INVESTIGATIONS OF INFRA-RED SPECTRA

Part III — INFRA-RED TRANSMISSION SPECTRA
Part IV — INFRA-RED REFLECTION SPECTRA

BY

WILLIAM W. COBLENTZ

WASHINGTON, D. C.

Published by the Carnegie Institution of Washington
December, 1906

CARNEGIE INSTITUTION OF WASHINGTON
PUBLICATION No. 65

PRE88 OF

JUDO A DETWEILEFt, INC.
WASHINGTON, 0. C.

CONTENTS.

PART III — INFRA-RED TRANSMISSION SPECTRA.

Page

Chapter 1 7-15

Introduction 7~9

Present method of investigation 9-10

Pleochroism , II

Apparatus and methods 12-13

Accuracy attainable 13

Preparation of the mineral sections 14-15

Scope of present investigation 15

Chapter II. — Transmission spectra of different minerals 17-7°

Group I : Minerals containing water of crystallization 17-4°

Group II : Minerals containing water of constitution 4O-53

Group III : Miscellaneous compounds 53~70

Sulphates 53-57

PART IV. — INFRA-RED REFLECTION SPECTRA.

Chapter 1 73-75

Introduction 73

Apparatus and methods 73-75

Chapter II. — Infra-red reflection spectra of various substances 77-106

Reflection of sulphates 77-79

Reflection of carbonates and silicates 80-90

Reflection of sulphides 91-93

Transparency to x-rays 95

Reflection of metals 95-99

Reflection of solutions 99-105

Summary 105-106

Tables of maxima 107-109

Appendixes 110-127

Appendix I : The temperature of the moon 110-115

Appendix II : Additional data 116-117

Appendix III : The emission spectrum of burning carbon disulpride,

etc i 18-120

Appendix IV : A vacuum radiomicrometer 121-125

Appendix V : Note on blowing fine quartz fibers 126-127

Index to minerals examined . 128

PREFATORY NOTE.

Parts I and II of "Investigations of Infra-red Spectra" were pub-
lished by the Carnegie Institution of Washington, October, 1905.

The present volume contains the detailed results of a continuation of
this work carried on at the National Bureau of Standards during the
past year.

4

PART III

INFRA-RED TRANSMISSION SPECTRA

CHAPTER I.

INTRODUCTION.

In the various determinations of the physical properties of matter,
such as electrical and thermal conductivity, specific inductive capacity,
coefficients of expansion and refraction, etc., the spectroscopic study of
the transparency of substances to radiant energy, and in particular to
heat-waves, has been left in the background, although this data is just
as necessary in meteorology, geology, and the allied sciences as is our
knowledge of any of the other so-called constants of nature.

The following research was undertaken to ascertain the cause of
absorption, and the connection between absorption and the structure of
crystals. In this and in a previous research it has been shown that cer-
tain absorption bands in the infra-red are due to particular groups of
atoms. The relation of these results to the question of the structure of
crystals will be obvious to the reader. For, if the crystal is composed
of molecules of, say, water and calcium sulphate, which separately have
characteristic absorption bands, then, if these molecules or certain
groups of atoms in them undergo no physical change when they com-
bine to form a crystal (of selenite in this case), one would naturally
infer that the absorption spectrum of the product will be the composite
of the absorption bands of the two constituents.

This phenomenon is different from the one in which Julius1 showed
that the absorption spectrum of a chemical compound is not the com-
posite of the bands of the constituent elements. Here the physical
molecule has been changed.

Just how these specific groups of atoms are placed with respect to the
crystallographic axes it is impossible to determine.

A practical question in optics, viz, the possibility of finding material
suitable for prisms, in certain regions of the spectrum in which our
present prism material is opaque or in which its dispersion is small, was
also kept in view.

The reflecting power of metals was included to fill a gap left in the
work of Hagen and Rubens done at the Physikalische Technische
Reichsanstalt. The reflection from minerals composing the earth's crust
has unexpectedly furnished us with data which will be useful to the
meteorologist. It may even aid in clearing up such an obscure question

1 Julius : Verhandl. Konikl. Akad. Amsterdam, Deel I, No. I, 1892.

8 INFRA-RED TRANSMISSION SPECTRA.

as the radiation from the moon. The exact determination of the data
presented involves the use of one of the most sensitive heat-measuring
devices yet produced, viz, the radiometer. The region examined is a
vast one compared with the visible spectrum, and has heretofore been
little explored.

From the intimate relation between refractive index and reflecting
power, the data obtained gives us some idea of the dispersion of the
substances examined. The bands of selective reflection, especially the
very sharp ones, such as the new one found in quartz at 12.5 /x, will be
useful as a monochromatic source of energy, for example, in inter-
ferometer work.

Many chemical compounds contain both oxygen and hydrogen, which,
on applying heat, pass off in the form of water. That the water is not
united so tenaciously as the other constituents is evident from the fact
that in many instances it can be more easily removed ; many salts give
up their water if exposed to dry air at ordinary temperature. From
the circumstance that many of these compounds are crystalline, the
water is said to be present as "water of crystallisation." The manner
in which the water exists in the crystal is not understood. By some it
is considered a part of the chemical molecule ; by others it is thought
that the molecules of water exist in their entirety among the molecules
which constitute the crystal. It is characteristic of water of crystalliza-
tion that it is expelled at a temperature far below red heat, and fre-
quently below 100° C. Another characteristic of minerals containing
water of crystallization is their property of reabsorbing water after it
has been removed. Copper sulphate is an excellent example ; on apply-
ing heat, the blue crystal becomes a crumbling white mass, which, if
permitted to stand in dry air, absorbs water and resumes its blue color
and crystalline structure.

On the other hand, the water which is given off only at a red, or even
a white heat, can scarcely be present in the compound in the same man-
ner as the water of crystallization, and is distinguished as "water of
constitution." In this case the water is not supposed to exist as such
in the mineral, but to result from the union of oxygen and hydrogen or
from the hydroxyl groups contained in the compound.

Of course there are minerals which contain water in both these

\

forms. In copper sulphate (CuSO4+5H2O), four molecules of water
are given off at 100° and the fifth at 200°, so that the latter is possibly
combined in some manner different from the former. In epsomite
(MgSO4-f-7H2O), six molecules of water are given off at 132°, while

METHOD OF INVESTIGATION. 9

the seventh is held more tenaciously, and is not given off until heated
to 210°.

However, in cases like these, where successive portions of water are
given off at different temperatures, it is difficult to make a distinction
between water of crystallization and water of constitution. In the min-
erals just quoted, however, it has been found that the heat of hydration
of the last molecule of water is different from that of the molecules of
water which pass off at a lower temperature, which confirms the belief
of a difference in the bonding in the two cases.

As a whole, the question of the association of the atoms of oxygen
and hydrogen in certain compounds is far from settled, while the diffi-
culties involved in its investigation are very great. The existing data
bearing upon the subject is practically nil.

PRESENT METHOD OF INVESTIGATION.

In a preliminary paper bearing upon this subject1 the writer described
the infra-red absorption spectra of two numerals, selenite (CaSO4-(-
2H2O) and brucite (Mg(OH)2), in which the atoms of oxygen and
of hydrogen are thought to be combined in a different manner. It was
the application of the results of previous work, in which it was abun-
dantly proven that certain groups of atoms have characteristic absorp-
tion bands. Hence, if the oxygen and hydrogen, which enter into the
composition of certain minerals, are united as they are in a molecule of
water, then one would expect to find the absorption spectra of such
minerals to be a composite of the bands of water and of the bands
caused by the other constituents.

The only previous investigation bearing on this subject, from the
standpoint of infra-red absorption spectra, is that of Konigsberger,2
who studied the pleochroism of several minerals, including selenite.
Unfortunately his plate of selenite was too thick, so that no energy was
transmitted beyond 2.5 /x. He calls attention to the fact that a small
absorption band at 1.5 /A coincides with that of water, from which it
would appear that in selenite the absorption of the "water of crystalliza-
tion" does not appear to be different from that of water.

Thinking that it would be fairer to select the large water bands at
3.0, 4.75, and 6 /x, as a criterion, the writer (loc. cit.) examined thin sec-
tions of several minerals and found, as had been anticipated, that min-
erals containing water of crystallization have large absorption bands

1 Physical Review, 20, p. 252, 1905.

- Konigsberger : Ann. der Phys., 61, p. 703, 1897.

io INFRA-RED TRANSMISSION SPECTRA.

coinciding with those of water. The selection of selenite (CaSO4-f
2H2O) and brucite (Mg(OH),) as types of water of crystallization and
of constitution was a fortunate one, for the two following reasons:
(i) The selenite curve showed all of the absorption bands of water in
their proper positions and intensities, except the 4.65 p band, which is
shifted and too deep for the thickness of water contained in the plate
under examination. The fact that the band is shifted caused me to
suspect that it may be complex, and, from the fact that it lies in the
region where the NCS radical and certain sulphur compounds have a
strong absorption band, that it may be due to the SO4 radical — a sur-
mise that since then has been unexpectedly verified. (2) Although the
brucite curve did not contain the water bands, thus showing the differ-
ence between water of constitution and water of crystallization, it con-
tradicted my previous work,1 in which it was shown that the OH radical
in the alcohols has an absorption band of 2.95 /A (true value is 3/u.).
From the brucite curve the conclusion was drawn that the OH radical is
inactive. Since then I have studied the chemical side of the question
more thoroughly, from which it appears that there is no marked differ-
ence2 in the activity of the OH radical in the hydroxides studied, and,
hence, that the brucite curve should have a band at 3.0 p, instead of the
one shifted to 2.5 p.. In other words, the present research, which is in
part the outcome of the foregoing discrepancies, has to a very unex-
pected degree furnished us with an abundance of new proof that certain
groups of atoms have definite absorption bands, which really means that
these groups of atoms, or "ions," enter into the various compounds in a
similar manner.

To sum up, in the present investigation the criterion for distinguish-
ing water of crystallization from water of constitution is the presence
of absorption bands at 1.5, 2, 3, 4.75, and 6 p., which is the location of
the absorption bands of water. If there are no other absorption bands
near by, then the intensity of these bands should be somewhat like that
of water, viz, the bands at 1.5, 2, and 4.75 /* are weak, while the bands
at 3 and 6//, are very strong.

A hydroxyl group will also cause an absorption band at 3 p.. Silicates
may have a small band shifting from 2.9 to 3.1 p., but since it is weak,
there is no danger of confusing it with the strong water band in the

same region.

1 Investigations of Infra-red Spectra, Washington, 1905 ; Phys. Rev., vol. 22, 1905.

2 From the chemical standpoint, however, the OH in alcohol and in H2SO4 is
more active, since it is replaceable by a metal (more acid) than the OH in brucite

, which is not replaceable (more basic).

PLEOCHROISM. II

PLEOCHROISM.

Since the present research involves the examination of minerals
which are not isotropic, the question presents itself whether it would be
better to examine them in natural or polarized light. In the former
case we obtain all the absorption bands, irrespective of the direction of
transmission, but their observed intensities will not necessarily be real.
In the latter case, where the intensity of the transmitted energy depends
upon the direction of vibration of the polarized rays with respect to the
axes of the crystal (pleochroism), the observed intensities will be real,
the number and position of the absorption bands will depend upon the
direction of observation, but the accuracy is less the farther one pene-
trates the infra-red, due to loss in polarizing the source of energy.

By these statements it is not intended to give the impression that
pleochroism is not worth considering. The work of Merritt1 and of
Konigsberger2 is especially valuable in demonstrating that absorption is
dependent upon the direction of vibration of the incident energy. The
visible spectrum is so narrow that it is difficult to separate the absorp-
tion bands, while, on the other hand, the infra-red is a vast and almost
unexplored region, and it is here that one would expect to resolve wide
absorption bands such as exist in the visible spectrum. The curves of
Merritt and of Konigsberger show this, especially the calcite curves,
which are entirely independent for the ordinary and extraordinary rays.
Nevertheless, in looking over the previous work, it seemed to the writer
that in the present state of our knowledge of absorption a greater
advance would be made by simply mapping the spectra of a great many
minerals for energy transmitted through them in its natural mode of
vibration.

The results have been most gratifying. If we had been content with
the examination of a few, easily obtainable, large-sized crystals like
colemanite, for pleochroism, the results would have been quite different.

The maxima of the absorption of water are constant in position for
the amorphous and the crystalline minerals studied, which would indi-
cate that in this case the transmission is not affected by the direction
of vibration.

The desirability of examining many minerals is illustrated in the
appended curves, in which it will be seen that frequently in one region of
the spectrum the water bands are obliterated by an increase in the gen-
eral absorption or by an adjacent absorption band, while in another part
of the spectrum the water bands are very distinct.

1 Merritt : Phys. Rev., 2, p. 424, 1894.

2 Konigsberger : Ann. der Phys., 61, p. 687, 1897.

12 INFRA-RED TRANSMISSION SPECTRA.

APPARATUS AND METHODS.

In the present investigation a mirror spectrometer, a rock-salt prism,
and a Nichols radiometer were used. With the exception of certain
improvements, the details in mounting and adjusting the different parts
have been described elsewhere,1 and it will be sufficient to add that the
spectrometer mirrors were 10 cm. in diameter and 50 cm. in focal
length, and that the rock-salt prism was an unusually fine one having
a refracting angle of 59° 59' 36" and faces 9 by 9 cm. area. The spec-
trometer slits were 0.3 mm. wide, or about 2' of arc. The time required
for the radiometer to reach a maximum deflection was about 20 seconds.
The different portions of the spectrum were projected upon the radi-
ometer slit by using the Wadsworth2 mirror-prism method and rotating
the prism-table. Although one doubles the error in making spec-
trometer settings, the method was the most convenient one in the
present work. The "heater" of a no Nernst lamp on an 8o-volt
storage-battery circuit was used as a source of energy. The sections
were mounted in the usual manner upon a wooden carrier which moved
in vertical ways between the source of energy and the collimator slit.

To protect the prism from moisture a small crystallizing dish of
phosphorus pentoxide was placed over it, and when not in use a large
glass crystallizing dish was inverted over the prism and drying mate-
rial. The edge of the large dish was ground to fit the prism-table and
prevented the entrance of moisture.

The spectrometer was remodeled from a small Fuess pattern having
mirrors 4 cm. in diameter. The focal length of the new mirrors was
selected as being more advantageous than the 35 cm. or 100 cm. mir-
rors, used in previous work, on account of the increased resolving
power over the former and the decreased atmospheric absorption which
is apparent in the latter. The advantages gained by the use of the
larger prism and mirrors is so much more than anticipated that it
seems worth while to mention it here. The whole was inclosed in a
tin box to protect the prism from air currents and from moisture.

The radiometer was of the usual form, with an additional improve-
ment which consisted in covering it with a piece of heavy brass tubing,
within which was a thick layer of hair felt. Although the radiometer
stood close to a large outside window and exposed to sunlight, it was
not affected by temperature changes. The vanes were of mica, covered
with copper oxide, which was applied by means of shellac or a thin

1 Phys. Rev., vols. 16 and 17, 1903; Investigations of Infra-red Spectra, Wash-
ington, 1905.

2 Wadsworth: Phil. Mag. (5), 38, p. 346, 1894.

ACCURACY ATTAINABLE. 13

layer of Canada balsam. This method of blackening the vanes was
found easier and more satisfactory than smoking them.

The sensitiveness was about 10 cm. deflection per square millimeter
of exposed surface, scale and candle at I meter. The weight was 10
milligrams. The vane was made heavy to avoid tremors.

ACCURACY ATTAINABLE.

The writer has been so frequently questioned on the accuracy of the
observations that a few remarks will be in order.

First of all, nearly all the substances examined have numerous
absorption bands, so that we can not find an extinction coefficient by
means of which we may determine their transmissivity in absolute
value for, say, a centimeter thickness. Again, different samples of a
given mineral will vary in homogeneity and in purity. Different parts
of a given specimen will vary in thickness, homogeneity, purity, and
polish, so that, on remounting the same sample before the spectrometer
slit, the relative values of transmission may differ by several per cent.
The result is that such a comparison has but little meaning, and is at
all times of minor importance. But, after once mounting the specimen
before the spectrometer slit, the important question is the relative trans-
parency of different regions of the spectrum and the accuracy of the
location of the maxima of absorption. This depends upon the accuracy
of the observations in any region of the spectrum. With a trust-
worthy instrument like the radiometer, and with the specimens securely
mounted before the spectrometer slit, in the present research it was
found that for a given spectrometer setting the variation of single
observations would differ from the mean by only about a tenth of i
per cent, so that only in the extreme infra-red, or in locating sharp
maxima, were the observations repeated. Consequently the reality of
many of the small maxima can be accepted with confidence. The
most important question, however, is the accuracy of the location of
the maxima, especially those which are sharp and well defined.

