FRONTISPIECE ,.. Apparatus for Analysis of Atmospheric Air, devised by Dr. Klas Sonden. The Composition of the Atmosphere With Special Reference to Its Oxygen Content BY FRANCIS G. BENEDICT Director of the Nutrition Laboratory of the Carnegie Institution of Washington WASHINGTON, D. C. PUBLISHED BY THE CARNEGIE INSTITUTION OF WASHINGTON 1912 CARNEGIE INSTITUTION OF WASHINGTON PUBLICATION NO. 166 3 H- 1* I* ISAAC H. BLANCHARD COMPANY NEW YORK CONTENTS. Paet I. Page An historical account of the development of methods for determining oxygen. 3 Early investigations on the composition of the air 4 The nitric-oxide eudiometer 7 The beginnings of accurate air-analysis 15 The foundations of modern air-analysis 29 Summary of historical digest 66 Part II. Analyses of atmospheric air made at the Nutrition Laboratory 69 Fundamental essentials of accurate air-analyses 71 Apparatus and technique used in this research 75 Detailed description of the apparatus 76 Reagents used 79 Plan and methods of research 82 Method of collecting outdoor air 82 Method of using the apparatus and results obtained 83 First routine and results obtained 83 Second routine and results obtained 88 Control analyses 93 Third routine and results obtained 94 Effect on oxygen absorption of high and low temperatures 96 Fourth routine and results obtained 98 Fifth routine and results obtained 100 Conclusions from results with fifth routine 105 Analyses of air collected on the Atlantic Ocean 106 Analyses of air from Pike's Peak 108 Analyses of street air 109 Analyses of subway air 110 Absorption of oxygen by potassium pyrogallate Ill Conclusions 114 Hi THE COMPOSITION OF THE ATMOSPHERE WITH SPECIAL REFERENCE TO ITS OXYGEN CONTENT PART I. AN HISTORICAL ACCOUNT OF THE DEVELOPMENT OF METHODS FOR DETERMINING OXYGEN. The interest in meteorology and aeronautics — an interest ever increas- ing and international in scope — and the recent discoveries in the atmos- phere of argon, helium, and their associated rarer gases, accentuate the fact that the present information regarding the oxygen content of atmos- pheric air, and, indeed, of the carbon-dioxide content, is far from satis- factory. The known sources of oxygen are very limited in number, for although it has been demonstrated that during certain periods of vege- tative growth oxygen is liberated, the amount thus supplied to the atmos- phere must of necessity be slight. On the other hand, the drafts upon atmospheric oxygen are constantly increasing. Taking into considera- tion those agencies that are directly or indirectly influenced by man, we see that since both the population of the world and the combustion of fuel are increasing enormously, this drain upon atmospheric oxygen must to-day be very much greater than, for instance, during the Stone Age. If, in addition, we consider the abstraction of oxygen by living organisms other than man, the oxidation of organic matter and decay, and the oxida- tion of iron, we find that all doubtless play an important role in decreasing the percentage of oxygen in the air. With these various agencies at work diminishing oxygen and producing carbon dioxide, it is to be expected that variations in the density of population, in the number of factories, in the distribution of vegetative tracts, and in the proportion of land and sea, would lead to like variations in the composition of the air. The near- est analogy to the atmosphere is sea-water, which, while vast in bulk, is known to have differences in composition at different depths and with different geographical distribution. If changes take place in the composition of the air, of what nature are they? Are they measurable by our present methods of chemical analysis? Do these variations bear any relation to the changing seasons, to growth of vegetation, to rain, snow, and similar meteorological conditions, and to geographical location? These are questions that have long been in the minds of scientists, and, indeed, are to-day still debatable. Alterations in climatic conditions have been ascribed by Arrhenius1 to relatively small changes in the carbon-dioxide content of the air. Thus it seems that on meteorological grounds alone a study of the composition 1 Arrhenius, Philosophical Magazine, 1896, pp. 237-276; also, Svenska Vetenskaps- Akademiens Forhandlingar, 1901, 58, pp. 25-58. 3 4 Composition of the Atmosphere of air is well worth undertaking. But such a study, especially the study of the oxygen content of the air, has a still higher value in its relation to human life. No one chemical element enters so extensively into vital processes as does oxygen. The human body may live without food for many weeks; it may live without water for several days; but without air or oxygen it can live for only a very few minutes. Not only is oxygen es- sential to life, but the purity of the air is of fundamental hygienic impor- tance. That this fact has been recognized is evidenced by the emphasis recently laid upon the necessity of an outdoor existence in combating tu- berculosis. Furthermore, a knowledge of the composition of the air is necessary for the solution of the important problems of the ventilation of houses, mines, rapid-transit subways, and railroad tunnels. Of still more special significance, and with a more intimate bearing upon problems in physiology, is the fact that the determination of the oxygen consumption of man and important quantitative determinations in respiration experi- ments depend usually upon an exact knowledge of the composition of the air taken into the lungs. It is peculiarly fitting, therefore, that a study of this subject should be made a part of the scheme of research carried out in the Nutrition Laboratory. The two forms of gas-analysis apparatus conceded by all experimenters to give the most exact results are the apparatus of Haldane in England and that of Sonden and Pettersson in Stockholm. Several forms of the Haldane apparatus are in the possession of the Nutrition Laboratory, and also a Sonden apparatus specially designed for the determination of carbon dioxide and oxygen in the air of the respiration chamber. This latter apparatus was devised by Dr. Sonden after a conference with the author in Stockholm four years ago, in which the various difficulties in the way of exact gas-analysis were carefully considered. With this apparatus the Nutrition Laboratory found itself in a position to carry out a more com- plete study of the percentage of oxygen in outdoor air than had hitherto been made. The ingenuity of Dr. Sonden and the technical skill of Miss Alice Johnson, of the laboratory staff, made such a study of atmos- pheric oxygen possible. EARLY INVESTIGATIONS ON THE COMPOSITION OF AIR. Although much has been written in recent years regarding the chemi- cal composition of the atmosphere, there exists nowhere, at least in Eng- lish, an historical account of the development of knowledge regarding the percentage of oxygen in the air; it is therefore deemed fitting to collect in this memoir the widely scattered records of the development of this most interesting subject. The tenacity with which the belief in the elemental nature of air was held is well exemplified by the fact that not until the latter part of the eighteenth century did scientists begin to appreciate the fact that air con- sisted of two or more gases. This recognition of the composition of air History of Air-Analysis 5 was unquestionably much retarded by the general acceptance of the phlo- giston theory advocated by Stahl. According to the supporters of the phlogiston theory, all air had the same composition, but was more or less supplied with the "combustible essence" phlogiston. As a matter of fact, we find that the earliest investigators whose experimental evi- dence subsequently showed air to be composed of two or more gases, namely, Scheele, Priestley, Cavendish, and Lavoisier, all firmly believed in the phlogiston theory. Even after numerous roughly quantitative experiments had been made in which it had been demonstrated that the volume of air decreased after oxidation of material in it, scientists were loath to give up the phlogiston theory, and air was said to be more or less phlogisticated. Although a diminution in volume was observed when air was exposed to certain substances, such as alkaline sulphides, moist iron filings, phosphorus, or nitric oxide, this loss was simply considered as due to a portion of the atmosphere which was not saturated with phlogiston. Scheele, noting the fact that the specific gravity of the air after ab- sorption by various reagents had not altered, concluded that the decrease in bulk could not be due, as he first supposed, to the absorption of phlo- giston, and that the atmosphere must of necessity consist of two distinct fluids. Although at first a strong supporter of the phlogiston theory, Lavoisier in 1777 enunciated the belief that the air consisted of two gases, one nitrogen (azote) and the other at first called dephlogisticated air, but finally known as oxygen; thus, for the first time, the definite existence of two distinct elemental components of the atmosphere was made clear. This observation soon led Lavoisier to the belief that all the phenomena of combustion could be explained on the basis of oxygenation without refer- ence to the existence of phlogiston. Cavendish did not accept this new conception of the composition of the air until a number of years later, and even then his acceptance was far from a complete surrender. The one man, Priestley, who, perhaps more than anyone else, illuminated our knowledge of the atmosphere by his discovery of oxygen, advocated the phlogiston theory until his death in 1810. It is thus clear that the dephlogisticated air of the earlier scientists was nothing more nor less than what we now call oxygen, and hence, although many of these writers considered the diminution in volume produced by the various reagents as an index of the amount of dephlogisticated air present, their observations have a certain historical value as indicating approximately the estimation of the amount of oxygen in the air by the methods then current. The earliest observations of the quantitative relationship between the dephlogisticated air and the phlogisticated air were undoubtedly made simultaneously by Priestley in England and Scheele in Sweden. l Scheele's 1 According to the published notebooks and laboratory records of Scheele (Carl Wil- helm Scheele, Efterlemnade bref ooh anteckningar, Utgifna af A. E. Nordenskiold, Stockholm, 1892), his experiments must have antedated Priestley by two or three years. 6 Composition of the Atmosphere experiments were wonderfully comprehensive and included a determina- tion of the decrease in volume of a confined mass of air not only when mixed with nitric oxide but when subjected to the action of alkaline sul- phides, moist iron filings with and without an admixture of sulphur, phos- phorus ignited and at room temperature, and precipitated ferrous hy- droxide. His experiments with alkaline sulphides were somewhat more extensive than with the other agents. In the first experiment he dissolved alkaline liver of sulphur in water, poured 4 ounces of the solution into a 24-ounce bottle, which he corked well, then reversed the bottle, and im- mersed its neck in a small vessel of water, keeping it in this position for a fortnight. At the end of this time, without removing the bottle from the water, he took out the cork, and the water at once rushed into the bottle. By noting the amount of water thus added he was able to demonstrate that in a fortnight, out of the 20 volumes of air in the bottle, 6 volumes were lost. In a second experiment he reports that 4 parts were lost out of 20, and at another time, when the bottle was corked for 4 months, there were 6 parts lost out of 20. As the result of the third experiment agreed with the first, the average of the three experiments shows a loss of approxi- mately 6 parts out of 20, or 30 per cent. Scheele also exposed phosphorus to a confined volume of air, allowing it to remain for 6 weeks, and found that one-third of the air was lost. Of particular significance in the light of the main purpose of this mem- oir is the fact the Scheele was the first to attempt a systematic study of the composition of the atmosphere over a lengthy period. In 1779 he com- municated to the Academy1 in Stockholm the results of his investigation of the preceding year. This communication was deemed of great impor- tance by Lavoisier.2 The method employed by Scheele was to expose a confined volume of air to the action of a mixture of 2 parts of iron filings and 1 part of pow- dered sulphur, to which had been added a small amount of water. This mixture produced in a few hours a diminution of the air greater than that obtained by the sulphuret of potassium in several days.3 Scheele used this method to determine the degree of salubrity of the atmosphere at different times of the year and to find out the proportion of "vital" air. From January 1 to March 23, 1778, atmospheric air was found to con- tain 27.3 per cent of oxygen; on March 23, 24.2 per cent; on April 19, 20, and 21, 30.0 per cent. During the months of May and June the quantity of vital air was between 24 and 27 per cent; on October 5, during a very heavy storm, it was found to be 30 per cent. From October 5 to Novem- ber 4 the quantity of vital air was 24 to 27 per cent; and on November 4 and 5, with the barometer very high, it was 24 per cent. From November 1 Scheele, Kongl. Vetenskaps-Academiens Handlingar, 1779, 40, p. 50. 'Lavoisier, Recueil des Memoires de Lavoisier, 3, p. 154; and Oeuvres de Lavoisier, 1862, 2, p. 715. 8 A figure of Scheele's apparatus is also given in F. Hoefer's Histoire de la chinfie, 2d ed., Paris, 1869, 2, p. 453. History of Air-Analysis 7 5 to 20, the quantity of vital air was 24 to 27 per cent; on November 20 30 per cent, and on November 21, 24 per cent. During December the quantity of vital air was constantly between 24 and 27 per cent. Scheele's investigation slowly but surely claimed the attention of scientists in other countries. In this connection it is of interest to quote the words of Dr. Joseph Black:1 But Scheele was the first person who, from a number of ingeniously contrived ex- periments, concluded by very fair reasoning that atmospherical air is a mixed fluid composed of about two parts of azotic gas, and one part of vital air or oxygen gas, along with a very small admixture of carbonic acid. The ingenuity and industry of this great Swede may properly be con- sidered as having started the investigation of the composition of the air — an investigation that has had almost the continuous attention of chemists for over 130 years. THE NITRIC-OXIDE EUDIOMETER. Contemporaneously with Scheele, Priestley in England published sev- eral volumes of his "Observations on Air." In 1772 Priestley observed that when nitric oxide, prepared a number of years before by Stephen Hales, was added to common air confined in a vessel over water, a diminu- tion in volume resulted. Experimenting in this way, Priestley showed that about one-fifth of the air combined with the nitric oxide and was ab- sorbed by the water.2 Priestley's discovery of oxygen, which he found by heating the red ox- ide of mercury, was made on August 1, 1774. He was so wedded to the phlogiston theory, however, that he could only consider this oxygen as dephlogisticated air; hence its elemental nature was never admitted by him. This discovery was shortly followed by researches on the relation- ship between oxygen and the vital processes. It was early believed that the vital processes were more active in oxygen-rich air than in air that was deficient in oxygen; this stimulated innumerable investigations of the purity or salubrity of the air, chiefly by means of the simple nitric-oxide reaction of Priestley. The attempts to measure quantitatively the oxy- gen in the air early led to the development of special forms of apparatus for these measurements ; as a matter of fact so extensively was the nitric- oxide test employed for studying the oxygen content of air, and so univer- sal was the belief that the larger the amount of oxygen in the air the better was the air, that the instrument was actually designated an eudiometer, i.e., a measurer of the goodness or salubrity of the air. As the dephlogisti- cated air supported respiration and combustion much better than ordi- nary air, it was natural to ascribe the healthfulness of the latter to the amount of dephlogisticated air present in it. 1 Black, Lectures on the elements of chemistry, 1st Am. ed. from the last London ed., 1806, 2, p. 344. See also Scheele, Efterlemnade bref och anteckningar, edited by A. E. Nordenskiold, Stockholm, 1892, p. 78. * Priestley, Experiments and observations on different kinds of air, 1775, 1, p. 111. 8 Composition of the Atmosphere In considering the heterogeneous results reported by Priestley, it is important to note that the values he obtained were all comparative rather than absolute. He supposed that all samples of air had different quan- tities of dephlogisticated air in them; and if he took one sample of good air, compared it with a sample of questioned purity taken at the same time and at another place, and found that they underwent the same contrac- tion in volume, he could assume that the two samples were equally pure. In other words, Priestley evidently failed to realize the significance of the contraction in volume. All of the observations of Priestley, then, were made distinctly upon the comparative rather than upon the absolute ba- sis. He was, indeed, somewhat disturbed by the fact that air which theo- retically was bad did not often show any deterioration.1 He says on this point : When I first discovered the property of nitrous air as a test of the wholesomeness of common air, I flattered myself that it might be of considerable practical use, and par- ticularly that the air of distant places and countries might be brought and examined to- gether with great ease and satisfaction; but I own that hitherto I have been rather disappointed in my expectations from it. My own observations have not, indeed, been many; but according to them the difference of the open air in different places, as indicated by a mixture of nitrous air, is generally inconsiderable; and I have reason to think that when very unwholesome air is conveyed to a great distance, and much time elapses before it is tried, it approaches, by some means or other, to the state of wholesome air. At least such I have found to be the case with the worst air that has at any time been sent to me in Wiltshire from distant manufacturing towns and workshops, etc., in them, where the air was thought to be peculiarly unwholesome. I am satisfied, however, from my own observations, that air may be very offensive to the nostrils, probably hurtful to the lungs, and perhaps also in consequence of the presence of phlogistic matter in it, without the phlogiston being so far incorporated with it, as to be discoverable by the mixture of nitrous air. I gave several of my friends the trouble to send me air from distant places, especially from manufacturing towns, and the worst they could find to be actually breathed by the manufacturers, such as is known to be exceedingly offensive to those who visit them; but when I examined those specimens of air in Wiltshire, the difference between them and the very best air in this country, which is esteemed to be very good, as also the difference between them and specimens of the best air in the counties in which those manufacturing towns are situated, was very trifling. Mr. S. Vaughan, senior, on his passage from Jamaica, brought me two bottles of air, one from the hold of the ship, intolerably offensive, the other the fresh air above deck in about 30' N.; but the difference between these specimens of air, and the air of Wilt- shire, was quite inconsiderable. I have frequently taken the open air in the most exposed places in this country at different times of the year, and in different states of the weather, etc., but never found the difference so great as the inaccuracy arising from the method of making the trial might easily amount to, or exceed. This recognition of at least the existence of a limit of accuracy for his method was unfortunately not seriously considered by many of his con- temporaries. 1 Priestley, loc. cit., 1779, 4, p. 269. History of Air-Analysis 9 The application of the nitrous-oxide method for determining the degree of phlogistication of the air as first brought out by Priestley immediately led to an extensive interest in this problem on the part of a number of in- vestigators. As Priestley's observations were but roughly quantitative, the Abbe* Felice Fontana1 in Italy constructed an instrument permitting a much greater accuracy in the measurement of the contraction in volume and determined the quantity of oxygen contained in air by absorption with nitric oxide, obtaining results showing from 18 to 25 per cent. Using this instrument, a number of investigators began studying the absorption due to nitric oxide, an absorption that was shortly to be explained by the oxygenation theory of Lavoisier. Slight minor modifications of the ap- paratus and method were made by many scientists who appeared almost inordinately2 occupied in the testing of air in various places. Prominent among the users of this instrument was Marsiglio Lan- driani,3 who in 1775 published a record of his investigation and first in- troduced the term "eudiometer," descriptive of the instrument devised by Fontana. In a letter to Priestley, dated at Milan, November 17, 1776,4 Landriani writes : Before you receive this letter I shall have sent you my eudiometer, together with a short memoir, explaining the use of the machine, in order to ascertain with exactness the wholesomeness of the air in any particular place. It is the same instrument that I made use of in my tour through Italy, in the course of which I have had the satisfaction of convincing myself that the air of all those places which, from the long experience of the inhabitants, has been reputed unwholesome, is found to be so, to a very great degree of exactness, by this instrument of mine, so that the theory seems to correspond very well to observation. In the mountains near Pisa I made trial of the air at different heights, beginning on the plain, and proceeding to the highest summits; and found a remarkable difference in the state of the air, every stratum being purer in proportion as I ascended. 1 Felice Fontana, Descrizioni ed usi di alcuni stromenti per misurar la salubrita dell' aria, Firenze, 1774. 2 For 20 to 30 years after Priestley's first discovery of the nitric-oxide eudiometer, it would appear from the innumerable references in the literature that every scientist of reputation, and many with no reputation, attempted air-analyses. Writing in 1912, one can but compare proportionally the number of those using the various forms of eu- diometer 130 years ago with those to-day using wireless telegraphic apparatus. Priest- ley has summed up the situation admirably in the following sentences from the preface to volume 3 of his "Experiments and observations on different kinds of air," 1777: "Those of my readers who may wish that I would still give a principal attention to this branch of experimental philosophy, will the less regret my discontinuing it, when they are informed with how much ardour and ability these pursuits are now prosecuted in very different parts of Europe. "I am also informed by my friend Mr. Magellan, who frequently visits, and has a very extensive correspondence with the Continent, so as to be well acquainted with the present pursuits of philosophers, and who has himself taken pains to instruct many in- genious foreigners in the best methods of making experiments of this kind, that many other persons, whose names are at present unknown to the public, are at this very time assiduously employed on the same subject." See also Johann Andreas Scherer, Geschichte der Luftgiiteprufungslehre, Vienna, 1785, 2. p. 74. 3 Marsiglio Landriani, Richerche fisiche intorno alle salubrity dell' aria, Milano, 1775. 4 Translated and printed in Priestley's Experiments and observations on different kinds of air, 1777, 3, p. 380. 10 Composition of the Atmosphere Among other early modifications of the nitric-oxide eudiometer are those of Magellan.1 These are also referred to in a postscript in a letter to Priestley,2 dated London, November 30, 1776, in which he writes: The other contrivances I want to show you are two new eudiometers to measure the degree of the salubrity of the air in different places. One of them, which I reckon the better, being the more simple and the neater of the two, is according to the original idea of your experiments on this subject; but in neither of them do I make use of any cock, both on account of its being difficult to be made, and likewise subject to be out of order. Those of Mess, the Chevalier Landriani and the Abbe Fontana seem to be liable to this inconvenience. The experiments already made in most parts of Italy by the former of those gentlemen, with his own eudiometer, deserve the greatest praises; and it is to be wished that philosophers would more generally apply themselves to this inter- esting subject of inquiry. Dobson's3 instrument was used for analyzing the air of "sea weed pods" and also, for comparison, the air of Liverpool. Lavoisier4 reports three experiments made with nitric oxide in which he found 25.3, 25, and 25.2 per cent of oxygen, respectively, an agreement which he says he had not dared to hope for. From these results, therefore, he concludes that the atmosphere, as he had previously announced, con- sisted of about 3 parts of mephitic air and 1 part of vital air. In 1777 Lavoisier made determinations of the quantity of vital air contained in the atmosphere and found it to be about 27.5 parts in 100, but maintains it is possible that this larger quantity of vital air depends upon the season. Undoubtedly this quest for air with the highest oxygen content led to innumerable analyses of atmospheric air in the latter part of the eight- eenth century which otherwise would not have been made. We find that Cavendish is reported as having made over 500 analyses of air by the nitric-oxide eudiometer before 1 790. 5 Similarly the method was employed "daily for three years" by de Saussure6 in a comparison of this method with the phosphorus eudiometer which was later to play an important role in air-analyses. While many observers early found difficulties in the nitric-oxide method and soon discarded it for other methods, it seems to have been reserved for the genius of Cavendish so to adjust the conditions of experi- mentation with this agent as to secure approximately accurate infor- mation regarding the proportion of oxygen and nitrogen in the air. His results are marvelously accurate when judged by analyses made by the most approved methods of modern times. Cavendish was more interested 1 J. Hyacynth de Magellan, Description of a glass apparatus for making mineral waters like those of Pyrmont Spa, Seltzer, Seydschnitz, etc., together with the descrip- tion of some new eudiometers, etc., London, 1777. 1 Priestley, loc. cit., 1777, 3, p. 379. • See Dobson's letter to Priestley, loc. cit., 1779, 4, p. 469. 4 Lavoisier, Memoires de l'Academie des Sciences, 1782, p. 486. See also Oeuvres de Lavoisier, 1862, 2, p. 503. 6 Wilson, Life and works of Cavendish, London, 1851, p. 227. 5 de Saussure, Journal de Physique, 1798, 47, p. 470. See also Gilbert's Annalen der Physik, 1799, 1, p. 505. History of Air- Analysis 11 in knowing the amount of the decrease in volume than the cause. With his keen insight and experimental technique he attacked the problem, and in 1783 published a paper describing a new eudiometer. By carefully noting the rate and the amount of nitric oxide used with this apparatus and particularly by adding the air to a previously measured amount of nitric oxide, he was able to make a more careful examination and test of air. Of particular interest in connection with this paper are his words regarding the tests of air on different days :* During the last half of the year 1781 I tried the air of near 60 different days in order to find whether it was sensibly more phlogisticated at one time than another; but found no difference that I could be sure of, though the wind and weather on those days were very various, some of them being very fair and clear, others very wet, and others very foggy. He also studied the air in different localities and compared the air of London with that of the country. Although Cavendish found some evi- dence to show that the air at Kensington was better than that in London, he nevertheless believed that the differences were no greater than the limits of experimental error, and taking the mean of all, that there was apparently no difference between them. The number of days compared was 20, the greater part of the samples being taken in cold winter weather, when there were a great many fires and but little wind to blow away the smoke. From figures given in Cavendish's notebook, Wilson2 concluded that these observations established that the percentage of oxygen in the air was 20.83. This figure has been made much of in discussions of the percentage of oxygen in the air, but an examination of Cavendish's data shows that the limit of error was very large, and that he could not possibly have been inside of 1 or 2 per cent of the total amount of oxygen involved ; hence a representation of his percentage of oxygen with four significant figures is without value. As Cavendish at that time paid no attention to the purity of the nitric oxide, although recognizing in a crude way the dif- ferences in quality and the possibility of combination with varying amounts of dephlogisticated air or oxygen, it is fair to conclude that the method could not possibly have had an accuracy closer than 2 per cent of the total. Cavendish was the first to establish that the composition of the air was essentially constant within the limits of his apparatus, and that it did not sensibly vary in different parts of the country. That it was possible to obtain these results with an instrument and a method with as great error as we now know them to have had is certainly most remarkable evi- dence of his skill as an experimenter. Among others who extensively used the nitric-oxide eudiometer must be mentioned Ingen-housz, who, in his researches on plant life and on his travels, used a portable apparatus of his own devising.3 1 Cavendish, Philosophical Transactions, 1783, 73, p. 126. 2 Wilson, loc. cit., p. 41. 3 Ingen-housz, Vermischte Schriften, 2d ed., Vienna, 1784, 2, p. 242. 12 Composition of the Atmosphere Van Breda1 in Delft made analyses of air by the nitric-oxide eudiom- eter on 195 days in 1781 and 1782, but the results have little quantitative value. In an address before the Academy in Barcelona on May 22, 1790, An- tonio de Marti2 announced the results of his experiments on the "Quan- tity of vital air in the atmosphere, and the different methods of measuring it." Among other means he used nitric oxide and reported uniformity in composition of the air. During some days of the year 1787, in which the common air experienced no varia- tion by means of nitrous air, since 100 parts of each were uniformly reduced to 99 or 00, I was desirous of making a comparative trial of the same common air by means of iron and sulphur, and I observed that of 100 parts of air there remained from 79 to 81 and that consequently, from 19 to 21 hundredths had disappeared. One of the most extensive contributors to our knowledge of the com- position of the air obtained by means of the nitric-oxide eudiometer was Alexander von Humboldt, and in his records we find recognition of the existence not of dephlogisticated air but of oxygen, accepting Lavoisier's newest theory with regard to the air. von Humboldt recognized the im- portance of testing the purity of the nitric oxide used, and determined the degree to which it would dissolve in a ferrous-sulphate solution. In 1798 von Humboldt3 had analyzed the air collected in a balloon journey by Garnerin and Beauvais at an altitude of 1303 meters and com- pared it with the air of Paris, using nitric oxide and ferrous sulphate. The air in Paris showed 27.6 per cent oxygen and the balloon air 25.9 per cent. These balloon samples are of general interest in that they sub- stantiate the simultaneous observations of de Saussure,4 who, using a nitric-oxide eudiometer, analyzed air taken from the tops of several moun- tains. Though the results are not expressed numerically, de Saussure concludes that the air of mountains is somewhat less pure in vital air than that of the neighboring plains and valleys. Although a large number of desultory investigations had been made by Ingen-housz in London, van Breda in Delft, Pickel in Wurzburg, Lam- padius in Freiburg, Lichtenberg in Gottingen, Scherer in Vienna, and Breze, all of whom used the nitric-oxide eudiometer, yet von Humboldt5 was dissatisfied with these researches, believing that analyses of the air should be accompanied by observations with regard to its elasticity, tem- perature, moisture, electricity, and clearness. He accordingly undertook an extensive series of observations with regard to the oxygen content of 1 Letter to J. Ingen-housz, Vermischte Schriften, 2d ed., Vienna, 1784, 2, p. 441. 1 This lecture was translated into French and printed in the Journal de Physique, 1801, 52, p. 173. Abstracted in Gilbert's Annalen der Physik, 1805, 19, p. 389. The lecture was also printed in English in the Philosophical Magazine, 1801, 9, p. 250. 3 von Humboldt, Journal de Physique, 1798, 47, p. 202. 4 de Saussure, Voyages dans les Alpes, Neuchatel, 1803, 1 , p. 427. 5 von Humboldt, Versuche liber die chemische Zerlegung des Luftkreises, Braun- schweig, 1799, p. 150. History of Air-Analysis 13 air in Salzburg, using a nitric-oxide eudiometer. The close proximity to the Alps also enabled him to study the composition of air on the top of mountains as well as that in the valley. Although we now know that von Humboldt's oxygen determinations were erroneous owing to the test used, namely, nitric oxide, nevertheless his observations and the tabular statement of his results have the com- pleteness which characterizes modern scientific research, and in many ways the thoroughness of the investigation commends itself to modern workers. Using the Fontana eudiometer, he made observations of the air in Salz- burg (1302 feet above sea-level), covering practically a whole year, the samples of air being taken for the most part from a garden on the south side of the city. Since this air might be considered contaminated, he compared it with that from the open country some distance from the city, but was never able to detect any difference. The results are expressed in tabular form at the end of his book, each value representing the average of 3 to 5 tests. The analysis of the Salzburg air gave a range in oxygen content from 23.6 per cent to 29 per cent. The author concluded that the oxygen con- tent can vary 5.4 per cent and does not always remain between 27 and 28 per cent. In observations on 144 days he found only 7 times that the oxygen rose above 28. 1 per cent. The average oxygen content was : per cent November 25.6 December 26.8 January 27.5 February 27.2 per cent March 26.9 April 27.2 Average 26.8 He came to the conclusion that in clear weather there is an increase in the oxygen, and as bad weather approaches there is a decrease. He also made comparative observations of mountain and valley air. With the assistance of a friend, samples were taken simultaneously at noon from the Geisberg (3890 feet) and from the valley. On December 18, 1797, the mountain air gave 23.6 per cent of oxygen and the valley air 26.2, or 2.6 per cent of oxygen more than the mountain air. On January 30, 1798, another observation was made, the results for the mountain air being 26.1, and for the valley air, 27.4 per cent. On March 4, 1798, another set of samples gave for mountain air 26.4 and for valley air 27.3 per cent. On March 11, 1798, the results for the two samples were identical, namely, 26.4 per cent. Assuming that the nitric oxide used by earlier experimenters was not below a certain degree of purity, von Humboldt computed the oxygen content of the air in several cities as follows: per cent Vienna 26.1 Gottingen 26.6 London 26.9 Florence 25.3 Delft 27.0 14 Composition of the Atmosphere He evidently wished to take into consideration not only the tempera- ture of the air examined, but the possibility of differences in expansion, and referred to one of his researches indicating such a difference. About the time that his book was published, von Humboldt started on a scientific expedition to Spanish America, and in a letter written by him to Delam^therie1 from Cumana, South America, he shows his intense interest in the composition of the air, and in the possible sources of con- tamination or alteration of the oxygen content, by citing his experiment with a sample of air which he had collected in a bottle from the crater of a volcano. After having determined the purity of his nitric oxide by means of ferrous sulphate, he found only 19 per cent of oxygen in this sample, while at the sea-level the oxygen content was 27.8 per cent. The idea of a geographical difference in composition of the air was also evidently present, since he cites the fact that he was able to analyze the air on board ship with as much ease as in his laboratory, and found that the sea air at 10° 30' "on a beautiful moonlight night" contained over 30 per cent of oxygen. So firm was von Humboldt's belief in the nitric-oxide eudiometer that he was in constant polemical discussion with Berthollet with regard to the relatively new phosphorus eudiometer. As facility in experimental technique was acquired by scientists in the new field of pneumatic chemistry, errors in the nitric-oxide eudiometer were early recognized and it is not surprising to find Seguin2 stating that this method has 20 different errors. Even von Humboldt,3 its most ar- dent supporter, pointed out that nitric acid of different strengths yielded nitric oxide of different character which combined with different amounts of oxygen from the air. A few years later4 he acknowledged completely the errors of the method, which he then discarded for the hydrogen eudi- ometer of Volta. Subsequently, Berger5 in Geneva laid especial em- phasis upon the errors of the nitric-oxide method and opposed von Hum- boldt's belief that the nitric-oxide solution in ferrous sulphate was a suitable reagent for accurate oxygen determinations, concluding that the phosphorus eudiometer was very much superior. Although the original nitric-oxide method of determining oxygen was destined to be relegated to the ranks of impracticable chemical operations, the experience with it naturally led to the employment by Davy6 of a solution of nitric oxide in ferrous sulphate. This means of testing the purity of the nitric oxide had been advocated by von Humboldt, and Davy first made use of the oxygen-absorbing power of such a solution to analyze air. Davy strongly criticized the old Fontana nitric-oxide eudiom- eter, but pointed out that 1 c.c. of a reasonably strong solution of ferrous 1 von Humboldt, Gilbert's Annalen der Physik, 1800, 4, p. 443. 2 Seguin, Annates de Chimie, 1791, 9, p. 293. 3 von Humboldt, Annales de Chimie, 1799, 28, p. 123. 4 von Humboldt and Gay-Lussac, Journal de Physique, 1805, 60, p. 129. 5 Berger, Journal de Physique, 1802, 56, p. 253. 