablB 2 (Caniinued).

Animal. Distance. Parts ASfOs per millioa.

Calf 3 SSW 6.3

Sheep 10 N 8.9

Sheep 4.2 NNE 5.0

Pig 1.7 SSW Trace

L represents a sample of the liver; B, of the bone; Lu, of the lungs; U, of urine;
H, of hair; K, of kidney; Br, of brain; S, of stomach. The animals represented bv
samples i to 42 were more emaciated on the average than those represented by sam-
ples 43 to 82.

The number of parts per million multiplied by seven-tenths gives the
number of gmins of arsenic trioxide to one himdred pounds of tissue. In
this way a comparison may easily be made with the recommendation of the
Royal Commission on Arsenical Poisoning:*

**In our view it would be entirely proper that penalties should be im-
posed under the sale of Food and Drugs Acts upon any vender of beer or
any other liquid food or of any liquid entering into the composition of
food, if that liquid is shown by an adequate test to contain i/ioo of a grain
or more of arsenic in the gallon ; and with regard to solid food — no naatter
whether it is habitually consmned in large or small quantities, or whether
it is taken by itself (like golden syrup) or mixed with water or other sub-
stances (like chicory or 'camos') — ^if the substance is shown by an ade-
quate test to contain i/iooth gmin of arsenic or more in the pound."

According to this recommendation a large number of the samples of
liver given in Table 2 would be considered as deleterious when taken as
food by human beings, and in one case a cow's liver contained forty-three
times the maximum amount allowed in food by the commission. Flesh
and also milk were found which exceeded the limit prescribed by the com-
mission.

The livers of animals, according to Table 2, contained from a trace to
63.12 parts of arsenic trioxide to one million parts of tissue. The maxi-
mum number was obtained in a case of acute poisoning which occurred
about three miles north of the smelter. This seems a strange case when it
is considered that, although the grass of the ranch has usually contained
a large amount of arsenic, at the time the cow died the percentage was
relatively low, being between thirty and forty parts to the million.

An effort has been made to trace a relation between the quantity of
arsenic ingested with the food, and that contained in the livers of the
animals, but this has been impossible. It is true that the amounts of
arsenic are larger on the average in cases where the animals were kept dose
to the smelter, but it is obvious that the condition of the animal is the
more important factor in determining the amount of arsenic retained by
the tissues.

^ Final Report Royal Commission on Arsenical Poisoning, p. 50 (1903).

ARSENICAL POISONING OF HERBIVOROUS ANIMALS. 935

The amounts of arsenic present in the organs of the animals is in many-
cases small, yet no smaller than might reasonably be expected in chronic
arsenical poisoning following the repeated and regular administration of
moderate doses of arsenic. Before the time of this investigation, very
little experimental work had been done on horses and cattle in which
accurate anal)rses had been made of the organs following arsenical poisoning.
In a case reported by the Russian Minister of the Interior, the ^ver of a
cow which had been fed considerable amounts of arsenic for a period of six
months, contained 0.13 part of arsenic per million. Spallanzani and
Zappa* fed a cow from 0.4 to 0.5 gram (6 to 8 grains) of arsenic trioxide
daily for 44 days and the following results in parts of arsenic per million
were obtained from an analysis of the viscera: — ^stomach, 19; liver, n;
kidneys, 4.5; spleen, 7.6; lungs, 3; muscles, 3.8.

In order to see how the results of Table 2 would compare with those
obtained from animals killed by arsenic, and also in order to secure data
as to the poisonous dose, horses were fed upon arsenic in different forms.
The doses given were large as the time for the experiments was very short.
A horse was fed upon flue dust containing a total of 20.65 P^r cent, of arsenic
calculated as trioxide, and 17.89 per cent, of soluble arsenic, also calculated
as trioxide. Considering only the soluble arsenic, the horse was fed two
grams of arsenic trioxide for eighteen days in addition to hay containing
about 0.0030 per cent. In the liver was found 3.5 parts, and in the kidneys
18.0 parts per million, an amount for the liver which was less than the
average of the values given in Table 2. A second horse was given 2.8
grams (o.i oz.) of arsenic trioxide in two doses, on the first day mixed with
bran, and on the second ingested as a drench. On the fourth day the
animal died, and on analysis the liver was found to hold 8.7 parts per
million. A third horse died on the third day after having been given two
doses of 7.5 grams each, one on the first, and one on the second day. The
liver contained 12.2 parts, while that of a sheep which had been fed arsenic
for some weeks contained 11.9 parts to the million.

In taking samples during the first few years of the case, the more emaci-
ated animals were usually selected, but beginning with January, 1906, a
larger number of those that were fat and in a seemingly good condition
were chosen. The result of this change of policy is shown in the table,
where the average content of arsenic for 1906 is much higher than for pre-
vious years. «

The elimination of arsenic probably begins very early and persists during
the whole period of its absorption. In the human subject it often appears
in the urine within five hours after ingestion, and may continue to be
eliminated for thirty days after the last dose is taken. This is tmusual,
however, fifteen days usually sufiicing to remove almost all of the arsenic

' Annql di AgricoUura, 131, 25.

93^ W. D. HAREINS AND R. E. SWAIN.

from the human system. The difficulty in finding in the tissues any con-
siderable amount of arsenic — an amoimt for example, sufficient to prove
on the basis of a chemical analysis alone that death was due to arsenical
poisoning, lies here. And where the case is one of chronic poisoning ex-
tending over a long period, and caused by a fairly constant amount of the
poison being ingested daily, the isolation of an amoimt approximating a
toxic dose is often impossible, simply because a really toxic dose was never
taken at one time, and what was taken was partly excreted by the kidneys
even before its absorption from the stomach and intestines was compkte.

The proof of poisoning is complete, (i) "when the symptons known to
be caused by the poison have been observed dining life ; (2) when the post-
mortem examination shows the presence of such lesions as it is capable of
producing, and the absence of other causes of death ; (3) when the toxic
agent is demonstrated to be present in the cadaver or dejecta of the animal
poisoned." It is not always possible to present evidence along all these
lines, for it has often occurred, even with a poison so prompt in its action
and of such certainty of detection as arsenic, that life may be prolonged
for a sufficient length of time to permit the total elimination of the poison,
and death results from its action by a continuation of the morbid processes
which it established. Again one of the symptoms of chronic poisoning
through arsenic is loss of appetite, so that often during the last few weeks
or days of life, little or no food is taken. Then where the poison accom-
panies the food, and is proportionate to it in amount as in the cases at issue
in this investigation, failure to take any considerable amount of food during
the last few weeks of life, stops the ingestion of the poison and allows the
system to expel all or nearly all the substance before death ensues. In
certain of the cases given in Table 2, animals apparently in a diseased
condition were slaughtered. This was true of cow 31, whose liver showed
6.8 parts of arsenic trioxide to the million, or a total of 0.5 grain for the en-
tire organ. It also carried 1.5 grains of copper. Though considerably
emaciated, the large quantity of food in the stomach showed that the ani-
mal had not lost its appetite, and this was substantiated by the statement
of the owner. A sheep from ten miles north of the smelter, on the other
hand, though slaughtered, was virtually in a dying condition, and had
evidently partaken of but very little food for some time. Only a slight
trace of arsenic could be detected in the liver, but a surprisingly large
amount of copper was present. Still the animal was undoubtedly suffering
from arsenical poisoning, and the reason so little arsenic and so much copper
were found is that arsenic is rapidly eliminated while copper is very slowly
excreted, being retained mainly by the liver. In some of the cases very
large amounts of copper were found in the fat, a part of which had a green-
ish tinge.

Moderate amounts of arsenic continuously administered, cause an in-

ARSENICAL POISONING OF HBRBIVOROUS ANIMALS. 937

crease in body weight and much increased storage of fat between the
muscles as well as around the kidneys. While this is true in general of all
animals it is notably true of herbivorous animals. In minute doses arsenic
improves the appetite and increases both the motions and secretions of the
stomach and duodenum; and since there is no considerable accumulation
of arsenic, due to its rapid elimination through the excretory channels,
medicinal or smaller doses may be administered daily over a prolonged
p>eri<xi without showing harmful effects.

Reliable data on the subject of the arsenical poisoning of live stock are
very meagre, and most of the statements fotmd in the usual text books are
so conflicting that a definite conclusion as to what may be considered a fatal
dose of arsenic for a horse, cow, or sheep, cannot be reached through them
alone. Much of the most reliable work has appeared in the chemical jour-
nals.

Spallanzani and Zappa ^ fed moderate amounts of arsenic continuously to
a ** Durham" cow for 46 days, when death resulted. From 0.5 to 3.0
grains (7.7 to 46.3 grains) of arsenious oxide were administered daily,
the dose being gradually increased to the maximum of 3 grams, when the
animal died. Spallanzani concludes from this and other experiments
that cattle will take without injury, over indefinite periods, doses of 0.5
to 0.7 gram (7.7 to 10.8 grains) of arsenic trioxide per day, and may indeed
increase in weight under it. They first show toxic symptoms with doses
of I gram (15.4 grains) per day. The maximum non-toxic dose for cattle
is given as about 0.00015 part of arsenious oxide per day for 100
parts body weight, or 10.5 grains per day for an animal weighing i,ooo
pounds.

The results cited in the last paragraph are well in accord with the re-
sults of the investigations of the writers as made on the animals of smelter
districts. A review of the literature of the subject reveals such great
discrepancies in regard to the fatal dose that it is almost impossible to
believe all of the results cited. On the one hand, we have the work of
Cameron' which shows that ten cows were killed by one dose for each
cow, of 8.4 grains of arsenic trioxide in the form of sodium arsenite. In
contrast with this case, which seems to be authentic, we have the state-
ment attributed to Hertwig* that he gave arsenic to eight different horses
in doses beginning with twenty grains but increasing to a dram, and
continued these doses for from 30 to 49 days with no bad effects, in fact,
"the condition was improved."

On account of the unsatisfactory state of the literature of this subject,
it was decided to inaugurate ftuther experiments to test the effects of

' AnnoH di AffricdUuta, 131, 25.

* Analyst, 1888.

» Vetennairian, 1843, p. 345.

93° W. D. HAKKINS AND R. B. SWAIN.

different doses upon cows and horses. However, at this time it was not
found possible to meet the expense of such an undertaking, so the work
was done upon sheep.

Results of Work onSheep, — Four of the healthiest sheep were chosen from
a flock of several hundred. They were fed upon local arsenic-free hay at
Palo Alto, California, and the doses given in starch capsules as follows :
Table 3.

DDK. WeiEht at

AtmdIc triozlde. beslnDfog. Fom of ■nenlc

t o. 181 gram twice a day 95 lbs. Arsenk trioride

3 0.133 gmm once a day S7.5 lbs. Soditun aTScnite

3 0.055 S'^tti once a day 115 lbs. Sodium aisenite

0.021

4 or gram oace a day 90.5 lbs. Sodium anenite

0.090

The results of this experiment are given graphically in Fig. i. Sheep
No. 4 was given daily doses of 0.02 1 gram for 35 days, when an increase to
0.090 gram was miade, because it was beUeved that upon the smaller
dose the sheep would not die before the conclusion of the experiment at the
end of ninety days. This was the only sheep that did not die, but that

Fig, 1. — Effect of arsenic trioxide on the weight of slieep.
N9TB.— Curve I is raised ten units of weight in order not to interfere with Curre^IV.

ARSENICAI^ POISONING OF HERBIVOROUS ANIMALS. 939

death would have resulted soon after the expiration of the time set for the
dose of the experiment is evident from the curve (IV) which shows a
rapid decrease in weight toward the end of the period.

The experiment shows that 46 milligrams (o . 7 grain) of arsenic trioxide
per day, administered in the form of arsenite of sodium, to 100 pounds of
body weight, is sufficient to cause the death of a sheep. The perfectly reg-
ular way in which the arsenic reacted upon the sheep as expressed in the
curves of the body weight, at least suggests strongly that the result was not
due to individual susceptibility.

The case of sheep No. 2 was an instructive one. It was given a dose of
o. 123 gram per day for twenty-five days, and lived eight days longer be-
fore death ensued. No food was taken during the last thirteen days, and
practically none for eighteen days, though fresh food was offered three
times each day. During the eighteen dajrs the animal was practically in a
comatose condition, suffering no pain, and reclining upon its side most of
the time. On dissection the intestines were found absolutely empty, since
an attack of diarrhea had lasted for eight days, while the stomach was
greatly distended and packed with solid food. Digestion had been ab-
solutely suspended for a long time, and decomposition of the stomach lining
had already begun. At the begirming of the feeding the animal weighed
87.5 pounds, and at the end 56 potmds, of which eight pounds was undi-
gested food packed in the stomach. A number of the doses of arsenic were
found undigested in this organ. To each million parts the liver contained
a trace, the tissue of the stomach 3.0 parts, and the brain 4.2 parts, cer-
tainly a peculiar distribution of the poison. Evidently little arsenic had
gone into the circulation from the stomach for a considerable period, so the
Uver had been able to eliminate most of the arsenic. The post-mortem
appearance of the organs of the sheep, taken as a whole, was that of acute
rather than long standing chronic arsenical poisoning.

The question of the amount of arsenic which will kill a farm animal, if
fed daily, is a very important one to the chemist who undertakes to
investigate the conditions existing in smelter regions. The effects de-
pend so greatly upon the conditions that even after such an extensive in-
vestigation as that carried out by the veterinarians, pathologists, bacteri-
ok>gists and chemists, upon the present case, no very definite statements
can be made in regard to this point. A study of Table i of the second paper
of this series will give some idea of the poisonous dose, for on almost all of
the ranches listed, animals have been supposed to die from arsenical poison-
ing. On the other hand, there is almost no place in the farming district
where some of the animals will not survive. As has already been indicated,
there is comparatively little sickness during the late spring and summer,
\mt by November a large number of animals are affected, if they are allowed
to[run upon the pastures.

940 W. D. HARKINS AND R. E. SWAIN.

The following table gives the average amounts of arsenic trioxide in the
grasses analyzed during the last three years:

Table 4. — Average Amounts of Arsenic Trioxtoe in Grass and Hay.

AssOs in parts per million. Grains to 25 pounds food.

« . ■ ■■■> — ^ / * \

Year. Grass. Hay. Grass. Hay.

1905 106 45 18.6 7.9

1906 155 42 21.7 7.4

1907 100 17.5

The column '^Grains to 25 pounds food" is supposed to represent the
amount of arsenic taken in a day's feeding, since this is the amount of dry
matter in the daily ration for this region. On the average, then, the daily
ingestion of arsenic is about 20 grains for grass, and 7 . 5 gmins for hay.
The amount varies from a minimum of i . 75 grains for young meadow grass
taken two miles south of the smelter, to 271.4 grains for a sample taken in
the smelter field one fourth mile from the old low stacks. The average for
the grass is interesting in comparison with the statement of Spallanzani
and Zappa, that one gram (15.4 grains) per day is the minimum amount
which can give rise to the toxic symptoms in cattle.

The fanners claim that animals which are shipped into the valley suc-
cumb more quickly than those that have lived for some time in the district,
and this is undoubtedly true. It might be assumed that this means the
animals able to survive are those of great individual resistance and tolerance
with respect to arsenic, an assumption which is true in part; but, in addi-
tion to this, there is little doubt that tolerance is gradually established to a
certain extent by the use of the poison, as is the case with human beings.

In order to see what proportion of the arsenic in the plants would be
soluble in the digestive juices of the animals, and thus act as a poison, two
digestion experiments were made. A sample of hay or grass, 300 grams,
was digested at 38° C. for two days with a glycerol extract of the mucus lin-
ing of the abomasum, the mixture being made acid with hydrochloric acid.
Then the liquid was made slightly alkaline with sodium carbonate, digested
five days with the glycerol extract of two pancreas glands, and toward the
end of the time putrefactive bacteria were added. The results were as
follows :

Grass. Hay.

5 mi. N. 4 taC N.

Percentage of soluble As^O, 0.0242 0.0058

Percentage of insoluble AsgOj 0.0019 o. 001 14

Total per cent. As^O, ... 0.0261 0.0069

Percentage of total arsenic which is soluble 92 . 6 83 . 6

The greater amoimt of arsenic is undoubtedly in a poisonous form.

The Distribution of Arsenic in the Organism, — Several animals were sam-
pled in such a ^vay that portions of nearly every organ were taken, but only
one set of data will be presented — for a case in which the distributioii

ARSENICA!, POISONING OF HERBIVOROUS ANIMALS. 94I

seemed normal in comparison with our other results, although the amount
of arsenic is lower than the average. On November 4, 1906, a horse was
killed and sampled. In general condition, the animal was unthrifty, and
its coat was very rough. The clinical symptoms were redness in the stom-
ach and intestines, congestion of the lungs and pleura, congestion of the
brain and bladder, slight congestion of the kidneys, catarrh of the intestines
and an enlarged spleen. The results of the analyses of the organs are pre-
sented in Table 5.

Table 5. — ^Thb Distribution op Arsenic in thb Tissues op a Horse in a Case op

Chronic Poisoning.

AsiOs in parts
No. Pood or organ. per million.

1 Grass No. i 45 -oo

2 Grass No. 2 107.00

3 Dust from hay 9190.00

4 Ulcer in nose 658 .00

5 Contents stomach wet 25 .00

6 Contents stomach dried 398 .00

7 Urine 5900

8 Hau: of tail 58.00

9 Liver 6 .00

10 Thyroid gland 6.00

ri Stomach 4. 70

12 Spleen 4.60

13 Pancreas 4.40

14 Small intestines 4.00

15 Brain 3.30

16 Spinal cord 2 . 60

17 Muscles 2.50

18 Iftmgs 2.20

19 Bones 2.20

20 Heart 2.10

21 Bladder i . 40

22 Kidney 1 . 40

23 Right parotid o . 80

24 Fat 0.70

25 Suprarenal 0.06

26 Fluid around heart 0.05

27 Blood 0.03

The horse had fed upon grass and hay containing from twenty to forty-five
parts of arsenic trioxide to the million for a period of some months, and for
three days had been eating grass containing 107 parts. For several years
it had been fed upon grass and hay containing arsenic in varying amoimts.
Dust shaken from the hay stack in the field where the horses were pastured
for three days, contained 9190 parts of arsenic trioxide.

Arsenic in Milk, — ^Ten samples of milk, most of them mixtures from sev-
eral cows, were obtained by milking directly into bottles provided with
glass stoppers. The analyses gave :

942 W. D. HARKINS AND R. B. SWAIN.

Table 6. — ^Arsenic in Milk.

Distance from

Parts As^O«

Grains to

No.

Date, 1906.

amelter.

No. of cows.

per million.

100 gallons.

Sept. 21

4-5 NNE

2.94

17.7

Nov. 3

3 E

0.47

2.83

Nov. 3

6 NNE

0.70

4.00

Nov. 3

4.5 NNE

0.18

1.08

June 28

4.5 NNE

1.40

8.42

June 27

5 E

1. 00

6.02

July 2

3 SE

4.20

25.28

June 28

3 N

340

20.47

June 23

3 W

5.70

34.30

The Arsenic Content of Ulcers of the Nose. — A complaint among the horses
. of smelter regions called the "sore nose" has been observed by the writers
in the Anaconda and Salt Lake regions. It is also reported that the same
disorder is fotmd in Great Falls, and among the horses of the smelting
region in Cornwall Of the persons who worked on the smelter stacks in
determining the arsenic content of the smoke, two became affected with an
arsenical rash upon the face, while the nostrils of the third were almost
closed by a swelling caused by the irritant action of the flue dust. In the
case of horses, one nostril may become closed almost absolutely on account
of an ulcer which forms on the lower portion of the nasal partition. Sev-
eral of these ulcers were taken from the nostrils of different horses and
analyzed with the following results:

No. Distance. Parts ASfO«. No. Distance. Parts AaiO».

1 • 2 miles S 254 4 2 miles S 902

2 4.5 miles NNE 587 5 3 miles SW 54s

3 3 miles SE 1015

Undoubtedly the highly arsenical dust from the hay and grass lodges in a
fold of the nostril and irritates the mucous membrane tmtil the nose scab is
formed. The dust from the hay of the ranch where samples 3 and 5 were
obtained, contained 9190 parts of arsenious oxide per miUion.

In order to see if the cases observed in Salt Lake could be due to this
cause, samples were taken in different parts of the district. These samples,
taken September 3, 1905, contained from ten to sixty parts of arsenic tri-
oxide to the million, results which would indicate that sufficient arsenic is
present to cause the observed effects.*

Arsenic in the Hair, — According to the evidence of Mann' arsenic is local-
ized in and eliminated by the hair. Large amounts of arsenic were found
in the hair of animals of the Anaconda region. The hair of the tail of one
horse contained fifty-eight parts of arsenic trioxide to the million, an
amount ten times as great as that found in the liver. The hair of a colt con-
tained 605 parts, the liver 4.4 parts, and the bone 13.2 parts, while the

* For other results from this district see preceding paper. Table VII.

* Report of the Royal Commission, p. 13; and Minutes of Evidence, Vol. i, p. 139^

ARSENICAL POISONING O^ HERBIVOROUS ANIMALS. 943

grass in the field where the colt was feeding contained only ten parts per
million. The hair of a filly pastured five miles north of the smelter gave
460 parts of arsenic. The case is more complicated than those investigated
in England, since an unknown fraction of the arsenic in the hair was un-
doubtedly deposited from the atmosphere. Nevertheless, the results are
striking and important.

Normal Arsenic, — ^Attempts have been made to show that the amounts
of arsenic present in the tissues of the animals of smelter regions represent
what may be called normal arsenic. This is certainly a perversion of the
conclusions of Gautier and Bertrand, for the results obtained by them were
of a totally different order of magnitude from those obtained in forensic
cases; and they found arsenic only in the thyroid, th)rmus, brain and skin.
Even these results have been criticized by Kunkel,' Hddlmoser,* Cemy,*
Stevenson and Mann.^ The latter writers claim that the arsenic present is
wholly adventitious.

No attempt was made by the writers to test Gautier's conclusions, but
about forty-five livers from Palo Alto, California, New York City, and
Missoula, Montana, were analyzed, using samples of from one hundred to
eight hundred grams. Using tests which would detect the presence of
I / 1000 mg. of metallic arsenic, in no case was arsenic found. This is suffi-
cient proof that the question of normal arsenic need not come into smelter
smoke investigations.

Symptoms. — The following three examples may be taken as typical cases
which together exhibit the range of important symptoms which have ap-
peared in coimection with arsenical poisoning in the Deer Lodge Valley.

Case (a). — ^The first example is that of a roan mare owned by a farmer
living about eight miles north of the smelter. She was eight years old,
weighed thirteen hundred pounds, was sleek and fat, and so far as outward
appearances went, perfectly sound. The owner stated in answer to ques-
tions, that the animal had fed rarely on pasture but almost entirely on hay,
of which she ate much more than the average ration. For nearly a year,
however, she had been failing in strength, and was no longer able to do an
ordinary day's work, profuse perspiration and total exhaustion following
any unusual exertion. An examination showed a "sore nose * ' scar in one
xiostriL

The animal was shot and the autopsy made immediately thereafter.
The urine was white in color and heavily sedimented; the heart "flabby"
and larger than normal ; the lungs were covered with a mattery adherent
exudate and were "flabby," due to the blocking of the bronchii with a

» Z. physiol, Chem,, 44, 511-529 (1905).

• Ibid., 33f 329-344 (1901).

•/6jU, 34»4o8 (1901).

^ Minutes of Evidence, Royal Commission on Arsenical Poisoning.

944 W. D. HARKINS AND R. E. SWAIN.

yellow caseous substance, and the escape of air into the tissue's; the liver
weighed 1 7 pounds, 7 pounds above normal ; the mucous membrane of the
stomach was reddened over certain areas in the fimdus ; the small intestines
showed a great many extensive diffuse red patches; the large intestines
were generally badly reddened, and here and there distinctly eroded. An
unmistakable odor of garlic was observed when the intestines were opened.
The mucous membrane of the uterus and bladder was notably reddened,
and the right ovary was gorged with a deep red gelatinous substance. The
bone marrow was of a deep yellow color, due to a quantity of yellow oil
which filled the interstices and retained its liquid character at ordinary
temperatures.

Case (b). — The second is a colt, one year old, which was posted two miles
south of the smelter on July 3, 1906. It began to appear unthrifty during
the fall of the preceding year while on the hill pastures, and at no time
thereafter did it show a normal growth. The autopsy showed that this
was a case of remote chronic rather than of acute poisoning. The secreting
mucous membranes throughout the body were reddened in patches, but
only slightly. The hair was shaggy and lusterless, and the whole organism
weakened and emaciated.

Case (c). — ^The last case is that of a colt, eleven months old, which had
developed normally during the sucking period of six months, when it was
weaned and removed to an adjoining pasture early in the month of March.
From that time until July ist it subsisted on hay from a stack in the field.
On the date mentioned, following an effort to rope it, the colt had a **&"
and died in convulsions. The stomach showed three distinct, crater-like
ulcerations and extensively irritated areas, especially in the folds- The
lungs were badly discolored as in necrosis or fatty degeneration. The
organ was certainly badly affected, and as soft as a partially decomposed
organ. The intestines were highly inflamed, often for five to eight inches
in one place, and the whole of the small intestine was covered with inflamed
patches.

The first of the cases cited is apparently one in which the arsenic showed
its usual stimulating action, strengthening the appetite, promoting the di-
gestion, causing the deposition of much intestinal fat and giving a sleek
appearance to the subject. The action was proceeding beyond this stage
however, and a breaking down of many of the organs was in evidence.
This was a case of progressive chronic poisoning, shown further by the
large amoimts of arsenic in the bone (20 . 67 parts per million).

The second case seems to be one of true chronic poisoning which evidenth'
took another course, the stimulating effect being constantly overshadowed
by the more destructive action of too large doses. The organism was too
weak to recover, even after loss of appetite reduced the ingestion of the toxk
agent to a minimiun. The stomach was nearly empty, and the liver carried

ARSENICAL POISONING OF HERBIVOROUS ANIMA13. 945

only very small amounts of arsenic, wliile the storehouses for arsenic in the
organism, the bone and hair, showed excessive amoimts. The analytical
results were: hair, 605 parts; bone, 13. 2 parts; liver, 3. 3 parts; and heart,
1 . 7 parts of arsenic trioxide per million.

The third case is one of recent chronic poisoning, leading rapidly to a
culmination in a way truly characteristic of the substance. The liver of
this colt contained 31 . 7 parts, and the kidneys, 2 . 4 parts of arsenic trioxide
per million.

Dtuing a considerable time the writers were associated in this work with
Dr. D. E. Salmon, of the Bureau of Animal Industry, who very kindly
prepared an outline of the more prominent symptoms of the animals as they
appeared to him, for use in this publication. These are given as follows :

Symptoms of the Chronic Arsenical Poisoning caused by the Vegetation

of Smelter Regions.

Horses. — i. Raised red line at the base of incisor teeth; 2, breath of a
garlic odor; 3, loss of spirit, vigor, and endurance; 4, falling of hair; 5,
retention of old hair; 6, ulcers of the nose; 7, weakness and impercepti-
bility of pulse; 8, erosions on the outer side of gums; 9, puffiness above
the eye; 10, rough lusterless hair; 11, partial paralysis of hind limbs; 12,
with more acute form: (a) Difficult breathing, (6) labored action of heart,
(c) dilation of pupils of the eyes, (d) partial paralysis of the diaphragm
and costal breathing.

Cattle. — I, Shrinkage of milk within a day or two after smoke has been
over pastures; 2, salivation and drooling; 3, constipation; 4, rough
scurfy coat; 5, eyes red, inflamed, and weeping; 6, loss of appetite; 7,
diarrhea when disease becomes more pronounced; 8, tucked up abdomen;
9, loss of flesh; lo, weakness, loss of vigor; 11, cough; 12, breath of gar-
lic odor; 13, droppings covered with mucus; 14, abortion and failure
to breed. ^

In an examination of the animals of a smelter district the chemist may be
greatly aided by a careful post-mortem examination, and by the histologi-
cal study of small specimens taken for this purpose. The most important
features to be seen in the sections taken for microscopical investigation
are: proliferation of the connective tissue cells, degeneration and
desquamation of the tubules in the kidneys, congestion or diapedesis, the
occurrence of hemorrhagic areas, and occasionally a total disintegration
of the cells. In some cases there is very marked fatty degeneration.
The kidneys show these symptoms more prominently than the other
organs.

The Anaconda case is of interest not only to the toxicologist, but also to

* The diminution in the human birth-rate was noticed during the Manchester
epidemic, and the results tabulated by J. Niven. See Royal Commission on Arsenical
Poisoning, Minutes of Evidence, Vol. II, Appendix 17, p. 196.

946 JULroS STIEGLITZ.

the industrial chemist and the metallurgist, since arsenic in a great number
of cases is a constituent of the metallic sulphides, especially those of copper.
Large amotmts of arsenic are given off by the smelters of Salt Lake, Utah;
Everett, Washington; Great Palls, and Butte, Montana; and from many of
the smelters of Germany, England, and other countries. In 1854 and 1875,
Haubner investigated the smelter smoke disease in the Freiburg district,
and various other cases have been studied to a slight extent.

In the decision of Judge Marshall of the Circuit Court of the United States
as made November 5, 1906, the conditions existing in the Salt Lake smelter
district are described in such a way as to make an interesting comparison
with the results of this series of papers. In speaking of the sulphur dioxide
he says: "This gas is heavier than air, and when cooled, falls to the ground
at a distance from the smelters dependent upon the air currents. When it
is brought in contact with moisture, either in the form of rain, freshly irri-
gated ground, or the moisture present in growing plants and the foliage
of trees, sulphurous or sulphuric acid is formed, which is destructive to
vegetation. Besides the emission of gas, some flue dust is emitted from the
smelters which contains perceptible quantities of arsenic resulting in the
death of horses and cows."

In conclusion, the writers wish to thank Dr. John Maxson Stillman for
the many suggestions which have been helpful in this work.

Thb Univbrsitt op Montana and Stanford Univbrsitt,

March i4« 1908.

NOTE Oir THE SOLUBIUTT PRODUCT.'

By Julius Stieolttz.
Received April 2, 1908.

Nemst' was the first to advance the theory that at a given temperature
the solubility of a difficultly soluble electrolyte in water or in aqueous
solutions of other electrolytes is dependent on a constant called the
solubility product, which is proportional to the concentrations of the icHis
of the salt, each raised to the power corresponding to the number re-
sulting from one molecule. The constant is an important one in the
theory of precipitation and solution and particularly useful in calcula-
tions of the solubility of a precipitate in mixtures that are not too con-
centrated. In Nemst's text -book on physical chemistry,' the relatioo
for a difficultly soluble binary salt — such as silver acetate — ^in water and
in solutions containing a salt with a common ion, is developed as follows:
calling the total concentrations of the difficultly soluble salt m^ and ■
in the saturated water solution and in the salt solution respectively,

^ Reported at the Chicago meeting of the American Chemical Society.

* Z. pkysik. Chem., 4, 372 (1889).

* Translation of the 4th German Edition (1904), p. 527.

NOTB ON THB SOI.UBII.ITY PRODUCT. 947

and the corresponding degrees of ionization a^ and a we have first by the
application of the isotherm of dissociation, i. e., on the basis of the law of
mass action for the two cases:

(Wo«o)* ^ K:Wo(/ — a^) (i)

and

(ma + x) ma = Km (i — a), (2)

in which (wa-f x) represents the total concentration of the common ion.
Then, on the basis of the theorem that the undissociated substance has
a constant solubility — derived from the law of heterogeneous or physical
equilibrium,

^0(1 — «o)=^ (! — «)• (3)

Combining the three equations we have

{m^a^y^ma (ma+x),

and as (m^ a^ )', the product of the concentrations of the ions in a saturated,
pure aqueous solution at a given temperature, has a definite value, we have
in general,

ma (ma + :r) - KsombUity product' (4)

Substituting the variable symbols C^^, C^HjCoo ^^^ ^CH,cooAg> ^^
y, can express the first principle for silver acetate by

CAgXCcH^OO XT / N

~ A — Jv-Ionixation.' (5)

VJCHtCOOAg '

and the principle of the constant solubility of the undissociated silver ace-
tate by

^CHaCOOAg *■ *^' (6)

Then for the constant solubility product we have simply

^Ag X CcHgCOO " ^Solnbility prodnct' ( 7)

itc in which C^^ and C^HjCoo ^present the concentrations of the silver and
% z acetate ions in any saturated solution of silver acetate, whether in aqueous
m > solution where C^^ « Cch,coo ^^ "^ ^^^ presence of other salts, where
■0.- they will usually have different values.

j:C It has long been known, however,^ that the ionization of strong electro-
^t t l3rtes does not conform to the law of mass action — ^in other words that equa-
^ I- tions (i) and (2) as used above to develop the theory of the solubility prod -
^ ^, uct do not agree with the facts in the cases of such electrolytes, e. g., salts,
, J xx for which the theory of the solubility product is especially important. The
^Tieal relation for a salt like silver acetate is that the proportion on the left
^;side of equation (5) grows larger with increasing concentration.* This diffi-
'. iculty was recognized by A. A. Noyes,' to whom we owe a large part of the

•^ ' Translation of the 4th German Edition (1904), p. 498.
■ Rudolphi, Z. physik. Chem., 17, 385 (1895).
• Ibid., 6, 241 (1890); 9, 613 (1892); x6, 125 (1895); 26, 152 (1898); 43, 336 (1903)

948 JULroS STIEGLITZ.

most exact experimental work on the solubility product, and Noyes at-
tempted to meet the difficulty by assuming that for all solutions in which
the concentration of the undissociated substance was kept constant, the
product of the ion concentrations is also a constant. Noyes calculated his
results on the basis of this assumption and used it even as a method of
determining the degrees of ionization of salts in the mixture. Arrhenius^
showed the error of such an assumption, which cannot be reconciled with
his principles of isohydric solution. The latter principle has been estab-
lished on a safe basis by the work of Arrhenius, Manson, Barmwater,
Archibald, McKay, Barnes and others, and has been found to hold, up to a
concentration of at least half -normal, for mixtures of salts of the same type
as well as for those of different types.' We find thus that the application
of the law of mass action in equations (i) and (2) to the question of solu-
bility is not justified by experience.

In an important paper published in 1899, Arrhenius* took up the investi-
gation of the second fimdamental principle involved in the solubility of
electrolytes, the widely and generally accepted theory of the constant sol-
ubility of the undissociated molecules, as expressed in equations (3) and (6).
Originally himself apparently inclined in common with practically all chem-
ists to accept* it as a matter of course, as a consequence of the law of
heterogeneous or physical equilibrium, but later led to question its correct-
ness, he tested its validity by determining the solubility of silver salts of a
number of organic acids in water and in solutions containing an increasing
excess of the corresponding sodium salts. Calculating the proportion of
ionized and non-ionized silver salt with the aid of the principle of isohydric
solutions, he found that as a matter of fact the molecular solubility is not
constant but decreases decidedly with the increasing concentration of the
total electrolyte.* A similar fact has long been known for the relative
solubility of gases like carbon dioxide, oxygen, etc., in water and in salt
solutions and Arrhenius's result should not have been unexpected.*

We thus find that the two fundamental equations on which the theory of
the constant solubility product originally was based, are both invalid and
we may well ask if this result does not thoroughly discredit and dispose of
the theory. Perhaps some such feeling led Arrhenius himself to refrsLin
from calculating the values for the solubility products for the silver salts
in his own experiments.

However, if we consider the true relations as now established for
the two cases of chemical and physical equilibrium, we have for thr

» Z. physik. Chem., 11, 391 (1893) and 31, 197 (1899).

* A. A. Noyes, Vol. IV, p. 31 1, Reports of the Congress of Arts and Science, Si. Lhus^
» Z. physik. Chem., 31, 197 (1899).