During the past few years the writer has repeatedly mounted and
adjusted his apparatus and has always found the location of certain
sharp absorption and emission bands, used as standards of reference, to
agree to 0.02 p., which is as accurate as our knowledge of the dispersion
of rock salt will permit. The adjustment of the zero of the instrument
was tested daily. It was found that the yellow helium line is narrower
and brighter and thus better adapted than sodium for making this
setting.

14 INFRA-RED TRANSMISSION SPECTRA.

PREPARATION OF THE MINERAL SECTIONS.

Whenever possible the crystals were split parallel to a cleavage plane,
since many crystals would break in grinding. The remaining ones were
ground thin in the ordinary manner. To accomplish this they were
mounted upon glass plates by means of pitch having a low melting-
point, so as to avoid disturbing the water of crystallization which is
expelled from some minerals at a low temperature. The sections were
cleaned in chloroform and mounted upon heavy cardboard having
rectangular holes about 4 by 10 mm. cut into them. That the crystals
were not affected by grinding and the cement was proven in a number
of cases where cleavage sections of the same mineral were examined.
Sometimes the sections were mere fragments 5 mm. long, so that two
pieces had to be placed end to end with a slight overlapping, but not
glued at the center. Since we are not concerned with total transmis-
sion, but simply with absorption bands, this does not interfere with the
investigation. Of course if a glue could have been found which is free
from large absorption bands, the sections could have been mounted upon
rock-salt plates and ground much thinner. As it was, the sections
were entirely free, except at the ends, which rested upon the cardboard,
which was then mounted upon the holder, which moved in vertical
ways before the spectrometer slit.

In the description of the different minerals, the direction of cutting
the sections is parallel or perpendicular to some crystallographic axis.
In many crystals there is usually a perfect cleavage parallel to one of
the crystallographic axes, which fact was taken advantage of in pre-
paring the sections. The cleaving was done by means of a small knife-
blade struck with a light piece of metal, for example, a file. It may be
of interest to note that with crystals, for example anhydrite, which have
perfect cleavage in several directions, it was necessary to strike the
knife-blade a sudden blow, while in others, like scolecite, a series of
light blows started the cleavage and produced large plates. A great
many minerals like copper sulphate and alum could not be used, because
a slight warming was sufficient to dehydrate them, when the surface
became rough and scattered the incident radiation.

In the present research the examination of the sections after being
prepared was overshadowed by the difficulty in obtaining specimens of
sufficient size and in grinding them thin enough. In my previous work
it was found that as a general rule the compounds examined increase
in their general absorption beyond 5 /*, so that the (liquid) films had
to be reduced to o.oi mm. In the present work it was impossible to
reduce the thickness to much less than o.i mm., with the consequent
result that many sections became opaque beyond 5 /JL.

SCOPE OF PRESENT INVESTIGATION. 15

The minerals were obtained from various dealers and from the U. S.
National Museum. They were selected according to composition. Of
the total number thus selected only about 40 per cent were obtainable
and these were oftentimes the least desirable to illustrate the points in
question. The expense involved is no small item, and I am very grate-
ful to Director Stratton for his generosity in meeting that question.

SCOPE OF PRESENT INVESTIGATION.

The principal object in this research was to gain information in
regard to the molecular structure of minerals containing oxygen and
hydrogen. To this end minerals were selected containing these con-
stituents. Previous work has shown that in organic compounds certain
radicals have characteristic absorption bands. One would therefore
naturally expect to find similar relations in inorganic compounds,
where more important radicals are to be found. Hence, the list was
extended so as to contain a series of minerals having the radicals CO2,
PO4, SO4 and OH. The results obtained serve the purpose very well
for demonstrating the rashness of attempting to establish "laws" from
present data for future investigators to refute, just as is being done at
present in more thoroughly explored fields, such as, for example, the
absorption of solids in solution.

In the study of the carbon compounds in organic chemistry, their
constitution has been established by the replacement of certain con-
stituents by organic radicals, by the preparation of a series of deriva-
tives, by vapor density determinations, or by studying their physical
properties in solution. Mineralogy is essentially the chemistry of
silicon compounds, to which, as yet, it has not been possible to apply
any of the above methods, so that the constitution of many minerals
has been derived from analogies with other compounds which are
better understood. Hence, in an investigation like the present, it is not
surprising to find exceptions to the tentative rules for classifying these
minerals.

The transmission curves are given without correction for reflection
from the surfaces. In most cases, as will be noticed from the reflection
curves, the correction is negligible. In one case, viz, stibnite (Sb2S3),
such a correction is worthy of notice. The transmission shows a uni-
form value of 45 per cent. To the writer this seemed a very transparent
substance until the greater surprise of a uniform reflecting power of
about 36 per cent was observed, which indicates that after correcting
for reflection this mineral (for the thickness examined) is practically
transparent.

CHAPTER II.

TRANSMISSION SPECTRA OF DIFFERENT MINERALS.

GROUP I: MINERALS CONTAINING WATER OF CRYSTALLIZATION.

All the minerals studied which are generally thought to contain
water of crystallization are collected in Group I. First of all, it will
be necessary to consider the absorption spectrum of water, which has

WO?

0 I Z 3 4 56789

FIG. i.— Water.

been studied by Julius,1 Paschen,2 Aschkinass,3 and others. All observ-
ers agree in their location of large absorption bands at the approximate
wave-lengths 1.5 /*, 3 /*, 4.75 /*, and 6 p.

1 Julius : Verhandl. Konikl. Akad. Amsterdam, Deel I, No. I, 1892.
'Paschen: Ann. der Phys. (3), 53, p. 334, 1894.
'Aschkinass: Ann. der Phys. (3), 55, p. 406, 1895.

2— c 17

i8

INFRA-RED TRANSMISSION SPECTRA.

All have found water extremely opaque to infra-red radiation, so that
the film had to be reduced to a few thousandths of a millimeter in order
to be able to study it at all. In fig. i are given the absorption curves
of water is found by Aschkinass. His values of the maxima are slightly
greater than those found by others, but this is simply due to a fault in
his calibration. The curves for the different thicknesses (0 = 0.05;

90 %

O I 2 3 4 5 6

FIG. 2.— Selenite (a and c) ; Anhydrite (b).

b = o.oi mm.) illustrate very well what we are to expect in the case of
minerals containing several molecules of crystal water. For a very
much thinner film (curve c = 0.001 mm.) by the writer1 the absorption
band at 4.75 //, has entirely disappeared.

SELENITE (CaSO4+2H2O) ; ANHYDRITE (CaSO«).

(Figs. 2, 3, and 4. Selenite: Monoclinic, cleavage parallel to b; t = 0.648 mm.
Anhydrite: Orthorhombic, cleavage piece parallel to c; t — 0.656 mm. From
Stassfurt, Germany.)

Of all the minerals studied containing water of crystallization these
two are the most conspicuous for demonstrating the effect of the presence

1 Phys. Rev., 20, p. 257, 1905.

WATICR OF" CRYSTALLIZATION.

of H2O. It would have been highly desirable to procure more minerals
which occur in the hydrous and anhydrous state, but none were obtain-
able. The artificially dehydrated minerals, such as copper sulphate, were
too opaque for examination after expelling all the water. The only
exception is selenite, the various curves of which will now be considered.
The anhydrite curve, b, is to be noticed first, from which it will be

90%

01 234 56

FIG. 3.— Anhydrite (d) • Selenite.

observed that there are small bands at 1.9, 3.2, 5.7, 6.15, and 6.55 p.,
and an enormous band at 4.55 p, which will be shown later to be due
to the SO4 ion.

Turning to the selenite curve (a, fig. 2), it will be noticed that it is
less transparent for the same thickness, 0.65 mm., and that in its general
trend it is similar to the curve for water. All of its absorption bands
coincide with water, with the exception of the 4.75 //, band, which is
shifted to 4.6 /*. The shifting of this band is due to the SO4 band at
4.5 fi, as was found on examining anhydrite.

The curve, c, for a thickness of 2.57 mm. is due to Konigsberger
(loc. cit.).

20

INFRA-RED TRANSMISSION SPECTRA.

In fig. 3 is illustrated the effect of dehydrating the selenite. This
was accomplished with difficulty on account of warping and shrinking
of the plate, which necessitated dismounting the specimens for each
heating. It was found necessary to clamp the specimen between two
metal plates in order to prevent it from warping and breaking. Curve a
gives the transmission of a clear piece having a thickness of 0.204 mm- >
curve b shows the transmission after partial dehydration.

7O%

S 6 7 8

FIG. 4.— Selenite.

It is of considerable interest in showing the permanence of the 4.55 /*
band. Beyond 6 ^ the transparency increases and becomes 1.5 times as
great as selenite at 6.9 p. The dehydrated section, which is an opaque
white mass, was then moistened with water and allowed to stand over
night to dry and set, just as in the case of plaster of Paris. The result
is shown in curve c, where the transparency at 6.9 p. has decreased to
almost the original value found for the transparent crystal. It will of
course be understood that the great opacity up to 5 p. for the dehydrated
section is due to the lack of homogeneity introduced in expelling the
water, which has the well-known property of scattering the radiation
of short wave-lengths. At 7/x the increase in absorption due to the
presence of water can not be doubted.

In fig. 4 the transmission curves of a very much thinner partially
dehydrated section are given. In curve a the section was subtrans-
parent and in curve b it was translucent. The curves are given to show
the increased transparency beyond the band of metallic reflection in
the region of 9 /x.

WATER OF CRYSTALLIZATION.

21

Curve c shows the metallic reflection band found by Aschkinass,1
and it will be noticed that the position of the maximum as found by
him at 8.69 /t is shifted from that indicated by the present transmission
curves. The same lack of coincidence of the maximum of the reflection
band with the minimum in the transmission curve will be found in
calcite and in mica. This is the first attempt to explore the region of
transmission near a reflection band and the explanation to be offered
later on for the discrepancy may not hold when more data have been
accumulated.

QUARTZ (SiO2) ; OPAL (SiOa+xH2O).

(Quartz: Cut perpendicular to axis; / = 1.30 mm. Opal: Massive, transparent;

t = 0.12 mm. Fig. 5.)

'00%

3456

FIG. 5.— Opal (b) ; Quartz.

The specimen of quartz examined was made for polariscopic work
and was perfectly clear. Its transmission curve (a, fig. 5) shows small
bands at 2.9 /x and 4.35 p. Curve c, which shows a series of bands at

1 Aschkinass : Ann. der Phys. (4) , i, p. 42, 1900.

22

INFRA-RED TRANSMISSION SPECTRA.

5.02, 5.3, 6, 6.26, and 6.65 /x, is due to Nichols,1 who used a section
0.018 mm. thick. Unfortunately his curve begins in the slight band at
4-35 A1; so that there is no check upon its accuracy of location.

Konigsberger2 examined smoky quartz and found bands at 3.05 and
4.05 /«,. Merritt,3 using polarized light, found bands at 3 and 3.6 /t for
the extraordinary ray, and bands at 2.9, 3.75, and 4.1 p. for the ordi-
nary ray.

For amethyst, which is violet quartz, of which the color has been
attributed to manganese, Konigsberger found an absorption band at
3.1 fi. As a whole, different observers agree in locating a small band
near 3 p. which will not interfere in considering water of crystallization.

60%

345

FIG. 6.— Heulandite.

to /•*//

Opal is quartz containing variable proportions of water, from 5 to 30
per cent. It shows no traces of crystallization. Neither is it considered
a solid solution, for the water contained is not a function of the vapor
pressure. The transmission curve of opal has the general outline of the
curve for water and contains the bands of water at 1.5, 2, 3, and 6/u,,
as well as the silicon bands at 4.2 and 5 p. The 3 and 6 p. bands are
the composite of the water and silicon bands in those regions.

(Section ground parallel to b; f = o.20 and 0.22 mm.; semi-transparent. Fig. 6.)

Heulandite belongs to a class of minerals called zeolites,4 which are
hydrated silicates of alumina, alkalies, and generally lime. To this class

1 Nichols : Phys. Rev., 4, p. 297, 1896.

* Konigsberger : Ann. der Phys. (3), 61, p. 687, 1897.

'Merritt: Ann. der Phys. (3), 55- P- 49, 1895.

1 Miers : Mineralogy, p. 483.

WATER OF CRYSTALLIZATION.

belong heulandite, stilbite, analcite, natrolite, and scolecite, which are
included in the present research.

In these minerals the water of crystallization is very loosely held.
It is impossible to distinguish between water of constitution and water
of crystallization. The water continues to be gradually expelled as the
temperature is raised and may be replaced by other substances, such
as NH3, H2S, or C2H5OH. The dehydrated crystals absorb definite
quantities of these substances as a sponge absorbs water, the process
being accompanied by the evolution of heat.

90%

10

O 2 3 4 5 6 7 8

FIG. 7.— Stilbite (a); Potassium alum.

Since the per cent of water present varies continuously with the
vapor pressure it is generally concluded that the water of the zeolites is
not analogous to the water of crystallization of most hydrated salts, but
resembles more nearly the intermixture which occurs in solid solutions.

In fig. 6 two curves of heulandite are given. The sections were about
the same thickness, and the great difference in transparency is due to
the fact that the crystals were of different homogeneity and trans-
parency. The curves show the presence of the water bands at 1.5, 2,

24 INFRA-RED TRANSMISSION SPECTRA.

3, and 4.75 /x, beyond which point the opacity becomes too great for
further exploration. The crystals obtainable were small, so that the
section was not quite long enough to cover the whole slit.

STILBITE (CaAl2Si6Ow+6H2O).
(Transparent section cut perpendicular to optic axis. f = o.ii. Curve a, fig. 7.)

The stilbite curve is to be noticed for its great transparency with all
the water bands superposed. There do not appear to be any important
bands belonging to the mineral itself.

POTASSIUM ALUM (K2SO4Al2(SO4)a-|-24HaO).
(Cut perpendicular to axis. t = 0.085 mm.; transparent. Curve b, fig. 7.)

The exploration of the spectrum extends to 3 /*. The water bands at
1.5 and 2 n are present. With 24 molecules of water and SO4 bands

01 a 3 4 5 6

FIG. 8.— Natrolite (a) ; Scolecite.

at 455ft, to be mentioned later, it would not be possible to penetrate
beyond 3 /*, unless a thinner section could be made.

NATKOLITE (Na2Al2SisO10+2H2O).

(Orthorhombic section ground parallel to m; subtranslucent ; f = o.u mm.

Curve a, fig. 8.)

The specimen from which this section was cut was a tuft of acicular
crystals. The section was made from the more massive part, and was
almost opaque for the thickness used. The curve (a) in fig. 8 shows
the water bands in their usual positions and proper intensities.

WATER OF CRYSTALLIZATION. 25

SCOLECITE (CaAl2SisO,0+3H2O).

(Monoclinic section split parallel to m; perfectly transparent; t = 0.565 mm.

Curve b, fig. 8.)

This mineral differs from the preceding in being a lime zeolite, and
in having one more molecule of water. Its greater homogeneity makes
it more transparent in the region of the short wave-lengths. It has the
general outline and the absorption bands of the water curve up to 5 /*,
where it is completely opaque. The 3 /* band is evidently complex.
Prof. S. L. Penfield kindly presented this specimen.

80 %r

23456

FIG. 9. — Analcite (a); Colemanite.

ANALCITE (Na2Al2Si4O
(Cut perpendicular to optic axis; subtransparent ; f = o.ii mm. Fig. 9.)

This specimen, made by Steeg & Reuter, was small and broken in
shipping, so that by placing two pieces end to end the spectrometer
slit was not quite covered. Nevertheless it shows a transparency farther
into the infra-red than is usual. As a whole, however, it is very opaque
(curve a, fig. 9), and the water bands at 1.5, 2, and 4.7 p. are quite
obliterated. There seem to be no other than water bands.

COLEMANITE (Ca2B8On+5H2O).
(Cut parallel to b; transparent; / = 0.268 mm. Curve b, fig. 9.)

This mineral is obtainable in large transparent crystals. Neverthe-
less, it is unusually opaque to infra-red radiation, so that very little

26

INFRA-RED TRANSMISSION SPECTRA.

energy is transmitted beyond 3 /A; even the water bands at 1.5, 2 and 3
are almost obliterated. As the examination proceeds it will be noticed
that this opacity appears to be a property of the borates.

CALCIUM CHLORIDE (CaCl2+6H2O).
(Film melted between rock-salt plates, t = 0.1 mm. Fig. 10.)

This compound comes in large hexagonal crystals. A specimen was
melted between two plates of rock salt, and stood over P2O5 for several
days, until the edges became white from dehydration. The substance

90%,

10

0 I Z 3 4 5678

FIG. 10. — Calcium chloride (a); Potassium ferrocyanide.

recrystallized in the meantime and when examined showed the water
bands in their usual position and intensities except the 3 p band, which
is shifted to 3.2 yu..