6 Davy, Journal of the Royal Institution, 1802, I, p. 45. History of Air-Analysis 15 sulphate saturated with nitric oxide would absorb 5 to 6 c.c. of oxygen. Comparative tests were made with phosphorus and alkaline sulphides. In his analyses of atmospheric air with a saturated solution of nitric oxide, he never found a change in the constituents. The air on October 3, 1800, on the sea, at the mouth of the Severn, with a strong west wind blowing over the Atlantic Ocean, contained 21 per cent of oxygen. Ex- actly the same amount was found in air brought from the coast of Guinea to Dr. Beddoes by two Liverpool surgeons. Davy's ingenious use of the ferrous-sulphate solution of nitric oxide was short-lived, for he himself found that the alkaline sulphides always gave a somewhat larger absorp- tion. Allen and Pepys1 employed the ferrous sulphate-nitric oxide method to analyze not only air, but some gaseous mixtures rich in oxygen. They report that outdoor air was continuously found to contain 21 per cent of oxygen. The last writer to make any considerable use of the nitric-oxide eudiom- eter was Dalton,2 who, comparing the Volta hydrogen eudiometer with the nitric-oxide eudiometer and the sulphide of lime absorption method, says: The nitrous gas eudiometer is of singular utility on many occasions. No other can exceed it in accuracy when mixtures contain very little, as one or two per cent of oxygen; or on the other hand when nearly the whole of the gas is oxygen. But when the mixture of gases contains from twenty to eighty per cent of oxygen, as in the case of common air, it is not the best when great exactness is required. THE BEGINNINGS OF ACCURATE AIR-ANALYSIS. The early historical interest attaching to the development and exten- sive use of the nitric-oxide eudiometer justifies the special treatment which has been given of this method; nevertheless other absorbents for oxygen were used. The large number of scientists experimenting daily with the newly discovered gas, oxygen, and acquiring information with regard to its properties, rapidly brought together a series of methods for determining this gas in air. While the nitric-oxide method has long been discarded as utterly worthless, the alkaline sulphides first employed by Scheele were for some time used in air-analyses, and phosphorus for absorb- ing oxygen, likewise employed by Scheele, has even to-day an extensive use. The explosion of air with hydrogen, first employed by Volta, has also withstood the critical attacks of over 100 years and to-day is much used. As Scheele found from his experiments with different absorptive agents variations in the oxygen content of air, so Lavoisier likewise, using various agents, was unable to find any satisfactory value for the percentage of oxygen, for we find in his reports varying values assigned to the oxygen ■ « ■ 1 Pepys, Philosophical Transactions, 1807, Part I, p. 247; Allen and Pepys, Philo- sophical Transactions, 1808, 8, p. 255. 2 Dalton, Philosophical Magazine, 1838, 3d ser., 12, p. 158. 16 Composition of the Atmosphere content of air. Lavoisier was infinitely more interested in chemical prob- lems involving combustion in the air than he was in the exact composition of air itself, and hence we find his records of air composition always inci- dental to other important data. Thus, in studying the properties of phosphorus,1 he reports that in a large number of experiments in which an excess of phosphorus was burned in a bell jar containing 109 pouces2 he found a diminution in volume of 20 to 21 pouces, that is, about one-fifth. Lavoisier also noted an increase in the weight of the phosphorus. Of particular interest is the experiment reported in his memoir on the respiration of animals,3 in which he heated mercury with a confined volume of oxygen. On heating 50 pouces of common air with 4 ounces of mercury, he found at the end of 12 days that the air was diminished one-sixth part. In the same year, 1777, in his re- port of the research on the combustion of a candle,4 he states that the air contains about one-fourth of its volume of "air pur et respirable." Fi- nally, in 1785, in a report of a study of the alteration of air by respiration,5 he maintained that air contains 25 per cent of oxygen. Lavoisier's experience with all known methods for absorbing oxygen justified his critical discussion of the subject6 in which he maintains that the eudiometers established on the principle which depends upon the great affinity between oxygen and phosphorus, or the alkaline sulphides, or the mixture of iron and sulphur, are much preferable to those of Priest- ley, Fontana, and Ingen-housz using nitric oxide. In the first place, it is always easy to have phosphorus, sulphur, and iron perfectly identi- cal, while nitric oxide always differs in composition. Second, the nitric oxide does not always absorb all of the vital air. Third, this gas is sus- ceptible of different degrees of oxygenation varying with the temperature and pressure, the rapidity of mixture, and the diameter of the vessel. Fourth, the nitric oxide is capable of mixing with nitrogen in all propor- tions. An excess of sulphuret of potash, iron, sulphur, or phosphorus may be used without affecting the results, but with nitric oxide an excess produces an error. Lavoisier finally decided to use sulphuret of potash, preparing the air sample over water and at the end of 15 to 20 days de- termining the contraction in volume. The importance of a general study of the oxygen content of the air on the earth is emphasized in the following paragraph from Lavoisier: II est a desirer que quelque physicien ait le courage d'entreprendre, par cette m6thode, une suite d'exp6riences sur Pair atmospherique recueilli dans difterents lieux, dans dif- 1 Lavoisier, Opuscules Physiques et Chimiques, 1773; Oeuvres de Lavoisier, 1, p. 643. 1 A cubic pouce equals 19.6 c.c. ' Lavoisier, Memoires de l'Academie des Sciences, 1777, p. 185; Oeuvres de Lavoisier, 2, p. 174. 4 Lavoisier, Memoires de l'Academie des Sciences, 1777, p. 195; Oeuvres de Lavoisier 2, p. 184. 6 Lavoisier, Recueil d. Memoires de Lavoisier, 3, p. 13; Oeuvres de Lavoisier, 2, p. 676. * Lavoisier, Recueil d. Memoires de Lavoisier, 3, p. 154; Oeuvres de Lavoisier, 2, p. 715. History of Air-Analysis 17 fe>entes saisons, dans differentes circonstances. On pourrait faire marcher ensemble des experiences correspondates par la combustion du phosphore. J'ai toujours eu le projet de me livrer a ces recherches, auxquelles j'6tais naturellement conduit par les ex- periences que j'ai faites sur la salubrity de l'air des salles de spectacle et des dortoires des hopitaux; mais je n'ai pu encore realiser mon projet. While Scheele first attempted a study of the effect of season and weather conditions on the oxygen content, we find Lavoisier empha- sizing the importance of recognizing the influence of geographical locality upon the composition of the atmosphere. Alkaline sulphides were used by Guyton1 to absorb the oxygen quan- titatively from the air; calcium sulphide was likewise employed by de Marti2 in his analyses of the air of Catalonia. Upon comparing the dif- ferent methods — nitric oxide, the Volta hydrogen eudiometer, phosphorus, moist iron and sulphur, and the alkaline sulphides — de Marti decided the last was the most satisfactory. The experiments were made during 1787, and his results are of especial interest, as they show his remarkable intuitiveness. Among other refinements, de Marti recognized the importance of sat- urating his absorbing solution with nitrogen before use. He says regard- ing his researches: The proof by sulphuret is that best calculated to ascertain the quantity of vital air contained in any gaseous fluid, since it will leave the mephitic air, and the other kinds of air which do not combine with it, without fear of any other gaseous substance being pro- duced, or any lost, except the quantity of vital air, which alone has an affinity with the sul- phuret, as I assured myself in 1787. A hundred parts of atmospheric air exposed to sul- phuret lost between 0.21 and 0.23; and as several other proofs on the same air, made with nitrous gas, had taught me that it experienced no sensible variation, I was then convinced that the air which we breathe in Catalonia is constantly composed of from 0.21 to 0.23 of vital air, and from 0.77 to 0.79 of azotic gas. To ascertain whether there might not be variations afterwards in the proportion of these two principles which constitute in the atmosphere that elastic substance on which our life chiefly depends, I continued my experiments by means of sulphuret. I repeated them so many times with atmospheric air, and on so great a number of days, that the uniformity in my results demonstrates not only the exactness of this method, but it seems to result from my observations made on the southern coast of this province: 1st, That the wind never caused the variation of a hundredth part in the respective quantities of vital air and azotic gas which compose the elastic fluid of our atmosphere, since I have always found that a hundred parts contained 79 of the latter and 21 of the former, without ever reaching 22. 2nd, That neither the moisture nor dryness of the atmosphere, nor the state of the latter in being more or less charged with exhalations, nor serene nor rainy weather, oc- casioned any difference. 3d, That the proportion of the quantities of the two same principles was equally constant during the days that Reaumur's thermometer stood at the freezing point, as well as during those when it indicated 24 degrees of heat. 1 Guyton, Chemisches Annalen, 1788, 1, p. 316; ibid, 1796, 1, p. 22. 2 de Marti, Journal de Physique, 1801, 52, p. 173; also printed in Philosophical Maga- zine, 1801, 9, p. 250. 18 Composition op the Atmosphere 4th, That I did not observe any variation in the air thus taken while the mercury of the barometer was very low, and when it exceeded 28 inches. In a word, during winter, in summer, in spring, and in autumn, in every month and at all hours, I found the air of my country, taken in the open fields, to be always composed of from 21 to 22 parts of vital air, and of from 78 to 79 of azotic gas. But though this proportion does not vary a hundredth part in the course of several months, and even years, may it vary a very small part, such as a thousandth part, which after a very long time may become sufficiently sensible to make the proportion of the vital air of the atmosphere experience a progressive or periodical increase or diminution? De Marti's researches, though carried out in 1787, were not translated into English until 1801, but his success with the sulphide of calcium evi- dently stimulated others to use the sulphides as reagents. Berger1 reports a series of experiments, using several forms of sul- phide. With potassium sulphide, he found 21.65 per cent of oxygen; with iron sulphide, 21.19 per cent; with calcium sulphide, 20.88 per cent; and with sodium sulphide, 20.38 per cent. The agreement in these deter- minations led Berger to conclude that these eudiometric substances ab- sorb from the air only one substance, namely, oxygen. Henderson,2 studying the changes which air undergoes as a result of respiratory processes, used "sulphuret of lime." On three days, June 16, 1803, June 18, 1803, and February 11, 1804, he analyzed common air and found 22 per cent of oxygen in all three cases. Gay-Lussac,3 using both the hydrogen eudiometer and the absorp- tion by alkaline sulphides, made analyses of air collected in a balloon. He found with the hydrogen-explosion method that the air at a height of 6636 meters had the same composition as that on the surface of the earth, and that at both places they gave 21.49 per cent of oxygen. With the alkaline sulphide solution, he found 21.63 per cent in the air brought down in the balloon, and maintained that this slight increase over 21.49 per cent was inside the limit of error of the apparatus. Julia de Fontanelle,4 while making a tour of Europe, analyzed over 50 samples of air in France at the foot and summit of Canigou, with an ele- vation of 2780 meters, on the Corbi£res, on the Clape, and on the plains of Roussillon and Narbonne; also in Spain on the plains of Figueras, Gironne, and Barcelona, and on the mountains of St. Jerome-D' Ebron, Mont-Joui, etc. Using calcium sulphide, he found constantly 21 per cent of oxygen, with slightly more oxygen at noon than at midnight. The last recorded use of the sulphide of calcium for air-analysis was made by Dalton,5 who, however, gives no results, and by Moyle,6 who analyzed the air of mines by this method in comparison with several others, including the long-discarded nitric-oxide method. 1 Berger, Journal de Physique, 1802, 56, p. 375. * Henderson, Nicholson's Journal, 1804, 8, p. 40. » Gay-Lussac, Annates de Chimie, 1804, 52, p. 75. 4 J.-S.-E.-Julia, Recherches historiques, chimiques et m£dicales sur l'air mare- cageaux, Paris, 1823. 6 Dalton, Philosophical Magazine, 1838, 3d ser., 12, p. 158. 8 Moyle, Annales de Chimie et de Physique, 1841, 3d ser., 3, p. 318. History of Air- Analysis 19 With the passing of the nitric-oxide1 and the alkaline-sulphide methods, we may consider the historical development of such methods as have sur- vived a century or more of keen analytical criticism. Since this article has to deal primarily with the development of the knowledge regarding the composition of the outdoor air and but secondarily with methods, we may now advantageously consider the chronological records of progress in air-analysis. In 1774, the brilliant Italian physicist, Volta,2 announced his eudio- metric method of employing the explosion of a confined volume of air with hydrogen by the electric spark. No results of his analyses are re- ported, but the process evidently attracted much attention, for we find that Cavendish,3 while working on the composition of water, published the following interesting statements : From the fourth experiment it appears that 423 measures of inflammable air are nearly sufficient to completely phlogisticate 1000 of common air; and that the bulk of the air remaining after the explosion is then very little more than four-fifths of the com- mon air employed; so that as common air can not be reduced to a much less bulk than that by any method of phlogistication, we may safely conclude, that when they are mixed in this proportion, and exploded, almost all the inflammable air, and about one-fifth part of the common air, lose their elasticity, and are condensed into the dew which lines the glass. Although Cavendish was in no sense appreciative of the fact that this series of experiments proved the accuracy of the Volta eudiometer for air- analysis, the results are surprisingly accurate. Shortly afterwards the method was adversely criticized by Seguin,4 who maintained that the apparatus gave only comparative results and could never be taken as an absolute measure, and by Berthollet,5who ob- jected to the complicated apparatus. The latter remarked on the prob- able contamination of the hydrogen by carbonaceous gases and pointed out that we do not as yet know enough about the specific weight of the two different gases. But in spite of this adverse criticism, the method was most carefully employed in analyzing samples of air brought from a great height by Gay-Lussac in a balloon flight.6 Later, von Humboldt and Gay-Lussac7 published a long research with the Volta eudiometer on the composition of air taken over the Seine under varying weather 1 It is interesting to note that in 1890 Wanklyn and Cooper resurrected the nitric- oxide method and enthusiastically recommended its use, reporting three analyses of pure air as giving 20.59, 20.54, and 20.67 per cent of oxygen respectively. Chemists have not accepted the method for modern use. (See Wanklyn and Cooper, Air-analysis, London, 1890, p. 35.) 1 Volta, Sopra Un Nova Eudiometro. Lettera al Signor Dottore Giuseppe Priestley, Como, 2 Settembre, 1777. Published in Collezione dell'Opere. del Cavaliere Conte Alessandro Volta, Firenze, 1816, 3, p. 177. Originally published, Scelta di Opuscoli interessanti di Milano, 1777, 34, p. 65. 3 Cavendish, Philosophical Transactions, 1784, 74, pp. 119-153. * Seguin, Annales de Chimie, 1791, 9, p. 293. 6 Berthollet, Memoires sur 1'Egypt publies pendant les Campagnes du Gen6ral Bona- parte. Paris, 1800(Annee8),p.2S4. 6 Gay-Lussac, Annales de Chimie, 1805, 52, p. 75. 7 v. Humboldt and Gay-Lussac, Journal de Physique, 60, p. 129. 20 Composition of the Atmosphere conditions. Although in the analyses of air collected by Gay-Lussac in the balloon, the ratio of hydrogen to oxygen was erroneously taken as 2.04 to 1, von Humboldt and Gay-Lussac used in this research the correct ratio, 2 to 1, which had but recently been established. In this paper von Humboldt acknowledges his error in advocating so strongly the nitric- oxide eudiometer in his contention with Berthollet a few years before. The analyses of the air were made on the day of collection, a sum- mary of the results being given. Table 1. — Results of a research on the composition of air, made with the Volta eudiometer, by Humboldt and Gay-Lussac. Date ■ Temper- ature. 180 Nov. 17 °C. 7.3 Nov. 18 4.5 Nov. 19 4.7 Nov. 20 10.0 Nov. 21 12.5 Nov. 22 6.7 Nov. 23 1.5 Nov. 24 8.5 Weather conditions. Wind. Nov. 25 Nov. 26 Nov. 27 Nov. 28 Dec. 1 Dec. 3 Dec. 5 Dec. 7 Dec. 13 Dec. 19 Dec. 23 10.6 3.3 1.6 1.3 4.1 2.3 4.2 3.1 9.6 2.2 1.0 Overcast ...Do Fine rain ...Do Overcast Cloudy; little rain Cloudy Rain Overcast Cloudy Frost Snow Fog Cloudy Rain Thick Fog Rain Overcast Heavy frost ; thick fog . E ESE Very strong, SW. to W, S SW SW W S SW E N N NNE E S SSW. NE. SE. . Oxygen. p. ct. 21.0 21.0 21.0 21.0 21.0 21.0 21.0 21.1 21.2 21.0 21.0 21.0 21.0 21.0 21.0 21.1 21.0 21.1 21.1 21.0 21.1 21.0 21.0 20.9 21.0 21.0 21.0 21.0 21.0 The authors conclude that they have shown, first, that the atmos- pheric air does not vary in composition; second, that there are 21 parts of oxygen in 100 parts of air; third, that there are no noticeable amounts of hydrogen present in the air. This investigation is the first extensive re- search into the composition of the atmosphere employing the hydrogen eudiometer, and the conclusions drawn by the authors are astonishingly correct when it is considered that the research was carried out over a cen- tury ago. That the errors in the apparatus were far greater than are permissible in modern research, especially when such fundamental de- ductions are to be made, should not in any way dim the brilliancy of the work of these investigators. Subsequently, the method was to have exten- History of Air-Analysis 21 sive use, be considerably increased in accuracy, and contribute materially to our knowledge of the oxygen content of the air. Henry1 analyzed atmospheric air frequently, using the Volta eudiom- eter. He reports that he was unable to satisfy himself "whether it con- tains 21 or 20 volumes of oxygen in 100, the proportion being mostly be- tween these two extremes." Simultaneously with the Volta eudiometer, another method of air- analysis was rapidly developed, which was also based upon the fundamen- tal observations of Scheele with regard to the absorption of oxygen from a confined volume of air by slowly or rapidly burning phosphorus. Scheele's experiments have already been cited, but in the English trans- lation of his book,2 we find Richard Kirwan criticizing adversely Scheele's results, maintaining that Lavoisier, when using the combustion of phos- phorus, never found more than between one-fifth and one-sixth of oxygen absorbed, while Scheele, it will be remembered, found a much larger contraction in volume. Kirwan also pointed out that Fontana had made experiments with phosphorus but found the diminution in volume much less than that found by Scheele. Later Lavoisier3 mentioned the fact that when employing the combustion of phosphorus he found the quantity of vital air contained in the atmosphere was about 27.5 parts in 100. Volta, in a letter to Priestley,4 wrote in a general way of his experience with "Bolognian phosphorus," showing that at the same time Scheele in Sweden, Lavoisier in Paris, and Volta in Italy were using phosphorus to absorb oxygen from the air. Rapidly burning phosphorus was also em- ployed by Achard,5 who described two eudiometers, one for nitric oxide and one for rapidly burning phosphorus. Dissatisfied with the incomplete descriptions and development of the earlier methods employing ignited phosphorus, Seguin6 in his memoir on eudiometry described accurately the methods used by Lavoisier and him- self, but gave no results. It is noteworthy that after the ignition of the phosphorus and the contraction in the volume of the air, they placed in the jar a little caustic alkali to absorb the carbon dioxide and the phosphoric acid. Seguin maintained that this method was very rapid and very exact. Simultaneously with his condemnation of the nitric-oxide eudiometer, Berthollet advocated the use of slow-burning phosphorus.7 In his ob- servations on eudiometry, he criticized severely the nitric-oxide eudiometer and the Volta hydrogen eudiometer, and stated that the use of alkaline sulphide is too long a process and that hydrogen sulphide is present. 1 Henry, Elements of experimental chemistry, London, 1829, 11th ed., I, p. 316. 2 Scheele, Experiments on air and fire, London, 1780, p. 202. 3 Lavoisier, Memoires de l'Academie des Sciences, 1782, p. 486; also in Oeuvres de Lavoisier, 1862, 2, p. 503. 4 Priestley, loc. cit., 1777, 3, p. 381. 5 Achard, Nouveaux Memoire de l'Academie Royale des Sciences et Belles Lettres, for the year 1778 (published 1780), p. 91. 6 Seguin, Annales de Chimie, 1791, 9, p. 293. 7 Berthollet, Memoires sur l'Egypt publics pendant les Campagnes du General Bona- parte. Paris, 1800, (Annee 8), p. 284. 22 Composition of the Atmosphere Berthollet's expressions regarding the existing condition of air-analysis methods is of special interest even at the present day: Depuis que Ton sait que l'air atmospherique est compose de gaz oxygene et de gaz azote, on a cherche" a determiner les proportions de ces deux gaz, et les variations que peuvent y survenir; mais on n'est point encore d'accord sur la methode qu'on doit preferer, et sur le r^sultat auquel on doit s'arreter. Believing the use of slow-burning phosphorus to be the best method, he passed a cylindrical stick of phosphorus into air collected over water in a glass vessel. At the ordinary temperature of Cairo it required about 2 hours for complete absorption, but in Paris he found it required 6 to 8 hours. A correction of one-fortieth for phosphorus vapor was recom- mended. In Cairo he found that the air contained generally 22 parts of oxygen, with a variation of hardly more than 0.5 part. A most interest- ing discussion of the factors affecting the composition of the atmosphere concludes his paper: En effet, comment peut-on concevoir que l'atmosphere continuellement agit^e par des mouvements qui la transportent rapidement, qui changent ses contacts et la renou- vellent, puisse varier considerablement d'un village a un autre: il y a cependant une exception a faire pour les lieux qui sont fort elevens audessus du niveau de la mer. La difference de pesanteur specifique entre le gaz oxygene et le gaz azote, qui, dans l'etat eiastique, n'exercent r6ciproquement qu'une tres faible action, explique celle qui a 6t& trouv^e dans leurs proportions. Parrot in Riga began active experimenting in phosphorus eudiometry in the latter part of 1799, and in 1800 described an apparatus which he called an oxygenometer.1 In a letter to Gilbert2 he emphasized the im- portance of temperature changes, noting that a change of 4° or 5° Reau- mur may produce a change of 1 to 2 per cent in the oxygen measurement. In his later experiments he found that the oxygen varied from 20.7 to 23 per cent. By applying a correction — not identical, however, with Berth- ollet's one-fortieth — the results were 22.25 and 24.72 per cent by volume. He concluded that the greatest variation was 2.5 per cent and that the greatest oxygen content of the air was about 25 per cent. His argu- ments for variation in the oxygen content of the atmosphere are of interest : Der Grund, den Berthollet fiir die Bestandigkeit des Sauerstoffgehalts angiebt, nam- lich die Bewegung der Luft, beweist allerdings, dass dieser Gehalt nicht sehr stark variieren kann, schliest aber Variationen von 2 bis 2| pC. nicht aus, es versteht sich, fiir sehr entfernte Orte und verschiedne Zeiten. Ein Wind, der 15 Fuss in einer Sekunde durchlauft, braucht etwa 5 Tage, um eine Strecke von 18° zu durchstreichen. Warum sollte z. B. vor einem Sudwinde die Luft in Schottland, Schweden, Norwegen, Russland nicht an Sauerstoff armer seyn, als 5 Tage nach dessen Entstehung, wenn z. B. eine iippige Vegetation, von vielem Sonnenscheine begunstigt, viel Sauerstoffgas in Italien, im nordlichen Afrika, in Griechenland entwickelt hat ? Warum sollte ein Ostwind, der uber Asiens Vegetation herkommt, nicht Europa mit mehr Sauerstoff versehen, als der Westwind, der uber das atlantische Meer herweht, wo er keine Sauerstoff-Entwickelung antrifft? 1 Parrot, Voigt's Magazine, 1800, 2, p. 154. 2 Parrot, Gilbert's Annalen der Physik,1802, 10, p. 193. History of Air-Analysis 23 F. Berger1 in Geneva, in a paper criticizing the Fontana eudiometer, mentions the phosphorus eudiometer as "introduced" by Giobert2 of Turin and "improved" by Spallanzani.3 A number of tests comparing the nitric-oxide with the phosphorus eudiometer all show the great advan- tage of the latter. Using both alkaline sulphides and the phosphorus eudiometer, he always found between 20 and 21 per cent. He analyzed air from the glacier of Mont Cervin and other glaciers, but found the air over the glacier no purer than air from the same height on a mountain. He concludes that the atmosphere is throughout the whole extent of equal composition and that the oxygen is very nearly one-fifth of the total air. The activity of numerous chemists in advocating various methods for absorbing oxygen led to the rapid accumulation of evidence in favor of the phosphorus and hydrogen eudiometers, but the latter seemed to have the most extended use. Biot,4 during a study of the air contained in the bladders of fishes, analyzed the air of two islands in the Mediterranean, Formentera and Iviza. They report that the analyses, which were made by the hydrogen- explosion method, showed consistently an oxygen content of 21 per cent. In a like research, i.e., the analysis of the air contained in the bladders of fishes, Configliachi6 used side by side the phosphorus and the hydrogen eudiometers, but apparently was more confident of results obtained with the latter. In this research, analyses were made of outdoor air from moun- tains, marshes, and grain fields. He concludes that his results, which are given in table 2, show the uniform composition of the atmosphere. Table 2. — Comparative study of the percentage of oxygen in atmospheric air, made by Configliachi. Air of lowlands. Mountain air. Grain field. Marsh. Pizzo Legnone (2642 meters) . . . St. Bernard (472 meters) Mont Cenis (2067 meters) Simplon (2006 meters) p.ct. 21.0 21.1 21.0 20.9 p. ct. 20.9 { 20.8 (20.8 \ 20.7 }20.6 20.8 p. ct. 21.0 20.9 21.0 21.0 20.9 21.9 Employing a phosphorus eudiometer, Vogel6 found in the air from the Baltic Sea between 20 and 21 per cent of oxygen, the latter figure never 1 Berger, Journal de Physique, 1802, 56, p. 253. 2 Giobert (Journal de Physique, 1798, 47, p. 197) analyzed the air of Vaudier and of Turin by the combustion of phosphorus. This refers probably to rapid rather than slow combustion. In Vaudier he found from 25 to 33 per cent of oxygen, but in Turin the variation was much less, being 26 to 28 per cent. 3 Spallanzani, in his Memoires sur la Respiration, Geneva, 1803, p. 101, states that the air we breathe contains 27 per cent of oxygen. 4 Biot, Gilbert's Annalen der Physik, 1807, 26, p. 459. 5 Configliachi, Journal fur Chemie und Physik, 1811, 1, p. 137. 8 Vogel, Gilbert's Annalen der Physik und der physikalischen Chemie, 1820, 6, p. 93. 24 Composition of the Atmosphere being attained. Kruger reported to Vogel the result of four analyses, made with a Volta eudiometer, of sea-air taken on a half mile from shore, in which he never found over 20.59 per cent of oxygen. The author's explana- tion of this low figure is that the oxygen had been absorbed by the water. Another investigation of air from the Baltic Sea was made by Hermb- stadt1 in 1821. Although his method of sampling is questionable, he asserts that he employed a "very exact" Volta eudiometer. Air taken 5 feet above the surface of the sea gave 21.5 per cent of oxygen; air 16 feet above the sea, 20.5 per cent; and air 24 feet inland from the shore, ex- actly 20 per cent. The author concludes that the larger oxygen content of air from the sea is due to the continuous evolution of oxygen from either the sea-water or marine life. Thomson,2 in Edinburgh, was occupied in air-analysis many years and, indeed, states that in 1801 he made experiments which showed that the composition of the air in Edinburgh was the same as that found by Davy and Berthollet elsewhere. In 1824 he made a new series of tests, employ- ing the Volta eudiometer. After much experimenting as to the proper volume of hydrogen to use, he found, as the average of 10 experiments, 79.9335 per cent of nitrogen and 20.0665 per cent of oxygen. Thomson's analyses and conclusions are so obviously dominated by the precon- ceived notion that air is a chemical compound consisting of four parts of nitrogen and one of oxygen, that his contribution has very little quanti- tative interest. Another Englishman, John Dalton, whose theoretical discussions were of great importance to chemists, also analyzed air upon a number of occasions. On January 8, 1825, 3 he found as a result of many experiments, 21.15 per cent of oxygen in air sampled in the country, the barometer being 30.9 inches, and the wind blowing very moderately from the north- east after 3 days of calm and a light frost. He states that ordinarily the atmosphere has only 20.7 to 20.8 per cent of oxygen. Dalton's concep- tion of the independent nature of each gas and the computed differences in composition of the air at different heights greatly stimulated research in this direction. In contradiction of his expressed views upon the solu- bility of gases in water, he collected samples by letting the water run out of a bottle and then corking the bottle. Frequently it was opened under water and allowed to stand several months. In the Philosophical Maga- zine, 1838, 12, p. 397, Dalton gives further experimental evidence to sup- port his view, but is evidently convinced that the theoretical compu- tations are not verified by experiment. Thus he states : From the experiments about to be related, I have reason to believe that the higher regions of the atmosphere are somewhat less abundant in the proportion of oxygen than 1 Hermbstadt, Journal fur Chemie und Physik, 1821, 32, p. 283. 2 Records of general science, by Robert D. Thomson, M.D., 1836, 15, p. 179. See also Journal f iir praktische Chemie, 1836, 8, p. 359. 3 Dalton, Annals of Philosophy, 1825. 10, p. 304. History of Air-Analysis 25 the lower, though the reverse might be expected from the enormous consumption of oxygen by daily processes on the surface of the earth, when we know of no proportionate consumption of azote. It appears, however, that the disproportion of the two elements at different elevations is by no means so great as theory requires; and therefore we must conclude the unceasing agitation of the atmosphere by currents and counter-currents is sufficient to maintain an almost uniform mixture at the different elevations to which we have access. His experimental evidence consists of analyses of samples from Mount Helvellyn (3000 feet), Snowdon (3570 feet), two balloon journeys made by Green at 9600 feet and 15,000 feet respectively, and three samples from Switzerland sent by Crewdson from Mer de Glace, the Simplon Pass, and the Wengern Alps. The results, together with an abstract of the many analyses of Manchester air made for comparison, are given in table 4. Referring to the results obtained from these analyses, Dalton says: The general conclusions, it seems to me, to be drawn from these experiments are, that the proportion of oxygen to azote in the atmosphere on the surface of the earth is not precisely the same at all places and times; and that in elevated regions the proportion of oxygen to azote is somewhat less than at the surface of the earth, but not nearly so much so as the theory of mixed gases would require; and that the reason for this last must be found in the incessant agitation in the atmosphere from winds and other causes. Numerous computations as to the composition of the atmosphere in higher strata, based upon Dalton's hypotheses, have been made from time to time by Babinet,1 Benzenberg,2 Bauer,3 Morley,4 and Hinrichs.5 The values computed by Morley and Hinrichs are given in table 3. Table 3. — Percentages of oxygen in high-strata air, as computed by Morley and Hinrichs. Oxygen. Oxygen. Height. Height. Morley. Hinrichs. Morley. Hinrichs. kilometers. p. ct. p. ct. kilometers. p. ct. p. ct. 0 20.96 21.00 10 18.31 18.43 1 20.68 20 15.92 16.07 2 20.41 30 13.90 3 20.14 40 11.86 4 19.87 50 10.25 9.83 5 19.60 60 7.52 6 19.34 70 4.7 7 19.07 80 2.2 8 18.82 90 0.7 9 18.56 100 4.69 0.3 1 Cited by Dumas and Boussingault, Annales de Chimie et Physique, 1841, 3d ser., 3, p. 258. 2 Benzenberg, Poggendorff 's Annalen der Physik und Chemie, 1834, 3 1 , p. 8. 3 Bauer, Poggendorff's Annalen der Physik und Chemie, 1868, 135, p. 135; also Zeitschrift fur analytische Chemie, 1869, 8, p. 397. 4 Morley, The American Journal of Science, 1879, 3d ser., 18, p. 168. 5 Hinrichs, Comptes rendus, 1900, 131, p. 442. 26 Composition of the Atmosphere Table 4. — Percentages of oxygen in analyses of air made by Dalton. Date. 1824 July 14 July 14 Nov. 23 1825 Jan. 8 June 8 June 10 Nov. 3 1826 May 14l May 14 May 182 May 182 May 18 May 18 July 1827 June 273 June 27 July 2 1828 Aug. 5 Aug. 5 1831 July 44 1832 July 26 July 27 1835 Aug. 21 B Aug. 21* Aug. 29B Aug. 296 Sept. 156 Sept. 155 No. of experi- ments averaged. 4 6 4 6 10 10 6 4 5 10 7 8 13 2 4 e4 4 64 4 64 Barom- eter. 28.0 30.94 29.90 30.30 28.76 26.20 16.8 Weather. Rain; wind SE., strong. Following a week of calm weather. Sunny and sultry; wind SW. Rainy; wind SW... . Wind NE., light .... Place. Wind SW., light Rain and fog; wind SW. Summit of Helvellyn (3000 ft.). Manchester Town air , .Do. ...Do Field near the town Town air. Summit of Snowdon, 3570 ft. above the sea. Country, 3 miles from Manchester. Summit of Snowdon . . Second bottle from Snowdon. Country , near Man- chester Town air Summit of Helvellyn . Town air Balloon voyage over Cheshire (9600 ft.). Town air , ....Do Town air Summit of Snowdon . Summit of Helvellyn . Collected from balloon; altitude, 15,000 ft. (first phial) . Town air Oxygen. Merde Glace, 6000 ft. above the sea. Simplon Pass, 6174 ft. Wengern Alps, 6230 ft. p. ct. 20.70 20.88 20.25 21.12 20.97 20.58 20.6 20.65 20.8 20.59 20.9 20.7 21.04 20.63 20.73 20.7 20.83 20.8 20.92 20.44 20.57 20.59 20.95 (20.2 )19.4 \ 19.98 1 19.53 20.45 20.11 1 Analyzed on May 28, 1826. 2 Analyzed on May 25, 1826. 3 Collected on June 26, 1827. 4 Analyzed on July 21, 1831, or 17 days later. s All analyzed in October, 1S35. 6 Duplicate. The explosion of hydrogen and air by the electric spark, as originally proposed by Volta, had been the only method of uniting hydrogen and oxygen in gas-analysis up to 1824, when Doebereiner1 announced his dis- covery of the catalytic action of platinum sponge. 1 Doebereiner, Schweiggers Journal fur Chemie und Physik, 1824, 42, p. 60. History of Air-Analysis 27 On a trip to South America in 1825, Boussingault1 made observations on the oxygen content of the air at various altitudes. In at least one analysis he used platinum sponge, for he reports that air taken in Novem- ber, 1826, at Mariquita, in the valley of the Magdalena, at an altitude of 548 meters, gave with platinum sponge 20.77 per cent of oxygen. Two analyses made with the Volta eudiometer gave results as follows : December 1826, Ibague" (1323 meters) 20.7 per cent. April 1825, Santa F6 de Bogota (2643 meters) 20.65 per cent. Boussingault concluded that his observations were not in accord with the Dalton hypothesis. Turner,2 in a paper read before the Royal Society of Edinburgh, 1824, reported his experiences with the use of spongy platinum on a mixture of air and hydrogen. Three experiments gave 21.8, 22.3, and 21.7 per cent of oxygen, respectively. Suspecting the purity of his hydrogen, he left an active ball of spongy platinum in contact with hydrogen over night and made 6 tests the next day. The results were 20.3, 20.3, 20.7, 21, 21.3, and 21.7 per cent of oxygen, respectively, the mean of these ex- periments being 20.88 per cent; he assumes 21 per cent of oxygen as the correct value. Table 5. — Percentages of oxygen obtained by Baumgartner in atmospheric air. Date. Oxygen. Date. Oxygen. Date. Oxygen. 1831 p. ct. 1831 p. ct. 1831 p.ct. Sept. 24 20.6 Oct. 4 21.1 Oct. 14 21.0 Sept. 25 21.4 Oct. 5 21.4 Oct. 15 20.9 Sept. 26 21.0 Oct. 6 21.3 Oct. 16 20.7 Sept. 27 20.9 Oct. 7 21.1 Oct. 17 21.0 Sept. 28 21.1 Oct. 8 20.9 Oct. 18 20.7 Sept. 29 21.2 Oct. 9 20.9 Oct. 19 20.8 Sept. 30 21.2 Oct. 10 20.9 Oct. 20 21.0 Oct. 1 21.4 Oct. 11 21.0 Oct. 21 21.0 Oct. 2 20.4 Oct. 12 20.7 Oct. 22 21.0 Oct. 3 21.3 Oct. 13 20.8 Degen3 in Stuttgart likewise used platinum sponge and found in out- door air 20.80, 20.88, and 20.89 per cent. Kupffer4 in Kasan, by using a Volta eudiometer and mixing 99 parts of hydrogen with 198 parts of air, found after explosion a residue of 171 to 172 parts, corresponding to 21 to 21.2 per cent of oxygen. The appearance of cholera in Vienna in 1831 led to an exhaustive study of the atmosphere by Baumgartner,5 who analyzed the air each day from September 24, 1831, to January 31, 1832, by means of the Volta eudiometer. Differences between analyses on the same sample seldom varied 0.2 per cent. 1 Boussingault, Annales de Chimie et de Physique, 1841, 3rd ser., I, p. 354. 1 Turner, abstracted in Boston Journal of Philosophy, 1825, 2, p. 238. 8 Degen, Poggendorff ' s Annalen der Physik and Chemie, 1833, 27, p. 557. * Kupffer, Annales de Chimie et de Physique, 1829, 41, p. 423. 5 Baumgartner, Medicinische Jahrbucher des k. k. osterreichischen Staates, 1832, 12, p. 83. 28 Composition of the Atmosphere The results for the first month, which are fairly representative of the remainder of the study, are given in table 5. The use of a metal to absorb oxygen was first suggested by Scheele,1 who employed metallic iron. A Spaniard, Luzuriaga,2 in 1784, used lead, but his results are not available. The first practical use of a metal as an oxygen absorbent in air-analysis was made by Theod. de Saussure,3 who employed lead shavings moistened with a very little water. After shaking them 3 hours, he found that the absorption was complete, de Saussure criticized the earlier methods, since so many divergent results were obtained by different workers, con- sidering the Volta eudiometer especially open to criticism, as there was always danger of an impurity in the hydrogen. He believed the lead method, though not so convenient, to be much more accurate. A state- ment of his results is given in table 6. Table 6. — Results obtained by de Saussure with the lead method. Place. Date. Weather. Oxygen and carbon diox- ide absorbed. Lake of Geneva .... July 18 Do Aug. 16 Street in Geneva Anc 25 Quiet and clear p. ct. 21.08 20.98 21.03 21.03 21.13 21.15 21.08 21.09 20.98 21.086 21.006 21.1 21.0 21.04 Clear; NE. wind Clear; light SW. wind Clear; light NE. wind Chambeisy Do Aug. 27 Aug. 27 Sept. 13 Sept. 13 Nov. 5 Nov. 21 Dec. 13 Dec. 24 Dec. 28 Dec. 29 Rainy; strong SW. wind .. ..Do Do. Clear; light NE. wind Lake of Geneva .... Chambeisy Do .. ..Do Overcast; calm Quiet; foggv Do. Do Do. Overcast; strong NE. wind Clear; strong NE. wind Lake of Geneva .... Average of all Partly overcast; light SW. wind . . . 21.05 .04 Carbon dioxide . . Oxygen in 100 parts of air . . . 21.01 1 Meadow 1 league from Geneva. From these results one concludes that de Saussure determined carbon dioxide and oxygen together and subsequently deducted the carbon diox- ide. His results show surprising constancy in the oxygen content of the air. Dupasquier,4 employing the alkaline ferrous hydroxide originally em- ployed by Scheele, found that normal air always gave 21 per cent. Some- what later, Brunner5 reverted to the precipitated ferrous hydroxide method, but reported no air-analyses. 1 Scheele, Air and Fire, London, 1780, p. 13. 2 See Kopp's Geschichte der Chemie, 1845, 3, p. 211. 3 de Saussure, Annalen der Physik, 1836, ser. 2, 8, p. 171. 4 Dupasquier, Annales de Chimie et de Physique, 1843, 3d ser., 9, p. 247. 