* Ibid., n, 396 (1893).

• Vide also A. A. Noyes, Reports, etc.. Vol. IV, p. 322 (1904).

• Vide also Nernst, loc. cit, p. 476.

NOTE ON THK SOLUBILITY PRODUCT. 949

equilibrium between the ions and the undissociated substance of a binary
electrolyte a proportion,

>"^Po«. ion X V^^Neg. ion >
CmoI

which grows larger with increase of concentration, * and for C|^qi the molecu-
lar solubility resulting from the equilibrium between the solid phase and
the solution, a value growing smaller with increasing concentrations of the
total electrolyte present.^ Now it is obvious that with a decreasing value
for Cjjoi and an increasing value for the whole proportion, the ion product,
^os. ion X Cijcg. ion possibly might remain constant or approximately
constant after all : it is clearly a question for rigorous experiment and calcu-
lation to determine whether the ion product does or does not remain con-
stant, that is, whether the values for the proportion and the molecular solu-
bility are inversely porportionate, the proportion growing larger to the
same extent as C^oj grows smaller with increased concentrations. Even
if it should not prove to be a real, natural constant, it might still be
found to be sufficiently constant to be of practical value and assistance in
the study of the reactions of precipitation and solution.

I have not found in the literature any discussion or investigation of the
subject from this point of view; although Arrhenius's paper was published
in 1899, even the new edition* of Nemst's text-book,*'as well as other* recent
editions of books on physical chemistry, such as Jones's ''Elements of Phys-
ical Chemistry," Mellor*s"Chemical Statics and Dynamics,"* as well as text-
books on the application of physical chemistry to analytical chemistry, do
not even refer to it or its extremely important conclusions, but develop the
solubility product as given above. I should except a statement made by
A. A. Noyes in his address before the Congress of Arts and Science at
St. Louis, in which the discrepancy shown to exist by Arrhenius between
the principle of isohydric solutions and the old hypotheses concerning solu-
bility is clearly pointed out and the need for further investigation on these
lines emphasized.*

Considerations^ -of the above nature led me then to complete the calcu.
lations of Arrhenius's experimental data on the solubilities of organic silve^

' Rudolphi, Loc. cit,

• Arrhenius, Loc. cit.

' Translation of the 4th German Edition (1904).

' P. 595.

• 1904, p. 231.

• Loc. cit.f page 321. Since the presentation of this paper, Professor Noyes has in-
formed me that work along these lines has been continued in his laboratory since 1904.

^ The theory of the constant solubility product formed an essential element in a
chemico -geological investigation carried out by me for Professor Chamberlin {vide a
forthcoming report, Carnegie Institution), and this study has resulted from that investi-
g ation.

950 JULIUS STIEGUTZ.

salts in the presence of an excess of the corresponding sodium salt, by calcu-
lating the value of the ion product in each experiment. To this end, the
degrees of ionization of the sodium salts of the fatty acids in the mixtures
used had to be calculated first, and this was done with the aid of the principle
of isohydric solutions. On account of the form in which Arrhenius's data
are presented, the following method was pursued : Arrhenius gives first a
table showing the ion concentrations and degrees of ionization of isohydric
solutions of silver and sodium acetate. From these data the corresponding
total concentrations for the isohydric solutions of the two salts Y?ere first
calculated and the results expressed in two curves in which the cube roots
of the molar concentrations and the degrees of ionization were used as co-
ordinates. The curve obtained for sodium acetate is rectilinear and that
for silver nitrate only slightly curved, so the necessary interpolations for the
further calculations were easily made. In his other tables, Arrhenius gives
the solubility of the silver salts and their degrees of ionization, and the
total concentration of the sodium salts, but not the corresponding degrees
of ionization. For instance, in the presence of o . 2667 mol. sodium acetate,
the solubility of silver acetate is 0.0203 mol., of which 50.4 per cent is
ionized. To find the degree of ionization of the sodium acetate in such a
mixture, I found from the curve for the silver salt that an ionization of
50 . 4 per cent, corresponds to a value of o . 73 for ^m and consequently the
molar concentration of the silver salt is (o. 73)* or 0.389 when it is considered
to be in its share of th^ water according to the isohydric principle. Then its
share of the liter of water must be o .0203/0 . 389 or 52 . 2 cc. This leaves
948 cc. for the sodium acetate and its concentration in 948 cc. is
0.2667/0.948 or 0.2813 molar. Then its degree of ionization is 69.6 per
cent., according to the curve. The concentration of the acetate ions in
each of the isohydric solutions, considered separately, is o . 1961 and o. i960
for the silver and the sodium acetate, respectively — shovnng thatt they
v^ere really isohydric. The agreement was not in all cases as close as this,
but with one or two exceptions mentioned below, it was satisfactory, the
concentrations of the common ion being usually within one per cent, of
each other.

In the given case, the total concentration of the acetate ion in the mixture
is therefore o. 196, that of the silver ion, as given by Arrhenius is 0.0102
and the value of the ion product in this experiment therefore :

^Ag XCcHjCoo'^O'O^o^ X0.196 or 0.00200.

All the calculations were made in this way and the following tables give
all the results obtained for the acetate, propionate, butyrate, valerate and
chloracetate of silver. In the calculations the curves for the acetates were
always used to find the degree of ionization for all these salts, following
Arrhenius's method.

NOTE ON THE SOLUBILITY PRODUCT.

951

In the tables the first column gives the molar concentration of the sodium
salt used, column 2 its degree of ionization, calculated in the way just de-
scribed, column 3 the concentration of the ionized part of the sodium salt.
Column 4 gives the total solubility of the silver salt, column 5 its degree of
ionization, column 6 the concentration of the ionized part of the silver salt
and column 7 the concentration of the undissociated part, which represents
therefore the molecular solubility of the silver salt. The last column gives
the value for the solubility product, C^gXCcH,coo- '^^^ value for C^ is
found in column 6, and Cch^oo *s ^^^ ^^"^ ^^ ^^^ acetate concentrations
given in columns 3 and 6.

Table i. — SttVBR Acbtatb at 18.6**.

Na-Acet.

looa.

io*Acet.

Ag-Acet.

lOOa'.

xo»Acct.

xoSMoL

lO^K.

• •

• • •

0.0593

72.0

42.7

16.6

182

0.0333

80.0

26.6

0.0474

68.1

32.3

15.1

190

0.0667

78.0

52.0

0.0384

64.4

24.7

137

190

0.1333

75.0

100. 0

O.Q282

58.4

16.4

II. 8

191

0.2667

69.6

186.7

0.0203

50.4

10.2

10. 1

200

0.5000

63.0

315 0

0.0147

42.8

6.3

8.4

202

Tablb 2. — Sn.vBR Propionatb

AT l8.2*>.

■

Na-Prop.

lOOa.

io>Prop.

Ag-Prop.

looa.

io*Prop.

loiMol.

io»K.

• ■

• •

0.0462

74.5

34.4

II. 8

118. 4

0.0167

82.0

»3.69

0.0393

72.1

28.3

II. 0

119. 4

0.0333

81.0

27.00

0.0345

69.6

24.0

10.5

122.4

0.0667

78.9

52.63

0.0258

65.5

16.9

8.9

118. 0

0.1333

75-3

100.37

O.O191

58.8

XI. 2

7.9

124.7

0.2667

69.8

186.2

O.OI31

50.6

6.6

6.5

127. 1

0.5000

63.2

316.0

O.OIOI

42.9

4-3

5.8

(137.7)

Table 3. — Silvbr Butyratb

AT l8.2^

Na.Butyr.

lOOa.

io»Butyr/

Ag-Butyr.'

looa.

io»Butyr/

loSMol.

xoSK.

• •

• • •

0.0224

81. 1

18.2

4.2

33.0

0.0066

86.0

56.8

0.0199

79.6

15.8

4.1

(49.8)

0.0164

84.9

13.9

0.0169

77.3

13.1

3.8

35.4

0.0329

83.0

27.3

O.OI31

73.5

9.6

3-5

35.4

0.0658

80.0

52.6

0.0091

67.7

6.2

2.9

36.4

O.1315

75.9

100. 0

0.0060

59.9

3.6

2.4

37.3

0.2630

70.1

184.4

0.0040

51. 1

2.0

36.8

0.4930

633

312. 1

0.0027

43-2

1.2

1.5

37.6

Tablb 4.—

-Silvbr Valerate .

AT 1 8.6**.

•

Ag-Val.

lOOa.

io«Val.'

Ag-Val.

XOOdU

io»Val/

loBMol.

xo»K.

• •

• • •

0.0095

87.3

8.3

1.2

6.9

0.0175

86.5

15. 1

a. 0047

81. 1

3.8

0.9

7.2

0.0349

84.0

293

0.0030

75-5

2.2

0.8

6.9

0.0689

80.1

55.9

0.0018

68.1

1.2

0.6

6.9

0.1395

75.8

105.7

0.0015

59.7

0.9

0.6

(96)

952 JULIUS STIHGLITZ.

Tabla 5. — SiLVBR Chi^ro-Acbtaib at 16.9"

Na-ClAcet.

lOOa.

io»ClAcet.'

Ag-ClAcet.

looa.

io»aAcet.'

io*MoL

xo»K.

• •

• ■

0.0644

71. 1

45.8

18.6

209.8

0.0333

79.8

26.6

0.0499

67.6

33-7

16.2

202.2

0.0667

78.1

52.1

0.0405

639

25.9

14.6

203.3

0.1333

74.8

99.71

0.0299

58.1

17-4

12.5

202.3

0.2667

69.4

185. I

0.0208

50.4

10.5

10.3

205.8

0.5000

63.2

316.0

0.0162

42.8

6.9

9 3

222.0

The experimental results of Nemst's determinations of the solubility of
silver acetate in the presence of sodium acetate and silver nitrate, respec-
tively, were also recalculated. Nemst used the isohydric principle in de-
termining the degrees of ionization of the salts in the mixtures, but con-
sidered the salts to ionize with about equal readiness. Such is not the'case,
if we accept Arrhenius's determinations of the ionization of silver acetate,
the values for which, it is true, were very largely obtained by extrapolation.
But these same values having been used in bringing the proof that the mo-
lecular solubility decreases with the increasing concentrations of the total
electrolyte, it seems most reasonable to use the values also for the deter-
minations of the solubility product of the same salt. In Tables 6 and 7 the
results of the recalculation of Nemst's data are tabulated, the degrees of
ionization being determined by the application of the isohydric principle in
the usual way — ^tlie two salts present being supposed to divide the water in
such a way as to give solutions containing the same concentration of the
common ion. The division of the water was rapidly ascertained by trial
calculations with the help of the curves of dissociation for the third root of
the concentrations. The columns in these tables have in part a different
significance from the columns of the previous tables : column i gives the
molar concentration of the sodium acetate (silver nitrate in Table 7)
used in excess, column 2 gives the portion of the water, in cubic centimeters,
in which all the sodium acetate (silver nitrate) is supposed to be dissolved
to form a solution isohydric with the solution of silver acetate in the rest
of the water. Column 3 gives the degree of ionization of the sodium acetate
(silver nitrate), column 4 the concentration of the common ion in this solu-
tion, 5 the total concentration of silver acetate, 6 its degree of ionization,
and column 7 the concentration of the common ion in the isohydric silver
acetate solution. The last column gives the values for the solubility prod-
uct.

Tablb 6. — SiLrVER Acetate at 16°.

Na-Acet.

CC. HjO.

1 00a.

io»Acet.'

Ag-Acct.

lOoa.'

io3Acet.'

KfiK.

• • •

• • ■ •

0.0603

70.8

0.0427

182.3

0.061

656

78.6

0.0731

0.0392

64.5

0.0735

185.4

0. 119

843

75.8

0. 1070

0.0280

59-7

0.1065

178.5

0.239

937

70.8

0.1738

0.0208

52.3

0.1727

188.2

AgNOs.

CC. HtO.

XOOa.

• ■ ■

• • • •

0.061

650

82.0

O.II9

825

78.4

0.230

944

74.0

NOTE ON THE SOLUBILITY PRODUCT. ' 953

Tabl^ 7. — Silver Acetate at 16®.

io*Ag/ Asr-Acet. looa.' io*Ag.' lo^K.

0.0603 70.8 0.0427 182.3

0.0770 0.0417 64.0 0.0763 204.4

O.II31 0.0341 58.6 O.II42 227.2

0.1803 0.0195 51.7 0.1809 182. I

Considering the data in the last columns of the tables, we find that the
value of the solubility product increases at most from five to ten per cent,
in mixtures down to a concentration of half -molar, * and in some of the
experiments (Tables 4, 5, 6 and 7), it shows practically no variation of
moment. All the variations may well be within the limit of errors of ex-
periment and computation in an investigation in which a large part of the
calculations are based on an extrapolated curve. Of the serious discrep-
ancies, the bracketed value in Table 3 (line 2) corresponds to an experi-
ment, the figures of which given by Arrhenius show, by the method of calcu-
lation used above, a decided divergence from the principle of isohydric
solutions — so there must be an error of observation or record in it. The
bracketed value in Table 4, last line, also corresponds to an experiment
in which there is probably some error, as is indicated by a consideration of
the values in the next to the last column of the table.

We find thus empirically that in the case of the silver salts of these or-
ganic acids the principle of the constant solubility product, faulty as its
original theoretical basis was, is sufiiciently in agreement with the observed
facts to prove of some practical value. This conclusion confirms the
results of the experimental data of others on the solubility of a salt in the
presence of other electrolytes having a common ion. Nemst's results have
already been mentioned. Findlay's^ experiments on the relative solu-
bility of lead iodide and sulphate, although the theoretical development
is open to the same criticism made above, were nevertheless calculated
correctly according to the isohydric principle and they agree with the theory
of a constant solubility product. In this case, the condition of the equi-
librium in the mixtures was determined not only on the basis of conduc-
tivities but also by measurements of electromotive forces. Bodlander's*
work on the solubility of calcium carbonate in water containing carbon di-
oxide in equilibrium with varying partial pressures of carbon dioxide

' A. A. Noyes, loc. cii., p. 322, pointed out that in the case of a solution saturated
simultaneously with thallous chloride and bromate, the product of the ion concentra-
tions of each is increased by about five per cent. Each salt gives an approximately y^
molar solution. This case was considered typical.

* Z. physik. Chem., 34, 409 (1900).

* Ibid., 35, 23 (1900). Vide also a paper by Stieglitz to be published in
the report of the Carnegie Institute of Washington, in which the data presented by
Bodl^nder are recalculated on the basis of more recent determinations of the constants
involved.

954 LAWRENCE J. HENDERSON.

led to excellent constants: the theoretical treatment started from the as-
sumption of a constant solubility product for calcium carbonate. In the
address referred to, A. A. Noyes^ mentions two or three other cases, such as
the solubility of lead iodide in the presence of potassitun iodide and of lead
chloride in the presence of potassium chloride, and of calcium hydroxide
in the presence of ammonium chloride, in which cases the theory is said to
be sustained approximately.

In view of these facts and also in view of the results of the complete cal-
culation of Arrhenius's data on the solubility of the silver salts, which re-
moved the last theoretical foundation for the solubility product constant,
we may well consider it for the present to be an approximate empirical
principle, much in the same way as so many other important principles
concerning electrolytes are rtill simply empirical, such as the isohydric
principle itself, and the various rules — Rjudolphi's, van't HoflF's, Kohl-
rausch's — expressing the equilibrium between strong electrolytes and their
ions. A great deal more exact work on the extent of the reliability of the
solubility principle will obviously be necessary to determine what the true
relations are. If it should be confirmed still further and firmly established,
the question of its theoretical bearing will become an interesting one —
particularly in its relation to the other empirical principles of solutions of
electrolytes.

Univbrsxtt op Chicago,
Chicago, III.

[Contribution prom ths Laboratory op Bioi<ogical Chemistry op the Harvard

MSDiCAL School.]

A DIAGRAMMATIC REPRESEnTATION OF EQUILIBRIA BETWEEN

ACIDS AND BASES IN SOLUTION.

By Lawrence J. Henderson.
Received March 37, 1908.

During a series of investigations concerning the adjustment of neutrality
in the animal organism,' it has been found convenient to construct dia-
grams representing the equilibria between bases and adds of diflFerent
ionization constants in solution of varying acidity and alkalinity. These
diagrams, simple consequences of the concentration law, and of the princi-
ple of isohydric solutions, seem to possess certain advantages over other
methotfs of presentation of the somewhat involved conditions. Espe-
cially are they useful to indicate immediately the adjustment of all possibk
equilibria of this sort, with at least a moderate degree of accuracy, and on
account of such practical usefulness they are here presented.

* Loc. cii., p. 322.

' For general conclusions and the literature see Henderson, American Jownal of
Physiology, May, 1908.

DIAGRAMMATIC REPRBSKNTATION OF EQUILIBRIA. 955

Let it be required to find the extent to which any add, HA, of any
ionization constant, *, present in aqueous solution together with a varying
amount of its salt of a strong base, for instance sodium hydroxide, exists
uncombined as free acid in the solution, when the hydrogen ion concen-
tration is equal to (H+).

The ionization of the acid may be expressed by the reaction,

HA - H+ + A-
whence

For all weak acids the concentration of undissociated molecules (HA),
is almost precisely equal to the total concentration of free acid, and the
concentration of the anions, (A"), is equal to the total quantity of salt,
NaA, multiplied by its degree of ionization, y. Thus one obtains the
equation

then

In this equation C is always greater than k. Under ordinary circumstances
its value may be stated as follows:

2k>C>k.
In any particular case it is an easy matter to estimate the value of C,
but for ordinary purposes very little error is involved in assuming equality
between C and k, especially in very dilute solutions. The effect of concen-
tration, however, is usually quite small.

HA
From equation i, there may be calculated the values of the ratio rj— ^

corresponding to any values of (H+) and C, and from the numbers thus

HA
obtained, the per cent, of free acid, tt a i vr a °^y ^ deduced. On the

accompanying diagram the results of such calculation are indicated.

Values of C are plotted logarithmically as abscissas ; per cents, of acid

HA
uncombined , TjxXiT~r f ^^ plotted as ordinates The several curves are

drawn to connect points corresponding to equal hydrogen-ion concentra"
lions, that is to say, to equal degrees of acidity and alkalinity, and these
concentrations are indicated on the curves.

The use of the diagram may be illustrated as follows for the case of car-

956

LAWRENCE J. HENDERSON.

5X10-3

Fig. I.

DIAGRAMMATIC REPRESENTsATION OF EQUILIBRIA. 957

bonic acid. The ionization constant of carbonic acid* is very nearly
3 X 10"^. In a decinormal solution of sodium bicarbonate the degree of
ionization is approximately 0.8. These numbers yield the equation

C = 3:2^0;^ ^^^^_^

0.0

On examination it appears that the abiscissa corresponding to
C - 3.8 X io~^iscut

A.t the ordinate.
Per cent.

the
ion

curve of hydrogen-
concentration.

0.3

2.7

10-9

io-«

20.7
72.3

10-7

963

10 5

99.6

10 <

That is to say, at a hydrogen ion concentration i X 10"^ iV, o . 3 per cent,
of all the carbonic acid must be present as free acid, and as acidity increases
this fraction increases, as defined by the above table, until when the
acidity is greater than i X 10""* iV hydrogen-ion concentration nearly all
the carbonic acid must be free.

Disregarding the incompleteness of ionization, the following table is
obtained from the diagram (C = 3.0 X 10"^).

Ordinate. Curve of hydrogen -

Percent. ion concentration. . . .

0.4 10"^

— 8
3-2 loT^

25.0 IO~^

77.0 10"^

97.0 IO~^

99.6 lO""**

Evidently the differences between this table and-the former one are unim-
portant for approximate estimations.

On the diagram, the letters above the lines of abscissas designate the
ionization constant of these acids as follows :

I^etter. Acid. k. Letter. Acid. k.

A Maldc' 1. 17 X loT"^ I Carbonic* 3 .04 X loT^

B Monochloracetic' • • i • 55 X id~^ J NaH^O/ 2.0 X lo"'^

C* Tartaric* 9.7 X 10"^ K Hydrogen sulphide* .-57

D Formic* 2 . 14 X io~"^ L Boric* i 7

B Lactic* 1 . 38 X lo""^ M Hydrocyanic* 1.3

F Aspartic* 6.9 X lo""^ N Alanine* 9.0

G Acetic* 1 .80 X io~5 o Glycocoli* 3.4

b Picolinic* 30 X 10"^ P Phenol* 1.3

* Walker and Cormack, /. Chem. Soc, 77, 20 (1900).
' Ostwald, Z. physik. Chem., 3, 418 (1889).
■ Walker and Cormack, /. Chem. Soc. *j*j^ 20 (1900).
^ Winkelblech, Z. physik. Chem., 36, 587 (1901).

• Private communication of Professor A, A. Noyes.

10-^

10-9

10-"

ID '^

IO-"

95^ LAWRENCE J. HENDERSON.

For convenience the ionization constants of certain add substances are
indicated as values of C on the diagram. In like manner, equilibria of
these acids, and of all other acid substances, except very strong acids,
whose case is obviously a special one, are defined by the diagram and may
be read off from it directly, correcting for the ionization of the salt, if
necessary.

By a somewhat different use of the diagram, equilibria in complicated
mixtures are defined, and also isohydric solutions made up of any acids
(with the exception noted) and strong bases.

In a solution made by neutralizing sodium hydroxide vrith phenol, boric
add, hydrogen sulphide, carbonic add, picolinic add, and acetic add, let
it be required to find the extent to which the base is combined with the
several adds, the hydrogen ion concentration being i X io~^.

The curve of hydrogen ion concentration i X lo"^ cuts the absdssas

Of At the ordinate of per cent

P Phenol 99 .9

I^ Boric add 98 . 5

K Hydrogen sulphide 64 . 5

I Carbonic add 25 .0

II Picolinic add 3.5

G Acetic add 0.4

Accordingly, in this solution, phenol is hardly at all combined with
sodium, and acetic acid is almost completely combined with it. The other
adds are partly free, approximately in the degrees indicated by the above
percentages. The above numbers serve also to define one series of isohy-
dric solutions made up of these acids and their sodium salts.

In like manner, all possible similar mixtures and all possible similar
isohydric solutions are defined by the diagram.

The case of bases of varying strengths in equilibrium with strong adds, is
strictly analogous. For the ionization reaction of a base

BOH = B+ + 0H-.

The following equation may be derived :

(OH-)=CX^.

This equation yields the diagram represented in Fig. 2, in form identical
with the diagram of Fig. i. The curves, however, represent hydroxyl ion
concent rations; the abscissas are as before, values of C plotted logarithmic-
ally, and ordinates indicate per cent, of base which is uncombined witfi
add. The diagram is to be used in precisely the same way as Fig. i, and
for convenience the values of k for certain bases are indicated approxi-
mately upon it as values of C.

DIAGRAMBfATIC REPRKSENTATION OF EQUILIBRIA.

959

BQJtL

2X10— »o

5X10-"
aXior-"

SXlOr-"

2XIO-"
I0r-I2

5Xio-»3

aXio-»3
i<r-»3

Fig. 2.

95o C. S. HUDSON AND F. C. BROWN.

These bases and their ionization constants are as follows :

Letter. Base. h.

A Diethylamine* i . 26 X io"~ ^

B Methylamine* 5.0 X icT"*

C* Trimethylamine* 7.4 X io~5

D Ammonia* 2.3 X 10""^

E Hydrazine* 2.7 X io~*^

F (CH,)j,SnOH* 1.7 X iq— ^

G >Tohiidine* 1.6 X xo"^

H Aniline' 4 9 X lo"""

I Alanine* 38 X lo"-"

J GlycocoU* 2.9 X 16""

K Sarcosine' 1.8 X id~"

L Aspartic add' 1.3 X io~""

M Betaine' .7.6 X lor"'^

Summary.

A diagram is presented which expresses the requirements of the concen-
tration law regarding the equilibrium in solution between strong bases and
acids of all strengths. A precisely similar diagram indicates the equi-
librium in solution between strong acids and bases of all strengths. (The
diagrams are not useful for solutions containing both strong bases and
strong acids in which the acidity or alkalinity is high.)

Tliese diagrams define with considerable accuracy the conditions of
equilibrium at all hydrogen- and hydroxyl-ion concentrations, between all
bases and all acids with the above-mentioned exceptions, and in all mix-
tures of such substances. They also define all isohydric solutions of such
substances in which this quality is dependent upon equality in concentra-
tion-of hydrogen- and hydroxyl-ions alone.

THE HEATS OF SOLUTION OF THE THREE FORMS OF

MILK-SUGAR.

By C. S. Hudson and F. C. Brown.
Received April 15, 1908.

Milk-sugar can be crystallized from solution in two forms, one of which
is a monohydrate, CijHjzOu.HjO, and the other an anhydrous modifica-
tion, C12H22O11, named ^-anhydrous milk-sugar. When either of these
crystalline milk-sugars is dissolved in water it changes partially to the
other form until a condition of dynamic equilibrium is reached in which
both forms are present in the solution. In distinction from the ^-anhy-
dride of milk-sugar there is an ,a-anhydride which is produced when
hydrated milk-sugar is heated at 125° to constant weight. This « anhy-

* Bredig, Z. physik. Chem,, 13, 289 (1894).

' Nemst, Theoretische Chemie, 3rd Edition, p. 497.

■ Winkelblech, loc. cit.

HEATS OF SOLUTION OF MILK-SUGARS. 96 1

dride is markedly hygroscopic, it dissolves in water with an evolution of
considerable heat, and its freshly prepared solutions are identical with
those of hydrated milk-sugar; these facts indicate that the a-anhydride
is unstable in the presence of water and changes immediately to hydrated
milk-sugar when dissolved. Measurements of the rate at which the
equilibrium between the hydrate and the /^-anhydride is approached
and of the proportions of the two forms that are finally present in the
solution have already been published.* The present research is a study
of the heats of solution of these three forms of milk-sugar under the
guiding hypothesis that when any one of them is dissolved in water there
occurs the incomplete or balanced reaction in solution between the hydrate
and the /^-anhydride. These heats of solution are not simple quantities,
because the dissolving of any form of the sugar is usually complicated
by the presence of a second heat effect due to the change to the stable
mixture of hydrate and /?-anhydiide. In order to distinguish the com-
ponent parts of these complex heat effects the following nomenclature is
adopted :

The initial heat of solution is the heat that is produced when any form
of the sugar is dissolved tmder such conditions that the subsequent change
to the stable mixture of hydrate and /^-anhydride is greatly retarded.

The final heat of solution is the total heat that is produced when any
form of the sugar dissolves to give a solution in which the stable mixture
is present.

The heat of passage of one form to the other is the heat that is produced
when a given quantity of the one form changes in solution to an equivalent
quantity of the other form.

To illustrate these definitions by an example, when hydrated milk-
sugar dissolves in cold water quickly no appreciable amount of the /?-anhy-
drous form is produced and the heat which is absorbed during the dis-
solving is the initial heat of solution of the hydrate. On the other hand,
if the hydrated sugar dissolves in alkaline water its partial change to the
^-anhydrous form is instantaneous and the heat absorbed in dissolving
is in this case the final heat of solution, which is obviously the sum of the
initial heat of solution of the hydrate and the heat of passage of that
portion of the hydrate which subsequently changes to the /^-anhydride.

The Method of Measuring the Heats of Solution, — The calorimeter that
was used is similar to one that has been described by A. A. Noyes.^ The
inner can is of brass, silver-plated and polished, and of a liter capacity.
It rests upon three corks in a polished tin vessel and this in turn is sup-
ported by corks within the outermost vessel which is double walled and
water-jacketed. A close covering of thick felt completely surrounds the

^ This Journal, 26, 1065- 10S2 (1904).
2 Z. physik, Chem., 43, 513-38 (1903)-

962 C S. HUDSON AND F. C. BROWN.

calorimeter. The temperatures were read from a certified Beckmann
thermometer graduated to htmdredtbs of a degree. The contents of the
inner can were stirred by a silver-plated brass propeller driven by an
electric motor. In most of the experiments on the heats of solution the
finely powdered sugar was loosely packed in a silver-plated brass tube of
50 cc. capacity which was suspended from the cover of the inner can.
An aluminum disk was cemented water-tight in the lower end of this tube
with paraffin and could be ptmched out at the proper time with a glass
rod inserted through the calorimeter coverings. This arrangement
allowed a sudden and controllable mixing of the dry sugar with the water
while the calorimeter remained closed.

fT'The Indiial Heat of Solution of Hydrated MUk-Sugar. — ^Pure hydiated
milk-sugar was prepared by recrystallizing once a very good quality of
commercial milk-sugar crystals, and it was dried to constant weight at
100°. This hydrate does not lose its water of crystallization at this
temperature even under prolonged heating. The determination of its
initial heat of solution was carried out at 20^ by the method that has just
been described and the resulting data are given in the following table.
In order to make certain that all the sugar dissolved, the stirrer was stopped
and the calorimeter opened immediately after seveml of the measure-
ments, but in no case could any tmdissolved sugar be seen.

Tablb I. — Initiai* Hbat op Soi^ution of Hydratsd Mojc-Sugar at 20®.

Mass of
water.
Grams.

Mass of
sugar.

Total water
equivalent.!

Temperature
change.

Heat of
solution.

Time to (Us-
•oWe. Minutes.*

999

31 26

1038

— 0.361

— 12, 0

984

26.13

1022

—0.304

— 12.0

999

26.10

1037

— 0.302

— 12.0

991

26.23

1029

—0.306

— 12.0

996

26.08

1034

—0.302

— 12.0

993

25.87

103 1

— 0.296

—II. 8

987

25.87

1025

— 0.302

— 12.0

998

11.23

103 1

—0. 128

—11.8

1003

29.60

1042

—0.343

— 12. 1

1000

37.07

104 1

— 0.400

— II. 8

Average — 12.0

The average of these ten determinations is — 12.0 gram calories per
gram for the initial heat of solution of hydrated milk-sugar at 20®. This
value is somewhat larger than those which others have found, Berthetot*

^ The water equivalent of the inner calorimeter and its attachments, including
the immersed portion of the thermometer was calculated to be 29.9 grams. The
specific heat of milk-sugar is 0.30, Magie, Physical Review, 16, 381 (1903).

' The sugar was assumed to be cempletely dissolved when the rate of c^hange of the
temperature reached the value that it had before the sugar was added. >

* Michanique Chimique, I, 545. '

HEATS OF SOlrUTlON OF MH,K-SUGARS. 963

gives — 10.2, Brown and Kckering^ — 11.5, Magie' — 11. 5. These differ-
ences are probably principally due to the fact that hydrated milk-sugar
has a large temperature coefficient of its heat of solution. !Magie has
found the molecular heat of hydrated milk-sugar to be 108 in the solid
state and 165 in solution, therefore the heat absorbed by the solution of
I gram must increase (i65-io8)/36o = 0.16 calotie for each degree rise
of temperature. According to this, the difference between the value here
found for the heat of solution and that given by Brown and Pickering
and by Magie can be explained by a difference of only 3 ° in the tempera-
ture of the experiments. As Brown and Pickering's measurement was made
at 16®, and probably also Magic's, while the present ones were performed
at 20°, it is evident that an excellent agreement has been obtained by the
different observers. Solubility measurements on hydrated milk-sugar
also show that the heat of solution increases with the temperature, for
if the solubilities at 15® and 25° are used to calculate the heat of solution
for this range by the well-known method of van't Hofif, — 1 1 . 5 calories
is obtained,* but if those at 0° and 15° are employed — 10.3 is found.
If the former value is considered to refer to 20° and the latter to 7.5° the
change in the heat of solution per degree is o.io calorie which agrees
with the coefficient that was fotmd above from the molecular heats.

In the data given in Table i there is no indication that the heat of
solution changes in value according as the sugar is dissolved in much
water or in Uttle* If such an effect were present the solutions would of
necessity also show considerable heat of dilution ; this fact was accordingly
made use of as a direct test. Fifty cc. of a semi-normal solution of milk-
sugar were put in the tube of the calorimeter and after constant tem-
perature was attained the bottom of this tube was punched out and the
solution allowed to mix with 800 cc. water, which was 9 grams of sugar
passing from half to three hundredths molal concentration. In the two
experiments rises of temperature of 0.001° and 0.003° were observed,
and they are so small that the absence of any considerable heat of dilu-
tion can be considered established. It follows from this that the heat of
solution of milk-sugar does not change with the concentration in dilute
solution (e. g., below 0.3 molal) and changes only very slightly in con-
centrated solutions.

As has been mentioned, hydrated milk-sugar fonns ^-anhydrous milk-
sugar slowly during many hours after its solution in cold water. In
order to make certain that the data given in Table i refer to the initial
heat of solution it must be shown that this chemical change in the solution
did not proceed far enough during the time of the measurement to give

» /. Chem, Soc.,71^ 783 (1897)-

* Pkys. Rev., 16, 381 (1903)-

• See page 970 below.

964 C. S. HUDSON AND F. C. BROWN.

any heat effect. Now in that experiment which required the longest
time, four minutes, 3 per cent, of the dissolved hydrate changed to the
/?-anhydride, according to previous measurements of the rate of this
change.* It will be shown later in this article that when the change
proceeds to equilibrium the heat effect is small, certainly not greater than
one calorie per gram of sugar transformed; therefore only three hmi-
dredths of a calorie at the most is the error that is due to the change of the
hydrated sugar after dissolving, and it can be neglected.

The Preparation of Pure P-Anhydrotis Milk-Sugar. — ^Three methods for
preparing /^-anhydrous milk-sugar have been published; in the first a
solution of any form of milk-sugar is boiled to dryness,' in the second
alcohol and ether are added to a hot solution to precipitate the anhy-
dride,' while in the third the solution is allowed to crystallize slowly at
95° from supersaturation.* The first and second of these methods give
an impure anhydride contaminated with varying amounts of hydrated
milk-sugar, but by following the directions given below which are based
upon the third method, it has been found possible to secure a good yield
of the /9-anhydride in large crystals which are colorless and free from
hydrate.

Commercial crystallized milk-sugar is dissolved in hot water to form a
saturated solution at 100^, which is then decanted into a copper beaker
and rapidly boiled until its boiling-point rises to 104-105°. The copper
beaker is then suspended in boiling water or steam for twenty-four hours.
A crust forms over the solution soon after boiling ceases which hinders
further evaporation and at the end of the twenty-four hours numerous
well-formed crystals are found hanging to the sides of the beaker and the
crust. The bottom of the beaker is covered with a compact mass of
minute crystals which are rejected, as it is impossible to free them from
the mother-liquor. It is well to roughen the surface of the copper beaker
in order that the crystals may adhere better. To free the crystals from
the solution, which is a thick syrupj an opening is cut through the crust,
the beaker removed from the heater and as much as possible of the hot
solution poured off ; the crystals are then removed from the beaker, pressed
between filter papers and immediately washed by decantation with
glycerol heated to 140°, followed by hot 95 per cent, alcohol and then by
ether.

In order to test the crystals that were prepared by this method for the
presence of traces of the hydrate, a finally saturated solution of hj^rated
milk-sugar was prepared at 20°, filtered to free it of all suspended particks,

* This Journal, 26, 1076 (1904).

' Erdman, Ber., 13, 1915-31 (1880).
»Tanret, BtUl. soc. chim. [3], 15, 354 (1896).

* Hudson, Z. physik. Chem., 44, 488 (1903).

HEATS Olf SOLUTION OF MILK-SUGARS. 965

•

and cooled to o®, at which temperature such a solution is known to be
supersaturated with respect to the hydrate but greatly undersaturated
with respect to the anhydride.* Several crystals of the anhydride that
were prepared as described above were then placed in this solution.
They dissolved completely without leaving a visible trace of insoluble
matter, which would not have been the case if the crystals contained
particles of the hydrate. This method of testing for the presence of the
hydrate is quite delicate and the result shows conclusively that the j8-anhy-
dride is pure. This test for purity can be applied to a few other sub-
stances for which such a test has heretofore been lacking, for example
to the /^-anhydrous forms of glucose, galactose and the other mutarota-
ting sugars.