POTASSIUM FERROCYANIDE (K4Fe(CN)«,+3H2O).
(Section split parallel to c; transparent; £ = 0.3 mm. Fig. 10.)

The water is so easily expelled from this mineral that the heater had
to be used on 70 volts. The transmission curve shows the water bands

WATER OF CRYSTALLIZATION. 27

at 1.5, 2, and 3 /*, while the next two bands are shifted to 5 and 6.2 ju.
respectively. Porter,1 by means of "reststrahlen," located a band at
4.84 it.

APOPHYLUTE (H7KCa4(SiO3)8+4/^H2O).
(Section cut parallel to c; transparent; t = o.il mm. Fig. n.)

This mineral is entirely opaque beyond 3 p. up to 8 fi, where there is
slight transparency. The water bands at 1.5 and 2 //, are almost oblit-
erated.

60%

3 4- 5 6 7 Q

FIG. it. — Apophyllite (a); Deweylite

10

II

DEWEYLITE

(Massive; whitish color; subtranslucent in section 0.08 mm. thick. From New

Rochelle, New York. Fig. u.)

In the curve, b, for this mineral the small absorption bands are oblit-
erated but the bands at 3 ^ and 6 /*, are very strong, which would make
it appear that the water is present in the form of water of crystalli-
zation.

THOMSENOLITE (NaCAlF8-f-H:>O).

(Monoclinic; cleavage parallel to c; transparent; f = o.75 mm. From Ivigtut,

Greenland. Curve a, fig. 12.)

This mineral was obtained as a mass of small crystals, from which
one was obtained having an area of about 4 by 5 mm. This crystal

1 Porter : Ap. Jr., 22, p. 229, 1906.

28

INFRA-RED TRANSMISSION SPECTRA.

was split into two inclined prisms, which, when placed end to end,
made a tight joint. The transmission curve, a, shows all the absorption
bands of water in their proper intensities. A new band occurs at 2.6 /i
which widens the one at 3 /u, but does not displace it. The specimen
suddenly becomes opaque at 6 p..

GISMONDITE (Hi(Na2Ca)Al2Si6Oi8+4H2O).

(Stalactitic mass; subtranslucent ; f = 0.245 mm. From County Antrim, Ireland.

Fig. 12.)

This specimen was not very homogeneous, but, in spite of its great
complexity chemically, it is fairly transparent. The only bands present
are those due to water at 1.5, 2, 3, and 4.7/4. The band at 3;* is very
wide, which suggests the presence of others, perhaps of silicon, at 2.9 /i.

60%

Z 3 4 5 6

FIG. 12.— Thomsenolite(a); Gisraondite.

QLJU

BLODITE (MgSCXNaSO^HjO).
(Cut perpendicular to axis; t = 0.10 mm. Curve a, fig. 13.)

This mineral shows all the water bands, the one at 2/x being con-
spicuous for its sharpness. The band at 4.6 is composite, due to the
SO4 band at 4.55 //,, as will be shown in discussing the sulphates, while
the 3 n band is also complex.

THAUMASITE (CaSiO3CaCO3CaSO«-r-i5H2O).
(Cut perpendicular to axis; transparent; t = 0.125 mm. Fig. 13.)
This mineral is too opaque for heat rays to be considered in demon-
strating the presence of water of crystallization. The water bands at

WATER OF CRYSTALLIZATION.

1.5 and 2)U, are well defined, considering the great opacity of the sub-
stance.

70%

50

40

to

"I"

c

10

h

10

3 4 5 6 7

FIG. 13.— Blodite (a); Thaumasite (b).

KIG. 14.— Hydrotalcite.

INFRA-RED TRANSMISSION SPECTRA.

HYDROTALCITE (Mg3Al(OH)6+3H2O).

(From Vernon, New York. Lamellar; massive; subtranslucent ; £ = 0.04 mm.

Fig. 14.)

The hydroxide group (OH), to be noticed later on, confuses matters
at 3 fji. The other water bands at 1.5, 2, and 4.75 /u. are visible in spite
of the great opacity, which becomes complete at 6 p..

a

FIG. 15.— Varicite (a) ; Wavellite (b and c).

VARICITE (A1PO«+2HSO).

(Massive; blue color; subtranslucent in thin section of 0.03 mm. From Lewiston,

Utah. Fig. 15.)

In this mineral the water bands at 1.5, 2, and 4.75 ju, are almost obliter-
ated. Those at 3 and 6 p are shifted and are no doubt composites.
From this and other minerals studied, having the PO4 group, there is no
marked evidence of an absorption band belonging to this group.

WAVELLITE (AlOH),(POO.+5HaO.

(Crystallized in rays of needles; section ground parallel to rays; subtranslucent.

From Arkansas. Fig. 15.)

The lack of homogeneity of the crystals renders this mineral very
opaque. The water bands at 1.5 and 2 /* are visible, in spite of the
complete opacity at 3 /u.. This great opacity appears to be a characteris-
tic of the phosphates, as has already been observed in the borates.

WATER OP CRYSTALLIZATION. 3!

VIVIANITE (Fes(PO4)2+8H2O).

(Monoclinic; section ground parallel to b; bluish color; translucent; t = 0.11 and

0.25 mm. Fig. 16.)

This section is practically opaque beyond 3 /*. The water bands at
1.5 and 2 p. are visible, although slightly shifted, which is due to general
transparency of the mineral. This is well illustrated in curve c, which
is curve b magnified 5 times. I am indebted to the late Prof. S. L.
Penfield for this mineral.

FIG. 16.— Vivianite.

MEU.ITE (Al2C«Oi2+i8H2O). (Fig. 17.)

This mineral occurs as waxy crystals of variable transparency in
European coal measures. It was selected on account of its carbon, and
also because of the large number of molecules of water, which is
expelled at a low temperature.

In fig. 17, curve a, is given the transmission curve of a mineral from
Steeg and Reuter, purporting to be mellite or "honeystone," / = 0.125
mm. It was the most flagrant exception to the rule that minerals having
water of crystallization are very opaque and have absorption bands in

32 INFRA-RED TRANSMISSION SPECTRA.

common with water. Evidently this needed further investigation;
and this was done with minerals purchased from various dealers, care
being taken to obtain them from different localities. Curves b and
c give the transmission for transparent yellow crystals, t=o.^ and
0.15 mm. respectively, which came from Arten, Thuringia. Curve d
is the transmission for a grayish-white translucent crystal, ground to
0.12 mm. in thickness, from Mallonka, Austria. Curve e is for a white
transparent crystal, 0.06 mm. in thickness, from Tula, Russia. In all
these curves it will be noticed there is no energy transmitted beyond

10

345678
FIG. 17.— Cassiterite (a) ; Mellite.

10

3/i, while from the very nature of the transmission curves the 1.5 band
is invisible and the 2 /* band is almost obliterated. As a whole the
theory that, in crystal-water, the bonding is the same as in ordinary
water has also been confirmed with mellite. A letter of inquiry to
Steeg and Renter in regard to the sample sent us revealed the fact
that cassiterite (SnCX) was accidentally substituted for mellite. The
fact that it was possible to detect the error would indicate that the
method of analysis is trustworthy, and in that respect it is fortunate that
the substitution occurred.

WATER OF CRYSTALLIZATION.
d-FRUCTOSE

33

(f = o.i mm. Fig. 18.)

The sugars are complex carbohydrates, the constitution of which is
not well known. They are rich in hydroxyl groups similar to the
alcohols, and hence one would expect to find a band at 3 /*. We would
also expect a band at 3.43 /* due to CH2 or CH3 groups, so that as a
whole the curve will be complex at 3^. The water bands at 1.5, 2,

80%

345

FIG. 18.— rf-Fructose.

and 4.5 /x, should appear in those compounds containing water of crys-
tallization. The region from 4 to 6/x ought to be one of great trans-
parency if previous results hold. It is a property of sugars that they
decompose and give off water when heated above the melting-point.

In rf-fructose (C6H12OG) the constitutional formula is written
CH2OH(CHOH)3COCH2OH. The curve (fig. 18) shows no absorp-
tion in the region of the water bands. There is a large band at 3.25 //,
and a small one at 5.85, while beyond 7/x, there is complete opacity.

3— c

34

INFRA-RED TRANSMISSION SPECTRA.

Fructose belongs to the anhydrous group of compounds, but is con-
sidered here in order to show the great contrast between it and other
sugars having water of crystallization.

J-GLUCOSE (DEXTROSE) (CeHiaO.+H2O).
(/ = o.i4 and 0.18 mm. Fig. 19.)

This compound differs from the preceding in having one molecule of
water of crystallization. The region of 3 to 4 ju, is very complex and
indeterminate, but the water bands appear at 1.5, 4.75 and 6ju, the

345
FIG. 19. — Dextrose.

to 14-)

latter being quite indistinct. A sample was heated at 110° for four
hours which is supposed to render it anhydrous. After standing over
P2O5 for several days, several crystals had formed in the glassy mass.
These were melted between plates of rock salt. Curve b shows, how-
ever, that the water bands still remain, and from the fact that the
amorphous material could not be entirely removed from the crystals
it is quite probable that there was still some of the hydrous material
present.

WATER OF CRYSTALLIZATION.

35

CANE SUGAR (CuH«Ou).
(Melted between rock salt; t = o.i. Figs. 20 and 200.)

In fig. 20 are given a series of curves for different samples of cane
sugar, which is supposed not to contain water of crystallization. It is
one of the most conspicuous exceptions to the rule, if it really be anhy-
drous. Curve a represents the transmission for barley sugar, which is
the name given to cane sugar after it has been melted, when it becomes
an amorphous mass. Curve b is another sample, partially dehydrated
and decomposed, called "caramel." It was light brown. Curve c is for

90%

60

70

60

<n
,5

1*0
<0

u

30

to

a

d

to /4 u,

FIG. 20. — Cane sugar.

a dark-brown caramel made by driving off some 15 per cent of the
water. In this sample it will be noticed that the 1.5 and 4.75 p, water
bands have quite disappeared, which is to be expected.

Curve d shows the transmission through a film (£ = 0.2 mm.) of
chemically pure sugar. It was an amorphous solid melted between
two plates of rock salt. It was originally crystallized from alcohol and,
hence, was free from water. The amorphous film was colorless, showing
that there was no decomposition in melting. Nevertheless the water
bands at 1.5, 2 and 4.75^ are present.

80%r

30 INFRA-RED TRANSMISSION SPECTRA.

Curve f, fig. 2oa, is for an amorphous film of beet sugar, which also
shows the 1.5 /x, band. Curves e and h are for two large, clear crystals
of rock candy, respectively 2 and 1.7 mm. in thickness, in which the 1.5/1,
band is unusually strong, which would indicate that there was no decom-
position in melting the other films. An analysis by Dr. J. C. Blake
showed that the rock candy contained 99.81 per cent pure sugar, while
the water content was about 0.05 per cent. Now, it has never before
been suspected that cane sugar has water of crystallization, and the
present disagreement with the results of chemists who have made a

life-long study of sugars in
general must be considered
with caution. A compari-
son with lactose, maltose,
and dextrose would indicate
the presence of a molecule
of water of crystallization.
There may be other ex-
planations for this excep-
tion. It has already been
noted that the sugars are
rich in OH groups, which
will confuse matters at 3 /x.
where the OH radical has a
characteristic band, so that
the effect of water must be
based upon the small bands
at 1.5 and 4.75 /x. In the
present work on mannite
(C6H8(OH)6) and in pre-
vious work1 on ethyl alcohol
(C2H5OH) and myricyl alcohol (C30H61OH), as well as on glycerin
and on phenols, there is no indication of absorption bands at 1.5, 2,
4.75, and 6/x,, so that their presence in the cane-sugar spectrum can
hardly be attributed to the OH group. In Puccianti's curves2 the alco-
hols show depressions at i-5/x,, but they are not very sharp, although
he used a larger dispersion and a thicker cell. As a whole, the question
is in an unsettled state with the results obtained by the method of
absorption spectra indicating the presence of a molecule of water of
crystallization in cane sugar which has heretofore been considered to be
composed entirely of water of constitution.

FIG. 2oa. — Cane sugar.

1 Investigations of Infra-red spectra, Washington, 1905.
3 Puccianti : Nuovo Cimcnto, II, p. 241, 1900.

WATER Of CRYSTALLIZATION. 37

MALTOSE (GaHaOu+H«O) ; LACTOSE (CuHaOu+HaO). (Fig. 21.)

These two compounds were melted between plates of rock salt, dried
over P2O5 and the edges sealed with Kohtinsky cement. The thickness

60%

2345
FIG. 2i.— Maltose (a) ; Lactose.

8//

5678
FIG. 22.— Raffiuose.

10

13 a.

INFRA-RED TRANSMISSION SPECTRA.

of the films was about o.i mm., and in the case of lactose the color was
slightly tinged, due to decomposition. The water bands are present as
noted elsewhere. The region from 3 to 4 p. is complex, with the maxi-
mum at 34ft.

RAFFINOSE (Ci8HaO
(Amorphous solid melted between rock salt; t = o.2 and 0.05 mm. Fig. 22.)

This sugar melts very easily and forms perfectly transparent solid
films. The transmission curve shows all the water bands, viz, 1.5, 2,

3, 4.75, and 6.05^. The
band at 3 p. is quite dis-
tinctly resolved into the sec-
ond component lying in the
region of 3.3 p. The trans-
parent region at n p. is just
the reverse of the alcohols,
which have an absorption
band at 11.3 /*.

GUM ARABIC (2C8H10OB+HaO).
(f = 0.1805. Fig. 23.)

Two samples were exam-
ined. The first was a film
of the gum dissolved in
water and placed between
two plates of thin micros-
cope cover-glass, which was
kept over P2O5 for several
days. The second was the
combination of two films
6//, which were spread over
cover-glass and dried over
P2O5 for several days. The

cracked and peeled off, so that it was necessary to place two facing
other. The water bands are of course visible, while the region at
is complex.

Fio. 23. — Gum arable.

films
each
3.5 fj.

RocHlXue SAi,T(C«H«O«KNa+4HaO).
(t— 1.27 and o.i mm. Fig. 24.)

Curve a shows the transmission through a transparent crystal section,
while curve b is for a solid film melted between two thin plates of
cover-glass. In the latter case the film was flaky, which decreased the

WATER OF CRYSTALLIZATION.

39

transparency. The crystal section was examined with difficulty because
of its low melting-point. Curve c shows the band at 4.75 yu, magnified
10 times. The water bands at 1.5, 2, and 4.75 //, are very apparent, while
the one at 3 /* is obliterated, as with the sugars.

BRUCINB (C28H2eNa
(Thickness: 0 = 0.05; b=i.i$ mm; c = o.i5 mm. Fig. 25.)

Brucine belongs to the alkaloids, which have a complicated and
unknown structure. As will be noticed from the curves it is very

50

40

C.

o
'w

30

c

<D

20

10

a

0/2345

FIG. 24.— Rochelle salt.

transparent to contain 4 molecules of crystal water. The 6 p. band is
the only one that is conspicuous, while the strong one at 3 p. is entirely
absent. The water is easily expelled at a low temperature. Curve b
is for a thick layer of amorphous material melted on a plate of rock
salt. It shows the general carbohydrate band at 1.7^, and the 3.43 /i
band characteristic of compounds rich in CH2 or CH3 groups. For
curve a the film was made by placing the solid powder between two
plates of rock salt, which were heated on a metal plate, and as soon as

4O INFRA-RED TRANSMISSION SPECTRA.

melting began, the rock salt, containing the film, was removed. In the
same manner the film for curve c was obtained, but here it was not
permitted to melt entirely, so that the film was only semi-translucent.
Nevertheless, it does not show the water bands at 1.5 and 3 /x. It is
difficult to conceive how all the water was expelled in these two cases.
Curve d is curve c magnified 10 times at 3 p..

Out of a list of 32 substances examined containing water, generally
in the form of water of crystallization, this is the first exception to the

90%

6769

FIG. 25.— Brucine.

rule tentatively assumed in this work. The second exception thus far
considered is cane sugar, which is not supposed to contain water of
crystallization. In fig. 40 Kieserite (MgSO4-)-2H2O) is to be noticed.
It has water bands at 1.5, 3, 4.55, and 6/u, the latter being complex
on account of the SO4 groups. In fig. 42, cadmium sulphate QCd
SO4-|-8H2O) is also to be noticed, since it contains the water bands at
1.5, 2, 3, and 4.6/4.

GROUP II: MINERALS CONTAINING WATER OF CONSTITUTION.