6 Brunner, Poggendorffs Annalen der Physik und Chemie, 1848, Erganzungsband , 2, p. 509. History of Air-Analysis 29 THE FOUNDATIONS OF MODERN AIR-ANALYSIS. About 1840, Bunsen in Marburg, working with that marvelous tech- nique that characterized all of his chemical observations, developed to the highest degree the explosion method with hydrogen for determining oxy- gen. A preliminary description of much of his technique was published in an article by Kolbe.1 Bunsen's apparatus is there described and to demonstrate the accuracy of the apparatus several analyses of atmos- pheric air were made in 1846. These analyses with slightly corrected figures are given again in detail in Bunsen's book,2 and are reproduced in table 7. Bunsen expressed the belief that the composition of the air could be determined much more accurately if the eudiometer readings were re- peated from hour to hour. One such analysis made on May 31, 1847, gave 20.964 per cent. Table 7. — Results obtained by Bunsen with the hydrogen-explosion method. Date. Oxygen. 1846 Jan. 9 Jan. 11 Jan. 13 Jan. 14 Jan. 18 Jan. 20 p. ct. \ 20.970 \ 20.963 20.927 20.914 20.950 With another eudiometer \ 20.906 } 20.928 \ 20.927 ) 20.927 Date. 1846 Jan. 22 Jan. 24 Jan. 26 Jan. 28 Jan. 30 Oxygen. p. ct. \ 20.919 1 20.880 20.921 I 20.943 j 20.927 I 20.934 \ 20.928 } 20.911 20.889 / 20.892 Date. 1846 Feb. 1 Feb. Feb. Feb. 5 8 Oxygen. p. ct. 20.840 20.859 20.925 20.940 20.937 20.952 20.953 These figures of Bunsen marked a great advance in accuracy, and it is important to note that Bunsen's method subsequently received exten- sive use by different investigators. An extract from a letter written by Bunsen to J. J. Berzelius on November 3, 1846, is of especial interest in connection with the main problem of this memoir.3 Ich habe zunachst meine Aufmerksamkeit auf einige Fragen tiber die Zusammen- setzung der atmospharischen Luft gerichtet und befinde mich im Besitz von mehr als 300 gleichzeitig in Marburg, Copenhagen, Reykjavik, in der Nahe des Polarkreises und auf dem atlantischen Ocean aufgefanger, und in zugeschmolzenen Glasgefassen be- wahrter Luft proben, die zu Analysen nach einer Methode ausreichen, deren Scharfe und Sicherheit kaum etwas zu wunschen ubrig lasst, wie die nachstehenden zur Priifung dieser methode in verflossenen Winter angestellten Versuche beweisen. 1 Kolbe, "Eudiometer, Eudiometrie" in the Handworterbuch der Chemie, Liebig, Poggendorff and Wohler, 1842, 2, p. 1050. 2 Bunsen, Gasometrische Methoden, Braunschweig, 1857, p. 77. • Gesammelte Abhandlungen von R. Bunsen, Leipzig, 1904, 2, p. 4. 30 Composition of the Atmosphere Unfortunately nowhere in Bunsen's subsequent publications do we find any record of the analyses of this large number of samples of air, and obviously the pressure of other work prevented his carrying out this inquiry. It is greatly to be regretted that with his masterful tech- nique such analyses could not have been made. Although Lavoisier had shown that when phosphorus was burned in air there was an increase in weight corresponding to the diminution in volume of the air, nevertheless no air-analyses were based upon gravi- metric determinations until the appearance in 1833 and 1834 of the unique method of Brunner1 in Berne. Brunner devised a plan of passing a volume of air through a tube that contained some suitable absorbent for oxygen which could be weighed. All previous determinations had been made over water, or occasionally mercury, upon relatively small volumes of air confined in glass tubes, eudiometers, etc., but with Brunner's method, a considerably larger volume of air could be used. Furthermore, it was possible by this process to measure likewise the amount of nitrogen remaining in the gas, and thus make a determination not only of oxygen by weight, but of nitrogen by volume. After a number of preliminary experiments made with iron and with copper, Brunner finally decided upon phosphorus as the best absorbent. With perfectly dry phos- phorus and a very moderate air-current, he found that oxygen was rapidly and quantitatively absorbed. In 1833 Brunner made a series of experiments in Berne in which he determined the average oxygen content of the air as 21.0705 per cent. The agreement was usually within 0.1 per cent, although occasionally the variation was as high as 0.2 per cent. Of interest, also, is the fact that he analyzed air taken on the Faulhorn on July 18, 19, and 20 of the same year; from 14 determinations he found the oxygen varying from 20.75 to 21.11 per cent, the average of all being 20.915. Eight years later, in July 1841, 2 Brunner made 7 experiments in the same manner as those made at Berne, and found ranges from 20.75 to 20.867 per cent, with an average of 20.821 per cent. The fact that this latter value is very much less than those found 8 years before is explained by Brunner on the ground that there was probably an error in the measurement of the size of the vessel used in the earlier experiments. Brunner's article is particularly valuable, as it contains a critical discussion of methods and of the limit of accuracy of the various methods that had been proposed for absorbing oxygen. While Brunner had successfully weighed the oxygen absorbed from the air, he had always measured the volume of nitrogen. In 1841 Dumas and Boussingault3 published a research which in plan was quite similar to that of Brunner, except that they not only weighed the oxygen but like- 1 Brunner, Poggendorff's Annalen der Physik und Chemie, 1833, ser. 2, 27, p. 1 ; also, ibid., 1834, ser. 2, 31, p. 1. 1 Brunner, Annales de Chimie et de Physique, ser. 3, 1841, 3, p. 305. 3 Dumas and Boussingault, Annales de Chimie et de Physique, ser. 3, 1841, 3, p. 257. History of Air-Analysis 31 wise the nitrogen. By using large glass vessels which could be evacuated and passing the air over heated metallic copper, they absorbed the oxy- gen by the copper, weighing the vessel before and after absorption, thus giving a true weight of the nitrogen left behind. The method was ob- viously best used by its illustrious devisers, since it was much more tech- nical and difficult to carry out than any previously suggested. By means of two apparatus, simultaneous experiments were made in 1841. The percentage volumes of oxygen found are given in table 8. These samples were all taken during very clear and beautiful weather, and as a control a sample was taken on May 29, 1841, during rain. The result was essen- tially that found during the clear weather, viz, 20.817 per cent of oxygen. Table 8. — Percentages of oxygen in air analyzed by Du mas and Boussingault. Date. Fi,rst. Sc(ion.d analysis. analysis. Average. 1841 Apr. 27 Apr. 28 Apr. 29 p. ct. 20.73 20.83 20.83 p. ct. 20.73 20.88 20.84 /). ct. 20.73 20.S6 20.83 Average . . 20.80 20.82 20.81 Although the experimental evidence of Gay-Lussac and von Hum- boldt, as well as the earlier observations of Boussingault in South America, agreed perfectly with the more recent work of Dumas and Boussingault, nevertheless, owing to the importance of the Dalton hypothesis, Dumas and Boussingault decided upon making some analyses of air taken from the Faulhorn. By previous arrangement with Brunner in Berne, a series of experiments was planned in which samples of air would be taken simul- taneously in Paris, Berne, and on the Faulhorn. These latter were col- lected by means of large evacuated glass balloons which were sent by Dumas from Paris. In Berne, Brunner operated with his method pre- viously described. The results of these comparisons, which represent the first cooperative investigation of any magnitude on the composition of the air, are given in table 9. Table 9. — Percentages of oxygen in air analyzed by Dumas and Boussingault, and by Brunner. Date. Paris. Faulhorn. Berne. July 20 July 21 July 24 Aug. 7 Average . . p. ct. 20.85 20.80 20.87 p. ct. 20.77 20.89 \ 20.76 I 20.67 20.78 p. ct. 20.80 20.70 20.78 20.77 20.84 20.78 20.76 While there is a difference between the samples in Paris and the sam- ples in Berne and on the Faulhorn, it is important to note that somewhat 32 Composition of the Atmosphere later some 50-liter samples, which were taken in Paris and analyzed with the very greatest care, showed variations equally as great; thus, on September 20 the percentage was 20.875, while on September 22 it was 20.709. Brunner's method was used successfully by Verver1 in Groningen in May and August 1838. As a result of 45 analyses, he found the average oxygen content of carbon-dioxide and water-free air to be 20.864 per cent. The oxygen percentages obtained by Verver are given in table 10. Table 10. — Percentages of oxygen in air analyzed by Verver. Date. Oxygen. Date. 1 Oxygen. Date. Oxygen. 1838. p. ct. 1S38. p. ct. 1838. p. ct. May 21 21.00 May 24 21.09 Aug. 2 20.96 May 21 20.70 May 24 20.88 Aug. 3 20.70 May 21 20.80 May 24 20.76 Aug. 3 20.80 May 21 20.85 May 25 21.10 Aug. 3 21.05 May 22 20.79 May 25 21.06 Aug. 3 20.90 May 22 21.00 May 25 21.02 Aug. 3 20.80 May 22 20.77 May 25 21.06 Aug. 5 20.60 May 22 20.95 Aug. 1 20.90 Aug. 5 20.70 May 23 20.93 Aug. 1 21.08 Aug. 5 20.65 May 23 20.91 Aug. 1 20.90 Aug. 5 20.64 May 23 20.85 Aug. 2 20.91 Aug. 5 20.90 May 23 20.90 Aug. 2 20.80 Aug. 5 20.90 May 23 20.94 Aug. 2 20.80 Aug. 6 20.70 May 24 20.82 Aug. 2 20.90 Aug. 6 20.70 May 24 20.67 Aug. 2 20.80 Aug. 6 20.70 Dumas and Boussingault were at first inclined to believe that since their results, those of Berger, Gay-Lussac, and von Humboldt, the earlier results of Boussingault, and the results of Brunner, all agreed so remark- ably, the composition of the atmosphere was uniform. Subsequent ex- periments made in combination with Brunner on the air from the Faul- horn, Paris, and Berne showed that there were also differences; they ac- cordingly modified their conception and said that although on the whole the composition of the air was constant, nevertheless there must be cer- tain variations which might occur. The demonstrated accuracy of the method of Dumas and Boussingault led to its extensive use by other observers. In 1841 and 1842 Lewy made analyses of air from the North Sea, the court of the Polytechnic School, Copenhagen, and the coast at Elsinore.2 In November and December of 1841, five analyses of air in Copenhagen were made, giving 20.82, 20.83, 20.78, 20.81, and 20.84 per cent of oxygen, respectively, with an average of 20.816 per cent. The author points out that these results agreed per- fectly with those obtained by Dumas and Boussingault in their analyses of the air collected in Paris and on the Faulhorn, as well as with those obtained by Stas in Brussels, Marignac in Geneva, Brunner in Berne, and Verver in Groningen. 1 Verver, Bulletin des Sciences Physiques et Naturelles en N6erlande, 1840, p. 191. 2 Lewy, Annales de Chimie et de Physique, 1843, ser. 3, 8, p. 425. History of Air-Analysis 33 During August 1841, on a journey from Paris to Copenhagen, he collected samples on the North Sea, the results being 20.46, 20.42, 20.45, and 20.43 per cent, with an average of 20.44 per cent. His low values may possibly be ascribed to an error in weighing the glass balloon used. Analyses were also made of air taken at Elsinore on the coast. Three samples taken on February 18, 1842, gave 20.84, 20.83, and 20.84 per cent respectively. The average was 20.837 per cent. On the return journey from Copenhagen to France, 5 samples of air taken on the sea were collected, the percentages of oxygen being 20.88, 20.91, 20.89, 21.01, and 20.84, with an average of 20.907. Lewy also reports analyses of samples taken at Guadeloupe and ana- lyzed in Paris. The results show abnormally high carbon-dioxide values and low oxygen contents. When computed on the basis of carbon- dioxide-free air, the oxygen content in 9 samples varies from 20.51 to 20.93 per cent. The author is inclined to attribute the exceptionally high carbon-dioxide content to the nearby volcanoes. In Geneva, Marignac1 found on three different days, in January and February 1842, 20.81, 20.80, and 20.77 per cent of oxygen, with an average of 20.799 per cent. Meanwhile, Stas2 in Brussels found in 12 different experiments during 1842 a minimum of 20.84 per cent and a maximum of 20.87 per cent, but the author points out two analyses with no error that could be accounted for in which the results showed 20.90 and 20.93 per cent. In general, however, Marignac, Lewy, and Stas, all using the method of Dumas and Boussingault, obtained results that agreed with those obtained by the latter investigators. In studying respiration, Marchand3 reported air-analyses made in Halle at 10 different times. The method was similar to that of Du- mas and Boussingault. The percentages of oxygen obtained were as follows: per cent. 1 20.99 2 20.97 3 20.98 4 20.90 5 20.96 per cent. 6 20.89 7 20.98 8 20.99 9 21.02 10 21.03 The average of these values, 20.97 per cent, was employed by him as indicating the composition of normal air. Two years later, Marchand4 published two analyses of outdoor air in Halle which were made by the hydrogen-explosion method with results as follows: 8 a.m., 20.920 per cent; 8 p.m., 20.912 per cent. In the attempt to establish some relationship between the com- position of the air and the invasion of cholera, at least two investi- gations on the oxygen content of the atmospheric air during a cholera 1 Marignac, reported by Dumas in Comptes rendus, 1842, 14, p. 380. 2 Stas, Comptes rendus, 1842, 14, p. 570. s Marchand, Journal fur praktische Chemie, 1848, 44, p. 1. 4 Marchand, Journal fur praktische Chemie, 1850, 49, p. 449. 34 Composition of the Atmosphere epidemic are reported in the literature,1 i.e., those of Baumgartner2 and Laskowsky.3 Employing Brunner's method, Laskowsky in Moscow made a num- ber of analyses of air during the cholera epidemic. The results, which are given in table 11, show a variation of 0.16 per cent of oxygen, which the author maintains is not far from the results of Dumas and Boussingault, who report variations of 0.19 per cent. He concludes that the air in Moscow during the time of the cholera was normal. Table 11. — Air-analyses made by Laskowsky in Moscow during the cholera in 1847. Date. Time. Oxygen. Date. Time. Oxygen. 1S47 Nov. 3 Nov. 3 Nov. 4 Nov. 4 Nov. 7 Nov. 7 Nov. 8 Nov. 8 Nov. 9 Noon Evening Noon Evening . . Noon Evening Noon Evening . . Noon p. ct. 20.88 20.76 20.73 20.85 20.89 20.S2 20.82 20.76 20.88 1847. Nov. 9 Nov. 10 Nov. 10 Nov. 11 Nov. 11 Maximum . Minimum . Average Evening . . Noon Evening . . Noon Evening . . p. ct. 20.77 20.S0 20.87 20.82 20.77 20.89 20.73 20.82 Adding oxide of manganese to the reduced copper of the Dumas and Boussingault method to increase its absorbing action, Deville and Gran- deau4 determined the oxygen in a number of samples of air for May and June of 1859, and found as an average 20.88 per cent. An ingenious use of an ammoniacal solution of copper chloride as an absorbent for oxygen was made by Doyere5 in connection with studies on respiration. As the result of several analyses by this and other meth- ods, Doyere reports that the percentage of oxygen in air is about 20.5 to 20.7, but the research does not inspire confidence. Employing an exceptionally accurate hydrogen eudiometer designed primarily for use in their classical respiration experiments, Regnault and Reiset reported in a preliminary communication in 18486 a large number of analyses that were made during the three preceding years, 1845, 1846> and 1847, in Paris, as well as near Dieppe. All of these analyses showed an oxygen content between 20.85 and 20.97 per cent. In the report of their experiments on the respiration of animals,7 the authors give a com- 1 The early observations of Davidson, who found 67 per cent of vital air in Mar- tinique during an epidemic of yellow fever, are only of historic interest. (Cited by Russell, Transactions of the Sanitary Institute, 1893, 13, p. 232.) 2 Baumgartner, loc. cit. 3 Laskowsky, Liebig's Annalen der Chemie und Pharmacie, 1S50, 75, p. 176. 4 Deville and Grandeau, Comptes rendus, 1859, 48, p. 1103. 5 Doyere, Annates de Chimie et de Physique, 1850, 3d ser., 28, p. 5. Among the innumerable observations on the composition of the air made by Scheele, it is interesting to note that he found that one-third of the air was absorbed by an ammoniacal solution of copper. (See Scheele, Efterlemnade bref och anteckningar. Edited by A. E. Nor- denskiold, Stockholm, 1892, p. 58.) 6 Regnault and Reiset, Comptes rendus, 1848, 26, p. 6. 7 Regnault and Reiset, Annales de Chimie et de Physique, 184S, 3d ser., 26, p. 299. History of Air-Analysis 35 plete description of their eudiometer, and report 6 air-analyses made from one sample of carbon-dioxide-free air, in order to show the accuracy of the apparatus. The results obtained were 20.936, 20.940, 20.932, 20.960, 20.946, and 20.941 per cent of oxygen, the variation being 0.028 per cent. Table 12. — Regnault's analyses of air collected in Paris. Date. 1847. Dec. 24 Dec. 24 Dec. 24 Dec. 28 Dec. 28 Dec. 29 Dec. 29 Dec. 30 Dec. 30 Dec. 31 Dec. 31 Dec. 31 Dec. 31 Dec. 31 Dec. 31 Place. Oxygen. Balcony, College of France .Do ...Do Top of Pantheon . . . Top of Pantheon (snow) Observatory, College of France Place de la Concorde Observatory, College of France \ .Do 184S. Jan. Jan. Do Do } Top of the Pantheon Choisy-le-Roi Esplanade de Vin- cennes . . ..Do Over a cornfield . $ of Court, College France Observatory, College of France p. ct. 20.987 20.952 20.957 20.999 20.962 20.963 20.956 20.939 20.953 20.948 20.939 20.930 20.984 20.967 20.949 20.959 20.966 20.945 20.947 20.980 20.992 20.913 20.934 Date. Place. Oxygen. 1S4S. Jan. 4 Jan. 6 Jan. 7 Jan. 8 Jan. 8 Jan. 9 Jan. 10 Jan. 11 Jan. 12 Jan. 12 Jan. 13 Jan. 14 Jan. 14 Jan. 15 Jan. 15 Jan. 15 Jan. 15 Jan. 15 Jan. 15 Jan. 15 Jan. 15 Observatory, College of France ....Do ....Do ....Do Choisy-le-Roi Observatory, College of France ...Do ....Do ...Do Choisy-le-Roi p. ct. 20.929 20.948 20.943 20.956 20.948 20.981 20.948 20.957 20.963 20.947 Observatory, College 20.970 of France 20.968 ...Do 20.952 Pantheon 20.953 Observatory, College of France 20.986 Versailles I 20.936 ....Do 20.922 ....Do 20.948 .Do .Do .Do .Do 20.954 20.992 20.993 20.998 20.952 It is to Regnault that we are indebted for the first extensive inter- national investigation on the composition of the air. This carefully planned cooperative investigation was in part disturbed by the political differences in Europe during 1848, but in spite of the difficulties encoun- tered, Regnault finally succeeded in getting together a large number of samples, which he analyzed.1 Air was collected in Paris and other parts of France, in Berlin, Madrid, Switzerland, on the Mediterranean Sea, on the Atlantic Ocean, in Ecuador, on the coast of Africa, in India, and on the Pacific and Arctic Oceans, thus representing by far the most extensive investigation on the composition of atmospheric air that has ever been undertaken. Indeed, it may be stated that no subsequent investiga- tion has approached it in completeness with regard to the geographical collection of samples. Regnault made arrangements to have these sam- 1 Regnault, Annales de Chimie et de Physique, 1852, 3d ser., 36, p. 385. 36 Composition of the Atmosphere pies collected on the 1st and 15th of each month at about the hour of true midday in each place. The samples were then sent to Paris, where they were analyzed by explosion with hydrogen in the apparatus used for analyzing the samples of Paris air. About 100 analyses were made of air collected in Paris or its suburbs, the larger number being taken at the observatory of the College of France. A part of the results are given in table 12, showing the variations on the different days and the accuracy of the analyses made on the same sample as indicated by the close agreement of the duplicate determinations. The smallest amount of oxygen found was 20.913 per cent, the largest 20.999 per cent, and the general average, 20.96 per cent. The extreme difference, 0.086 per cent, is greater than the errors resulting from the experiments themselves, for this rarely exceeds 0.02 per cent. Regnault concludes that the absolute change is so small that one can easily attrib- ute this to local alterations which would be frequently found in the center of large cities. He notes further that the variations found at different hours of the same day are no smaller than the variations between days. The results of the analyses of air collected at other points in France are given in table 13. Table 13. — Regnault's analyses of air collected in France outside of Paris. Montpellier. Lyon. St. Martin-aux-Arbres, IS ormandy. Date. Oxygen. Date. Oxygen. Date. Oxygen. 1848. Feb. 1 . . . . Feb. 15 ... . Mar. 15 Apr. 1 . . . . Apr. 15 ... . p. ct. \ 20.929 I 20.948 20.962 \ 20.959 ( 20.968 20.940 20.952 1848. Feb. 1 . . . . Feb. 15 ... . Mar. 15 ... . p. ct. 20.966 20.930 20.918 1848. Feb. 29 ... . p. ct. 20.952 Table 14. — Regnault's analyses of air collected in Berlin. Date. Oxygen. Date. Oxygen. Date. Oxygen. 1848. Feb. 1 . . . . Feb. 15.. .. Mar. 15 ... . Apr. 1 . . . . Apr. 15 ... . May 1 . . . . May 15 June 15 ... . July 1 . . . . July 15 ... . p. ct. 20.967 20.959 20.956 20.958 20.903 \ 20.941 ( 20.953 20.937 { 20.943 I 20.933 20.908 20.903 1848. Aug. 16 Sept. 15 ... . Oct. 1 . . . . Nov. 2 . . . . Nov. 15 ... . Dec. 1 . . . . Dec. 15 ... . 1849. Jan. 1 . . . . Feb. 3 . . . . p. ct. 20.910 20.943 20.976 20.986 20.936 20.946 20.996 20.919 20.981 20.962 20.973 1849. Feb. 15.... Mar. 15.. .. Apr. 1 . . . . Apr. 16. . . . May 3 . . . . May 17 June 3 . . . . June 15 ... . July 1 . . . . p. ct. 20.993 20.980 20.962 20.947 20.976 20.967 20.967 20.966 20.998 Samples were also collected at noon in Berlin by Magnus from Feb- ruary 1, 1848, to July 1, 1849; at the Madrid Observatory in 1848; and at History of Air-Analysis 37 different places in Switzerland; the results are reported in tables 14, 15, and 16, respectively. Table 15. — Regnault's analyses of air collected at Madrid. Date. Oxygen. Date. Oxygen. 1S4S. Feb. 15 ... . Mar. 1 May 15 June 15 ... . July 15 ... . Aug. 1 . . . . p. ct. 20.922 20.953 20.973 20.916 20.924 20.974 1S48. Aug. 15 ... . Sept. 1 . . . . Sept. 15 ... . Oct. 1 . . . . p. ct. 20.963 ( 20.982 \ 20.964 20.975 20.970 Table 16. — Regnault's analys es of air collected in Switzerland. Date. Place. Oxygen. Date. Place. Oxygen. 1848. Jan. 15 Feb. 1 Feb. 1 Apr. 1 Apr. 1 June 15 July 1 July 15 Aug. 1 Aug. 15 • Mont Saleve \ Obs. of Geneva. . . j Mont Saleve . . j Obs. of Geneva Mont Saleve Obs. of Geneva . ..Do p. ct. 20.940 ; 20.953 20.946 20.935 20.963 20.957 20.920 20.928 20.956 20.903 20.935 20 937 | 20.961 1848. Sept. 1 Sept. 6 Sept. 8 Oct. 1 Nov. 1 Nov. 15 Dec. 1 Dec. 15 1849. Jan, 15 Feb. 1 Feb. 15 Montanvert (valley of Chamounix) . . . Mont Buet (Savoy) . Do p. ct. 20.924 20.963 20.930 20.981 20.969 20.990 20.955 20.913 20.959 20.993 20.982 Obs. of Geneva Do Do Do . . ....Do Do Do ....Do Do ....Do The percentage of oxygen in these different samples of air collected in various parts of Europe ranged from 20.903 to 21, that is to say, it had approximately the same limit of variation as the air collected in Paris. Regnault fortunately secured the cooperation of a number of travelers, and hence was able to extend his observations considerably. The results are given in tables 17 to 20. Table 17. — Analyses of air collected on the Mediterranean Sea. Date. Place. Oxygen. Date. Place. Oxygen. 1851. May 4 May 20 May 21 May 22 May 22 Mav24 May 25 May 27 May 27 At sea, SE.of Minor- ca, lat. 38° 18' N.; long. 1°16' E Toulon harbor Do p. ct. 20.970 20.912 20.931 20.951 20.970 20.950 20.960 20.854 20.872 20.979 1851. May 28 June 8 June 5 June 9 June 27 June 28 June 29 June 30 Toulon harbor At sea, 70 miles NNE. of Algiers . At sea, S. of Minor- ca, lat. 39°0'N.; long. 1°32' E. . . . Toulon harbor .... . ..Do p. ct. 20.935 20.961 20.420 20.395 20,927 20.982 20.928 20.955 20.964 ..Do .. ..Do ....Do ...Do. ....Do j Do .Do ....Do 38 Composition of the Atmosphere Table 18. — Analyses of air collected on the Atlantic Ocean, between Liverpool and Vera Cruz. Date. 1848. Apr. 1 Apr. 26 May 11 May 2 May 7 Place. At sea, lat. 34° 21'; long. 24° 40'. Near St. Domingo, lat. 20°; long. 72° 30' . Vera Cruz Entrance to Gulf of Mexico, lat. 21° 50'; long. 88° 40'. Oxygen. p. ct. \ 20.922 ( 20.920 \ 20.920 '/ 20.918 20.962 20.953 Bay of Vera Cruz 20.965 Table 19. — Analyses of air collected at different parts of the East Indian Ocean. Date. 184S July !. 5 Sept. 15 184C Jan. I. 15 Feb. 1 Mar. 8 Mar. 15 Mar. 24 Aug. Dec. 25 15 1850 Mar. 19 Place. Oxygen. Bay of Goree (Senegal) Lat. 33° 40' S.; long. 16° 15' E. Long. 78° 38' E.; lat. 2° 29' S Gulf of Bengal; lat. 9° 4' N.; long. 83° E.1 . On the Ganges, near Calcutta, noon.2 Calcutta On the Hoogly River (East Indies), opposite Ked- gerre, lat. 21° 53' N Mayotte, off Mozambique Simons Bay (Cape of Good Hope) Mers-el-Kebir (coast of Africa) p. ct. 20.S96 20.843 20.854 20.975 \ 20.460 I 20.453 \ 20.390 I 20.387 20.866 i 20.920 - 20.921 ( 20.928 20.910 20.936 20.870 1 The aircontained 0.057 per cent of carbon dioxide. 2 A note in connection with this sample says on March 8 there was a sudden invasion of cholera and no samples were taken until March 15. The weather was excessively foggy during the night, and the fogs did not disappear during the day; the air was full of putrefying vegetable and organic matter, and there were many dead bodies in the river. The air contained 0.133 per cent of carbon dioxide. Two exceptions to the usual results appear in samples of air taken in Toulon Harbor on May 27 at 8h 30m a. m., duplicate analyses giving 20.854 and 20.872 per cent of oxygen, respectively. These numbers are appreciably less than the minimum which was obtained in the air of Paris* but the air collected on June 5 at 11 p. m. in the port of Algiers gave even lower results, these being 20.420 and 20.395 per cent, respectively. The author does not question the sealing of the tubes, as it was done by a person who had had experience in his own laboratory. In Ecuador two samples were taken; one which was collected in the village of Guallabamba on August 3, 1848, at 8h 15m in the morning gave 20.960 per cent; the other, taken on the summit of Pichincha — a mountain higher than Mont Blanc— on May 15, 1849, at 12h 45m p. m., gave 20.949 and 20.988 per cent, respectively. Among the analyses of air collected on the East Indian Ocean (see table 19), only two show a composition very different from that of normal History of Air-Analysis 39 air. The analysis of air collected on February 1, 1849, in the Gulf of Bengal, gave 20.460 and 20.453 per cent of oxygen. The notes which accompany this sample do not present anything of distinct interest. The air collected on the Ganges on March 8, 1849, shows 20.390 and 20.387 per cent of oxygen, and Regnault maintains that the conditions set forth in the notes accompanying this sample explain this anomalous case. Four other samples taken in 1852 at different points in the Pacific and South Atlantic Ocean gave 21.015, 20.935, 20.950, and 20.963 per cent of oxygen, respectively. Samples of air from the Arctic Ocean were also analyzed, these being given in table 20. Table 20. — Analyses of air collected on the Arctic Ocean. Date. Place. North latitude. East longitude. Oxygen. 184S. 0 i o p. ct. June 1 Cape Farewell m 10 39 14 20.91 June 20 July 1 Oct. 15 Whale Island 67 70 73 5 20 52 55 9 55 30 90 12 20.91 20.92 20.93 Black Hook Port Leopold Oct. 15 ....Do 73 52 90 12 20.93 Nov. 15 ....Do 73 52 90 12 20.85 1849. Jan. 15 Port Leopold 73 52 90 12 20.91 Mar. 1 ....Do 73 52 90 12 20.89 Mar. 15 ....Do 73 52 90 12 20.86 Mar. 15 ....Do 73 52 90 12 20.86 Apr. 15 .. ..Do 73 52 90 12 20.90 Apr. 15 ....Do 73 52 90 12 20.94 May 1 .. ..Do 73 52 90 12 20.93 May 1 ....Do 73 52 90 12 20.91 Aug. 1 ....Do 73 52 90 12 20.87 Aug. 15 ....Do 73 52 90 12 20.94 Aug. 15 ....Do 73 52 90 12 20.94 Regnault's conclusions are: (1) that the air of our atmosphere usually presents variations in composition which are sensible but very small, for the percentage of oxygen varies generally only from 20.9 to 21.0, although in certain cases, apparently more frequent in the warm countries, the proportion of oxygen may fall to 20.3 per cent; (2) that the average per- centage of oxygen contained in the air of Paris during the year 1848 was 20.96. The importance of a geographical study of the composition of the atmosphere was so keenly felt that the French Academy commissioned Lewy1 to make analyses of air while on a voyage to South America. The analyses were made by the method of Regnault and Reiset. Pre- vious to taking this journey, he made several analyses of air in France. On September 6, 1847, Lewy found as a result of three analyses of air in Paris, 21.018, 21.015, and 21.008 per cent, with an average of 21.014 per cent. In Havre, he found on November 22, 1847, that in three analyses the percentages of oxygen were 20.895, 20.880, and 20.888, with an average of 20.888 per cent. 1 Lewy, Annales de Chimie et de Physique, ser. 3, 1852, 34, p. 5. 40 Composition of the Atmosphere The analyses of the atmospheric air collected on the Atlantic Ocean and the Caribbean Sea are reported in table 21, the results giving an average of 0.046 per cent for carbon dioxide and 21.03 per cent for oxy- gen. On examining his figures, Lewy found that the composition of the air collected during the day differed from that collected during the night, the air of the day being richer in carbon dioxide and oxygen. He accord- ingly averaged his results for both the day and the night, obtaining the following values: for the day, carbon dioxide 0.053 per cent, oxygen 21.06 per cent; for the night, carbon dioxide 0.0346 per cent, oxygen 20.97 per cent. Table 21. — Analyses of air collected on the Atlantic Ocean and the Caribbean Sea. Average of 3 analyses of each sample. Date. Latitude N. Longitude W. from Paris. Carbon dioxide. Oxygen. 1847. o / O ' p. ct. p. ct. Dec. 1 47 30 10 5 0.0488 21.05 Dec. 4 47 00 13 0 .0334 20.96 Dec. 8 35 40 20 35 .0550 21.06 Dec. 17 22 5 39 0 .0577 21.06 Dec. 18 21 45 41 3 .0335 20.96 Dec. 18 21 9 42 25 .0542 21.06 Dec. 19 20 35 43 35 .0339 20.96 Dec. 26 15 49 64 28 .0529 21.06 Dec. 28 14 6 70 4 .0509 21.06 Dec. 30 12 5 76 0 .0514 21.06 Dec. 31 .0377 21.01 Lewy concludes that the difference between the oxygen and carbon dioxide in the atmospheric air on the ocean and in normal air becomes more noticeable the farther away one goes from the land. He further says that it is impossible to attribute this difference to errors in analysis, as he maintains that two analyses of air do not differ by more than 1 part in 10,000. According to his interpretation of the results, this difference is caused by the fact that the water of the ocean gives off carbon dioxide and oxygen. During the day the surface of the sea becomes warmed by the sun's rays and a part of these gases is dissolved; during the night, on the contrary, this influence is not felt. Results of analyses of air samples collected in New Granada on a trip from Santa Marta to Montserrat are given in table 22, while the results obtained from air collected in 1850 in Bogota, at a height of 2645 meters, with an average barometer of 565 mm., are reported in table 23. The high values for carbon dioxide are attributed to the presence of volcanoes 3 or 4 hours distant from Bogota. In a general resume of the subject, the author concludes that by ex- amining these results (tables 21 to 23), one sees that the composition of the air up to a height of about 3000 meters is nearly the same in the New History of Air-Analysis 41 World as in the Old World, but not absolutely constant. The increase in the percentage of carbon dioxide, which rose as high as 0.49 per cent — 14 or 16 times the normal percentage — is attributed by Lewy to the in- fluence of large forest fires and the presence of volcanoes. Table 22. — Analyses of air from New Granada. Date. 1848. Jan. 25 Feb. 7 Feb. 18 Mar. 3 Mar. 29 Aug. 5 Aug. Apr. Aug. July July Place. Elevation. Average of 3 analyses of each sample. Carbon dioxide. Oxygen. Santa Marta . . . Mompox Magdalena River ...Do Honda Ambalema Esperanza Guaduas Santa Ana Bogota Montserrat .... meters. 0 38 242 282 396 996 998 2645 3193 p. ct. 0.046 .031 .033 .046 .032 .112 .245 .031 .123 .050 .052 p. ct. 21.02 21.05 21.03 21.00 20.99 20.55 20.33 21.00 20.54 21.03 20.99 Table 23. — Analyses of air from Bogota (2645 meters). Date. Carbon dioxide. Oxygen. Date. Carbon dioxide. Oxygen. 1850. p. ct. p. ct. 1850. p. ct. p. ct. Mar. 7 0.0386 21.02 Sept. 2 0.1704 21.03 Apr. 12 .0366 21.00 Sept. 3 .1585 21.02 May 8 .0361 20.99 Sept. 3 .4896 21.03 May 9 .0382 20.99 Sept. 3 .4904 21.03 June 15 .0419 21.00 Sept. 4 .1326 21.03 July 24 .0425 21.02 Sept. 4 .0865 21.00 Aug. 19 .0504 21.01 Sept. 8 .1283 21.02 Aug. 23 .0481 21.02 Sept. 9 .0751 21.03 Sept. 1 .0618 21.02 Sept. 10 .0458 21.03 Sept. 2 .0765 21.02 Sept. 12 .0471 21.03 Sept. 2 .1629 20.97 Oct. 3 .0475 21.02 The hydrogen-explosion method and the weighing of the amount of oxygen absorbed by a metal or by phosphorus involved an exceedingly elaborate technique and restricted the number of observations that could possibly be made by one man in a day. In 1851 there appeared the de- scription of a method of absorbing oxygen by means of an alkaline solu- tion of pyrogallic acid. Liebig1 reported the results of 11 analyses, with percentage values as follows: 20.99, 20.89, 21.03, 20.95, 20.77, 20.92, 20.90, 20.80, 20.75, 20.76, 20.93. With this method he maintained that in an hour half a dozen analyses could be carried out. Gallic acid could also be used instead of pyrogallic acid, although, according to Liebig, the absorption required a longer time than with pyrogallic acid, i.e., lh to 2 hours instead of as many minutes. Liebig gives the results of 8 analyses 1 Liebig, Annalen der Chemie und Pharmacie, 1851, 77-78, p. 107. 42 Composition of the Atmosphere obtained with gallic acid with percentages of oxygen as follows: 20.59, 20.69, 20.97, 20.52, 21.35, 20.80, 20.78, and 21.19. This method was destined to become one of the most important meth- ods for technical gas-analysis that had ever been devised. It is interesting to note that at this date (1912) practically all of the exact gas-analysis apparatus employ the alkaline solution of pyrogallol. Almost imme- diately after the introduction of the method it was found that under certain conditions the interaction of air with pyrogallic acid and potas- sium hydroxide resulted in the formation of slight quantities of carbon monoxide. This for a time discouraged observers from using the method. Subsequently conditions were so adjusted as to minimize this apparent error. Using a modification of the Regnault method, Frankland and Ward1 published 6 analyses of air as an indication of the accuracy of the ap- paratus. The results given are 20.880, 20.888, 20.883, 20.867, 20.868, and 20.876 per cent, with an average of 20.877 per cent, the greatest difference being only 0.021 per cent. Frankland in 1861,2 reporting the results of some air samples taken by himself on an excursion to the top of Mont Blanc, cites analyses of air collected during a balloon ascension in August 1852, which were made by Dr. Miller. At a height of 18,000 feet the oxygen percentage was found to be 20.88; a sample taken at the same time at the surface of the earth gave 20.92 per cent.3 Frankland's analyses of the air from Mont Blanc were made by ab- sorbing the carbon dioxide by means of caustic potash, and determining the oxygen by explosion with electrolytically prepared hydrogen. Speci- mens were taken at Chamounix (altitude 3000 feet), at the Grands Mulets (11,000 feet), and at the top of Mont Blanc (15,732 feet). At Chamounix the percentages of oxygen found were 20.892 and 20.870, and of carbon dioxide, 0.063. At the Grands Mulets, two samples showed 20.793 and 20.765 per cent of oxygen and 0.111 per cent of carbon dioxide. At the summit of Mont Blanc, on August 21, at 8h 45m a.m., two samples gave 20.950 and 20.951 per cent of oxygen respectively, and 0.061 per cent of carbon dioxide. The averages of all the results are, therefore, as follows : Place. Carbon dioxide. Oxygen. Chamounix p. ct. 0.063 .111 .061 p. ct. 20.894 20.802 20.963 Grands Mulets Summit of Mont Blanc . . p 1 Frankland and Ward, Quarterly Journal of the Chemical Society, London, 1854, 6, 197. 2 Frankland, Quarterly Journal of the Chemical Society, London, 1861, 13, p. 22. 3 Julius Hann in his Lehrbuch der Meteorologie, Leipzig, 1901, p. 9, states that sam- ples of air were collected in a balloon journey made by Welsh; according to the analyses made by Miller, the oxvgen content on the surface of the earth was found to be 20.92 per cent; at 4100 meters, 20.89; at 5500 meters, 20.75; and at 56.80 meters, 20.89 per cent. History of Air-Analysis 43 Two series of analyses of the air of Madrid were made by Torres Muiloz.1 The percentages of oxygen are very small, and those of carbon dioxide usually very high, the latter ranging from 0.02 to 0.09 per cent. The oxygen was determined in part by cuprous ammonium chloride and in part by potassium pyrogallate. The percentages of oxygen obtained are given in table 24. Table 24.— An alyses of air collected in Madrid by Torres Munoz. Oxygen outside the walls, in March. Oxygen within the walls, in April. p. ct. p. ct. p. ct. p. ct. 20.71 20.70 20.70 20.78 20.79 20.74 20.70 20.69 20.77 20.69 20.77 20.70 20.77 20.81 20.75 20.78 20.73 20.79 20.70 20.80 20.75 20.78 20.69 20.73 Another investigator, Russell,2 using a modified Bunsen apparatus, reports in 1868 an analysis of air as containing 20.845 per cent of oxygen. One of the first to use the pyrogallic-acid method for the analysis of atmospheric air was Schiel,3 who in 1857 determined the oxygen content of air at an altitude of 2330 feet on the boundary between Kansas and Colorado, finding 20.91 per cent as an average of three experiments. The importance of Liebig's discovery of the use of potassium pyrogal- late as an absorbing agent for oxygen was early recognized by Speck,4 who submitted the method to numerous tests in connection with his physiological researches. Speck recommends the use of barium hydroxide in place of potassium hydroxide to prevent the formation of carbon mon- oxide. The analyses with barium hydroxide and pyrogallic acid agreed well with the Bunsen explosion method. Duplicate analyses by the barium pyrogallate method gave results as follows: Date. I. II. 1866. Sept. 23 Oct. 10 Oct. 11 p. ct. 20.98 20.93 20.93 p. ct. 20.97 20.96 20.98 One of the most extensive investigations of the oxygen content of atmospheric air was that carried out by R. Angus Smith, of Manchester, 1 Estudios quimicos sobre el aire atmosferico de Madrid, por D. Ramon Torres Munoz de Luna, Madrid, 1860. A translation of this entire paper was made by de Chambry and published in Annales d'Hygiene Publique et de Medicine legale, 1865, seriesr2, 15, p. 337. 2 Russell, Journ. Chem. Soc, London, 1868, newser., 6, p. 140. 3 Schiel, Annalen der Chemie und Pharmacie, 1857, 103, p. 120. 4 Speck, Schriften der Gesellschaft zur Beforderung der gesammte Naturwissenschaft zu Marburg, 1871, 10. 44 Composition of the Atmosphere England. His results, which were first published in a condensed form,1 were subsequently given more in detail as a part of a larger publication.2 Using for the most part the explosion apparatus of Bunsen, although making several ineffective attempts to secure accurate results by means of Liebig's pyrogallic-acid method, Smith made an enormous number of analyses of outdoor air in connection with his investigation of the ven- tilation of houses. In his book, which contains the best collection of the literature and analyses of outdoor air thus far published in English, he reports over 100 determinations of the oxygen content of outdoor air. Firmly impressed with the belief that the presence of putrefying organic matter requires a material draft upon the oxygen of the air, Smith made a comparative study of the air collected at the front door of the labora- tory and in outhouses in the near neighborhood. His results are given in table 25. Table 25. — Determinations made by Smith of oxygen in outdoor air and in the air of outhouses in Manchester. Date. Air from front door of labo- ratory. Air from outhouses. Date. Air from front door of labo- ratory. Air from outhouses. 