The specific gravity of pure ^-anhydrous milk-sugar at 20° is 1.59,
of the hydrated form 1.54.

The Initial Heat of Solution of fi- Anhydrous Milk-Sugar, — The initial
heat of solution of this form of the sugar was measured in the same manner
that has been described in the case of the hydrate. The sugar sample
marked one was a year old, the other two were freshly prepared. The
results are given in Table 2.

TabLtB 2. — Initiai* Heat of Solution of /?-Anhydrous Milk-Sugar at 20°.

Time to
Mass of Mass of Total water Temperature dissolve. Sample Heat of
water. sugar. equivalent. change. Minutes, number, solution.

Grams.

716 13-7 749 —0039 2 I — 2.1

902 29.6 939 — 0.063 2 I — 2.0

736 15.4 768 —0.051 2 2 —2.5

464 8.1 497 —0.038 I 3 —2.3

686 13. 1 710 — 0.051 3 3 — 2.3

These experiments give — 2.3 calories per gram as the heat of solution
and since the chemical change that occurs in these solutions is the reverse
of that in the solutions of the hydrate it proceeds too slowly to affect the
measurements, and the value obtained is the initial heat of solution.
The values for the heat of solution of this anhydride that have been
found by previous observers are all larger than the above, undoubtedly
because some hydrate was present in their samples; Brown and Picker-
ing* found — 5.4, Magie and Hudson* — 3.6.

^4 Precipitated Milk-Sugar Showing No Mutarotation. — If a solution of
any form of milk-sugar is kept for a day at room temperature or boiled
and cooled, to allow the reversible chemical change that occurs in such

* Tms Journal, 26, 1071-1074 (1904). The concentration of ^-anhydride in a
saturated solution of the hydrate at 20° is only about one- quarter the initial solubility
of the anhydride at 0°.

* J. Chem. Soc, 71, 783 (1897).

■ Princeton Univ. BuU., April, 1902.

966 C. S. HUDSON AND F. C. BROWN.

solutions to reach equilibrium, and a mixture of strong alcohol and ether
is then added, a crystalline precipitate is formed which does not show
mutarotation when it is dissolved in water as do the other forms of milk-
sugar. Tanret* who discovered this precipitate in 1896 regarded it until
recently as a new form of milk-sugar, the stable modification to whidi
the other forms revert in solution, ascribed to it the formula C^HaOu. JH,0,
and named it ^-lactose. It is now generally accepted that this form is a
mechanical mixture of hydrated and /9-anhydrous milk-sugar in about the
proportions in which they are present in solution in equilibrium. The
view that this form of milk-sugar is such a mixture was expressed six
years ago' and direct evidence was given to support it ; but as the publica-
tion in which this evidence appeared had small circulation the data ate
presented again, using, however, the more accurate values of the heat of
solution that are now at hand. If the alcoholic precipitate is a mixture
of hydrated and ^-anhydrous milk-sugar Its initial heat of solution will be
intermediate between those of its constituents. The exact value can be
calculated from the fact that the two forms must be present in the mixture
in the same proportion in which they occur in the stable solution or other-
wise the precipitate would show mutarotation when dissolved. It has
been shown from solubility measurements' that this proportion at room
temperature is 1.5 parts ^-anhydride to each part hydrate; the initial
heat of solution of such a mixture is therefore calculated to be — [12.0 -f-
(2.4)(c.5)]h-2.5 = — 6.2 calories. Magie and Hudson* found by ex-
periment the value — 6.5 calories for Tanret's alcoholic precipitate.
The agreement of these values is clear evidence that this precipitate is a
mechanical mixture of hydrated and j9-anhydrous milk-sugar.

The Heat of Passage in Solution betiveen the Forms of Milk'Sugar.—
The establishment of equilibrium between the two forms of milk-sugar
in solution proceeds so slowly at room temperature that no noticeable
heat effect due to the transformation ordinarily occurs. Although the
rate of production of heat in these solutions under usual conditions is
thus too slow to admit of its measurement, the velocity of the chemical
change can be enormously accelerated by suitable catalytic agents,
especially an alkali, and the heat is then produced very quickly and can
be measured.*

In the following experiments milk-sugar hydrate was dissolved quickly
in water at 20° and immediately after its complete solution a small
quantity, usually 0.5 cc, of tetranormal sodium hydroxide was added

^ Loc. cU.

» Princeton Univ. BuU., April, 1902.
» This Journal, 26, 1074 (1904).

* This is the method that was used by Brown and Pickering (loc. cit.) in measuring
several such heat effects in the carbohydrate group.

HEATS OF SOLUTION OP MILK-SUGARS. 967

and the immediate rise in temperature measured. An equal quantity
of the alkali was then again added and the accompanying rise in tem-
perature again noted. The difference between the first rise and the
second is taken to be due to the partial change of the hydrated milk-
sugar to the /?-anhydride. This conclusion is not strictly correct, and if
theory were exactly adhered to the addition of the alkali that was made
to the already alkaline solution should have been made to a neutral
solution of milk-sugar which had reached equilibrium by long standing,
but as several experiments showed that the two methods of addition
of the alkali gave the same heating effect the more convenient one was
subsequently followed. To make certain that the first addition of alkali
caused a sufficiently rapid attainment of equilibrium, the change of
rotation of several solutions was observed in the polariscope immediately
after 0.5 cc. of alkali was added ; the rotation was not quite constant at
the end of one minute but did not change at all after two, proving that
the catalysis was sufficiently rapid. As from five to twenty minutes
elapsed in the experiments given below between the dissolving of the
hydrate and the first addition of the alkali it is necessary to correct for
the amount of hydrate that was transformed to /?-anhydride during this
interval; this quantity was calculated from the previous measurements
of the velocity of the reaction at 20° and it is recorded in column five.
The values in column four are calculated from the equilibrium constant
of the reaction, 40 per cent, of the sugar in stable solutions being hydrate
and 60 per cent, ^-anhydride at 20°.

TablS 3. — Hbat of Passage in Solution of Hydratbd Milk-Sugar to ^-Anhydride

AT 20°.
Mass of hydrate

Water.
Grams.

Total
water
equiv-
alent.

originally
dissolved.

finally
changed.

Changed

before
catalysis.

First
rise.

Second
rise.

Diff.

Heat of
passage.

1002

1040

29.6

17.8

0.5

0033

0.016

0.017

1 .02

999

1038

313

18.8

0.6

0.028

0.012

0.016

0.91

999

1037

37.1

22.3

I.I

0.029

0.008

0.021

I 03

1000

1038

22.5

13.5

0.9

0.022

O.OII

0.90

1004

1042

32.9

197

1.4

0.027

O.OIO

0.017

0.97

It appears from these measurements that the heat of passage is small,
the average value being closely one calorie of heat developed for each
gram of hydrate that changes in solution to ^-anhydride. Brown and
Pickering^ give +0.19 as the heat developed when i gram of dissolved
hydrate changes to the equilibrium mixture, but this is evidently a mis-
print and should read 0.25, which would give 0.4 calorie for the heat of
passage, a value not greatly different from the one here found. It is
exceptional that the formation of this anhydride should develop heat,

* Loc. cit.f p. 782.

966 C. S. HUDSON AND F. C. BROWN.

solutions to reach equilibrium, and a mixturc of strong alcohol and ether
is then added, a crystalline precipitate is formed which does not show
mutarotation when it is dissolved in water as do the other forms of milk-
sugar. Tanret^ who discovered this precipitate in 1896 regarded it until
recently as a new form of milk-sugar, the stable modification to which
the other forms revert in solution, ascribed to it the formula CuHj,Ou.iH,0,
and named it /?-lactose. It is now generally accepted that this form is a
mechanical mixture of hydrated and ^-anhydrous milk-sugar in about the
proportions in which they are present in solution in equilibrium. The
view that this form of milk-sugar is such a mixture was expressed six
years ago^ and direct evidence was given to support it ; but as the publica-
tion in which this evidence appeared had small circulation the data are
presented again, using, however, the more accurate values of the heat of
solution that are now at hand. If the alcoholic precipitate is a mixture
of hydrated and /^-anhydrous milk-sugar its initial heat of solution will be
intermediate between those of its constituents. The exact value can be
calculated from the fact that the two forms must be present in the mixture
in the same proportion in which they occur in the stable solution or other-
wise the precipitate would show mutarotation when dissolved. It has
been shown from solubility measurements" that this proportion at room
temperature is 1.5 parts /3-anhydride to each part hydrate; the initial
heat of solution of such a mixture is therefore calculated to be — [12.0 -h
(2.4)(f.5)] -^2.5 = — ^.2 calories. Magic and Hudson* found by ex-
periment the value — 6.5 calories for Tanret's alcohoUc precipitate.
The agreement of these values is clear evidence that this precipitate is a
mechanical mixture of hydrated and /^-anhydrous milk-sugar.

The Heat of Passage in Solution hehveen the Forms of MUk-Sugar. —
The establishment of equilibrium between the two forms of milk-sugar
in solution proceeds so slowly at room temperature that no noticeable
heat effect due to the transformation ordinarily occurs. Although the
rate of production of heat in these solutions under usual conditions is
thus too slow to admit of its measurement, the velocity of the chemical
change can be enormously accelerated by suitable catalytic agents,
especially an alkali, and the heat is then produced very quickly and can
be measured.*

* Loc. cit.

» Princeton Univ. Bidl., April, 1902.
"This Journal, 26, 1074 (1904)-

* This is the method that was used by Brown and Pickering (loc. cU.) in measurioi
several such heat effects in the carbohydrate group.

HEATS OF SOLUTION OF MILK-SUGARS. 967

and the immediate rise in temperature measured. An equal quantity
of the alkali was then again iadded and the accompanying rise in tem-
perature again noted. The difference between the first rise and the
second is taken to be due to the partial change of the hydrated milk-
sugar to the )9-anhydride. This conclusion is not strictly correct, and if
theory were exactly adhered to the addition of the alkali that was made
to the already alkaline solution should have been made to a neutral
solution of milk-sugar which had reached equilibrium by long standing,
but as several experiments showed that the two methods of addition
of the alkali gave the same heating effect the more convenient one was
subsequently followed. To make certain that the first addition of alkali
caused a sufficiently rapid attainment of equilibrium, the change of
rotation of several solutions was observed in the polariscope immediately
after 0.5 cc. of alkali was added ; the rotation was not quite constant at
the end of one minute but did not change at all after two, proving that
the catalysis was sufficiently rapid. As from five to twenty minutes
elapsed in the experiments given below between the dissolving of the
hydmte and the first addition of the alkali it is necessary to correct for
the amount of hydrate that was transformed to j9-anhydride during this
interval; this quantity was calculated from the previous measurements
of the velocity of the reaction at 20° and it is recorded in column five.
The values in column four are calculated from the equilibrium constant
of the reaction, 40 per cent, of the sugar in stable solutions being hydrate
and 60 per cent, /^-anhydride at 20°.

TablS 3. — Hbat of Passagb in Solution of Hydratbd Milk-Sugar to /?- Anhydride

AT 20°.

Total

Mass of hydrate

Water.
Grams.

water
equiv-
alent.

originally
dissolvea.

finally
changed.

Changed

before
catalysis.

First
rise.

Second
rise.

Diff.

Heat of
passage.

1002

1040

29.6

17.8

0.5

0.033

0.016

0.017

1 .02

999

1038

313

18.8

0.6

0.028

0.012

0.016

0.91

999

1037

37.1

22.3

I.I

0.029

0.008

0.021

1.03

1000

1038

22.5

135

0.9

0.022

O.OII

O.OIl

0.90

1004

1042

32.9

19-7

1-4

0.027

O.OIO

0.017

0.97

It appears from these measurements that the heat of passage is small,
the average value being closely one calorie of heat developed for each
gram of hydrate that changes in solution to /?-anhydride. Brown and
Pickering* give +0.19 as the heat developed when i gram of dissolved
hydrate changes to the equilibrium mixture, but this is evidently a mis-
print and should read 0.25, which would give 0.4 calorie for the heat of
passage, a value not greatly different from the one here found. It is
exceptional that the formation of this anhydride should develop heat,

* Loc. cit., p. 782.

970 C. S. HUDSON AND F. C. BROWN.

The Final Heat of Solution of Hydrated Milk-Sugar Calculated from lis
Solubility, — By the principles of the osmotic theory of solutions it is
possible to calculate the heat of solution of milk-sugar from the change
of its solubility with the temperatiu-e. The final solubilities of the hydrate
are known at 15° and 25° and the final heat of solution is also known at
20°, so that a comparison of the concltisions of the two independent
methods of measurement is possible. The values^ for the solubility aie
49.7 millimols per 100 grams water at 15° and 63.4 at 25°, and the molec-
ular heat of solution (Q) is, from the well-known formula of van't Hoff,
log. nat. 63.4/49.7 = Q/(i/288-i/298), equal to — 4130 calories. Since
the molecular weight of hydrated milk-sugar is 360, the final heat of solution
per gram is — 11.5 calories. This calculated value agrees perfectly
with that given by the calorimetric measurement, — 11.4.

Summary.

1. Each of the three forms of milk-sugar shows two heats of solution
according as the rate of the balanced reaction that occurs in milk-sugar
solutions proceeds very slowly or very rapidly. In the former case the
initial heat of solution is observed, in the latter the final heat of solution,
which is the sum of the initial heat of solution and the heat genetated
by the rapidly progressing reaction, called the heat of passage.

2. The following data on these heat eflFects at 20° have been deter-
mined with a calorimeter.

Hydrated milk-sugar. a>Anhydride. 0-Anhydridc.

Initial heat of solution — 1 2 . o cal. gram +7-3* — 2.3

Final heat of solution — 1 1.4 +7-9 — 2 • 7

Heat of passage to /^-anhydride +1.0 +1.0

3. The equality of the heat of passage of the hydrate and the a-anhy-
dride is further evidence supporting the accepted view that the a-anhy-
dride passes instantly to hydrate when dissolved, which then slowly
builds ^-anhydride according to the balanced reaction

^12-"'24^12 -< -"-2^ ' ^12 "22^11

(Hydrate. ) 0?- Anhydride.)

4. Evidence from two independent sources shows that the initial
heat of solution of the hydrate increases about o. i calorie per degree rise
in temperature.

5. The dilution of strong milk-sugar solutions (0.5 molal) causes a very
slight development of heat, which cannot be due to any change in the
equilibrium of the balanced reaction with concentration because the heat
development at room temperature on dilution is instantaneous while any
change of the balanced reaction at this temperature would be very slow.

* This Journai., 26, 1072 (1904).

' From the work of Jorissen and Van der Stadt, loc. cit.

NEW INSTRUMENT FOR REDUCING GAS V01,UMES. 97 1

The heat of solution is independent of the concentration when this is less
than about 0.3 molal.

6. Pure ^-anhydrous milk-sugar has been prepared by an improved
method of slow crystallization, and a delicate test shows that it is free
from hydrate. Its specific gravity at 20° is 1.59, that of the hydrate being

I-54-

7. The present data on the heats of solution confirm quantitatively

the view that the crystalline substance which alcohol and ether pre-
cipitate from cold stable milk-sugar solutions is not a pure substance as
was first supposed but is a mechanical mixture of the hydrate and the
^-anhydride. It does not show mutarotation when redissolved in water
because the two substances are present in it in the same proportions
approximately in which they occur in stable solutions.

8. When the temperature of a stable solution of milk-sugar is suddenly
changed a slight thermal lag is observed in its rotatory power, which
indicates that the hydration is slightly increased with rise of temperature
between o® and 100°. The direction and magnitude of this lag agree
with the conclusions drawn from the observed value of the heat of passage.

9. The final heat of solution of hydrated milk-sugar is calculated from
the solubilities at 15° and 25® to be — 11.5, agreeing with the calorimetric
measurement at 20°, which gives — 11.4.

UNXVBB8XTT OP ILLINOIS, ^

Urbana, III.

A NEW mSTRXTMENT FOR KEDUCIIf 6 GAS VOLUMES TO STAin>ARD

COITDITIONS.

Bt Grant T. Davis.
Received April i, 1908.

The necessity for making a considerable number of accurate measure-
ments of gas volumes has led to the working out of the following device for
their reduction to standard volume. Since all that is necessary for the
physical reduction of a gas to standard volume, is to subject it to a definite
pressure, a water column of variable length was first tried, but this was
discarded in favor of a column of known length with a fixed scale. The
length of the column of mercury (Lm) necessary to compress a gas to stand-
ard volume at temperature "^" is foimd by the formula, V — ^ / . \^,

^ 760(1 + at)

the conditions being such that i;— V; then /) = 76o (i -\-at) -f /)'and Lm"
p — 760. The length for the water column was taken as Lm X 13 . 59, and
correction was made for the expansion of water with rise in temperature.
The apparatus for use with gases which can be collected over water con-
sists of a piece of iron pipe about two meters long, fitted with a T near the
top and an elbow at the lower end. The elbow is closed by a rubber

972 GRANT T. DAVIS.

Stopper carrying a glass T, one arm of which is fitted with a stopcock, and

the other connected to the water supply. The T at the upper end of the

pipe carries a small reservoir at the top to prevent overflow in case of loo

hasty filling, and the ='de-opening is closed by a rubber stopper carr>-inf

a siphon with a very short inner leg, and

the long leg empt>'ing into the sink, oras

shown in the figure. To the lower part o!

the pipe is attached a scale, so graduated in

deerees that the column of water fmm the

n of the short leg of the siphon to

aduation is the calculated length (or

mperature represented by the gradua-

Attached to this "temperature" scale,

liding "pressure" sale whose divisions

. 6 mm. long.

ivoid the error which would be intro-
by using tap water of a temperatute
nt from that of the room, the system
ttles "A" "B" is introduced. "A"
ns distilled water at the room tem-
ire. Tap water being admitted to
the water from "A" is forced into
>paratus, and no temperature cban^

ase the apparatus the burette ccn-
H the gas is cormected to the T at the

i a of the pipe and the stopcock opened.

If "^are must be taken that there is no air in

^*"'' the connecting tube. The graduation 760

on the pressure scale is placed opposite that point on the temperature scale
which corresponds to the observed temperature, water is admitted to the
pipe until the siphon begins to act, and while a slow stream continues to
flow, the meniscus in the burette is brought level with that graduatkm of
the pressure scale which corresponds to the observed barometric pressure,
and the corrected volume of the gas read off. The podtion represented
in the figure is that for a gas observed at a temperature of 21 .4°, and
barometric pressure of 754 mm.

Table "A" gives the length of water column necessary to compress a
gas to the standard volume for temperatures from 15° to 34°. Table "B"
offers a comparison between the standard volume as calculated, and as
observed by means of this instrument.

A LECTURE TABLE DOWN-DRAFT. 973

Tablb "

A."

•9

126

•9

132

138

■7

89.

144

150

99.

157

104.

163

no.

•

169.

1x5.

176.

121.

183.

Tablb " B."

(Cal.)

V(OU.)

750.5

42.0

37.7

750.5

41.6

• 13

37.1

742.1

41.9

.11

36.1

742.1

31.6

.16

27.1

741 2

26.

30.6

26.3

733.2

21.

74.0

64.58

645

736.4

22.

6.2

5.4

748 4

19.

50.2

45.

45-2

760.5

18.

49.6

45.

45 4

760.0

21.

417

37-

38.8

Uhivbrsitt op Illinois,

Urban A, Ii

LL.

LECTURE

U-DRAFT.

TABLE

DOW

Bt Wm. L. Dudlbt.

Received March a6, 1908.

A powerful and reliable down-draft on a lecture table is of great service
and is much superior to a hood behind the lecturer since by its use ex-
periments with the most disagreeable and poisonous gases can be made
safely in full view of the audience.

My experience in the past has been that such down-drafts are rare,
in fact, I have never had one that was usable until I installed in the
lecture theatre of Furman Hall, the new chemical laboratory, the arrange-
ment herein described.

The requirements of a good down-draft are (i) certainty of action
regardless of the weather or temperature, (2) sufficient suction to permit
of moderate freedom of action on the part of the experimenter which is
not possible if the air current is so slow as to require a closed chamber to
prevent the escape of gases into the room, (3) the draft tube with its cover
being so arranged that it can be located in or near the center of the lecture
table flush with the table top so as to offer no obstruction, and (4) a
cover to the draft tube which will not stick nor become fast from corrosion.

974 ^*™' L. DUDLEY.

All of these requireineiits are amply met by our down-draft airaoge-
ment which will be readily understood by reference to the accompanying
sketch and description.

-E!9-«-

A (Fig. i) is the top of the lecture table. B is a two-inch cast iron soQ
pipe which drops vertically through the lecture table and the floor under
which it passes at right angles to a flue. The end of the horizontal pipe
is cbsed at D by a plug, through the center of which passes a brass pipe,
E, one-eighth of an inch inside diameter, extending about 1 2 inches beyond
the vertical pipe. The pipe E is connected to a compressed air pipe.

The hub of the vertical pipe B is fitted with an ordinary two-inch
brass "wash-tray plug," without a stopper, set flush with the table top
by letting it into the wood. Connected with the crossbars in the wash-
tray plug is a chain, the other end of which is attached to the bottom of a
thin flat brass plate, F, 2 ■/„ inches in diameter, which serves as a cover.
This cover fits flush with the wash-tray plug which has had an offset
turned in the face, of sufficient diameter to receive the cover-plate. A
finger lift is also cut in the plug, thus the cover plate can be easily slipped
aside when the down-draft is in use and the chain prevents its being
detached. The valve controlling the supply of compressed air is located
under the table at a place most convenient to the experimenter.

In using the down-draft, I perform the experiments inside of a glass
cylinder six inches in diameter and open at both ends, as shown in Fig.
2. At one end, the cylinder is 8i inches in diameter and at the other, 10
inches in diameter. The vessel containing the gas or volatile substance
is placed on an adjustable support over the down-draft pipe, and the
glass cylinder is put around it with its larger end resting on the table.

PRECIPITATED ANTIMONY SUI,PHIDE. 975

The top of the vessel, for convenience, should be about on a level with
the top of the cylinder. When the vessel is opened all of the fumes which
come out are sucked down into the down-draft pipe.

With an air pressure of 25 pounds, the suction in the down-draft tube
is 2J inches of water pressure, and the velocity of the air going through
the six-inch cylinder was found by careful anemometer measurement to
be 135 feet per minute, which gives a velocity of about 1200 feet per
minute in the two-inch down-draft pipe.

PuRMAN Hall. Vandbrbilt Univbrsity,
Nashville, Tbxnbssbb.

PURITY AND VOLATILITY OF PRECIPITATED ANTmOlfY

SULPHIDE.

* Bt i^bwis a. Yoxrrz.

Received April 10, 1908.

This work was taken up to discover whether pure antimony trisulphide
could be obtained by precipitation by hydrogen sulphide. The volatility
of the product of precipitation when heated in an inert atmosphere came in
more or less incidentally in the course of the work as a check on the purity
of the sulphide. Practically the only impurity considered however was
the chloride.

Most of the work on the determination of antimony by precipitation as
sulphide, as is too often the case in analytical work in general, is empirical
The investigator will fall upon a method that approximates accurate results;
then by shifting the conditions here or there he finally arrives at a set of con-
ditions that will give results within the limits of experimental error. He
then announces his method as an accurate one though often it is accurate
merely because of a series of compensating errors.

It might be said in anticipation that such was found to be the case in the
method of determining antimony by precipitation as sulphide in the pres-
ence of hydrochloric add, heating the dried sulphide in an atmosphere of
carbon dioxide, and weighing.

Purity.

(i) As a preliminary experiment in the investigation of the purity of the
precipitated sulphide, two samples of antimony trichloride of approximately
1 . 5 grams each were dissolved in 250 cc. of water and 30 cc. of concen-
trated hydrochloric acid, the antimony precipitated as completely as possi-
ble by hydrogen sulphide gas, the volume of the liquid then increased to
500 or 600 cc, warmed nearly to boiling, and saturated with the gas. The
sulphides were washed with water saturated with hydrogen sulphide till
ID cc. of wash water gave no test for chloride, or but a faint trace. It
should be noted that it is very difiScult to wash the precipitated sulphide
free from chloride, from 1000 to 1500 cc. of water being required in this and

976 LEWIS A. YOUTZ.

in subsequent cases where precipitation was made in the presence of hydro-
chloric acid. These precipitates were tested qualitatively, and notable
quantities of chloride were fotind in each.

The method of testing for chloride in this case as in all subsequent cases
was to fuse the sulphide in sodium carbonate in a porcelain crucible, acidify
the fusion with sulphuric add after taking up with water, filter out the
sulphide produced by the acidification and test the filtrate for chloride after
the removal of any hydrogen sulphide in the solution. Blanks were run in
every case, as soditun carbonate absolutely free from chbrides was found
extremely difiScult to obtain or to make. Testing for chlorides by making
a solution of the trisulphide by means of concentrated sulphuric or nitric
adds, was not thought wise, owing to the liability of volatilization of the
chloride by the acid, and also owing to the interference in precipitation of
chloride by silver nitrate in the presence of antimony. A crudble was
tested for chloride by fusion of sodium carbonate in it. No chloride was
found. The glaze did not contain chloride.

(2) Thinking that taking a smaller amount of antimony chloride to start
with, so as to give less of the sulphide to wash, might effect the amount of
chloride in the end, 0.2215 gram of metallic antimony was dissolved in
aqua regia, the solution boiled with excess of hydrochloric add till the
nitric add was decomposed, ending with about 50 cc. of 20 per cent, hydro-
chloric add. This was then diluted with about 400 cc of water and the an-
timonic acid treated with hydrogen sulphide till complete predpitation of
antimony pentasulphide. The predpitate was then washed till no further
test for chloride was fotmd in the filtrate. A sample of the predpitate
showed abundant chloride present.

(3) o . 2400 gram of metallic antimony treated as the preceding, but the
predpitate dried to constant weight at 110°, was heated for 0.5 hour
to 240-250° in an atmosphere of carbon dioxide. This predpitate,
assuming it to be antimony trisulphide calculated over to antimony, indi-
cated o . 2397 gram antimony. It will be noted that the sample of sulphide
was heated to 250° in an atmosphere of carbon dioxide, and the results
showed only o. 12 per cent, error, yet on testing a strong test for chloride
was indicated. The antimony was oxidized to the pentad form and pre-
cipitated as pentasulphide, with the thought that the sulphur freed from
this sulphide, when transformed to the black modification and triad anti-
mony sulphide, might act to transform any chloride of antimony to the sul-
phide at the higher temperature.

(4) A sample of 0.3251 gram of antimony sulphide predpitated from a
hydrochloric add solution as usual and dried at no® to constant weight
gave 0.3 per cent, chlorine equivalent to 1.45 per cent. SbOCl, supposing
the chloride to be in this form under the conditions.

(5) To try the effect of tartaric acid on reducing the amotmt of adherent

PRECIPITATED ANTIMONY SULPHIDE. 977

chloride, a sample of antimony sulphide was prepared from metallic anti-
mony as in the previous cases but 5 grams of tartaric acid were added before
precipitation. The precipitate was washed as usual and dried at no®.
0.2954 gram of this sulphide gave, on testing, 0.51 per cent, chlorine equiv-
alent to 2.47 per cent. SbOCl. Another portion of this heated o . 5 hour
at 240° gave o . 242 per cent, chlorine, which, if considered as Sb^OgClj as is
the case when pure antimony oxychloride is heated to 250®, is equivalent
to 1 . 80 per cent. Sb405Cl2. 247 grams of SbOCl by loss of SbCl, to become
Sb^OgClj by calculation give 184 grams of Sb405Cl2, which points strongly
to the impurity being Sb^OjClj after being heated to 250®. Plainly here
the tartaric acid did not reduce the amoimt of chloride as impurity.

(6) Another sample treated as in the preceding case, but with the tar-
taric acid omitted, gave chlorine o . 66 per cent, on the sulphide dried at
1 10°, but 0.45 per cent, after being heated to 240® for 0.5 hour.

(7) A sample from o . 2680 gram of antimony dissolved as usual in aqua
regia, precipitated by hydrogen sulphide, washed as usual, precipitate dis-
solved in ammonium sulphide, reprecipitated by acetic acid, dried at iio*^,
heated in intervals of o . 5 to 1.5 hours in carbon dioxide for a total of nine
hours, showed o . 10 per cent, chlorine yet present which is equivalent to
o . 9 per cent. Sb^OjClj.

These samples of results represent only a few of a much larger number of
experiments carried out with varying amounts of antimony and hydro-
chloric acid present. In no case was it found possible to produce a pre-
cipitate of antimony sulphide, either the trisulphide or the pentasulphide,
even approximately free from chloride, nor was it possible to remove the
chloride by washing, though a point could be reached after long washing
ivhere the wash water contained but the merest trace of chloride. Yet
these samples contained as high as 2 .47 per cent, of chloride calculated as
SbOCl, and in one case after heating for o . 5 hour in a carbon dioxide stream
to 250®, showed 4 .05 per cent, calculated as Sb405Cl2.

The tartaric acid was without apparent effect in reducing the amount of
chloride retained by the precipitate. Further, it was surprising that after
dissolving the precipitated sulphide in ammonium sulphide free from chlor-
ide, and reprecipitating by acetic acid also free from chloride, the sulphide
yet contained chloride, and after heating for 9 hours to 250 degrees there
was still present o . 10 per cent, chlorine.

Volatility of the Precipitated Sulphide.

A large number of tests of the volatility of these sulphide precipitates
were carried out under var3dng conditions of precipitation. Uniformly the
precipitates were heated in an atmosphere of carbon dioxide at a temper-
ature of approximately 250® after drying to constant weight at usually
105-110*^ in air.

The precipitated sulphides were collected and washed in a Gooch cruci-

978

LEWIS A. YOUTZ.

ble with asbestos filter. After drying the precipitates in the crucible, they
were in each case placed in a small beaker covered with a watch glass per-
forated to admit a thermometer and carbon dioxide delivery tube. The
beaker and crucible so arranged were then heated in an asbestos oven
after the removal of the air from the beaker by carbon dioxide.
A few only of the tests will be given as illustrations:
(i) 1.5 grams of antimony chloride were dissolved in water and enough
hydrochloric acid added to give a clear solution, the antimony precipitated
by hydrogen sulphide, filtered, washed till the wash water was free from
chloride and dried in air at iio^. This sample was heated at varying
intervals of from i to 5 hours in an atmosphere of carbon dioxide at a
temperattue of 250% altogether for 34 hoiu^. The loss on approximately
I gram of the precipitate averaged i.i mg. per hour though the loss per
hour was somewhat irregular. The last hour it lost 2.5 mg.

(2) A sample of 0.1931 gram metallic antimony was dissolved in aqua
regia, excess of nitric add decomposed by boiling with hydrochloric add,
diluted and predpitated as usual. After heating for i hour at 230° the
sulphide calculated over to antimony indicated 0.1936 gram antimony;
results were thus quantitative so far. After heating for 1.5 hours more
the weight was reduced by 5.8 mg., or something over 2 per cent. loss.

(3) To test the volatility of the sulphide free from chloride a gram of
tartar emetic purified by oystallization was dissolved in water and pre-
dpitated after addification with a small amount of sulphuric add and a
larger amount of acetic add. The precipitate after drying at no**
weighed 0.4942 gram. After heating for 3 hours at 250° it weighed
0.4938. Loss 0.4 mg. Thus the sulphide free from chloride is practically
non-volatile.

(4) Another sample of 0.18 10 gram of antimony was dissolved and
predpitated as usual. The sulphide was then dissolved by ammonium
polysulphide and reprecipitated with dilute sulphuric add. This was
repeated three times. Finally the precipitate was washed, thoroughly
dried and heated to 250° as usual. After the first large loss due to the
sulphur from the pentasulphide, 2.5 hours' heating in three intervals
caused a loss of 1.7, 2.6, 2.7 mg. successively.

As was shown by a quantitative test for chloride in another sample
similarly treated, the remarkable thing here is the persistency with
which the chloride is retained by the predpitated antimony chloride,
for it is the presence of chloride that causes the volatility.

These illustrations are ample to show the impossibility of getting con-
stant weight at 250® with chloride present.

Is the method accurate? It is almost useless to discuss this question
for the method of predpitating antimony as sulphide, either trisulphide
or pentasulphide, transforming to the black modification, and volatiliza-

SEPARATION OF THE RARB EARTHS. 979

tion of the excess of sulphur in an atmosphere of carbon dioxide at 250®
has long been used with good results for accuracy in the hands of many
analytical chemists.

But merely to bring the fact freshly to our minds again, I give here a
dozen or more determinations made by starting.with metallic antimony,
with the temperature, time of heating in carbon dioxide atmosphere,
and the acid used.

Wt.ofSb.

Temp.

Time, hn.

8b found.

O.1931

HCl acid

227®

0.5

0.1936

0 . 2040

II (1

245^

0.5

0.2037

0.2287

i< It

240®

0.5

0.2283

0.3685

II II

250®

0.5

0.3692

0.1414

II it

245 **

0.5

O.1415

0.2400

II II

240^

0.5

0.2397

0.2586

HCl and Tar. acid

240®

0.5

0.2584

0.2405

II (1 II i 1

240®

0.5

0.2409

0.2035

II II II II

250 *»

0.5

0.2031

0.2060

II II II II

245^

0.5

0.2066

0.1592

II II II II

250**

0.5

0.1596

0.2214

II II II II

240®

0.5

0.2221

This is sufficient to show that quantitative results may be obtained.
In no case is the error over 0.3 per cent, and usually from o.i to 0.2 per
cent, even when calculated to metallic antimony, which is clearly close
enough for any ordinary analytical work. To be sure, if the samples were
to be heated much longer than 0.5 hour the error would rapidly in-
crease and soon would become large owing to the volatility of the chloride.

It is not surprising that the results should be quantitative when the
sulphide is heated only for a short time just long enough to volatilize the
stilphur either free or combined as pentasulphide and to transform the
sulphide to the black modification, for the equivalent molecular weights
for SbjSa — SbOCl and Sb^OjClj are 168.2, 171.45 and 157.9 respectively and
thus not very widely apart, so that as far as the weight of the precipitate
is concerned even several per cent, of SbOCl or even Sb405Cl2 as an im-
purity would give a weight practically the same as though the antimony
were all in the form of sulphide.

I am greatly indebted to Mr. Alva G. Austin for carrying out much
of the detail of this work.

Chemical I«ABORATORXBS« I^awrbncb UxtzvBssiTY,

Applbton, Wis.

A SCHEME FOR THE SEPARATIOlf OF THE RARE EARTHS.

C. JA1CB8.

Received April a, 1908.

In this communication, a comparatively simple scheme for the separa-
tion of mixtures of the rare earths is offered in the hope that it may prove of

980 C. JAMES.

value to any one who is desirous of entering this very interesting field of re-
search. The conclusions presented are the results of several years' in-
vestigation upon the various methods proposed for fractionation and
of personal trial of the applicability of many other compounds which
had not previously been employed for the separation of these elements.
Only those methods which proved valuable will be mentioned.

It might be as well to state in the beginning for the benefit of those
who have had little cause to study this subject, that there is no quanti-
tative method for the separation of any of the rare earths. The true
members of this family comprise those elements that are included in the
cerium and yttrium groups and are characterized by their trivalency
and by the fact that they form oxalates that are insoluble in dilute adds
and in cold ammonium oxalate solution.

The nearest approach to a quantitative separation is found in tlK
case of cerium. This is due to the fact that the properties of the metals
of all these earths and their salts, with the possible exception of cerium
itself, vary among themselves by very minute differences.