The first part of this group contains minerals in which the oxygen
and hydrogen are present in the form of the univalent hydroxyl radical.
OH, and are known as hydroxides. The hydroxides when heated also
yield water, but it is a characteristic that they must be strongly heated
before they are decomposed. From previous work, already noticed, we
would expect an absorption band at 3 /* for hydroxides.

WATER OF CONSTITUTION.

MANGANITE, MnO(OH).
(Orthorhombic; ground parallel to m; opaque; t = 0.08 mm. Fig. 26.)

As will be noticed in curve a of fig. 26, manganite has no sharp bands
throughout the whole region to 9^. There is a light indication of a
band at 3 ^ and another at 6.2 /*.

23*56789

FIG. 26.— Mauganite (a) ; Gothite (b}\ Bauxite (c).

10

GOTHITE, FeO(OH).
(Section ground parallel to b; £ = 0.15 mm. Curve b, fig. 26.)

The appearance of this mineral is opaque black, metallic luster, and
as a whole would not impress one as being quite transparent to 9 /A.
It transmits about 20 per cent of the energy throughout this whole
region, which is in marked contrast with the almost complete opacity
of iron. The 3 ^ band is well marked, while a second one occurs at 9 //..
This specimen was kindly presented by the late Prof. S. L. Penfield.

30

C

IT,

to

7

V

4-56
FIG. 27.- Turquoise (a) ; Lazulite.

BjU

BAUXITE (AUO(OH)O.
(Chocolate-brown color; / = o.u mm. Curve c, fig. 26.)

In this mineral the 3 /* band is wide and complex. There are other
bands at 4.7, 5.8, and 7/t.

42 INFRA-RED TRANSMISSION SPECTRA.

TURQUOISE (AlPCuAKOH^+H-O).
(Massive, light green; subtranslucent in section of o.i mm. Curve a, fig. 27.)

This mineral is supposed to contain a molecule of water, although
Miers writes the formula without it. The OH groups will cause a
band at 3 /u,, and in general the substance is opaque so that it is difficult
to decide that question.

The region at 3 p. is complex, with the maximum at 3.3 p.. There are
other bands at 5.1, 5.3, 5.6, 6.3, and 7.6 p, all of which are small. There
seem to be no bands belonging to PO4.

LAZULITE, (MgFe)AU(OH)i(PO)2.

(Massive; blue by transmission; ^ = 0.19 mm. Curve b, fig. 27.)
The 3 ju, band is prominent, while a new band appears at 4.3 /*.

/ 2 * s G 7

FIG. 28.— Hydrargillite (a); Diaspore (*) ; Datolite (c).

9JU

HYDRARGILLITE (A1(OH)3).
(Stalactitic mass; t = o.2 mm.; subtranslucent. Curve a, fig. 28.)

The section obtainable was only about 6 mm. long, which did not
cover the entire slit; hence the capacity is not quite so great as it
appears.

There is only one large band, at 3 /x, which is no doubt due to the
OH radicals.

DIASPORA A1O( OH).
(Cleavage piece parallel to b; t = o.iS. Curve b, fig. 28.)

The specimen was transparent, with a few patches that were trans-
lucent. The specimen came from Professor Penfield. The band at 3 /A
is wider and not so well defined as in the preceding. There are smaller
absorption bands at 1.9, 5.8. and 6.7 /*.

WATER OF CONSTITUTION.

43

DATOLITE (Ca(BOH)SiCu).
(Section ground to 0.23 mm. Curve c, fig. 28.)

In this curve the OH band at 3 p. is shifted to 2.8 /*. The other bands
present are at 2, 3, 5.3, and 6.3 /*. Although the specimen was full of
cracks, it is rather transparent for a boro-silicate.

AZURITB (2CuCO3.Cu(OH)2).
(Translucent; section parallel to c; £ = 0.35 mm. thick. Fig. 29.)

The section was ground from a group of crystals all of which were
apparently piled parallel to c. There are absorption bands at 2.5, 3.05,
3.58, and 4.05 ju. and complete opacity beyond 5.5 /*.

The 3.05 /A band is shifted slightly from the position common to
substances containing OH groups.

a

4-56789

FIG. 29. — Brucite (a and b) ; Azurite.

<0

BRUCITE (Mg(OH),).

(From Woods Mine, Lancaster County, Pennsylvania. Foliated; massive; lamina
parallel to c; { = 0.05 and t = o.ig mm. Fig. 29.)

As mentioned elsewhere, the fact that brucite did not show an
absorption band at 3 //,, which is the maximum of the hydroxyl group
in a series of organic compounds previously studied, made a reexamina-
tion highly desirable. Of course, as will be noticed in curve b, by
straining matters one might have concluded that the band at 2.5 /*,
which evidently is complex, containing the 3 /* band. In the present

44

INFRA-RED TRANSMISSION SPECTRA.

examination the cleavage section was perfectly transparent and much
thinner than the previous one. Furthermore, the dispersion is greater,
so that as a whole this large band is resolved into three components with
maxima at 2.5, 2.7, and 3^,. Beyond this region there are no marked
bands, while beyond 12 ^ there is complete opacity. The curve shows
depressions at 5, 7, 7.7, 8.2, 9.7, and 10.8 ju, but as a whole the curve
is conspicuous for the complete absence of sharp absorption bands,
except at 2.7 p, and is similar to glass and the micas.

PORTLAND CEMENT.
(t — 0.17 mm. Fig. 30.)

The specimen examined was ground thin from a piece made for
tensile-strength tests, and, in making it, was mixed with 21 per cent of
water. The chemical constitution of Portland cement is unknown.
According to Le Chatlier1 the action of water causes the formation

I a 3 4- 67

FIG. 30.— Portland cement.

of a hydrous silicate (2(SiO2CaO)5H2O) and a hydrous aluminate
(Al2O34CaO.i2H2O).

From the present curve it will be noticed that the substance is very
opaque, with a large absorption band at 3 p, which would seem to indi-
cate hydroxyl groups. From the curve of mellite and alum it appears
that Portland cement is too transparent beyond 3 ^ to contain 17 mole-
cules of water of crystallization, as indicated in the aforesaid formula?.

MANNITE (CoH»(OH)e).
0 = 0.05. Curve a, fig. 31.)

This substance is a hexahydric alcohol. The film examined was a
semi-transparent (due to crystallization) solid, between rock-salt plates.
The transmission curve has the general characteristics of the alcohols,

JLe Chatlier: Annales des Mines, Feb., 1888.

WATER OF CONSTITUTION.

45

which have a wide band from 3 to 4 /A and a transparent region from
4-5 to 5-5 /"-, followed by opacity beyond 6/x,.

It is known that for alcohols the large band in the region of 3 /A
is the composite of two bands situated at 3 and at 3.43 /A.

CHLORAL HYDRATB (CC1SCH(OH)2). (Curve b, fig. 31.)

The constitution of this compound is uncertain. For the present
examination the crystals were melted between plates of rock salt, and
placed over P2O5 to dry and recrystallize. The transmission curve
shows ail the water bands ; although it is not considered water of
crystallization, curve c shows the region at 3 /* after the film stood over
P,O5 for six days and had crystallized into an opaque mass. Chloral

to

3 4 S 6 7 8

FIG. 31.— Mantiite(a); Chloralhydrate.

hydrate has the following bands: 1.5, 2, 3, 3.2, 3.7, 4.7, 5.1, 5.7, 6.2,
7.2, and 8 p. Curve c is b after crystallizing several days.

In the foregoing dozen minerals containing hydroxyl groups there is
only one, or possibly two, viz, manganite and datolite, which appear to
be exceptions to the rule that hydroxyls have a band at 3 p. It will, of
course, be understood that this decision is also based upon previously
studied compounds referred to in the text.

Several other minerals in which the constitution is less understood,
but which are supposed to contain hydroxyl groups, for example ser-
pentine and epidote, will be considered later on, where it will be found
that the 3 /n band is also very evident.

46

INFRA-RED TRANSMISSION SPECTRA.

PREHNITE (H2Ca2Al2(SiO4)s).

(Stalactitic; light green, translucent, crystalline mass; £ = 0.78 mm.

Curve a, fig. 32.)

We have now to consider a series of hydrous silicates in which the
oxygen and hydrogen are supposed to exist as such in the molecule. It
will be noticed that the SiO2 radical does not give sharp bands that are
in a fixed position — e. g., at 2.9/x,.

Prehnite has bands at 2.9, 4.15, 5.96, and 7.3/4, but none coincide
with the water bands. It so happens that in this mineral the bands
coincide very closely with those found in quartz.

i 2 4- 5 6

FIG. 32. — Prehnite (a); Hydronephelite.

(HNa2Al2(SiC>4)s+3H2O).

(From Litchfield, Massachusetts. Massive dark-gray material ; subtranslucent in

section of 0.12 mm. Curve a, fig. 32.)

This substance is supposed to contain water of crystallization, but the
general outline of the transmission curve does not show it. Possibly it
is an hydroxide. The general appearance of the material did not appear
to warrant a further inquiry into this apparent exception to the rule.
There is a wide band at 2.9 ^. Water is expelled at a high temperature.

PECTOUTE (HNaCa2(SiO3)3).

(Translucent sections; / = i.o and 0.25 mm. Curves a and c, fig. 33. From Bergen

Hill, New Jersey.)

Curve a is for a monoclinic prismatic cleavage piece, while curve c is
for a mass of needles all of which came from the same piece. The two

WATER OF CONSTITUTION.

47

do not give coincident results, but there being no bands near those of
water, it did not seem worth while to make further inquiry.

CHLORITOID (H2(MgFe)Al2SiO7).

(Variety, masonite; cleavage piece parallel to c; transmits blue-green; t = o.2$-

Curve b, fig. 33.)

This mineral is very opaque to infra-red radiation. There are sev-
eral small absorption bands, viz, at 2.3, 2.6, 3.3, 5, 6, and 6.3 /*, as in the
preceding. The point of interest is the absence of bands at 1.5 and 3 /*.

/ 234-567 6JU

FIG. 33.— Pectolite(a); Chloritoid

CUNOCHLORE (HsMgsAlaSiaOis) J PfiNINlTS (HsMgFe)6Al2Si8Oi»).

(Clinochlore: t = o.o8. Peninite: curve b, t = o mm. Fig. 34.)

These two minerals belong to the chlorite group, which is related to
the micas, to be noticed presently. These minerals contain about 12
per cent of water, which is given off at a high temperature. Their
transmission curves are unusually similar to those of the micas. There
are absorption bands at 2.9, 5.9, 6.3, 7.1, and 7.8^. The frequently
recurring bands at 2.9 and 5.9 p, with the metallic reflection band at 8.5
and 9.02 fi, remind one of harmonic series.

It may be that these bands belong to a slowly conveying spectral
series, if not a harmonic series, for the first band oscillates between the
values 2.8 and 2.95 /*, while the next band shifts from 5.6 to 5.9 /A, as
will be noticed in the micas.

As already mentioned, the constitution of many minerals is still doubt-
ful. Clinochlore is one of these. From investigations made by Clarke
and Schneider, it was inferred that the hydroxyl groups MgOH and

48

INFRA-RED TRANSMISSION SPECTRA.

A1(OH)2 are present.1 In the present examination there is no indica-
tion of a band at 3 //,, which is a characteristic of the OH radical.

0

FIG. 34.— Clinochlore(a); Peninite.

TOURMALINE

(Cut parallel to optic axis. Fig. 35.)

These curves are due to Merritt (loc. cit.), and show the variation in
transmission for the ordinary ray (plotted o-o-o), for the extraordinary
(x-x-x), and for unpolarized light. The important absorption bands
are at 1.28 and 2.82 /«,. The latter is to be noticed in considering the
effect of SiO2, SiO3, and SiO4 groups.

MICA; MUSCOVITE, HiKAU(SiO«)»; BIOTITE, (H,K)i(Mg,Fe)«AU(SiO«),.

(Cleavage parallel to c; curve a, muscovite; * — 0.02 and 0.04 mm; curve b,

biotite, £ = 0.03 mm. Fig. 35.)

The constitution of the micas is involved in a greater or less degree
of uncertainty. They are all silicates of aluminium, and either K, Na,
Li, or of Fe and Mg. All the micas yield water upon ignition, but it is
uncertain whether this water of constitution is due to the presence of

1 Clarke and Schneider : American Journ. Sci., 40, 405, 1890.

WATER OF CONSTITUTION.

49

H or OH. There is no deep, wide absorption band at 3 p., so that,
judging by the present method of examination, there can not be any
hydroxyl groups present.

Muscovite has a deep, narrow band at 2.85 /*, and smaller bands at
J-9> 3-6, 5-6, 5.9, 6.3, and 7.1 /*. Biotite has small bands at 2.8, 5.9, 6.2,
and 6.7 /z, and indications of bands at 5.6 and 7 to 8 ju,. It has also a
band in the visible spectrum.

Both micas have a large opaque region at 9 to n, followed by a
transparent region at 12 p, beyond which there is again complete opac-
ity. Rubens and Nichols1 have found metallic reflection bands at 8.32,

J 4 5 6 9 9 10 II

FIG. 35.— Muscovite (a) ; Biotite (b) ; Tourmaline (c).

a

9.38, 18.40, and 21.25 p. It will thus be noticed that the bands located
by reflection and by transmission do not coincide, as is to be expected,
since we are in regions of anomalous dispersion.

The spectrum energy curves, found by the writer2 using a rock-salt
prism, and a radiometer having windows of the same material, never
showed a sharp depression at 2.8 p., while those found by Stewart,8
using a fluorite prism and a radiometer having an inner window of
muscovite mica and an outer one of fluorite, contained a conspicuous
depression in this region. The discrepancy was explained on the

Rubens and Nichols: Ann. der Phys. (3), 60, p. 418, 1897.
- Nichols and Coblentz : Phys. Rev., 17, p. 267, 1903.
8 Stewart : Phys. Rev. 13, p. 261, 1901.

4— c

5O INFRA-RED TRANSMISSION SPECTRA.

assumption of a larger dispersion, and a longer column of air traversed
by the rays, which would intensify the CO2 band at 2.75 //,.

From the muscovite curve, however, it will be noticed that the air
(CO2) band would be intensified by that of mica at 2.85 /*.

SERPENTINE (H4(Mg,Fe)3SiO,).

(Massive; subtranslucent ; curve a, t = 0.225 mm; curve b, f = o.o8 mm. Fig. 36.

From Montville, New Jersey.)

In serpentine the water is chiefly expelled at red heat. According to
Miers,1 talc and serpentine are to be regarded as basic and not hydrated

10

0 I 23456789

f

FIG. 36.— Talc (c) ; Serpentine.

silicates, since they part with their water only at a high temperature.
From the curve of serpentine, which shows a large absorption band at
3 /t, if the results from the present method of examination are to be
trusted, it would appear that there are hydroxyl groups present.

From the results obtained by Clarke and Schneider2 it was inferred
that the Mg is present as MgOH, and hence the formula is written
H3(MgOH), Mg2(SiO)2. It has also been written with two mag-

1 Miers : Mineralogy, p. 439.

2 Clarke and Schneider : Amer. Jour. Sci., 40, p. 308. 1890.

WATER OF CONSTITUTION.

nesium hydroxyl groups. Serpentine has absorption bands at 1.4, 3, 5,
6.6, 74, 8.1, and 8.5 /*.

(Subtransparent; f = 0.06 mm. Curve c, fig. 36.)

In talc, water is expelled at red heat ; from its stability with acids it is
considered an acid metasilicate. From the investigations of Clarke and
Schneider (loc. cit.) no hydroxyl groups were inferred, and the trans-

IOO°fo

10

0 I ? 3 4 5

/

FIG. 37.-Epidote.

mission curve does not show a band at 3 ju,. Talc has absorption bands
at 5-6, 5-95> and 7.15/4.

EPIDOTB (Ca,(A10H) (Al,Fe).(SiO).).

(Cleavage piece parallel to c for curve a, t = o.2i; section perpendicular to axis

for curve b, t = o.og. Fig. 37.)

The chemical formula of this mineral is variously written, and it may
contain hydroxyl groups. The iron may be present as FeOH.

52 INFRA-RED TRANSMISSION SPECTRA.

In the present examination curve a is for a section transmitting a
brown color, while curve b was light green. Both show the hydroxyl
band at 3 /*,, while additional bands occur at 4.3, 4.7, 5.0, 5.3, 5.6, 5.9,
6.6, and 7.4 /u,, many of which are in common with those of quartz.

SODIUM METAPHOSPHATE (NaPo3).
(Fused transparent glass; t = 0.37. Fig. 38.)

This substance was examined in connection with the following phos-
phates to learn the behavior of phosphorus in a compound. There is a
wide absorption band at 4 and a second at 6 p..