1863. p. ct. p.ct. 1863. p. ct. V- ct. Dec. 1 20.90 20.80 Dec. 19 20.87 Dec. 10 20.96 20.85 Dec. 21 20.92 20.56 Dec. 11 20.98 20.79 Dec. 21 21.02 20.79 Dec. 11 20.90 20.72 Dec. 21 20.88 20.64 Dec. 15 20.90 20.87 Dec. 21 20.91 20.94 Dec. 15 20.02 20.76 Dec. 21 21.01 20.67 Dec. 17 20.96 20.59 Dec. 22 20.96 20.53 Dec. 17 20.78 20.85 Dec. 22 20.92 20.71 Dec. 17 20.83 20.90 1864. Dec. 18 20.91 20.21 Feb. 26 21.01 20.66 Dec. 18 20.92 20.58 Feb. 24 21.05 • « • Dec. 18 20.87 20.74 Feb. 20 20.98 Dec. 18 21.02 20.40 Feb. 20 20.99 .... Dec. 18 21.00 20.77 Feb. 20 21.01 Dec. 19 20.83 20.99 Feb. 20 20.94 • . • • Dec. 19 20.98 20.70 Dec. 19 20.88 20.82 Average 20.943 20.70 Dec. 19 21.01 20.46 A very extensive examination of the air of London was carried out by Smith in November 1869. These analyses give us a method of judging of the accuracy of Smith's method and the agreement of duplicates. They are in part reported in tabic 26. The probable influence of weather conditions, especially moisture and fog, led Smith to investigate the changes in oxygen content as affected by this factor. Five analyses are reported of air taken near the labora- tory in Manchester during wet weather, the results being 20.90, 21.01, 21.01, 21.05, and 20.96 per cent of oxygen, respectively, with an average 1 Smith, Memoirs of the Literary and Philosophical Society ofjManchester, 1864-65, ser. 3, 3, p. 5. 2 Smith, Air and rain; the beginnings of a chemical climatology. London, 1872. History of Air-Analysis 45 of 20.98 per cent. In dry and foggy weather, when the smoke of Man- chester hung over the town, the results were as follows: Oxygen percentage. Near center of town j 9088 At laboratory j i^jj At laboratory, afternoon 20.91 At laboratory, forenoon 21.01 At laboratory, afternoon 20.82 Average 20.91 Table 26. — Determinations made by Smith of oxygen in London air. Place. Islington, Duncan Terrace Hoxton, Hoxton Square Dalston, Albion Road Hackney, near Hackney Station . Clarendon Square, Somers Town Alpha Road and Grove Road . . Average Parks and open places. Near Belsize Park Kennington Park Chelsea Hospital, gardens near river . . . Vauxhall Bridge, near river Houses of Parliament, terrace Hyde Park, Sloane Street 20.91 Middle of Hyde Park 21.03 Average First Second analysis. analysis. p. ct. p. ct. 20.86 20.81 20.85 20.82 20.90 20.91 20.82 20.85 20.90 20.89 20.S7 20.80 21.02 21.00 20.96 20.92 20.91 20.91 20.90 20.91 20.96 20.93 20.91 20.94 21.03 20.98 1 Average. p. ct. 20.835 20.835 20.905 20.835 20.895 20.835 20.857 21.010 20.940 20.910 20.905 20.945 20.925 21.005 20.95 In a very dense fog, which Smith states was a rare experience in Man- chester, and which made the eyes smart and walking difficult, he found 20.82 and 20.89 per cent of oxygen. In the yard back of the laboratory he regularly found somewhat less oxygen than in the front, his results being 20.80, 21.01, 20.94, 20.84, 21.09 per cent, respectively, with an average of 20.936 per cent. He summarized his results as follows : Oxygen percentage. In very wet weather, in front of the laboratory 20.98 At all times (average of 32 experiments) 20.947 Behind the laboratory, in medium weather 20.936 In foggy frost 20.91 In outhouses 20.706 If we except the air from outhouses, we find an average variation in these analyses of 0.07 per cent. During 1863 to 1865 Smith made an extensive examination of the air at both the summit and base of a number of mountains in Scotland. While the altitudes were by no means as great as those of the Alps, and hence the results, which are given in table 27, can not contribute exten- 46 Composition of the Atmosphere sively to the question of the Dalton hypothesis, nevertheless they are extremely interesting as showing careful attention to a very important problem. Table 27. — Determinations made by Smith of oxygen in the air from mountainous districts of Scotland. Place. Ben Nevis Do Do Do Do Lochin-y-gair (Balmoral) Do Ben Ledi Do Do Ben Voirlich Do Ben-na-bourd Ben Lomond Summit. Base. p. ct. 20.91 20.96 20.94 20.88 21.01 20.94 20.95 20.98 20.97 20.97 21.01 21.03 20.94 p. ct. 20.93 20.91 20.89 20.80 21 21.02 20.87 20.88 21.18 20.95 Place. Ben Lomond . . Do Ben Muich Dhu Do Do :-.-.- Do Do Do Ochill Hill Do Moncrieffe Hill . Average . , Summit. Base p. ct. 21.08 20.91 21.00 21.07 21.02 20.99 20.93 21.01 21.05 21.07 20.93 20.98 p. ct 20.94 For comparison, Smith analyzed a large number of samples of air from districts in Scotland with but slight, if any, elevation. (See table 28.) Of particular significance is the extremely low value found near Inverness, which Smith attributes to some impurity arising from the water. Table 28. — Determinations made by Smith of oxygen in air from districts not mountainous. Place. Shore at Lossiemouth . . . Do Inverness, at Moray Frith Do Do Inverness, behind the town1 Sea-shore, Oban Edinburgh, Prince's street Do Edinburgh, Calton Hill. . Aboyne Do Do Aberdeen, sea-shore2 Do Do Oxygen. Average . p. ct. p. ct. 21.05 .... 20.95 21.00 20.89 ■ > ■ ■ 20.89 20.86 20.88 20.88 20.98 • . • > 20.99 • . . . 20.92 20.94 20.95 20.94 20.95 • • » • 21.02 20.96 21.05 21.01 21.07 21.04 Place. Errol, marshy ground3 Do Caledonian Canal (near Inverness)4 Balmoral Do Taynuilt (near Oban) . . Do Braemar-on-the-Dee5 . . Huntly Mar Forest6 Do Do Do Forest near Braemar . . Total average . . . Oxygen. Average. p. ct. p. ct. 20.91 20.96 20.94 20.88 20.88 21.00 20.90 20.92 20.86 20.89 21.18 « . • • 21.03 21.04 21.02 21.08 20.88 21.66 20.87 .... .... 20.96 1 Very clear weather. 2 Wind from sea, N. : evening. 3 Windy and cloudy. 4 Cloudy and windy, SW. 5 Cloudy. 6 Rain and sunshine. Although unquestionably slightly affected by the carbon dioxide in the exhalations of the inhabitants and in the production of furnaces and stoves in Perth, the results of Smith's analyses of air collected from a History of Air-Analysis 47 number of alleys and narrow streets in this city are also given here (see table 29). Table 29. — Determinations made by Smith of oxygen in air from the worst places in Perth.1 Place. Close, 70 South street Close, 44 Pomarium Do Do Weaver's Close, Pomarium . . Do St. Paul's Close Do Long Close, off George street Oxygen. p. ct. 20.87 20.92 20.94 20.93 20.96 20.94 20.96 20.99 20.94 Place. Long Close, off George street Close, 28 Watergate Close, 82 South street Hewat's Close, 148 South st. Do Close, 44 Meal Vennel Average Oxygen. p. ct. 20.90 21.02 21.01 20.97 21.00 20.90 20.95 1 Results of determinations made on air which was obviously indoor have been omitted. The results found by Smith in Scotland in 1863-5 are summarized by him as follows: Av. p. ct. of oxygen. Sea-shore and heath 20.999 Summit of hills 20.98 Base of hills 20.94 Places not mountainous > 20.978 Inferior parts of town (favorable, i.e., windy weather) 20.95 Lower places, marshy, etc 20.922 Forests 20.97 Total average 20.96 During a visit to Switzerland in August 1864, Smith collected a num- ber of samples of air and determined the oxygen. In giving the results, he includes for comparison a series of four analyses of air taken in the following month among some brushwood at Reddish, near Manchester. (See table 30.) Table 30. — Determinations made by Smith of oxygen in air from marshy or confined places, Switzerland. Place. Sion, upper valley of the Rhone, Switzer- land, over water and marshy grass (morn- ing) Sion, over water and brushwood (morning) Oxygen. p. ct. 20.86 f21.0n 20.94 21.05 21.02 20.96 20.94 20.95 20.83 21.00 20.90J Average. p. ct. 20.95 Place. Lauterbrunnen . Chamounix, Montanvert Verdin, in the Sologne . . Vouzerou Oxygen. (Average. Reddish, near Manches- ter, England, among brushwood p. ct. ( 20.94 ) ] 20.97 [ ( 20.95 ) j 21.03 ? 20.99 $ \ 20.97 } 20.90 S i 21.01 ( 20.90 20.92 20.98 20.95 20.90 p. ct. 20.953 21.01 20.95 20.937 48 Composition of the Atmosphere Again, in the winter of 1869 a large number of analyses were made of air collected in both the congested portions and the open parts of the city of Glasgow. The results are given in table 31. Table 31. — Determinations made by Smith of oxygen in air from Glasgow. Place. Congested sections. Buchanan street, near Western Club Exchange, front of Union street Miller street, Argyle street Argyle street, near Queen street St. Enoch's Square Cross Blackboy Close, Gallowgate Gallowgate, between Kent street and bar- racks A close, High street Armour street, near barracks Kirk street Coulter's lane, Abercromby street A small court, Tobago street Oswald street, Dalmarnock road Average Open sections. Tennant street, St. Rollox Charles street Middleton place : Castle street, near the cathedral Dobbies Loan, near poorhouse New City road, near Abercorn street Blythswood Square Renfrew street Newton Terrace, Sauchiehall street University, Gilmour Hill Quay, near Broomielaw Bridge Anderston quay Sharpe's lane, Stobcross street Finnieston quay Pointhouse pier Average Total average Oxygen. First analysis. p. ct. 20.92 20.89 20.88 20.90 20.96 20.92 20.90 20.90 20.88 20.87 20.86 20.88 20.88 20.85 20.87 20.90 20.92 20.94 20.92 20.93 20.91 20.94 20.90 20.95 20.88 20.95 20.90 20.98 20.90 20.98 Second analysis. p. ct. 20.91 20.90 20.88 20.93 20.90 20.91 20.85 20.88 20.87 20.86 20.88 20.87 20.88 20.90 20.89 20.93 20.92 20.95 20.87 20.91 20.92 20.95 20.92 20.99 20.92 20.90 20.90 20.97 20.92 21.01 Average. p. ct. 20.915 20.895 20.880 20.915 20.930 20.915 20'.875 20.890 20.875 20.865 20.870 20.875 20.880 20.875 20.880 20.8890 20.915 20.920 20.945 20.895 20.920 20.915 20.945 20.910 20.970 20.900 20.925 20.900 20.975 20.910 20.995 20.9293 20.9092 Smith also cites two analyses made of air collected by a friend in the West Indies. In one taken on the North Atlantic, latitude 43° 5' N., longitude, 17° 12' W., at a point 18 feet above water, at 2h 30m p. m. on a fine day, he found 21.01, 21, and 20.97 per cent of oxygen, respectively, the average being 20.99 per cent. In a sample taken at St. John's, An- tigua, on April 11, 1865, at 9 a. m. on a showery morning, three analyses of the same sample gave 20.96, 20.91, and 21 per cent, with an average of 20.95 per cent. Smith considers the difference between these two analy- ses as significant, and discusses the possibility of an influence on races, History of Air-Analysis 49 and sections of the same race, of small variations in the amount of oxygen in the atmosphere. Finally, Smith distinguishes between pure and impure air (such as that collected in outhouses) and maintains that pure air deviates from 21 by 0.065 per cent. Hinman,1 using an explosion apparatus, made analyses of air freed from carbon dioxide, obtaining on April 25, 1874, 20.94 and 20.93 per cent of oxygen, and on April 26, 1874, 20.94 and 20.92 per cent of oxygen, respectively. Using Bunsen's method, A. R. Leeds2 analyzed many samples of air collected in July, August, and September of 1876, near Hoboken, New Jersey, at the Centennial Exposition in Philadelphia, and at several places in the Adirondack Mountains. His results are given in table 32. Table 32. — Determinations of oxygen made by Leeds in samples of outdoor air. Date. Location. 1876. July Aug. Aug. Aug. Aug. Aug. Sept. 4 2 11 29 30 31 1 Stevens Institute . .Do Do Do Do .Do .Do Oxygen. p. ct. 20.957 20.957 20.821 20.843 20.954 20.934 20.942 20.952 20.957 Date. 1876. Sept. 7 Aug. 15 Aug. 18 Sept. 26 July 17 July 21 Location. Oxygen. Stevens Institute j Centennial grounds ....Do Stevens Institute . . Keene Flats, Adi- rondack^ Mount M a r c y (summit). p. ct. 20.932 20.944 20.962 20.918 20.915 21.029 20.928 20.926 In testing a gas-analysis apparatus, using copper immersed in an ammoniacal solution of ammonium chloride to absorb oxygen, Schlosing3 found 20.80 and 20.96 per cent oxygen in two samples. In the decade between 1879 and 1889 an unusually large number of researches on the composition of the air appeared, each of far-reaching importance. Prominent among these are the papers of von Jolly, Morley, Kreusler, Vogler, and Hempel. No paper since that of Regnault stimulated so much subsequent re- search on the composition of the air as did the publication of von Jolly's4 investigation. Working with extreme care in an attempt to weigh the gases in atmospheric air, he found in 1876 changes in oxygen content amounting to 0.5 per cent. Placing in the bulb of an air thermometer a copper spiral which could be heated by an electric current to incandes- cence, von Jolly was able to absorb the oxygen by the heated copper and thus determine the percentage of ox3'gen. His apparatus was ex- tremely ingenious and also very complicated. His results are given in table 33. 1 Hinman, American Journal of Science, 1874 (3), 8, p. 188. 2 Leeds, Annals Lyceum of Natural History, New York, 1878, p. 193. 3 Schlosing, Chemisches Centralblatt, 1869, p. 678. 4 von Jolly, Wiedemanns Annalen, N.F., 1879, 6, p. 520. 50 Composition of the Atmosphere Table 33. — Determinations of oxygen made by von Jolly. Date. 1877. June 13 June 18 June 24 June 27 June 31 July 3 July 17 July 19 July 27 Oct. 12 Oct. 14 Oct. 15 Oct. 16 Barometer. Wind. Oxygen. mm. p. ct. 714.03 w. 20.53 717.7 N. 20.95 716.8 NE. 20.73 718.7 NE. 20.65 718.1 NE. 20.69 716.9 E. 20.66 713.1 S. 20.64 713.9 sw. 20.56 719.9 NE. 20.75 715.7 E. 20.78 720.9 NW. 20.86 719.3 E. 20.83 723.3 E. 20.75 Date. 1877. Oct. 21 Oct. 23 Oct. 27 Oct. 31 Nov. 2 Nov. 10 Nov. 13 Nov. 20 Min. Max. Av .. Barometer. Wind mm. 723.0 710.6 721.5 714.2 724.1 718.2 707.0 708.9 714.03 721.5 E. NW. N. W. NE. SE. W. NW. W. N. Oxygen. p. ct. 20.84 20.84 21.01 20.85 20.91 20.56 20.67 20.65 20.53 21.01 20.75 Since these results agree with the variations found earlier by him, von Jolly concludes that the highest oxygen percentage is accompanied by a polar wind. He opposes Regnault's contention of an approximately constant composition of the air and in conclusion makes the following interesting and important statements : Ob von Jahr zu Jahr die Schwankungen stets in gleichen Grenzen erfolgen, und ob im Mittel der Sauerstoffgehalt in jedem Jahre der gleiche ist, wird erst durch eine ausge- dehntere Beobachtungsreihe sich feststellen lassen. Zunachst ist es wahrscheinlich , dass ebenso wie die Dauer der Polar — und Aequatorstrome an gleichem Orte nicht jedes Jahr die gleiche ist, auch kleine Differenzen im mittleren Sauerstoffgehalte sich von Jahr zu Jahr werden geltend machen. Auch wird man aus den Beobachtungen zweier Jahre schliessen diirfen, dass trotz der reicheren Vegetationsdecke sudlicherer Breitegrade die Oxydationsprozesse (vielleicht in Folge der hoheren Temperatur) die Reduktionsprozesse iiberwiegen, wahrend umgekehrt der reichere Gehalt an Sauerstoff der Polarstrome ein Zurucktreten der Oxydationsprozesse gegen die der Reduktion fur die nordlicheren Ge- genden ausdriickt. A series of analyses of air samples taken at the observatory in Palermo in 1879 were reported by Macagno,1 who was one of the first observers to use potassium pyrogallate systematically in air-analysis. (See table 34.) Table 34. — Results of analyses of atmospheric air made by Macagno. Date. Oxygen. Carbon dioxide. Date. Oxygen. Carbon dioxide. 1879. p. ct. p. ct. 1879. p. ct. p. ct. Feb. 20 20.879 0.021 May 31 20.017 0.033 Feb. 28 20.891 .048 June 10 20.894 .041 Mar. 10 20.715 .025 June 20 20.918 .043 Mar. 20 19.994 .025 June 30 20.915 .043 Mar. 31 20.888 .022 July 10 20.977 .020 Apr. 10 20.910 .021 July 20 20.984 .076 Apr. 20 20.880 .064 July 31 20.899 .039 Apr. 30 20.898 .045 Aua;. 10 20.910 .028 May 10 20.913 .005 Aug. 20 20.888 .030 May 20 20.902 .049 Aug. 31 20.895 .039 1 Macagno, Chemical News, 1880, 41, p. 97. History of Air-Analysis 51 Average results for February, March, April, and May (with rainfall) were for oxygen, 20.717 per cent, and for carbon dioxide, 0.033 per cent. For June, July, and August (without rainfall) the average results were for oxygen, 20.920 per cent, and for carbon dioxide, 0.039 per cent. Macagno emphasizes the low percentage of oxygen found during the sirocco wind, the extremely low percentages on March 20 and May 31 being due to this cause. The percentages of oxygen obtained during the sirocco were as follows: 1879. per cent March 20 19.994 March 21 20.008 March 22 20.064 April 15 19.998 1879. per cent May 29 20.021 May 30 20.032 May 31 20.017 Commenting on these results, Macagno writes: Here we have seven determinations of the most important element of air during that singular wind with its heat and dryness,1 rendering so troublesome the medium in which we are always bathed, and in all cases the want of oxygen is very evident. Simultaneously, but independently of von Jolly, E. W. Morley,2 of Cleveland, engaged in the accurate analysis of air with a view to study- ing the variations in composition and formulating a hypothesis which he hoped would be subsequently verified with an improved apparatus. Morley maintained that the descent of cold air from higher regions brought with it air poorer in oxygen, presupposing the correctness of the Dalton hypothesis. The apparatus3 used was a modification of that of Frank- Table 35. — Determinations of oxygen in atmospheric air collected by Morley after sudden depressions of temperature. Oxygen. Oxygen. Date. Average tempera- Date. Average tempera- ture. First Second ture. First Second analysis. analysis. analysis. analysis. 1S7S. °F. p. ct. p. ct. 1879. op p. ct. p. ct. Dec. 28 20.98 20.96 Feb. 16 26.3 20.95 1879 Feb. 20 18.9 20.87 20.87 Jan. 2 7.6 20.91 20.92 Feb. 26 22.3 20.45 20.50 Jan. 2 7.6 20.90 20.89 Feb. 27 12.8 20.77 20.80 Jan. 3 -7.2 20.90 20.91 Mar. 15 22.8 20.88 20.84 Jan. 3 -7.2 20.96 20.97 Mar. 15 22.8 20.84 20.86 Jan. 6 13.5 20.97 Mar. 16 25.3 20.92 20.92 Jan. 10 9.6 20.96 Mar. 17 24.5 20.89 20.90 Jan. 28 37.4 20.96 Apr. 3 25.0 20.77 20.79 Feb. 1 18.5 20.96 Apr. 3 .... 20.85 20.87 Feb. 1 18.5 20.94 20.94 Apr. 4 27.1 20.80 20.80 Feb. 2 19.8 20.91 20.93 Apr. 4 27.1 20.88 20.85 Feb. 2 19.8 20.82 20.80 Apr. 5 28.2 20.77 20.77 Feb. 15 11.1 20.88 20.86 Apr. 5 28.2 20.86 20.82 1 During the sirocco wind the relative humidity of air (determined by the psychrom- eter) is diminished to 30°, 24°, 20°, and even 18°. 2 Morley, American Journal of Science and Arts, 1879, 18, p. 168. 3 Morley, loc. cit.; also Amer. Chem. Journ., 1881, 3, p. 1. 52 Composition of the Atmosphere land and Ward,1 in which the hydrogen-explosion method is employed. Most of the samples were analyzed in duplicate, the difference at times being 0.05 per cent. An abstract of his results is given in table 35. The most noteworthy observations in this series are the extremely low values occasionally found for oxygen. On this point Morley says : On Sept. 16, 1878, two very careful analyses of the same sample gave 20.49 and 20.46 per cent of oxygen. * * * Within the time covered by the analyses now published, there were several well-marked great and sudden depressions of temperature, and the fig- ures show the falling off in the proportion of oxygen in the air at these times to be as well marked as the depression of temperature. The deficiency is not proportionate to the de- pression of temperature ; this could not be expected. In a second communication,2 Morley reports a very large number of analyses made with even greater care, comparing the results with the meteorological conditions which existed at the time. The investigation extended from January 1, 1880, to April 20, 1881, analyses being made nearly every day except during the vacation months of July, August, and September 1880. Usually duplicate analyses of the same sample were made and occasionally several samples were taken on the same day. For painstaking care and extent the work is marvelously complete. A part of his results are given in table 36, these being fairly indicative of the ac- curacy of the work, and the agreement of duplicate analyses. Table 36. — Determinations of oxygen in atmospheric air made by Morley. Date. Oxygen. Date. Oxygen. Date. Oxygen. First analysis. Second analysis. First analysis. Second analysis. First analysis. Second analysis. 1881. Jan. 1 Jan. 1 Jan. 2 Jan. 2 Jan. 3 Jan. 4 Jan. 5 Jan. 6 Jan. 7 Jan. 8 Jan. 9 p. ct. 20.949 20.936 20.954 20.952 20.959 20.954 20.958 20.956 20.958 20.946 20.954 p. ct. 20.939 20.940 20.960 20.947 20.957 20.951 20.958 20.953 20.948 20.960 20.962 1881. Jan. 10 Jan. 11 Jan. 12 Jan. 13 Jan. 14 Jan. 15 Jan. 16 Jan. 17 Jan. 18 Jan. 19 Jan. 20 p. ct. 20.966 20.957 20.963 20.958 20.957 20.960 20.952 20.959 20.950 20.956 20.958 p. ct. 20.967 20.959 20.956 20.957 20.961 20.961 20.952 20.957 20.952 20.956 20.960 1881. Jan. 21 Jan. 22 ! Jan. 23 Jan. 24 Jan. 25 Jan. 26 Jan. 27 Jan. 28 Jan. 29 Jan. 30 Jan. 31 p. ct. 20.948 20.938 20.955 20.959 20.952 20.969 20.959 20.953 20.960 20.968 20.960 p. ct. 20.950 20.933 20.947 20.958 20.947 20.967 20.959 20.961 20.954 20.964 20.962 While Professor Morley cites numerous instances of meteorological conditions accompanied by decreases in oxygen which are consistent with his hypothesis, there are fully as many days of low oxygen when he ad- mittedly is unable to explain the fall as a result of meteorological change. Furthermore, he repeatedly cites falls in oxygen amounting to but 0.01 per cent as significant. Morley concludes that there is no connection between the deficiencies in oxygen and the direction of the wind at the 1 Frankland and Ward, Quart. Journ. Chem. Soc. (London), 1854, 6, p. 197. 2 Morley, American Journal of Science, 1881 (3), 22, p. 417. History of Air-Analysis 53 time of taking the sample and "that the theory that deficiencies in the amount of oxygen in the atmosphere are caused by the descent of air from an elevation fairly well agrees with the facts." In another paper1 Mor- ley discusses in detail the improbability of the von Jolly hypothesis. The experimental researches of Morley and von Jolly stimulated, among other writers, Vogler,2 who, in a theoretical presentation of the subject, maintains that neither von Jolly's hypothesis nor that of Mor- ley explains the anomalies as well as does the conception of a separation of the gases of the air under conditions of high pressure. During a period of minimum barometer he maintains that the rapidly moving currents of air thoroughly mix the atmosphere, so that there is no difference in the oxygen content, but when there is a period of high barometer the air is quiet and there is a separation of the gases, with a high oxygen content near the earth. Table 37. — Determinations of oxygen made in air analyzed by Kreusler. Date. Oxygen. Date. Oxygen. Date. Oxygen. ' Date. Oxygen. 1883. p. ct. 1883. p. ct. 1883. p. ct. 1883. p. ct. Jan. 5 20.882 Apr. 2 20.925 July 12 20.927 Oct. 30 20.879 Jan. 7 20.905 Apr. 5 20.881 July 18 20.900 Nov. 2 20.899 Jan. 8 20.927 Apr. 9 20.991 July 25 20.924 Nov. 5 20.888 Jan. 9 20.914 Apr. 19 20.867 July 30 20.945 j Nov. 8 20.922 Jan. 11 20.899 Apr. 23 20.923 Aug. 4 20.930 Nov. 12 20.895 Jan. 13 20.925 Apr. 26 20.909 Aug. 7 20.936 Nov. 15 20.909 Jan. 14 20.900 Apr. 30 20.873 Aug. 10 20.906 Nov. 19 20.908 Jan. 17 20.900 May 3 20.896 Aug. 13 20.895 Nov. 22 20.903 Jan. 20 20.886 May 7 20.909 Aug. 17 20.934 Nov. 26 20.918 Jan. 21 20.917 May 10 20.919 Aug. 20 20.910 Nov. 29 20.901 Jan. 22 20.902 May 13 20.912 Aug. 24 20.933 : Dec. 3 20.876 Jan. 23 20.884 | May 16 20.896 Aug. 29 20.917 Dec. 6 20.900 Jan. 25 20.913 May 21 20.928 Sept. 2 20.886 Dec. 10 20.916 Jan . 26 20.915 May 24 20.914 Sept. 10 20.937 Dec. 13 20.904 Jan. 27 20.912 May 29 20.906 Sept. 14 20.883 ! Dec. 17 20.936 Jan. 28 20.984 June 1 20.927 Sept. 17 20.923 Dec. 20 20.902 Jan. 29 20.895 June 4 20.918 Sept. 20 20.888 Dec. 24 20.911 Jan. 30 20.914 June 7 20.926 Sept. 24 20.890 Dec. 27 20.907 Jan. 31 20.908 June 12 20.937 Sept. 27 20.985 1884 Feb. 4 20.926 June 15 20.917 Oct. 1 20.911 Jan. 11 20.937 Feb. 7 20.937 June 18 20.911 Oct. 4 20.916 Jan. 19 20.922 Feb. 10 20.900 June 20 20.894 Oct. 7 20.885 Jan. 23 20.941 Feb. 13 20.909 June 25 20.894 Oct. 11 20.924 Jan. 24 20.914 Feb. 16 20.899 June 28 20.926 Oct. 15 20.928 Jan. 27 20.925 Feb. 19 20.911 July 3 20.906 Oct. 18 20.921 Jan. 30 20.925 Feb. 22 20.895 July 6 20.918 Oct. 22 20.900 Feb. 7 20.936 Mar. 24 20.884 July 9 20.904 Oct. 25 20.916 By far the most complete collection of the literature on the composi- tion of atmospheric air thus far published is to be found in the admirable article by Kreusler.3 Using a eudiometer similar to that employed by 1 Morley, American Journal of Science, 1881 (3), 22, p. 429. 2 Vogler, Chemisches Centralblatt, 1882 (3), 13, p. 556. 3 Kreusler, Landwirtschaftliches Jahrblicner, 1885, 14, pp. 305-378. This article, although published in a somewhat inaccessible place, has been most helpful in preparing the material for this memoir. I have freely drawn upon the material Professor Kreusler has collected and wish to express here my appreciation of his paper. 54 Composition of the Atmosphere von Jolly, Kreusler made a series of analyses of atmospheric air in Bonn. These observations covered the whole year, averaging about 8 or 9 ex- periments each month, March excepted. About the middle of the year, certain improvements were made in the apparatus. His results are ab- stracted in table 37. As a result of his experiments, Kreusler found that the oxygen varied from 20.867 to 20.991 per cent as the extreme limits, but with certain ex- ceptional cases eliminated, the variations were from 20.88 to 20.94 per cent. The 99 observations in the year 1883 gave an average result of 20.911 per cent. The average values for each month were as follows: January 20.910 February 20.911 March April 20.910 May 20.910 June 20.917 July 20.918 September 20.913 October 20.909 November 20.905 August 20.920 December 20.906 The slightly higher values found during the warmer months are ex- plained by Kreusler as being due to the assimilative activity of vegetation. In referring to earlier analyses, Kreusler considers all experiments in which the variations were under 0.1 per cent as normal results, those be- tween 0.1 and 0.15 abnormal, and those over 0.15 per cent as most ab- normal. Of the 1025 observations that he has been able to find in the literature which are comparable, but 15 can be classified under the head of "abnormal" and 22 under the head of "most abnormal." The ob- servations of von Jolly in Munich in 1878 play an important role in this subdivision, for his results give 12 normal, 3 abnormal, and 6 most abnor- mal; Kreusler points out that while in the case of the other observations 97.2 per cent are normal and only 1.2 and 1.6 are abnormal and most ab- normal, respectively, the experiments of von Jolly give but 57 per cent normal, with 14.3 and 28.6 per cent, respectively, as abnormal and most abnormal. Obviously, therefore, von Jolly has found much greater vari- ations than all the other observers combined, and consequently Kreus- ler takes exceptions to his conclusions. Furthermore, the average oxy- gen content found in von Jolly's experiments was 20.75 per cent, which is very much lower than both the earlier and later observations. After considering the errors in von Jolly's experiments, Kreusler dis- cusses the anomalous observations of Brunner, Lewy, Regnault and Lewy , Liebig, Macagno, and Morley, and says in conclusion : Ich glaube hiermit den Nachweis geliefert zu haben, class die ja lange Zeit herrschend gewesene, neuerdings aber mehrfach wieder in Zweifel gezogene Annahme einer inner- halbenger Grenzen konstanten Zusammensetzung der atmospharischen Luft that- sachlich noch zu Recht besteht, insofern alle entgegengesetzten Beobachtungen bis jetzt nicht geniigend verbiirgt scheinen. In testing his extremely complicated apparatus for gas-analysis, Geppert1 made a number of air-analyses, employing the hydrogen ex- 1 Geppert, Die Gasanalyse und ihre physiologische Anwendung nach verbesserten Methoden, Berlin, 1885, p. 96. History of Air-Analysis 55 plosion for the determination of oxygen. A sufficiently large sample was taken to eliminate the errors incidental to working with minute quantities of gas such as would be obtained from blood; the results given in table 38 were obtained for the percentage of oxygen. The extremely low value of 20.68 is attributed by the author to a defect in the particular eudiometer used for this single determination. The author points out that if one wishes to make exact gas-analyses it is desirable to test the eudiometer previously with normal air. Table 38. — Percentages of oxygen determined on atmospheric air by Geppert. Series 1. Series 2. Series 3. Series 4. Series 5. Series 12. p. ct. p. ct. p. ct. p. ct. p. ct. p. ct. 20.910 20.848 20.885 20.948 20.88 20.96 20.928 20.863 20.904 20.912 20.91 20.97 20.929 20.837 20.912 20.93 20.97 .... .... .... 20.68 20.91 Contemporaneously with Morley and von Jolly, but working entirely independently, Hempel, employing potassium pyrogallate as the absorb- ing agent, began a research on the composition of the air.1 In 1877 he found, as the result of many analyses, differences so great as to be ex- plainable only on the ground of imperfect technique. Subsequent de- velopment of apparatus and method yielded a procedure so accurate that duplicate analyses of each day's sample did not vary from one another by more than 0.02 per cent. On five different days in the fall of 1877 he found 20.89, 20.76,20.96, 20.91, and 20.90 per cent of oxygen, respectively. In 1879, analyses were made in April and May as follows: p. ct. Apr. 24 21.16 Apr. 25 20.91 Apr. 26 20.92 Apr. 27 20.83 Apr. 28 20.87 Apr. 29 20.70 p. ct. Apr. 30 20.83 May 1 20.82 May 3 20.55 By means of the improved apparatus, an interesting comparative series of analyses was made, samples of air being collected in July 1883, simultaneously by Professor Hempel in Dresden and Professor E. Hagen on the steamer between Liverpool and New York, all samples being taken at 8 a. m. The results are given in table 39. Table 39. — Determinations of oxygen in atmospheric air, collected at sea and in Dresden. Date. At sea. Dresden. Date. At sea. Dresden. 1883. July 22 July 23 July 24 July 25 July 26 p. ct. 20.94 20.80 20.88 20.91 20.95 p. ct. 20.93 20.92 20.86 20.91 1883. July 27 July 28 July 29 July 30 p. ct. 20.87 21.09 20.91 21.01 p. ct. 20.92 20.97 21.01 20.95 Hempel, Berichte der deutschen chemischen Gesellschaft, 1885, 18, p. 267. 56 Composition of the Atmosphere Using identically the same apparatus, Oettel in Hempel's labora- tory determined each day the carbon dioxide and oxygen of Dresden air from October 12, 1884, to December 24, 1884. Oettel's results are ex- pressed in the form of a curve not easily reproduced, but table 40 shows his results from November 8 to 18. The duplicate analyses permit an estimate of the accuracy of the method he used. The figures also show the general extent of the variations he observed in the oxygen and carbon- dioxide content of the atmosphere. Table 40. — Determinations of oxygen in Dresden air, made by Oettel. Date. Oxygen and carbon dioxide. Carbon dioxide. Date. Oxygen and carbon dioxide. Carbon dioxide. 1884. Nov. 8 Nov. 9 Nov. 10 Nov. 11 Nov. 12 Nov. 13 p. ct. \ 20.84 \ 20.85 \ 20.87 1 20.89 ) 20.88 J 20.86 J 20.90 , 20.91 \ 20.74 ) 20.75 \ 20.77 I 20.75 p. ct. 0.036 .037 .039 .041 .050 .055 .0389 .0391 .040 .044 .044 .048 1S84. Nov. 14 Nov. 15 Nov. 16 Nov. 17 Nov. 18 p. ct. \ 20.81 1 20.78 \ 20.82 I 20.79 \ 20.80 } 20.83 \ 20.86 I 20.84 \ 20.92 } 20.92 p. ct. 0.049 .054 .038 .041 .0416 .0430 .040 .037 .040 .044 A few months later Hempel, in defending the use of potassium pyrogal- late from the criticism raised by Kreusler,1 published further experiments on air.2 In one sample of air he reports, as the result of four determina- tions in which carbon dioxide and oxygen were collectively absorbed, 20.936, 20.938, 20.938, and 20.938 per cent, respectively— an agreement that is striking. Table 41. — Determinations of oxygen in atmospheric air, made by Hempel. Date. 1885. Feb. 3 Feb. 6 Feb. 7 Feb. 8 Feb. 9 Feb. 10 Oxygen and Carbon Average i dioxide. dioxide. oxygen. p. ct. p. ct. p. ct. \ 20.960 0.035 20.920 ) 20.955 \ 20.970 .035 20.939 ) 20.980 \ 20.945 .035 20.917 I 20.960 ( 21.001 .034 20.962 I 20.991 \ 20.969 .034 20.940 / 20.979 \ 20.961 .034 20.926 { 20.958 Date. Oxygen and c b carbon r. .r dioxide. I dloxlde" 18*5. Feb. 11 Feb. 12 Feb. 13 Feb. 14 Feb. 15 p. ct. \ 20.950 "/ 20.944 \ 20.965 ') 20.968 \ 20.958 1 20.974 \ 20.932 , 20.946 \ 20.975 ( 20.971 p. ct. 0.037 .035 .036 .034 .035 Average oxygen. p. ct. 20.910 20.932 20.930 20.905 20.938 Beginning with February 3, 1885, Hempel made duplicate analyses of air nearly every day until March 28. The results for the first half of 1 Kreusler, loc. cit. 2 Hempel, Berichte der deutschen chemischen Gesellschaft, 1885, 18, p. 1800. History of Air-Analysis 57 February are given in table 41 as illustrative of the accuracy of his work and of the magnitude of the fluctuations experienced by him. An examination of all of his results shows a minimum of 20.877 per cent, a maximum of 20.971 per cent, with a difference of 0.094 per cent, and an average of 20.93 per cent of oxygen. Hempel points out that his results compare favorably with those of Kreusler, who found 20.911 per cent, and of Morley, who obtained 20.949 per cent, each investigator using a wholly different method. Obviously to three such skilled experimenters as Morley, Kreusler, and Hempel, the uncertainty regarding the general question as to the constancy, or lack of constancy, of the oxygen content of the air was somewhat disconcerting, and it is not surprising that we find them in 1886 engaging in a cooperative investigation. Morley in Cleveland, Kreusler in Bonn-Poppelsdorf, and Hempel in Dresden collected samples at the same time, making due allowances for geographical location. In addition, Hempel secured the aid of Pusinelli, who took samples in Para, Brazil, and of Schneider, who simultaneously took samples at Tromso in Norway. Thus the times for collecting; were: *& Dresden 2h 38m p.m. Tromso 3 00 p.m. Cleveland 8h 18m a.m. Para 10 31 a.m. Bonn 2 12 p.m. Kreusler,1 who made the determinations with his glowing copper-wire eudiometer, published his results independently shortly before Hempel's paper2 appeared. Kreusler abates somewhat his criticism of the pyrogallic-acid method, but still adheres to von Jolly's copper eudiometer. His results need not here be reproduced in full, as they are in part incorporated with those of Hempel and Morley in table 42. Kreusler found percentages of oxygen ranging from 20.907 to 20.939, substantiating his earlier findings and be- lief that the oxygen fluctuations, in spite of changes in meteorological conditions, are small. The average is 20.922 per cent, which is a little higher than the average of his earlier results, i. e., 20.911. Hempel's summary of the complete investigation includes the results of both Kreus- ler and Morley. A specimen set of records from April 1 to April 11, 1886, will serve to show both the agreement of duplicates and the variations experienced in various places. The samples from Tromso, Dresden, and Para were all analyzed alike, in that both oxygen and carbon dioxide were simultaneously absorbed. Oxygen alone was determined in the samples collected in Bonn and Cleveland. The Tromso analyses usually showed an agreement within 0.01 per cent on the same day; on several occasions the difference was 0.03 per cent, but the average for each day showed a fluctuation ranging from 21 per cent down to 20.90 per cent. It should be borne in mind that the 1 Kreusler, Berichte der deutschen chemischen Gesellschaft, 1887, 20, p. 991. 2 Hempel, ibid., p. 1864. 58 Composition of the Atmosphere results obtained from the Tromso samples were for oxygen plus carbon dioxide and not for oxygen alone. Table 42. — Comparative study of the oxygen content of air, made by Kreusler, Hempel, and Morley. Date. Oxygen and carbon dioxide. Oxygen. Tromso. Dresden. Para. Bonn. Cleveland. Found. Average. Found. Average. Found. Average. Found. Found. Average 1886 p. ct. p. ct. p. ct. p. ct. p. ct. p. ct. p. ct. p. ct. p. ct. Apr. 1 20.95 20.93 20.94 20.94 20.93 20.94 20.91 20.91 20.91 20.93 20.90 20.90 20.90 Apr. 2 20.94 20.94 20.94 20.96 20.96 20.96 .... 20.93 20.93 20.93 20.93 Apr. 3 .... 20.93 20.93 .... .... 20.91 20.91 20.92 20.92 Apr. 4 20.94 20.94 20.95 20.95 20.95 20.91 20.92 20.91 20.93 20.93 20.93 20.93 Apr. 5 20.95 20.94 20.95 20.93 20.94 20.94 20.94 20.92 20.93 20.93 20.92 20.94 20.94 20.94 Apr. 6 20.97 20.97 20.97 20.92 20.94 20.93 20.95 20.95 20.95 20.92 20.93 20.93 20.93 Apr. 7 20.96 20.96 20.96 20.89 20.90 20.90 20.93 20.92 20.93 20.92 20.93 20.93 20.93 Apr. 8 .... .... .... 20.90 20.90 20.90 20.91 20.91 20.91 20.91 Apr. 9 20.97 20.95 20.96 20.91 20.91 2oii 20.96 20.96 20.96 20.90 20.93 20.93 20.93 Apr. 10 20.96 20.95 20.96 20.93 20.92 20.93 20.95 20.95 20.95 20.93 20.93 20.93 20.93 Apr. 11 .... .... 20.92 20.91 20.92 20.91 20.91 20.91 20.92 20.92 20.93 20.93 The Dresden samples showed very slight variations, the results for the same day usually being very close, and commonly exactly alike. In one instance only was there a difference of 0.02 per cent. The average for different days ranged from 20.96 per cent down to 20.88 per cent; the maximum and minimum did not coincide with those for Tromso. For the Para samples the agreement for individual days was likewise very close, although in all not so many samples were analyzed. The agreement was within 0.01 or 0.02 per cent, the difference in no instance being more than 0.02 per cent. The average for the day showed fluctu- ations from 20.99 to 20.86 per cent. The analyses made in Tromso, Dresden, and Para, therefore, since they all represent the content of oxygen plus carbon dioxide, are directly comparable with one another, the averages being respectively: For Trom- so, 20.946 per cent; for Dresden, 20.928 per cent; and for Para, 20.923 per cent. On the supposition that the carbon-dioxide content remained on the average 0.03 per cent, and assuming that it was somewhat higher in the day period than in the night, Hempel computed the average oxygen content by deducting 0.03 per cent, and found for Tromso, 20.92 per cent; History of Air-Analysis 59 for Dresden, 20.90 per cent; and for Para, 20.89 per cent, respectively. In the same month, therefore, the oxygen content in the neighborhood of the poles was somewhat higher than in the neighborhood of the equator. In the oxygen values found in Bonn, only a single observation for each day was given, or possibly these figures represent the average for each day. They varied from 20.94 to 20.90 per cent. The values for the Cleveland samples, which also represent the oxygen content alone, showed usually an exact agreement between duplicates on the same day, and in only a few instances was there a discrepancy of 0.01 per cent. The averages ranged from 20.95 to 20.90 per cent. The total averages for these two places are: For Bonn, 20.922 per cent; for Cleveland, 20.933 per cent. The total average value obtained by the analysis of 203 different air samples, collected in five different places, and analyzed by three different methods, is 20.91 per cent of oxygen. On an expedition to Cape Horn to observe the transit of Venus, Muntz and Aubin1 made analyses of air taken at Orange Bay, using the eudiometer of Regnault and Reiset as modified by Schloesing. Their results are given in table 43. The authors conclude that the average oxy- gen percentage in the air taken near Cape Horn is less than the value found by Regnault in Paris, but essentially that found in different parts of the world; they believe, however, that the proportions of oxygen and nitrogen are subject to variations which are within the narrow limits es- tablished by Regnault. Table 43. — Determinations of oxygen in atmospheric air, made by Muntz and Aubin. Date. Oxygen. Date. Oxygen. Date. Oxygen. 1SS3. p. ct. 1883. p. ct. 1883. p. ct. May 10 20.86 May 17 20.97 July 22 20.90 May 11 20.92 May 18 20.85 July 22 20.95 May 12 20.93 May 19 20.88 July 29 20.89 May 13 20.89 May 19 20.72 Aug. 1 20.78 May 14 20.95 May 22 20.87 Aug. 2 20.79 May 15 20.90 May 24 20.84 Aug. 2 20.83 May 16 20.72 July 5 20.83 Phosphorus as an absorbent of oxygen again made its appearance in air-analysis in the hands of Ebermayer,2 who analyzed air from forests in 1885. He found in free atmospheric air 20.82 per cent of oxygen. Breslauer3 analyzed air in Brandenburg at least once a moDth, some- times as frequently as two or three times each month from January to December. As an average of 20 analyses he gives 20.934 per cent, with a minimum of 20.895 and a maximum of 20.955 per cent.4 1 Muntz and Aubin, Comptes rendus, 1886, 102, p. 421. 2 Ebermayer, Fortschritte a. dem Gebiete der Agrikultur-Phys., 9, p. 229; abstracted in Chemisches Centralblatt, 1886, 17, p. 770. 3 Breslauer, Die chemische Beschaffenheit der Luft in Brandenburg a. H., Berlin, 1886. 4 Breslauer, Deutschen Medizinal-Zeitung, 1885, 6, p. 1. 