Bearing these observations in mind, the following scheme is presented :
The. mineral is decomposed, either by hydrochloric acid, sulphuric add,
potassium bisulphate, sodium hydroxide or hydrofluoric acid. When
hydrochloric acid is used, the whole is evaporated to dryness to render
the silica insoluble. It is then warmed with a little concentrated hydro-
chloric acid, after which the mass is treated with water and filtered.
The filtrate may then be treated with either oxalic acid or ammonium
oxalate. If the liquid contains considerable mineral acid, ammonium
oxalate is to be preferred. When sulphuric acid or potassium bi-
sulphate has been used to break up the mineral, it is necessary to
stir with cold water to obtain the solution of the desired elements.
Fusion with sodium hydroxide and washing with water gives a residue
of oxides. These are dissolved in hydrochloric acid. Hydrofluoric
acid decomposes many minerals, such as columbates, tantalates, etc,
in the cold, giving a residue of rare earth fluorides while silicon, colum-
bium, tantalum, etc., go into the solution. The insoluble fluorides are
decomposed by means of sulphuric acid.

In the sulphate or chloride solutions obtained by one of the above
methods, the earths are then precipitated by anunonium oxalate or oxalic
acid as mentidned above.

Having obtained the earths in the form of oxalates, they are treated
as follows :

Zirconium and Thorium, — Should the rare earth oxalates contain
these elements they may be separated by boiling with a solution of
ammonium oxalate when the whole of the zirconium and nearly all
of the thorium pass into solution. The residue is filtered off and washed

SEPARATION OF THE RARE EARTHS. 98 1

with ammonium oxalate solution. On the addition of an excess of hy-
drochloric acid to the filtrate, thorium oxalate alone is precipitated,
while the whole of the zirconium is held in solution by the oxalic acid
produced by the action of the hydrochloric acid on ammonium oxalate.

As zirconium oxalate is soluble in an excess of oxalic acid, this reagent
alone may be used in the absence of thorium. In this connection it
should be remembered that the oxalates of cerium and of the earths
of the yttrium group, are somewhat soluble in hot concentrated ammo-
nium oxalate solution so that varying amounts go into the solution. Tho-
rium, however, can be separated easily from the yttrium earths which
remain with it, by means of the double sulphate of thorium and potassium,
which is insoluble in a solution of potassium sulphate, while the corre-
sponding compounds of the )rttrium group are soluble. The crude thorium
oxalate is converted into sulphate and the cold solution is stirred with
solid potassium sulphate. Sulphate solutions of the earths are the
best to work with for this purpose, for if other compounds are used,
care must be taken to keep the solution from being too concentrated
since members of the yttrium group may be precipitated also.

After the removal of the yttrium earths, cerium earths are still present
and thorium may be separated from these by means of the solubility
of its oxalate in ammonium oxalate or, according to Glaser, by the solu-
bility of thorium oxalate in ammonium acetate;^ also by the method
of Wyrouboff and Vemeuil in which thorium is precipitated by hydrogen
peroxide.' These methods are given under the separation of cerium
and thorium, because under usual conditions they separate together
from the rest of the rare earths.

In the next step, there are three alternatives which depend upon the
composition of the oxalates as approximately determined from the solu-
bility of the double sodium or potassium sulphates in potassium or sodium
sulphate solution, viz,, (a) if containing 20 per cent, or more of the
yttrium earths and only a trace of thorium; (b) if containing 20 per
cent, or more of the yttrium earths together with thorium; (c) if con-
taining less than 20 per cent, of the yttrium earths.

(a) When the material consists of 20 per cent, or more of the yttrium
earths and practically no thorium, the oxalates are converted into sul-
phates by mixing with strong sulphuric acid and carefully igniting until
fumes of sulphuric acid are no longer evolved. The residue is then pow-
dered and dissolved in ice-cold water. The resulting sulphate solution
is stirred with solid sodium sulphate which throws down the double
sulphates of sodium and the cerium earths.'

• This Journal, x8, 782.

• Bull soc. chim. [3], 19, 219.

' Another method, which is simpler if the oxides will dissolve in acid, is to ignite

982 C. JAMES.

This precipitate contains some of the yttrium group of earths, whik
small amotmts of samarium, gadolinium and europium remain in solution.
The precipitate is separated by filtration and washed with a solution
of sodium sulphate. The insoluble double sulphates consist chiefly of
cerium, lanthanum, praseodymium, neodymium, samarium, europium
and gadolinium together with small amounts of the yttrium earths in
which the terbium, dysprosium and holmium contents are considerably
increased. These double sodium sulphates constitute Fraction A of
the table. The filtrate, on the addition of an excess of oxalic rcid, throws
down the oxalates of terbium, dysprosium, holmium, 3rttrium, erbium,
thulium, ytterbium and scandium together with some samarium, europium,
and gadolinium. This precipitate forms Fraction B of the table.

(6) The composition of the oxalates in this case is very similar to
(a), the only difference being the thorium content.

The sulphate or chloride solution is treated with sodium sulphate
and the insoluble double sodium cerium group sulphates, forming Fraction
i4, filtered off. Since the sodium thorium sulphate is somewhat soluble
in sodium sulphate solution, the filtrate is saturated with potassium
sulphate, when the remaining thorium is precipitated as thorium potas-
sium sulphate, insoluble in potassium sulphate solution. After separa-
ting the precipitate, the filtrate is treated with an excess of oxalic add,
the insoluble oxalates being filtered off and washed. This material is
added to Fraction B.

(c) In this, the third and last alternative, the oxalates consist almost
entirely of the cerium metals, and it is best to start the work of separation
from the point A.

Cerium and Thorium. — ^The next operation consists of separating
cerium together with thorium, if the latter is present, from the other
elements forming Fraction A. This is best carried out by treating the
nitrate solution with an excess of zinc oxide and potassium permanga-
nate. If the material is in the form of the insoluble double sodium sul-
phates, it should be boiled with sodium hydroxide in excess. The result-
ing hydroxides are filtered off, well washed with hot water and dissolved
in nitric acid. In dealing with oxalates that contain large amounts of
lanthanum, praseodymium and neodymium, it is necessary only to ignite
when the oxides, so obtained, will readily dissolve in nitric add. As a

the oxalates and dissolve the oxides so obtained in hydrochloric acid or nitric add,
dilute the solution and treat with solid sodium sulphate until the double sulphates are
precipitated. However, when oxalates rich in oerium are ignited, the oxides wfaicfa
are formed dissolve with great difficulty in hydrochloric or nitric add and so in this
case it is better to treat the oxalates with sulphuric add as mentioned above. Extra
care must be taken when the chlorides or nitrates in solution are stirred with solid
sodium sulphate, for if the solution is too concentrated, metals from the yttrium gronp
will be predpitated also.

SEPARATION OF THE RARE EARTHS,

983

rule, when cerium is present in large amounts, since the oxide dissolves
with great diflftculty, the oxalates are converted into sulphates. The
sulphate solution is then poured into fairly strong and boiling sodium

984 C. JAMES.

hydroxide. The rare earth hydroxides formed under these conditions
filter rapidly, and after washing with boiling water, are dissolved in
nitric acid. The nitrate solution obtained by any of the above methods
is neutralized, stirred rapidly by a motor and an excess of zinc oxide
added. On the addition of potassium permanganate, cerium peroxide
is precipitated and the addition is continued until the liquid after con-
tinued stirring remains red. This method leaves a little cerium in solu-
tion, which is separated later. The precipitate, consisting of cerium
and manganese peroxides together with thorium and a small amount
of lanthanum, praseodymium and neodymium, makes Fraction C on
the diagram. The filtrate is saturated with sodium sulphate, causing
a precipitate of the double sulphates of sodium with lanthanum, praseo-
dymium, neodymium, etc. This, after filtering and washing with sodium
sulphate solution, constitutes Fraction D. To this last filtrate, contain-
ing small amounts of the yttrium group, an excess of oxaUc acid is
added. The oxalates so obtained are united with those forming B,

Thorium. — Although thorium is not now considered as a rare earth,
methods for its purification do not seem to be altogether out of place.
The cerium peroxide contains large amounts of manganese, which is
first removed by dissolving in strong hydrochloric acid and precipitating
the earths by means of solid sodium sulphate until no more insoluble
double sulphates separate. The liquid is then filtered and the precipi-
tate washed with sodium sulphate solution. A portion of the thorium
remains in the filtrate, but this can be separated either by means of
oxalic acid or else by stirring with solid potassium sulphate.

The cerium sodium sulphate is boiled with an excess of sodium hy-
droxide, the residue filtered off, washed with boiling water and dissolved
in nitric acid. The nitrate solution is then neutralized by means of
ammonia, after which peroxide of hydrogen is added* and the whole
boiled for a few minutes. Some of the filtered solution should then
be tested by treating vdth an equal volume of hydrogen peroxide and
boiling and the process repeated until no precipitate is thus obtained.
The filtrate is reserved for the preparation of pure cerium. The thorium
precipitate, having the composition Th407N205, is very imptire and may
have a yellow or even an orange color after standing for a short time.'

The crude thorium obtained above, may be purified by treating the
nitrate solution with an excess of warm ammonium oxalate. The solu-
ble portion can then be converted into the oxalate and treated again
with warm ammonium oxalate.

' WyrouboflP and Vemeuil, Bull. soc. chim. [3], 19-20, No. 6 and Chem. Neu:s,
77, 245.

' It is highly important that the peroxide of hydrogen be free from phosphoric
add; otherwise an insoluble cerous phosphate may be thrown down with the thorium.

SEPARATION OF THE RARE EARTHS. 985

Thorium can be separated from cerium^ by the solvent action of ammo-
nium acetate on thorium oxalate. This as well as the ammonium oxalate
method can be applied to the cerium precipitate. The whole is dis-
solved in strong hydrochloric acid and precipitated by means of oxalic
acid. Glaser says: "Thorium is separated best by converting the oxa-
lates into sulphates, the greater part of the free acid neutralized with
ammonia, the solution boiled and boiling ammonium oxalate added in
excess. After a short time (as soon as oxalates of the cerium metals
have formed but before the liquid has cooled), a solution of ammonium
acetate is added. When cold, the entire cerium group is precipitated
as oxalates while thoria remains in solution. After prolonged standing,
best over night, the insoluble oxalates are removed by filtration F; in the
filtrate, precipitate thoria with ammonia in excess, filter and wash."
All the thorium precipitates are accumulated at E on the diagram.
Another treatment or two, with hydrogen peroxide in neutral nitrate
solution, gives a very good thorium product.

Thorium can be obtained very pure in the following manner: Thorium
hydroxide is first prepared by adding a slight excess of ammonium hy-
droxide to a solution of thorium nitrate and washing well the precipitate
thrown down. This is then added to a solution of acetylacetone in
absolute alcohol and the mass heated on the water bath for a short time,
after which it is filtered and allowed to crystallize. The acetylacetonate
is then placed in a very small retort and carefully distilled in a vacuum.
The portion that condenses in the neck of the retort and in the receiver is
dissolved in concentrated nitric acid, boiled for a short time, diluted with
water, filtered and precipitated by means of oxahc acid. Thorium oxide
obtained by igniting this oxalate is absolutely snow-white even after long
ignition. A determination of the equivalent gave an atomic weight of
232.3. Thorium may also be purified by the sulphate method as fol-
lows: Anhydrous thorium sulphate is dissolved in ice-cold water until
the liquid is saturated.-

The filtered liquid is then heated to 20° C, when nearly pure thorium
sulphate separates. It is then dehydrated and the treatment repeated
two or three times.

Cerium. — F is the starting point for the preparation of pure cerium.
If WyroubofF and Vereneuil's method for separating thorium has been
used, the purification is easily carried out by slightly modifying the zinc
oxide and potassium permanganate method. If other methods have
been used it is best to convert the oxalates, etc., into nitrates. The
liquid is neutralized by ammonia and treated with hydrogen peroxide
in order to remove the last of the thorium. To the solution of cerium

* Glaser, This Journal, 18, 782.

' According to Urbain ammonium acetate aids the solution.

986 C. JAMES.

nitrate, freed from thorium, sufficient ammonium nitrate is added to form
the double salt. An excess of potassium permanganate is run in and
only enough cream of zinc hydroxide added to precipitate most of the
cerium, being sure to leave some in solution, for otherwise praseodymium
and neodymium will accompany the precipitate. The whole is then
heated by steam, filtered and washed with water containing a little ammo-
nium nitrate. The precipitate is dissolved in concentrated hydrochloric
acid, the solution diluted and the cerium precipitated with oxalic acid, G.
The oxalate obtained by this method usually contains a little zinc and
manganese, the latter coloring the oxide brown, so for the final purifica-
tion the oxalate is treated with a slight excess of sulphuric acid and the
whole heated until the fumes of sulphuric add are no longer given off.
The resulting sulphate is dissolved in cold water and the filtered solution
heated on the water bath. The sulphate that separates is washed with
boiling water. This material should give an oxide with only a pak
yellow tint.*

The mother-liquor is treated with oxalic acid, the oxalate which is
thrown down being worked up with the next lot. The filtrate from the
cerium peroxide still contains cerium, which is removed by adding an ex-
cess of zinc oxide and more potassium permanganate should the color be
discharged. The precipitate obtained here is mixed with F. The
filtrate is precipitated with an excess of oxalic acid and the insoluble
oxalates, consisting of lanthanum, praseodymium, neodymitun, etc,
constitute Fraction H.

Lanthanum, etc. — Fractions H and D contain lanthanum, praseo-
dymium, neodymium, samarium, europium and gadolinium together
with small amounts of th& yttrium earths and some cerium.

These are best separated from each other by the fractional crystalliza-
tion of certain double nitrates, such as those formed by the rare earth
nitrates with ammonium, magnesium, manganese or nickel nitrate. For
the separation of lanthanum and praseodymium the double ammonium
nitrates are by far the best and for separating praseodymium from neodym-
ium the manganese salts are to be preferred. However, where one is
working on the large scale it is better to start with the double magnesium
nitrates, as the more soluble portions crystallize more readily than is the
case with the double ammonium nitiates.

The double magnesium nitrates,^ 2[M-"(N08)J.3[Mg(N03)J + 24HjO,
are prepared by dissolving the rare earth oxides in a known amount of
nitric acid. An equal amount of nitric acid is then neutralized by mag-

* Praseodymium is a very common impurity fotmd in cerium. This has been
pointed out many times by different investigators, so that if very pure cerium is re-
quired, one must be on his guard against this substance.

' Demarpay, Compt. rend., 130, 1019.

SEPARATION OI^ THE RARE EARTHS. 987

nesium oxide, after which the two solutions are mixed and evaporated
until upon blowing on the surface, small crystals form. Water is sprayed
over the surface and the whole allowed to cr3rstallize for about twenty-
four hours. The mother-liquor is then poured off and evaporated further,
while the crystals are heated with water until dissolved, the correct
amount to use being soon learned by experience. Both fractions are
again allowed to crystallize for a like period, the concentration of the
solutions being such that half of the solid separates on cooling. Two
fractions have thus been obtained and in subsequent fractionations the
more soluble moves in one direction and the less soluble in the opposite.
After the crystallization of the second series is complete the liquid from
the most soluble portion is poured off and evaporated, while the liquid
from Fraction / is used as the solvent for the crystals forming Fraction //,
adding water or evaporating as may be necessary. The least soluble
portion. Fraction /, is again dissolved by heating with water. The above
is repeated many times. When the fractions at either end become too
small to work they should miss one crystallization and then be added
to the next lot. After a few series of crystallizations, the least soluble
portion becomes very light colored, later growing nearly colorless, and
finally takes a faint green tinge. When the fractions at this end no longer
show the characteristic absorption bands of neodymium, they should be
placed aside and mixed together according to the amount of praseodym-
ium contained therein; in other words, fractions of the same color are
united.

The most soluble portion changes very rapidly. It soon takes a yellow
color and shows a samarium spectrum together with the bands of dyspro-
sium, holmium and erbium. Sometimes at this stage the liquid refuses to
crystallize or else a precipitate may form. If either of these things
happens it is best to dilute with water and saturate with solid sodium
sulphate to separate the yttrium earths and impurities that have ac-
cumulated and interfere with the crystallization. The insoluble double
sodium sulphates are converted back to the double magnesium nitrates
in the same manner as already described and the solution will be found to
crystallize readily on evaporation. The filtrate from the double sodium
sulphates is precipitated with an excess of oxalic acid and the oxalates
of the yttrium group which are thrown down are added to lot B. The
neodymium bands finally become very weak in the jnost soluble frac-
tions and these are set aside for the preparation of samarium, europium
and gadolinium. After the samarium has been separated in this manner
the more soluble portion of the remaining fractions rapidly turns to a
beautiful amethyst and when this occurs it is separated from the rest as
neodymium. After the process has been continued a little longer it will
be found that the material has been split up into four groups according

988 C JAMES.

to the order of their solubilities. Commencing with the least soluble we
have:

1. Lanthanum and praseodymium.

2. Praseodymium and neodymium.

3. Neodymium.

4. Samarium, europium and gadolinium together with small amounts
of terbium, dysprosium, etc.

Lanthanum, — Lanthanum and praseodymium are best separated from
each other according to the method of Auer von Welsbach^ which con-
sists of the fractional crystallization of the double ammonium nitrates
of the type M"'(NOj)3.2(NH4N08) -f 4H,0. These compounds are
crystallized from water containing nitric acid to the extent of one-tenth
the weight of the dissolved solid. To prepare the double salts, the oxides
are dissolved in the required amount of nitric acid and for every three
parts of acid required for the oxides two additional parts are neutralized
by ammonium hydroxide. The resulting solutions are mixed, filtered
and evaporated imtil small crystals form on blowing over the surface of
the liquid. A little water is sprayed over the surface and the whole set
aside for twenty-four hours. The process of fractionation is then carried
out similarly to the double magnesium nitrates. By this method lan-
thanum ammonium nitrate is soon obtained perfectly colorless and a
saturated solution gives no praseodymium absorption spectrum even when
observed through very thick layers. The lanthanum ammonium salt
does not enclose anything like the amotmt of mother-liquor that the
double magnesium compound does. Both cerium and praseodymium
accumulate in the more soluble portion.

The colorless lanthanum salt is dissolved in water, the solution acidified
and precipitated by means of oxalic acid. This oxalate is treated with a
slight excess of concentrated sulphuric acid and the whole gently ignited
until all free acid has been driven off. The sulphate is powdered and
dissolved in water at about i ° until the liquid is saturated, after which
it is filtered, placed in a water bath and gradually raised to 32°. The
solution soon changes to a solid mass which is placed on a Buchner funnel
and washed with hot water. The few grams that remain in solution are
thrown out by means of oxalic acid. The crystallized sulphate may be
rendered anhydrous and submitted once again to the sulphate method.

This lanthanum gives a fine white oxide.

Praseodymium, — ^There are two sources for praseodymium, firstly,
from the more soluble portion obtained from the purification of lanthanum
and secondly, from those fractions of the double magnesium nitrates
which show a strong praseodymium spectrum. These are not mixed but
treated separately.

* Monatsh. Chem., 6, 47 "^

SEPARATION OF THE RARE EARTHS. 989

In the first case the crystallization is carried on until no more colorless
crystals separate, praseodymium accumulating in the more soluble frac-
tions together with a little cerium.

In the second case the double magnesium salts are converted into the
corresponding manganese compounds which are finally fractionally
crystallized^ from nitric acid of sp. gr. 1.3. In order to do this the mag-
nesium double salts are dissolved in water, the solution acidified and the
rare earths thrown down by oxalic acid. The oxalates obtained are
washed, dried and ignited to oxides. The oxides are then dissolved in a
known amount of nitric acid. An equal amount of nitric acid is then
neutralized by manganese carbonate, after which the two solutions are
mixed. A precipitate of manganese peroxide is sometimes obtained at
this point but it is easily removed by adding a little oxalic acid and warm-
ing. The least soluble portion, from this fractional crystallization that
no longer gives any neodymium bands in the spectroscope, is dissolved
in water, acidified and thrown down with oxalic acid. The fractions of
praseodymium ammonium nitrate that are free from lanthanum are also
dissolved and precipitated by oxalic acid. The praseodymium oxalate
from the two sources is then united. This material may be impure, owing
to the presence of cerium.

Cerium can be separated in several different ways. One method con-
sists in treating the nitrate solution with potassium permanganate and a
little sodium carbonate. A separation is obtained according to Wyrouboff
and VerneuiP by adding a solution of sodium acetate to a solution of the
nitrates and precipitating the cerium by hydrogen peroxide. The above
methods throw down a certain amount of praseodymium also, so the
precipitate should be worked up again.^

Neodymium, — Pure neodymium is obtained by continuing the crystal-
lization of the neodymium magnesium nitrate obtained somewhat earlier.
After a few more series of crystallizations the liquid assumes a beautiful
bluish lilac color which is seen better when some of the solution is diluted
with water. On observing the spectrum the absorption bands in the blue
stand out clearly. When the solution contains samarium or p?;aseodym-
ium, these weaker neodymium bands are usually a little hazy. An
excellent test of the purity of neodymium is found by observing the
color of the oxide which is blue only when pure.

Samarium and Europium. — Samarium, europium and gadolinium are
contained in the mother-liquors which are obtained during the fractiona-
tion of the double magnesium nitrates. The solutions are evaporated

* Lacombe, Bidl. soc. chim. [3], 31, No. 10 and Chem. News, 89, 277.

* BtUl. soc. chim. [3], 19, No. 6 and Chem. News, 77, 254.

* Meyer and Koss, Ber., 35, 672, recommend magnesium acetate in place of sodium
acetate.

990 C. JAMES.

and the residue fractionally crystallized from nitric acid^ of 1.3 sp. gr.
The addition of the isomorphous bismuth magnesium nitrate aids enor-
mously in the separation of these elments as Urbain and Lacomb' have
shown. Its solubility places it between samarium and europium and it
also assists in the crystallization of the sirupy mother-liquors inasmuch
as it carries down with it the more crystallizable portions. Samarium
is obtained from the least soluble fractions. Europium is separated
from the excess of bismuth magnesium nitrate which is found between
the samarium and gadolinium fractions. The bismuth is thrown down
by hydrogen sulphide and the mother-liquor precipitated by means of
oxalic acid.

Gadolinium. — The fractions between europium and dysprosium, etc,
consist mainly of gadolinium magnesium nitrate. These solutions are
acidified and the gadolinium thrown down as oxalate. This is then
washed and ignited to oxide. The resulting oxide is converted into the
nickel nitrate' of the type 2Gd(N08)3.3Ni(N03), + 24H2O, which is then
fractionally crystallized from nitric acid of density 1.3. Terbium is left
in the most soluble portion.*

Fraction B contains oxalates of terbium, dysprosium, holmium, >i;trium,
erbium, thulium, ytterbium and scandium. This material is converted
into the anhydrous sulphate, the latter dissolved in cold water and poured
over an excess of barium bromate.* The whole is well stirred and placed
on the water bath. After the double decomposition is complete, t. e.,
when the clear liquid gives no precipitate with barium bromate solution
after diluting and boiling, the mass is filtered and evaporated until a
drop removed on the end of a glass rod nearly solidifies when stirred on a
watch glass. A little water is then sprayed on the surface and the whole
submitted to fractional crystallization. The absorption spectrum soon
shows that a rapid change is taking place. Small amounts of samarium
and gadolinium are rapidly separated in the least soluble portion while
the next fractions contain terbium and give oxides of a deep red-brown
color. Dysprosium and holmium are more soluble than terbium.
Yttrium places itself between holmium and erbium.®

* Demarcay, Loc. cit.

' CompU rend., 137, 792 and 138, 84.

• Urbain, Compi. rend., 140, No. 9.

^ The writer is applying the bromate method to a complicated mixture of samarium,
gadolinium, terbium, dysprosium and holmium and is obtaining interesting results.

* James, This Journal, 30, 182 and Chem. News, 97, 61.

• The writer, in a previous paper describing the serial order of the bromates, made
an error in the position of yttriun. This element, as stated above, ranges itself in the
fractions between holmium and erbium. The cause of the error was due to the fact
that the material on which the serial order was worked out contained only a little
yttrium and a fair amount of gadolinium. At the time, gadolinium was not suspected
and the colorless bromate crystals were supposed to be due to yttria.

SEPARATION OF THE RARE EARTHS. 99 1

The most soluble portion contains erbium, thulium and ytterbium.

Terbium, Dysprosium and Holmium, — ^These elements are extremely
difficult to separate. Operations with the bromates are still in progress.
Terbium separates in the least soluble together with samarium and
gadolinium. The separation of dysprosium from holmium is extremely
slow and so much so that the bromate method is of no value for this
work. Urbain recommends the fractionation of the double nickel nitrates
from nitric acid of density 1.3 to separate samarium and gadolinium from
terbium, dysprosium and holmium. The more soluble portion, con-
sisting of terbium, etc., with some gadolinium, is converted into the
simple nitrate and fractionally crystallized from concentrated nitric acid
in the presence of bismuth nitrate. Terbium collects with the bismuth
nitrate in the fractions between gadolinium and dysprosium. Fractional
crystallization of the ethyl sulphates gives dysprosium. Small amounts
only of dysprosium and terbium have been separated. The preparation
of pure holmia has not yet been accomplished.

Yttrium. — ^Yttrium is obtained from the bromate fractions between
holmium and erbium. Some fractions show holmium bands in addition
to erbium. It is most easily prepared from the fractions that are free
from holmium. The earths are precipitated from dilute boiling solutions
of the bromates by the addition of boiling potassium hydroxide solution.
The hydroxides are filtered off, washed and converted into the nitrates.
Yttrium is then separated by the method of Muthmann and Rolig* as
follows: The concentrated neutral nitrate solution is boiled and a thick
cream of magnesium oxide added until the liquid no longer gives the
absorption bands of erbium. The fractions that contain holmium in
addition to erbium can be put through the same process. Yttrium
oxide obtained by the bromate and magnesium oxide methods is snow-
white and absolutely free from terbium, etc.

Erbium, Thulium, Ytterbium and Scandium. — ^These elements are
separated from each other by the continued fractionation of the most
soluble portion of the bromates. The erbium solutions become a beauti-
ful rose tint. Thulium collects between erbium and ytterbium. The
mother-liquor contains ytterbium with a very little scandium. The
neutral solution is saturated with potassium sulphate when scandium
potassium sulphate separates as it is insoluble in potassium sulphate
solution. The filtrate on the addition of oxalic acid gives a precipitate
of ytterbium oxalate. "

The fractionation of the least basic earths is still being carried on with
the object of preparing pure thulium for a determination of the atomic
weight and also to confirm Urbain's lutecium.

In conclusion I again thank the Welsbach Company for large amounts

* Ber., 31, 1718.

992 CHARLES F. MABERY AND J. HOWARD MATHEWS.

of material received through the courtesy of Dr. H. S. Miner. I also
tender my thanks to the Christiania Minekompani, of Christiania, Nor-
way, for many mineral specimens for examination.

New Hampshire College, Durham, N. H.,
March 5, 1908.

[Contributions from thb Chbmical Laboratory of Case School of Applied

Science. No. 2.]

ON VISCOSITY AND LUBRICATION.

By Charles P. Mabery and J. Howard Mathews.

Received April 7, 1908.

Excepting the work done in this laboratory during the last fifteen years in
determining the composition of American petroleum, so far as we know,
no attempts have been made to ascertain the composition of lubricating
oils with reference to the hydrocarbons or even the series of hydrocarbons
of which they consist.

Neither the composition of the oils nor the source of the petroleum from
which the various products were manufactured have been relied on as a
means of distinguishing^ differences in quality or durability, except a
general distinction between straight hydrocarbon oils and compounded
oils. Until comparatively recently, the refiner had to rely for high vis-
cosity on mixtures of animal or vegetable oils with oils separated from
petroleum, and the latter were obtained from Pennsylvania, Ohio, or other
similar natural oils ordinarily referred to as paraffin oils, since they con-
tained the solid paraffin hydrocarbons, C^ll^^_^^. Naturally, the refiner
became convinced of the superiority of his compounded oils over straight
hydrocarbon oils, and this idea has been maintained so persistently, it
still prevails very generally with consumers of lubricating oils.

But within the last ten years, other varieties of petroleum have been
found to yield lubricating oils with superior viscosity and wearing quali-
ties which makes it no longer necessary to rely on compounded oils either
for use on bearings or in cylinders. This is of especial importance with
reference to cylinder oils, for it is well understood that the conditions of
high temperatures and highly heated steam in cylinders lead to saponifi-
cation of the animal or vegetable oil used in compounding, with conse-
quent corrosion of the cylinder. As is well known, castor oil is one of the
very best lubricating oils, especially for durability, but its general use is
precluded by its high cost. It is now possible to prepare straight hydro-
carbon oils fully equal in viscosity and wearing qualities to castor or any
other high viscosity vegetable oil.

Viscosity is generally accepted as a standard of value in classifying
lubricating oils, but it is not certain that it is reliable as indicating the
durability and wearing qualities of oils differing widely in composition.
There is little doubt that a confirmation of viscosity by chemical data

ON VISCOSITY AND LUBRICATION. 993

and frictional durability tests may be depended on to give accurate
information for commercial use. The viscosity of lubricating oils has
received much attention and several methods and forms of appa-
ratus have been suggested for its determination, but for the most
part of arbitrary construction and comparison, and differing so
essentially that determinations made with different instruments are not
easily and readily comparable. As thus determined, viscosity is but
an arbitrary standard based on an assumption that the outflow of a
liquid through an orifice, influenced as it is by several physical
conditions, is a correct measure of surface viscosity between bearing sur-
faces. It is merely a relative comparison with an oil arbitrarily selected
as a standard, or with water or by means of a metal apparatus arbitrarily
constructed. But with the use of water, evidently, the conditions of tem-
perature must be stated in the results, for the viscosity of water is quite
different at different temperatures, and slight variations in temperature
have likewise an important influence on the viscosity of oils.

It is, therefore, necessary to know the temperature coefficient of water,
and it would be interesting to follow out a series of observations with an
homologous series of hydrocarbons, although the possibility of such an
investigation is almost precluded by the immense labor necessary in sepa-
rating in an acceptably pure form the individual hydrocarbons. We have
on hand, members of the different series C^H^^.^,, C^H,^, C^H^^..^,
C^H^ , and we have made a series of observations on some of the indi-
vidual hydrocarbons. Interesting results have also been obtained on the
viscosity of mixtures, showing the influence of the hydrocarbons of the
different series.

In attempting to arrive at a series of determinations which should avoid
the errors in methods in which differences in specific gravity, and accurate
observations of temperature are neglected, it was evidently inexpedient
to use any of the commercial methods, especially since, as explained above,
the data afforded by those methods are merely empirical, and with no
definite relations to a common standard. The well-known method of Ost-
wald was selected, therefore, as best suited for these determinations, and
the apparatus employed needs no detailed description. In this method a
definite volume of liquid flows through a capillary tube under a definite
head. In the calculation, the pressure under which the liquid flows
through the capillary, is corrected for its density in the Ost aid forn^ula :*

' Ostwald-I uther, Physico-Chemische Messiingen., p. 20.

V « The viscosity of the liquid examined,

S "- Density of the liquid examined;

t — Time of outflow of the liquid examined;

ijo " The viscosity of the standard liquid;

Sb ** Density of the standard liquid;

to ■" Time of outflow of standard liquid;

17 - i?o S//So<D.

994 CHARI.ES p. MABERY AND J. HOWARD MATHEWS.

The values thus obtained, express the ratio of the viscosity of the liquid
under examination to a standard liquid used for reference. Water is the
standard liquid most commonly chosen, and the values of i) are referred to
as "Specific Viscosities," t. e. , the ratio of viscosity to that of water at that
particular temperature. The advantage of using water consists in the
ease with which it may be obtained sufficiently pure, and in the fact that
the value in absolute units for the viscosity of water is the best known of
any liquid, and specific viscosities may be converted easily into absolute
units. The specific viscosities obtained at different temperatures are not
comparable, since they express only the ratio at the particular tempera-
ture chosen and take no account of the change in volume of the apparatus,
especially in the size of the capillary. To compare the results obtained at
different temperatures, it is necessary to convert the values into absolute
units, using the known values for the viscosity of water at the temperature
used.

From interpolation and extrapolation of the result obtained by Thorpe
and Rodger* we obtained the values 17 = 0.01007 at 20*^ C. and 17=
0.004625 at 60 ^^ C, where 17 is the coefficient of viscosity in absolute units,
i. e., dynes per square centimeter. By multiplying the values obtained
at 20° and 60® by these numbers, comparable results are obtained.

Constant temperature was maintained by placing the viscosimeter in a
glass thermostat through which observations could be made, and which
was supplied with water from a larger thermostat maintained at a constant
temperature by means of an electric thermo-regulator. The water
was pumped from the larger to the smaller thermostat by means of a
small lift-pump operated by a hot air engine. The temperature in the
glass thermostat was held at 20° ( ±0.02) ; since the viscosity of the oils,
like that of most liquids, changes about 2 per cent, per degree, this small
fluctuation is negligible.

The measurements at 60°, were made in a glass thermostat of about
10 liters capacity in which the temperature was maintained by super-
heated steam injected at the bottom through a small orifice. The steam
could be easily regulated and the temperature readily maintained at 60°
(±0.02). In observations on the paraffin hydrocarbons, it was found
that viscosity increases with some regularity in the homologous series
with decreasing percentages of hydrogen.

These hydrocarbons were obtained by long-continued fractional separa-
tions under systematic conditions, and their identity was shown by anal-
ysis and critical examination ; but it is doubtful whether the homologues
can be completely separated even by very prolonged distillation unless
much larger amounts of material are used than is possible on a labora-
tory scale.

' Proc. Roy. Soc., 1894; Z. physik. Chem., 14, 361.

ON VISCOSITY AND LUBRICATION. 995

Tablb I

(20°).

Hydrocarbon.

B. p.

Sp. gr.

Specific viacosity,

CyHift

98-icx)°

0.724

0.51

CrH,,

125^

0.735

0.60

CjoHa*

172-173°

0.747

0.96

CioHa

1 74-1 75 <»

0.753

0.95

CjoHa

163**

0.745

0.89

C..H,'

209-210°

0.762

1-25

C|lHlB

212-214"

0.769

1.49

CiftHaa

158-159° (50 mm

0.793

2.79

C„H„']1

155-158^

((

0.796

2.75

^18^34

174-175''

0.799

3-35

CjgHjg

199-200°

0.813

5 97

Sp. gr.

Specific viscosity.

0.781

10.88

0.841

21.23

0.775

8.51

0-835

15.63

In Table t, it will be observed that viscosity increases somewhat irregu-
larly with every increment of CHj, and that the change is greater with the
increase in molecular weight. Since the proportion of hydrogen to carbon
apparently influences materially the value of viscosity, it seemed desirable
to compare the viscosity of hydrocarbons of different series. In Table 2,
are given the values for hydrocarbons with the same boiling points, but
members of different series.

Table 2 (60°).

Series. B. P.

CnH2n+2 294-296° (50 mm.)

C«H2«-j 294-296° "

Cnn2n^2 274-276° "

CnU^n 274-276° "

The greater viscosity of the hydrocarbons poorer in hydrogen, is clearly
shown. In comparing the viscosities of the two hydrocarbons boiling at
294^-296° it will be observed that the difference is greater than the differ-
ence between the viscosities of the two hydrocarbons boiling at 274^-276°.
This demonstrates the influence of a decreasing percentage of hydrogen
since*inTthe first set, the change is'^from 2«+a to 2«— 2, whereas in the
second set, the change is only from 2«+2 to 211. Both viscosity and
specificTgravity increase with the decreasing hydrogen. Another possible
influence must not be overlooked, however, namely, the internal structure
of the different hydrocarbons. It is reasonable to assume that the
straightljor open-chain structure of the paraffin hydrocarbon C^H^.^.^
behaves differently under the stress of internal forces on which viscosity
depends, from the ring or cylic structure, which must be accepted for the
other series, until more is definitely known concerning their constitution.
Certainly this is plainly shown in lubrication, where the paraffin hydro-
carbons are of comparatively little value.