10

0 I Z 6 d S to 14- U.

FIG. 38.— Sodium metaphosphate (a) ; Meta phosphoric acid (b) ; Ortho phosphoric acid (c).

META PHOSPHORIC ACID (HPO3).
0 = 0.25 mm. Curve b, fig. 38.)

This substance comes in transparent cylindrical sticks like KOH. It
was melted between a plate of rock salt and one of glass. The latter
became loose on cooling, and was removed. The compound is too
opaque for examination beyond 3 /A.

ORTHO PHOSPHORIC ACID
(t = 0.12 mm. Curve c, fig. 38.)

This compound is formed from HPO,, and also from P2O5, standing
in moist air.

SULPHATES.

53

In both cases the curves b showed the water bands at 1.5 and 2/t,
beyond which there was complete opacity.

GROUP III : MISCELLANEOUS COMPOUNDS.

It has been noticed elsewhere that in the examination of selenite the
large absorption band at 4.55 /* is shifted, and too deep to belong to
water ; also that it was suspected to be due to the SO4 groups of atoms.

The band has further been noticed in discussing selenite and anhy-
drite (fig. 3), and in thaumasite and blodite (fig. 13). Under the
present heading will be discussed the absorption spectra of simple sul-

80%

E
•n

c

0/234567

FIG. 39.— Barite(a); Glauberite.

phates formed by the combination of a metallic oxide and sulphur
trioxide, all of which show a band at 4.55 p., while several have another
band in common at 6.5 //,.

SULPHATES.
BARITE (BaSCy).

(Orthorhombic; section parallel to c; translucent; ^ = 0.25. Curve a, fig. 39.

From Cheshire, Connecticut.)

The barite curve is quite opaque, due in part to numerous cracks.
There are bands at 3, 4.6, 6.2, and 6.5 p. Konigsberger's curves show
that the band at 4.6 ju, is wide, and shifts for different directions of vibra-
tion of polarized light.

54

INFRA-RED TRANSMISSION SPECTRA.

GLAUBERITE (Na2SO*CaSOO.
(Monoclinic; cleavage section parallel to c; transparent ; t = 1.26. Curve b, fig. 39).

This is a mixture of two sulphates, of which the CaSO4 band at 3.2
and 4.55 /A have already been noticed. Glauberite has bands at 3, 4.55,
5.6, and 6.2 /*.

THENARDITE (Na2SO4).

(Borax Lake, California. Orthorhombic ; ground parallel to c; f = 2.2 mm.

Curve a, fig. 40.)

This mineral was too brittle to grind successfully. It shows large
bands at 3.1 and 4.55 p.

0/234567

FIG. 40. — Theuardite (a) ; Kieserite.

KlESERITE(MgSO4).
(Stassfurt Mines. Massive; subtranslucent ; ( = 0.22 mm. Curve b, fig. 40.)

This mineral, like selenite, shows both water and SO4 bands. It is
very opaque. The water bands at 1.5, 3, 4.55 (complex with SO4),
and 6.05 p are prominent. There are two small bands at 1.8 and 7^,
respectively.

CELESTITE (SrSO<).

(From Lampasas, Texas. Cleavage parallel to c; transparent; t = 0.67 mm.

Curve a, fig. 41.)

Celestite has absorption bands at 3.2, 4.55, and 6.4/1,. The 4.55 /u,
band is conspicuous for its depth.

ANGLESITE (PbSO<).

(Monte Poni, Sardinia. Translucent; section parallel to m; ( = 0.7 mm.

Curve b, fig. 41.)

The section used was not quite long enough to cover the slit. The
lack of transparency was in part due to the fact that on account of its

SULPHATES.

55

brittleness the section was not highly polished. The band at 4.55 ]«. is
complex, with a deeper one at 5 p. The same is true of the 6.4 /u, band,
which is complex, with a stronger band at 6.7 /A. There are slight
depressions at 1.9 and 3.2 p,.

CADMIUM SULPHATE
(Curves a, b, c, d, fig. 42; t = o.g& and o.i mm.)

This was an artificially grown crystal, and was perfectly transparent.
As a whole, it is very opaque to infra-red radiation. The band at 3 p is
complex, as is also the one at 4.6^, which is shown in curve d, the
vertical scale of which is magnified 10 times. The water bands at 1.5
and 2 /* are well defined, considering the opacity of the substance.

70%

23456
FIG. 41.— Celestite (a) ; Ans;lesite.

SULPHURIC ACID (H2SO«).
(Curve e, fig. 42.)

The great opacity of this substance makes it difficult for examination.
A film, 0.08 mm. in thickness, was pressed between two plates of cover-
glass, and stood over P2O5 for several days. It shows no water bands,
while the trend of the curve at 4.5 p makes it impossible to decide on
the question of the SO4 band. There is a band at 3.6 /* which is broad
and shallow.

It is difficult to decide whether the frequently recurring bands at 6.2
and 6.5 fj. are due to SO4 ; but it is a fact that they are very conspicuous

100%

56 INFRA-RED TRANSMISSION SPECTRA.

in the sulphates examined, especially for K, Ba, Ca, and Pb, which have
a common band at 6.5 /*,.

The sulphates of the metals K, Rb, and Cs have been compared by
Tutton,1 who has shown that both as regards crystalline form, specific
gravity, thermal expansion, and corresponding refractive indices the

Rb salt lies between the K
and Cs salts. These mono-
valent elements occupy con-
secutive positions in the
even series of Mendeleef's
table. The next even series
is Ca, Sr, and Ba, the trans-
mission curves of which are
before us. The following
odd series contains Mg,
which just precedes Ca and
Cd, which lie between Sr
and Ba. In other words,
the elements lie in the order
Mg, Ca, Sr, Cd, Ba, while
the maximum of the ab-
sorption band occurs in the
order 4.5, 4.55, 4.6, 4.6,
4.63 p. Whether this is a
true shift, with increase in
molecular weights, needs
further examination. The
data presented are certainly
very suggestive of a real
shift, with increase in mole-
cular weight of the metal. This same shifting of the maximum, with
increase in molecular, will be noticed in the reflection curves of SrSO4
and BaSO4 ; in the former the maxima are at 8.2, 8.75, and 9.05 /j., while
in the latter the maxima are shifted to 8.34, 8.9, and 9.1 /A.

Drude2 has shown that in the ultra-violet the absorption band is due
to the sympathetic vibrations of particles which have a charge and mass
identical with the "ion" (or "corpuscle"), while, in the infra-red, the
absorption bands are due to particles which have a mass of the order
of magnitude of the molecule. From this standpoint one would expect

0123

FIG. 42.— Sulphuric acid («) ; Cadmium sulphate.

1 See Miers, Mineralogy.

2 Drude: Ann. cler Phys. (4), 14, p. 677, 1904.

MISCELLANEOUS COMPOUNDS.

57

to find a shift of the maximum toward the longer wave-lengths, as we
increase the atomic weight of the element which is attached to the
radical.

MISCELLANEOUS COMPOUNDS.

POTASSIUM BICHROMATE (K2CrO7).
(Cleavage parallel to m; ^ = 0.85 mm. Fig. 43.)

It was found that this substance furnished beautiful, highly polished
cleavage pieces, which are quite transparent to infra-red radiation. The
curve shows a complex depression at 2.9 to 3.2 /*, and bands at 5.4, 6,
6.65, 7.7, and 8.7 /*.

Porter1 located a reflection band at 10.3 p.

70%

60

50

IV

34567
KIG. 43.— Potassium dichromate.

SO/A

POTASSIUM CHLORATE (KClO8).
(Large transparent crystal; t = 0.47 mm. Curve a, fig. 44.)

The absorption bands in this substance are strikingly similar to those
of potassium dichromate, especially the 5.25 ^ band, which is shifted to
5.4 p. in the latter. There are bands at 3.2, 3.5, 5.25, 6.28, 6.93, and
8.1 /t, beyond which there is complete opacity.

1 Porter : Astrophys. Jour., 22, p. 229, 1905.

58 INFRA-RED TRANSMISSION SPECTRA.

APATITE (Ca5F(PO4)8).

(From Kragerve, Norway. Massive; subtranslucent ; f = o.i2 mm.

Curve b, fig. 44.)

There are no strong bands in this mineral, and none are found to be
common with substances containing the PO4 radical. The bands are
small and occur at 2.9, 3.9, 4.85, 5.85, 6.4, and 6.85 /tt.

01234-56

FIG. 44.— Potassium chlorate (a) ; Apatite.

GARNET (Ca«(Fe,Mg)liAla(SiO«)8).
(f = 3.25 mm. Curve a, fig. 45.)

This garnet was wine-red in color, showing absorption bands in the
yellow-green and in the green-blue parts of the spectrum.

The transmission curve is extraordinary, having a wide absorption
band extending from 1.2 to 2.6 //,, and complete opacity beyond 4.5 ju,.

MONAZITE, (Ce,La,Di)PO4.
(t = 1.04 mm. Curve b, fig. 45.)

This brown-colored mineral was partly transparent to visible rays,
and is equally opaque to infra-red rays. A large band extends from
3.5 to 5.5 p., and opacity beyond 6.5 //,.

MISCELLANEOUS COMPOUNDS.

59

(From Parker Shaft, Franklin, New Jersey. Semi-transparent; f = o.o8 mm.

Curve c, fig. 45.)

The sample was a mixture of calcite and willemite. The section
examined was quite pure willemite. The curve shows a depression at
2.9 /A and bands at 5.9 and 6.75 p..

50°/c

O

2 3 4 S 6

FIG. 45.— Garnet (a); Monazite (d) ; Willemite.

a

COLLODIUM (C,HTO2(ONOa)«).

(f = o.oi mm. Curve a, fig. 46.)

The film of this compound was made by painting a layer of the com-
mercial "collodion" over a rectangular hole, 4 by 10 mm., in a piece of
cardboard. After the volatile solvent had evaporated the film was
dried over P2O5 for a week.

The constitution of this compound is unknown. It will be noticed
that the transmission curve is entirely different from those rich in CH2
or CH3 groups or from those found for the benzine derivatives.

There are absorption bands at 1.65, 2.9, 3.5, 6.05, 7.3, 7.85, 8.7, 9.2,
9.5, 10.0, 10.8, 11.7, 12.3, 12.7, and 13.6^.

AMMONIUM CHLORIDE (NH4C1).
(Translucent; t = o.6 mm. Curve b, fig. 46.)

Section made from a lump of the commercial material. It is very
opaque, with bands at 1.65, 2.2, and 2.95 /x, the latter being in the region
where NH2 compounds and NH3 have a band. Porter (loc. cit.) found
a reflection band at 344/*.

6o

INFRA-RED TRANSMISSION SPECTRA.

so

80

70

eo

50

+0

/o

O I 2 J f 5 6 8 9 /O II /2

FIG. 46.— Collodium (a) ; Ammonium chloride.

10

Z 45

FIG. 47.— Tartaric acid.

6 tO

MISCELLANEOUS COMPOUNDS.

61

TARTARIC ACID (C*H6Oo).
(Cleavage parallel to a. Curve a, fig. 47.)

In fig. 47 curve c is for a film melted and crystallized between thin
cover-glass, £ = 0.09 mm., while curve a is for a cleavage section,
/ = 0.32 mm. Curve b is for a film melted between plates of rock salt.
This compound has the properties of a divalent alcohol and of dibasic
acids. The curves are characteristic of alcohols, having an absorption
band from 2.9 to 4 /*, followed by a transparent region at 4.6 /x,, which is
characteristic of carbohydrates.

to

01?. 4 6 8

FIG. 48.— Phloroglucin (a) ; Malic acid.

PHLOROGLUCIN (C8H8(OH)8+2H2O).

(f = O.O2. Fig. 48.)

This belongs to the group containing water of crystallization.
Apparently it is an exception to the rule. The sample turned light
brown on melting, and it is quite probable that the H2O was partly
expelled. The compound belongs to the phenols, hence one would
expect a band at 3 to 3.2 /*, the latter being the characteristic of benzine.
There are small water bands at i .45 and 2.05 /u,.

62

INFRA-RED TRANSMISSION SPECTRA.

MALIC ACID (C2H3(OH) (CO2H)a).
(f = o.o8. Curve b, fig. 48.)

This compound belongs to the trivalent dibasic acids (CnH2n_,,O6),
and, like the acids previously studied, is very opaque, with wide absorp-
tion bands at 3.5, 5.9, and 7.2 p..

SODIUM BIBORATE (Na2Br4Oi).
(Fused glass, transparent, ground to t = o.i2 mm. Curve a, fig. 49.)

This compound is formed on fusing borax (Na7B4O7), which con-
tains 10 molecules of crystal-water. It is difficult to decide whether or
not the bands at 3 and 4.8 p. are due to H2O. There are other bands at
1.9 and 3.7/1.

90%

80

70

60

' 50

V)

(0

h

30

20

IO

IO to 14- u

3 3 •* -5 6 7 8

FIG. 49-— Sodium biborate (a) ; Silver nitrate.

SILVER NITRATE (AgNOa).
(£ = 1.65 mm. Curve b, fig. 49.)

This crystal was ground, but not highly polished. Its transmission
curve is marked for its sharp bands at 2.7, 3.6, 4.1, 4.78, and 5.65 /x.

MISCELLANEOUS COMPOUNDS. 63

SPHALERITE (ZnS).

(Cleavage piece, t= 1.53 mm.; transparent; slightly yellowish tinge.

Curve a, fig. 50.)

The transmission curve is marked for its extraordinary transparency
from 5 to 12 n, interrupted by slight depressions at 1.6, 11.2, and 13.2/4.
There is a wide band from 2.7 to 3.3/t, and complete opacity beyond 15/1.

SILVER CHLORIDE (AgCl).
(Vitreous, t = o.8 mm. Curve b, fig. 50.)

This is the only known substance for which no absorption bands have
yet been found in the infra-red. It seems to increase in transparency
with increase in wave-length, but is opaque for the "reststrahlen" at 53
and 6 1 p., as one would expect from its analogy to NaCl and KC1.

56 7 B S IO II 12 13 /•* IS

Fio. 50,— Sphalerite (a) ; Silver chloride.

ORTHOCLASE (KAlSi3O8).
(Yellowish tinge; subtransparent ; t = 0.07 mm. Curve a, fig. 51.)

This mineral belongs to the feldspar group. It has bands at 2.85,
4.7, 5.7, and 6.28 p., which are bands of silicates.

Curve c, t = 0.23 mm., is for a cleavage piece from a specimen of
unknown composition, and was simply marked "Feldspar." It has the
general outline of, but the bands are less marked than in, curve a.

AMPHIBOLE (CaMg3(SiO8)i).

(Silky gray color, probably tremolite; ground parallel to m; translucent;
# = 0.07 mm. Curve b, fig. 51.)

This curve is similar to orthoclase, with bands at 2.8, 4.8, 6, 7.4, and
8.2 11. The 6 /A band may be the mean of the two found in orthoclase,
at 5.7 and 6.28 /*,, respectively.

64

INFRA-RED TRANSMISSION SPECTRA.

SO

c
o

in
tn

u)

C

L
h-

3O

10

23456 8

FIG. 51.— Orthoclase (a) ; Amphibole (b) ; Feldspar (c).

Fio. 52.— Oligoclase (a) ; Orthoclase (*) ; Zircon (c) ; Glass (d).

MISCELLANEOUS COMPOUNDS.

OUCOCLASK

(From Bakersville, North Carolina. Cleavage piece; not polished; transparent;

f = 1.25 mm. Curve a, fig. 52.)

This mineral has no marked absorption bands. There is a depression
in the transmission curve at 3.3 /A.

ORTHOCLASE (var. ADULARIA) (KAlSinOs).

(From St. Gothard, Switzerland. Cleavage parallel to c; perfectly transparent;

* = O.I5 mm. Curve b, fig. 52.)

The transmission curve shows bands at 2.0, 3.2, 3.9, 4.8, and 5.6^,
and is in marked contrast with the orthoclase curve.

ZIRCON (ZrSiCu).

(Transparent; f = 3 mm. Curve c, fig. 52.)

This specimen was not long enough to cover the slit. It shows bands
at 2.1, 3.1, and 3.6 /x, and complete opacity beyond 5 /A.

90

80

•o

60

50

JO

10

\

\

\

IZ 13

14

O I ^ 3 <? 5 6 7 8 9 10 II

pIG. 53.— Glass (a) ; Albite (b) ; Bnstatite (c).

GLASS.

(Microscope cover-glass; f = o.og mm. Curve d, fig. 52; curve a, fig. 53.

t = 0.00 1 mm.)

The transparency of glass is very great up to 2 p. The curve shows
a depression at 3.2 /*, and bands at 5.6 and 6.25 /*, which are common to
SiO8 compounds. For this thickness glass is transparent to 8 /*.
5— c

66

INFRA-RED TRANSMISSION SPECTRA.