60 Composition of the Atmosphere In connection with his research on the influence of the ingestion of food on metabolism, Magnus-Levy1, in Zuntz's laboratory, described a new form of gas-analysis apparatus in which phosphorus is used to absorb oxygen. To demonstrate the accuracy of his apparatus he reports the determinations of the oxygen percentages in a series of 16 air-analyses, as follows: p. ct. p. ct. p. ct. p. ct. 20.90 20.90 20.73? 20.88 20.93 20.88 20.81 20.86 20.89 20.97? 20.90 20.90 20.89 20.82 20.90 20.88 Leduc2 in 1890, recognizing the discrepancy between the results of Regnault on the one hand and of Dumas and Boussingault on the other, sought to explain the difference on the grounds of slight errors in Reg- nault's determinations of the densities of the gases. One year later Leduc,3 having questioned the accuracy of absorbing oxygen by copper, inasmuch as nitrogen combined with the hydrogen of the reduced copper, used Brunner's method, in which the oxygen is absorbed by phosphorus, and weighed both the air and the nitrogen remaining. Two experiments carried out with the greatest care gave 23.244 and 23.203 per cent of oxy- gen by weight, or 21.024 and 20.987 per cent by volume, respectively. The author obtained the same results by determining the densities of nitrogen and oxygen. Wanklyn and Cooper,4 using the pyrogallic-acid method, obtained on three samples 20, 20.8, and 20.5 per cent of oxygen, respectively. By explosion with hydrogen they obtained 21.34, 20.94, and 21.34 per cent of oxygen. Using an alkaline ferroso-tartrate solution for an absorbent, de Ko- ninck5 found as the mean of four determinations 21.00 per cent of oxygen. Laulanie6 in 1894 published the description of a eudiometer in which the oxygen was determined by absorption with phosphorus. He claimed very accurate results, with a constant oxygen percentage of 20.9. Using a modified Bunsen apparatus, Schaternikow and Setschenow7 made analyses of laboratory air in Moscow, and report 12 analyses of out- door air. The oxygen percentage ranged from 20.874 to 21.036, with an average of 20.962 per cent. By passing a measured amount of air and nitric oxide through hydri- odic acid, Kreider,8 in New Haven, Connecticut, determined the free iodine thus liberated by means of 1/10 normal arsenious acid by titration. He gives a series of results with outdoor air on two samples collected over 1 Magnus-Levy, Archiv fur die gesammte Physiologic, 1894, 55, p. 1. 2 Leduc, Comptes rendus, 1890, 1 11, p. 262. 3 Leduc, Comptes rendus, 1891, 1 13, p. 129. 4 Wanklyn and Cooper, Air analysis, London, 1890, p. 35. 5 de Koninck, Chemical News, 1891, 64, p. 45. 6 Laulanie, Archives de Physiologie, 1894, 26, p. 739. 7 Schaternikow and Setschenow, Zeitschrift fur physikalische Chemie, 1895, 18, p. 56.3 8 Kreider, Zeitschrift fur anorganische Chemie, 1896, 13, p. 423. History of Air-Analysis 61 water; 11 analyses made on the same sample of air gave percentages of oxygen varying from 20.91 to 21.19, with an average of 21.08, and on another day 6 analyses of the same sample of air gave percentages varying from 20.96 to 21.03, and averaging 20.99. The author also reports two analyses of air by pyrogallic acid, showing 20.93 and 20.88 per cent of oxygen, respectively. Employing a reaction first utilized by Winkler1 to determine the oxy- gen dissolved in water, Chlopin2 utilized the interaction between man- ganous salt and oxygen in the presence of potassium iodide to determine the oxygen content of gaseous mixtures by titration. The results of a series of analyses of the air taken in the court of the Hygienic Institute in Dorpat are given in table 44. Table 44. — Results of analyses of outdoor air, made by Chlopin. Date. Oxygen. Eate. Oxygen. 1S97. Feb. 25 Mar. 4: Sample a . . . . Sample b ... Mar. 8: Sample a. . . . Sample b ... Mar. 13 Mar. 19 p. ct. 20.86 20.87 20.89 20.88 20.94 20.95 20.90 1897. Apr. 2 Apr. 10 June 1 June 2 June 5 Average . . p. ct. 21.08 20.97 20.87 20.91 20.84 20.905 A second paper by Chlopin3 reports analyses of outdoor and room air, but does not indicate which are those for outdoor air. Samojloff and Iudin,4 describing a method of gas-anafysis in which a modified Bunsen apparatus was used, have published a considerable num- ber of analyses of outdoor air. Using explosion with hydrogen to de- termine the oxygen, they found 20.97, 20.97, 20.96, 21.00, 20.99, 20.99, 21.01, 20.99, 20.96, 20.92, 20.97, and 20.98 per cent. Two analyses with pyrogallic acid gave 20.91 and 21 per cent of oxygen, respectively. In 1902, Krogh, of Copenhagen, on a voyage to the island of Disko, west of Greenland, latitude 70° north, undertook a number of analyses of atmospheric air. While the prime object of his research was to study the carbon-dioxide tension in natural waters and especially in the sea, these determinations were combined with direct analyses of the atmospheric carbon dioxide and oxygen. The analyses were made with a Haldane apparatus, the burette of which contained only 10 c. c. The accuracy is given as 0.005 to 0.01 per cent and the numerous double determinations generally agree within these limits.5 The extensive series of oxygen de- 1 Winkler, Berichte der deutschen chemischen Gesellschaft, 1887, 21, p. 2843. Also ibid., 1889, 22, p. 1764. 2 Chlopin, Archiv fur Hygiene, 1899, 34, p. 71. 3 Chlopin, Archiv fur Hygiene, 1900, 37, p. 329. 4 Samojloff and Iudin, Le Physiologiste Russe, 1901, 2, p. 171. 5 Krogh, Meddelelserom Greenland, 1904, 26, p.333. 62 Composition of the Atmosphere terminations obtained at this time are reported in a second article dis- cussing the abnormal carbon-dioxide percentage of the air in Greenland, and the general relations between atmosphericandoceanic carbon dioxide.1 On 22 days samples of air were taken from a number of different places in Greenland, 47 determinations of oxygen being made. Excepting one observation of 20.84 per cent, which was attributed by the author to a possible analytical error, the results exhibit variations from 20.92 to 21.015 per cent. Omitting the very low value of 20.84 per cent, the average of the determinations is 20.960 per cent. Of particular signifi- cance is the fact that analyses made on the same samples showed ex- tremely high values for carbon dioxide, ranging at times from 0.025 to 0.07 per cent. In a private communication from Dr. Krogh, he reports that a series of experiments made by him in Greenland in 1908 showed oxygen per- centages ranging from 20.895 to 20.980, with an average of 20.945. The unusually high carbon-dioxide percentages of former years were not ob- tained, although two observations gave 0.055 per cent. Dr. Krogh also writes that in 1907 and 1908 Dr. Lindhard of Copenhagen made obser- vations in northeast Greenland (Denmark Haven) using the identical modified Pettersson apparatus described by Dr. Krogh in a former paper. He reports that Lindhard's results would be liable to about 0.001 per cent error, and they agreed perfectly with those found by himself on the west coast. Lindhard generally found about 0.035 per cent of carbon dioxide, but on one or two days it was below 0.03 per cent, and on 5 days out of 23, 0.04 per cent or more. The maximum value found was 0.062 per cent. Dr. Krogh, commenting upon his own determinations of oxygen, writes that they may have an error of several hundredths of a per cent, as the absolute accuracy may be much affected by dirt accumulating in the burette and by variations and gradual displacement of the contained water. He says: "I do not think that it is at all possible to determine oxygen with great absolute accuracy except by analyzing dry and in a perfectly clean burette." It should be borne in mind that Dr. Krogh's original investigation dealt simply with the carbon-dioxide tensions in air and in water, and the oxygen determinations were quite incidental; likewise, the oxygen deter- minations in 1908 were made in connection with experiments with the respiration apparatus for determining the gaseous exchange of Eskimos. As one of the foremost investigators in gas-analysis, Krogh's experiences are doubly valuable.2 1 Krogh, Meddelelser om Groenland, 1904, 26, p. 409. 2 On a recent trip to Copenhagen, I was accorded the privilege of seeing a new gas- analysis apparatus devised by Dr. Krogh in which the conditions outlined by him above are fully realized. Unfortunately, at the time of going to press with this report, Dr. Krogh has not completed satisfactorily his experiments with this apparatus. History of Air-Analysis 63 By means of the apparatus described by Atwater and Benedict,1 Miss Charlotte R. Manning2 made a number of air-analyses at Middle- town, Connecticut, the details of which are given in table 45. In 1904 Chandler3 reported a series of analyses made of 50 samples of air collected in the New York subway, the content of oxygen ranging from 20.3 to 20.8 per cent. Determinations were also made of the oxygen in 9 samples of surface air; the results ranging from 20.6 to 20.9 per cent. Table 45. — Determinations of oxygen in atmospheric air, made at Wesley an University in Middletown, Connecticut. Date. Direction of wind. Oxygen. I. II. ill. 1903. Nov. 25 Nov. 27 Nov. 30 Dec. 4 Dec. 5 Dec. 7 w. w. sw. NE. SW. NNW. NW. W. W. p. ct. 20.91 20.95 21.09 21.05 21.01 20.97 21.01 20.93 21.01 p. ct. 20.92 21.00 21.03 20.97 21.00 20.99 20.99 20.92 21.00 p. ct. 20.93 20.97 20.90 Pecoul and Gizolme4 published in 1903 the description of their method of air-analysis, in which they employed a unique absorption apparatus and potassium pyrogallate. They also published the results of an analy- sis of an air sample collected in Paris at Place Lobau, the percentage of carbon dioxide obtained being 0.03 and of oxygen 20.85. Utilizing the extremely ingenious manometer and compensating de- vice of his earlier apparatus, Pettersson and Hogland5 so modified this apparatus as to determine not only the carbon dioxide but also the oxygen, using sodium hydrosulphite as the absorbing agent. They report the average of all the oxygen determinations made by them in Stockholm in October, November, and December 1889 as 20.940 per cent. Unfortu- nately, although the authors promised further details and additional re- sults, no published report has as yet appeared; nor did a personal visit to Stockholm result in obtaining further information. That this method was extremely promising was foreseen by Jaquet,8 who, in connection with the development of his most ingenious respiration apparatus, felt the imperative need of exact oxygen determinations. Em- ploying an apparatus modeled after the design of Pettersson, but using 1 Atwater and Benedict, Carnegie Institution of Washington Publication No. 42, 1905. 2 Unpublished data from laboratory note-book. 3 Chandler, The air of the subway, New York, 1904 (privately printed). 4 Pecoul and Gizolme, Annales d'Observatoire Municipal (Observatoire de Mont Souris), Paris, 1903, 4, p. 184. 6 Pettersson and Hogland, Berichte der deutschen chemischen Gesellschaft, 1889, 22, p. 3324. 6 Jaquet, Verhandlungen der naturforschenden Gesellschaft in Basel, 1904, 15, p. 252. 64 Composition of the Atmosphere potassium pyrogallate as the reagent, Jaquet analyzed a number of samples of outdoor air and reports the results in support of his con- tentions regarding the accuracy of his apparatus. These are given in table 46. Table 46. — Results of air-analyses made by Jaquet. Sample No. Carbon dioxide. Oxygen. 1 2 3 4 5 6 Average . . p. ct. 0.02 .03 .032 .032 .038 .032 .031 p. ct. 20.93 20.935 20.928 20.928 20.94 20.94 20.934 Staehelin, using the same apparatus as Jaquet, reported in 1907 J that in 51 analyses of air in Basel the average oxygen content was 20.90 per cent. The average value was found only in 22 samples. Twice the oxygen was less than 20.89 per cent and eight times larger than 20.91 per cent. The extreme values were 20.875 and 20.94 per cent. Using the gas-analysis apparatus in Basel, Gigon2 found with po- tassium pyrogallate that the oxygen content of the air near the laboratory varied between 20.87 and 21.2 per cent. These fluctuations he is inclined to attribute to the fact that the laboratory is in the center of the city and near large factories. In connection with the use of the Zuntz-Geppert respiration apparatus, many analyses of air have been made and occasionally such analyses are reported. Few have greater interest than those published by Durig and Zuntz3 in connection with the description of one of their trips into the high Alps. Table 47. — Results of air-analyses made by Durig and Zuntz. Date. Place. Carbon dioxide. Oxygen. 1903. Aug. 14 Aug. 19 Aug. 23 Aug. 31 Sept. 6 Col d'Olen p. ct. 0.04 .03 .02 .02 .03 p. ct. 20.88 20.86 20.89 20.88 20.87 .. ..Do Capanna Margherita .. ..Do .. ..Do In 1903 they analyzed free air at Col d'Olen and at the Capanna Margherita on the summit of Monte Rosa. The results are given in table 47. The authors state that these results, which agree with their numerous 1 Staehelin, Verhandlungen der Naturforschenden Gesellschaft in Basel, 1907, 19, p. 9. 2 Gigon, Archiv fur die gesammte Physiologie, 1911, 140, p. 517. 3 Durig and Zuntz, Archiv filr Anatomie und Physiologie, Physiologische Abtheilung, Supp].Bd.,1904,p.421. History of Air-Analysis 65 analyses in Berlin, substantiate the belief that the atmosphere has a con- stant constitution up to an altitude of 4600 meters. In a private communication Professor Durig writes that in the two expeditions in 1903 with Zuntz and in 19061 with Frau Durig and others, he made over 100 analyses of air. The percentage of oxygen was always between 20.87 and 20.96. On the Teneriffe expedition in 1907 he found from 20.87 to 20.98 per cent — limits almost exactly those experienced in Vienna and on Monte Rosa. In describing his extremely accurate and ingenious gas-analysis ap- paratus, Haldane2 has published the results of a number of air-analyses. Four analyses of the same sample of air gave the following percentages: ***. Oxygen. Carbon dioxide. 1 2 3 4 Average 20.930 20.926 20.931 20.924 0.025 .030 .035 .030 .030 20.928 The author concludes that 20.93 per cent may be taken as the true per- centage of oxygen in pure air. Absorbing the oxygen from dry air by means of heated yellow phos- phorus, Watson,3 working in Guye's laboratory in Geneva, determined the oxygen in air collected in Geneva and on some of the nearby moun- tains. In describing his method, he includes a few preliminary results which are given in table 48. Table 48. — Analyses made by Watson of air collected in Switzerland. Source ot air. Laboratory, Geneva (alt. 300 in.) : July 11, 1910, 4 p.m July 12, 1910, 5 p.m May 19, 10 a.m Saleve (alt. 1300 m.), May 19, 10 a. m Rochers de Naye (alt. 2045 m.) : May 19, 10 a.m May 19, 5h30ma.m Oxygen. p. ct. p. ct. 20.96 20.93 20.98 20.95 21.02 21.04 20.95 20.93 21.02 21.04 .... III. p.et. 21.03 1 Durig, Archiv fiir die gesammte Physiologie, 1906, 1 13, p. 213. 2 Haldane, in The investigation of mine air, by Foster and Haldane, London, 1905, p. 113. See also, J. S. Haldane, Methods of air analysis, London, 1912, p. 44. 3 Watson, Journal of the Chemical Society, August, 1911, p. 1460. 66 Composition of the Atmosphere SUMMARY OF HISTORICAL DIGEST.1 While the earliest studies of the composition of the atmosphere can hardly be considered as giving results of quantitative significance, these researches stimulated greatly the study of chemistry in general and air- analysis in particular, the great interest in the composition of the atmos- phere leading to the rapid development of many methods of analyses. Seldom has a philosophical instrument or a chemical process attracted so much attention as did the eudiometer, which utilized the reaction be- tween nitric oxide and air. Although soon discarded for methods better founded scientifically, the apparatus nevertheless was a ready and port- able means for increasing the interest of investigators and diffusing a knowledge of the composition of the air. The successors of this method, 2. e., methods involving the use of absorbents like alkaline sulphides or phosphorus, or employing explosion with hydrogen, all of which depended upon volumetric measurements, soon demonstrated the difficulties in air- analysis — difficulties which taxed the ingenuity and the patience of prac- tically all the prominent chemists. One figure in this early history of air-analysis shines out above all others — that of the scholarly, isolated Scheele. That Scheele may rightly be designated as the pioneer in the study of the chemistry of the air few who examine the literature can deny. His results, while admittedly of no quantitative significance, do nevertheless imply a knowledge of the chem- istry of the air, of its composition, and of the possibilities of change in its composition, which was expressed no more clearly by other writers many years later. Eudiometric observations were exclusively relied upon during the first 50 years of the development of air-analysis, but later gravimetric methods were introduced by Brunner and Dumas in which the oxygen was ab- sorbed by copper or phosphorus, and was subsequently weighed. Then there followed a return to the hydrogen-explosion method, which was ad- vanced to the highest degree of accuracy by Bunsen, Regnault, Frankland and Ward, and Morley. Meanwhile the interesting method of Liebig, employing an alkaline solution of pyrogallic acid, and the copper eudio- meter of von Jolly made their appearance. In all of these earlier researches we find that while the chemical proc- esses involving the absorption of oxygen from the atmosphere were capa- ble of innumerable refinements, the grossest errors were due to purely 1 In compiling the historical material in this book I have been greatly aided by Miss B. Clark, librarian of the William Ripley Nichols Library of the Massachusetts Institute of Technology; and Mrs. Austin Holden, of the Library of the American Academy of Arts and Sciences; and my thanks are especially due to Dr. J. S. Billings and his associate, Dr. Henryk Arctowski, of the New York Public Library. The facilities of the library of the Harvard Medical School, the Library of Congress, and the Surgeon General's Library have also been freely drawn upon. I am also much indebted to Dr. E. P. Cathcart, of Glasgow, 1911-1912 Research Associate of the Carnegie Institution of Washington, attached to this laboratory, for hia painstaking and critical reading of the entire manuscript of this book. History of Air-Analysis 67 mechanical reasons, chiefly to the solubility of gases in water, the diffi- culties of physical measurements, the lack of knowledge concerning the physical properties of gases, the inadequate and incorrect calibrations of the glassware then in use, improper temperature control, and the imperfect preparation of the hydrogen — these factors affecting more or less the accu- racy of the data obtained with the earlier methods. As the more popular chemical processes for the determination of oxy- gen in the air have varied materially — the eudiometric method first being used, then the gravimetric, and finally the eudiometric method again — similarly we find that the prevailing opinion has fluctuated with regard to the constancy or lack of constancy in the composition of the air. When the eudiometer was first used it was firmly believed that the oxygen percentage varied enormously, and, indeed, that the salubrity of any climate was directly proportional to the amount of oxygen present. Just at this time Cavendish, although using an imperfect apparatus, made a remarkable series of experiments, coming to the conclusion that the composition of the air was constant; in other words, that there were no fluctuations that were measurable on his instrument. Then followed the development of the law of gases and of union by volume, with the measurement of the oxygen and nitrogen in the air as approximately 1 to 4, which led to the belief that the air was a chemical compound, having the formula N40. This belief, however, was soon dis- carded, inasmuch as it was found possible to separate the nitrogen and oxygen by mere physical processes, particularly that of diffusion. Evi- dence began to be accumulated to demonstrate that the percentage of oxygen in the air was not sufficiently constant to justify the use of the formula N40; indeed, there appeared to be considerable variation in the composition of the air. As experimental work progressed, however, the variations began to grow less. In a long series of investigations, covering 50 years, no variations in the oxygen content greater than 0.15 per cent were found, save in desultory observations made under conditions that do not inspire the greatest degree of confidence. The only exception was the interesting research in 1887 of von Jolly in Munich, who, by ab- sorbing the oxygen in his copper eudiometer, found much greater varia- tions than had formerly been obtained. Independently and simulta- neously, but employing a somewhat different form of apparatus with the highest grade of technique, Morley in Cleveland found similar results, although the fluctuations were much smaller than those found by von Jolly. Morley's experiments continued over a period of several years, ultimately resulting in the belief that the oxygen content of the air was affected by downward currents, particularly following a sudden drop in temperature. The researches of Morley and von Jolly stimulated further study and were followed by the cooperative investigations of Morley, Kreusler, and Hempel, which showed that under proper control the fluctuations formerly found in part disappeared. 68 Composition of the Atmosphere Finally, as an indication of the present-day conception of the com- position of the atmosphere, the following, written by F. W. Clarke1 in 1908, may be cited: In a roughly approximate way it is often said that air consists of four-fifths nitrogen and one-fifth oxygen, and this is nearly true. The proportions of the two gases are almost constant, but not absolutely so; for the innumerable analyses of air reveal variations larger than can be ascribed to experimental errors. A few of the better determinations are given in the subjoined table [table 49], stated in percentages by volume of oxygen. They refer, of course, to air dried and freed from all extraneous substances. Some of these variations are doubtless due to different methods of determination, but others can not be so interpreted. Hempel, comparing his analyses of air from Trom- soe, Norway, and Para, Brazil, infers that the atmosphere is slightly richer in oxygen near the poles than at the equator, an inference that would seem to need additional data before it can be regarded as established. The most significant variation of all, however, has been pointed out by E. W. Morley.2 As oxygen is heavier than nitrogen it has been supposed that the upper regions of the atmosphere should show a small deficiency in oxygen, as compared with air from lower levels; although analyses of samples collected on mountain tops and from balloons have not borne out this suspicion. It is also supposed that severe depressions of temperature, the so-called "cold waves," are connected with descents of air from very great elevations. Morley's analyses, conducted daily from January, 1880, to April, 1881, at Hudson, Ohio, sustain this belief. Every cold wave was attended by a deficiency of oxygen, the determinations, by volume, ranging from 20.867 to 21.006 per cent, a difference far greater than could be attributed to errors of measurement. Air taken at the surface of the earth seems to show a very small concentration of the denser gas, oxygen. Table 49. — Determinations of oxygen in air, in percentage by volume. Analyst. Locality. Number of analyses. Mini- mum. Maxi- mum. Mean. p. ct, 20.960 20.924 20.943 20.970 20.922 20.930 20.92 20.89 20.864 20.933 V. Regnault R. W. Bunsen R. Angus Smith Do 100 28 32 34 45 46 41 28 20 45 p. ct. 20.913 20.840 20.78 20.80 20.901 20.877 20.86 20.72 20.90 p. rt. 20.999 20.970 21.02 21.18 20.939 20.971 21.00 20.97 20.95 Heidelberg Manchester Mountains of Scotland Near Bonn U. Kreusler W. Hempel Dresden . . Do Tromsoe .... Do Para . A. Muntz and E. Aubin E.W.Morlev Cape Horn Cleveland, Ohio 1 Clarke, Data of geochemistry, U. S. Geological Survey Bulletin 330, 190S, p. 38. 2 Morley, American Journal of Science, 1879, 3d ser., 18, p. 168; 1881, 22, p. 417. PART II. ANALYSES OF ATMOSPHERIC AIR MADE AT THE NUTRITION LABORATORY. From the standpoint of pure physiological chemistry, the importance of an exact knowledge of the composition of the air entering the lungs is obvious when one attempts to consider the various means for studying the respiratory exchange. Every living individual is continually taking into the lungs air of a certain composition, which on leaving the lungs has a different composition. By knowing the volume of air passing through the lungs and the change in composition, important deductions with re- gard to the metabolic processes can be made. Furthermore, in certain lines of physiological experimenting, it is customary to confine a subject inside an air-tight chamber through which a current of ventilating air is passed, the changes in composition of the air inside the chamber being accurately measured. Obviously, here again it is necessary to know the exact oxygen content of the air entering the chamber. Although recognizing that the evidence thus far accumulated shows slight differences in the percentage of oxygen in the air, experimenters for the most part have been content to assume a constancy in this factor for air supplied to a respiration chamber or entering the lungs through nose-pieces or a mouth-piece in an apparatus requiring special appliances for breathing. Singularly enough, however, while assuming a certain degree of constancy, investigators have been at variance in regard to the value to be assigned for the oxygen content of the air. In examining the literature one finds, even in recent researches, variations in the assumed composition all the way from 20.88 to 20.96 per cent. With the Zuntz respiration apparatus and with the Chauveau and Tissot apparatus, the changes in composition of the air passing through the lungs are very great, so that this difference in assumed composition is not of as great magni- tude as it is with other forms of apparatus. For example, air entering the lungs may be assumed to contain 20.93 per cent of oxygen, and de- terminations of the air leaving the lungs may show an oxygen content of 16.93 per cent, or a difference of 4 per cent. Consequently an error of 0.04 per cent in the assumed composition of the air entering the lungs would only make an error of 1 per cent in the total oxygen determined. One of the most important and promising methods of studying the respiratory exchange is that elaborated by Professor Jaquet of Basel, and extensively used by both Staehelin1 in Berlin and by Grafe2 in Heidel- 1 Staehelin and Kessner, Charite-Annalen, 1909, 33, p. 1. 2 Grafe, Zeitschrift fiir physiologische Chemie, 1910, 65, p. 1. 69 70 Composition of the Atmosphere berg. This apparatus is essentially on the Pettenkofer-Voit principle, in that a current of fresh air constantly passes through the chamber. The air leaving the apparatus is analyzed, the change in composition being assumed to have resulted from the metabolic activity of the subject inside the chamber. In his modification of the Pettenkofer-Voit appara- tus, Jaquet has included the determination of the oxygen in the out- going air-current, thus obtaining data regarding the amount of oxygen used, as well as the carbon dioxide produced. With a large ventilation the oxygen deficit may be very small in amount; conversely, the smaller the ventilation, the larger will be the deficit. Unfortunately, most work- ers with this apparatus, although recognizing the fact that the oxygen deficit should be large rather than small, in practice frequently do not heed it and many experiments have been reported in which it is but 0.50 per cent. Under these conditions, therefore, it can be easily seen that an error in assuming the oxygen content of the incoming air may be of con- siderable moment, for each 0.01 variation may make a difference of 2 per cent in the determination of the oxygen absorbed. It might further be said at this point, although it is not necessarily germane to this discussion, that the small oxygen deficit so commonly used by workers with the Jaquet apparatus is likewise enormously affected by analytical errors in determining the oxygen in the out coming air. The Jaquet system is simple, and has many advantages in its favor, but in using it an exact knowledge of the composition of the air entering as well as of that leav- ing the chamber is of fundamental importance. A knowledge of the exact oxygen content of the air inside the respira- tion chamber is also of great importance in using the Regnault-Reiset type of respiration apparatus. For example, assuming that the air in a chamber containing 1000 liters has an oxygen percentage of 20.93, the oxygen con- tent would be approximately 209.3 liters. An error of 0.1 per cent in the determination of the oxygen would therefore result in an error of prac- tically 1 liter in the total amount of oxygen residual in the chamber, so that with an oxygen consumption of approximately 15 liters per hour, the error in the determination of the total oxygen consumption might easily amount to one-fifteenth, or approximately 6 per cent. One of the most important uses of the determination of the oxygen content of the air in the chamber, however, is not so much to obtain an exact knowledge of the amount of oxygen present as to indicate in an elaborate and complicated apparatus the presence of a leakage of air into or out of the system. This point was brought out in a former publica- tion.1 In a closed-chamber apparatus, it is obvious that with an initial volume of 1000 liters of air there can be no change in the nitrogen content by virtue of metabolic processes,2 so that from hour to hour it should re- 1 Atwater and Benedict, Carnegie Institution of Washington, Publication No. 42, 1905, p. 93. 2 Krogh, Skandinavisches Archiv fur Physiologie, 1906, 18, p. 364. Apparatus and Methods 71 main essentially the same. If, however, there is a leakage of air out of the system and a consequent replenishment of oxygen to maintain con- stancy in volume, obviously there would be a loss in nitrogen. On the other hand, if there is a leakage of air into the system, less oxygen will be required to obtain constancy in volume, and the percentage of nitrogen will continually increase. By making, at stated periods, determinations of the nitrogen in the residual air, it is possible not only to detect when there has been a leakage of air into or out of the system, but also from the results obtained to compute easily the magnitude of this leakage.1 This principle has been used recently by Roily2 in making experiments with a small respiration apparatus. At this point it should be stated that throughout this discussion it is considered for convenience that the air used for the determinations of oxy- gen is free from water and carbon dioxide. Furthermore, since the proportion of argon and the rarer gases in the atmosphere does not play any role in this research, no special recognition is made of the presence of 0.94 per cent of argon. It is therefore assumed that the air consists only of nitrogen and oxygen, and that after the absorption of the oxygen the residual gas is pure nitrogen. While there are a large number of methods for determining the carbon dioxide produced by the body, the determination of the oxygen consump- tion is at best a very difficult procedure. When the Regnault-Reiset type of apparatus in this laboratory has been thoroughly tested and shown to be air-tight, air-analyses are unnecessary; nevertheless for long-con- tinued experiments, periodical, accurate determinations of the oxygen in the air residual in the chamber are important. Consequently, the physiological importance of knowing the constancy or lack of constancy in the composition of the air justified the study of this problem by the Nutrition Laboratory. FUNDAMENTAL ESSENTIALS OF ACCURATE AIR-ANALYSES. Although the gravimetric determination of oxygen in air was especially successful in the hands of Dumas and Brunner, it is too time-consuming to be practicable for metabolism experimentation, and hence there has been a general trend in the last 30 years toward the volumetric determina- tion of oxygen by absorption with either phosphorus or potassium pyro- gallate. An apparatus for the determination of oxygen in physiological laboratories, to be successful and practical, should have first an efficient absorbent for oxygen, i. e., the last traces of oxygen should be readily absorbed; second, no by-products of the chemical reaction should be given off into the residual gas, thereby increasing its volume; third, tem- perature changes in the apparatus during the process of an analysis should 1 For an elaboration of this theory and its successful application, see Atwater and Benedict, loc. cit., pp. 88-89, and 93. 2 Roily and Rosiewicz, Deutsches Archiv fur klinische Medizin, 1911, 103, p. 58. 72 Composition of the Atmosphere be fully compensated or readily corrected; fourth, barometric changes in pressure taking place during an analysis should be fully compensated; fifth, there should be an equal tension in the gas before and after absorbing oxygen in the final measurement; and sixth, the contraction in volume as measured should be due only to the absorption of oxygen. Absorbents for oxygen. — Of the numerous absorbents for oxygen, in- cluding phosphorus, potassium pyrogallate, and, more recently, sodium hydrosulphite, there seems to be but little choice with regard to the com- pleteness of absorption. Although both phosphorus and potassium pyrogallate are affected somewhat by low temperature, when properly handled they absorb oxygen completely. While the same is true of sodium hydrosulphite, certain difficulties in the way of handling this re- agent have precluded its general adoption by chemists.1 Formation of by-products. — The absorption of oxygen is invariably an oxidative process. Usually the products of oxidation are non-gaseous, particularly when phosphorus and metallic absorbents are used. On the other hand, it has been claimed that in the interaction between oxygen and potassium pyrogallate a small amount of carbon monoxide is formed. This militated greatly against the use of potassium pyrogallate in the earlier stages of its introduction, but in more recent years a study of its composition has led to changes in the general method of using this re- agent so that the formation of any measurable amount of carbon monoxide has been practically precluded; hence as satisfactory results can be obtained with potassium pyrogallate as with phosphorus. Correction for temperature changes. — From the time when the sample of gas is first measured until after the absorption of either carbon dioxide or oxygen and its subsequent measurement, there should be no material alteration in the volume of the gas due to temperature. Modern ap- paratus corrects for these temperature changes by means of a compensa- ting vessel or pipette of the same size and in the same temperature en- vironment as the vessel used for measuring the sample. Frequently both vessels are immersed in a water-bath which is constantly stirred to secure temperature equilibrium. Barometric fluctuations. — While usually of slight moment, inasmuch as the analyses can be readily carried out in a few minutes, barometric changes should also be taken into consideration. Particularly is this the case in exact gas-analysis when the period of contact between the air and the various reagents must be longer than in the ordinary technical analysis. In some analyses it may require 30 minutes or more for com- 1 The use of sodium hydrosulphite, employed by Tobiesen (Skandinavisches Archiv fur Physiologie, 1895, 6, p. 278), has been more recently brought to the attention of physi- ologists by Durig, of Vienna. (See Biochemische Zeitschrift, 1907, 4, p. 65.) The necessity for preventing the corrosive action of this reagent on glass has called for a certain technique that is not readily acquired; for example, Durig coats the inside of his pipette thinly with gutta percha. The absorption coefficient of the solution of this reagent is extremely high, but as yet the chemical has not come into general use, al- though it can now be readily purchased in a pure form. Apparatus and Methods 73 plete absorption; during this time there may be an appreciable barometric change. Practically all modern gas-analysis apparatus provides for this change in temperature by adjusting the compensating pipette so as to take care not only of the changes in temperature but likewise changes in barometric pressure. Under these conditions it can be safely said that within reasonable limits all changes in temperature and pressure are readily compensated by the modern compensating pipette of the best forms of gas-analysis apparatus. Tension of aqueous vapor. — The varying percentages of water in sam- ples of air and the differences in tension of the aqueous vapor above the absorbing solutions make it necessary to insure that the tension of gases before and after absorption remains exactly the same. This is best se- cured in practically all modern gas-analysis apparatus by saturating the gas with water-vapor both before and after absorption. A satisfactory method for this is the placing of a few drops of water upon the surface of the mercury which is ordinarily used as the liquid for inclosing and measuring the sample. Under all conditions, therefore, the gas as meas- ured is saturated with water-vapor at the temperature of the water-bath. If in the compensating pipette both water-vapor and a slight excess of water are present, then the tension of aqueous vapor is exactly the same in the compensating pipette and in the measuring pipette. Contraction in volume as a measure of the oxygen absorbed. — The most difficult condition in gas-analysis apparatus is to make sure that the con- traction in volume as measured is due only to the absorption of oxygen. The usual procedure in measuring the gas is to read the top of the mer- cury meniscus; obviously, this reading gives not only a measurement of the gas to be analyzed, but of the water-vapor, and also of the liquid water used to insure constancy in the tension of aqueous vapor. The ab- sorption of the gas to be measured changes the level of the mercury, rais- ing it materially; it is assumed that all of the liquid water adhering to the walls of the tube is removed by the mercury as it rises, and that when the mercury meniscus is again read the decrease in volume is due only to the absorption of gas, the volume of liquid water present in the tube above the mercury remaining essentially unchanged. The difficulties experi- enced in proving this assumption have been practically insuperable, and it has been necessary to resort to a reading of the water meniscus, which is at best very unsatisfactory. Fortunately for purposes of compari- son, when essentially the same gas — as atmospheric air, for instance — is analyzed day after day, it is possible to arrange the conditions so as to make the amount of water adhering to the walls of the tube practically constant as the level of the mercury changes. Theoretically, therefore, the best method for analyzing gases is to measure them absolutely dry both before and after absorption in a perfectly dry and clean pipette over absolutely dry and clean mercury. These conditions Dr. Krogh has succeeded in securing in his new gas-analysis apparatus, which unfortuna- tely has not as yet been described. 