If, then, the lower series furnish lubricators with greater viscosity, the
addition of a member of a higher series should give a mixture lower in
' Of approximately this composition.

996 CHARLES F. MABERY AND J. HOWARD MATHEWS.

viscosity. Observations were therefore made on mixtures of pure hy-
drocarbons of the different series with reference to variations of viscosity,
and the results of these measurements are given in Tables 3 and 4.

Table 3 (20®).
Influbncb op a Solid Paraffin Hydrocarbon.

Specific
Hydrocarbon. B.P. Sp.Gr. TtBcosit7.

(a) Penn. distillate CiJIih-^i filtered 312-314° (50 mm.) 0.868 87.42

(6) Same cooled to —10° and filtered 312-314° " 0.868 88.16

of same B.P 312-314° • 0.868 82.30

(ci) Penn. distillate CnHin cooled to — 10°

and filtered 276-278° " 0.861 37.57

The amounts of solid paraffin hydrocarbons added in these experiments
were all that the oils could hold in solution at that temperature. Although
no appreciable changes appear in specific gravity, there were material
changes in viscosity. Table 4 shows that the diminution in viscosity still
holds at a higher temperature, but in a less marked degree even when a
larger portion of the paraffin hydrocarbon is introduced.

Tablb 4 (60**).
Influence of a Soud Paraffin Hydrocarbon.

Specific
Hydrocarbon. B. P. Sp. Gr. viacosity.

(a) CicHm— 3 294-296° (50mm.) 0.841 21.23

(6) C«H2«+2 294-296° " 0.781 lO.Sii

(d) Pa. hydrocarbon CuHan+a 274-276° " 0.775 8.51

(e) c + 5 per cent, of i o . 831 15 . 16

South American Oils.

Two well-fractioned distillates from South American petroleum of unde-
termined series, but doubtless poor in hydrogen, were examined at 20®
with the following results :

Table 5.

B. p. sp. gr. Specific visoosity.

Distillate 1 155-160° (50 mm.) 0.884 S-H

Distillate 2 215-220° " 0.896 19-57

So far as viscosity is an indication of lubricating value, it is e\adent that
these distillates are inferior to the Pennsylvania distillates of the same
gravity. The heaviest Pennsylvania distillate examined, with lower
gravity than the South American distillates, viz. o . 08687.. had a viscosity of
88.16 as compared with the numbers 8. 14 and 19.57 for the latter. But
on the other hand if we consider the properties of distillates taken at the
same temperatures, the South American distillates have a much greater

ON VISCOSITY AND LUBRICATION. 997

gravity and viscosity. It is evident therefore that neither gravity nor
boiling point can be depended on for lubricating value unless the source of
the oil is known. The method of manufacture has also much to do with
relation of specific gravity and lubricating value.

Valuation of Lubricating Oils.

The various standards which have been proposed for the valuation of
lubricating oils, are based on their physical properties, especially on spe-
cific gravity and viscosity. Both specific gravity and viscosity are, how-
ever, unreliable, unless the source and composition of the oil are known,
and unless viscosity is still further defined by frictional tests on bearings
under definite conditions which demonstrate the wearing quality of the oil.
As is well known, many lighter oils have a greater viscosity than other
heavier oils. With these limitations, the property of viscosity has a direct
relation to lubricating value.

An ideal lubricator is evidently one which holds two bearing surfaces
at a suflScient distance from each other to prevent friction between them
and at the same time has the least possible amount of internal friction so
that the work necessary to overcome the friction of the oil particles upon
each other may be reduced to a minimum. If the oil has too great an
internal friction, considerable mechanical energy is expended in over-
coming this friction and in conversion of mechanical energy into heat,
which should be avoided as far as possible. The choice of a lubricator
must, therefore, depend upon the weight to be supported and upon the
speed desired. At a high speed, an oil of small internal friction should be
chosen, but for slow heavy work, an oil of greater viscosity must be used
to support the weight, and because of the slow speed, the greater internal
friction of the oil is of less consequence. The viscosity of an oil then, with
a knowledge of its composition and with the aid of frictional tests, gives a
direct measure of its usefulness as a lubricator under any given conditions,
since viscosity is merely another term for internal friction.

But viscosity measured at ordinary temperatures may lead to erroneous
conclusions concerning true lubricating value for higher temperatures.
This would be especially true for light loads and high speeds where the
temperature is considerably higher than the surrounding temperature.
The viscosity of all liquids decreases with rise of temperature, but not to
the same extent for all liquids. In general, the decrease is about 2 per cent
per degree centigrade, within ordinary ranges of temperature. The fol-
lowing table shows the dependence of viscosity on temperature for two
Pennsylvania distillates.

Tablb 5.

viscosity in Viscosity in Change in

dynes per dynes per viscosity

HydTocarbon. B. P. sq. cm. at ao'^. sq. cm. at 60^. per degrees C.

C«H2«-2 312-314® (50 mm.) 0.8803 0.1320 2.12%

C«H2« 27&-278® (50mm.) 0.3783 0.0723 2.02%

998 CHARLES F. MABERY AND J. HOWARD MATHEWS.

The first oil shows a greater variation per degree than the latter of the
C^H,^ series. The coefficient of the more mobile C^H,^^, distillate with
the same boiling point at temperatures above its melting point, should be
still smaller. Small differences in the temperature coefficient of viscosity
should also be expected in the same series when members of difTerent bdi-
ing points are compared. It may readily occur then that two oils have
quite different viscosities at high temperatures while at lower tempera-
tures the value should be the same.

The viscosity of a hydrocarbon oil may be changed, as is commonly done,
by the addition of viscous vegetable oils such as castor. Such blended
oils, made to meet specifications for a required viscosity, are in common
use. For certain kinds of work, such oils are doubtless serviceable, but
probably have less durability than mineral oils of equivalent viscosity.
In cylinders where the oil comes in contact with superheated steam, the
vegetable constituents must be saponified to a greater or lesser extent , with
consequent danger of corrosive action by the free add. There is in fact
at present, less necessity for the use of compounded oils to obtain desirable
viscosity and wearing qualities since the recent development of oil terri-
tory which yields heavy products.

Durability tests on compounded oils and pure distillates by a frictional
machine in the Case mechanical laboratory have shown the superior fric-
tional qualities of the mineral oils.

Allusion has been made to the fact that viscosity alone caimot always
be relied on to give a correct estimate of the wearing qualities of a lubri-
cating oil on a bearing. Much time has been devoted to the wearing qual-
ities of oils in this laboratory, and some of the results were presented in a
paper at the New York meeting of the American Chemical Society, Decem-
ber, 1906. The machine employed in these tests was constructed by Prof
C. H. Benjamin for the mechanical laboratory of Case School of Applied
Science. It consists of a bearing in two sections so arranged that the
upper section can be raised, and when in position can be weighted by
means of strong springs on the axle to be lubricated. Attached to the
upper section of the bearing is a lever arm extending outward over the
platform of a scale, and the outer end rests on a support standing upright
on the scale platform by which the scale registers by changes in weight the
varying pressure caused by changes in friction. An oil cup is connected
with the bearing in such a manner that a regular supply of the oil to be
tested is permitted to flow over the journal. A thermometer is also in-
serted in the bearing close to the journal for the purpose of indicating the
changes in temperature during a test. An examination of a lubricating
oil by this method, consists in allowing the oil to run on the axle at the rate

ON VISCOSITY AND LUBRICATION. 999

of six to eight drops per minute for two hours, noting the temperature and
pressure on the balance every five minutes. At the end of this period the
flow of oil is stopped and the journal allowed to run until the oil ceases to
lubricate, which is immediately shown by greatly increased frictional
pressure and a large rise in temperature. The crucial test of the oil is the
time it can support the axle after the flow is stopped.

The number of revolutions was set at about 500 per minute, and an
occasional reading of the speed was taken to be sure that the axle was
revolving under fairly constant conditions. The pressure on the axle ex-
erted by the spring set on the upper half of the bearing was equivalent to
1300 pounds. In practice, evidently every oil should be tested under an
equivalent of the load and speed it is expected to carry. The results of
several of the oils tested are given below. Every oil was run for
two hours with continuous lubrication, and the time stated in the table
is reckoned from the end of the hour when the flow was stopped,
to the time when the oil refused to lubricate, which was indicated by
sudden and large increase in temperature and frictional load. Such a
dumbility test affords a fair comparison of the ability of the oil to support
friction.

Having on hand a variety of hydrocarbons of the series C^H^^^j,, C^H^,,,
^rJ^2H—2 ^^^ ^n^2H—4 which havc accumulated during the years one of
us has been occupied in ascertaining the composition of American petro-
leum, an opporttmity was afforded to ascertain by frictional tests whether
hydrocarbons of a different series exhibit differences in wearing capacity
corresponding to variations in viscosity. For the purpose of comparing
the frictional qualities of these hydrocarbons with the requirements of
lubricators in actual use, several vegetable oils were included, viz., castor,
sperm, and rape oils, which are the best lubricators of their class, and often
used in compounding with petroleum oils as an aid to the viscosity of the
latter. Some results of the best compounded oils are also included for
comparison. The following elements were used in the calculation of

results:

Constant of lever arm (determined by independent observations) 13.656 lbs.

Radius of lever arm 31.625 inches.

Radius of journal 1.61 "

Load 1,300 pounds.

II 62s
Scale reading — 13.625 ■ \

Formula: Coefficient of friction = — •

1300

The real tests of lubricating capacity depend on temperature, measure
of friction, and the time the oil continues to lubricate after it ceases to
flow on the journal. The ktter observations are given in the column
headed "Test of durability.'* It is interesting to observe that the life

lOOO

CHARLES F. MABBRY AND J. HOWARD MATHEWS.

•

M M 1-

to 2

0 c

! S"8"8

• • ■

■

• • ■

•

0 0 0

0 0

000

•r

■

»o

^ .

M (S M

N M

W CO cO

000

> • •

0 0

• •

000

• ■ •

0 0 0

0 0

000

N i-« to

^ «

to »o r*

000

0 0

000

• • ■

•

• • •

000

0 0

000

0 0

000

■j

•^ «o <*h

10 0

0 0 "O

w 00 vO

<N VO

to « w

M M M

« M

N « «

•

000

0 0

OOP

r>. 0 rO

10 0

0 0 "O

NO 00 00

0 *0 M

\0 C4

M M M

M H

»N M M

•

•**

000

0 0

000

I CO

vO 0 to
r^ 00 !>.

RR5c85

_ >»«

OS-

0 0 0

0 0

0 »0 Q

W M M

»o -^

•0

*" io-

r- ^ -K

j; r^ 0" 0^ 0" 0;
■a 2 g S^ 2 2

On m C) C«

•

0 W X.

.2 -^ <N

^ %.• v-/ ^^

n «3>»

00 to »0

^"^ oa

to* 00 ""tf

> 0 ON 0 N «^

^ &§

fO 00 ON

m H

1 W M W

Si >»

M 00 rO

2--

NO VO «s

r>. Tf •-• « S*

C >

00 00 O^

Ov O* On On 00

CD ^

odd

•

0 0

■

• •
000

• • •

r* ri* ^^

CCS

s c a

0 0 0

»0 irj rO

%«• >,„^ ».»-/

0 0 0

^ ■* 2v

l^ *« fO

N ro «

4 i »

• •

I y-N /-N

• > *
• •

:'21

. 73 "d

: as.

: E a

■

•0

a 0 0

: 8 8

- J

c S c

■0 "«*•

ill

3 g B

' 0 0

2 p 0

^ c

• nS nS

Wi H k<

'0 -d '0

5 t

S..S.5

X K S

! PS

THE COLORED SALTS OP SCHIFP'S BASES. IOC I

of the individual hydrocarbons increases with the decrease in hydrogen,
and in a similar ratio to the increase in specific viscosity. Doubtless the
most valuable quality of an oil is its ability to reduce friction to the small-
est value; of the hydrocarbons compared in this test, the one with least
hydrogen, of the series C^H^^..^ seems to show the best eflSciency, as it
also shows the greatest durability. Of the vegetable oils compared, the
castor shows the longest life, but the sperm shows the least coefficient of
friction, as it does also the lowest temperatures. The hydrocarbon C^H^^.^
stands ahead of the vegetable oils in durability, and the equal of the sperm
in temperatiilies and coefficient of friction. In the above table, tempera-
tures are selected at the beginning of the test, at the end of two hours, and
when the oil ceases to lubricate. The coefficient of friction is also calcu-
lated at these points. This method appears to be capable of giving valu-
able information concerning the nature of lubricating oils. Like all tests
of such oils, the results are relative referring to some common standards.
Evidently in practice, these standards must be based on the work required
of the oil, taking into account certain data, such as speed of th e journal and
the load carried. The coefficient of friction as used above represents the
fraction of one pound for each potmd of load on the journal which is sus-
tained by the oil in use.

With reference to the comparative value of the various series of hydro-
carbons in petroleum from which lubricating oils are prepared, it must be
understood that the series C^H,^^, has a low lubricating value; this
was shown above in tests of specific viscosity. Lubricating oils pre-
pared from Pennsylvania petroleum consist for the larger part of the
series C^H^^ and series CJlI^„_^. Those from the heavier oils consist
largely of the series C^H^^., and*the series C^H,^.^.

As to the structural constitution of the series of hydrocarbons in lubri-
cating oils, at present nothing whatever is known. The series C^H^^^,
is doubtless represented by the straight chain or its isomers, all with the
open chain. The series poorer in hydrogen, no doubt have the cyclic ring
structure of the form of condensed benzene rings in part, or condensed
methylene rings, single larger rings than are at present recognized, or
smaller rings with condensed side chains.

CX^EVBLAND, OHIO.

THE COLORED SALTS OF SCHIFF'S BASES.

n. THE HYDROCHLORIDES OF BASES FORMED BY CONDENSING /-AMINO-

DIPHENYLAHINE WITH AROMATIC ALDEHYDES.

By p. J. Moors and R. G. Woodbrzdob, Jr.

Received April lo, 1908.

In a paper recently published by one of us, in collaboration with Mr. R.
D. Gale,^ attention was called to the fact that bases of the general formula

» Tms Journal, 30, 394 (1908).

I002 F. J. MOORE AND R. G. WOODBRIDGE, JR.

■CH,

R — CH = N-< >— N

^ Vh,

add successively two molecules of hydrochloric acid to form salts which
differ markedly in color. While the bases themselves are, for the most
part, of a light yellow or orange color, the salts containing one molecule of
acid are dark red, while those containing two molecules are light or pak
yellow, resembling the free bases in this respect.

The present investigation was tmdertaken in order to determine whether
the behavior described was not still more general, and for this purpose,
to study the color of the salts of bases formed by condensing />-aminodi-
phenylamine with aromatic aldehydes. These condensation products
have the general formula

R— CH=N— / N— NH— CjHj,

In the paper above referred to by Moore and Gale, a preliminary state-
ment was made to the effect that these bases add one molecule of hydro-
chloric add to form dark red salts, but that they do not add a second
molecule of the acid. This statement requires modification. The be-
havior of these diphenyl compounds is, in fact, entirely analogous to that
of the dimethyl compounds already described. In this case, however,
the dihydrochlorides are formed with more difficulty. If one molecule of
the base be treated in ethereal solution with two molecules of dry hydro-
chloric acid dissolved in ether, the resulting precipitate is dark colored,
consisting of the monohydrochloride. If, now, the solution be kept satu-
rated with hydrochloric acid by passing a current through it for several
hours, the precipitate first formed gradiiklly becomes lighter in color, and
in some cases, practically white. When this precipitate is analyzed, it is
found to contain two molecules of acid. Of the dihydrochlorides studied,
all but one are sufficiently stable to retain the two molecules of hydro-
chloric acid when dried in a vacuum desiccator containing caustic soda.
The exception is the salt of the base formed by condensing />-aminodiphenyl-
amine with cinnamic aldehyde. This salt is quite Hght in color when
first prepared, but after standing some days in the desiccator, it becomes
dark, and as the analytical data clearly show, gradually loses a whole
molecule of acid.

It is true of all the dihydrochlorides that their color becomes much
darker on standing, even in dry air, and some of them become quite deep
red again- We are at some loss to account for this. It may be due to
the fact that there are two dihydrochlorides, one red and the other >'ellow.
Another explanation would be that some acid had been lost, resulting in
the formation of sufficient monohydrochloride to account for the cotor
observed. Inspection of the analytical data presented below will show

THE COLORED SALTS OF SCHIFF'S BASES. IOO3

that any such loss is too small to clearly show itself in the analyses. This,
however, is not conclusive, as we know, for example, that a very extensive
blackening of silver chloride has very little influence upon its composition,
and thus, in this case also, the presence of a small amount of the salt con-
taining only one molecule of acid might account for the phenomenon. The
behavior of the cinnamic aldehyde compound above referred to, would
certainly point, by analogy, to a loss of acid as the cause of the change in
color. If two kinds of salts are present, whatever their nature, it may
prove possible, by means of solvents, to isolate one or both of them, and
some experiments in this direction are in progress.

Experimental Part.

The /)-aminodiphenylamine used in this investigation was prepared
from diphenylamine by treatment with nitrous acid, forming the nitros-
amine. The latter wasthen subjected to the Fischer-Hepp rearrangement,
forming the para nitroso compound, and this was finally reduced with
ammonium sulphide. So prepared, the substance forms almost colorless
needles of a pearly luster melting at 75°.

Benzylidene p-aminodiphenylamine. — ^This compound was first prepared
by Heucke* by the action of />-aminodiphenylamine upon benzaldehyde .
The base is of a pale yellow color.

The monohydrochloride is blood-red. Calculated for CigHi8N3,HCl; CI,
II .43; found, 13.48.

The saturated salt, prepared by subjecting the ethereal solution to the
action of hydrochloric acid gas for some hours, is nearly white when first
precipitated. When kept in a desiccator over caustic soda it turns reddish.
The analysis speaks, however, for the presence of two molecules of acid in
the salt. Calculated for Ci9H,5N2,2HCl: CI, 20.53; found, 19.95 and
19.90.

Salicylidene p-aminodiphenylamine, — ^This compound was also prepared
by Heucke. The base is light yellow.

The red hydrochloride has the following composition: Calculated for
Ci^ioON2,HCl : CI, 10.91; found, 11. 21.

The dihydrochloride is light yellow. Calculated for C^H^fi^^a^QX:
CI, 19.63; found, 20. 59.

Piper onylidene p-aminodiphenylamine. — ^The base is of a light yellow
color and melts at 116®.

Red hydrochloride. Calculated for Cj^HieOjNjjHCl: CI, 10 07; found,
10.24.

The saturated salt is light yellow. Calculated for C2oHieOjNj.2HCl:
CI, 18.22; foimd, 18.49.

1 Ann., 355, 189 (1889).

I004 P* J* MOORE AND R. G. WOODBRIDGB; JR.

Anisylidene p-aniinodtphenylamine. — ^This substance crystallizes from
alcohol in silver-gray scales of a pearly luster which melt at 105 **. Its
composition was confirmed by a nitrogen determination. Calculated for
C20H18ON,: N, 9.30; found, 9.48.

The monohydrochloride is red, but of a lighter shade than most of the
others hitherto examined. Calculated for CjiH^ON^HCl : CI, 10.47;
fotmd, 12.42.

The dihydrochloride as we obtained it, was of a pale pink color. Calcu-
lated for Cj^,gONj,2HCl: CI, 18.90; found, 19.57 and 19.75. Although
these results are a tittle high, we should hardly be justified in ascribing
this to a tendency to add three molecules of acid as does anisylidene p-
aminodimethylaniline.

Cinnamylidene p-aminodiphenylamine, — ^This compound was prepared
by condensing />-aminodiphenylamine with cinnamic aldehyde. The base
crystallizes from alcohol in brilliant yellow scales and melts at 145^. Its
composition was checked by a nitrogen determination. Calculated for
CjoH„Nj: N, 9.42; found, 9.37.

When an ethereal solution of this base is treated with dry hydrochloric
acid, the resulting precipitate is extremely dark in color, almost black.
Calculated for Cj^jHijNjjHCl: CI, 10.59; found, 10.55.

An excess of hydrochloric acid finally yields a product of much lighter
color, which, when first filtered oflf, is pale pink. This substance, however,
when dried in a vacuum desiccator over sulphuric acid and caustic soda,
rapidly turns dark. A sample of this product, after standing in the dedc-
cator for two days, still retained a perceptible odor of hydrochloric add
and contained 16 . 14 per cent, of chlorine. A dihydrochloride should con-
tain 19 . 10 per cent. The same material, after standing for five days, was
nearly black, and no odor of hydrochloric add was noticeable. This mate-
rial contained 1 1 .06 per cent, of chlorine, which is only slightly in excess of
that required for a monohydrochloride, as indicated above.

Summary.

When />-aminodiphenylamine is condensed with aromatic aldehydes,
the resulting bases show the same curious color phenomena in the forma-
tion of their salts which are exhibited by the analogous compounds of
/>-aminodimethylaniline ; that is, they unite with one molecule of add to
form dark colored salts and with two molecules to form light colored ones.
The salts containing two molecules of add are less easily formed than in
the cases hitherto studied, and, in a single instance, the second molecule
of acid is readily lost.

MA88ACHU8BTTS IKSTITUTB OP TBCHKOLOOT

BosTOK, April 7, 1908.

THE ENDO- AND EKTOINVERTASE OF THE DATE. IOO5

THE ENDO- AND EKTOINVERTASE OF THE DATE.^

By a. B. Vinson.
Received April x, 1908.

In a recent paper the writer* called attention to the fact that the invertase
of the date could not be extracted from the active tissues by solvents until
the fruit ripened. In other words the invertase changes suddenly from an
endo to an ektoenzyme.' The invertase of the green date thus corresponds
very closely, in its behavior towards solvents, to the invertase of certain
yeasts,* of immature PeniciUium,^ of Monilia Candida,^ and of various parts
of growing plants,' notably the rootlets of seedlings; to the urease in the
torula causing alkaline fermentation of urine;' to Buchner's* zymase of
beer yeast ; and finally to several zymogens, notably that of diastase which
occurs in the scutellum of ungerminated grain*® and which hydrolyzes
soluble starch only, the diastase of translocation of Brown and Morris. ^^
It probably corresponds also to the zymogen or proferment of the pro-
teolytic enzymes observed by Vines" in yeast, beans, and other plant
tissues, and possibly mistaken by him for vegetable ereptase. After pre-
senting the general facts relating to the insoluble invertase of the green date
the application of a new hypothesis will be made which seems to harmonize
these various phenomena without asstmiing differences in permeability of
the cell wall.

The action of green date tissue in inverting cane sugar, while no active
extract can be prepared, might be considered as due to the living cells,
tmder the influence of cane sugar, secreting invertase" or activating some

^ Read at the Chicago meeting of the American Chemical Society.
« Bot, Gat., 43, 393 (1907).

' M. Hahn, Z. Biol., 40, 172; see also Green, Soluble Ferments, Cambridge^
i899i P- 116; Oppenheimer, Die Fermente, Leipzig, 1903, p. 72.

* Beer yeast, O'Sullivan and Tompson, /. Chem. Soc., 57, 873 (1890); 61, 593
(1892); Fembach, Thesis, Paris, 1890, through Pottevin and Napias; Pottevin and
Napias, Compt. rend. soc. bioL, 50, 237; Milk sugar yeast, lactase, £. Fischer, Ber.,
27, 3481, 1894, etc.

• Penicilium glaucum, Bourquelot 1886, through Green, p. 115; P. duclauxi, Bull,
soc. mycol., 8, 147 (1891).

' Hansen, through Fischer and Lindner; E. Fischer and Lindner, Ber., 28, 3034
(1895); Buchner and Meisenheimer, Z. physiol. Chem., 40, 167; Fischer, Z. physiol,
Chem., 26, 77 (1898).

' Brown and Morris, /. Chem. Soc., 1893, 633; Brown and Heron, Proc. Roy. Soc,
1880, 393; J. O'Sullivan, /. Chem. Soc. Proc., 16, 61; Abst. CerUb., 1900, 773; /. Chem.
Soc., 77, 691, 1900.

• Sheridan Lea, /. Physiol., 1885, 136.

* Ber., 30, 117, mo, 2668; 31, 209, 568, 1084, 1090, 1531; 32, 127.
*• Reed, Ann. of Bot., 1904, 267.

" /. Chem. Soc., 1890, 458; Vines, Ann. of Bot., 1891, 409.
^^ Ann. of Bot., 16, 10; 17, 237, 597; 18, 289; 19, 149; 20, 113.
" Kffront, p. 69-70.

I006 A. E. VINSON.

zymogen, either soluble or insoluble, already existing in the cell; that is,
the inversion was due to some activity of the living protoplasm.* This
point, however, was readily answered in the negative by a short series
of tests in the presence of ordinary killing reagents. The inversion
curves for equal weights of green and of ripe date pulp were deter-
mined and found to be practically identical. Similar amounts of
green and of ripe pulp from the same sample as was used for the
blank, were allowed to act upon equal quantities of cane sugar under
the same conditions, excepting that i per cent, of chromic add, V,
per cent, of picric acid and 2 per cent, of formaldehyde were added to the
respective samples. If the inversion by the green date were due to living
protoplasm, its inverting power would be inhibited to a much greater degree
than that of the ripe date, where the invertase is known to exist in its solu-
ble form. In all cases there was considerable inhibition, but the curves for
both green and ripe pulp were approximately parallel and of the same order.
These results were fully confirmed by later experiments, designed to break
down the resistance of the plasmotic membrane, by treating the green pulp
with chloroform, ether, toluene, and other reagents. These killed the pro-
toplasm but left the inverting power of the tissues unaffected. The invert-
ing action of chromic acid under the same conditions was determined. The
following abridged table shows the effect in a general way of these proto-
plasmic poisons on the rate of inversion.

» Note. — ^Whatever our personal beliefs concerning the mechanism of biochemical
processes may be, it is not always an easy matter to differentiate clearly between
enzymatic and protoplasmic agencies. This is abtmdantly exemplified in the litera-
ture; thus Hugo Fischer (Cent, Bakt., 1903, 453) says:

''Unrichtig ist es aber darum auch, enzymatische Vorgange nun als 'rein chemische'
grundsgitlzich von denen zu trennen, die man nicht in zellenfreier Ldsung sich abspielcn
lassen kann; die Eiweissynthese ist gerade so gut ein chemischen Vorgang wie die
alkoholische Gariing. Denn, dass das Agens der letzteren sich in Wasser last, das der
ersteren nicht, ist zwar fflr die Laboratoritmistechnik wesentlich, viel weniger aber
fflr die theoretische Physiologic." Bokomy (PflUger's Arch., 90, Heft 11-12, autoabst.
Centb. Baki.) "Die wirkliche Bntscheidung iiber Protoplasma oder Enzymnatnr
konnte hier wie immer nur durch den Nachweis der bestimmten Organization oder des
Fehlens einer solchen hierbei gefflhrt werden. * * * * Die Ldslichkeit in Wasser
spricht gegen die Protoplasmanatur, da das Protoplasma nach den bisherigen
Beobachtungen nie ein wirkliche Ldsung darzustellen scheint. Doch ist auch be
einer Ldsung, wenn dieselbe eine, ' micellare" ist. Organization, das ist bestimmte
spedfische aneinanderreihung der Moleciile mdglich." M. Hahn (Ber., 31, 200) quotes
Neumeister (Lehrb. d. physid. Chem., 137): "Fflr die tierischen Zdlen nimmt Neu-
meister an dass die cellulare Verdauimg ohne Enzyme lediglich dtirch eine dgenartige
Thatigkeit des lebenden Protoplasma zu Stande kommt." Wills (Review of Buchner's
book, Die Zymasegarung, Centb. BakLy 1903, 464): "Die Tatsache, dass die beiden
Erscheinungen augenscheinlich durch dieselben Momente beinflusst werden, legt den
Gedanken nahe, dass auch die Reduktionswirktmg auf der Wirktmg enzymatige Kdrper
beruht, der fflr gewohnlich in der Leibessubstanz der zellen dngeschlossen ist."

THE ENDO- AND EKTOINVERTASE OF THE DATE. IOO7

Ei^FECT OP Protoplasmic Poisons on thb Ratb op Inversion op Cane Sugar by

Green and by Ripe Date Tissues.

No poison. Chromic acid. Picric acid. Formaldehyde.*

Time. Ripe. Green. Ripe. Green. Alone. Ripe. Green. Ripe. Green.

Nov. 27 +4.81 +6.02 +4.81 +5.41 +4.75 +4.72 +6.43 +4.65 +5.73

Nov. 28 +2.65 +1.63 +2.59 +1.09 +4.61 +5.12 +5.83 — 0.06 +0.93

Nov. 29 +0.10 — 0.56 +1.03 +0.07 +4.20 +3.61 +5.08 — 1. 00 — 0.67

Nov. 30 — 1.76 — 1. 81 — 0.66 — 0.85 +4.04 +3.32 +4.67 — 1. 81 — 1.36

Dec. I — 2.32 — 2.25 — 1.93 .. +3-95 +293+4-23 — 2.10 — 1.99

Dec. 3 — 2.07 — 2.24 — 1. 10 — 0.98 +4.14 +1.91 +2.81 — 2.01 — 1.92

Dec. 12 —2.45 — 2.44 —1.32 — 1.95 +4.13 — 1.29 —0.85 — 2.05 —1.99

Dec. 22 . . . . . . . . . . — 2 .04 — 1 .95

Since the in vertase becomes soluble at the time of ripening and the tannin
becomes insoluble at the same period, it would appear entirely possible
that, on crushing the green fruit, the soluble tannin, which is confined to
specific cells, would precipitate the invertase and prevent its extraction.
It has been shown by Brown and Morris* that a decoction of hops will not
extract diastase from malt until after the tannin has been removed by
treatment with hide powder. They find also that the diastase of the hop
strobile cannot be extracted with water and that those plants whose aque-
ous extracts show weak diastatic powers are rich in tannin.^ This is
especially true of Hydrocharis.^ Warcolier* belives the presence of ab-
normal quantities of starch in bruised apples to be due to the action of
escaped tannin on the diastase. These considerations led the writer to
investigate the relations between tannin and invertase in general, and
especially the possibility of the phenomenon of insoluble invertase in the
green date being due to soluble tannin. I have not been able to connect
the tannin with enzymic action, although its disappearance at the time the
invertase becomes soluble and its presence in those parts of plants where
metabolism is rapid, as in the tips of shoots like those of the rose, seem to
be suggestive.

The first point to determine is whether tannin in relatively large amounts
is in any way inimical to the action of invertase. The effect of added
tannin was studied both with date pulp and date extract. Twenty grams
of ground date were disintegrated with 100 cc. of water, varying amounts of
tannin added, then 500 cc. of sugar solution. The values in the following
tables are the readings given on the sugar scale of the saccharimeter, when
double the normal weight of the sugar solution was used. The variations
in initial reading are due to inversion before weighing off the quantities for
polarization.

* Trans. Inst, of Brewing, 6, 94 (1893).
' /. Chem. Soc., 63, 604, 640.

■ Ibid., 653.

* Compt. rend., 141, 405 (1905).

I008 A. E. VINSON.

Effect of Tannin on Rats of Inversion of Cane Sugar by Date Tissue.

No tannin. o.as gram. 0.50 gram. 0.75 gram. t gram.

• Initial reading +14.5 +14.2 +12.8 +13.2 +13.0

24 hours later — 4.0 — 4.0 — 4.7 — 4.7 — 4.9

The presence of added tannin seems to slightly increase the rate of inversion
rather than retard it.

The extracts used in the following table were made from 250 grams of
ground date with 400 cc. of water and with 200 cc. of water and 200 cc of
glycerol respectively. In the experiments 500 cc. of sugar solution were
treated with 50 cc. of date extract.

Effect of Tannin on the Rate of Inversion of Cane Sugar by Date Extract.

Water extract. Glycerol extract.

No tannin, x gm. tannin. No tannin, i gm. tannin.

Initial reading +14 -9 +14.2 +15.2 +14 -3

24 hours later — 5.2 — 5.5 — 5.0 — 6.0

The effect of tannin on the solubility of the enzyme, however, is a very
different matter, as shown by the references cited above. A short series
of experiments was planned to study this effect on the solubility of date
invertase in both water and glycerol. 250 grams of ground date (Amari
and Amhat mixed, one year old) were treated with 400 cc of water, with
and without the addition of 5.0 gmms of tannin, and also with 200 cc. each
of water and glycerol with and without tannin. 50 cc. of each of the ex-
tracts were then added to 500 cc. of sugar solution and the rate of inversion
determined as before.

Effect of Tannin on the Solubility of Date Invertase in Water and Glycerol

AS Determined by the R^vte of Inversion.

Water alone. Water + glycerol.

Date. without tannin. With tannin. Without tannin. With tannin.

June 21 +14.85 +15.8 +15.2 +152

June 22 — 5.2 +14.6 — 5.0 — 1.3

June25 —56 +9-5 —51 —5-5

The press residues appeared about as active as the original ground date.
The glycerol solution, being somewhat turbid, was allowed to settle several
days, the clear upper portion decanted off, and the solid residue separated
and washed in a small centrifugal. 50 cc. of the clear upper portion gave
a result almost identical with that obtained from 50 cc. of the unsettled
solution. The solid residue from the entire extract showed only a weak
inverting action.

This solubility of the taimin enzyme compotmd in glycerol, as compared
with its behavior toward water alone, seemed so unusual that I thought it
desirable to change the conditions somewhat. Accordingly two samples,
each containing 250 grams of the same date material and 5 grams of added
tannin, were treated with 200 cc. of water, the one receiving 200 cc, of

THE ENDO- AND EKTOINVERTASE OP THE DATE. IOO9

glycerol at once and the other receiving a similar quantity two days later.
Both were macerated 24 hours longer, pressed, and the turbid extracts
allowed to stand in long narrow tubes for two weeks. The sample which
had stood two days with the tannin before the glycerol was added, settled
out clear in a few hours ; but the other, which received the glycerol at once,
remained slightly cloudy, tending rather toward opalescence. Considerable
residue had separated out in both cases, from which the supernatant
solution was carefully siphoned. These solutions were then tested as before,
using 50 cc. of both the clear upper portion and the turbid lower portion,

Eppbct op the TiiiS OP Adding Glycerol on the Solubility op Dats Invertase

IN THE PRBSBNCS OP TanNIN.

Glycerol added at once. Glycerol added after two days.

* ' " » * 1*1. ^

Date. Upper portion. Lower portion. Upper portion. Lower portion.

July 16 +152 +14-3 +'5-9 +157

July 17 — 5.6 — 5.6 +11. 1 +4-5

July 18 — 5.65 — 6.0 +6.7 — 2.7

July 19 :.. +3-7 —4-3

July 24 — 5.0 — 6.0

These results indicate a marked solubility of the tannin-invertase com-
pound in glycerol, especially when the glycerol was added at once. If the
inversion had been due to a suspended insoluble invertase compound the
first two columns of the table would show a marked difference, such as is
apparent in the case where glycerol was added later. Prom these obser-
vations, we may conclude that if green date contained any soluble invertase
its extraction would not be seriously hindered by the tannin liberated when
the fruit was crushed, provided the glycerol was added at once. Numerous
samples of green dates from several different invert-sugar varieties, treated
in this way, have given negative results, although the press residues them-
selves were always very active, even after prolonged washing with water.
While this test in itself should sufGice to demonstrate that the invertase of
the green date is not prevented from going into solution by the soluble
tannin present, the distribution of the tannin in the fruit is such that it can
be dissected away readily and leave a considerable mass of taimin-free
tissue. For this purpose one of the larger varieties, such as Rhars or
Amraayah, serves best. The taimin in the ovulary of the date exists dur-
ing all stages of its growth in a layer of large tannin cells just beneath the
skin and easily visible to the naked eye, so that it can be pared away as
readily as the peel of any other fruit. After removing the seed, the fibrous
envelope, which also contains tannin, is easily scraped out. By carefully
comparing the inverting power of equal weights of pulp from the outer tan-
nin-bearing and the inner tannin-free portions of the same sample, no ap-
preciable difference could be detected. Extracts of the two portions were
invariably inactive. This shows, beyond doubt, that the insolubility of the

lOIO A. E. VINSON.

invertase in the green date is not the result of the presence of soluble tannin,
and the change in the solubility of the invertase and the disappearance of
soluble tannin are not directly connected, although they coincide very
closely in time.