In fig. 53, curve a, is shown the transmission curve or ordinary soft
glass which has been blown into a bulb of such a thickness that it
showed interference colors. This is the thinnest film yet examined.
There is one large complex absorption (reflection) band extending from
8.5 to 10.7 p., the maximum being at 9.7 p.

ALBITE (NaAlSi8O8).
(/ = o.i4. Curve b, fig. 53-)

This mineral is more transparent than orthoclase (fig. 51, curve a).
The absorption bands at 2.9, 5.7, and 6.3 fi coincide with those of ortho-
clase.

ENSTATITE (MgSiO8).
("Bronzite." t = i mm. Curve c, fig. 53.)

This specimen was a mixture of several crystals. The transmission
curve shows a large absorption band at 1.85^, and smaller depressions
at 2.9 and 5.2 p.

~1

0 / .? 3 •? S 5 89

FIG. 54.— Ebonite (a) ; Rubber.

EBONITE.
(t — o.i mm. Curve a, fig. 54.)

In fig. 54 is shown the transmission curve of ordinary ebonite, which
has been ground thin, so that it was subtranslucent. It is quite trans-
parent to 14 p., with absorption bands at 3.4, 5.9, 6.9, 8.3, 9.1, and 10 p.,
which are in coincidence with bands belonging to carbohydrates.

PARA RUBBER.
(Subtranslucent; t = 0.27 mm.)

In curve b, fig. 54, is given the transmission curve of a thin piece of
sheet rubber. It is very opaque, but shows a large band at 3.3 p..

STIBNITE (Sb2S3).
(Cleavage piece parallel to b; £ = 0.45, 0.98 and 4.9 mm. Fig. 55.)

This is one of the most remarkable substances examined. It has a
metallic luster, which is highly splendent on fresh cleavage surfaces.

MISCELLANEOUS COMPOUNDS. 67

It is, of course, opaque to the visible rays, but, for the thickness used, it
transmits 45 to 50 per cent of the infra-red from 1.5 to 12^. This
great transparency to heat rays seem to be a property of the sulphides,
as will be noticed in sphalerite (ZnS). In this connection is to be
noticed that selenium is also very transparent to heat rays.

From the reflection curve it will be noticed that the above thickness
of stibnite is practically transparent, instead of transmitting only 45 per
cent, as indicated above. From curves a and b, it will be noticed that
there is practically no difference in the transmission, although one is
twice as thick as the other. The greater transparency of the thicker
film at 7 p is not an error in observation, as was found on repeating the

3456 8 10 II

FIG. 55.— Stibnite ; Brookite (rf) ; Rutile (e).

work. In curve c, where the thickness is 4.9 mm., the absorption
(selective reflection ?) band beyond 15/4 becomes quite evident. In
this region the reflection curve suddenly decreases, indicating a region
of anomalous dispersion.

BROOKITE (TiO2).
(t = 0.37 mm. Curve b, fig. 55.)

The specimen examined was subtranslucent, and by transmission was
of a grayish tinge. The curve shows no bands except a depression
at 3.1 ^

RUTILE (TiO2).
(t = 0.2S mm. Curve c, fig. 55.)

The section of this mineral was ground from a long, flat needle, which
transmitted a deep red. It is more opaque than brookite, from which it
differs in crystalline form.

68

INFRA-RED TRANSMISSION SPECTRA.

POTASSIUM NITRATE (KNO»).
(f = o.32. Curve a, fig. 56.)

The section of this substance was made from the commercial material.

It is conspicuous for its large absorption band extending from 6.5 to
8.5 ft, and a slightly smaller band at 12 p..

There are absorption bands at 3.6, 4.0, 4.7, 5.65, 6.5 to 8.5, 9.6, 10.2,
12.0, 13.1, and 14.3 fji. Some of these bands coincide with sharp bands
in silver nitrate. The reflection curve (x-x-x-x) shows a maximum
at 7.15/1.

AMBER.
(t = 0.8 mm. Curve b, fig. 56.)

Amber is a fossil resin. The section examined is too opaque to give
us any information.

-? £ 6 8 9 to II 12

FIG. 56.— Potassium nitrate ; Amber (b).

15 16-fi.

FLUORITE (CaF2).

In fig. 57 are shown the transmission curves of several pieces of
fluorite, examined several years ago, but never published. They are
given to show the difference between a perfectly white fluorite, curve a,
and a greenish tinged variety, curve b. The latter shows a slight
absorption band at i.4/i. In curve a, t = 2$ mm., and in curve b,
t = 4.75 mm., from which thickness it will be observed that if we cor-
rect for reflection, white fluorite is perfectly transparent, while the green
variety is likewise beyond 2.5 p. This transparency has been observed
by Rubens and others, and continues beyond 6 p., where it decreases to
zero at 10 /*,.

From the curves it will be seen that for certain work on radiation the
green fluorites are just as serviceable as white ones, which latter are
becoming difficult to obtain.

MISCELLANEOUS COMPOUNDS.

ROCK SAI.T (NaCl).
(/ = 3.n mm. Fig. 57.)

The surfaces of this cleavage section were slightly blurred from
moisture, and the curve is given to show its great effect in the visible
spectrum, which becomes negligible in the infra-red.

CALCITIC (CaCO8).
(f = o.i4 mm.; cut 70° to axis. Curve a, fig. 58.)

The section used was a small piece, 5 by 6 mm., made for polarimetric
work. It was too short to cover the whole slit, hence the actual trans-

_U)

« 7C

C
ID

U

60

50

,<

FIG. 57.— Rock salt (c) ; Fluorite.

mission is greater than 55 per cent, indicated on the curve. The curve
is remarkable for its transmission of heat-waves. After passing through
the opaque region at 7/x, the section is again fairly transparent to 14 ft.
There are bands at 3.44, 3.93, 4.6, 5.7, 6.5 to 7.5, 8.5, 9.4, 10, 10.7, 11.3,
12.2, 12.5, and 13. 2 /x. The region of great absorption at 3.4 to 3.9,
6-5 to 7.5, and 15 ;«, are closely harmonic.

It will be noticed that the CO, band at 4.28 is lacking, while the 4.6 p.
band, found in CO, is small, from which it appears that CO2, which is

-.,

INFRA-RED TRANSMISSION SPECTRA.

so prominent as a gas, is differently bonded in calcite and magnesite.
The same thing has been noticed in Azurite.

MAGNESITE (MgCO3).
(Massive; subtranslucent; / = o.op. Curve b, fig. 58.)

This mineral is more unusual than calcite. It has the transparent
region of calcite, at 5 u, preceded and followed by almost complete
opacity. The bands at 3.3 and 3.8 /* are probably in common with
calcite.

J

!W\/U

,5 6 7 8 3 IO II

FIG. 58.— Calcite (a); Magnesite.

13

CASSITERITE (SnO:).
(/ = 0.085 mm.; transparent. Curve a, fig. 17.)

This section, when bought, was thought to be mellite; but its trans-
mission curve is so entirely different from what one would expect to
find for substances containing water of crystallization that it was sus-
pected to be another mineral. Specimens of mellite were purchased,
and found to be unusually opaque to heat rays. Upon inquiry from
the maker it was found that they had accidentally substituted cassiterite
for mellite. The fact that it was possible to detect the error would
indicate that the method of analysis is trustworthy. There are small
absorption bands at 1.3, 3, 5.8, 6.6, 7.3, 8.25, 9.7, and 10.4 p..

PART IV

INFRA-RED REFLECTION SPECTRA

CHAPTER I.

INTRODUCTION.

Previous work on this subject has been rather disconnected, and the
present investigation is only preliminary to an extended inquiry into the
subject. Nichols1 compared the reflecting power of quartz with silver,
and found that from the visible to 8 p. only a small amount is reflected,
while at 8.5 and 9.02^ there are two intense reflection bands which
have a reflecting power almost as great as silver. This work was con-
tinued by Rubens and Nichols,2 by using successive reflections from sev-
eral surfaces of the same mineral, who found the location of the maxima
of the selectively reflected rays("Reststrahlen")of a series of substances,
including quartz, mica, rock salt, sylvite, crown and flint glass, sulphur,
alum, and calcite. By using a wire grating they were able to extend
their observations to 61 p., the longest heat-waves yet discovered.

In 1900, Aschkinass,3 using a rock-salt prism, compared the reflecting
power of a series of minerals, including marble, calcite, selenite, and
alum. He found that marble and calcite have strong reflection bands
at 6.7 and 11.4^, while selenite has a band at 8.69^. The 6.67 band
of marble is broad, and no doubt complex ; and in the present work it
will be noticed that the calcite band is also double, with the stronger
component at 6.6 p..

Recently Porter,4 using the method of Rubens and Nichols, found
the selectively reflected rays of a series of compounds for the region
extending to 10 p. By this method he was able to locate maxima which
are very small and would not appear by a single reflection. For
example, potassium ferrocyanide has a series of medium-sized bands
which were found in the present examination of absorption spectra, one
of which also appears in his reflection spectrum of this compound.

APPARATUS AND METHODS.

The apparatus for the absorption work was used in the present exam-
ination. To this end the Nernst heater was replaced by a sliding carrier
containing the reflecting mirrors (fig. 59). The carrier consisted of a
vertical sheet of metal which, by means of cords and pulleys, was moved

1 Nichols : Phys. Rev., 4, p. 297, 1897.

2 Rubens and Nichols: Ann. der Phys. (3), 60, p. 418, 1897.
'Aschkinass: Ann. der Phys. (4), I, 42, 1900.

4 Porter : Astrophys. Jour., 22, p. 229, 1905.

73

74 INFRA-RED REFLECTION SPECTRA.

back and forth in a horizontal direction in a very accurately milled slot.
Two openings, each having an area of 2 by 3 cm., were cut into the sheet
of metal, over which and facing the spectrometer slit were attached the
mirrors. The silver mirror was, of course, permanently attached over
one of the openings in the carrier, and it was a simple matter to attach
the previously prepared reflecting surfaces of the minerals over the
other opening. This brought the reflecting surfaces of the silver mirror
and the mineral into the same plane, and limited the exposed area to the
size of the opening in the carrier. The silver mirror was chemically
deposited upon a plane glass plate, and was new.

Occasionally the reflecting surface of the mineral was simply a cleav-
age plane — for example, selenite and celestite.

The angle of incidence
upon the mirrors was 25°.
Since we are finding the
ratio of the reflecting power
of the mineral to that of

Spectrometer slit

silver, this large angle,
Mineral which was unavoidable on
account of the construction
of the spectrometer, will not
affect the results apprecia-
h eater bly. The values given are

FIG. 59.— Arrangement of apparatus for reflection work, sliglltlv higher than the ab-
Spectrometer iiot shown. n

solute reflecting power, but

the correction is only about 2 per cent1 at the maximum, and since so
much depends upon the condition of the reflecting surface of the min-
eral, the correction has not been made, except for the metals.

The spectrometer slits were 0.3 mm., or about 2" of arc, as in the
preceding work. The radiation from the Nernst "heater" was thrown
upon the reflecting mirrors by means of a mirror having a focal length
of 15 cm. and an aperture of 12 cm.

As mentioned elsewhere, in a majority of minerals the general absorp-
tion increases so rapidly beyond 5 ju, that unless one can reduce the thick-
ness to o.oi mm. it is impracticable to examine them. It is not a diffi-
cult matter to reduce liquid films to this thickness. On the other hand,
to grind a solid to even o.i mm., and then mount it so that it will be free
from glue and free from its support, is not practicable in many cases.
The reflection method is not much more advantageous, because larger

1 Hagen and Rubens: Ann. der Phys. (4), ir, 873, 1903, have found the reflect-
ing power of silver to be about 98.5 per cent throughout the spectrum beyond 2ft,

APPARATUS AND METHODS. 75

surfaces will be required. Hence the present examination is limited to
those minerals which afforded surfaces I by 2 cm. or larger.

Before discussing the results it will be well to recall some of the
known facts in regard to absorption, reflection, and dispersion. In case
of normal dispersion the index of refraction decreases with increase in
wave-length. For a transparent substance this means a decrease in
reflecting power, with increase in wave-length. In the region of an
absorption band the refractive index will be abnormally decreased on
the side toward the short wave-lengths, while it will be abnormally
increased on the side of the long wave-lengths. Consequently the reflec-
tion curve, which is now a function of the refractive index and the
extinction coefficient, will be abnormally decreased on the side of short
wave-lengths and abnormally increased on the side of the long wave-
lengths.

These facts are well illustrated in the present series of curves. The
same is to be observed in Aschkinass's curves. He makes no com-
ments, however, although the minerals were examined in connection
with the question of anomalous dispersion. In comparing the present
curves with the transmission curves it will be noticed that the maxima
do not coincide. This is to be expected, for in the transmission curves
in the region of great absorption and anomalous dispersion the loss of
energy is mostly by reflection, hence the maximum will be abnormally
broadened and shifted. This is well illustrated in the transmission
curve of potassium nitrate, which has a large opaque region extending
from 6.5 to 8.5 ju,. Before examining this region by reflection the
writer expected to find this region to be a complex of several bands, as
in calcite, or at least of one large band with a maximum at about 7.5 /*.
The reflection band, however, is found at 7.18/11, and the asymmetry in
the transmission curve is to be attributed to the slight loss of energy, by
reflection on the side of the short wave-lengths and to a correspond-
ingly great loss on the side of the long wave-lengths.

Before discussing the reflection curves it may be added that the crys-
talline minerals were usually cleavage pieces cut (but not always pol-
ished) in the same direction as for the transmission work. A series of
sulphates will first be noticed.

CHAPTER II.

INFRA-RED REFLECTION SPECTRA OF VARIOUS SUBSTANCES.

SULPHATES.

(CaSO*+H2O).

Cleavage piece1 cut parallel to b axis. The curve a, fig. 60, shows a
uniform decrease in reflection to 7 ju., then an abnormal decrease to 7.8 /*.
The maximum occurs at 8.7^ (Aschkinass, 8.69 /A), and from its lack
of symmetry may be complex. In fact, from what follows one would

70%

6 7 a 9 IO

FIG. 60.— Selenite (a); Barite.

12

expect another band at 9.1 ;«,, which is of frequent occurrence, and is
harmonic with the SO4 band at 4.5 /*.

In a recent research by Koch,2 who used a reflection band as a source
of energy in an interferometer, there are indications that the band of
selenite at 8.7 ju is complex.

1 In the present discussion "cleavage piece" signifies that the natural unpolished
surface was used.

'Koch: Ann. der Phys. (4), 17, p. 658, 1905.

77

INFRA-RED REFLECTION SPECTRA.

BARITE (BaSO«).
(Massive. Curve b, fig. 60.)

The reflection decreases normally from 4 to 6.8 p., then abnormally to
7.8 p.. There are maxima at 8.35, 8.9, and 9.1 ft, followed by a trans-
parent region to 12 p., where the present observations cease. The reflec-
tion appears to increase at 11.5 p..

30

o

+3
u

cr

/o

6789
FIG. 61.— Celestite (a); Kieserite.

10

U

CELESTITE (SrSCu).
(Cleavage parallel to c. Curve a, fig. 61.)

The reflection curve is similar to the preceding, with maxima at 8.2,
8.76, and 9.1 p.. The region beyond g p. is evidently complex, with a
possible maximum at 9.3 p..

*

KIESERITE (MgSO4+2H2O).
(Massive. Curve b, fig. 61.)

This mineral is hydroscopic, and does not take a high polish. The
two small maxima occur at 8.7 and 9.25 p., respectively.

FIG. 62.— Glauberite (a) \ Ferrous sulphate.

GLAUBERITE (NaSCXCaSO*).
(Ground parallel to c. Curve a, fig. 62.)

^i

The maxima in this curve are not well denned, particularly the one
at 9.1 p. and the one at 8.3 p.. The band at 8.7 p, is not so sharp as in
selenite.

SULPHATES.

79

FERROUS SULPHATE (FeSC»4+7H2O).
(Curve b, fig. 62.)

This crystal dehydrated so rapidly that the surface became rough.
There is one well-defined maximum at 9.1 /* which is harmonic with the
SO4 band at 4.55 p.

ALUM (K2S(XA12(SO4)8+24H2O).
(Curve a, fig. 63.)

Aschkinass (loc. cit.) found a small reflection band at 9.05^. The
present examination was made to compare it with that of other com-
pounds having the SO4 radical. The maximum was found at 9.1 /*.

6 6

FIG. 63.— Alum (a) ; Anhydrite.

IO

12/A

ANHYDRITE (CaSO4).
(Curve b, fig. 63.)