74 Composition of the Atmosphere Fig. 1. — Sonden air-analysis apparatus. The two pipettes A and B, and the reagent containers C and D, are immersed in water in the glass tank. Stop-cocks a, b, c, and d, permit intercommunication of all parts. Cabron- dioxide percentages are read directly on A, and oxygen percentages directly on B. Mercury reservoirs F and E are connected with pipettes A and B respectively. The manometer M aids in securing equal tension in A and B. Apparatus and Methods 75 APPARATUS AND TECHNIQUE USED IN THIS RESEARCH. The importance of securing the highest degree of accuracy in these analyses led to a critical examination of all the forms of gas-analysis ap- paratus of unusual accuracy now in use. Special attention was given to the apparatus of Haldane,1 Chauveau,2 and Sonden-Pettersson.3 After a careful personal examination had been made of practically every form of exact gas-analysis apparatus in existence, it appeared that the poten- tialities for exact analysis were greatest with the Sonden apparatus. Ac- cordingly I visited Stockholm to make arrangements for securing such an apparatus, but on my arrival was much discouraged to find that none thus far constructed was sufficiently exact for the research proposed. Through the geniality and interest of Dr. Klas Sonden, however, I was able to spend considerable time with him in discussing the conditions to be met; as a result, he designed and superintended the construction of the apparatus herein to be described, which, we believe, fulfills perfectly all of the conditions outlined except the last, i. e., constancy in volume of the liquid water above the mercury. Since Dr. Sonden had previously spent a great deal of time in ex- perimenting with the hydrogen method for determining oxygen and the results had been unsatisfactory, it was decided to use, as the absorbing reagent, a strong solution of potassium pyrogallate, potassium hydroxide being used to absorb carbon dioxide. To insure thoroughly controllable temperature conditions, the entire apparatus, including both the two measuring pipettes and the containers for the reagents, is immersed in a water-bath, nothing but capillary tubing being exposed to the room tem- perature. The apparatus is constructed entirely of glass, and neither the reagent nor the gas is in contact with any other material. The gas volumes are measured over mercury to avoid the solubility of the gases in water, and a few drops of water above the mercury in the measuring pipette insure saturation with water-vapor at the temperature at which the gases are measured. In using the apparatus a volume of air is first measured, the tempera- ture, pressure, and tension of aqueous vapor being exactly equal to a confined volume of air in a compensating pipette. The air sample is then passed into a strong solution of potassium hydroxide by means of which the carbon dioxide is absorbed. The gas is next returned to the original measuring-vessel and the apparent volume arbitrarily adjusted so as to be the same as before the absorption of the carbon dioxide. There is then a slightly decreased pressure of the confined air due to the volume of car- 1 Foster and Haldane, The investigation of mine air, London, 1905. See, also, J. S. Haldane, Methods of air-analysis, London, 1912. 2 Chauveau's apparatus is described in detail by Tissot, Traite de physique biolo- gique, 1, pp. 709-723. 3 Pettersson, Zeitschrift fur analytische Chemie, 1886, 25, pp. 467 and 469; Pettersson and Palmquist, Berichte der deutschen chemischen Gesellschaft, 1888, 21, p. 21-29; Sonden, Zeitschrift fur Instrumentenkunde, 1889, 9, p. 472. 76 Composition of the Atmosphere bon dioxide absorbed. The volume of the air in the compensating pipette is also adjusted to exactly the same pressure and temperature as that in the measuring-vessel, and by measuring the apparent increase in volume of the air in the compensating-pipette, a direct measure of the carbon dioxide absorbed is obtained. The carbon-dioxide-free air is now passed into the potassium pyrogallate to absorb the oxygen. When the air is again returned to the measuring-pipette, the volume is adjusted so that the tension of the residual gas (nitrogen and argon) is exactly the same as the tension of the gas in the compensating-pipette. The amount of con- traction in volume may then be directly read as the percentage of oxygen in the air. DETAILED DESCRIPTION OF THE APPARATUS. A somewhat diagrammatic representation of the various parts of the apparatus is given in fig. 1 ; the apparatus as actually in use is shown in the frontispiece. The main features of this apparatus are two calibrated measuring-pipettes A and B; two reagent containers C and D; a delicate manometer M, and a series of glass stop-cocks permitting intercom- munication between all parts. The whole apparatus is substantially mounted upon a heavy block of marble. The pipettes and reagent containers are immersed in a glass tank filled with water, which is supported at the bottom by a metal ring support firmly fastened to two nickel-plated uprights (81 cm. high), while a similar ring holds the upper part. In the bottom of the glass tank are four holes, through which the ends of the various pipettes and reagent containers pass, these passages being made water-tight by rubber connections. The water in the tank is stirred by allowing a current of air to bubble through it. Under these conditions, therefore, the tempera- tures in the apparatus are uniform throughout, and while a thermometer is suspended in the water-bath, the temperature readings made with it are not essential. The two measuring-pipettes A and B, the two reagent reservoirs C and D, and the manometer M are connected by capillary tubing and glass stop-cocks, so that all five members are fused together into one whole. Stop-cock a connects pipette B either with stop-cock b, which controls the entrances to the two analytical reagent reservoirs, or with stop-cock d, which, in turn, connects with either the outdoor air or with the right side of the manometer M. Stop-cock c connects pipette A with either the outside air or with the left side of the manometer. Stop-cocks / and e at the bottom provide communication between the leveling reservoirs F and E and their respective pipettes A and B. The best quality of stout- walled rubber tubing should be used for connecting the leveling bulb with the stop-cocks/ and e. This is important, since we find that many kinds of rubber tubing contaminate the mercury, so that a thin coating of sul- phide, which interferes seriously with the accurate reading of the meniscus level, forms on the top. Apparatus and Methods 77 Since the changes in the level of the mercury in pipette A are slight (these changes corresponding to the carbon-dioxide percentages of normal air) the leveling bulb F is not usually moved. On the other hand, the leveling bulb E is hung alternately on the upper and lower hooks in order to expel air from pipette B into the solution in either C or Z), or out into the air at the completion of the analysis. Minor changes in level of the mercury in the two graduated pipettes are produced by pressure on the rubber tubing with the delicate screw-cocks P and N. By shutting the glass stop-cock / below and screwing in the cock P, pressure can be pro- duced against the rubber tubing so as arbitrarily to adjust at will the level of mercury in the capillary portion of A. A similar adjustment of the mercury level in pipette B may also be made by means of the screw-cock N. In the manometer M a small drop of light petroleum oil serves as an index. When the stop-cocks c and d are removed so that there is atmos- pheric pressure on each side, this globule should stand in the exact cen- ter of the manometer if the apparatus is properly leveled. All stop-cocks are well ground, perfect in fit, and lubricated by a thin layer of mutton tallow. The two calibrated pipettes A and B may be designated respect- ively the compensating and the measuring pipettes, although measure- ments are actually made in both. Compensating pipette. — The compensating pipette is used not only for the final adjustment of the volume of gas in the measuring pipette B, but also for reading directly in its lower graduated portion the percentage of carbon dioxide. The volume of this pipette from the zero mark to the glass stop-cock is GO cubic centimeters. The graduations demand a spe- cial discussion. In all measurements made with this pipette, the gas in both A and B is under definite, though slight, decrease in tension. If the volume of gas in the measuring pipette has been decreased one-thou- sandth by the absorption of carbon dioxide by the potassium hydroxide, to adjust the air in the compensating pipette to the same tension, the vol- ume of air must be expanded one-thousandth by lowering the mercury a certain amount in the graduated portion of the tube. In the graduation of this pipette, therefore, due cognizance has been taken of the altera- tion in pressure in the pipette B. The graduations are so arranged as to make the instrument direct reading, the level of the mercury in pipette A indicating the percentage of carbon dioxide in the air sample. This instrument was primarily designed to study the percentage of oxygen in the air-content or the ventilating current of a respiration cham- ber. Under these conditions the carbon dioxide may at times be nearly 1 per cent of the air and the deficiency in oxygen approximately the same. The pipette A is so constructed that percentages of carbon dioxide as great as 1 per cent can be measured. Each small division indicates 0.01 per cent and tenths of divisions are readily estimated by the eye, so that records may be obtained with three significant figures when the percent- age of carbon dioxide is greater than in normal air. 78 Composition of the Atmosphere For the purpose of this discussion, however, we have to deal only with small percentages of carbon dioxide — never over 0.08 — and the special extended graduation of this pipette is only of incidental interest. The variations usually found in the carbon-dioxide content of outdoor air are so small that all adjustments of the mercury-level can be made by the adjusting screw P without disturbing the leveling-bulb F. Measuring pipette. — The measuring pipette B has two bulbs, the zero mark being placed slightly below the lower bulb. Between the two bulbs is a constricted portion which is graduated and represents that part of the pipette corresponding to from 19.5 to 21.0 per cent of the total volume. This constriction favors the accurate determination of oxygen, since after the gas is absorbed from a sample of air and is again drawn into the pipette, the mercury-level must be raised a sufficient amount to correspond to the volume of oxygen absorbed. This is usually between 19.5 to 21.0 per cent in experiments with the respiration chamber, and for outdoor un- contaminated air is generally not far from 20.940 per cent. The gradua- tions are such that each scale division corresponds to 0.01 per cent, and hence direct readings may be accurately made to 0.001 per cent. To avoid errors in parallax, the graduation marks on both pipettes completely circle the glass. The total content of pipette B is 60 c. c. A few drops of water are placed in both pipettes A and B and the air is thus continually saturated with water-vapor. The upper end of each pipette is connected by capillary glass tubing to the various stop-cocks, while the lower end passes through a hole in the bottom of the glass reservoir and is there connected with the adjusting screws P and N, the stop-cocks / and e, and the leveling-bulbs F and E, respectively. Special water-tight closures are necessary where the end of the pipette passes through the glass. These are shown in detail in fig.l. After the absorption of oxygen, the level of the mercury in pipette B indicates the percentage of oxygen in carbon-dioxide-free air. When ordinary outdoor air with a carbon-dioxide content of but 0.030 per cent is to be considered, it matters but little, so far as the expression of the percentage of oxygen is concerned, whether the air is carbon-dioxide-free or not, since in one case the proportion would be 20.93 c. c. of oxygen in 100 c. c. of air and in the other 20.93 c. c. in 100 c. c-0.03 c. c, or 99.97 c. c. of carbon-dioxide-free air; thus the percentage would not be measur- ably affected. When, however, there may be 0.8 per cent of carbon di- oxide in air taken from a respiration chamber, it becomes a matter of some moment whether the percentage is of carbon-dioxide-free or of carbon- dioxide-containing air. After the absorption of the carbon dioxide, the mercury-level in the compensating pipette is so adjusted that the air in this vessel is under slightly decreased tension — the decrease in tension being equivalent to the volume of the carbon dioxide absorbed. The final adjustment of the Apparatus and Methods 79 gas in the measuring pipette likewise produces a corresponding decreased tension. As the pipette is graduated in percentages, it obviously makes no difference whether the gas measured is at atmospheric or less than at- mospheric pressure, provided the residual gas is at the same pressure. For this reason, then, when the level of the mercury in the measuring pipette B is adjusted after the oxygen is absorbed, no change is made in the level of the mercury in the compensating pipette A. The reading for the oxygen, therefore, represents the percentage of oxygen in carbon- dioxide-free air. Reagent containers. — To insure rapid and efficient absorption, the reagent should be contained in a vessel capable of exposing a relatively large area of reagent to the gas. Since the relation between the area and the volume of gas depends in large part upon the amount of gas to be absorbed, in the case of carbon dioxide, of which no more than 1 per cent is ever present, the area may be very much less than with oxygen of which over one-fifth is absorbed. The forms of reagent containers found most advantageous for this type of apparatus are shown as C and D in fig. 1. As will be seen, the general structure of both vessels is the same, the notable difference being that D contains a large number of short glass tubes which increase greatly the absorbing surface. To prevent the tubes from falling out, the opening at the bottom is somewhat more con- stricted than in C. Each container is so constructed that when filled with the reagent the level inside and out is the same; special marks not only on the inner capillary tube but also around the outer glass envelope of the chamber aid in introducing the proper amount of reagent and in setting the level of the reagent in the capillary. A short length of glass tubing fused to the chamber projects above the level of the water in the tank and, being open to the air, insures atmospheric pressure on the outer surface of the reagent. This tube serves to introduce the various reagents for absorbing the gases to be determined. The lower ends of both cham- bers project through holes in the glass bottom of the tank, the same pre- cautions for water-tight closure observed in the case of the pipettes A and B being also here taken. REAGENTS USED. The solution for absorbing carbon dioxide. — The solution for absorbing the carbon dioxide is prepared by dissolving 2400 grams of stick potassium hydroxide in 1750 c. c. of water. After the solution becomes cold it is decanted to remove all sediment, and the reservoir C is filled through the tube at the top. This is best accomplished by inserting a 20 cm. length of small-sized glass tubing in the open tube and attaching a funnel to it by means of a short bit of rubber tubing. This elongated funnel con- ducts the reagent well down into the chamber and avoids choking the passage with the somewhat viscous liquid. The level of the liquid inside and out should be essentially the same, the usual height being shown in 80 Composition of the Atmosphere fig. 1. Care should be taken to prevent dust from entering the tube or liquid, as it is liable to accumulate on the inside of the reservoir and ulti- mately cause difficulty in reading the exact height of the liquid in the capillary tubing. The solution for absorbing oxygen. — Shortly after Liebig announced his discovery of the fact that potassium pyrogallate absorbed oxygen quanti- tatively, the criticism was raised by Calvert, Boussingault, and Cloez1 that potassium pyrogallate, when reacting with oxygen, gave rise to the formation of a certain amount of carbon monoxide, the evolution of this gas naturally vitiating the results obtained by this method of analysis. Several years later a further caution was published by Hempel to the effect that one should not use potassium hydroxide "purified by alcohol" in the preparation of the potassium pyrogallate.2 In this defense of the use of the pyrogallate solution, he also pointed out that if proper attention was given to the concentration of the solution there need be no fear of the evolution of carbon monoxide. When the formulas for preparing this solution are examined, it is found that the main differences noted are in the proportion of water pres- ent. When "a 60 per cent solution of caustic potash" is stated, one im- mediately has to determine what is meant by "a 60 per cent solution." Even when the weight of potassium hydroxide is given in the formula, difficulty is experienced owing to the differences in water-content of the substance, stick potassium hydroxide frequently containing as much as 25 per cent of water. Haldane's formula alone obviates this difficulty, as he requires a fully saturated solution of potassium hydroxide with a specific gravity of 1.55. Earlier experience with Haldane's formula, however, showed that with a low room-temperature the material solidi- fied; and with so delicate an apparatus as that of Sond£n it seemed un- desirable to introduce a reagent that might solidify and possibly burst the glass container. Since the prime object of this research was a comparative study of the oxygen content of the air, and since certain fundamental defects in the apparatus prevented deductions regarding the exact absolute value, we modified slightly Haldane's solution as follows: A solution of potassium hydroxide was prepared by dissolving 500 grams of stick potassium hydroxide, not purified by alcohol, in 250 c. c. of water. Usually the specific gravity of the resulting solution was 1.55. During the progress of this research, several shipments of stick potassium hydroxide were used, and the varying water-content of the chemical is shown by the fact that it was frequently necessary to add more potassium hydroxide to bring the solution to the desired density. To 135 c. c. of this saturated solution was added a solution of 15 grams of pyrogallic acid 1 Calvert, Comptes rendus, 1863, 57, p. 873; Cloez, Comptes rendus, 1863, 57, p. 875; Boussingault, Comptes rendus, 1863, 57, p. 885. s Hempel, Berichte der deutschen chemischen Gesellschaft, 1887, 20, p. 1865. Apparatus and Methods 81 in 15 c. c. of distilled water. By means of the funnel and tube, the mixed solutions were then carefully introduced into chamber D. One such charge of potassium pyrogallate was found to be sufficient to make approximately 30 air-analyses. This solution takes up oxygen rapidly and has a high absorptive ca- pacity. It has been assumed that since the solution was so much more concentrated than Hempel's, his assertion that no carbon monoxide was developed with his weaker solutions held true in this case also, particu- larly as Haldane states that with his extremely concentrated solution no traces of carbon monoxide are found. Furthermore, certain evidence here presented seems to support this view. When a known sample of air is analyzed a number of times, the percentage of oxygen at the beginning of the series does not differ from that found at the end, even when as many as 30 analyses are made with the one charge of potassium pyrogallate. It seems reasonable to suppose that if carbon monoxide were formed, a somewhat different amount would be produced after the first, second, or third analysis than after the twenty-eighth or twenty-ninth. On the other hand, it is not impossible that in the production of carbon monoxide there may be an extremely small quantitative relationship between the oxygen absorbed and the disintegration of the pyrogallic acid, so that the carbon monoxide given off may remain strictly proportional to the amount of oxygen consumed. Since in each of these analyses exactly the same amount of oxygen is absorbed, there still may be a slight constant factor present; consequently, it is necessary to take into consideration the fact that in all of these analyses there may be traces of carbon monoxide produced. In that case the tendency would be to make the percentage of oxygen slightly too small. Although throughout the whole research a slightly modified Haldane solution was used, subsequent experiments with the Haldane formula show a somewhat larger oxygen percentage. This increased percentage may be due to two causes: (1) the actual ab- sorption of more oxygen, or what is more probable, (2) the formation of less or no carbon monoxide. It remains a fact, nevertheless, that the solution as used is without question suitable for a comparative study of the oxygen percentage of the atmosphere. 82 Composition of the Atmosphere PLAN AND METHODS OF RESEARCH. The report of the research carried out with this apparatus may be sub- divided into several parts as follows: (1) The main study of the comparative oxygen-content of uncontami- nated outdoor air under all conditions as to wind direction and strength, temperature, cloud formation, barometer, and weather, including rain, snow, fog, and mist. (2) A study of the influence of the temperature of the reagent upon its absorptive power. (3) An examination of samples of air collected on the North Atlantic Ocean between Montreal and Liverpool, and between Genoa and Boston. (4) Analyses of air obtained from the top of Pike's Peak. (5) Analyses of air taken in the crowded streets of Boston. (6) Analyses of air taken in the Boston and New York subways. (7) An experimental research with various absorbents for oxygen. Before proceeding to a description of the routine of air-analysis with this apparatus, a discussion of the method of procuring samples of un- contaminated outdoor air is advisable. By far the greater number of analyses were made of air collected near the laboratory, and a permanent installation was made to secure convenient and accurate sampling of air. It is unnecessary at this point to describe the methods of sampling em- ployed when samples were taken at some distance from the laboratory. METHOD OF COLLECTING OUTDOOR AIR. As the laboratory is located near a large power-house, it was feared that in spite of all precautions there would be a contamination of the at- mosphere due to the products of combustion from the large furnaces. On the other hand, as the prevailing winds are from the southwest and the power-house is north of the laboratory, it seemed probable that during the prevailing winds the contamination should not be perceptible. In order to provide for a possible variation in composition on two sides of the building, arrangements were made for taking samples on both the east and west sides. The sampling arrangements are as follows: A standard ^s-inch brass pipe (7 mm. internal and 10 mm. external diameter) was extended out from the west wall of the laboratory, 4.8 meters from the ground and at a distance of 2.2 meters from the wall. The end was pointed downward so as to prevent clogging by water, ice, or dirt. This pipe was then brought into the laboratory, conducted to the sink near the gas-analysis apparatus, and a water-suction pump so connected as to suck continuously a current of air from outdoors through the pipe. The intake tube of the gas-analysis apparatus was attached to the air-pipe, so that it was possible to have a continuous stream of fresh outdoor air passing by the analysis apparatus. Similarly, a second pipe was carried out from the east side of the building at the same distance from the ground as the pipe on the west side. This pipe was also con- Comparative Air-Analyses 83 nected with the gas-analysis apparatus, so that by turning a valve the water-suction pump could be connected with either the west or the east pipe and fresh outdoor air drawn from either side of the building at will. In all analyses care was taken to have the suction pump in full operation for several minutes before taking the sample, thus insuring a complete sweeping out of the pipe by fresh outdoor air. With all analyses simul- taneous records have been made of the weather, the direction and strength of the wind, the outdoor temperature, and the barometer. METHOD OF USING THE APPARATUS AND RESULTS OBTAINED. The best procedure in the use of this apparatus is by no means obvious from an inspection of its construction; furthermore, as errors appeared in the technique and in the apparatus, the routine has been altered funda- mentally on several occasions. Inasmuch as the results first obtained are so completely in harmony with many of those of the earlier inves- tigators, it seems advisable to give our entire series here, even though we know that the results of the first two years are open to objection, owing to slight errors which disappeared as the routine became more perfected. While the investigation was started primarily to study the oxygen- content of the outdoor air, it was necessary to determine beforehand the carbon dioxide, since an alkaline absorbent for oxygen was employed; hence practically all the analyses are accompanied by simultaneous de- terminations of the carbon dioxide in the air. In the especially exact ap- paratus designed by Sonden and Pettersson, the carbon dioxide is de- termined to the third or fourth significant figure, but as the amounts of carbon dioxide that were to be used in our apparatus might at times reach 1 per cent, it was impossible to secure this degree of fineness in the cali- bration of the carbon-dioxide pipette, hence readings can be taken only to one-thousandth of 1 per cent. Consequently, since other methods are better adapted for securing accurate carbon-dioxide determinations, little stress has been laid upon the determinations made in connection with this research, although they are probably accurate to within 0.002 in all cases. At one time during the research it was found that the potassium hydroxide reagent chamber C was broken, so it became temporarily necessary to absorb simultaneously the carbon dioxide and oxygen; hence for a short period the percentages represent the percentage of oxygen plus that of carbon dioxide. As soon as possible the reagent chamber was repaired and the research was then continued in the usual way. FIRST ROUTINE, AND RESULTS OBTAINED. The earlier results in this series are especially interesting as indicating how it is possible by constanc}r in routine to secure duplicate results on practically all samples. Furthermore, it is interesting to note that if the research had been discontinued at the end of the second year, the results could easily have been taken as verifying completely those of the earlier investigators, who showed that fluctuations in the oxygen content of the atmosphere are to be expected, slight though they may be. 84 Composition of the Atmosphere The apparatus described was first set in order in the latter part of February 1909. After considerable preliminary experimenting with room air and with air from a respiration chamber, a series of analyses of out- door air was begun on April 5, 1909. The first routine employed for de- termining the carbon dioxide and oxygen is as follows: Outline of first routine. — It is necessary in the first place to make sure that all the capillary tubes communicating with the different reservoirs are filled with nitrogen and not with air. For this purpose a blank analy- sis is made in which quantitative accuracy is not required. In making this analysis, the air in the apparatus is first passed into the potassium pyrogallate several times until thorough absorption of both carbon dioxide and oxygen is assured; it is then allowed to flow into the potassium hy- droxide and repeatedly drawn back and forth by means of the leveling bulb E until the air in all of the capillary tubes is replaced by nitrogen, the air being intermittently sent into the potassium pyrogallate to absorb the slight traces of oxygen picked up in its passage through the capillary tubing. The level of the potassium hydroxide in the capillary tube is then brought to a definite point by lowering the mercury in the pipette B, the final adjustment being made by the screw-cock N; a decreased ten- sion is thus produced which raises the reagent to the desired mark. When this point is reached the stop-cock b is turned 180 degrees to communi- cate with the chamber D. We now have pure nitrogen in the capillary tube leading from the potassium hydroxide reservoir to the stop-cock b. Since all the other capillaries are likewise filled with pure nitrogen, pressure is applied at screw-cock N to bring the potassium-pyrogallate solution to a definite mark on the capillary tube before shutting off the stop-cock b. After turning stop-cock a 180 degrees, the excess of nitrogen is then ex- pelled through the stop-cock d by means of the leveling bulb E. Between the stop-cock d and the pipe coming from the outside of the laboratory is a three-way stop-cock, one side of which is opened to the room air. When taking the sample of air for analysis, this three-way stop-cock is so turned as to allow direct communication between the gas- analysis apparatus and the sampling pipe through which a suction pump draws a current of outside air. Under these conditions, by lowering the mercury reservoir E, mercury runs out of the pipette B and air is drawn in through the capillary stop-cock d. The leveling bulb is again raised and the pipette repeatedly swept out by pure air. When a thorough and accurate sweeping out of the nitrogen is insured and the pipette is full of uncontaminated outdoor air, the sample is ready to be measured. The mercury is finally lowered to a mark somewhat below the zero mark on the pipette. The suction pump is then stopped and the three-way stop-cock between the gas-analysis apparatus and the sample pipe (not shown in either figure) turned so as to communicate directly with the room air. By raising the level of the mercury to the zero mark on the bottom of the pipette B, a slight amount of air is expelled and the Comparative Air-Analyses 85 level is set exactly at zero by the fine adjusting screw N. The stop- cock c is then turned so that the compensating pipette A communicates directly with the room air, and by means of the finely threaded adjust- ment screw P, the level of the mercury in A is also brought to the zero point. Under these conditions, therefore, the gas in both pipettes is at the same temperature and pressure. To test this the stop-cocks c and d are simultaneously and cautiously turned and the two reservoirs A and B placed in communication with the manometer; there should be no move- ment of the oil-drop on the scale, thus indicating constancy in tempera- ture and barometric conditions. The stop-cocks c and d are again turned so as to cut off the manometer, and the sample is ready for analysis. The gas is first sent into the potassium hydroxide to absorb the carbon dioxide. This is done by turning the stop-cock a 180 degrees and subse- quently turning stop-cock b so as to allow the air to enter the capillary leading to C. It is next slowly passed into the potassium hydroxide twice by raising and lowering the mercury leveling-bulb E. The air is then drawn back into B, the level of the reagent is again set, and the level of the mercury in the pipette B is brought down to the original zero-mark. Under these conditions there is a slightly diminished pressure in the pi- pette B, due to the decrease in volume it has sustained by the absorption of the carbon dioxide; consequently, before connecting pipette A to the manometer, the pressure of the air inside A, which is now acting as the compensating vessel, must be somewhat decreased. This decrease in pressure is obtained by lowering the mercury column by an amount which is estimated as being approximately the percentage of carbon dioxide in the air. For ordinary outdoor air this is not far from 0.03 per cent. Under these conditions, therefore, when the stop-cocks c and d communicating with the manometer are turned there will be a slight de- flection of the oil globule on the scale to the right or to the left, if the pressure is greater in either A or B. For example, if the movement of the petroleum globule is toward the right, it signifies that the pressure in the pipette A is greater ; accordingly, it is necessary to lower the mercury in A until the petroleum drop is exactly in the center. When the manometer shows that the temperature and pressure are the same in both pipettes, a direct reading of the percentage of carbon dioxide may be obtained by noting the fall of the mercury level in A. To obtain a second reading, the manometer stop-cocks are again turned 90 degrees, and then, by properly turning the stop-cocks a and b, the air is once more sent into the potassium-hydroxide solution. The level of the reagent is again set and after communicating with the manometer the reading is taken. The second reading is usually identical with the first, as the amount of carbon dioxide to be absorbed is extremely small. At this point the carbon dioxide has been completely removed from the gas in the apparatus, but the capillary tube between the stop-cock 6 and the level of the potassium hydroxide now contains air instead of nitrogen. 86 Composition of the Atmosphere By properly adjusting the stop-cocks a and b the air sample is slowly sent into the potassium pyrogallate pipette and allowed to remain one minute. After being drawn back and forth twice, it is left over the potassium py- rogallate for another minute, again drawn back and forth twice, next sent to the potassium hydroxide pipette three times, and finally into the po- tassium pyrogallate three times. When the air leaves each separate pipette for the last time the level for the potassium hydroxide and the potassium pyrogallate respectively are exactly set. Under these con- ditions there will be a marked difference in the level of the mercury in the pipette B, owing to the absorption of the oxygen. In adjusting the level of the mercury in this pipette, instead of drawing it to the zero point, it is brought back until it remains in the upper part of the grad- uated portion of the pipette. If outdoor air is being analyzed, and the composition is known with considerable exactness, it can usually be set not far from 20.9. Then by communicating with the manometer, and noting whether the oil index moves to the right or to the left, the mercury in the pipette B may be raised or lowered as necessary, without altering in any way the level of the mercury in the pipette A, until a position is finally obtained which indicates constancy in temperature and pressure conditions exactly like those of the air in the compensating pipette A. A reading of the percentage of oxygen is now taken. After turning the manometer stop-cocks, the air is sent into the potassium pyrogallate three times after each reading, and readings are taken until they agree within 0.002 of each other. The routine outlined was followed with practically no modification from April 5 up to Nov. 3, 1909. The details of an analysis made on April 5 at llh45m a. m. and carried out with this routine are given in table 50. Table 50. — Results obtained on sample of outdoor air with first routine, April 5, 1909, llh45m a. m. Reading. Carbon dioxide. Oxygen. First Second Third Fourth Fifth p. ct. 0.029 0.031 p. ct. 20.S93 20.911 20.923 20.929 20.929 Results with first routine.— The total results for the first stage in the de- velopment of this method, namely, from April 5 until November 3, 1909, are given in table 51. It should be noted that these residts are not of selected analyses, but include the records of every analysis made during this time, including both good and bad. In the analyses from May 28 to June 3 it was necessary to absorb carbon dioxide and oxygen simultaneously, owing to the fracture of the potassium-hydroxide chamber C previously referred to. An examination of the data shows that on the whole there is no material difference between the analyses made of air taken from the Comparative Air-Analyses 87 Table 51. — Analyses of outdoor air made at the Nutrition Laboratory.1 Series Date. 1909. Apr. 5 Apr. 6 Apr. 8 Apr. 9 Apr. 10 Apr . 12 Apr. 17 Apr. 19 Apr. 20 Apr. 21 Apr. 22 Apr. 23 Apr. 28 Apr. 29 Apr. 30 May 1 May 3 May 8 May 10 May 12 May 13 May 14 May 17 May 18 May 20 May 21 May 22 May 22 May 24 May 25 May 26 May 27 May 28 May 29 June 1 June 2 June 3 Oct. 18 Oct. 19 Time. llh45ma.m. 4 00 p.m. 11 00 a.m. 11 00 a.m. 5 00 p.m. 10 00 a.m. 12 00 noon 10 00 a.m. ii 66 a.m. a.m 3h00mp.m. a.m. p.m. Tem- pera- ture. DC. 16.4 11.0 S.3 6.8 11.0 13.7 12.1 23.7 8.5 6.8 14.0 14.7 10.9 13.3 2.7 6.1 11.0 23.3 21.9 23.2 26.0 10.1 13.0 13.9 8.5 5.5 28.0 19.0 31.2 14.3 12.2 21.6 23.7 Ba- rom- eter. mm. 764.10 761.35 758.50 756.50 760.65 774.50 767.50 767.15 752.50 765.35 769.90 756.30 760.30 760.00 770.50 763.65 Wind. Light SW . . . E SW. gale . . . Weather. CdtT Oxy- ide. I Sen- SE. w. s . Pleasant .Do. .Do. Rain S .., s. .. sw NE. Pleasant West side. E.. W. N. W. Pleasant Pleasant N. E.. ....Do. Rain . , 753.50 766.60 N SW Rain Pleasant 760.50 764.85 SW w Pleasant . . ..Do 764.00 757.75 761.00 769.65 770.15 766.00 764.00 756.70 762.85 769.10 766.70 756.70 752.60 762.00 764.30 760.55 SW. w.. E... E. NE NE. strong . W NW NW S N N NW SE SW Rain . . . Pleasant 771.45 NW. Stormy . Pleasant Pleasant p. ct. 0.031 .032 .028 .026 .029 .030 .027 .030 .027 .029 .030 .029 .030 .031 .032 .030 .031 .031 .031 .032 .032 .030 .030 .032 .030 .028 .028 '.028 .029 .029 .031 .028 .030 .029 .029 .029 .031 .029 .028 .030 .029 .031 p.cl. 20.929 20.921 20.920 20.921 20.922 20.918 20.919 20.920 20.921 20.920 20.920 20.901 20.911 20.921 20.921 20.922 20.920 20.921 20.920 20.901 20.902 20.919 20.921 20.920 20.921 20.922 20.920 20.9ii 20.919 20.921 20.920 20.920 20.912 20.911 20.901 20.921 20.902 20.902 20.922 20.920 ; 20.921 20.920 Pleasant Cloudy . Pleasant 20.950 20.950 20.949 20.940 20.949 .031 .032 .028 20.920 20.919 20.919 East side. Car- bon diox- ide. p. ct. 0.029 .028 .029 .028 .026 .030 .030 .031 .029 .030 .029 ;.030 .032 '.029 '.030 .030 .030 .032 '.029 .031 .028 .034 .032 .028 .032 .030 .029 .029 '.030 .028 .030 '.028 .029 .030 .029 Oxy- gen. p. ct. 20.920 20.941 20.880 20.921 20.931 20.921 20.919 20.920 20.920 20.921 20.921 20.931 20.902 20.920 20.922 20.920 20.920 20.902 20.9i9 20.920 20.921 20.920 20.920 20.910 20.918 20.920 20.919 20.919 20.9i0 20.920 20.902 20.920 20.920 20.919 20.920 1 In all of the analyses of air made in this research, the temperature of the water bath varied from 17° to 21° C. and was usually not far from 19° or 20° C. Subsequent experiments (see p. 97) showed that tem- perature variations do not materially influence the results. 