It remains yet to show whether green date tissue, in the act of inverting
cane sugar, does or does not secrete invertase into the surrounding medium
as some yeasts are known to do, or whether a pre-existing soluble proinver-
tase, a zymogen, such as Pantanelli* has lately shown to exist in Mucor,
was rendered active. To disprove this, portions of tannin-bearing and of
tannin-free date tissue were allowed to act on cane sugar solution several
hours. After the sugar was inverted, the juice was pressed off and more
cane sugar added to the extract, but no further inversion took place even
after standing several da3rs. Thus no appreciable quantity of invertase
passes from the green date into the surrounding medium during the process
of inversion ; the inversion must take place within the cell.

This brings us face to face with the long-accepted theory governing all
these cases; namely, that it is a matter of the impermeability of the cell
wall to the enzjmie. This theory was first advanced in 1871 by Hoppe-
Seyler, ' who had observed that the invertase could be extracted from yeast
cells after treating them with alcohol, the assumption being that such treat-
ment broke down the resistance of the protoplasm to the passage of the
enz)rme. This same theory has worked well in a great number of cases and
has been used by Lea* in explaining the solubility of urease after treating
the torula of urine with alcohol; by Buchner* for the release of zymase
from yeast cells by grinding with sand and kieselguhr; by Albert* for the
same result by treatment with acetone ; by Pottevin and Napias* to ex-
plain the discrepancies between Pembacb and O'SuUivan with regard to
the liberation of invertase by various yeasts; and by many others, but
always to explain the same phenomenon of an enzyme not being yielded to
the solvent, until after some special treatment of the cells containing it.
Later I shall discuss some of these cases in detail.

For the purpose of this investigation we may consider the plant cell to be
made up of two layers. The outer cell wall is composed chiefly of hemi-
cellulose, true cellulose, lignin, and cutin, the lignin, according to Konig,*
increasing with the age of the cell and at the expense of true cellulose.
This outer wall acts much like a filter, allowing practically any substance
\n solution to pass. Within this outer cellulose wall and closely adherent

* Proinvertase reversibilita dell invertasi nei Mucor, Rend, Accad. Lincei, RomCt
15, 1 Sem., 587 (1906); Abst. in Bot, Centb., 105, 245 (1907).

' Ber., 1871, 810. Report of the Rostock meeting by Victor Meyer.
' Loc. cit.

* Ber,, 35, 2378 (1902).

* Landw. Ver. Sto., 65, 55, 65 (1907).

THE ENDO- AND EKTOINVERTASE OF THE DATE. lOI I

to it, probably secreting it, lies the outer protoplasmic layer, the plasmotic
membrane or Hautschicht.^ So long as the protoplasm is living, this
plasmotic membrane acts Uke a semipermeable membrane, allowing the
free passage of water but tenaciously holding most other substances. On
the death of the protoplasm the plasmotic membrane loses its semipermeable
properties and allows various substances to pass it freely; that is, the pro-
toplasm no longer plasmolyzes. It would seem then, that any treatment
which killed the protoplasm would liberate the enzyme, and that appears
to take place in many cases, although not in all.^

At this point we must take into consideration the work of Iscovesco* on
the passage of colloids through colloids as applied to cell specificity. Isco-
vesco has shown that one colloid may penetrate another when they bear
electric charges of the same sign, but not when the signs are opposite. The
sign may be changed by changing the chemical composition of the sur-
roimding medium. It is thus plausible that treatment with various agents
might change the sign of the charge borne either by the protoplasmic wall
or by the enzyme, and thus render the cell permeable.

PantanelU* also has recently studied the permeability of the cell wall by
observing the effects of colloids on the formation and excretion of invertase,
and by comparing the permeability of the wall to certain salts at the time
when enz)rme secretion was at its maximum. He holds that the secretion
of the enzyme into the surroimding medium is a function of the hving cell,
being made possible by autoregulation of the permeability of the proto-
plasmic cell wall, and that this change is reversible. Thus he finds the in-
vertase of yeast and of Mucars to be the only true ektoenzymes, since
in others the escape of the enzyme from the cell is conditioned by the djring
of the cell itself. Some colloids, as gum arable and peptone, in the culture
medium stop the intracellular formation and secretion of invertase. The
secretion is at a maximum during the time of maximum fermentation, and
at the same time the permeability of the cell wall to certain salts is also
greatest. In the case of Mncor stolonifer, the appearance of the invertase in
the surrounding medium is due rather to the enzyme being given off by the
dead tissue, because it corresponds in time with spore formation and is
wanting in the earlier stages of development, at which period cane sugar
as such is taken up. The general facts here correspond closely with those
for the date as well as for PenicUlvwm, the invertase passing into the solu-
tion readily at the time of maturity. In the case of Mucor there seems to
be no adequate ground for supposing that increased permeability of the
cell wall to inorganic salts would indicate increased permeability to inver-

* Jost, Pflanzenphysiologie, Vorl. 2.

' Buchner and Meisenheimer, Z. physiol. Ckem,^ 40, 167.
» Compt. rend. soc. biol., 62, 625.

* Ann. di Bot., 3, 113 (1905); Abst. in BoU Centb., 105, 185

IOI2 A. E. VINSON.

tase, although it might be true. There is, however, no ground to believe
that the invertase, prior to its time of secretion, did not exist in an insolu-
ble state. While I am not in a position to disprove impermeability in every
case, it certainly does not conform.to the facts for the green date.

It is significant that observations heretofore on the persistency with
which cells often retain the enz)ane have been made, with few exceptions,
on unicellular plants. In these cases the cell walls are more resisting than
those of soft complex tissues, and a process similar to Buchner's would be
required to break open any considerable number of them. The conditions
in the interior of a green date, however, are quite different. Here the
heavy cellular walls fotmd in the yeasts and other tmicellular plants aie
not needed for protection; therefore grinding and crushing by ordinary
means should allow the escape of at least detectable amounts of invertase.
Especially is this true, since the cells of a complex tissue are more or less
torn and cannot glide about freely among one another like yeast cells.
The juice pressed from such bruised tissues contains a very large percentage
of all the water present in the fruit, and with it, sugar, protein, and other
soluble cell constituents, which are otherwise held by the semipermeable
plasmotic membrane. Under these conditions it is hardly possible to be-
lieve that invertase exists in solution in the cell sap. The retention of a
great amount of invertase by the ripe date pulp which caimot be removed
by prolonged washing with water, a similar partial retention of invertase by
the rootlets of seedlings and of catalase by cured tobacco, also speak against
retention by impermeability.

Another fact which points to the insolubility of the invertase is the be-
havior of green date tissue toward such substances, as ether, chloroform,
toluene,* etc., with regard to the subsequent solubility of the enzyme. It
was mentioned above that, on the death of protoplasm, the plasmotic mem-
brane ceases to be semipermeable and allows the passage of most substances
as freely as does the cellulose wall. If the enzymes in general are in solution
in the cell sap, then by treatment with protoplasmic poisons they should
be released. The instances already dted of various enzymes being released
by treatment with alcohol, acetone, toluene, etc., might well be due to this
effect. Fischer and Lindner,' however, failed to liberate invertase from
Monilia Candida by this process. The green date also fails to give up its
invertase after killing with these substances.

Samples of ground green date, from which the inactive juice had been
largely removed by pressure, were treated over night in closed vessels with
acetone, ether, toluene, and chloroform, respectively. After exposing in
the air imtil all traces of the added material had disappeared, they were
macerated with water for two days and the aqueous extracts tested for

* E. Fischer, Z, physioL Chem., 26, 75 (1898).
' Ber., 28, 3034.

THE BNDO- AND EKTOINVERTASE OF THB DATE. IOI3

inverting power with negative results in all cases. The residues were ex-
tracted seven days longer with very frequent changes of water, and their
inverting power determined . It did not seem to be in any degree impaired .
A microscopic examination showed the protoplasm drawn away from the
cellulose wall and collected in small dense masses within the dead cell. A
second test on the inner tannin-free portion of Rhars date from which the
sugar had been extracted with water was made by digesting two weeks in
a large volume of ether. After decanting the ether, the pulp was dried in
the air and then soaked in water several days. The extract obtained in
this way was found to be inactive, but the residue showed very active
properties. An attempt was also made to liberate the invertase by heat.*
A portion of the inner tannin-free material was extracted to remove sugar
and insoluble materials, dried in vacuo over sulphuric acid and finally
heated to 145-150*^ C. in a hot air oven for 45 minutes. The activity of the
pulp was much impaired, but no active extracts could be prepared from it.
In all these cases the protoplasm must have been killed and more or less
broken down, so that the retention of the invertase could not have been
due to impermeability of the cell wall. Considering all these facts we find
any other explanation, than that the invertase is insoluble, very difficult to
maintain.

The acceptance of insoluble invertase is not so easy because the concept,
enzyme, carries with it the idea of solubility. This undoubtedly follows
from the use of the term enzyme synonymously with soluble ferment or
unorganized ferment, in contradistinction to organized ferment which in-
volves the idea of living matter and is necessarily insoluble.

Before further consideration of the subject it will be well to recall the con-
dition under which an enzymic action may take place. Barring tempera-
ture, alkaline or acid reaction, etc., the essential condition is to establish
molecular cdntact. There are four cases:

First, soluble enzyme and soluble substance to be acted upon; reaction
follows. Invertase and cane sugar.

Second, soluble enzyme and insoluble substance to be acted upon. Dias-
tase and starch.

Third, insoluble enzyme and soluble substance to be acted upon. Green
date invertase or artificially precipitated invertase and cane sugar.

Fourth, insoluble enzyme and insoluble substance to be acted upon;
no reaction. Diastase of ungerminated grain and starch granules, and prob-
ably many enzymes which have been recognized as existing in the zymogen
state.

Insoluble Compounds of Invertase.

That invertase in the green date exists in an insoluble condition but re-
tains all the catal)rtic properties of the soluble enzyme seems entirely possi-

* Buchner, Ber., 30, 117, 120.

IOI4 A. K. VINSON.

ble. Loew^ has shown catalase also to exist in two modifications. The in-
soluble catalase he believes to be a nucleoprotein compound of the soluble
form. The writer has already called attention to the fact that tannin
does not retard the action of date invertase, although in aqueous solutions
the tannin-invertase compound may be filtered off. The tannin-enzyme
precipitate, whatever its nature, retains its original catalytic properties.
A similar result was observed in the case of the lead subacetate precipitate
of date extract . A number of samples of date extract — cane sugar solution
mixture was prepared as follows : 40 cc. of glycerol extract of Birket ei
Haggi date were mixed with 1000 cc. of sugar solution, samples of 85 cc.
each were removed with a pipette, placed in 100 cc. flasks, i cc. of lead sub-
acetate added to each flask and the mixed contents diluted to 100 cc.
Half the samples were filtered at once and the remaining half allowed to
stand in contact with the precipitate. One sample from each of the sets
was polarized from time to time. Those which stood in contact with the
lead precipitate were inverted rapidly, while those from which the precipi-
tate had been removed bv filtration were unaffected. The retardation is
scarcely more than would be expected from the poisonous effect of the lead
on the enzyme as noted by Bokorny for metallic salts in general.

Ratb of Invbrsion op Cane Sugar by the Lead Subacetate Date Extract

Precipitate.

Hours. Filtered. Unfiltered.

Initial reading +13. 96 +13. 96

18 +14.02 +11.64

42 +14.02 + 9.62

90 +14.08 + 4.80

234 +13.70* — 2.22

330 +13.35 — 496

The chemical nature of the tannin and lead subacetate precipitates is not
known. They may be true chemical compounds with the enzyme itself,
such as are formed with the other protein matter, or the removal of the
enzyme may have been mechanical, due to adsorption. In either case the
enzyme was removed from the solution, but ceased its catalytic activities
only after the removal of the precipitate. It is not improbable that com-
pounds of the enzyme could be formed which, while insoluble, would retain
all the catalytic properties of the original, provided the substrat was soluble.
In other words, the chemical nature of the enzyme molecule as a whole
could be changed easily so as to affect its solubility without affecting that
portion in which the catalytic properties reside.

» Catalase, U. S. Dept, Agr. Report, 68, 1901.

' The filtered mixture did not remain sterile but became quite turbid after about
one week, hence the slight inversion after that time. Had the slight inversion noticed
been due to invertase not precipitated by the lead subacetate, its effect would have
been most marked in the first hours of action instead of first becoming noticeable
after ten days. Microscopic examination showed great ntxmbers of cocci^

THB ENDO- AND EKTOINVERTASE OF THE DATE. IOI5

Hedin* has observed that casein solution in 0.2 per cent, sodium carbon-
ate is digested by contact with tr)^sin adsorbed in charcoal. He finds
that the casein actually takes up trypsin from the charcoal and that the
trypsin can then be removed by filtration, retaining its proteolytic prop-
erties. This he believes to be due to a non-adsorbable compound of the
enzyme with the substrat. Such a compound does not seem to exist
between cane sugar and date invertase, because the lead subacetate re-
moves all the invertase, even in the presence of cane sugar. If such an
invertase-cane sugar compound does exist, it is carried down completely
with the lead subacetate precipitate.

Hedin* has also studied the selective adsorption of the enzymes and has
separated a- and j9-pro tease of ox spleen by adsorption in charcoal. Appli-
cation is then made of this principle to explain why sometimes more
enzyme is obtained by Buchner's process than by others. He suggests
that probably kieselguhr and perhapsalso the cell residues act adsorptively.
One might infer from this, that in the case of the green date, the inver-
tase was held back by simple adsorption. The retention of the enzyme
by the green cells, however, is too complete to admit of explanation in this
way. Furthermore, if adsorption by the cell residues were responsible,
we should not expect so profound a change in the adsorption powers at
the time of ripening. I am inclined to believe that the retention of the
enzyme as an endoenz^one up to the moment of ripening, has a deeper
physiological significance than to be merely accidental to our methods of
extraction. 1 do not contend that no adsorption takes place, for it
probably does ; but I do not believe that adsorption in the sense used by
Hedin could take place almost momentarily and completely.

Insolubility of the invertase during the green stages thus seems to be the
only theory which explains the actual facts observed. To determine the
mechanism of this insoluble state, however, will require much further in-
vestigation. We suggest, tentatively, two general ways in which it could
be accomplished by the plant. The enzyme may exist in combination with
some other vsubstance and move about freely in the cell sap without being in
solution. This hypothetical compound might take the form of dense, mi-
nute, or even ultramicroscopic particles, or it might be indefinitely expanded
until it occupies the same limits as the medium in which it is suspended,
similar to the way caseinogen in milk is supposed to be expanded by the
help of calcium phosphate. Along this line, but seemingly less probable,
is the recent theory held by Hofmeister and Jacoby.* They believe the
endoenzyme is localized in \'acuoles which are kept isolated by a thin layer
of colloid. I fully agree with this idea in so far as it concerns localization,

* Biochem. Jour., i, 484 (1906).
» Ibid., 2, 81.

• Oppenheimer, Die Fermente, p. 73.

IOl6 A. E. VINSON.

but rather in the manner to be described later under the second possibflity.
Such a compound would be under the direct control of the cell and could be
broken down, leaving the free enzyme in true colloidal solution and capable
of passing the protoplasmic layer, which, heretofore, had acted as a filter.
In this way there could exist simultaneously both soluble and insoluble
enzyme exhibiting the same properties.

The second possibility is that the enzyme is held in some insoluble com-
bination by the protoplasm. This theory was first proposed by E. Fis-
cher for the invertase of Monilia, and seems to the writer to be the most
tenable of all, after certain modification. As is well known the protoplasm
is exceedingly complex, histologically, being made up of innumerable fine
threads and granules. Certain of the granules may be the seat of one
enzyme and others of another. This accomplishes the same purpose as the
vacuoles suggested by Hofmeister and Jacoby. The enzyme-protoplasm,
compound which I have assumed to exist, whether ii be simple adsorption
or a loose chemical combination does not necessitate living protoplasm.
The writer therefore suggests the following theory for those cases where
endoenzymes exist.

The enzyme is in combination with some constituent of the protoplasm.
This combination modifies the enzyme molecule so as to render it insoluble,
without, however, affecting its catalytic properties, provided the substrat is
soluble. The enzjone-protoplasm compound may or may not break down
on the death of the protoplasm. The protoplasm may and usually does
liberate the enzyme about the time of maturity of the cell. Slow decom-
position, autolysis, or external chemical or physical influence may render
the enzyme soluble. The latter possibly act by destroying the integrity of
the cell, thus allowing more intimate mixing of the contents and the conse-
quent liberation of the enzjmie before its proper time.

Zymase.

. The theory outlined above seems to explain adequately the phenomena
observed with zymase. Buchner* advances and successfully answers three
objections which might possibly be raised against his theory of fermentation
without living yeast cells. Firstly, the fermentation may have been due to
bacteria or yeast cells. Secondly, the evolution of carbon dioxide may have
been due to some process other than alcoholic fermentation. Thirdly, the
fermentation may have been due to small pieces of living protoplasm pass-
ing through the filters.

The idea that zymase exists within the cell as an insoluble compound is
not at variance with the observed facts, nor does it invalidate the theory erf
cell-free or life-free fermentation. If Hans Buchner* had maintained that

' Lecture before Deutschen Chem. Gesell., Ber., 31, 568.
' MUnchener medic. Wochenschr., 1897, 322.

THE ENDO AND EKTOINVBRTASE OI^ THE DATE. IOI7

the fermentation was due to a ferment held in combination by the proto-
plasm instead of attributing it to particles of the Hving protoplasm, his
objections would not have been answered so easily. Microscopic examina-
tions have shown the entire protoplasmic content of many of the cells to be
forced out into the juice by Buchner's process. Thus his results with yeast
cells may be explained in two ways other than assuming soluble Z3anase to
exist within the unbroken cell. Some of the insoluble compound may have
been easily comminuted till it passed the finest filters. The fact that Buch-
ner sometimes observed 90 per cent, of the activity of his juice to be re*
moved by filtration, speaks for this theory. He attributed the fact to
adsorption. Other observers* have found the juice to be rendered inactive
by similar treatment. Another explanation of Buchner's results may be
that by destroying the integrity of the cell, some other substance, probably
a protease which has been held isolated by the living protoplasm, comes
into contact with the enz)mie-protoplasm compound and splits it up. The
fact that zymase is rapidly destroyed by a protease after it is extracted from
the cell and that it is shielded by adding another protein, such as serum
albumin , favors this view.' Furthermore the green date which yields no solu-
ble invertase when killed by acetone, chloroform, etc. , was found by Slade,
while working upon the identification of the enzymes of the date, to contain
no, at least only occasionally, traces of protease.

Vegetable Ereptase.

The work of Vines* on the vegetable protease requires a somewhat differ-
ent interpretation when we consider the probability of an insoluble protease.
All the phenomena that Vines attributes to vegetable ereptase in his
earlier papers can be explained quite as readily by assuming an insoluble
protease. As he states, at that time he had never gotten the peptic re-
action unaccompanied by the tryptic, but frequently the latter without
the former.* The peptic reaction is determined by the disappearance of
a clot of fibrin, an effect which could never be brought about by
an insoluble enzjmie. The ereptase reaction is determined by the
production of tryptophane from peptone, which could take place
easily with an insoluble enzyme. His main indication of the existence
of ereptase depends upon the difference in the time of appearance of the
two reactions.* A clot of fibrin is suspended over ground bean; no reac-
tion takes place till about the 9th day, after which the clot quickly disap-
pears. The same bean material gave the tryptophane reaction with pep-
tone at once. This is exactly what would occur if the bean contained an

* Stavenhagen, Ber., 30, 2422.
' Hardin, Ber., 36, 715.

* Lak. cU,

* Vines, Ann. of Bot,, 1905, 169, 175.
^ Ibid., 20, 113, 118 (1906).

IOl8 A. E. VINSON.

insoluble protease which became soluble later. Another indication he finds
in the different effect of sodium carbonate on the two reactions when veast
was used.* The tryptophane reaction took place readily with a small
amotmt of yeast and relatively large amount of sodium carbonate. On the
other hand, a fibrin clot was first appreciably attacked in the presence of
2 per cent, sodium carbonate when 20 per cent, of yeast was present. This
is to be expected if an insoluble protease were present which was slowly
passing into the solution or even if a soluble protease existed within and was
escaping slowly from the yeast cells. Even though the speed of reaction by
which sodium carbonate destroyed the protease were very great, some tryp-
tophane would be produced from the peptone, since the chances for molec-
ular contact are very great, because both substances are soluble. On the
other hand, contact between protease and fibrin clot would take place only
with the molecules on the surface of the latter, a relatively small number,
and furthermore the enzyme must travel through a layer of sodium carbon-
ate solution before reaching the fibrin. Thus only when protease molecules
in overwhelming number are present can there be any appreciable effect on
the fibrin clot.

Recently Vines* has succeeded, however, in isolating a purely7peptic
enzyme from hemp seed. This admittedly renders probable the
existence of vegetable ereptase, but does not answer the foregoing
objections to his former experiment?.

Summary.

The invertase of the date remains insoluble in all ordinary solvents
throughout its green stages but becomes readily soluble on ripening. The
change in the behavior of the invertase toward solvents coincides very
closely in point of time with the passage of the tannin into the insoluble
form. Tannin in relatively large amounts does not retard the action of
date invertase either in the extract or in the pulp. Soluble tannin, however,
hinders the Solution of date invertase in water but the invertase can be ex-
tracted by glycerol, provided the glycerol is added at the same time the
tannin is added. Green date invertase cannot be extracted by crushing
and macerating the green fruit with glycerol, therefore the invertase is not
rendered insoluble by the escape of soluble tannin on crushing the tarmin
cells. This conclusion is confirmed by the behavior of the invertase in the
tarmin-free portion of the date after the tannin-bearing tissues have been
completely removed. There is no direct connection between the change
in the state of the tannin and that of the invertase.

The inversion of cane sugar by the green date pulp is not accomplished
by the living protoplasm liberating a soluble ferment from a pre-existing
insoluble zymogen or by rendering active a proferment. The rate of inver-

' Ann, of Bot,, 289, 305 (1904).
' Ibid,, aa, Z03.

THE ENDO- AND EKTOINVERTASE OF THE DATE. IOI9

sion by equivalent amounts of green and ripe pulp of the same variety are
practically identical. Protoplasmic poisons — picric acid, chromic add,
and formaldehyde — retard the action of green and ripe pulp but to approxi-
mately the same degree. If the living protoplasm were in any way con-
nected with the inversion, the retardation would be greatly intensified in
the case of green date. Green date tissue is not rendered inactive by
soaking in ether, chloroform, acetone, etc.

The inner tissues of the green date are composed of relatively tender
walled cells, which, unlike those of yeast and unicellular plants, are fixed
and cannot glide about freely among themselves; consequently by ordinary
grinding, appreciable numbers are broken up and jrield their soluble con-
tents to water or glycerol. The press juices contain most of the water in
the fruit and large amounts of soluble substances which are usually retained
by the healthy and unbroken semipermeable plasmotic membrane. These
juices and extracts are invariably free from invertase until the fruit ripens,
although the press residues are always very active.

Treatment of the tannin-free green date tissue with chloroform, ether,
toluene and acetone causes the protoplasm to collect in granules but does
not alter the behavior of the invertase towards solvents. Moreover, the
invertase is not liberated by heat. This treatment should destroy the
semipermeable nature of the plasmotic membrane. From these observa-
tions, the theory of the impermeability of the cell wall for the invertase of
green date is untenable.

Enzymic action will take place whenever either enzyme or material to be
acted upon is soluble ; that is, molecular contact must be established . Tannin
removes the invertase of green dates from solution, but inversion is not
checked unless the precipitate is filtered off. Date extracts precipitated by
lead subacetate still invert cane sugar, but on f^moving the precipitate
inversion is stopped. It is then possible to invert sugar by means of in-
vertase artificially rendered insoluble.

In the place of impermeability of the cell wall to the enzyme, the writer
proposes the following theory. It is highly probable that green date in-
vertase and possibly other endoenzymes are held in an insoluble combina-
tion by some constituent of the protoplasm. In some cases this combina-
tion may be broken down and the enzyme pass into solution while the
protoplasm is living, but in others the combination may persist, even after
the death of the protoplasm. The enzyme may be rendered soluble also
by external chemical or physical influence. These probably act by destroy-
ing the integrity of the cell and allowing the contact of substances which
have been localized by the living protoplasm. On maturity of the tissues
the enzyme is generally liberated, possibly by autodigestion or other pro-
found change in the protoplasm.

In order to establish the impermeability of the cell wall to the enzyme in

I020 FRANK T. SHUTT AND A. T. CHARRON.

any given case it must be shown that the enzyme is in solution in the cell
sap and not held in combination by the protoplasm.

UNIV. OF AUZOlf A AGK. BZF. STA.,

Tucson, Akizona.

NOTE Oir THE DYER METHOD FOR THE DETERMmATION OF

PLANT FOOD Df SOILS.'

BT PS,ANK T. SHUTT and a. T. CHARROlf .

Received March 24, 1908.

During the past twenty years or so the problem of soil fertility has
received much attention from chemists, physicists and biologists, and
as a result there is to-day a clearer understanding as to what constitutes
productiveness in a soil — its several factors and their relative importance —
than there was a quarter of a century ago. We did not then recognize
that the chemical data formed but one factor in soil diagnosis ; that texture,
moisture-holding capacity, bacterial life, drainage, precipitation, etc,
must all be considered before drawing any conclusion as to a soil's prob-
able productiveness. It was this neglect that caused doubt on all sides
to be entertained as to the practical value of soil analysis, and indeed the
agricultural chemist himself had well-nigh abandoned all hope of being
able to interpret his own data for the benefit of the farmer. Though we
are yet far from having soil diagnosis on a thoroughly satisfactory basis,
a distinct advance has been made and the writers feel that one of the most
prominent and valuable features of this diagnosis lies in the determination
and recognition of the soil's store of more or less available plant food.

In so far as the soil's crop-producing power or the need of special fertiliza-
tion can be determined by chemical means, the estimation of the plant
food present in a condition more or less available for crop use must be
a matter of considerable importance. The stores of insoluble, inert plant
food, no matter how large, can be of but little value to the growing crop;
it is rather those small percentages of potash, phosphoric acid, etc., that
are at once capable of being utilized by plants that serve to measure the
soil's immediate ability to sustain vegetable life. In these two classes
we have represented what might be termed latent and potential fertility,
though, of course, no strong line of demarkation can be drawn between
them — ^the former being always converted into the latter, gradually but
with varying rapidity, according to a number of conditions which we
need not now discuss.

In 1894 ^^' Bernard Dyer, working on soils from the Rothamsted
Experiment Station, the history of which as regards manuring and crop
yields was well known, proposed the use of a i per cent, solution of
citric acid as a solvent for the available phosphoric acid and potash.

* Read at the Chicago meeting of the American Chemical Society.

DYER METHOD FOR PLANT FOOD IN SOILS. I02I

His analytical data were well in accord with the field results and from the
percentages obtained he was able to establish minimum and maximum
limits regarding the necessity and non-necessity of phosphatic and potassic
fertilizers. Further, he found that the sap of the rootlets of a number
of farm plants had an average acidity approximately equivalent to that
of I per cent, citric acid solution, thereby in a large measure confirming
the correctness oi his deductions respecting the value of the solvent
proposed.*

Since that date, various solvents have been proposed, more especially
in the United States. Several of these were experimented with in the
laboratories of the Dominion Experimental Farms, Ottawa, but as they
appeared to lack that evidence which had been brought forward to sup-
port the Dyer method, it has been the latter process that has always been
employed in our official soil investigations. Dyer's limits have not been
found to answer invariably, but this is only what might have been ex-
pected; the general nature of the soil must be taken into account, its
geological origin and the climatic conditions of the district considered.
On the whole, however, the method and proposed limits have proved
very satisfactory. Possibly it is only a tentative m/ethod, but the fact
remains that as data accumulate from workers in different parts of the
world, its claims to represent the amounts of more or less readily assimi-
lable mineral plant food in the soil receive corroboration.

Within the past two months the writers, in working on certain virgin
prairie soils from the Northwestern provinces of the Dominion, have
thought it desirable to ascertain how far the results might be affected by
certain modifications of the original method as described by Dyer. These
modifications were, in the reduction of the time during which the solvent
was in contact with the soil and in the reduction of the volume of solvent
used per imit of air-dried soil, to obviate the necessity of evaporating
large volumes of solution. Dyer's method calls for an extraction period
of seven days, during the first six of which the mixture is kept constantly
agitated, looo cc. of solvent being used per loo grams of soil.

The soil sample was of a composite character, made up of 8 samples
collected (at Tisdale, Sask.) on as many dates between June 15th and
October 20, 1907. It represented the native prairie soil to a depth of 8
inches. The soil had been *' broken" in 1906, and this season (1907) bore
a crop of wheat. It might be described as a rich, black loam in which
sand predominated and typical of large areas in the Canadian North-
western wheat belt. Analysis showed it to contain large percentages of
semi-decayed vegetable matter and nitrogen, a characteristic feature of
these prairie soils.

The analysis of the air-dried soil, according to the method of the A.
* J. Chem. Soc, 65, 1 15-167 (1894).

I022 PRANK T. SHUTT AND A. T. CHARRON.

O. A. C, using 1.115 sp. g*"- hydrochloric acid as the solvent, afforded
the following data :

Per cent.

Moisture 6 . 26

Organic and volatile matter i3-34

Insoluble matter, clay, sand, etc 68. 49

Oxide of iron and alumina 8.68

Lime (CaO) i .04

Magnesia (MgO) o. 86

Potash (K,0) o 58

Phosphoric acid (P^OJ o. 19

Carbonic acid, etc. (undetermined) o . 56

100.00

Nitrogen, in organic matter o. 45

From these results we may conclude that as regards the stores of latent
plant food, the soil is one. very well supplied with nitrogen and potash;
that phosphoric acid is present in fair amounts and that lime exists in
quantities sufficient for crop needs.

In the subjoined tabular statement the results obtained by the modifica-
tions of the Dyer process already referred to, are presented, the per-
centages of lime being given in addition to those of phosphoric acid and
potash. All the data are averages from at least two closely concordant
determinations:

Influence op Time of Digestion.
(Solvent: 1% Citric Add Solution.)

Phosphoric add Potash I«ime

(PjOft). (K,0). (CaO).

Per cent. Per cent. Per cent.

100 grams soil in 1000 cc. 7 days 0.02287 0.03*818* 0.5320

100 grams soil in 1000 cc. 5 hours 0.01807 0.03958 0.5210

100 grams soil in 500 cc. 7 days 0.01999 o 03355 0.2718

100 grams soil in 500 cc. 5 hours 0.01599 0.03089 o. 2285

Comparing first the data from the analyses made when using 1000 cc
of the solvent, it will be noticed that reducing the period of digestion from
7 days to 5 hours materially decreased the amount of phosphoric acid
dissolved, but did not similarly affect the potash — the percentages of
the latter being practically identical for both periods. The lime falls off
but slightly with the shorter digestion.

Employing 500 cc. as the volume of solvent, the phosphoric acid per-
centages present practically the same differences as those already noted—
in other words, slightly higher amounts were obtained from the longer
digestion. Five hours' digestion gave a somewhat lower percentage of
potash than 7 days, but the difference is almost within the limit of experi-
mental error. The lime dissolved is markedly less for the shorter period.

OBSERVATIONS ON THE ASSAY O^ TKlrLURIDE ORES. IO23

Inpi^usncb op Volumb of SolvSnt.
(Solvent: 1% Gtric Add Solution.)

Phosphoric acid Potash Lime

(PjOb). (K,0). (CaO).

Per cent. Per cent. Per cent.

100 grams soil in 1000 cc. 7 days 0.02287 0.03818 0.5320

100 grams soil in 500 cc. 7 days 0.01999 0.03355 0.2718

100 grams soil in 1000 cc. 5 hours 0.01807 0.03958 0.5210

100 grams soil in 500 cc. 5 hours 0.01599 0.03089 0.2285

In this table are given the same data as in the preceding, but they are
arranged to more readily show the influence of the volume of solvent. No
detailed discussion of these data is necessary; a glance at the figures is
sufficient to make apparent the only conclusion that can be reached, viz.,
that reducing the volume of solvent used, materially decreases the per-
centages of phosphoric acid and potash obtained. In the case of lime, the
smaller volume extracted but one-half that taken out by the larger volume.
Clearly the influence of volume of solvent is decidedly greater than that of
the period of extraction.

No doubt soils of a different character would not yield results that
would fall exactly into line with those here recorded, that is, as to the
effect of varying the period of extraction and the volume of the solvent on
the amount of mineral matter dissolved, but the general trend would, we
believe, be the same. It seems highly desirable that further data on this
important question of available plant food should be obtained from widely
different points and correlated with the field results. These analytical
data should be, as far as is practicable, strictly comparable, and to this
end it is evidently necessary that the details of the process as regards
period of extraction and volume of solvent, should be carried out alike by
all workers. There seems to be every reason for adherence, in these re-
spects, to the time and volume as given by Dr. Dyer in his original account
of the process.

Dominion Bxpbrimbntal Farm,
Ottawa, Canada.

SOME OBSERVATIONS ON THE ASSAY OF TELLURIDE ORES.

By Gborob Borrowman.
Received April 4, 1908.

Much has been written concerning the assaying of telluride gold and
silver ores, yet the literature affords striking contradictions as well as
statements imsupported by experimental data. However, there seems
to be agreement in the opinion that tellurium is the cause of serious
irregularities. Mr. A. L. Davis, in Tech. QiiarU, Vol. XII, sums up the
situation as follows: "As to the percentage of loss sustained in work,
whether by scorification or crucible method, many experiments carried
out upon the foregoing lines indicate to me that nothing definite can be

I024 GEORGE BORROWMAN.

laid down in regard to it. Every ore, every slag, every scorificatioD,
every cupel, let alone the temperature at which the assay is carried on,
has some efiFect upon the loss and these make too many unknown quan-
tities to arrive at any conclusion."

It was the purpose of the writer in the work here recorded to ascertain
a few facts concerning the efiFect of tellurium in the crucible assay for
gold. The points investigated were: loss in the slag; loss in cupellation;
the efiFect of variation in temperature of fusion.

The ore selected for the experiments was a high-grade telluride gathered
from various parts of Colorado, rich in silver, gold and tellurium. With
the sulphuric acid test it gave a very strong indication of the latter element
and analysis showed the presence of 10.5 per cent, chiefly as hessite and
sylvanite. The gangue was silicious, consisting of quartz, feldspar and a
little calcite. The ore had a reducing power of about 1.5.

The material was ground very fine, first through 120 mesh sieve, then
in agate till all passed through bolting-cloth. In all the work i/io A. T.
was taken for each assay, the samples being weighed out on a quantitative
balance instead of on the usual pulp scales. All fusions were made in a
muffle at about 1200® excepting in Series No. 4; all the fluxes were passed
through a 40-mesh sieve mixed with the ore till thoroughly homogeneous.
In short, the greatest care was taken to eliminate all variations except
the one studied. ,

Experiments were first conducted to learn the influence of tellurium
in carrying gold into the slag. Mr. C. H. Fulton, in This Journal, 20,
586, records data to show that slag losses in telluride fusions are much
greater than in ordinary gold ore crucible work. Messrs. Hillebrand and
Allen, BvUetin No. 253, U. 5. Geological Survey, find losses insignificant,
and commenting on the work of Mr. Fulton suggest that his losses might
have been due to the use of iron nails in the fusions.