Cleavage pieces pasted upon cardboard to form an area sufficiently
large for reflection. In comparison with selenite, it will be noticed that
the large maximum of the latter is here found at 8.6 ju, followed by a
small band at 9.1 p. It seems a little surprising that the latter band
should be so small if it is a harmonic of the large band (of CaSO4) at

4-55 /*•

DIASPORA AIO(OH).

(Cleavage piece parallel to b. Curve a, fig. 64.)

This is the most remarkable mineral yet examined. For the region
up to 8 p. it has no marked absorption bands. Beyond this point condi-
tions are entirely reversed. There is a broad reflection band extending

8o

INFRA-RED REFLECTION SPECTRA.

from 9.4 to 10.3 p,, with maxima at 9.45, 9.8, and 10.2 p,. The region
from 12 to 13 /A is complex, with a maximum at 12.8 p. In addition to
this are sharp maxima at 8.55, 13.8, and 14.6 /A.

QUARTZ (SiO2).
(Curve b, fig. 64.)

Section cut perpendicular to optic axis. The examination was made
to compare the maxima of the silicates studied. The maxima (at 8.48
and 9.02/1,) agree well with those found by Rubens and Nichols, using
a wire grating. The enormous reflecting power of 95 per cent at 9 /*

1 00 of c

3 9 10 II I

FIG. 64.— Diaspore (a); Quartz.

a

is

16JU.

has been found only in quartz, and is slightly higher than found by
others. In the silicates to be noticed presently it will be found that the
bands found in quartz are shifted to new positions for each new com-
pound examined.

A new band which is unusually sharp and narrow was found at
12.5 p.. It is so very narrow that it seems more appropriate to call it a
line instead of a band.

Trowbridge measured the reflecting power of quartz at 9, 10, II, 12,
and 12.78 ju,, and found a rapid decrease to 12 /*, while the last reading is
much higher, which increase he interpreted as being due to the band at

CARBONATES AND SILICATES.

81

20.75 /*• From the present curves, which include more observations in
the same region, it will be noticed that his last observation was in the
midst of a reflection band, beyond which the reflecting power decreases
to a low value as it approaches the 20.75 /x, band.

30%

56789

/•

FIG. 65.— Potassium uitrate (a); Enstatite.

POTASSIUM NITRATE (KNO»).
(Curve a, fig. 65.)

This compound shows a sharp maximum at 7.15^. As was noticed
elsewhere, the great shift of this maximum, as compared with the band
formed in the transmission curve, is due entirely to the great difference
in the reflection on both sides of the transmission curve, which is given
without correction for reflection.

ENSTATITE (MgSiOs).
(Curve b, fig. 65.)

This mineral is a mass of fine crystals. The reflecting power is low ;
there is a small maximum at 9.1 /*.

o:

789/0
FIG. 66. — Calcite (a) ; Magnesite.

//

CALCITE (CaCO»).
(Curve a, fig. 66.)

Reflection from a natural cleavage face of Iceland spar. The trans-
mission curve shows a wide complex band extending from 6 to 8/t.
6— c

82

INFRA-RED REFLECTION SPECTRA.

The reflection is complex, with maxima at 6.6 and 6.85 p,. Aschkinass1
found a band at 6.69 p. for Iceland spar, while his curve for marble is
flat, and evidently complex.

MAGNESITE (MgCO3).
(Massive. Curve b, fig. 66.)

The band of selective reflection is similar to that of calcite, and con-
sists of two maxima at 6.5 and 6.8 p., respectively.

60 <%

50

V

it-
CD
CE

20

10

\

6739
FIG. 67.— Microcline (a) ; Talc.

IO

II

MICROCLINE (KAlSiaOs).
(Cleavage piece parallel to c; polished. Curve a, fig. 67.)

The reflection band from 9 to IO/A is complex. There are sharp
maxima at 8.85 and 9.95 p., with a probable band at n /*.

TALC

(Massive. Curve b, fig. 67.)

There are sharp maxima at 9.05 and 9.75 /*. Beyond this point to
12 p. there is great transparency.

* Aschkinass, loc. cit.

MICAS. 83

MUSCOVITE MICA (H2KAl3(SiO08).
(Curve a, fig. 68.)

Cleavage piece, clear white; a thick piece shows a slight brownish
color. The micas are the only minerals examined in which the maxima
do not agree with those found by previous observers. From the fact
that quartz was examined at the same time, and no disagreement was
observed with other observers, the discrepancy is probably due to the
kind of mica used.

Rubens and Nichols (loc. cit.) found reflection bands at 8.32 and
9.38 /*.

600/0

6

769
FiG. 68.— Muscovite (a); Biotite.

!O

II

In the present curve there are bands at 9.2, 9.7, and 10.2 p.. The
absence of the 8.32 /* band was also noted by Rosenthal,1 whose reflec-
tion curve does not show any sharp maxima.

BIOTITE MICA, (H,K)2(Mg,Fe)2Al(SiOO».
(Curve b, fig. 68.)

The biotite curve has maxima at 9.3, 9.6, and 9.85 p., the latter being
unusually sharp. The reflecting power of the micas is high in compar-
ison with some of the silicates. The position of the maxima is variable,

1 Rosenthal: Ann. der Phys. (3), 68, p. 791, 1899.

INFRA-RED REFLECTION SPECTRA.

and is no doubt to be attributed to the variable composition of the
mineral.

DEWEYUTE (H4Mg*(SiOO«+4HaO).

(Curve a, fig. 69.)

This mineral has a complex region of selective reflection extending
from 9.5 to 1 1 /A. There are maxima at 9.65 and 10.5 /*.

o ^0

IO

CL,

56 789/0

FIG. 69.— Deweylite(a); Apophyllite.

APOPHYLLITE

(Cleavage parallel to c, fig. 69.)

There is the usual region of reflection common to the silicates. There
are maxima at 9.15 and

SODIUM METAPHOSPHATE (NaPO3).
(Curve a, fig. 70.)

There is a single band at 8.0 ^ which is harmonic with the band at 4 /*.
In other regions the reflection compares with glass and the vitreous sub-
stances examined.

6 733 IO //

FIG. 70. — Sodium metaphosphate (a); Tourmaline.

TOURMALINE

(Curve b, fig. 70.)

The reflection was formed from a natural crystal face. There are
maxima at 7.5, 8.0, 9.2, 9.7, 10.2, and 10.7 //,.

SILICATES.

SERPENTINE

(Curve a, fig. 71.)

The reflecting power of this mineral is low. The maxima are at 9.7
and 10.5 /A.

COLEMANITE (Ca2B8Oii+5H2O).
(Cleavage piece parallel to b. Curve b, fig. 71.)

From the transmission curve it will be noticed that this mineral is
very opaque. Yet it has no metallic reflection bands before arriving at
7 p. Its reflecting power of only i per cent at 4 /A is unusually low.
There are maxima at 7.3, 7.6 ( ?), 9.4, 10.6, and 11.2 p..

6 8 9 IO n /<?

FIG. 71.— Serpentine (a); Colemanite.

AUJITE (NaAlSi8O8).
(Curve a, fig. 72.)

Section ground parallel to a cleavage plane. The curves of soda and
potash feldspars are unusually similar, as was found for transmission.
The maxima occur at 8.7, 9.7, and 10 /z.

6 B 9 10

FIG. 72. — Albite (a) ; Orthoclase.

I3ju.

ORTHOCLASE (var. ADULARIA) (KAlSisOs).
(Curve b, fig. 72; cleavage parallel to c.~)

The first maximum is shifted to 8.8 p., while the second is a broad
band, with a maximum at 9.8 p..

86

INFRA-RED REFLECTION SPECTRA.

AMPHIBOLE (CaMg3(SiO»)«)-

(Curve a, fig. 73.)

The general trend of the reflection curve is similar to quartz. There
are maxima at 8.7, 9.45, 10.03, and 1 1 P-> tne first and last being small
and not well defined.

5O°/o

56 769/0

FIG. 73. — Amphibole (a) ; Sodium biborate.

SODIUM BIBORATE

(Curve b, fig. 73.)

This substance is a glass made by fusing borax. The surface was
well polished ; nevertheless the reflection bands are weak. The maxima
are wide and occur at 7.5 and 10 /*. From a comparison with the trans-
mission curve it appears that the great opacity of this glass is due to
some other property than to bands of selective reflection.

7 B 9 IO

FIG. 74.— Datolite(a); Hydrotalcite.

/3/J.

SILICATES.

DATOLITE (Ca(BOH)SiO«).
(Curve a, fig. 74.)

The reflection curve shows a series of small, sharp maxima at 8.8,
9.2, 9.5, 10, and 10.8 p. The reflection was from a natural crystal face.

HYDROTALCITE (Mg3Al(OH),+3HIO).
(Curve b, fig. 74.)

The results from this mineral are rather disappointing. There is but
one reflection maximum at 9.9 /*, and this one is not very large. The
transmission curve shows great opacity.

30 <%

c
_o
Y>

u

/o

456 e 9 /O II 12

FIG. 75.— Natrolite (a) ; Apatite.

N ATROLITE ( Na2 AUS i3O10 + HjO ) .
(Curve a, fig. 75.)

Reflecting face ground parallel to m. There are maxima at 9.05,
9.5, and 10.05 P-> beyond which point the reflecting power remains
unusually high to 12/1.

APATITE (Ca5F(PO4)s).
(Curve b, fig. 75.)

The reflecting power gradually decreases to 7.5 /*, beyond which point
there are maxima at 9.12 and 9.65 /*.

GLASS.
(Curves a, b, c, fig. 76, and curve a, fig. 78.)

Glass is of such a variable composition that if the various metallic
oxides other than of silicon had an effect upon the reflection bands one
would expect to observe it. For example, crown glass 0381 contains
68.7SiO2+i3.3PbO-f i57Na2O+2ZnO, while a flint silicate 857 con-
tains 2i.9SiO,+78PbO, and a lead glass contains 4.62PbO+8K2O
+45Si02.

The following samples show that the variation in the silicate content
of the glass has little effect upon the maxima, except that of intensity.

The curves show a slight curvature at 8.6 to 8.8 /*, a sharp maximum

88

INFRA-RED REFLECTION SPECTRA.

at 9.2 to 9.3 ju, and a third band at 9.7 p. Curve a, fig. 76, gives the
reflection for a monochromatic red glass, No. 2745, made by Schott &
Co., of Jena ; curve b is for a plane parallel interferometer plate ; curve c
is for a piece of fluorescent uranium glass, while curve a, fig. 78, is for
a piece of ordinary plate glass.

The fluorescent glass has an unusually low reflecting power. All of
the samples have a uniform reflecting power to 7.0 /*,, then a sudden
decrease, followed at 8 ju, by a large band of selective reflection. For

FIG. 76.— Glass.

the transmission curve of glass (fig. 53) it will be noticed that there is a
decrease in the transparency beyond 15 p. Hence one would expect to
find the reflection curve to be irregular as observed, being abnormally
high just beyond the reflection band at 9.3 p., and abnormally low on
approaching the second large absorptive band which lies beyond 15/x.
The glass plates were from 3 to 8 mm. thick, hence opaque beyond 4 /x,
so that the reflection values are not influenced by energy reflected from
the second surface, i. e., the reflection is from only one surface.

SILICATES.

89

Pfund1 has published a reflection curve of glass, which is entirely
lacking in these small bands, due to the fact that only six spectrometer
settings were made between 8 and I2/*, while nineteen settings were
made in the present work.

GRANITE.
(From Barre, Vermont. Fig. 77.)

Granite is a mixture of quartz, feldspar, and mica, and hence gives a
reflection spectrum which is the composite of these minerals. In the
present specimen the mica plates were partly broken off the surface in
polishing.

25%

O
U

<*-

<D

6 69

FIG. 77. — Granite (a); Porcelain.

PORCELAIN (GLAZED).
(Curve b, fig. 77.)

The specimen was a flat fragment of a glazed white porcelain dish.
As is well known, porcelain ware is made from kaolin, which is the
decomposition product of aluminous minerals, especially the feldspar of
granitic rocks.

The reflection curve shows maxima at 8.5 and 9.25 p..

PECTOLITE (HNaCa2(SiO8)8).
(Curve b, fig. 78.)

The specimen examined was spindle-shaped, consisting of two masses
of needle-shaped crystals grown end to end. The reflecting surface

1 Pfund : Astrophys. Jour., 24, p. 25, 1906.

INFRA-RED REFLECTION SPECTRA.

was ground parallel to the needles. The reflection curve is of the usual
form and shows maxima at 9.4, 10.3, and 10.8/1.

40%

c
o
'-p
o

<a
ir

>o

e 9 /o
FIG. 78.— Glass (a); Pectolite.

WILLEMITE (Zn2SiO«).
(Curve a, fig. 79.)

The reflecting surface was ground from a massive specimen. The
transmission curve becomes opaque very abruptly at 9 fi. This is fol-
lowed by a sudden rise in the reflection curve at IO/A. There are three
strong maxima at 10.1, 10.6, and n /A, respectively, followed by a small
band at n.6p.. From the location of the maxima it is evident that the
bonding of the SiO2 in willemite is different from that of quartz.

56 7 8 9 /O

FIG. 79.— Willemite (a) ; Vaiicite.

VARICITE

(Curve b, fig. 79.)

The reflecting power is low, and decreases uniformly from 4 to 7.5 p.,
where it decreases abruptly to 8.5 p.. There are maxima at 9.25 and

SULPHIDES.

9.7 p.. The latter is not very well defined, so that, from the fact that it
coincides with a band found in apatite, one can hardly infer that it is
due to the PO4 radical.

SULPHIDES.

STIBNITE (Sb2S8).

(Large fresh cleavage piece parallel to b, perfectly plane without striations. Area

i by 1.5 cm. Curve a, fig. So.)

The reflecting power of the sulphides of Zn, Pb, Fe, and Sb are
worthy of notice. They are known for their metallic luster, especially
stibnite (Sb2S3). Their reflecting power in the infra-red is equally
conspicuous for its high value, which is uniform throughout the region
examined to 12 p.. Stibnite is the most marked in this respect. By
transmission it was found to be unusually transparent, the absorption of
a thin film, 0.45 mm., being about 45 per cent throughout the whole
region to 12 p., beyond which there appears to be another absorption
band. The reflecting power was found to be about 35 per cent to 1 1 p,

soy,

3 4 s 6 7 a 9 10 n n a
FIG. 80.— Stilbite(a); Pyrite (b) ; Galena (c); Sphalerite (rf).

is/t

so that after eliminating the reflection the mineral is almost transparent.
This is well illustrated in curve b, fig. 55, which is for a section 10 times
that for curve a. It will be noticed the true absorption is only about 5
per cent. It is difficult to decide from the present date whether the
band beyond 14 p. is due to absorption or to selective reflection. The
reflection curve drops in the region of 13 ju,, which would indicate a
selective reflection just beyond this point.

The observations were repeated on different specimens, and no differ-
ence greater than experimental errors was found in the reflecting
power, so that it can be definitely stated that the reflecting power is
higher in the visible spectrum, and drops to a lower value beyond 12 p..
This, as well as the other sulphides, was found to be opaque to Roentgen
rays.

92 INFRA-RED REFLECTION SPECTRA.

The refractive index of stibnite, found by Drude, for sodium light is

449 (5-1?)-

The value calculated from the well-known Fresnel formula :

for the region from 4 to 8 /x, where the reflecting power is R = 36,
w = 4.o, and n2=i6. The dielectric constant, determined for me by
Dr. N. E. Dorsey by the static method, by placing thin sections in a
small air condenser, is about 8 to 9, which is only about one-half the
computed value, -•= n2 = 16.

The difference is so great that it is not permissible to consider this
substance a transparent non-conductor. Of course, if there is a large

reflection band beyond 15/11 (fig. 55) the above assumption is not true.
The proper test would be to find the reflecting power for Hertz waves.

Konigsberger and Reichenheim1 examined a series of natural sul-
phides by means of what may be called spectrum energy screens or
filters, viz, plates of quartz, fluorite, etc., which transmit only certain
regions of the spectrum. From the thickness of the plates of the min-
erals examined and the observed energy transmitted they computed
the reflecting power. The method can not be as accurate as the present
one, which involves a direct measurement of the reflecting power ; and
this may account for their higher values. For stibnite they observed a
reflecting power of 47.6 per cent for the region 0.5 to 4.0 /A, and 43.9
per cent for the region 1.6 to 4.0 /z. Of course, it is possible that their
sample had a higher reflecting power. The electrical conductivity was
of the order icr15 (Hg= i), and they concluded, as was found in the
present work, that the Maxwell relation does not hold true.

(From Rio Marina, Island of Elba. Curve b, fig. 80.)

The reflecting surface of one specimen was formed by grinding a
large crystal face, which was found would not take a high polish.
This specimen gradually increased in reflecting power from 25 per cent
at 2 /u, to 34 per cent at 12 //.. The low reflection at 2 p is evidently due
to lack of polish.