2 In this analysis the mercury in the compensation tube was inadvertently brought back to 0 after the carbon dioxide had been absorbed. west side of the building and those from the east side. In view of these results it was not considered necessary to draw samples from both sides of the building, and only those drawn from the west side were analyzed for the remainder of the research. Aside from certain values which are ob- viously erroneous, it would appear that the average percentage of oxygen was not far from 20.92, as indicated by this technique and under these Con- di tions. The fluctuations in the percentage of carbon dioxide are those commonly experienced and represent nothing unusual, save that on days when the wind was blowing directly from the thickly settled por- tion of the city a much higher carbon-dioxide content than at other times 88 Composition of the Atmosphere would be expected instead of the constancy indicated. The laboratory is so situated that the wind from the southwest would come from the more residential portion of the city, while the wind from the northeast and east would come from the business and factory part of the city. The power-house of the Harvard Medical School is located exactly north (100 meters) from the laboratory, but aside from this, no large factories or other smoke-producing buildings are nearer than 600 meters. Errors in first routine. — In October 1909 it was found that if the trans- fer of gas from pipette B to the potassium pyrogallate was continued for some time, there was usually a steady, though slight, increase in the per- centage of oxygen, this increase amounting to from 0.001 to 0.002 per cent for each repetition of the routine. Furthermore, the increase continued until the percentage of oxygen had risen from 20.92 to 21 and over, when it could no longer be read accurately, as the graduations extended no farther. It was believed at that time that this increase was due to the distillation of water from the pipette B over into the solution of potassium hydroxide. The theory was that the strong alkali had a tension of aque- ous vapor considerably less than that of water, and that each time that the air was sent over into the potassium pyrogallate it carried with it a slight amount of moisture; this moisture was retained by the strong alkali until all the water was gradually distilled over. SECOND ROUTINE, AND RESULTS OBTAINED. Since it seemed desirable to minimize as much as possible the trans- fers of air from the measuring pipette into the strong alkali, the first rou- tine was modified somewhat, and a second routine adopted on November 3, 1909. The following changes were made in the method : The absorption of the carbon dioxide was unchanged, the variation in the routine being chiefly in the determination of oxygen. After the carbon dioxide was absorbed, the gas was sent into the potassium pyro- gallate and allowed to remain for 10 minutes. It was then withdrawn and after being sent into the potassium hydroxide was again returned to the potassium pyrogallate and allowed to remain 5 minutes. Then the first reading was taken. This procedure, i.e., once into the potassium hydrox- ide and a 5-minute sojourn over the potassium pyrogallate, was carried out three times, readings being taken as each routine was concluded. Results ivith second routine. — The second routine was followed almost without change from November 3, 1909, until February 15, 1911, sam- ples being taken only on the west side of the laboratory. A sample analy- sis made on November 4, 1909, is given in table 52. The detailed results for this series of analyses are given in table 53, in which are incorporated, likewise, the temperature of the outdoor air, the barometer, and data regarding the wind and weather, as well as the times at which the analyses were made. These analyses were continued over a period of more than a year, the summer months only being excepted. Comparative Air-Analyses 89 Table 52. — Results obtained on sample of outdoor air with second routine, November 4, igog, gb/om a. m. Readin*. Carbon dioxide. Oxygen. First p. ct. 0.035 0.036 p. ct. 20.888 20.909 20.910 Second Third Table 53. — Analyses of outdoor air made at the Nutrition Laboratory. Series 2. [All samples were taken from the west side of the laboratory.] Date. 1909. Nov. 4 Nov 5 Nov. 6 Nov. 8 Nov. 10 Nov. 11 Nov. 12 Nov. 13 Nov. 15 Nov. 16 Nov. 17 Nov. 18 Nov. 22 Nov. 23 Nov. 24 Nov. 29 Dec. 9 Dec. 21 Dec. 22 Dec. 27 Dec. 28 Dec. 29 Dec. 30 Dec. 31 1910 Jan. 1 Jan. 3 Jan. 4 Jan. 5 Jan. 6 Jan. 6 Jan. 7 Jan. 7 Jan. 8 Time. 9h10ma.m. 3 12 p.m. 10 00 a.m. 9 00 a.m. 11 25 a.m. 9 16 a.m. 8 47 a.m. 8 56 a.m. 8 57 a.m. 8 34 a.m. 10 34 a.m. 9 45 a.m. 9 14 a.m. 8 46 a.m. 3 31 p.m. 9 37 a.m. 10 34 a.m. 9 30 a.m. 8 32 a.m. 11 16 a.m. 2 37 p.m. 2 52 p.m. 10 24 a.m. 9 27 a.m. 8 57 a.m. Temper- ature. "C. "4.0 8.7 7.8 11.7 17.8 11.7 11.7 6.9 14.9 4.8 6.8 17.9 0.0 3.0 1 (') (') (') (') — 6 —12 — 8 8 45 a.m. — 3 8 40 a.m. i 1 9 21 a.m. ! — 14 8 50 a.m. ! — 15 8 58 a.m. 6 a.m 6 p.m 8h43ma.m. •2 49 p.m. 8 39 a.m. 2 — 4 Barom- eter. mm. 753.25 763.60 774.95 779.90 778.00 767.05 775.55 769.90 769.90 751.75 766.00 752.50 762.00 764.70 760.00 760.77 756.90 749.00 749.10 752.50 753.00 755.55 766.95 769.65 761.85 775.00 783.30 761.35 759.65 756.00 758.10 778.90 Wind. Weather. Pleasant . Cold, raw SW sw w Very little, if any No wind SW., very little . NW Brisk SW Brisk NW SE SW Strong NE. NE W SW NW Cold, raw Pleasant . ....Do. .. ....Do. .. Overcast ....Do. .. Pleasant Rain . . . Overcast Rain . . . W. W. Very little, SW. Light WNW. . NW W SW. SW. NW W.. SW. SW. NW. NW. W . Overcast; rained night before, sun out part of time.1 Snow.-sleet Stormy Pleasant Pleasant, sunny Sunny, bright . . {Bright, sunny"- snow storm day before, about 30 cm. snow. Pleasant Foggy . . Sunny . . Pleasant Sunny . . Cloudy . Bright . Snowing Rain . . . .Do. Heavy rain Pleasant ...Do. . Carbon dioxide. Oxygen. p. ct. p. ct. 0.036 20.910 .029 20.932 .033 20.949 .035 20.942 .035 20.940 .039 20.920 .035 20.9i,5 .034 20.920 .033 20.923 .035 20.930 .034 20.930 .030 20.920 .031 20.931 .035 20.930 .034 20.927 .035 20.923 .033 20.922 .030 20.931 .030 20.933 .030 20.921 .031 20.923 .033 20.930 .034 20.930 .033 20.925 .034 20.920 .029 20.922 .031 20.921 .027 20.921 .031 20.923 .030 20.919 .029 20.910 .031 20.900 .032 20.920 .028 20.921 .031 20.921 029 20.910 .032 20.923 .041 20.903 .043 20.902 .029 20.919 .031 20.922 .028 20.911 .030 20.912 .031 20.913 .030 20.913 .030 20.921 .029 20.921 .032 20.919 .042 20.933 .035 20.932 .034 .036 .040 20.920 .032 20.930 .034 20.929 .031 20.932 .031 20.922 'Below zero. 90 Composition of the Atmosphere Table 53. — Analyses of outdoor air made at the Nutrition Laboratory. Series #.— Cont'd. [All samples were taken from the west side of the laboratory.] Date. 1910. Jan. 10 Jan. 17 Jan. 18 Jan. 19 Jan. 24 Jan. 25 Jan. 26 Jan. 27 Jan. 28 Jan. 29 Jan. 31 Feb. 1 Feb. 2 Feb. 4 Feb. 5 Feb. 7 Feb. 8 Feb. 11 Feb. 12 Feb. 17 Mar. 21 Mar. 24 Apr. 1 Apr. 4 Apr. 5 Apr. 7 Apr. 8 Apr. 11 Apr. 13 Apr. 15 Apr. 18 Apr. 21 Apr. 26 Apr. 28 Apr. 29 May 2 May 3 May 4 May 5 May 6" May 6 May 7 May 9 May 10 May 11 May 12 May 13 May 16 Time. 4h00r 3 49 8 54 8 35 3 51 8 36 8 50 8 56 9 00 111 lo 9 00 'p.m. p.m. a.m. a.m. p.m. a.m. a.m. a.m. a.m. a.m. a.m. 9 54 a.m. 10 05 a.m. 3 49 p.m. 2 32 3 06 2 44 3 40 p.m. p.m. p.m. p.m. 11 15 a.m. 8 50 a.m. 2 40 p.m. 2 38 8 27 3 50 8 10 9 12 8 36 2 01 9 12 10 22 7 59 8 19 11 48 8 26 9 24 8 42 p.m. a.m. p.m. a.m. a.m. a.m. p.m. a.m. a.m. a.m. a.m. a.m. a.m. a.m. a.m. 9 42 a.m. 2 31 p.m. 3 35 p.m. 10 21 a.m. 8 37 a.m. 8 45 8 32 9 32 9 07 10 09 8 33 9 31 9 29 12 20 a.m. a.m. a.m. a.m. a.m. a.m. a.m. a.m. p.m. 2 39 p.m. 9 50 a.m. 8 42 a.m. 8 15 a.m. 9 16 a.m. 10 17 a.m. 8 28 a.m. 3 29 p.m. 10 53 2 35 2 36 a.m. p.m. p.m. Temper- ature. °C. 2 3.0 5.0 5.0 5.0 3.0 3.0 2.0 3.0 3.0 0.0 3 0.0 1.0 0.0 -10 4 - 4 0.0 0.0 10.0 13.0 16.4 16.0 19.0 18.7 10.2 '8.3 10.2 20.2 10.9 16.6 15.3 '9.6 12.0 10.0 12.4 '9.8 "9.2 lh'.i 20.5 15.8 17.8 16.8 14.9 13.3 12.9 16.5 Barom- eter. mm. 774.00 768.70 760.00 759.00 765.30 764.00 760.20 74S.40 752.85 743.85 762.45 762.05 765.25 753.15 752.65 765.75 765.40 778.00 747.00 766.70 768.75 707.55 765.75 762.85 75L00 750.10 754.50 76L45 765.00 759.00 766.50 76o!6o 763.90 762:66 765.90 768.40 764.90 764i90 766125 766!6o 760.00 757.25 755.85 755.00 756.50 761.90 761.65 772.40 Wind. Weather. NW E SE NW S.W Light NE. W SE sw NE NW NW. NW. NW. Strong W. . . Strong NW. SW SE NW. ... NE. . . . Light W. NW. ... SW. . . . No wind SW. w. . NW. Pleasant Sunny Rain Sunnv Cloudy , ...Do Pleasant Cloudy Sunny, bright. Rain Cloudy, light snowstorm night before. Pleasant . ..Do Snow in morning; sunny and pleasant at 3h49m p.m. Pleasant . . . Not sunny . Cold, raw day, no sun. Snowing and raining, raw day. Cold, raw day, misty. Pleasant, sunny and bright. . ... Do ...Do Cloudy . ... Do Sunny Slightly overcast, very warm. Very cloudy .... Cloudy NW. ... SW. . . . w No wind SE Light SE . . . Strong NW' Light NW . Very pleasant . Pleasant, sunny Sunny, bright . Rained a little . Pleasant but not sunny. Rainy No wind . Light SW NW Light SW Pleasant, cool, and windy. Very pleasant; sunny. Cloudv ...Do Cloudy Cloudy' SW. w SE Brisk NW. SE. SE. SE SE No wind Very pleasant; sunny. Pleasant Rainy . . Pleasant Cloudy Pie asant sunny. , ...Do ...Do Very pleasant and Carbon dioxide. Oxygen. p. ct. p. ct. 0.029 20.922 .031 20.923 .033 20.921 .030 20.922 .031 20.919 .030 20.919 .029 20.923 .031 20.918 .030 20.919 .034 20.920 .033 20.921 .029 20.923 .028 20.918 .029 20.923 .030 20.921 .032 20.919 .032 20.920 .029 20.919 .032 20.920 .031 20.913 .031 20.912 .028 20.919 .027 20.911 .027 20.910 .028 20.931 .030 20.931 .030 20.932 .033 20.932 .030 20.933 .030 20.940 .029 20.941 .029 20.932 .030 20.932 .029 20.933 .032 20.939 .030 20.930 .030 20.940 .031 20.942 .030 20.943 .028 20.921 .028 20.933 .027 20.921 .029 20.921 .029 20.940 .029 20.931 .029 20.932 .031 20.931 .030 20.933 .030 20.930 .027 20.960 .029 20.931 .030 20.922 .033 20.923 .033 20.920 .031 20.933 .031 20.922 .030 20.921 .028 20.922 .029 20.920 .033 20.933 .030 20.930 .029 20.922 1 Apparatus leaked. All stopcocks were removed and greased. Comparative Air-Analyses 91 Table 53. — Analyses of outdoor air made at the Nutrition Laboratory. Series 2 — Cont'd. [All samples were taken from the west side of the laboratory.] Date. Time. Temper- ature. Barom- eter. Wind. Weather. Carbon dioxide. Oxygen. 1910. May 17 May 18 May 20 May 21 May 23 May 24 May 26 May 28 May 31 June 1 June 2 June 4 June 6 June 7 June 8 June 9 June 10 June 11 June 13 June 14 June 16 June 17 June 18 June 20 June 21 June 22 June 23 June 24 June 25 June 27 June 28 June 29 June 30 Sh50ma.m. 9 49 a.m. 11 01 a.m. 11 5S a.m. 2 14 p.m. 2 57 p.m. 3 57 p.m. 3 09 p.m. 3 42 p.m. 4 42 p.m. 10 30 a.m. 11 42 a.m. 3 16 p.m. 2 14 p.m. 3 24 p.m. 1 12 p.m. 2 14 p.m. 1 19 p.m. 3 08 p.m. 1 49 p.m. 2 52 p.m. 9 09 a.m. 10 00 a.m. 2 24 p.m. 3 27 p.m. 3 17 p.m. 4 18 p.m. 3 05 p.m. 4 10 p.m. 2 40 p.m. 3 31 p.m. 4 34 p.m. 5 27 p.m. 2 10 p.m. 3 16 p.m. 9 11 a.m. 10 14 a.m. 8 35 a.m. 9 37 a.m. 2 28 p.m. 3 37 p.m. 8 46 a.m. 9 48 a.m. 8 35 a.m. 9 35 a.m. 10 37 a.m. 9 05 a.m. 1 11 p.m. 2 22 p.m. 3 25 p.m. 8 23 a.m. 10 25 a.m. 11 30 a.m. 8 33 a.m. 9 22 a.m. 9 03 a.m. 10 03 a.m. 9 10 a.m. 10 11 a.m. 11 15 a.m. 9 16 a.m. 10 10 a.m. 11 10 a.m. 2 02 p.m. 2 54 p.m. 10 06 a.m. 11 05 a.m. 8 55 a.m. 10 07 a.m. 12 31 p.m. 1 23 p.m. 8 38 a.m. 12 43 p.m. "C. 22.8 15.i 23.0 13.7 20.0 28!6 18!6 18.3 17.0 19.i i.4.7 16.4 14.0 20I1 23.6 12!3 ii!2 24^2 27.1 19!7 i7i8 23^9 26\9 30.9 36!l 2f)!2 24 !s 26!5 25.6 2L9 24!3 28!6 mm. 770.00 758.55 762.80 761.30 763.00 750.70 762.40 747.40 748.20 751.80 764.65 757.95 701.70 705.10 765.25 706.45 762.50 763.15 765.45 761.45 752.90 759.50 758.95 759.50 755.90 761.00 767.35 759.65 753.75 756.75 758.85 Light SW. Pleasant, warm and sunny. p.ct. 0.034 .033 .034 .031 .029 .028 .029 .031 .032 .033 .032 .030 .030 .031 .031 .031 .032 .034 .032 .029 .031 .032 .030 .030 .028 .033 .031 .031 .030 .028 .028 .028 .028 .029 .030 .028 .031 .031 .030 .030 .031 .030 .028 .033 .032 .032 .034 .030 .032 .033 .030 .029 .028 .030 .028 .030 .029 .030 .028 .030 .029 .030 .031 .031 .033 .030 .030 .033 .031 .032 .031 .031 .031 p. ct. 20.922 20.925 20.923 20.921 20.923 20.911 20.911 20.921 20.912 20.913 20.923 20.902 20.920 20.715 20.743 20.911 20.910 20.941 20.942 20.940 20.940 20.920 20.922 20.930 20.931 20.930 20.931 20.930 20.933 20.931 20.921 20.932 20.930 20.931 20.933 20.931 20.930 20.933 20.931 20.941 20.941 20.921 20.921 20.930 20.921 20.923 20.920 20.921 20.933 20.932 20.951 20.920 20.921 20.921 20.918 20.930 20.931 20.929 20.910 20.931 20.909 20.933 20.930 20.921 20.923 20.939 20.939 20.933 20.931 20.921 20.923 20.923 20.923 SE. . Rain Pleasant, sunny Rainy, misty . . . Cloudy Light SE No wind Light SE Pleasant No wind SE Cloudy part of time; sunny. Strong NW .... SW Cloudy SE Very pleasant . . NW. . Rainy day, but not raining when analyses were made. W SW Pleasant, sunny SW. . Pleasant, sunny Strong NE NE. . SW Changeable, cloudy and sunny. Pleasant, sunny No wind No wind NE No wind Cloudy, showers SW Sunny, pleasant . .Do SW. . w. .. Pleasant; not sunny. Cloudy No wind Light, E Sunny, pleasant Light, SE Pleasant SW Pleasant, but not sunny. Cloudy NW. . NW. . . . Pleasant and sunny. W 92 Table 53. Composition of the Atmosphere -Analyses of outdoor air made at the Nutrition Laboratory. Series 2 — Cont'd [All samples were taken from the west side of the laboratory.] Date. Time. Tl mper- ture. Barom- eter. Wind. Weather. Carbon dioxide. Oxygen. p. ct. 20.923 20.933 20.934 20.930 20.951 2b!921 20.942 20.941 20.960 20.900 20.932 20.930 20.928 20.930 20!932 20.931 20^953 20.941 20.941 20.923 20.924 20.920 20.941 20.939 20.950 20.949 20.953 20.943 . . . . 20.940 20.942 20^932 20.932 20^94 1 20.951 20.940 20.943 20.921 20.940 20.939 20.941 20.943 20.932 20.942 20.943 20.922 20.940 20.890 20.943 20.942 20.900 20.941 20.942 20.862 20.941 20.943 1910. July 1 Oct. 17 Oct. 18 Oct. 19 Oct. 20 Oct. 21 Oct. 24 Oct. 25 Oct. 26 Oct. 27 Oct. 28 Dec. 2 Dec. 6 Dec. 7 Dec. 8 Dec. 9 Dec. 10 Dec. 12 Dec. 13 Dec. 14 lh35mp.m. 2 34 p.m. 3 22 p.m. 3 09 p.m. 4 08 p.m. °C. 27.8 18^7 18.1 7.7 9.9 9.6 11.8 14.1 8.6 9.1 9.7 10.9 9.7 9.5 8.5 9.2 10.0 -7 -2 -0.7 -7i3 'b'.o -1.6 0.2 mm. 757.30 763.00 758.70 765.85 759.50 757.50 750.30 752.15 748.50 755.50 757.15 760.00 757.40 761.75 765.80 766.95 765.45 770.00 760.50 Light, SE Cloudy p. ct. 0.029 .031 .030 .030 .030 .029 .032 .041 .029 .027 .027 .032 .029 .029 .030 .029 .028 .031 .032 .028 .030 .030 .032 .031 .030 .029 .030 .030 .027 .027 .032 .030 .032 .027 .029 .031 .030 .029 .030 .028 .030 .028 .032 .031 .032 .031 .029 .031 .030 .033 .030 .028 .028 .027 .027 .030 .028 .028 .031 .031 .031 .028 .029 .029 .028 .029 .029 .030 .030 .030 .032 .030 .030 .030 .030 .032 .031 8 53 a.m. 9 40 a.m. 11 25 a.m. 2 36 p.m. 11 53 a.m. 3 05 p.m. 4 16 p.m. 9 00 a.m. 9 45 a.m. 2 37 p.m. 9 18 a.m. 10 25 a.m. 11 33 a.m. 2 34 p.m. 3 36 p.m. 8 49 a.m. 9 45 a.m. 10 44 a.m. 8 56 a.m. 10 09 a.m. 11 37 a.m. 2 35 p.m. 2 31 p.m. 3 23 p.m. 9 07 a.m. 9 57 a.m. 3 08 p.m. 11 27 a.m. - 10 24 a.m. 11 30 a.m. . 12 14 p.m. SW No wind NE Warm and pleas- ant. Rain. Pleasant NW Cool and pleas- ant. Light, SW . Strong NW. . NW. . SW. . . . SW. . NW. NW Light snow- storm, first of the winter. Bright, sunny . . Light, NW 2 10 p.m. 3 00 p.m. 9 07 a.m. 10 03 a.m. 11 37 a.m. 12 23 p.m. 2 15 p.m. 3 05 p.m. 4 13 p.m. 10 13 a.m. 9 25 a.m. 10 13 a.m. 9 41 a.m. 10 30 a.m. 11 18 a.m. Light, W W Bright, sunny . . Light, W Pleasant Bright, sunny . . Bright, sunny . . Light, NW SW Comparative Air-Analyses 93 Table 53. — Analyses of outdoor air made at the Nutrition Laboratory. Series 2 — Cont'd. [All samples were taken from the west side of the laboratory.] Date. 1910. Dec. 15 Dec. 19 Dec. 20 Dec. 22 Dec. 23 1911. Jan. 18 Jan. 20 Jan. 21 Feb. 10 Time. 9h43ma-m. 10 42 a.m. 11 04 a.m. 11 53 a.m. 2 20 p.m. 9 41 a.m. 10 37 a.m. 11 32 a.m. 2 04 10 38 11 38 a.m. 9 42 a.m. p.m. a.m. 3 10 p.m. 4 14 p.m. 9 31 a.m. 10 44 a.m. 11 52 a.m. 3 29 p.m. 9 17 a.m. 0.0 6.4 0.8 Temper- ature. Barom- eter. °C. 2.4 mm. 74S.65 'o\6 747.00 "i.i 770.20 "L9 771.60 'b!6 767.25 766.35 754.60 75S.00 Wind. Light, SE. , Brisk SW... Light, NW. Weather. Light, SW Not bright and sunny. Light, W Rained in a.m., but not raining when analyses were made. NW Bright and sunny Light, SW Light, W Pleasant and sunny. Pleasant Cloudy . Pleasant Carbon dioxide. p. ct. 0.028 .030 .030 .032 .031 .032 .030 .031 .032 .032 .032 .032 .031 .031 .033 .030 .029 .029 .030 .029 .027 .030 .028 .028 Oxygen p. ct. 20.938 20.941 20.921 20.933 20.962 20.831 26!942 20.831 20.932 20.931 20.923 20.952 20.960 20.960 20.950 20.961 20.961 20.951 20.969 20.960 20.961 20.964 20.961 As an examination of the results showed frequent marked altera- tions in oxygen content, numerous experiments were made with a view to changing experimental conditions, an accurate Haldane apparatus being often used. iThe variations persisted, however, and while the year's work from November 3, 1909, to February 15, 1911, fully substantiated the observations of earlier experimenters, by no stretch of the imagination could a relationship be established between the oxygen percentages and meteorological conditions, nor could any adequate explanation be found as to their cause or causes. The significant fact that there was no cor- responding alteration in the carbon-dioxide content — this factor remain- ing constant under all conditions of wind direction — led us to the belief that the oxygen percentage also approximated constancy, and that the discrepancies appearing in the results might be attributed to errors in either the technique or the apparatus. CONTROL ANALYSES. To determine the source of error it was necessary to make duplicate analyses on another apparatus exactly similar in shape; consequently Rudolph Grave, of Stockholm, was commissioned to construct a second apparatus. In the fall of 1910 this apparatus reached Boston, but un- fortunately had been utterly demolished in transit. Since it was then too late in the season to secure a third apparatus for the winter's work, the lack of control analyses was a very serious drawback. Finally, a cylin- der of compressed air was secured from the compressed-air plant of the Laboratory of Mechanical Engineering at the Massachusetts Institute of 94 Composition of the Atmosphere Technology for the purpose of making control analyses of the air in the cylinder. The steel cylinder employed, which had formerly been used for compressed oxygen, was repeatedly exhausted by a vacuum-pump and outdoor air admitted. It was assumed that the inner walls of the cylin- der would not absorb oxygen from the air rapidly. It was furthermore assumed that air stored in the large compressed-air chamber of the In- stitute of Technology would be thoroughly mixed and have a fairly con- stant composition. Employing precisely the same technical routine, samples of the cylinder air were frequently analyzed as a control on the analyses of the outdoor air. The results of these analyses made between December 3, 1910, and February 9, 1911, are given in table 54. Table 54 — Analyses made at the Nutrition Laboratory of air confined in a steel cylinder. Series 1. Date. Time. Carbon dioxide. Oxygen. Date. Time. Carbon dioxide. Oxygen. 1910. p. ct. p. ct. 1910. p. ct. p. ct. Dec. 3 9h34ma.m. 0.033 20.950 Dec. 28 9h38ma.m. 0.032 20.920 Dec. 6 2 35 p.m. .033 20.942 10 44 a.m. .032 20.911 3 37 p.m. .032 20.943 11 43 a.m. .033 20.951 Dec. 7 4 48 p.m. .034 20.941 3 37 p.m. .032 20.949 Dec. 8 4 03 p.m. .034 20.943 1911. Dec. 9 2 56 p.m. .032 20.940 Jan. 21 2 32 p.m. .031 20.951 Dec. 10 10 14 a.m. .031 20.937 3 43 p.m. .033 20.950 Dec. 13 12 02 p.m. .034 20.938 Jan. 23 9 37 a.m. .032 20.951 Dec. 15 11 52 a.m. .031 20.922 10 52 a.m. .033 20.952 Dec. 22 2 31 p.m. .033 20.922 Jan. 31 10 03 a.m. .032 20.962 3 43 p.m. .033 20.930 11 30 a.m. .033 20.975 4 52 p.m. .034 20.933 Feb. 9 .034 20.913 Dec. 23 10 50 a.m. 11 58 a.m. 3 24 p.m. .034 .032 .032 20.912 20.920 20.932 As the simultaneous analyses of outdoor air and cylinder air pro- gressed, it soon became apparent that there was some intimate relation between the fluctuations in oxygen content of the outdoor air and the fluctuations observed in the oxygen content of the cylinder air. Further- more, while a steady slight decrease in the oxygen percentage of cylinder air might have been expected, as a matter of fact the fluctuations were such as to indicate at times apparent increases. This was conclusive evidence that in spite of all precautions and delicacy of manipulation, the observed fluctuations in the oxygen content of outdoor air might well be due to errors in technique. THIRD ROUTINE, AND RESULTS OBTAINED. The fluctuations in the oxygen content of the cylinder air led to the belief than an error was introduced by the distillation of water from the measuring pipette over into the strong alkali. A series of test experiments, which occupied several weeks, almost to the exclusion of regular air- analyses, finally resulted in an alteration in the routine on February 15, Comparative Air-Analyses 95 1911. A further change was made at this time which was due to the fact that at the end of an analysis the potassium pyrogallate was in contact with pure nitrogen in the capillary tube inside the chamber, while the reagent in the outer part of the reagent chamber was in contact with air. This was remedied by the attachment of a double bulb to the tube through which the chamber is filled. This double bulb provided a seal, one bulb containing water and the other air. When a sample of air was forced into the pyrogallate vessel, the reagent, rising on the outside of the inner tube, forced the gas above it into the first bulb, thereby expelling the water which it contained into the second bulb. The modified routine adopted on February 15, 1911, was as follows: After absorbing the carbon dioxide, the air was sent back and forth into the potassium pyrogallate five times, being left there each time 10 seconds. The first reading was then taken. Following this the air was again sent into the potassium pyrogallate for 10 seconds and a second reading taken. This procedure was carried on until the readings were essentially constant. About this time a change was also made in the carbon-dioxide routine. As it appeared unnecessary to make two readings of this factor, the first reading was omitted, although the procedure for transfer of air to and from the potassium-hydroxide pipette was not altered in any way. With this routine, therefore, the air was passed from the measuring pipette into the potassium pyrogallate a maximum of eight times. A sample analysis made with the third routine is given in table 55. Table 55. — Results obtained on a sample of outdoor air with third routine, February 20, 1911, J^ 25m p. m. Reading. Carbon dioxide. Oxygen. First Second Third Fourth p. ct. 0.028 p. ct. 20.930 20.941 20.949 20.952 This routine was followed for but two days in the regular air-analyses; the few results are given in table 56. During the period from February 16 to March 10 the cylinder air was likewise more or less continually analyzed, the third routine being used. (See table 57.) Table 56. — Analyses of outdoor air made at the Nutrition Laboratory. Series 3. Date. Time. ' Tempera- ture. Barom- eter. Wind. Weather. Carbon dioxide. Oxygen. 1911. Feb. 20 Feb. 21 p.m. 4h00m 4 25 2 15 2 40 °C. —5 —2.8 mm. 755.00 761.75 NW. Snowing all day . . p. ct. 0.028 .031 .028 .028 .030 p. ct. 20.951 20.950 20.952 20.940 20.941 W.. Sunny, pleasant . . 96 Composition of the Atmosphere Table 57. — Analyses made at the Nutrition Laboratory of air confined in a steel cylinder. Series 2. Temper-) Temper-] 1 Date. Time. ature of Carbon water- dioxide. Oxygen. Date. Time. ature of water- Carbon dioxide. Oxygen. bath. bath. 1911. °C. p.ct. p. ct. 1911. °C. p. ct. p. ct. Feb. 16 8h30ma.m. 19.4 0.034 20.951 Feb. 27 10h30ma.m. 17.1 0.033 20.929 19.6 .034 20.950 11 00 a.m. 17.0 .032 20.933 11 15 a.m. 19.6 .035 20.941 12 00 noon 26.0 .032 20.943 2 15 p.m. 19.4 .036 20.951 3 30 p.m. 24.S .030 20.932 19.7 .035 20.953 25.2 .029 20.939 Feb. 18 2 15 p.m. 20.4 .033 20.952 25.7 .030 20.931 20.6 .033 20.949 Feb. 28 8 45 a.m. 16.7 .033 20.910 Feb. 20 19.6 20.1 .035 .033 20.938 20.943 9 25 a.m. 9 45 a.m. 16.8 16.9 .034 .033 20.930 20.930 5 00 p.m. Feb. 21 17.5 17.3 .033 .033 20.931 20.932 10 15 a.m. 11 05 a.m. 16.8 26.0 .035 .032 20.931 20.922 17.9 .034 20.935 11 25 a.m. 26.3 .028 20.942 25.9 .032 20.942 11 50 a.m. 25.9 .030 20.941 11 45 a.m. 25.3 .033 20.941 Mar. 1 2 35 p.m. 16.8 .033 20.921 25.3 .032 20.950 3 10 p.m. 16.8 .034 20.921 25.4 .031 20.950 3 35 p.m. 16.5 .034 20.920 3 45 p.m. 17.2 .033 20.900 4 00 p.m. 17.0 .033 20.920 4 20 p.m. 17.3 .033 20.933 17.2 .033 20.921 5 00 p.m. 17.3 .035 20.931 Mar. 2 4 00 p.m. 16.4 .033 20.882 Feb. 22 8 25 a.m. 17.4 .034 20.927 16.5 .035 20.911 9 30 a.m. 26.1 .035 20.920 16.7 .031 20.901 9 55 a.m. 26.1 .032 20.939 Mar. 3 8 30 a.m. 18.1 .034 20.932 10 50 a.m. 25.3 .035 20.930 9 00 a.m. 18.1 .033 20.923 11 25 a.m. 25.5 .034 20.941 9 30 a.m. 18.1 .033 20.929 Feb. 23 8 30 a.m. 17.7 .034 20.929 10 00 a.m. 18.1 .033 .... 9 00 a.m. 16.8 .035 20.931 10 30 a.m. 18.0 .034 20.930 9 35 a.m. 17.2 .033 20.930 11 00 a.m. 17.8 .034 20.931 10 30 a.m. 26.2 .032 20.922 Mar. 4 2 20 p.m. 19.6 .035 20.929 10 55 a.m. 26.0 .029 20.949 2 45 p.m. 19.7 .034 20.930 11 30 a.m. 25.9 .031 20.941 3 15 p.m. 19.S .034 20.921 Feb. 24 12 00 noon 8 25 a.m. 25.8 16.6 .030 .033 20.939 20.920 Mar. 6 1S.5 18.7 .034 .031 20.920 20.940 8 50 a.m. 16.3 .034 20.932 18.6 .034 20.919 9 20 a.m. 16.1 .034 20.921 Mar. 7 8 40 a.m. 18.5 .032 20.910 10 20 a.m. 26.4 .033 20.938 18.5 .034 20.920 10 50 a.m. 26.1 .032 20.930 9 25 a.m. 18.6 .034 20.912 11 15 a.m. 26.0 .031 20.930 3 15 p.m. 1S.S .033 20.912 Feb. 25 2 30 p.m. 17.2 .033 20.921 18.4 .033 20.910 3 00 p.m. 17.3 .033 20.920 18.5 .034 Feb. 27 16.4 16.6 .032 .033 20.930 20.941 Mar. 10 8 40 a.m. 18.6 17.'.i .031 .033 20.930 20.921 9 50 a.m. EFFECT ON OXYGEN ABSORPTION OF HIGH AND LOW TEMPERATURES. An examination of the results in table 57 shows an approach to con- stancy, but the variations are still too wide to be permitted in this study. During this period it is seen that the temperature of the water-bath had a wide variation, ranging from 16.1° C. to 32.2° C. Owing to the fact that a number of investigators with wrhom I had conferred in a recent European tour had suggested that the absorption of oxygen might be profoundly affected by changes in temperature, the wide range was artificially pro- duced for the special purpose of studying this particular point. From the data in table 57 the volumes for the various temperatures have been rearranged so as to give the average results at approximately an average temperature. These results have been incorporated in table 58. An ap- parently constantly decreasing percentage of carbon dioxide is not con- sidered here, since subsequent experiments show that the decrease is not an inevitable accompaniment of increased temperature. The oxygen de- terminations show a difference of 0.01 per cent between the value at 17.1° C. and that at 25.8° C, but an insignificant difference between the Comparative Air-Analyses 97 value at 25.8° C. and that 6 degrees higher. The low value at 17.1° C. may in part be attributed to the well-known slow absorbing action of potassium pyrogallate at low temperatures. From these results we may safely conclude that temperature changes in the reagent are without appreciable effect upon the absorption of oxygen within the limits men- tioned, i.e., between 17.1° C. and 31.8° C. Table 58. — Analyses made at the Nutrition Laboratory of air confined in a steel cylinder, using high and low temperatures of water-bath. Tempera- ture of water-bath. Carbon dioxide. Oxygen. Temper- ature of water- bath. Carbon dioxide. Oxygen. Temper- ature of water- bath. Carbon dioxide. Oxygen. °C. p. ct. p.ct. 'C. p. ct. p. ct. °C. p. ct. p. ct. 17.8 0.033 20.931 25.9 0.032 20.942 32.2 0.032 20.940 17.3 .033 20.932 25.3 .033 20.941 31.3 .030 20.932 17.9 .034 20.935 25.3 .032 20.950 31.4 .029 20.933 17.2 .033 20.900 25.4 .031 20.950 31.8 .030 20.943 17.3 .033 20.933 26.1 .035 20.920 31.9 .031 20.941 17.2 .035 20.931 26.1 .032 20.939 31.9 .028 20.949 17.4 .034 20.927 25.3 .035 20.930 31.9 .030 20.949 17.7 .034 20.929 25.5 .034 20.941 31.7 .031 20.941 16.8 .035 20.931 26.1 .032 20.922 17.2 .033 20.930 26.0 .029 20.949 16.6 .033 20.920 25.9 .031 20.941 16.3 .034 20.932 25.8 .030 20.939 16.1 .034 20.921 26.4 26.1 .033 .032 20.938 20.930 Avg. 17.1 .033 26.0 25.8 .031 .032 20.930 31.8 .030 20.927 20.937 20.941 Errors in the third routine. — Two fundamental alterations in routine were introduced at this time. Miss Johnson called my attention to the fact that after absorbing the oxygen, to connect the pipette B with the manometer and then subsequently to send the air into the potassium pyrogallate was illogical for the following reason: When the stop-cock a is turned so as to communicate the gas in B with the manometer, there is in B only nitrogen. On the other hand, in the capillary between the stop-cocks a and d there is air. During the short time required to set the manometer and take the reading, there is unquestionably a slight diffusion of air from the capillary into the chamber B. Subsequently, when this gas is passed into the potassium pyrogallate, there is a further contraction in volume which results in a considerable increase in the apparent percentage of oxygen. Probably the increased percentages found when the gas was repeatedly passed into the reagents were due to this fact rather than to the distillation of water. It was ap- parent, therefore, that before connection with the manometer is made, the gas must be sufficiently in contact with the potassium pyrogallate to ab- sorb all of the oxygen, and that no subsequent passage of the gas in A into the reagent should be made. Accordingly the routine was so changed as to secure these conditions. 98 Composition of the Atmosphere Shortly prior to this, it was believed that when the mercury was raised and lowered in pipette B, the amount of water adhering to the walls of the lower bulb and graduated portion would vary, and that these differ- ences might play a very important role in determining the percentage of oxygen. Experiments were accordingly made to find the oxygen per- centage in air more or less dry, but these were unsuccessful. Finally, it was decided that if an excess of water was present in the pipette the same relative amount of water would probably adhere to the glass walls each time. A large number of experiments were made to determine the amount of water which should be added to obtain a clear meniscus for reading and to insure constancy in the amount of water in the pipette. At first minute quantities of water were used, the attempt being made to secure only enough to saturate the gas with moisture. It was assumed that when the mercury was lowered the liquid water would adhere to the inside of the upper bulb, so that only the mercury would enter the constricted portion of the pipette and the lower bulb, and that no liquid water would be pres- ent. Under these conditions, however, it was found very difficult to set the meniscus, and the following routine was finally decided on: FOURTH ROUTINE, AND RESULTS OBTAINED. Outline of fourth routine. — The nitrogen resulting from the previous analysis was stored temporarily in the potassium-pyrogallate pipette, the capillary tube leading to the carbon-dioxide absorption chamber C also being filled with nitrogen. The stop-cock a was next removed and the mercury in the pipette B raised up through the capillary to the stop-cock. Water was then added and the mercury simultaneously lowered until there was a layer of 17 mm. of water above it in the capillary tube. The stop-cock a was again put in place, the nitrogen withdrawn from the pyrogallate reservoir, and the sample taken and analyzed as usual. Results with fourth routine. — The analyses made with this routine were continued from March 28 to April 10, 1911. The results are given in table 59. When these results are compared with those of earlier experi- ments, it will be noted that notwithstanding the differences in tempera- ture, barometer, direction of the wind, and other conditions, the oxygen determinations show a striking uniformity, the variations previously found practically disappearing. This new routine was also used for the analyses of cylinder air, which provided an excellent control of the apparatus and method. The results for the period between March 28 and April 14, 1911, are given in table 60. Here again it is seen that variations in the oxygen percentage are rare, and as used the method may be assumed to give constant results. Comparative Air-Analyses 99 Table 59. — Analyses of outdoor air made at the Nutrition Laboratory. Series 4. Date. Mar. 28 Time. llh30ma.m. 12 00 noon 2 25 p.m. 3 27 p.m. Mar. 29 Mar. 30 Mar. 31 Apr. 1 Apr. 3 Apr. 4 Apr. 5 Apr. 6 A pr 7 Apr. 8 Apr. 10 10 01 10 31 10 59 11 28 11 56 2 45 10 21 10 52 8 28 8 58 9 00 9 28 9 57 10 27 11 05 11 37 12 10 8 26 8 54 2 24 2 53 10 06 10 35 11 06 8 43 9 13 9 06 9 32 10 01 11 21 11 49 12 19 8 45 9 15 9 45 3 32 a.m. a.m. a.m. a.m. a.m. p.m. a.m. a.m. a.m. a.m. a.m. a.m. a.m. a.m. a.m. a.m. p.m. a.m. a.m. p.m. p.m. a.m. a.m. a.m. a.m. a.m. a.m. a.m. a.m. a.m. a.m. p.m. a.m. a.m. a.m. p.m. Temper- ature. °C. 7.6 7.4 9.6 5.7 4.5 2.6 4.2 2.i 3.6 14.0 16.8 8.6 8.6 9.9 Barom- eter. mm. 748.40 753.85 751.90 739.40 746. io 757.45 770.85 777.70 775.40 759.50 757.00 754.40 769.00 772.70 772.40 Wind. Weather. Light, SW N Brisk NW. Brisk W . W. Light, E. Light, SE Light,' sw! Light, W. sw".'.'.'.'. SW., mod- erate. Light, W. Light, W. Pleasant Sunny, pleasant Sunny at times, then dark again. Sunny, pleasant Sunny at times, then dark. Sunny, pleasant . . . Sunny, pleasant . Sunny, pleasant. Damp, raw; no sun Snow and sleet last night. Snow on ground; raining. Pleasant, sunny . . . Sunny, bright. Bright, sunny Sunny, pleasant Sunny, pleasant Carbon dioxide. p. ct. 0.034 .031 .033 .030 .032 .030 .030 .030 .029 .031 .030 .029 .027 .030 .029 .029 .030 .028 .031 .027 .029 .031 ,031 .031 .031 .031 .032 .032 .032 .033 .031 .032 .029 .030 .028 .028 .030 .031 .030 .031 .029 Oxygen. p. ct. 20.938 20.940 20.949 20.942 20.921 20.928 20.938 20.932 20.935 20.939 20.939 20.941 20.938 20.938 20.938 20.938 20.932 20.939 20.925 20.931 20.938 20.940 20.930 20.932 20.931 20.933 20.923 20.929 20.928 20.940 20.939 20.937 20.930 20.931 20.930 20.938 20.938 20.928 20.935 20.932 20.930 Table 60. — Analyses made at the Nutrition Laboratory of air confined in a steel cylinder. Series 3. Date. Time. Carbon dioxide. Oxygen. Date. Time. Carbon dioxide. Oxygen. 1911. p. ct. p. ct. 1911. p. ct. p . ct. Mar. 28 9h06ma.m. 0.038 20.930 Apr. 11 4h04mp.m. 0.034 20.919 9 25 a.m. .034 20.922 4 36 p.m. .033 20.912 10 00 a.m. .033 20.921 Apr. 12 8 20 a.m. .035 20.917 10 30 a.m. .032 20.921 9 06 a.m. .033 20.913 4 22 p.m. .036 20.922 9 40 a.m. .033 20.918 Mar. 29 8 30 a.m. .034 20.919 10 48 a.m. .036 20.913 9 20 a.m. .036 20.920 11 25 a.m. .032 20.915 12 10 p.m. .034 20.919 11 57 a.m. .033 20.915 3 48 p.m. .032 20.918 2 38 p.m. .034 20.920 Mar. 30 8 35 a.m. .031 20.910 3 13 p.m. .034 20.912 9 17 a.m. .033 20.918 3 53 p.m. .031 20.918 9 47 a.m. .032 20.919 Apr. 13 8 34 a.m. .036 20.910 11 10 p.m. .034 20.920 9 05 a.m. .036 20.909 Apr. 5 11 43 a.m. .033 20.920 9 36 a.m. .034 20.912 Apr. 6 9 57 a.m. .033 20.919 10 13 a.m. .036 20.911 Apr. 7 10 29 a.m. .031 20.917 11 06 a.m. .034 20.914 10 57 a.m. .033 20.911 11 52 a.m. .035 20.920 Apr. 10 2 54 p.m. .035 20.912 Apr. 14 8 56 a.m. .034 20.903 Apr. 11 9 12 a.m. .033 20.910 9 27 a.m. .033 20.904 9 46 a.m. .033 20912 10 02 a.m. .035 20.913 3 00 p.m. .035 20.916 10 35 a.m. .033 20.910 3 32 p.m. .034 20.910 11 16 a.m. .034 20.912 100 Composition or the Atmosphere FIFTH ROUTINE, AND RESULTS OBTAINED. A final change in the routine was made on April 15, 191 1 . When using a layer of 17 mm. of water in the capillary tube of pipette B it was found that variations in the percentage of oxygen in cylinder air were occasion- ally experienced which were somewhat greater than it was believed the method should allow. A few preliminary tests indicated that more uni- form conditions of moisture in the pipette could be obtained by another procedure, and the following routine has since been adopted: Outline of fifth routine. — The nitrogen remaining from the earlier analysis having been stored over the potassium hydroxide and potassium pyrogallate, the stop-cock b is closed and the plug of stop-cock a is with- drawn. The pipette B is first completely filled with mercury. Water is then introduced through the open stop-cock a, exactly as in the pre- ceding routine, and the mercury in pipette B simultaneously lowered, thus drawing the water down through the capillary tube into the pipette. This is continued until the mercury is lowered to the zero-point, and sev- eral times the amount of water previously used has been introduced. The whole interior of the pipette is thus thoroughly drenched with water. The mercury is again raised, expelling all of the visible water, so that none remains above the mercury, filter paper inserted in the cock opening being used to remove the water. The stop-cock is then inserted, and the ni- trogen is drawn from the two absorption pipettes, the levels of the potas- sium hydroxide and potassium pyrogallate are set, and the analysis pro- ceeds in the usual way, i.e., the carbon dioxide is absorbed by sending the gas twice into the potassium hydroxide, taking but one final reading; the air is next passed into the potassium pyrogallate for 5 minutes, then with- drawn, and passed into the potassium hydroxide once. It is again drawn back and passed into the potassium pyrogallate for 1 minute, then into the potassium pyrogallate for 5 minutes, and into the potassium hydrox- ide once. The level of the latter reagent is now set. The gas is finally sent into the potassium pyrogallate for 1 minute, the level is set, and the reading taken. All the capillaries now being filled with nitrogen as at the beginning of the analysis, the contraction in volume represents the per- centage of carbon dioxide in the air and the percentage of oxygen in carbon-dioxide-free air. This routine, which has been followed without any variation since April 15, 1911, has proved eminently satisfactory. Results with fifth routine. — The results of the analyses of uncontami- nated outdoor air between April 15, 1911, and January 9, 1912, as given in table 61, are conclusive in showing that no fluctuation of any magnitude occurs in the percentage of oxygen in air. Certain values lower than the average may be almost invariably attributed to the fact that after new potassium pyrogallate is placed in the reagent vessel the first analysis is inclined to show a somewhat low percentage of oxygen. When it is considered that all of the determinations, with the single exception of a short series from September 15 to 25, 1911, are reported, the accidental variations are indeed inconsiderable. Comparative Air-Analyses 101 Table 61. — Analyses of outdoor air made at the Nutrition Laboratory. Series 5. Date. Time. Temper- ature. Barom- eter. Wind. Weather. Carbon dioxide. Oxygen. 1911. Apr. 15 Apr. 17 Apr. 22 Apr. 24 Apr. 25 Apr. 26 Apr. 27 Apr. 28 Apr. 28 Apr. 29 May 1 May 2 May 3 May 5 May 6 May 19 May 20 May 22 May 23 May 24 May 25 May 26 June 1 4h25mp.m. 4 56 p.m. 11 45 a.m. 3 00 p.m. 3 28 p.m. 4 12 p.m. 4 48 p.m. 11 47 a.m. 12 21 p.m. 11 15 a.m. 11 55 a.m. 8 55 a.m. 9 33 a.m. 10 13 a.m. 8 45 a.m. 9 25 a.m. 8 58 a.m. 9 34 a.m. 10 05 a.m. 3 09 p.m. 3 54 p.m. 4 31 p.m. 8 31 a.m. 9 12 a.m. 9 57 a.m. 2 25 p.m. 2 58 p.m. 3 43 p.m. 8 45 a.m. 9 22 a.m. 2 53 p.m. 3 36 p.m. 8 26 a.m. 9 14 a.m. 9 53 a.m. 11 26 a.m. 12 10 p.m. 2 32 p.m. 3 03 p.m. 10 37 a.m. 11 13 a.m. 11 54 a.m. 8 54 a.m. 9 43 a.m. 10 29 a.m. 2 33 p.m. 3 04 p.m. 8 45 a.m. 9 24 a.m. 10 09 a.m. 10 48 a.m. 9 23 a.m. 10 09 a.m. 10 47 a.m. 11 49 a.m. 12 27 p.m. 1 04 p.m. 3 44 p.m. 8 34 a.m. 9 33 a.m. 10 14 a.m. 9 15 a.m. 9 57 a.m. 11 20 a.m. 11 55 a.m. 8 40 a.m. 9 18 a.m. 9 59 a.m. 3 52 p.m. 8 28 a.m. 10 02 a.m. 10 09 a.m. 10 48 a.m. 12 47 p.m. 8 55 a.m. 10 18 a.m. 2 41 p.m. 3 19 p.m. °C. 11.8 10.4 10.0 ' 3.6 12.2 12.4 18.i 22.9 24.5 24.5 25.6 25.6 16.7 22.0 19.7 13.5 ' 9.i 18.6 20.4 20. i 21.4 14. i 19.3 33.2 13.9 12.2 12.9 19.2 17.5 21.6 mm. 757.15 758.50 757.30 766.05 765. i 5 766.80 772.80 774.70 770.40 766.25 763.90 761.15 756.55 752.85 750.90 752.35 760'.6o 768.70 767.10 770. io 762.40 762.65 763.60 766.10 763.25 769.40 769.45 766.40 760.50 764.00 753.20 Light, SW. ( Cloudy, dark, J. a. m.; bright, ( sunny at 4 p.m Sunny, pleasant .... Do p. ct. 1 0.031 r .031 .030 .030 .030 .031 .029 .032 .033 .032 .031 .030 .031 .031 .031 .030 .031 .030 .031 .030 .031 .031 .030 .033 .032 .029 .032 .032 ! .030 r .031 .028 .030 .031 .028 .029 .028 .031 .030 .029 .029 .030 .028 .030 .030 .031 .031 .031 .031 .031 .029 .029 .031 .032 .032 .032 .031 .032 .032 .033 .033 .033 .031 .030 .031 .032 .029 .028 .031 .029 .028 .030 .029 .030 .029 .029 .030 .028 .028 p. ct. 20.941 20.939 20.938 20.940 20.940 20.937 20.931 20.938 20.940 20.943 20.942 20.932 20.940 20.942 20.941 20.939 20.930 20.939 20.937 20.930 20.939 20.938 20.929 20.938 20.932 20.931 20.937 20.939 5 20.938 20.941 20.939 20.941 20.943 20.931 20.933 20.939 20.942 20.939 20.938 20.921 20.929 20.932 20.933 20.940 20.937 20.937 20.938 20.928 20.932 20.943 20.930 20.930 20.938 20.931 20.938 20.940 20.940 20.942 20.929 20.929 20.931 20.929 20.940 20.940 20.943 20.944 20.942 20.942 20.947 20.940 20.938 20.941 20.938 20.940 20.941 20.939 20.948 20.942 Light, SW. Light, SW. Brisk NE. . . Light, SE . Sunny, pleasant .... Light, SE. . Sunny, pleasant .... Light, S. . . Sunny, pleasant .... N Sunny, pleasant .... N Sunny, pleasant .... N Sunny, pleasant .... N Sunny, pleasant .... Light, S. . . Sunny, pleasant .... Light, SW. ("Rain night before. ) clear but not ) sunny at time (_ of sampling. Light, S. . . Brisk W . . . Sunny at times, then dark. Brisk W . . . Brisk W. .. Light, SW. N Sunny, pleasant. . . . Light, SW . SE Cloudy SE E Cloudy, cold, raw . . Brisk SE. . . Light, SE. . Strong NE Cold, raw, not sunny. Brisk NE . . Light E. . . . Cold, not sunny . . . Dark ; fog or smoke . Brisk SE. . . Light SW.. Cloudy, no rain Brisk W . . Sunny, pleasant 102 Composition of the Atmosphere Table 61. — Analyses of outdoor air made at the Nutrition Laboratory. Series 5 — Cont'd. Date. Time. Temper- ature. Barom- eter. Wind. Weather. Carbon dioxide. Oxygen. 