In the investigation of this point a series of assays was first made on
the raw ore, using the following charge :

Ore i/io A. T.

Litharge 90 g.

Silica 10 g.

Argols 2}g.

Sodium bicarbonate 30 g.

Salt cover.

Another series was then made on ore from which the tellurium had
been removed. In removing it use was made of the fact that nitric acid
dissolves tellurium and silver in a telluride and leaves the gold in the
residue. Each sample was boiled with nitric acid (1.27) till no further
action was apparent ; the residue was washed and dried and fluxed tmder
the same conditions as maintained in the first series. Enough silver

OBSERVATIONS ON THE ASSAY OF TELLURIDB ORES. IO25

was added to the de-tellurized series to replace what had been removed
by the acid. The slags from each series were then carefully ground and
assayed, using the following charge :

Slag

Litharge 30 g.

Argols 2 g.

Salt cover
The buttons from both the ores and slags were cupelled. The results

are set down in the following tables:

Sbribs No. I.

Ore.

No.

Mg. gold found.

" —

Oz.

gold per ton.

Slag.
Gold found.

16.14

161. 4

None

16.18

161. 8

( c

16.18

161. 8

< (

16.23

162.3

< (

Av. .

16.18

161. 8

Series No. 2.

Ore.

SUg.
Gold found.

No.

Mg. gold found.

Oi.

gold per ton.

16.10

161.00

Trace

16.16

161.60

None

16.20

162.0

Trace

16. II

161. 10

None

Av 16.14 161. 4

From the above data it is obvious that the influence of tellurium in
carrying gold into the slag is very slight, if there is any; the amount
recovered was in no case weighable.

The effect of tellurium in cupellation was next investigated, and in-
cidentally the functions of litharge in the fusion. On the assumption
that the tellurium in a crucible charge is oxidized by the litharge .and
carried into the slag, a series of charges was made up varying this con-
stitutent from 90 to 180 grams. Theoretically the amount of tellurium
carried into the button as a lead alloy should be in inverse ratio to the
amotmt of litharge in the fusion. A series of four samples was made in
duplicate. The buttons of one series were dissolved in nitric acid and the
tellurium determined by precipitating it in hydrochloric acid solution with
sulphur dioxide. The other series was cupelled for gold. The two
duplicates are considered as one series, i. e., No. 3.

Series No. 3.

No.

Grams.
PbO in charge.

Mg.
Te in button.

gold fonnd.

Oz.
gold per ton.

lost

16.10

161.00

3
4

120
150
180

287.5
298.9

176.8

16.24
16.18
16.15

162.24
161.80
161.50

Average

... 16.17

161.70

I026 GEORGE BORROWMAN.

All of the beads of the above series were frosted, showing the presence
of tellurium, notwithstanding the great excess of litharge in the fusion.
It would then seem that litharge in reasonable amount is inadequate to
slag off all the tellurium in this ore. The loss of gold, however, was in-
considerable, the average of the series varying but o.oi mg. from the
average of Series No. i in which the tellurium had been removed previous
to fusion. These results are at variance with published statements re-
garding tellurium in cupellation. Messrs. T. J. Eager and W. W. Welch,
however, found that no loss occurred in the presence of the element up to,
lo per cent. The amounts in the above experiment are many times that.
A bead from an assay, duplicate of No. i in the above series which was
also the same as those of Series No. 2, was tested quantitatively for
tellurium and 6.9 mg. were found ; so that it may be present even in the
bead in considerable amount without there being more than a negligible
loss of gold.

The work was concluded with some experiments to learn the effect of
variation of temperature of fusion on gold recovered and the amount of
tellurium carried into the button. Some assayers recommend an ex-
tremely high temperature. Mr. R. W. Lodge, however, in his "Notes on
Assaying,'* states that he believes high temperatures breakup some tel-
lurium compounds in the slag with a consequent alloying of the element
with the lead.

A series of four was run in duplicate consisting of charges made up
exactly as those of Series No. 2. The temperature was varied between
800° and 1600°. As in the previous experiment the buttons from one set
were cupelled for gold, the other dissolved in nitric acid and assayed for
tellurium.

Series No. 4.

No. Temperature. Wt. of buUons. Gold found. Gold per ton. Tellurium found.

• Mg. 0«. Mg.

800° C.

30.5

16. 10

161.00

258.2

icxw^'C.

32.5

16. 16

161.60

150.0

1250® C.

32.8

16.20

162.00

195 -6

1600° C.
raee

32.8

16.16

161.60

215.9

Ave]

16.16

161.60

The irregularities in the amounts of telliuium could not be accounted
for, yet it seems improbable that the amount varies directly with the
temperature. As to the gold the yield is greatest at about 1200°, though
the average of the series is but slightly below that of Series No. i,

Summary.

The foregoing data seem to warrant the conclusion that tellurium as
the cause of irregularities in crucible work has been overestimated. Slag
losses are no greater than in ordinary gold ores; the element may be
present in the lead button in relatively large amounts with no consider-

THE DETERIORATION OF COAL. I027

able percentage of loss. The presence of tellurium in the bead does not
necessarily imply a loss of gold in the cupellation though, of course, a
frosted bead would not be permissible when silver is to be determined.
In high-grade tellurides when silver is to be estimated, the writer suggests
a preliminary treatment with nitric acid, with subsequent precipitation
of the silver as chloride which may be dried and added with the residue
from the acid treatment to the fluxes in the crucible. Variation in
temperature of fusion does not seem to be of great moment though the
data above are most favorable to a temperature of about 1200°.

The average obtained from Series No. i is higher than that obtained
in the others showing a loss due to tellurium in fusion and cupellation,
yet the variation is small, the average of No. i being not more than
0.24 per cent, higher than the lowest average, that of Series No. 2. The
members of the various series differ among themselves in some cases
considerably, but perhaps not more than would be expected in any high-
grade ore, owing to lack of homogeneity of sample. In the opinion of the
writer, irregularities in high-grade tellurides are due more to this lack of
homogeneity than to tellurium. It is conceivable that in some ores the
ratio of gold to tellurium might be much less and hence the percentage
of loss greater. In such cases a preliminary treatment with nitric acid
to remove the tellurium would obviate the difficulty.

Unxvbrsztt op Nbbraska,
Lincoln, Nbb.

THE DETERIORATION OF COAL.

By S. W. Parr and W. P. Wheblbr.
Received April 7, 1908.

In cooperation with the State Geological Survey and the Engineering
Experiment Station of The University ot Illinois, certain facts have de-
veloped which bear directly upon the behavior of coal. They are of con-
siderable moment and should be taken into account in any study of this
material. The first pertains to a deterioration which cannot be ascribed
to weathering processes, but rather to the simple fact of the release of
the material from the conditions which have surrounded it in the seam.
This has been recognized in a rather indefinite way from time to time,
but without data to substantiate the fact.^

The following items are given in support of this theory of loss. In the
summer of 1900 twenty-nine samples of coal were collected at the face
of the vein, quartered in the usual manner, placed in galvanized iron
cans with screw cap and tire -tape seal, exactly as described by the Coal
Testing Plant of the United States Geological Survey.^ They were

* Tms Journal, 28, 650 (1906).

» Bulletin No. 261, of the Coal Testing Plant, U. S. G. 5.

I028

S. W. PARR AND W. F. WHBBLHR.

shipped directly to the laboratory, where they were transferred at oaoe to
one-quart jars of the "Lightning" or Putnam type, the coal being suflBdent
to practically fill the jars. This transfer was made with as much as posa-
ble of the original moistiu-e retained in the coal. The ** Lightning" jars
were chosen because from extended experience with sodium peroxide, this
jar was fotmd to be the only container having a perfect seaL Twenty-
one other samples were collected and sealed in the ordinary Mason jars.

After standing in the laboratory for about ten months, twenty-six of
the "Lightning" jars, upon opening, showed a slight pressure of gas
which ignited with a strong blue flame, burning from one-half to six
inches in height above the top of the jar. Upon covering and retesting,
these jars would reignite for two or three successive times. None of the
samples sealed in the Mason jars would so ignite.

Two points are to be noted here, namely, that the coal content very
nearly filled the jar and that the enclosure in the Putnam jar is practically
that of a continuous glass seal, while the Mason jar is quite different as
to the security with which the gasket is held and, in addition, has a large
metallic surface exposed to the transmission of gases.

Another test pertains to the enclosing of the ai^-dried samples of the
same coal in Putnam jars for more than eighteen months with the dry
coal occupying about one-quarter of the jar. Upon opening, all of these
jars showed a very positive evidence of the absorption of oxygen as
indicated by the extinguishing of a lighted match. Analysis showed
the presence of less than 1.5 per cent, oxygen and less than 2 per cent, of
carbon dioxide.

Tablb No. I. — Loss m Calorific Valub during Transit.

B.T.U. of ash.

Test.

water and sni-

B.TU.

No.

Locality.

Sisc of coal.

When sampled.

phnr-frec coaL

Lo^

Westville

li inch screenings

same day as mined

14684

1 (

< 1 II

7 days after mining

14627

Springfield

II it

same day as mined

14478

i 1

it 1 1

4 days after mining

14351

127

Herrin

i) inch screenings

same day as mined

14658

f <

II II

6 days after mining

14553

105

Westville

3 inch nut

same day as mined

14768

f (

3 " "

7 days after mining

14586

182

Springfield

3 " ••

same day as mined

14655

1 (

3 " "

4 days after mining

14461

194

Herrin

3 " "

same day as mined

1 475 1

1 1

3 " ••

6 days after mining

. 14682

Another series of results related to the calorimetric determination of
freshly mined coal and similar determinations made upon the coal after
shipment in the cars in the ordinary manner. Six cars of coal were
sampled at the mine while in the process of loading and calorimetric

THE DETERIORATION OF COAL. IO29

determinations made on the samples so collected. After arrival at the
University, the cars were sampled from the wagons as the cars were im-
loaded.

In the table above, the results are tabulated and attention is called
to the fact that in each case there is a uniform drop in fuel values as be-
tween the freshly mined coal and that which had been subjected to trans-
portation conditions.

Another series of results is given in Table No. 2, which is a comparison
of values obtained at the St. Louis Coal Testing Plant of the United
States Geological Survey, and samples from the same or near-by Illinois
mines as determined in this laboratory. All conditions of operation were
duplicated as nearly as possible, including the type of calorimeter, which
was of the Mahler-Atwater design. Still, the results here, when reduced
to the "ash, water and sulphur-free basis," were imiformly lower than
those obtained at St. Louis. The only explanation seemed to lie in the
fact that the samples here were held in laboratory storage longer than
was the case with the St. Louis samples, our heat values being determined
on the average after about 10 months of such storage. The extremes of
difference lie between 1.6 per cent, and 3 per cent., but it is quite suflS-
cient in amount to be a distm-bing factor in basing conclusions on the
behavior of coals of this type.

Tablb No. 2. — Comparison of U. S. G. S. with III. Gbol. Surv. Values.

U.S.G.S.
No.

111. G. s.
Lab. No.

I«ocality.

B.T.U.
per lb. ash,
water and sul-
phur-free coal.

Differ.

ence

B.T.U.

Percent,
of dif.
ference
in B.T.U.

lU.

O'Fallon

14567

95,96,97

" (a)

14214*

353

—2.4

111.

Marion

14561

330

" («)

14335

226

—1.6

111.

Staunton

14615

91,92,93,94(0)

" («)

13933*

682

—4-7

in.

West Frankfort

14647

364

" (6)

14332

315

— 2.2

111.

E. of Springfield

14464

81,82

" (a)

14020*

444

—3-1

ni.

Centralia

14587

167,168,169

" (b)

14257*

330

—2.3

Til.

Herrin

14558

323,325

" (a)

14267*

291

— 2.0

lU.

La Salle

14722

393

" (6)

14440

283

—1.9

(a) Samples not from same mine, but from adjacent mines.

(6) Samples from the same mine.

In further

testing this matter of age, fresh

samples were collected by

us for

comparison with our

own samples of

10 months'

Standing, and

* Average

of several samples

from neighboring mines.

1030 S. W. PARR AND W. F. WHEELER.

calorimetric determinations made as before. Table No. 3 gives the
results of this comparison, which showed that these new values are uni-
formly higher upon the fresh samples than upon the old. The variations
lie between 1.3 per cent, and 3.4 per cent.

Table No. 3. — Comparison of Values for Fresh and old Samples by Illinois

Geological Survey.

B T.U. DiflTer-

in. G. s.

Lab. No.

Locality.

per lb. ash,
water and sul-
phur-free coal.

ence

in
B.T.U.

Per cent, of

difference

in B.T.U.

421

Du Quoin, fresh

14386

307,308,309

old (a)

14009*

377

—2.6

460 and 1088

Herrin, fresh

14647

323,325

* ' old(a)

14285*

362

—2-5

460

Clifford, fresh

14615

325

old (6)

14213

402

2.7

462

Marion, fresh

14781

330

. " old(6)

14335

446

540,740,741

Springfield, fresh

14468*

81,82

old (a)

14020*

448

—3.1

557

Westville, fresh

14550

332

old (6)

14054

496

3-4

558

Himrod, fresh

14564

333

old (6)

14087

477

3-3

Eldorado, fresh

14857

317

old(6)

14597

278

1-9

358

old (a)

14662

195

1-3

1114

Harrisburg, fresh

H93I

315

old (5)

14622

309

— 2.1

3 miles E. of Eldorado, fresh

15131

359

old(6)

14939

192

1.3

1119

Mary\'ille, fresh

14450

418

old (6)

14134

316

— 2.2

1121

Norris City, fresh

14658

316

old(6)

14322

336

2.3

(a) Samples not from same mine, but from adjacent mines.
(6) Samples from the same mine.

A third comparison was also made as between freshly collected and
determined samples here and those made by the St. Louis Fuel Testing
Plant; the results are given in Table No. 4. Here the differences are
equally distributed between those of a plus and minus character. To
determine, if possible, what effect the time element might have on these
variations, the two extremes were selected and by correspondence, with
the chemist of the Geological Surv'^ey at St. Louis, it was found that
^ Average of s^v^ral samples from neighboring mines.

THE DETERIORATION OF COAL.

IO3I

Illinois No. 7, with an extreme variation of -f 1.7 per cent., showed its
calorimetric value to have been determined upon the sample twenty
days after collection at the face of the vein. Our own sample, which,
as nearly as we could determine, was analyzed ten days after collection,
showed a higher value as indicated. The other extreme, Illinois No. 9,
with a drop in our results of — 2.4 per cent., was found by the records
of the Fuel Testing Plant, to have been but six days old. The exact age
of our own sample is not definitely known, but it was not less than from
ten to fifteen days old. The gther variations are small and might be
accounted for on other grounds than that of age of the sample.

Table No. 4. — Comparison op New U. S. G. S. Samples witn New Samples of the

III. State Geol. Surv.

U. S. G. S. 111. 9. G. S.
Lab. No.

111.

No.
3

B.T.U.
per lb. ash. Per ceat. of

water and sul- Difference difference
phur-free coal, in B.T.U. in B.T.U.

111. 4*

III.

111.

111. 14

111. 16

Locality.
Marion

Troy

" (6)
Collinsviile

" (a)
Staunton

" (a)
K. of Springfield
" (a)
Herrin

- (a)
Zeigler

- (6)

(a) Samples not from same mine, but from adjacent mines.

(b) Samples from the same mine.

In Table No. 5, the average differences as between old and fresh samples
by the two laboratories is assembled, and does not need explanation.

Table No. 5. — Averages op Tables Nos. 2, 3 and 4.

17 Illinois Geological Survey samples compared with 8 United
States Geological Survey samples Average 365 B. T. U.

Illinois Geological Survey samples analyzed 6 months to x or

year after collection ; United States Geological Survey analyses 2.5 per cent, lower,
made soon after collection.

111.

462

1118

723,724,725

735,736,737

540,740,741

459,460,1088

419,420

1 456 1

1 478 1

4-220

+ 1.5

14439

I4I68

— 271

—1.9

14373

I462I'

+248

+ 1.7

I46I5

14260'

—355

—2.4

14464

14468*

+ 4

+0.03

14558

14647'

+ 89

+0.6

1 4601

14463

-138

—0.9

17 Illinois Geological Sur\xy samples analyzed 6 months to i Average 356 B. T. U.
year after collection, compared with 16 similar samples analyzed or

within two weeks after collection. 2.4 per cent, lower.

16 Illinois Geological Survey samples analyzed within 2 weeks Average 29 B. T. U.
after collection, compared with 7 United States Geological Sur- or

vey samples analyzed soon after collection. 0.2 per c^t. lower.

^ Average of several samples from neighboring mines^

1032 S. W. PARR AND W. F. WHEELER.

It seems evident from these results that the drop in values occurs within
the first two or three weeks after the coal is broken out of the seam, but
the rapidity and extent as to any given length of time, is not easily deter-
mined and can be inferred only from such data as are found in the tables.
Presumably, it is due in the main part to exudation of combustible gases
consequent upon the releasing of the coal from conditions of pressure
and sealing in the vein.

' A second series of facts relates to a secondary process which undoubtedly
begins after the breaking out of the cgal from the seam, namely, the
oxidation of compounds in the coal. This is, perhaps, more properly
designated under the name of weathering. Advantage has been taken
of the fact that in a number of mines, old pillars have been standing,
and samples have been procured from these, after properly cleaning the
surface, and comparing the results with samples obtained at the freshly
worked faces of the mine. In Table No. 6 the results are given for such
samples from pillars twenty-two and twenty-seven years old. The pillar
coal shows a loss in comparison with the fresh coal of approximately 2.5
per cent. In the same table are given also the results for samples from
another mine which have been subjected to various conditions, including
submerging for one year, and the analysis of coal which has been exposed
to the weather for one year. The difference is inappreciable as between
the outer surface of the pile and that of the interior.

Table No. 6.

Six other weathering tests have been conducted on smaller samples
by Mr. N. D. Hamilton. Each sample was subjected to different con-
ditions, namely, submerged; exposed to the weather; exposed to a dry

Test No.

Material.
Belleville, Illinois

B.T.U. per
lb. referred

to aah,

water and

sulphur-free

Drop in heat
units com-
pared with
initial
values.

Fresh face sample

14785

Pillar coal, 22 years exposure
Equality, Illinois

14372

413

Fresh face sample

15188

Pillar coal, 27 years exposure
Westville, Illinois

14754

434

li inch screenings i week from mine

14627

3 inch nut, i week from mine

14586

i} inch screenings, submerged

I week

after mining, for i year

•14588

From surface of 15 ton pile, i

year

ex-

■

posure

14241

347

From throughout 15 ton pile, i

year

ex-

posure

14264

324

Four weeks after mining

14410

178

THE DETERIORATION OF COAL.

1033

atmosphere, at a rather elevated temperattire ; and a duplicate of the
latter with frequent drenchings with water. A charting of the results,
which is more or less characteristic of all the tests, is given in Fig. i,

Oakwood Nut 'and Slack.

SAMPLto One Day after Mining.

■

15000

r i

•

—

0
0

•^N

'%^

z 14000

•^^

—

'x,

•^

■*"

.^K-

' ^

III

"" 1

•

Outdoor exposure.

flO

ATBSTOlZO'r DRY.

— —At 85>o I20T wcTTEO orrtH

13000

^%A

»DI

•IL.

►r\^

1 .

0 1 234 5 6789 10

Time of Exposure- Months.

Fxo. z.

which is sufficiently clear to be self-explanatory. While the results of
this series of tests are not conclusive, they point to the fact that sub-
merged coal is without loss so far as oxidation processes are concerned;
that exposure to a dry atmosphere is quite as conducive to the loss of heat
values as exposure to weather and that, in general, these calorific losses
are largely overestimated and probably, on an average, do not exceed
3 or 4 per cent, in amount.

A continuation of these tests upon carload lots is now being made under
conditions of outdoor exposure, housing in bins, and in the submerged
conditions, with some accompanying experiments intended to develop,
if possible, the conditions which result in spontaneous combustion.

UNIVBR8ITT OF iLLZMIOSf

.URBANA, III.

I034 GSORGB O. ADAMS AND ALPRED W. KIMBALL.

STUDIES oir DIRECT nesslerizahon of kjeldahl digestates

m SEWAGE ANALYSIS.

Bt Gborob O. Adams and Alfred W. Kimball.
Received April lo, 1908.

Of late much has been written concerning the. great advantage of
direct readings of Kjeldahl digestates over the distillation method.
The greater accuracy of the method, the more simple technique, the
kss bulky and bothersome apparatus, and the great saving of time are
among the chief advantages claimed for the. method. To test the truth
of these claims, a series of parallel determinations were made at the
Lawrence Experiment Station under the direction of Mr. H. W. ClarL

Previous Methods. — Several procedures have previously been advanced,
each one claiming to obtain satisfactory results under local conditions.
Kimberly and Roberts* determine the total imoxidized nitrogen by
adding nitrogen-free sulphuric acid to a measured amount of sewage
and digesting until colorless. The digestate is transferred to a 50 cc.
flask| cooledi and made up to the mark. Twenty-five cc. of this mixture
are transferred to a 100 cc. flask, a 25 per cent, caustic soda solution
added nearly to neutralization, cooled and more caustic soda added
until a flocculent precipitate appears, when 2 cc. of a 10 per cent.
sodium carbonate solution are added to precipitate the calcium present.
The whole is then made up to 100 cc, shaken thoroughly, and allowed
to stand six hours, when a portion of the supernatant liquid is pipetted
into a 50 cc. Nessler tube, made up to 50 cc, nesslerized and read. They
found it necessary to use caustic soda f rea from organic matter because
otherwise turbid tubes were obtained.

Whipple' altered the procedure by diluting the digestate to 250 cc,
treating an aliquot portion of this solution with an equal amount of 5
per cent, caustic soda solution, and substituting filtering through filter
paper washed free from ammonia in place of the long period of settling.
He found ordinary "purified stick" caustic soda caused no trouble
from turbid tubes.

The Lawrence sewage is a strong domestic sewage low in calcium and
magnesium content. It was therefore not necessary to consider the
calcium in our experiments so no sodium carbonate was used. The
method of Whipple was tried but did not give satisfactory results. The
amount of free sulphuric acid present in the Kjeldahl digestate varies
so that it is impossible to take an aliquot portion after dilution, add an
equal amount of 5 per cent, caustic soda solution, and have an^n^herc
near the same excess of caustic soda present in the different digestates.

' Kimberly and Roberts, Jour. Infect. Dis., 1906, 2, p. 109.
* Whipple, Tech. Quart., 1907, 162.

DIRECT NKSSLERIZATION OF KJELDAHIv DIGESTATES. IO35

As a result of this difficulty a large number of turbid tubes were ob-
tained, due to the presence of too large an excess of caustic soda.

Method.

To overcome this difficulty the following method was devised:
Reagents. — ^The same as those used in ordinary Kjeldahl determinations
of organic nitrogen, with the addition of a 5 per cent, solution of caustic
soda.

Method. — Fifty cc. of the sample (or more if the nitrogen content is
low) are put in a Kjeldahl flask, diluted sodium carbonate added, and
boiled down to about 20 cc. to remove the free ammonia. Then 5 cc.
of nitrogen-free sulphuric acid (1.84) is added and the sample digested
until colorless. The digestate is transferred to a 250 cc. flask, diluted
to about 100 cc., and a 50 per cent, caustic soda solution added almost
to neutralization. After cooling, a 5 per cent, solution of caustic soda
is added in slight excess, the sample made up to 250 cc. and mixed
thoroughly. This is filtered through a filter paper washed free from
ammonia, 10 cc. of the filtrate are pipetted into a Nessler tube, made up
to 50 cc. with ammonia-free water, mixed by shaking, nesslerized, and
read after fifteen minutes.

Discussion.

Turbidity. — ^The Lawrence sewage being very low in calcium and
magnesium content, the chief difficulty experienced from turbid tubes
was due to the presence of too great an excess of caustic soda in the
neutralized digestate. All tubes having an excess of less than 0.05 gram
caustic soda in 50 cc. gave turbid tubes due to incomplete precipitation
of magnesium. Most tubes having an excess greater than 0.20 gram in
50 cc. were also turbid. The excess of caustic soda must therefore be
between 0.05 and 0.20 gram per tube to obtain clear tubes. Kimberly
and Roberts say that if potassium permanganate is used to complete the
digestion, turbid tubes will be obtained. Since the use of that salt is not
necessary for completing the digestion of ordinary sewages, it is not used
at this laboratory. In this laboratory a high-grade commercial caustic
soda costing six cents per pound is used for all work and no difficulty was
experienced from turbid tubes unless the excess of caustic was outside the
limits above mentioned. The neutralization of the whole digestate
rather than a small portion thereof reduces somewhat the correction
due to the blank and the smaller portion taken for nesslerization allows
of a greater excess of caustic in the whole digestate without causing
turbid tubes.

Accuracy. — Of 90 parallel determinations, 65 per cent, of the direct
determinations are lower, 32 per cent, are higher, and 3 per cent, the
same as the distilled determinations.

1036 GEORGE O. ADAMS AND ALFRED W. KIMBALL.

Comparison op Rssults by Distillation and Dirbct Methods.

Kjeldahl Nitrogen (Organic), Parts per 100,000.

Con- Sewmge
Regular sewage. Settled aewage. Septic sewage, tact and +50%

, t , * s * « * trickling city

Unfilt. Pilt Unfilt. Pilt. Unfilt. Pilt. effla. water.

Numberof samples. .. . 13 11 13 13 65 24 5

Av. distilled 1.27 0.56 0.97 0.59 0.83 0.46 0.52 0.90

Av. direct 1.29 0.58 0.93 0.57 0.88 0.46 0.47 0.82

Max. difference +0.22 +0.22 — 0.30 +0.15 +0.10 +0.11 — 0.19 --0.13

Min. difference . — 0,01 +0.02 +0.01 0.00 0.00 — 0.03 0.00 — 0.02

Av. difference +0.02 +0.02 — 0.04 — 0.02 +0.05 0.00 — 0.05 — 0.08

Av. % difference 1.6 3.6 4.1 3.4 6.2 0.0 9.6 8.9

Kimberly and Roberts, from 24 determinations, obtained results by
the direct method, 41 per cent, of which were lower than the distilled,
33 per cent, higher, and 16 per cent, the same as the distilled determina-
tion.

Whipple in the same manner by the direct method, obtained 41 per
cent, lower, 6 per cent, higher, and 53 per cent, the same as by the distilla-
tion method. Prom this it would seem as if the results obtained by the
direct process are as a rule a little lower than they should be.

Time Necessary. — ^The chief advantage advanced for the direct method
is the great saving of time accomplished by its use. That is to say, it
can be done in less time by transferring the digestate to a flask, making
up to a definite volume, mixing thoroughly, transferring an aliquot
portion of this mixture to another flask, adding caustic soda almost to
neutralization, cooling, then adding an excess of caustic soda, mixing
thoroughly and either (i) allowing to stand for several hours, or (2)
filtering through filter paper which must be washed free from ammonia,
than by adding an excess of caustic soda and distilling two tubes.

In this laboratory there are eight stills available. It was found that if
there were only four determinations to be made, the time required to
distil the digestates was practically the same as by the direct method,
but if there were more than four determinations to be made, by using the
eight stills they could be done approximately twice as fast as by the
direct method.

Furthermore, when the direct method is used, one's entire time. must
be directed to the digestates, whereas while distilling, something else
may be done while the digestates are being distilled. This still further
reduces the actual time of the distillation method as compared to the
direct.

Apparatus and Technique. — ^While the direct method does away with
the use of stills, on the other hand it makes necessary an increased hand-
ling of graduated flasks which are bothersome and bulky. It also calls
for a large amount of nitrogen-free water which is a disadvantage, as

STUDIES OF INCUBATION TESTS. IO37

nitrogen-free water is not always available in large quantities in a sewage
laboratory. In this laboratory two gallons and a half of nitrogen-free
water can be made in about four and one-half hours. Starting with the
water in the flasks cold, eight Kjeldahls can be distilled in about fifteen
minutes. Figuring on this basis the cost of distilled water necessary
for the direct method by Kimberly and Roberts' procedure is about one-
fifth greater than by the distillation process, and by Whipple's procedure,
about three times as great. If many determinations are to be irade by
the latter method the saving in the cost of water used would in a short
time pay the cost of a still.

In the direct method the digestate has to be made up to volume at
least twice and a definite amotmt measured twice, while in the distillation
method but one measurement is necessary. The chance for error in
manipulation is therefore four times as great by the direct as by the
distillation method.

Conclusions.

The direct Kjeldahl method undoubtedly has its own place in sewage
work, but it does not seem as if it should take the place of the distillation
method in a permanent sewage laboratory handling many samples be-
cause of :

1. The greater amotmt of bothersome and bulky apparatus necessary.

2. The large amount of nitrogen-free water required.

3. The greater chance for error in manipulation.

4. The necessity of having the excess of caustic within narrow limits
to avoid turbidity, this practically requiring a rough titration of each
determination.

5. The greater length of time required for the determination.

The method, however, is without doubt an excellent substitute for
the distillation method in a temporary laboratory where it is necessary
to incur the least possible expense for apparatus or in a small laboratory
where but a very few determinations are to be made daily.

Laboratory of the Lawrrnce Bxpbrimbnt Station,

Lawrbncb. Mass.

STUDIES OF INCUBATION TESTS.

By H. W. Clark and Gborob O. Adams.
Received April 10, 1908.

For the past seven years incubation tests have been made in the labora-
tories at the I^wrence Experiment Station to determine the stability of
the effluents of contact and trickling filters. These studies have shown
that the development of odor is, perhaps, the surest proof of putrescibility.
Oxygen consumed and oxygen dissolved tests before and after incubation
are of value but are sometimes contradictory. The so-called methylene-

1038 H« W. CIvARK AND GEORGE O. ADAMS.

blue test has been used to a considerable extent, but studies have shown
that according to this test, some samples are putrescible, which, judging
from the odor test only, are stable. In following out the work with this
methylene-blue test, quite a number of dyes have been experimented with
to find, if possible, one which would decolorize only when putrefaction as
determined by odor, had also taken place. The dyes tested in this way
were indigo carmine, methylene green, congo red, alizarin blue, add
brown, alkali blue, add violet, aniline yellow, curcumein, ponceau red,
martins yellow, methyl orange, tropaeolin, coccinine, toluidine blue and
azo blue. Of these dyes, indigo carmine and methylene green are more
readily decolorized than methylene blue, as will be shown later. Congo
red, methyl orange and tropaeolin were the only others affected, but were
too stable to use as an indicator of putrescibility in an incubation test.
The results with indigo carmine and methylene green, however, looked
promising enough to study.

During these studies, a test for hydrogen sulphide in the sample incu-
bated, was made in connection with the other putrescibility tests. The
test used is based on the formation of methylene blue from dimethyl-
^-diamino-benzene sulphate in the presence of hydrochloric add, hydrogen
sulphide and ferric chloride, and by this method approximately o.oi part
of hydrogen sulphide per hundred thousand can readily be determined.

Samples of the effluents tested were incubated at 80^ F. with methykne
blue, methylene green and indigo carmine in bottles with tightly fitting
stoppers, a blank bdng incubated at the same time. In each case enough
dye was added to the sample tested to give it a dedded color. The amount
of dye used, however, within reasonable limits is not important, since
samples with twice the amount of dye usually used were decolorized in the
same period of time. Hydrogen sulphide was determined before and after
incubation and the time required to decolorize the three dyes by incuba-
tion of the samples noted.

The procedure followed for the determination of hydrogen sulphide was
as follows: Standards were made up in loo-cc. Nessler tubes from hydro-
gen sulphide water, the strength of which was determined by titration
with iodine, and 2 cc. of strong hydrochloric add containing Va P^^ ^°^
ferric chloride and i cc. of a i per cent, solution of dimethyl-/>-diamino-
benzene sulphate. With small amotmts of sulphide about thirty min-
utes are required for a good color to develop. Reagents were added to
the proper amount of the samples to be tested, these samples being, also,
in loo-cc. Nessler tubes and the colors developed in the samples compared
with the standard colors. It was foimd that the same set of standards
might be used for several weeks without change, and it was also found
that equal amounts of reagents must be added to standards and to samples
in order that the same shade of bliie be obtained.

STUDIES OF INCUBATION TESTS.

1039

The following table shows the results of incubation of nineteen samples
which did not putrefy. In these samples hydrogen sulphide did not form
or odor develop, and with two exceptions, none of the three dyes was
decolorized. Following this is a table showing the results of incubation of
samples, all of which did putrefy.

Table Showing Results of Incubation of Non-Putresciblb Samples.

Number Hydrogen sulphide.
Effluent of Farts per ick),ooo.
of daysincu- ■

Filter No. bated. At start.

233
248

175

235

233

235

251^

136*

248

251

175

135

233

235
248

235
248
176

175

5
5
5
5
6

6
6
6
5

5
5
6

6
6

5
6

o.oi

O.OI

0.03

0.00

O.OI
O.OI
O.OI

0,01

O.OI

0.00

0.04
0.00

O.OI

0.00

O.OI

0.00

O.OI
O.OI
O.OI

After
incubation.

O.OI

0.00
0.00
0.00

O.OI

0.00

O.OI

0.00

O.OI

0.00

O.OI

0.00

O.OI

0.00
0.00
0.00

O.OI
O.OI

0.00

Methylene blue.
Methylene green.
Indigo carmine.

Not decolorized

Appearance,
odor, etc.

No change

Decolorized in four

days
Decolorized in four

days
Not decolorized

Very slight in-
crease in odor
Very slight in-
crease in odor
No change

((

«i

The results of the incubation of the 26 samples which may be said to
have been putrescible are shown in the table on the n.xt page.

Examining this second table, it will be noticed that in every case if one
dye was decolorized the other two were also, and that the average time
required for indigo carmine to be decolorized was 2 days; for methylene
green, 2V2 days and for the methylene blue, nearly 4 days. In every
sample tested after incubation there was an increase in the amount of
hydrogen sulphide present, but no close relation could be noted between
the amount of hydrogen sulphide formed and the appearance and odor
after incubation.

It is probable that the hydrogen sulphide is largely formed from the
decomposition of albuminous corpoimds in the effluents tested. Several
samples shown in the above table were, as designated in the table,
incubated with the addition of one part sulphur, as potassium sulphate,

* One part sulphur as potassium sulphate added before incubation, had no effect.

1040

H. W. CLARK AND GEORGE O. ADAMS.

Tablb Showing Results of Incubation of Putrsscibls Samplss.

Days required to decolorize

Effluent 4 »

of Methylene.

filter Indiso . »

No. carmine. Green. Blue.

•

Hydrogen

sulphide.

Start. Parts

per 100,000.

After incubating.

Hydrogen sul-
phide. Parts
Days, per 100,000.

1
Appearance.

Odor.

176

0.03

0.40

sl. black

str.

176

O.OI

0.05
0.10

• ■ ■ •
« ■ • •

d.
d.

221

black

str.

221

black

str.

221

black

str.

221

0.03

0.20

black

str.

221

0.02

0.28

black

str.

221

0.05

0.25

blark

str.

233

• • ■ •

233

■ • ■ •

233

■ 3

O.OI

0.05

• • • ■

247

black

str.

247

black

str.

247

black

str.

247

black

str.

247

0.02

0.80

sl. black

str.

247

}

O.OI

0.06
0.30

• ■ • •

black

str.

247

}

O.OT

3
5

0 25
0.25

black
black

str.
str.