1911. June 2 June 3 June 5 June 6 June 7 June 8 June 9 June 10 June 12 June 13 Sept. 26^ Sept. 27 Sept. 28 Sept. 29 Sept. 30 Oct. 2 Oct. 3 Oct. 4 Oct. 5 Oct. 6 Oct. 7 Oct. 11 Oct. 12 3h59mp.m. 8 34 a.m. 9 14 a.m. 9 59 a.m. 8 59 a.m. 9 42 a.m. 11 55 a.m. 2 52 p.m. 3 33 p.m. 10 31 a.m. 11 13 a.m. 12 19 p.m. 3 08 p.m. 3 47 p.m. 12 32 p.m. 2 18 p.m. 2 58 p.m. 8 30 a.m. 9 15 a.m. 9 55 a.m. 11 50 a.m. 3 17 p.m. 3 57 p.m. 10 45 a.m. 11 55 a.m. 2 34 p.m. 3 11 p.m. 3 51 p.m. 4 27 p.m. 9 10 a.m. 9 48 a.m. 10 35 a.m. 2 40 p.m. 3 19 p.m. 8 34 a.m. 9 18 a.m. 10 27 a.m. 12 16 p.m. 2 43 p.m. 3 30 p.m. 4 31 p.m. 8 56 a.m. 9 34 a.m. 8 35 a.m. 9 09 a.m. 9 45 a.m. 4 12 p.m. 9 07 a.m. 9 44 a.m. 10 19 a.m. 8 46 a.m. 9 27 a.m. 10 33 a.m. 11 53 a.m. 8 51 a.m. 9 30 a.m. 9 13 a.m. 9 50 a.m. 8 48 a.m. 9 27 a.m. 8 45 a.m. 10 12 a.m. 10 51 a.m. 3 26 p.m. 12 03 p.m. 12 37 p.m. 8 34 a.m. 2 30 p.m. 9 10 a.m. 10 00 a.m. °C. 24.7 23.4 15.4 12.5 16.4 20.5 25.3 23.7 12.7 15.7 17.0 12.4 i'i.8 ' 8.9 11.3 13.6 ".8.4 13*6 8.0 20.4 mm. 759.40 765.55 767.00 766.20 768.45 767.65 760.50 757.30 755.35 764.00 766.55 764.55 762.90 758.30 770.30 760.00 752.00 758.75 762.65 762.00 753.60 p. ct. 0.029 .029 .028 .030 .029 .028 .028 .027 .030 .028 .029 .030 .030 .029 .028 .029 .029 .027 .029 .029 .028 .029 .029 f .027 I .029 J .029 \ .029 / .031 V. .027 .046 .032 .030 .027 .029 .029 .028 .033 .029 .028 .031 .037 ( .031 f .029 .032 .032 .032 .029 .030 .032 .030 .033 .039 .033 .031 .032 .031 .030 .031 .032 .031 .028 .031 .031 .033 .035 .030 .027 .029 .028 .028 .031 .027 p. ct. 20.937 20.943 20.939 20.940 20.937 20.938 20.939 20.939 20.942 20.942 20.941 20.941 20.949 20.943 20.943 20.949 20.942 20.939 20.943 20.942 20.939 20.942 20.943 20.949 20.928 20.934 20.939 20.938 20.941 20.940 20.929 20.939 20.958 20.952 20.939 20.941 20.948 20.930 20.929 20.936 20.935 20.940 20.893 20.930 20.930 20.942 20.949 20.941 20.947 20.932 20.932 20.94 i 20.938 20.942 20.942 20.943 20.940 20.939 20.940 20.938 20.928 20.931 20.941 20.934 20.941 20.940 20.941 20.925 20.929 Brisk W. . . . Sunny, pleasant. . . . Light SW . . Brisk SE . . . Brisk NE . . Rains Light E. . . . Clear, bright Brisk NE. . . Sunny, pleasant .... Light E Sunny, pleasant .... Light NE . . / Warm, sunny I early ; dark when ( samples were I taken . Light SE . . . Rained a. m.; no thunder. Light thunder ; heavy rainfall. Light NW. . C Not raining; ) rained early; < heavy thunder- j storm night L before . Dull, dark Light SW. . Brisk NW.. Sunny, pleasant .... Brisk NW.. Sunny, pleasant. . . . Mod. NW.. Light W . . . Sunny, pleasant. . . . Brisk SE . . Dark ; rainy day . . . Brisk SW... Brisk NW. . Showers; not rain- ing now. Light SW . . Mod. NW.. Pleasant Rain ; cold ; raw .... Light W . . . Pleasant 1 Active experimenting began again on September 15, 1911, but the results for the first few days were obviously influenced by some as yet undetermined factor and are not here included. These are the only omissions in the entire series. Comparative Air-Analyses 103 Table 61. — Analyses of outdoor air made at the Nutrition Laboratory. Series 5 — Cont'd. Date. 1911. Oct. 12 Oct. 13 Oct. 17 Oct. 18 Oct. 26 Nov. 1 Nov. 13 Nov. 14 j Nov. 15 Nov. Nov. 17 20 Dec. 4 Dec. 9 Dec. 12 Dec. 13 Dec. 14 Dec. 15 1912 Jan. 2 Jan. 3 Jan. 4 Jan. 6 Jan. 9 Time. 3h32>"p.m. 8 48 a.m. 9 28 a,m, 4 32 p.m 9 55 a.m. 10 33 a.m. 11 05 a.m. 3 14 p.m. 3 23 p.m. 9 39 a.m. 10 14 a.m. 10 51 a.m. 9 12 a.m. 10 04 a.m. 10 38 a.m. 8 55 a.m. 9 44 a.m. 2 25 p.m. 2 59 p.m. 3 39 p.m. 4 23 p.m. 12 12 p.m. 2 13 p.m. 2 46 p.m. 3 31 p.m. 9 49 a.m. 10 28 a.m. 11 04 a.m. 11 39 a.m. 2 21 p.m. 12 01 a.m. 3 00 a.m. 10 12 3 57 12 01 3 00 8 47 12 01 3 00 10 29 12 01 3 00 9 13 9 48 9 04 a.m. p.m. a.m. a.m. a.m. a.m. a.m. a.m. a.m. a.m. a.m. a.m. a.m. 3 59 p.m. 2 45 p.m! 4 40 10 15 2 00 3 25 4 00 8 30 9 30 10 10 8 45 9 20 2 45 p.m. a.m. p.m. p.m. p.m. a.m. a.m. a.m. a.m. a.m. p.m. Temper- ature. °C. ' 9.6 11.5 12.5 14.7 7.5 1.4 ".8 5.8 4.6 4.9 4.7 —3.0 —2.5 7.8 16.6 14.6 6.6 ' 3.8 ' 3.6 —3.0 —9.6 ' 0.6 -^6.6 Barom- eter. Wind. mm. 761.70 768.10 762.80 767.00 763.15 764.30 773.20 74 9. is 764.70 761.20 765.15 765.30 773.i5 773.00 770.30 766.55 758.35 771.70 767.25 763.50 758.75 764.00 737.50 744.45 Brisk W . . Strong NE Brisk SE . SW Light NW. Brisk NW! Light W . n".'.'.'.'.'. Light W Brisk W Light W Light NE Light N SW Light SW Light SW Light W . Light SE!! Light N ., Light NE Brisk NW. Brisk W Strong W . . Average Weather. Beautiful clear day Cloudy, fine mist . Rained all day Sunny, pleasant . . . Cold, raw Sunny, pleasant. Sunny, bright ! Bright, sunny; light fall of snow evening before, followed by heavy rain . Pleasant Cold, clear Cold, raw Cold, raw; no sun Cold, raw Quite foggy . . . Not very foggy. Dark ; light misty rain. Fair Fair Sunny, pleasant Cloudy Cloudy Cloudy, unsettled . . Fair Fair Cloudy, unsettled . Light snowfall i n night; rains now. Pleasant Carbon dioxide. Unsettled ; light snow flurries. Sunny; clear, cold. Rain early a.m. . Now colder, clearing Clear and cold . . p. ct. 0.029 .030 .030 .031 .030 .031 .030 .032 .031 .029 .028 .028 .021 .029 .029 .030 .028 .026 .029 .029 .027 .026 .028 .030 .029 .029 .030 .028 .030 .031 .031 .034 .035 .033 .034 .030 .031 .029 .029 .028 .028 .030 .029 .029 .030 .028 .030 .027 .032 .038 .031 .031 .029 .031 .031 .029 .031 .029 .032 .029 .028 .031 .029 .029 .028 { .031 Oxygen. p. ct. 20.937 20.941 20.938 20.939 20.930 20.936 20.939 20.931 20.937 20.930 20.941 20.939 20.938 20.938 20.940 20.935 20.939 20.940 20.942 20.939 20.941 20.941 20.940 20.934 20.939 20.940 20.940 20.939 20.939 20.941 20.939 20.931 20.931 20.932 20.935 20.938 20.940 20.939 20.941 20.939 20.938 20.939 20.940 20.937 20.940 20.930 20.937 20.938 20.905 20.932 20.919 20.930 20.938 20.938 20.931 20.930 20.931 20.938 20.933 20.933 20.939 20.940 20.940 20.941 20.933 20.938 A result so strikingly at variance with all earlier researches must be subjected to most critical examination and control. It is possible, for example, that the apparatus was mechanically so constructed that the manometer is more readily set at zero when the mercury in the pipette B is at 20.94 per cent — a condition difficult to conceive, but nevertheless not absolutely impossible. 104 Composition of the Atmosphere For the best proof of the accuracy and sensitiveness of the apparatus we must again turn to our analyses of cylinder air. These were made in parallel with the analyses of outdoor air and continued from April 15 to June 7, 1911. The results are given in table 62. Table 62. — Analyses made at the Nutrition Laboratory of air confined in a steel cylinder. Series 4. Date. Time. Carbon dioxide. Oxygen. Date. Time. Carbon dioxide. Oxygen. 1911. p.ct. p. ct. 1911. p.ct. p. ct. Apr. 15 9h06ma.m. 0.036 20.909 May 26 llh37ma.m. 0.031 20.919 9 52 a.m. .033 20.919 12 15 p.m. .033 20.911 10 24 a.m. .033 20.911 3 27 p.m. .035 11 09 a.m. .034 20.920 3 58 p.m. .036 20.915 11 46 a.m. .033 20.912 May 27 8 28 a.m. .035 20.899 12 15 p.m. .035 20.911 9 04 a.m. .034 20.908 2 36 p.m. .034 20.918 9 40 a.m. .035 20.912 3 06 p.m. .034 20.920 2 03 p.m. .036 20.909 3 51 p.m. .034 20.919 2 42 p.m. .034 20.911 Apr. 17 8 51 a.m. .036 20.905 3 25 p.m. .034 20.913 9 31 a.m. .034 20.921 4 07 p.m. .033 20.910 10 16 a.m. .036 20.918 May 29 8 30 a.m. .034 20.902 10 57 a.m. .035 20.921 10 39 a.m. .033 20.911 Apr. 19 9 32 a.m. .035 11 30 a.m. .033 20.909 2 27 p.m. .036 20.918 2 02 p.m. .032 20.908 2 52 p.m. .035 20.919 2 40 p.m. .035 20.909 Apr. 22 2 00 p.m. .030 20.905 May 31 9 21 a.m. .048 20.903 3 03 p.m. .032 20.921 12 02 p.m. .033 20.914 3 42 p.m. .034 20.922 2 31 p.m. .033 20.909 Apr. 24 8 55 a.m. .035 20.915 3 15 p.m. .033 20.913 9 40 a.m. .036 20.909 4 29 p.m. .037 20.910 10 27 a.m. .036 20.922 June 1 8 23 a.m. .033 20.906 11 04 a.m. .036 20.919 9 18 a.m. .034 20.907 Apr. 25 9 08 a.m. .037 20.913 9 55 a.m. .033 20.912 9 50 a.m. .036 20.920 10 39 a.m. .035 20.907 10 28 a.m. .036 20.921 11 31 a.m. .032 20.903 May 2 10 37 a.m. .034 20.919 12 16 p.m. .036 20.912 May 19 3 01 p.m. .033 20.918 12 53 p.m. .034 20.912 May 20 11 15 a.m. .037 20.932 June 2 10 43 a.m. .032 20.908 2 20 p.m. .037 20.910 11 35 a.m. .033 20.908 3 11 p.m. .037 20.909 2 22 p.m. .033 20.905 3 52 p.m. .038 20.913 June 3 10 23 a.m. .035 20.910 May 22 12 37 p.m. .037 20.910 2 28 p.m. .031 20.912 3 03 p.m. .034 20.907 3 14 p.m. .031 20.919 3 48 p.m. .010 20.910 3 57 p.m. .032 20.920 May 23 10 54 a.m. .035 20.919 June 5 8 24 a.m. .037 20.914 11 40 a.m. .034 20.910 9 02 a.m. .036 20.920 12 20 p.m. .033 20.919 9 43 a.m. .034 20.911 May 24 9 22 a.m. .034 20.909 10 30 a.m. .034 20.918 10 48 a.m. .033 20.910 11 10 a.m. .033 20.917 11 35 a.m. .035 20.910 4 18 p.m. .033 20.914 12 11 p.m. .033 20.907 June 6 8 30 a.m. .032 20.912 2 48 p.m. .032 20.909 9 10 a.m. .033 20.912 May 25 8 41 a.m. .032 20.909 9 54 a.m. .031 20.913 9 19 a.m. .033 20.908 June 6 2 29 p.m. .033 20.913 11 29 a.m. .033 20.918 June 7 8 46 a.m. .031 20.910 12 13 p.m. 036 20.910 9 47 a.m. .032 20.909 3 00 p.m. .030 20.913 10 33 a.m. .034 20.919 3 46 p.m. .035 20.910 11 13 a.m. .034 20.918 4 20 p.m. .035 20.912 11 53 a.m. .033 20.920 May 26 8 14 a.m. 9 36 a.m. .033 .036 20.910 20.912 3 56 p.m. .034 20.914 Inasmuch as the analyses indicated that the air had remained ap- proximately constant throughout several months, it was believed that the conditions inside the cylinder were not such as to cause any material oxidation. When the air-analyses were resumed in the fall, after a sum- mer of unprecedented heat in Boston, the oxygen percentage was found to have materially decreased. Excluding a few analyses made between September 15 and September 25, we have the results given in table 63. Comparative Air-Analyses 105 Table 63. — Analyses made at the Nutrition Laboratory of air confined in a steel cylinder. Series 5. Date. Time. Carbon dioxide. Oxygen. ! 1 Date. Time. Carbon ' dioxide, i Oxygen. p.ct. 1911. p. ct. p.ct. 1911. p.ct. Sept. 25 3h16mp.m. 0.032 20.872 Nov. 15 llh26ma.m. 0.031 20.870 3 58 p.m. .034 20.879 1 11 59 a.m. .031 20.871 4 39 p.m. .034 20.892 12 34 p.m. .032 20.871 Sept. 26 11 02 a.m. .036 20.S89 Dec. 2 9 27 a.m. .032 20.859 11 44 a.m. .035 20.891 10 07 a.m. .030 20.863 Sept. 27 9 29 a.m. .030 20.881 10 50 a.m. .031 20.878 Sept. 28 11 00 a.m. .035 20.885 11 27 a.m. .032 20.878 Sept. 29 11 15 a.m. .034 20.882 12 03 p.m. .032 20.876 Sept. 30 10 19 a.m. .033 20.883 2 35 p.m. .026 20.875 Oct. 2 10 41 a.m. .033 20.883 3 15 p.m. .030 20.872 Oct. 3 10 10 a.m. .03S 20.888 3 57 p.m. .033 20.873 10 52 a.m. .039 20.8S0 4 53 p.m. .031 20.873 2 25 p.m. .038 20.870 Dec. 4 9 10 a.m. .032 20.880 3 03 p.m. .040 20.873 3 41 p.m. .032 20.878 Oct. 4 2 42 p.m. .032 20.881 4 19 p.m. .032 20.875 Oct. 5 2 22 p.m. .034 20.887 1912 Oct. 6 2 34 p.m. 4 03 p.m. .032 .032 20.881 20.S80 Jan. 6 .032 .032 20.860 20.860 12 10 p.m. Oct. 25 8 54 a.m. .021 20.873 Jan. 9 10 00 a.m. .033 20.872 10 10 a.m. .034 20.873 10 40 a.m. .033 20.862 Oct. 26 8 44 a.m. 9 21 a.m. .031 .035 20.858 20.861 11 15 a.m. .030 20.863 Of especial importance are the percentages for December 2 and 4, 1911, as they establish the absence of any effect upon the determination of car- bon dioxide and oxygen resulting from the temperature of the reagents. The percentages found were constant, independent of an alteration of over 10° C. in the temperature of the water-bath, which ranged during these analyses between 20.4° C. and 32.5° C. These experiments with im- proved technique substantiate fully the observations on p. 97. CONCLUSIONS FROM RESULTS WITH FIFTH ROUTINE. From the results of anatyses given in tables 61 and 62, it will be seen that the apparatus gives constant oxygen percentages for outdoor air; for any particular day or for a period of 3 or 4 days, the apparatus also gives constant results for the oxygen content of the cylinder air, although these are measurably lower than those for outdoor air. These facts dem- onstrate that the constant readings are not due to any peculiarity in the construction of the apparatus. Furthermore, since it is obvious from an examination of the results that there is a steady, though slight, decrease in the oxygen content of the cylinder air, the evidence in favor of both the accuracy and the sensitiveness of this apparatus becomes conclusive. On the basis of this series of observations, then, we may state that the uncontaminated air of Boston is of constant oxygen content irrespective of conditions of weather, humidity, temperature, barometer, wind direc- tion, and season. It may further be stated that the percentage of carbon dioxide does not undergo any material alteration under these conditions. And, finally, from the evidence thus secured, we may assert that in uncontaminated outdoor air the approximate percentage of carbon di- oxide is 0.031 and oxygen 20.938. 106 Composition of the Atmosphere ANALYSES OF AIR COLLECTED ON THE ATLANTIC OCEAN. Since numerous writers have found noticeable differences in air col- lected over the sea as compared with that collected on the land, a number of samples for analysis were secured of air taken over the Atlantic Ocean. I am indebted for these samples to Mr. Harold L. Higgins, of the Nutri- tion Laboratory, who most caref ully collected them, together with supple- mentary climatic data, on a sea trip between Montreal and Liverpool in November 1910. The glass sampler used was similar in form to that described by Regnault,1 and consisted of a cylindrical tube 40 mm. in diameter and 245 mm. long, with a capacity of approximately 200 c. c. To each end of the tube was attached a short piece of glass tubing, which was drawn out in a capillary, so that when the sampler had been filled it could be sealed in the flame of a candle or an alcohol lamp. The samples were always taken on the windward side of the vessel, the sam- pling tubes being filled by aspiration with the mouth for a few minutes. A water-seal was provided by drawing the air first through the sampler, then through a gas-washing bottle containing water and attached by a short piece of rubber tubing, thus preventing all contamination by ex- pired air. After sufficient aspiration the tubes were quickly taken to the stateroom and sealed, labeled, and packed in a specially constructed carrying case. No precautions were taken to dry the air before it entered the sampling tubes. Though the samples were taken in November 1910, the analyses were not made until a year later, i.e., October 6 and 7, 1911. The results are presented in table 64. Aside from the constancy in the oxygen percentages, the most striking feature of the results in this table is the extraordinarily low percentage of carbon dioxide in the samples collected on November 7 and 10. In fact, all of the carbon-dioxide values are lower than normal.2 Table 64. — Analyses of ocean air collected between Montreal and Liverpool. Date of collec- tion. Time. Lat. N. Long.W. Temp, of sea- water. Wind. Weather. Carbon dioxide. Oxygen. 1910. Nov. 7 Nov. 8 Nov. 9 Nov. 10 Nov. 11 Nov. 12 p. m. 12h 30m 12 15 12 10 12 15 12 35 12 25 O * 47 23 45 56 47 22 49 21 50 42 51 10 o / 59 4 51 15 42 58 34 39 25 33 15 56 "F. 44' 48 48 57 55 55 SE.; mod. to fresh . NW.;mod.. NW.; mod.. S. to SW. mod. S. to SW.; mod. only. SW.; mod. to fresh only. Average Misty to foggy; overcast. Overcast; sea almost calm. Clear, sun n y ; moderate sea. Overcast, misty ....Do ....Do p. ct. 0.003 f .021 t .023 .027 .006 .021 .024 p. ct. 20.940 20.930 20.932 20.939 20.940 20.933 20.940 20.936 1 Regnault, Annates de Chimie et de Physique, 1852, ser. 3, 36, p. 385. 2 The injunction of Regnault (Annates de Chimie et de Physique, 1852, ser. 3, 36, p. 392) to collect samples of dry air was unfortunately not followed. As the determination of oxygen percentages was the first consideration in this stud}', no especial thought was given in taking the samples to the determinations of the carbon dioxide. Comparative Air-Analyses 107 A second study of ocean air was made possible through the kindness of Mr. Thorne M. Carpenter, of the Nutrition Laboratory, who carefully collected a large number of samples on a return voyage from Genoa to Boston in June 1911. Profiting by a discussion of the problem with Dr. Krogh in Copenhagen, Mr. Carpenter took special precautions to collect several samples of air dried over phosphorus pentoxide. Usually the samples were taken by aspiration with the mouth, but in collecting the dry samples it was found that the phosphorus pentoxide powder was me- chanically carried to the mouth, with consequent discomfort; a rubber bulb was accordingly used. The samples were collected between June 10 and 25, 1911, but it was inexpedient to analyze them before October 14 to 26, 1911. The results of these analyses are given in table 65. Table 65. — Analyses of ocean air collected between Genoa and Boston. Date of Time. Lat. Long. Barom- Wind. General Condition of Car- bon Oxy- collection. N. eter. conditions. sample. diox- ide. gen. 1911. o / o / mm. p. ct. p. ct. June 10 June 12 llhooma.m. 746.0 754.1 Bay of Genoa .... Naples; cloudless sky; temp, on Moist . . Moist . Moist . . 0.034 .011 .016 20.938 20.931 20.933 Almost head wind ; strong. deck 20-2 1° C. June 15 11 45 a.m. 12 00 noon 38 17 9 04E 1st day out from Naples; moderate Moist . . Moist . . .022 .019 20.933 20.925 head sea. .019 20.931 June 16 11 40 a.m. 12 10 p.m. 37 17 1 27E 753.8 Fair Temp, on deck 21.5° C; steamer going with wind; Moist . . Moist . . Moist . . .010 .018 .012 20.930 20.935 20.932 11 50 a.m. calm sea. .017 20.938 June 17 11 40 a.m. 35 58 5 50W 752.0 Verylittle.if any; fresh breeze pre- viouseven- ing. Clear early morn- ing; clouded over at noon; very calm sea; temp, on deck, 20.7° C. Moist . . .018 20.929 June 18 36 40 13 11W 754.0 Temp, on deck in sun, 20.7° C. Moist . . .022 .017 20.929 20.931 June 19 37 25 20 43W Moist . . .018 .016 20.936 20.928 .012 20.932 June 20 10 00 a.m. 761.0 Blowing from sea toward the shore across ship Both samples taken a t Ponte Delgada, Azores, on steamer, anchored a b o lit one - half mile from shore; temp, on deck, 20.5° C. Moist . . .019 .013 20.931 20.929 June 21 12 15 p.m. 40 0 32 05W 756.5 Consider- Night before clear; Dry.... .031 20.930 able SW. during night, sea Dry.... .032 wind dur- became rough; Dry .031 20.931 i n g night 6 a. m., clouded Moist . . .011 20.929 before. over; 7 a.m., fog; 11 a. m., quite clear. Moist . . .018 20.938 June 22 41 (17 40 11W Changed from W. to Clear afternoon before; windy; Dry.... Moist . . .030 .013 20.932 20.932 SW.; high moist in morning. Moist . . .018 20.927 breeze. Moist . . .019 20.932 June 23 11 40 a.m. 42 18 47 42W 753.0 High SW. Very rough sea Moist . . .018 20.929 e ve ning evening before Moist . . .017 20.930 12 30 p.m. before; and during night; Dry. . . . .031 20.937 continu e d early morning, wind in heavy rain (hail) ; morning. clear about 7 to 8. June 24 42 49 55 19W 751.0 Moist . . .027 .028 20.931 20.931 Moist . . .022 .023 20.932 20.936 Dry.... .032 20.934 June 25 42 42 63 33W 759.0 Moist . . Dry .014 .033 20.936 20.929 Average 20.932 108 Composition of the Atmosphere The wisdom of taking samples dry is seen from these results, since in all dry samples the percentage of carbon dioxide was found to be always normal. The oxygen percentages again show a striking uniformity and constancy, irrespective of geographical location, weather conditions, etc. ANALYSES OF AIR FROM PIKE'S PEAK. The interesting expedition to the top of Pike's Peak made by Haldane, Yandell Henderson, Douglas, and Schneider, in the summer of 1911, was utilized in that these gentlemen kindly consented to collect samples of air for this research. The apparent constancy in composition of the cylinder air during the early half of 1911 led to the belief that air samples stored in steel cylinders would not undergo a material loss of oxygen; and obviously samples collected in this way would give opportunity for in- numerable analyses. Consequently three small steel cylinders, fitted with proper valves, and a strong bicycle pump were sent to Professors Haldane and Henderson. Both these gentlemen questioned seriously the advisability of using this method of sampling, and fortunately insisted upon having the usual glass samplers sent to them, in which they collected additional samples. The analysis of the air from Pike's Peak was not taken up until the fall of 1911. At this time it was found that during the sum- mer the oxygen percentage of the air in the control cylinder had changed from 20.918 to 20.880, this not inconsiderable change possibly resulting from the extreme heat of the summer, which had been abnormal for this section. Since there had been a change in the oxygen content of this air, it was seen that no reliance could be placed upon the constancy in compo- sition of air stored in steel cylinders, so that any results which might be obtained with air collected in this way on Pike's Peak would be vitiated. Furthermore, it could not even be assumed that cylinders filled on the same day and under the same conditions would be equal in oxidation ; hence any variations from the normal oxygen content found at sea-level could not reasonably be ascribed to a persistent regular oxidation in the cylinders. Table 66. — Analyses at Nutrition Laboratory of air from summit of Pike's Peak. [Air collected and stored in steel cylinders.] Data. Collected Aug. 6, 1911, 5 p.m. Wind, moderate N.W. : Weather clear. Carbon dioxide. Oxygen. p. ct. p.ct. 0.034 20.915 ! .032 20.918 .031 20.914 .n;;r, 20.923 .034 20.922 .036 20.928 .038 20.929 Data. Collected Aug. 8, 1911,5 p.m. Wind, moderate N.W. Weather clear. Carbon dioxide. p. ct. 0.031 .032 .032 .035 .033 .032 Oxygen. p. ct. 20.881 20.883 20.880 20.886 20.887 20.889 The samples of air collected in the steel cylinders were analyzed be- tween September 20 and October 5. While the results have but little value, they are given in table 66 as a further demonstration of the inade- quacy of this method of preserving air samples. The very .large differences in the oxygen content of the two cylinders bears out the belief that the Comparative Air-Analyses 109 oxidation may be very irregular. The percentage of carbon dioxide is slightly higher than normal, but whether this is due to oxidative processes in the cylinder, to organic matter from the lubrication and rubber hose of the bicycle pump, or to an actual condition of the air, the results do not show. The samples collected in glass tubes were analyzed October 9 and 10, 1911, the results in table 67 being obtained. Although some samples stored in glass have a strong tendency to decrease in the percentage of carbon dioxide, these analyses show a percentage which is approximately normal. With the exception of the results obtained for August 14, the oxygen percentages are also normal. While apparently there is a slight diminution in the percentage of oxygen, the average, 20.927 per cent, is obviously affected by the results for August 14, and since at least one analysis on each of the three days showed a percentage of 20.930 or over, it seems hardly probable that this apparent slight decrease is significant. Table 67. — Analyses made at the Nutrition Laboratory of air collected on summit of Pike's Peak— 4312 meters. [Air collected and stored in glass samplers.] Date of collection. Time. Barom- eter. Wind. Weather. Carbon dioxide. Oxygen. 1911. Aug. 11 Aug. 11 Aug. 12 Aug. 14 Aug. 14 GhOOmp.m. 6 00 p.m. 9 00 p.m. 9 00 p.m. 10 00 a.m. 8 30 a.m. 8 30 a.m. 5 30 p.m. 5 30 p.m. mm. 459 459 459 459 458 462 462 462 462 Strong W. . . ....Do ....Do ....Do ....Do Gentle N. . . ....Do Very light NE. f 1 hr. after snowstorm with I much lightning. j- Somewhat clearer; 32° F, Partly sunshine, partly driving clouds; beginning to clear and wind fall- ing. ("Beautiful clear day; day -< and night before clear and (^ warm. p. ct. ( 0.027 \ .032 I .030 I .028 .029 I .031 j" .029 .031 .029 p. ct. 20.928 20.936 20.932 20.930 20.930 20.911 20.923 20.932 20.921 20.927 Average ANALYSES OF STREET AIR. While the air in the vicinity of the laboratory would be expected to be somewhat contaminated with carbon dioxide and consequently deficient in oxygen, it is obvious that the contaminating factors are not of sufficient magnitude to affect perceptibly the analytical results. It became a question of interest, however, as to how far one must go into the heart of the city to secure air of less than normal oxygen content. Two samples were therefore collected, in containers fitted with excellent glass stop-cocks, from a crowded business street. The results of the analyses are given in table 68. The percentage of oxygen was slightly, though almost imperceptibly, less than that in normal air, while the carbon dioxide was slightly higher than the average normal. Since by reference to table 61 it can be seen that the samples analyzed at the laboratory on the same date showed 20.939 per cent of oxygen and 0.029 per cent of carbon dioxide, it can be 110 Composition of the Atmosphere safely asserted that this apparatus was sufficiently sensitive to show even the slight contamination produced by the congestion of population in a narrow city street. It is also remarkable that under these conditions the carbon-dioxide increment and oxygen deficiency were not very much greater. Observations such as these tend to demonstrate the extent of the diffusion of gases and the establishment of equilibrium by air-currents. Table 68. — Analyses made at the Nutrition Laboratory of air collected on a crowded business street in Boston. Date. Time. Barometer, Temper- ature. Place. Carbon dioxide. Oxygen. 1911. Nov. 14 Nov. 14 p. m. lb30m 1 40 mm. 769.55 769.55 °C. 2.7 2.7 In Washington st. between Summer and Avon sts . . . . In Washington st. between Summer and Franklin sts . p. ct. 0.031 .032 .032 .034 p. ct. 20.930 20.929 20.929 20.929 ANALYSES OF SUBWAY AIR. Although foreign to the major question here studied, namely, the composition of uncontaminated air, it was also of interest to find out to what extent the air was vitiated in the modern "tube" or "subway" so extensively used for suburban passenger traffic. Two samples taken si- multaneously at the Park Street station in the Boston subway gave the results presented in table 69. Table 69. — Analyses made at the Nutrition Laboratory of air collected at the Park Street station of the Boston subway. Date. Sample. Time. Carbon dioxide. Oxygen. 1911. Oct. 25 Oct. 25 I II 9h30ma.m. 9 30 a.m. p. ct. 0.064 .062 .065 p. ct. 20.903 20.898 20.897 The samples were taken immediatedly after the "rush hours" were ended and when the oxygen content of the air in the subway might be ex- pected to be at a minimum. The fall of approximately 0.03 per cent in oxygen is accompanied by a rise of 0.032 per cent in the carbon dioxide. Here again one is divided between appreciation of the extraordinary sen- sitiveness of the Sonden apparatus and surprise that the circulation of air in the subways can be so good and the diffusive power of air so ex- tended that the increases in carbon dioxide and decreases in oxygen are of such slight amount. Thanks to the kindness of Dr. E. F. DuBois and Dr. Warren T. Cole- man, two specimens of air were collected from the subway in New York City. The results, which are given in table 70, differ in no wise from those found in the Boston subway and show an increase in carbon dioxide and decrease in oxygen infinitely less than one would normally have ex- pected. Comparative Air-Analyses 111 Table 70. — Analyses made at the Nutrition Laboratory of air collected in the subway, in New York City. Date. Time. Place. Carbon dioxide. Oxygen. 1911. Nov. 21 Nov. 21 6h03mp.m. (rush hour) 5 52 p.m. Grand Central subway station .... p. ct. 0.061 .071 p. ct. 20.900 20.897 The results of this series of experiments have a particular value, inas- much as they show clearly that a decrease in oxygen is accompanied by an approximate increase in carbon dioxide. While the measurement of car- bon dioxide has been taken as an index of good or bad ventilation, the fact that the proportion of oxygen is actually lowered by an increase in the carbon dioxide has never before been clearly demonstrated. As a result of this study, however, knowing both the constancy of oxygen in outdoor air and the sources of carbon-dioxide production, and knowing also that with the carbon-dioxide production there must likewise be an oxygen consumption, we can safely state that for every one-hundredth per cent of increase in carbon dioxide there will be approximately one-hundredth per cent decrease in oxygen.1 It will be seen, therefore, that since there are a number of simple and accurate methods for determining carbon dioxide, the time-consuming and complicated determinations of oxygen are entirely unnecessary, as the determination of the percentage of car- bon dioxide in the air establishes the approximate percentage of oxygen. ABSORPTION OF OXYGEN BY POTASSIUM PYROGALLATE. Since the constancy of the oxygen content of the air has been demon- strated in this research, it has been possible to make a more accurate study than heretofore of the conditions affecting the absorption of oxygen by potassium pyrogallate. Many investigators have experimented with potassium pyrogallate as an absorbing agent for oxygen; but inasmuch as there appeared to be fluctuations in the oxygen content of the air, they have sought for the maximum absorption capacity and minimum produc- tion of carbon monoxide without attempting to control the absorption of oxygen. Particularly confusing has been the fact that the potassium hydroxide as found on the market varies greatly in its water content, the amount ranging from 5 to 25 per cent. To investigators on this subject, the experience of Haldane and Hem- pel has been of the most interest. As already stated, the use in this research of Haldane's saturated solution of potassium hydroxide for dis- solving the pyrogallol did not seem wise with so delicate an apparatus as ours. Should there be a complete solidification of the reagent in chamber D, considerable time would be required to get the solution again in con- xThe one inexplicable phenomenon is the abnormally high percentage of carbon dioxide found in the air of Greenland by Krogh. 112 Composition of the Atmosphere dition for use; this would result in delay, with always a possibility that the expansion might fracture the glass vessel. Furthermore, as the slightly more dilute solution ultimately used for the routine analyses was so much stronger than the solution recommended by Hempel, it was thought that the coefficient of absorption rather than the completeness of absorption would be sacrificed by employing a less concentrated solution.1 In the supplementary study made of the comparative value of the various potas- sium pyrogallate solutions, the formulas recommended by both Haldane and Hempel were tested as to their absorptive powers. In all, four dif- ferent solutions were used: (1) Hempel's solution, prepared according to his formula. Five grams of pyrogallol were dissolved in 15 c. c. of water and mixed with 120 grams of potassium hydroxide dissolved in 80 c. c. of water. Stick potas- sium hydroxide not purified by alcohol was used in all the solutions. (2) Hempel's solution, prepared by a modified formula. Inasmuch as stick potassium hydroxide contains from 5 to 25 per cent of water, instead of using 120 grams of stick potassium hydroxide as in the first solution, only sufficient was used to be equivalent to 120 grams of anhydrous potassium hydroxide. To maintain the proper proportion of water, the amount in the stick caustic potash was included in the 80 grams required by Hempel's formula. As the particular lot of stick caustic potash in use at that time was found by alkalimetry to contain 92 per cent of potassium hydroxide, 130 grams of this chemical was mixed with 70 c. c. of water and subsequently the 5 grams of pyrogallic acid added. (3) The potassium pyrogallate solution used throughout this research. This is described on p. 80. (4) Haldane's formula, requiring a saturated solution of potassium hydroxide in water, with a specific gravity of 1.55. The solution is made in the proportion of 1 gram of pyrogallic acid to 10 c. c. of the potassium- hydroxide solution and hence has the greatest density of all the solutions. For purposes of comparison, both outdoor air and cylinder air were analyzed in this study, exactly the same routine being followed in all analyses, and with all four solutions. (See p. 100.) The results are in- corporated in table 71. The results of these two series of analyses point conclusively to marked differences in the results obtained with solutions of varying strength. These differences may be attributed either to an incomplete- ness of absorption or to the formation of carbon monoxide. That in- completeness of absorption by the weaker solutions can account in any measure for the differences here observed seems hardly probable when the analyses for January 26, 1912, are considered. The results obtained on that date show that extending the time during which the air was in contact with the reagent for an additional 12 minutes made barely an 1 For a criticism of the use of potassium pyrogallate as an absorbent of oxygen, see B. Tacke, Archiv fur die gesammte Physiologie, 1886, 38, p. 401. Comparative Air-Analyses 113 appreciable increase in the percentage of oxygen. Furthermore, for fear that simple contact might be inefficient, as a further precaution, in cer- tain analyses on January 31 and February 1, the gas was passed into the potassium pyrogallate 10 times at the end of the regular routine, but with no appreciable increase in oxygen percentage. Table 71. -Results of comparative study of oxygen absorption by potassium pyrogallate solutions of varying concentration. Date. Outdoor air. Cylinder air. Hem pel solution. I. Hempel "solution. II. Regular solution. Haldane solution. Hempel solution. I. Hempel solution. II. Regular solution. Haldane solution. 1912. p. ct. p. ct. p. ct. p. ct. p. ct. p. ct. p. ct. p. ct. Jan. 6 .... .... 20.933 20.939 20.940 .... .... 20.860 20.860 .... Jan. 9 .... .... 20.940 20.941 20.933 .... .... .... 20.872 20.862 20.863 .... Jan. 26 20.848 20.849 20.843 120.857 120.852 120.854 120.852 20.759 '20.773 Jan. 27 20.952 20.958 20.950 20.950 20.952 20.879 20.878 20.875 Jan. 30 20.910 20.917 20.918 20.910 20.911 20.828 20.830 20.832 Jan. 31 20.949 20.958 220.953 220.960 220.975 Feb. 1 > . . . .... .... 20.958 220.961 220.954 .... .... .... .... Average 220.954 20.851 20.913 20.938 20.956 20.766 20.830 20.863 20.877 1 These were given an additional 12 minutes in the potassium-pyrogallate solution. 2 After absorbing the oxygen in the regular way, the air was sent back and forth into the potassium- pyrogallate solution 10 times before the reading was taken. Since the obvious inference is that there must have been a slight for- mation of carbon monoxide with the weaker solutions, one questions immediately if it is certain that even with the strong Haldane solution there may not be traces of carbon monoxide. Haldane's experience in detecting minute amounts of carbon monoxide goes a long way, however, in establishing faith in his assertion that not the slightest trace is found with the use of his concentrated solution. 114 Composition of the Atmosphere CONCLUSIONS. (1) Apparatus for gas-analysis. — The Sonden apparatus here de- scribed fulfills all conditions essential to exact gas-analysis, save that the gas is not measured dry in a dry pipette over mercury. In spite of this one drawback, the technique has been developed so as to insure a con- stancy and sensitiveness found as yet in no other form of gas-analysis apparatus. (2) Reagent for the absorption of oxygen. — Experimentation with all forms of absorbents for oxygen, including several strengths of potassium pyrogallate solution, leads inevitably to the conclusion that the Haldane potassium-pyrogallate solution is the most efficient agent thus far recom- mended. The analyses of those investigators employing phosphorus or sodium hydrosulphite do not lead one to believe that for efficiency they can supersede the Haldane solution. (3) The constancy of the oxygen percentage in outdoor air. — The results of analyses of air taken near the laboratory showed no material fluctua- tion in oxygen percentage during a period extending from April 15, 1911, to January 30, 1912. This constancy was maintained in spite of all possible alteration in weather conditions, changes in barometer, ther- mometer, humidity, and wind direction and strength; furthermore, the experiments were made before, during, and after the vegetative season. The average result of 212 analyses showed 0.031 per cent of carbon dioxide and 20.938 per cent of oxygen. The analyses of air collected over the ocean, at two different times of the year, and on the top of Pike's Peak, gave essentially similar results. The average results of all the analyses made in this research of outdoor air are summarized in table 72. Table 72. — Summary of analyses made at the Nutrition Laboratory of outdoor air. Number of analyses. Carbon dioxide. Oxygen. Oxygen1 (corrected). Air near laboratory 212 7 36 9 p. ct. 0.031 p. ct. 20.938 20.936 20.932 20.927 p. ct. 20.952 20.950 20.946 20.941 Ocean air (Montreal-Liverpool) . . . Pike's Peak 1 A correction of 0.014 has been added to the average results obtained in this research to make them comparable with results secured with the Haldane solution. (See p. 113.) (4) Tha absolute oxygen content of outdoor air. — While this research has dealt mainly with comparative values, certain fundamental difficul- ties in method and technique prevent deductions with regard to the ab- solute oxygen content of outdoor air. The use of the Haldane concen- trated potassium-pyrogallate solution would seem to preclude the pos- sibility of the formation of measurable amounts of carbon monoxide, but we always have to deal with the possible error in the water adhering to the pipette when the change in level of the mercury is made. Since the contraction in volume is assumed to be only that due to absorbed oxygen, and since unquestionably some water is confined between the Comparative Air- Analyses 115 glass and the mercury, the contraction as measured is invariably too large by the volume of the water so held. Unfortunately no quantita- tive measurements of this water are possible with the Sonden apparatus. In this study, however, we have aimed to secure constancy in the amount of water thus trapped, knowing that the absolute amount could not be measured. The analyses of both outdoor and cylinder air with the potas- sium pyrogallate employed in this research, as well as the analyses made with Haldane's strong solution, showed invariably that the correction of + 0.014 should be added to the results given to make them comparable with analyses made with the Haldane solution. The atomic weights of but few of the chemical elements are known to 1 part in 2000, and hence it may now rightly be said that air is a physical mixture with the definiteness of composition of a chemical compound. (5) While the combustion of fuel and the vital processes of men and animals result in a local increase in carbon dioxide and decrease in oxy- gen on the one hand, and vegetable growth results in a decrease in carbon dioxide and increase in oxygen on the other, the extraordinary rapidity with which the local variations in the composition of the air are equalized is accentuated by the observations on street air, which show but the slightest trace of an oxygen deficit. The ratio between the increment in carbon dioxide and the decrease in oxygen leads naturally to the conclusion that carbon-dioxide deter- minations may be taken as excellent indications of the oxygen content and thus the necessity for elaborate and time-consuming oxygen determin- ations disappears. For every 0.01 per cent increase in the atmospheric carbon dioxide, one may safely assume a corresponding decrease in the percentage of oxygen. Nutrition Laboratory of the Carnegie Institution op Washington, Boston, Massachusetts, February 19, 1912. 3 +(c, %i!l'jllK<,?,',,,U«RARY WH Jfljcl"