247

O.OI

I 50

black

str.

247

0.02

0.35

black

str.

247

}

black

str.

248

• • • ■

251

O.OI

0.15

• • • •

str.

251

O.OI

0.20

• • ■ •

251

O.OI

4
5

0.08
0.22

• • ■ ■

• • • ■

d.
d.

251

Not de-
colorized

O.OI

0.40

« « ■ •

and these showed no increase in the hydrogen sulphide formed above
that formed in duplicate samples without the addition of the sulphate;
on the other hand, a i per cent, solution of peptone in distilled water,
seeded with one drop of sewage, developed 2 . 5 parts of hydrogen sulphide
by five days' incubation. However, if putrefaction occurring is great,
inorganic sulphates may be reduced as happened in the following experi-
ment: Twelve bottles were filled with sewage. To six of them i part
of sulphur as potassium sulphate, was added and incubated for one,two,
three, four, seven and eight days, respectively. The hydrogen sulphide
formed is shown in the following table, and it will be noticed that after
three days' incubation there was a rapid development of hydrogen
sulphide in the samples to which potassium sulphate had been added.

NOTE. I04I

Hydrogen Sulphide.

(Parts per 100,000).

Days incubated.

Sewage only.

Sewage + sulphate,

Start

0.00

0.04

O.IO

0.30

0.35

1.20

O.IO

2.50

O.IO

2.50

One or two samples of filter effluents decolorized all of the dyes in four days
without the production of hydrogen sulphide or odor, and, on the other
hand, one sample showed a considerable development of hydrogen
sulphide and odor without decolorizing in 7 days. These two or three
results were, of course, abnormal and simply show that absolute reliance
cannot be placed on incubation results obtained by the methylene-
blue test; in fact, these studies have shown (i) that the degree of put-
resdbility of such effluents as experimented with can probably be better
estimated by odor and appearance after incubation than by the time
required to decolorize dyes; (2) the hydrogen sulphide formed comes
very largely from albuminous compoimds in the effluents and the
amount formed is, to some degree, a measure of the putrescibility of
the sample tested; (3) on the whole, it would seem that if a pu-
trescibility test of the methylene blue kind is to be adopted, equally
good results can be obtained in a shorter time by the use of indigo car-
mine or methylene green.

IfABOKATORY OP THB I^AWRBNCB EXPXRIMBVT STATIOK,

Lawrbmcb, Mass.

NOTE.

The Quantitative Determination of Arsenic by the Gutzeit Method. — In
the issue of Chemical A hstracts for April 10, 1908, p. 976, is an abstract of a
note by T. F. Harvey on the estimation of arsenic by the Gutzeit test.
As this immediately follows the abstract of an article by Sanger and
Black on the quantitative determination of arsenic by the Gutzeit
method, the casual reader may be led to infer that Sanger and Black
were anticipated by Harvey in the method published by them.

I have already called the attention of the editor of the Journal of the
Society of Chemical Industry to the misleading nature of Harvey's
article, and Mr. Harvey himself has assured me that it is quite clear to
him that his work had not come to our notice. The Harvey method,
however, is merely a quantitative treatment of the ordinary Gutzeit test,
while the paper of Saneer and Black not only introduces a different

1042 NKW BOOKS.

principle into the procedure, but also includes a detailed study of the
conditions of the reaction . Chari^BS Robert Sanger.

HA1.VARD UNIVBRSITY, CAMBRIDOB, MASS.,

April 30, X908.

NEW BOOKS.

Lehrbuch der Gerichtlichen Chemie. Baumbrt, Dbnnstbdt und VoigtlAnder*
In zwei Banden. Zweite ganzlich uragearbeitete Auflage. 8**-xvi, 490. Braun-
schweig. F. Vieweg und Sohn. 1907. Price, 12 Marks, bound 13 Marks.

The first volume of the new edition of this manual by Dr. Baumert
of the University of Halle is devoted to the detection and determination
of poisons and noxious substances in the cadavet and excretions, in foods
and beverages, household articles, water, air and soil and to chemico-legal
problems in general. Volume II will be written by Drs. Dennstedt and
Voigtlander, of Hamburg, and will be confined to the methods for the
examination of inks, writings, signatures, forgeries, etc., and to the ex-
amination of blood, blood and spermatic stains and materials of a similar
nature.

Dr. Baumert is entitled to the thanks and gratitude of analysts for
having placed in their hands a manual of legal chemistry truly worthy of
the name. Although a book of only 490 pages, it is a marvel of com-
pactness and thoroughness. A reader, glancing over the table of con-
tents, would be apt to form the opinion that the treatment, in general,
must be incomplete, elementary and unsatisfactory, but upon careful
study it becomes apparent that this is not true and that we have here one
of those rare cases where an author has been able to do justice to his sub-
ject in remarkably few words, and that contrary to the verbosity of so
many German writers we have in this book an exceptionally terse style.

While it is evident that the manual has been written to meet the needs
of German chemists, the discussions are of such a nature and the reference
to legal points and practice of such a character that it may be consulted
with profit by all experts. At the present time this little book is unique
in its field, being much more than a manual of determinative toxicology.

The author confines himself strictly to the chemistry of the materials
discussed, all questions involving physiological eflTects, etc., being avoided
so far as possible on the ground that such questions are not legitimately
those of the chemist but rather of the medical expert, and that when the
chemist has reported that in his judgment a substance is or is not present
his work is done. Any subsequent questions as to whether the material
found caused death, or could have caused death or was present contrary
to law are not within the province of the chemico-legal expert.

The introduction is devoted to a very brief statement of fundamental
facts relating to poisons and noxious substances, much space being

NEW BOOKS. 1043

saved by avoiding any extended discussion, and by quoting from
Robert's "Kompendium." Following the introduction, three chapters,
forty-nine pages in all, are devoted to the further presentation of general
information for the guidance of analysts wishing to qualify as chemico-
legal experts. For the beginner these pages are invaluable and even the
experienced expert may read the suggestions of the author with profit,
although the legal requirements of the German Empire are not deviated
from. The suggestions made and facts here presented relate to such
important topics as:

Chapter I. General rules to be followed in chemico-legal examinations;
preliminary tests, planning the methods of analysis and the subsequent
drawing up of reports and the statement of opinions; the laws govern-
ing the fees of experts in the German Empire. Chapter II. Poisonous
materials found in foods, beverages, household articles, toys, etc., and
finally in Chapter III, the author de^ribes the methods for the testing
and purification of reagents and apparatus in greater detail and more
thoroughly than in any other manual of like scope with which the re-
viewer is familiar.

Part II, comprising the bulk of the volume, is devoted to the properties,
methods of separation, identification and determination of such sub-
stances as the chemical expert may be called upon to search for. The
common poisons are treated at length and as a rule in each case the methods
given are many and varied, and the references to the original articles
full, complete and down to date. Cross reference to other parts of the
manual greatly facilitate the work of the analyst. If any comment may
be made it is that possibly too great a choice is given, but this, on the
whole, can scarcely be called an adverse criticism.

In the chapters devoted to inorganic substances in addition to the
discussion of poisons, the legal questions arising concerning precious
metals, their alloys, jewelry and counterfeit money are taken up.

An exceedingly valuable feature adopted by the author is to discuss in
separate paragraphs under each substance treated, the materials in which
the substance is to be found, the choice of methods to be employed and the
general questions the chemist is called upon to answer. In these respects
this manual is far in advance of any other. The methods of presentation
and the nature of the information given may best be described by illustra-
tions taken at random, for example. Silver — ^uses in the arts — ^forms met
with in commerce — separation of silver from other elements — ^identifica-
tion of silver — ^identification of silver compounds — determination of
silver — the examination of hair, textiles, papers, etc., for silver — ^the
recognition of silver stains — hair dyes, pharmaceutical silver prepara-
tions. Phosphorus — ^uses in arts — ^forms in 'commerce — poisoning by
phosphorus — distribution in the cadaver — detection of phosphorus —

I044 ^^^ BOOKS.

examination of the urine for phosphorus — identification of phosphorus
compounds — determination of free phosphorus — analysb of commercial
phosphorus — ^identification of phosphorous add — ^phosphorus-containing
compounds of commerce, matches, and methods for the study of match-
making material and of the finished products with reference to the ques-
tions arising under the laws of the German Empire. Or take chapters of
great interest at the present time to many American chemists, those
devoted to sulphur dioxide and its salts, and to alcohol, here the sub-
heads may be summarized as — Sulphur dioxide — ^general properties, iden-
tification, recognition in the air, examinations of foods for sulphites, deter-
mination of sulphites in wines, beer, etc., their detection in flesh, in fats —
the identification and determination of sulphur dioxide in plants, the in-
vestigation and effects of flue gases, smelter fumes, etc. — and — Alcohol —
properties, detection — determination — alcoholic beverages, analytical meth-
ods of German revenue service, the recognition of denatured and renatured
alcohol — tables giving percentage composition of alcohols, etc. — detec-
tion of methyl alcohol in beverages, etc., amyl alcohol, etc., etc. It will
be seen from these illustrations that the book is much more than a manual
of chemical toxicology.

The chapters devoted to alkaloids, glucosides and other substances of
vegetable origin are complete, down to date, and so well arranged that
after glancing over the book one is able to find at once just the information
one wishes both as to separation methods and identity tests. The color
reactions are all tabulated and so arranged as to render consultation easy.
Here again the analyst is given a variety of methods with the opinion of
the author as to the choice under given conditions, an excellent system
of cross references being introduced to aid in comparing the reactions
given by different substances.

An excellent index covering both author's names and subject matter
completes the book.

An appendix is devoted to such of the laws of the German Empire as
the expert chemist must be familiar with and with a few tests and
methods of investigation inadvertently omitted in the text. The book
is so well written and the methods otherwise so judiciously chosen that
it is a matter of surprise that the author makes so little use of the micro-
scope, an instrument absolutely indispensable in chemico-legal examina-
tions. E. M. Chamot.

Benedikt-Ulzer, Analyse der Fette und Wachsarten. Fifth, revised edition, by Ferdi-
nand Ui^zBR, P. Pastrovich and a. Eisbnstbin. Large octavo, xiii + 1168
pages, 113 figures in text. Berlin: Julius Springer. 1908. Price, M. 26.50.

The first edition of Benedikt's AnsAyse der Fette appeared in 1886.
Out of it have grown two monumental works which serve as the standard
guides to the analysis of fats and waxes in the English and German hn-

NEW BOOEB. 1045

guages. Since 1886 the progress in fat analysis and in the fats and oils
industries has been astonishing, and the book tmder consideration bears
striking evidences of this fact. In 1886 many of the physical and chemical
methods for the examination of fats had been worked out in essentially the
same form in which they are applied to-day. For example, the method of
determining "titer," practically dates from Rtidorff's work in 1856; Reich-
ert's value and Hehner's value date from 1879; Merz's acid number from
1880; Hubl's number and K5ttstorfer's number from 1884. On the other
hand, since 1886, have developed such important factors in this field as the
acetyl number (1887) ; Hehner's method for the determination of glycerol
(1889); Twitchell's method for the determination of rosin (1891); Wolf-
bauer*s work on the "titer" test (1894) ; Twitchell's method of saponifica-
tion (1898); Connheim's ferment method of saponification (1902); and
many others. The edition of 1886 of Benedikt's work was scarcely one-
fourth the size of the volume under consideration. Benedikt died in 1897,
just before the third edition was published, and it speaks well for his care
and foresight that the general plan of presentation has been followed from
the first edition to the last. It is a most difficult matter to bring an old
book up to date, but this has been done admirably in the present instance.
The subject matter has been divided into two main divisions, the first (550
pages) devoted to the general analysis of fats and waxes and examination
of technical products of the fat industries, the second (591 pages) covering
the natural fats and waxes and their examination. The first part is written
in collaboration with P. Pastrovitch, Director of the Oleomargarine, Candle
and Soap Works, "Salvator" in Vienna; the second with A. Eisenstein,
assistant in the Technological Industrial Museum in Vienna. The com-
ponents of fats and waxes, chemical and ph3rsical properties, determination
of ph3rsical constants and the qualitative and quantitative analysis of these
substances and their impurities and unsaponifiable constituents and the by-
products of the manufactures into which they enter, are treated at length.
Methods of chemical control receive considerable attention. While the
work is not primarily designed as a treatise on the technology of fats, the
industries based on fats and waxes are entered into to a considerable extent.
The descriptions of individual oils, fats and waxes are complete and ade-
quate. Throughout the book, errors and misprints are very few. Foot-
note references are especially complete and the absence of footnote com-
ments makes reading easy. The indexes are good and the typography,
printing and paper all that could be desired. The work still stands as the
best in German on the analysis of fats and waxes. W. D. Richardson.

Detection of the Common Food Adulterants. By Edwin M. Brucs. New York: D. Van
Nostrand Co. 1907. Cloth, i2mo. vii + 84 pp.

This little book has been prepared by the author as a simple qualitative

manual for food inspectors, and for teachers and students of chemistry.

1046 NEW BOOKS.

The qtialitative tests presented, comprise, with a few additions, the
principal ones given in Bulletin No. 65, of the Bureau of Chemistry, U. S.
Department of Agriculture, and in Leach's *'Food Inspection and
Analysis." Frequently, the descriptive language has been condensed.
While brief introductory notes are given stating the principal adulterants
to be found in the-several classes of foods treated in the respective chapters,
there are few cautionary and explanatory notes — a defect in a text for
beginners in this field of applied chemistry. At a few points, the work is
not up to date. Thus, in speaking of the doubledyeing test, on page 35,
the author states that ''nothing but coal-tar wiU color in this second
dyeing;*' whereas, it is now well known that lichen dyes also possess this
power. The chapter on honey does not mention the newer tests for in-
vert sugar. The book will doubtless, however, serve well the purpose for
which it was written. There is a good general index and an index of
tests bv authors, so that convenience of reference is secured.

Wm. Frbar.

Medico-Physical Works, being a translation of Tradaius quinque medico-physici by
John Mayow. Alembic Club Reprints, No 17. Chicago: The University of
Chicago Press. 1908. pp. xxiii 4- 331. Price, $1.36, post-paid.

"How true it is that the value of truth is not absolute; there is a time
and place for everything, including a new truth. If a discovery is made
before its time, it withers up barren, without progeny, as did Mayow's."
Thus wrote Sir Michael Foster in his lectures on the history of physiology.
It is astonishing to learn how adequately some of our present views on
chemistry and physiology are foretold in the writings of Mayow, whose
observations were allowed to remain unappreciated for nearly a century
and until Lavoisier had contributed his researches on oxidation. The
existence and functions of oxygen were foreshadowed in Mayow's nitro-
aerial spirit which he recognized as that part of the atmosphere which
supports combustion ; it is present in nitre and enters the blood in res-
pimtion. With a few verbal changes Mayow's description of the mecha-
nism of respiration might serve as a text-book account of the physical
features of the process to-day. The fimdamental characteristics of
muscular metabolism were also clearly appreciated, and that at a time
when the nature of gases was obscure. "We may then suppose," wrote
Mayow, physiologist and chemist, in the essay on respiration (1668),
"that nitro-saline particles {i. e., oxygen) derived from the inspired air
constitute the one kind of motive particles, and that these, when they
meet the others, the saline-sulphurous particles (t. e., combustible sub-
stances) supplied by the mass of the blood and residing in the motor
parts, produce the effervescence from which muscular contraction re-
sults" (p. 208).

NEW BOOKS. 1047

If the discussions on animal spirits, which Mayow identified with his
nitro-aerial particles, are less suggestive to-day, they are none the less
interesting as a record of contemporary chemical and physiological prog-
ress. As a specimen of these early views the following quotation is of
interest: "If the stomach be quite empty of food, its internal membranes
are, as is probable, pinched by the nitro-aerial partieles, and hunger
seems to arise from this."

It is a pleasure to have a thoroughly readable English translation of
these classic papers by Mayow. They can be recommended as enter-
taining specimens of scientific imagination and critical acumen, as well as
striking illustrations of an appreciation of the experimental method long
before the modem period of discovery in chemistry.

LAFAvmTE B. Mendei*.

Descriptive Biochemie mit besonderer Berticksichtigung der chemischen Arbeitsmethoden.
By Dr. Sigmund FrAnkei<. Dozent f. med. Chemie an der Wiener Universitat,
Wiesbaden: J. F. Bergmann. 1907. pp. 639. Price, 17 Marks.

This book contains a description of the substances occurring in the ani-
mal body together with the methods for their isolation, their synthesis, and
their quantitative determination and also their degradation products.
Spedal chapters are devoted to the ferments and to the chemistry of the
organs, secretions and excretions. The book is intended to serve as an aid
to workers in physiological chemistry. In the preparation of the book, the
literature up to the end of the year 1907 was consulted and numerous cita-
tions are made.

The facts of physiological chemistry are given in the book in the most
compact sort of way but not to the detriment of the subject. Some ex-
ception, however, may be taken to the very free use made of abbreviations
of names of common things which will require the reader to learn quite a
number of abbreviations devised by Frankel himself. In some parts of the
book there appears evidence of haste in the preparation of the manuscript
as shown by inaccuracies of statement and incorrect formulas. The book
contains a vast amount of valuable information brought up to practically
the latest date, and is a rich mine to physiological chemists.

John Marshai^l.

Studies in Plant and Organic Chemistry and Literary Papers. By Helen Abbott
Michael (Helen C. DeS. Abbott), with Biographical Sketch. Cambridge, Mass.:
The Riverside Press. One vol., pp. 423. 1907. Price, $2.50 net.

Although the subject of this appreciation had a "genius" for music
she deserted it to study Helmholtz's work on optics. From physics
her "interest ran to zoology and the dissecting of animals." Next she
enters a medical college and passes "the first year's examination in
chemistry, anatomy and physiology with a record of one hundred in

1048 NEW BOOKS.

each branch." Her investigative turn of mind then finds expressioo
in working out several phytochemical problems and in delivering public
lectures on such broad subjects as "plant analysis as an applied science,"
"the chemical basis of plant forms," etc.

Not content with what she can attain on this side of the Atlantic,
she goes to Europe in search of a laboratory in which to pmrsue her
phytochemical studies. However, she returns to America and resumes
the study of medicine. After her marriage she starts on a trip around
the world. A short residence at Worcester is soon interrupted
and residence is taken up in the Isle of Wight where a private chemical
laboratory is equipped. Here she works jointly with John Jeanprftre
as she had previously worked with Trimble in Philadelphia.

Returning once more to America she delivers her last lecture on a phyto-
chemical subject. Again she goes abroad, but this time "her interests"
are * 'enlisted in wider fields, " i, e., she writes about the Austrian peasant
and kindred topics. A third time she matriculates at Tufft's and wins her
medical degree in 1903. After a short medical practice she died Novem-
ber 29, 1904. Edward Krbmhrs.

Life and Scientific Activity of IV. A. Menshutkin. By N. Mbnshutkin. Published
by M. Frolovaia, 6 Galemaia Street, St. Petersburg (In Russian).

A detailed biography and review of the work of the late N. A. Menshut-
kin by his son. Considerable space is devoted to telling of the struggles
of the Russian students for liberty of assembly, etc., and of the faculties
of the University and Polytechnic Institute for autonomy.

H. M. GORDIN.

Neue Capillar- und Capillaranalytische Untersuchungen. By Fribdrich Goppbls-
ROBDBR, Basel: Emil Birkhauser. pp. xiv -f 81 pp.+ 52 tables. Price, 6 Marks.

This interesting little book is a concise report of original work in a field
which the author has made peculiarly his own. He gives a list of his eight
earlier publications upon the same subject, the first of which appeared in
1861.

As the facts Goppelsroeder has based so much work upon are, peAaps,
not widely known, it may not be amiss to state them briefly. When strips
of filter paper (cotton, linen or other fabrics may be used, but filter paper
is generally preferable) are hung with their ends dipping in liquids or in
solutions, capillary force causes the liquids to creep upward to definite
heights which are diiBFerent for diflFerent substances, as in capillary tubes.
The effects of adsorption are also apparent and the result, when several
substances are in the solution, is the formation of bands or zones of different
widths, each zone containing mainly some one of the dissolved substances.
Numerous qualitative separations can be brought about in this way, and
corroborative color tests can be applied to the several bands on the pftpcf-

NEW BOOKS. IC49

The eighty pages of text seem brief for the amount of fact they contain
and a stupendous number of separate observations are condensed into the
fifty-two tabks, some of which are large, opening out like maps. The fol-
lowing are some of the titles of chapters:

I. The effect of diflFerent kinds of filter paper upon the height to which

liquids ascend IV. The effect of the length of paper immersed on the

height to which liquids ascend VII. Influence on the ascension of a

mordant action on the fibers. . . . .VIII. Capillary analysis of the extracts
from separate zones which were obtained by a preliminary capillary anal-
ysis. . . .IX. Sensitiveness of capillary analysis X. Capillary analjrt-

ical tests of water solutions of alkaloids XI. Capillary experiments

with members of different homologous series of organic substances

XII. Capillary experiments with water solutions of inorganic salts

XIV. Capillary experiments with milk, with skimmed milk and with each
diluted.

In some cases the results are gratif3ring, for instance potassium bichro-
mate— sulphuric acid gave a positive test for strychnine on the filter paper
strip when only one part of strychnine was present in i ,600,000 of the solu-
tion, while the same reagents did not detect one part of strychnine in 25,000
in the solution itself. Similar results were obtained with other alkaloids.
In other cases the results are not so satisfactory. For instance, in one hour,
pure milk rose 14. i cm., diluted with lo per cent, water it rose 14.7
cm., and diluted with 20 per cent, water it rose 16.4 cm. The difference
in these heights is enough to make one hopeful but hardly enough to justify
much reliance on the method for detecting watered milk. Capillary anal-
ysis in its present stage of development is an art rather than a science.

There is much of value to analysts in the book and it certainly should be
in every reference library, S. Lawrence BigeU)W.

A Course of Practical Organic Chemistry. By T. Slater Pricb, D.Sc., Ph.D., F.I.C.,
AND Douglas Twiss, M.Sc., A.I.C. London: Longmans, Green & Co. 1907.
xiii + 239 pp. Price, $1.20.

Both authors are connected with the Chemical Department of the Bir-
mingham Mimicipal Technical School, Dr. Price being the head of the
department and Mr. Twiss one of the lecturers.

The occasion for publishing the book, and the field it aims to fill are set
forth in the preface as follows: "The recent revision of the Board of Edu-
cation syllabus for Practical Organic Chemistry has naturally created

the necessity for a text book which will cover the complete sylkibus

The present book really consists of an amplification of the notes which have

been given to our own students who comprise (i) those working for

the Board of Education examinations, and (2) for the B.Sc. degree,"
the classes being, "with few exceptions, held only in the evening."

lOjO NEW BOOKS.

The subject matter is divided into three Parts or Stages, conesponding
presumably with those of the Board of Education syllabus, Stage I appar-
ently covering Fatty Chemistry, and Stages II and III, Aromatic. Each
of these stages is further subdivided into chapters, the contents of which
appear to have been decided by genetic rather than by structural considera-
tions.

It is, of course, a difficult matter for one not thoroughly familiar with
the local conditions which this book aims to meet, to pass judgment upon
how well it is likely to fulfil its mission. It is designed primarily to enable
students to pass certain fixed examinations, and only those on the ground
can tell how well it is succeeding.

As a laboratory text-book of organic chemistry for students in this
country, it cannot be recommended. The style is too much that of the
"cook book" type, and in the arrangement of the matter, representatives of
widely diflFerent classes of compoimds are grouped together because they
happen to be obtained from the same substance. Formic acid, for this
reason, appears in the chapter on bibasic acids. In several cases, the prep-
aration of a substance is given in one part of the book and its character-
istic tests and reactions in quite another. Many pages are devoted to
separations of a special list of selected organic compounds, the scheme for
which is given in detail.

A laboratory text-book should be something more than a collection of
preparations. It should be so arranged as to be a useful adjunct to the
lecture work, illustrating by practical examples the methods of preparation
and classes of compounds discussed in the lecture course. In a subject of
such limitless detail as organic chemistry, it is very important that the
basis of classification should appear clearly in any laboratory book, so that
the student should see this at once by the preparations given. The re-
viewer is also of the opinion that the working out of elaborate schemes for
identifying one of a limited group of compounds, or separating a mixture of
several of them, is undesirable from many points of \dew. The student
should be taught the fundamental tests and reactions upon which all this
depends, and then work out his own scheme of identification or separation,
advancing from simple to more complex mixtures. This stimulates his
interest, sharpens his powers of observation, and develops self-reliance,
while the following out of a scheme devised by somebody else to cover a
limited number of substances does not give him the same chance to test his

own strength.

The student working through the book will, to be sure, gain a ver}- good
practical knowledge of the more important methods of preparing organic
compounds, as well as an intimate acquaintance with a considerable number
of typical organic substances, but if taken up in the order given, he will
find it necessary to re-assort this information quite extensively before it

NEW BOOKS. 105 1

will run parallel to any of the lines of classification along which the subject
of organic chemistry is generally developed.

As stated before, however, since the book aims to cover only a special
field and is designed primarily for "home consumption,*' the above criti-
cisms should not be construed too harshly. There are, on the other hand,
many excellent features. The preparations selected are typical, the details
are given clearly, and much more space is properly accorded characteristic
reactions and anal3^cal tests than is customary in such books. The book
is in attractive form, and the proof has evidently been read with great care.

Marston Taylor Bogert.

Practical Methods for the Iron and Steel Worki Chemist. By J. K. HSBss, Ph.C,
Chief Chemist for the Carnegie Steel Company, New Castle, Pa. pp. 60. Easton,
Pa.: The Chemical Publishing Co. 1908. Price, |i. 00.

The author has compiled from various sources, methods for the anal-
ysis of such materials as iron ores, coke, coal, slags, irons and steels, fire-
bricks, cements, boiler waters, fats, bearing metals, and chimney and
producer gases. These have been modified to conform to his experience
and it is stated that all procedures have been carefully tested, and that
the directions as given are intended particularly as a guide to analysts
of limited experience. For such readers the author also describes some
of the essential features of a works laboratory, and gives directions for
the general conduct of laboratory work, the preparation of reagents,
or standards, and adds a collection of useful tables.'

In his endeavor to make this a '^practical" manual, the author has
made his directions so concise as to approach, if not to pass, the danger
point, especially in a work designed for inexperienced analysts. The
volume is of interest as an expression of opinion on the part of one who
is familiar with the demands made upon the laboratory of an iron or
steel works, regarding methods best adapted for use in such a laboratory.

H. P. Talbot.

A Laboratory Outline for Determinations in Quantitative Chemical Analysis. By
AlbbrtP. Oilman, S.B.,A.M., Professor of Chemistry, Ripon College, pp. 88.
Easton, Pa: The Chemical Publishing Company. 1908. Price, 90 cents. •

The procedures described include a considerable range of gravimetric
analyses and the volumetric determination of iron by the permanganate,
dichromate, and stannous chloride methods. Each procedure is accom-
panied by a series of questions to be answered by the student, and a page
upon which it is apparently intended that the student shall record his
observed data.

It is, unfortunately, impossible to commend this little volume. It is
badly written, the procedures are not accurately described, and many
of them are unreliable, as the author states with singular frankness but

I052 NEW BOOKS.

without excuse for their presentation. The number of typographical errors
is not creditable to either the author or the publishers.

H. P. Talbot.

Analyiii of Mixed PaintB, Color PigmentSi and Varnishes. By Cupfokd Dybr Hollsy
and E. P. Ladd. John Wiley & Sons, New York. Price, $2.50.

Prof. Ladd's contribution to the present volume is a dissertation on
Mixed Paints in general, with particular reference to their truthful label-
ing and to the experience of North Dakota in legislation to compel such
labeling. In his discussion of this matter, he is fair and his arguments
are convincing.

Part II of the book upon the analysis and testing of paints, by Prof.
HoUey, treats of the subjects from the standpoint of one who has recently
been called upon to analyze a great number of the paints, both good and
bad, that are now on the market. The discussion and the methods of
analysis recommended, are more complete and satisfactory when pigments
are dealt with than in the case of the vehicles. It may be said with
fairness, that this is the best work that has yet appeared on the anal3rsis
of the pigments of the present day.

A chapter is devoted to the Practical Testing of Paints. This most
important subject has been taken up by the North Dakota Government
Experiment Station, and an accoimt is given here of their methods of
operating.

The book is certainly of value to all who are interested in the subject.

Parker C. McIlhinEy.

Commercial Organic Analysis. By Ai^frsd H. Ai^i^sn, F.I.C, F.CS. Vol. Il-Part
III. Third Edition. Revised by the Author and Arnold Rowsby Tankard, P.C.&
P. Bhddston's Son & Co., Philadelphia. 8vo. 547 pp. Price, $5.00.

A comparison of the present book with the parts of the former edition
which related to the same subjects, brings out forcibly the fact that upon
these branches of analytical chemistry, a tremendous amount of work
has been done in the interval between the two editions. The subjects
treated are, The Aromatic Adds, Resins, and Volatile or Essential Oils.
The statement on the cover, that the subject of Phenols is also treated,
is misleading; this subject is really treated in Part II of Volume 11.

There is, inevitably, in the discussion of such subjects as those to
which this book is devoted, a certain lack of connection or logical sequence
between its several parts. The properties, uses and anal34ical necessi-
ties of such materials as organic adds, resins and essential oils vary so
greatly among themselves that a systematic or connected treatment
of their analytical chemistry is a very difficult task. The present work
is not faultless in this respect, but the great amount of information con-
tained, makes up for its somewhat disjointed composition. The work

RBCENT PUBLICATIONS. TO 5 3

of bringing up to date the various subjects discussed, has in the main
been well done. 1 Parker C. McIlhinEy.

Chemical Reagents, their purity and tests; a new and improved test based on and
replacing the latest edition of Krauch's "Die Prtifung der chemischen Reagentien
atif Reinheit." By £. Mbrck. Authorized translation by Henry Schenck, A.B.
(Harvard). New York: D. Van Nostrand Company. 1907. vii -|- 250 pp.
Price, $1.50 net.

It would be superfluous to describe the arrangement and treatment
of subjects in Krauch's well-known book. The English translation is
from the fotuth edition, which was published in 1905, and it is to be
noted that no literature references of a later date are given. German
idioms are conspicuously absent, and for this novelty (as we may fairly
say) as well as for his uniformly clear phraseology, the translator is to
be commended. In the main, also, he adheres to the chemical termi-
nology and spelling which are at present, adopted by the leading American
and English journals, but which so many chemists still disregard or
are ignorant of. We refer more particularly to the endings — in, -ine
and -ol. As, hematoxylin, iodeosin (but eosine is given on page 88),
hematein and phenolphthalein, all of which are non-basic; but brucine,
diphenylamine and aniline which are basic. His usage in regard to
-ol, is not quite so uniform. Pyrogallol and resordnol are correctly
given, with the older names in parenthesis, but glycerin is preferred
to glycerol (given in parenthesis). So, also, phlorogludn and furfural
(furfurol) — ^for furol — are given. But all these points are, no doubt,
of minor importance and detract little from the value of the book. The
tests given are certainly delicate enough for all but the most critical
work, and with this qualification, the book can be recommended to all
chemists. It is not too much to hope that in some future edition of this
or of a similar work, even more delicate tests, which will satisfy the de-
mands of the highest accuracy, will be given for at least the commoner
reagents. C. E. Waters.

RECENT PUBLICATIONS.

AND8RSON, J. W. : Refrigeration. An elementary text-book. New York: Ix>ng-
mans, Green & Co. 1908. 242 pp. I2.25.

Bavink, B. : Natttrliche nnd kanstliche Pflanzen nnd Tierstoffe. Kin Ueberblick
fiber die Portachritte der neneren organischen Chemie. Leipzig: B. G. Teubner.
1908. M. 1,25.

Bbnbdikt-Ui^br. : Analyse der Fette nnd Wachsarten. 5 umgeab. Aufl. bearb.
von F. Ulzer, P. Pastrovich und A. Eisenstein. Berlin: J. Springer. 1908. M.
28,60.

Blancha&d, Arthur a. Syntliatic Inorganic Chemiatiy. New York: John
Wiley & Sons. 1908. 90 pp. i2mo. |i.oo.

BoCKMANN, Pribdrich: Celluloid, its raw material, manufacture, properties,

I054 RECENT PUBLICATIONS.

and uses, etc. Translated from the 3rd rev. German ed. by Charles Salter. New
York: D. Van Nostrand Co. 1908. 113 pp. i2mo. $2.50.

Cain, J. Cannki,i<: The Chemistry of the Diazo-Compomids. New York: Long-
mans, Green & Co. 1908. 172 pp. I3.00.

Chamberlain, Arthur H.: The Conditions and Tendencies of Technical Sdn-
cation in Germany. Syracuse, N. Y. : C. W. Bardeen. 1908. 108 pp. 50c.

Clappbrton, G.: Practical Paper-Making: a manual for paper-makers and
owners and managers of paper mills, etc. 2nd ed. revised and enlarged. New York:
D. Van Nostrand Co. 1907. 226 pp. i2mo. $2.50.

DuPONT DB Nbmours, K. I., Powder Co.: Useful Information for Practical Men,
compiled for E. I. DuPont de Nemours Powder Co. 1908. 216 pp. $1.

Fischer, E.: Organische Syntheae und Biologie. Berlin: J. Springer. 1908.
M. I.

GiLMAN, A. F.: Quantitative Chemical Analysis. Easton, Pa.: Chemical Pub-
lishing Co. 1908. 88 pp. 90C.

GoBRENS, Paul: Introduction to Metallography. Translated by Fred Ibbotson.
New York: Longmans, Green & Co. 1908. 214 pp. 12.50.

Hbess, J. K. : Practical Methods for the Iron and Steel Works Chemist. Easton,
Pa. : Chemical Publishing Co. 1908. 60 pp. |i.oo.

Hill, Leonard: Recent Advances in Physiology and Bio-Chemistry, by Leonard
Hill and others. New York: Longmans, Green & Co. 1908. 742 pp. $5.00.

HoLLEY, Clifford D. and Ladd, E. F.: Analysis of Mixed Paints, Color Pig-
ments, and Varnishes. New York: John Wiley & Sons. 1908. 238 pp. i2mo.
I2.50.

Landauer, John: Spectrum Analysis. Authorized translation by J. Bishop
Tingle. 2nd edition rewritten. New York: John Wiley & Sons. 1908. 236 pp.
8vo. I3.00.

LoEB, Jacques: A New Proof of the Permeability of Cells for Salts or Ions*
(Preliminary communication). Berkeley, Cal.: University of California Press. 190S.

Mairk, Frederick: Modem Pigments and Their Vehicles. New York: John
Wiley & Sons. 1908. 266 pp. i2mo. (2.00.

PoiNCAiR^, Jules Henri: The Value of Science. Authorized translation by G.
Bruce Halsted. New York: Science Press. 1908. 147 pp. i2mo. $1.25.

Standage, H. C: Decoration of Metal, Wood, Glass, etc., edited by H. C
Standage. New York: John Wiley & Sons. 1908. 228 pp. i2mo. $2.00.

Stansfield, Alfred: The Electric Furnace: its evolution, theory and practice.
New York: Hill Publishing Co. 1908. 211pp. 8vo. |2.oo.

Talbot, Henry P. : An Introductory Course of Quantitative Chemical Analysis,
with explanatory notes and stoicbiometrical problems. 5th ed. rewritten and revised.
New York : The MacMillan Co. 1908. 176 pp. t^-S^.

Whipple, George C: Typhoid Fever, its causation, transmission and pre-
vention. New York: John Wiley & Sons. 1908. xzxvi -j- 407 pp. 50 figs, lama
Cloth, I3.00.

' /

;

• )

) • H"

.^ «

'^Cotz'r^.

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Books

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "Journal of the American Chemical Society"

See other formats