Curve b is for a natural cubical crystal face which was i by 1.3 cm.
in area. It was quite plane, and had a high polish, except at one end,
where there were a few striations. A perfect crystal would have a

1 Konigsberger and Reichenheim : Centralblatt fur Mineralogie, Jahrg. 1905, p.
465-

SULPHIDES. 93

slightly higher reflecting power. The specimen has a slightly higher
reflecting power in the visible, and beyond 3 p a constant reflecting
power of about 32 per cent1 to 14^.

According to Reichenheim,2 the electrical conductivity is variable for
different specimens on account of impurities, so that no comparisons
can be made with the computed reflecting power, such as have been
made by Hagen and Rubens3 for metals.

(GALENA (PbS).
(Cleavage piece, surface 2 by 2.5 cm. ; quite plane. Curve c, fig. 80.)

The reflecting power appears to be slightly higher in the visible than
in the infra-red, where it is constant at 31 per cent to 14^. No doubt
the reflecting power would be slightly higher for a perfect specimen.
It was found impossible to grind a surface that was not full of small
depressions, due to chipping of the surface.

SPHALERITE (ZnS).
(Curve d, fig. 80.)

The specimen examined was a dark mass of crystals. The surface
showed several cracks. The reflecting power is no doubt low, even for
a perfect surface, as will be noticed from its uniformity throughout the
spectrum to I5//,. No large, clear specimen of this material was
obtainable. The refractive index in the visible is 2.369 (Na), while
computing from the reflecting power at 5 to 10 /JL would indicate a value
of n = 1.85 for R = 0.09.

SULPHUR (S).
(Reflection from large crystal face. Curve b, fig. 81.)

The reflection curve is low and uniform throughout the spectrum.
It is interesting to note that the absorption bands at 8 and 12 p. (found
in a previous examination) are too small to affect the reflection curve.
In this connection it will be noticed that all the sulphides examined,
except ZnS, have a much higher reflecting power, while the zinc sul-
phide reflection is practically the same as that of pure sulphur.

1 Konigsberger and Reichenheim : Centralblatt fur Mineralogie, Jahrg. 1905, p.
465, "by extrapolation from the curve for the visible spectrum found a reflecting
power of 30 per cent."

* Reichenheim : Inaug. Diss., Freiburg, 1906.

"Hagen and Rubens: Ann. de Phys. (4), u, p. 873, 1903.

94

INFRA-RED REFLECTION SPECTRA.

CARBORUNDUM (SiuciuM CARBIDE) (SiC).
(Large hexagonal plate, naturally highly polished. Curve a, fig. 81.)

This is an artificial product obtained from a carborundum furnace.

The reflection curve is the most remarkable one yet discovered. The
selective reflection bands of quartz at 8.5 to 9.03 ^ stand second in the
order of intensity.

Since absorption, reflection, and refraction are intimately connected,
the reflection curve gives us an idea, qualitatively, of the dispersion of
this mineral. It will be noticed that the reflection curve drops abruptly
from a fairly constant value at 9 p, to a very low value at 10 /x, while
beyond 13 /j. it remains abnormally high. This is exactly what is found
for the refraction curve, in the region of anomalous dispersion, and to
illustrate this point the carborundum curve is the best example yet
observed. In this connection it will be noticed the reflection curve of
quartz decreases more uniformly throughout this whole region.

FIG. 81.— Carborundum (a) ; Sulphur (b] ; Graphite (e).

GRAPHITE (C).
(Natural mineral from Siberia. Curve c, fig. 81.)

The reflecting power of various forms of carbon, such as "gas-
carbon" and anthracite, have been examined by Aschkinass,1 who found
the reflection to increase from 30 per cent at 8 p. to 53 per cent at 26 ft
for the former, and a uniform reflection of about 13 per cent throughout
this same region for the latter.

Aschkinass: Ann. der Phys. (4), 18, p. 373, 1905.

TRANSPARENCY TO X-RAYS. 95

The present sample, which had a high polish, increases uniformly in
reflecting power from 40 per cent at 2.5 //, to 60 per cent at 12 fi. It
thus appears that the presence of silicon in carborundum lowers the
reflecting power up to 9, beyond which it becomes abnormally high.

The results obtained from the present examination of reflection
spectra demonstrate a number of important facts. The first one is the
dependence of reflection upon absorption. The second point worthy
of notice is that the region of selective reflection begins beyond 7/x,
while for the majority of substances studied the region of greatest
activity is from 8 to io/x. Whether this coincidence in the grouping of
reflection bands (of minerals other than those containing SiO2) is fortu-
itous is unknown. A knowledge of their dielectric constants might
aid in deciding this point. In considering this question of the selective
reflection beyond 7 p., it will be recalled that the "general absorption"
of many substances was found1 to increase beyond this point.

The results as a whole show that there are not such definite bands,
whether found by reflection or by absorption, in the silicates as one
would expect. In other words, the silicon radical seems to be differ-
ently bonded in each mineral. Possibly there are several radicals — SiO2,
SiO3, SiO4 — one or more of which are present in each mineral, or even
in different specimens of the same mineral. This would explain the
lack of constancy of the occurrence of the bands of quartz at 3 p..

The investigation has added one more radical which has definite
absorption bands in the infra-red, viz, SO4, which has harmonic bands
at 4.55 and 9.1 /*.

TRANSPARENCY TO X-RAYS.

The majority of these silicates and sulphides were examined under
X-rays, and all but graphite were found to be opaque. Of course, the
samples were large, which means that in thin sections, no doubt, many
of the silicates would be as transparent as glass.

REFLECTING POWER OE METALS.

The reflection power of various metals and alloys, which can be
easily produced in the form of concave mirrors, has been measured by
Hagen and Rubens.2 The list does not include Co, Zn, Cd, Al, Sn,
Pd, and Ir, the reflecting power of which in the form of plane mirrors
is herewith presented.

1 Infra-red Investigations, Washington, 1905.

2 Hagen and Rubens: Ann. der Phys., 8, p. I, 1902; n, p. 873, 1903.

96 INFRA-RED REFLECTION SPECTRA.

The present list can not, of course, be of practical use, since the sur-
faces tarnish, but from a theoretical standpoint they are of considerable
importance. For example, Hagen and Rubens established relations
between the reflecting power and the electrical conductivity of the
metals studied. One would, therefore, expect similar relations for
closely related metals in the Mendeleef's series.

For example, one would expect the reflecting power of cobalt to be
of the same order as that of nickel, and a similar relation between zinc,
and copper, and palladium and platinum. From the present examina-
tion it will be noticed that such a close parallelism exists in all cases
where the actual condition of the reflecting surface, i. e., its polish, is
negligible. Unfortunately in the present list only zinc and cobalt take
a high polish which is quite permanent. Cadmium also takes a high
polish, but tarnishes in a day or two. Tin can not be given a high
polish ; palladium is of a similar nature, while aluminum always retains
a hazy white surface. As a result, in the shorter wave-lengths the
reflecting power is lower than normal, and rises steadily to 8 or 10 /*,
where it assumes a constant value which can no doubt be interpreted
as real.

The specimens examined were about 3 by 4 cm. area. They were
ground plane, then polished with Vienna lime and stearin oil. The
silver mirror was finally prepared by "buffing," and had a fine surface.

An attempt was made to use silver-on-glass mirrors, but even those
that transmitted only blue light were found to differ as much as 2 per
cent in reflecting power, while the best silver-on-glass mirror reflected
about 0.5 per cent less than the one of pure silver at 5 ju, to io/t. A
mirror of pure silver was therefore used as a standard of reference. It
consisted of a thick (0.5 mm.) sheet of the pure metal soldered on a
heavy plate of brass.

The method of observation consisted in placing the standard silver
mirror and the comparison mirror upon the carrier before the spectro-
meter slit, as in the preceding work, and obtaining the ratio of the
deflections. This gives the reflecting power relative to silver, and is
slightly higher than the absolute reflecting power, since silver is not a
perfect reflector. The absolute reflecting power of the metals, given
in Table I, were found by multiplying the relative values by the reflect-
ing power of massive silver, given in the first column of the same table.

The reflecting power of Ni and Pt are quoted from Hagen and
Rubens (loc. cit.) to show their close relations with Co, Pd, and Ir,
respectively. The angle of incidence was about 25°, and was permis-
sible, since it is well known that the reflecting power increases but
slightly up to this angle. Of course, the assumption is tacitly made

METALS.

97

here that the change in reflecting power with angle of incidence is the
same for all the metals examined, and, since we are rinding a ratio, will,
therefore, affect alike the numerator and denominator of the fraction.
Any error thus introduced could be only a fraction of a per cent, which
is as accurate as the variation in the polish of different samples of the
same metal will permit. The low reflecting power of most of the
metals examined in the region of I p, is due more to lack of polish and
planeness of surface than to a possible transparent region such as
obtains in silver, in the ultra-violet. Palladium is lower in reflecting
power than platinum; and it is barely possible that it would have a
slightly higher value if a better surface could be produced. The speci-
men was made by soldering a o.i mm. sheet upon a heavy plate of brass.

TABLE I. — REFLECTING POWER OF METALS.

[Absolute values.]

«

a

OJ

a

U

pM ^^

W

<U

^^

B)

^

0)

%

a

V

tn
1? 3

S-o

•—•a

cd

oJ a

S> 0

•z-e,

tf. 3

_ OJ

(massive, sh

1/5

CS

%

a

3

a

•*j

w)

4-*

K

s

_3

a

6C
at

W

N™^ .

Is

U!
(A

a!

s~*,

•M

tn

1

ii

a

fc a

•4-"

13
fi
o

o
a

• r*

a

a

13
_«

II

Sfl

.i: *

'§

•d

^

35

«

o

N

<

H

fL,

s

M

o

1. 06

964

72.0

67-5

794

73-8

54-o

74-8

72.9

79-4

70.8

I 71

q? 3

71. S

QI .O

80.8

cq i

79 3

79-5

84 7

85.0

/
3.01

973

88.7

/ * \s

767

7

95-5

88.3

68.6

t s \j

875

88.8

91.4

930

3 96

97-7

925

80.7

96.2

91.4

71.7

88 i

9'-5

93-3

95-2

5-24

97-3

947

86 2

97.2

938

76-7

90.4

93-5

94.2

95 9

6.75

985

948

92.7

97 2

95-2

80.3

933

95 5

94-7

97.0

8.02

988

95 °

948

98.0

96.9

83.2

94-7

95-1

948

978

938

989

95 6

964

98 i

974

87.0

953

95 4

95 6

98.4

10.49

990

95-8

96.8

984

96.9

87.0

966

95-9

95-8

98.3

12.03

989

95 7

96.6

98.3

97-3

86.9

965

96-5

96.1

98.2

T i oo

QQ O

QC fi

95

QQ.O

q6 4

99.0

7 ' T^

Considerable difficulty was experienced in casting a homogeneous
plate of cadmium. Success was finally attained by melting it in a thin
copper mold. When cool the mold was torn off and the ( i cm. thick)
plate filed and ground plane. In fig. 82 it will be noticed that its
reflecting power suddenly rises to a constant value beyond 5 /*,.

The sheet of cobalt was about 0.5 mm. thick, and permitted consider-
able filing and grinding. However, it was found impossible to prepare
a surface that was free from pores. This probably explains its devia-
tion in reflecting power from that of nickel, out to 10 p, where it reflects
more than nickel, as it should, since its electrical conductivity is higher.

rs v

98

INFRA-RED REFLECTION SPECTRA.

The aluminum was a sheet of commercial material. It took a high
polish. Its reflecting power is unusually high beyond lo/x, and is
known to be practically a perfect reflector for heat waves at 25 //,.

Little can be said concerning tin. It was found impossible to give it
a polish, although the melted surface on cooling was very bright. From
its electrical conductivity it ought to have a reflecting power of the
order of nickel and platinum.

\^^^

-JO-fc^

678

FIG. 82.- -Pure metals.

Zinc is the most interesting of all the metals studied. It takes an
unusually high polish which is quite permanent. Its color is peculiar.
It seems to have a low reflecting power in the visible spectrum, which
rises suddenly to a maximum beyond 4 p, and in this respect compares
favorably with silver, which is the highest and most serviceable reflector
known for the visible and the infra-red spectrum. The electrical con-
ductivity, as well as the reflecting power, of zinc and of cadmium are
close in agreement.

The indium mirror was obtained by polishing a sheet about 3 by
2 by o.i cm. It was not free from scratches, but took a high polish. It
has a slightly higher reflecting power than platinum. Its electrical con-

METALS.

99

ductivity appears to be unknown, which is also true of the purity of the
specimen examined.

The metals, with the exception of aluminum, were obtained from
Kahlbaum and from Heraeus. Pure cobalt is, of course, practically
unobtainable, and the specimen examined probably contained from 1.5
to 2 per cent nickel.

Hagen and Rubens (loc. cit.) have computed the absorption of the
metals from the electrical conductivity by means of the formula

100 — R= 3 ~5 where R is the reflecting power, c is the reciprocal of

the resistance of a conductor I m. long and I sq. mm. area, in ohms,
and I = wave-lengths in /* = o.ooi mm. They found a slight varia-
tion in the observed and computed values of 100 — R, the maximum
being about 0.5 per cent at 12 p. It must be said, however, that if
they had selected the wave-length i— 10.49 P, where in many cases
the value of R is frequently the same as for 12/1, the discrepancy
would be larger, and of the same order as that observed in the present
results. In the present work no attempt was made to attain the accu-
racy of these two investigators, for the reason that the nature of the
material would not permit it. The difference in the observed and com-
puted values of 100 — R is given in the following table, using the values
of the electrical conductivity as found by Jager and Diesselhorst.1

TABLE II.

Metal.

TOO—/?.

Observed.

Computed.

Cd

1.8
3-4

?r 5

L7
13-

2-7

2.87
3.28
3-48
2-59
3-55
1.89

Co

Pd

Zn.

Sn

Al

The agreement in the observed and computed values (excepting tin)
is as close as one can expect from the nature of the metals examined.

REFLECTING POWER OF SOLUTIONS.

It is well known that in the visible and in the ultra-violet the position
of the maximum of absorption of a solid is generally not affected when

1 Jager and Diesselhorst, quoted in Landolt and Bornstein, Tabellen.

JOO INFRA-RED REFLECTION SPECTRA.

in solution. Stenger1 found that it is only when a change in the aggre-
gation conditions or in the solving process is accompanied by a change
in the physical molecule that a change occurs in the absorption spec-
trum. One would expect similar conditions in the infra-red. In
Appendix IV of the first volume of this investigation a preliminary
examination was made of the transmission of several solids in solution,
and it was shown that the method is feasible for infra-red work. The
present examination by reflection is only preliminary, and was under-
taken primarily to learn whether possibly some of the sulphates, which
in the solid state have a single sharp maximum at 9.1 /A, really have
several bands, say at 8.6 and 9.6 /x, which for some unexplained reason
are merged into one. (There seems to be no reason why the sulphates

^

Spectrometer slit "/X Plane mirror

(Vertical) ir> i '[j -f—

' ^ // i '

ILjA-^-iJ) Vessel for liquids

\ 'wgTin itirriS and position of
I comparison mirror

\ I

\ I
\ i
\
\ t

i.'srnst heater

FIG. 83. —Arrangement of apparatus for reflection from solutions.

of Cd, Co, Ni, and Cu should have a single band, while those of Ba, Sr,
and Mg should have several bands.) Of course one would hardly
expect this to be the case, but preconceived ideas are often deceptive.
The constantly recurring bands of the sulphates at 4.5 to 4.6 /* and

9.05 to 9.2 fj, reminds one of similar conditions in compounds containing
CH3 groups. The adjustment of the apparatus is shown in fig. 83.
On account of the difficulty of adjusting the liquid to the level of the
silver comparison surface no attempt was made to obtain the absolute
reflecting power accurately.

SULPHURIC ACID (H=SO4).
(Concentrated. Fig. 84.)

This substance was not examined for various concentrations, as
explained in Appendix II.

The maxima occur at 8.6, 9.55, 10.42, and 11.35 /x. The fact that the

8.6 [ji and 9.52 /x, band of H2SO4 occurs in the hydrous and anhydrous

1 Stenger: Ann. der Phys. (3), 33, p. 578, 1888.

SOLUTIONS.

101

sulphates examined shows that they are not present exclusively in those
sulphates containing water of crystallization. In other words, they are
due to the SOj radical, and not due to hydrous sulphates, as might be
inferred from the study of sulphuric acid. Concentrated sulphuric acid

7 8 9 10

FIG. 84.— Sulphuric acid.

\Z

contains SCX, and it is interesting to note that the 10.4^ band, which
in a previous examination of gases was found in SO2, is one of the
strongest here, and, as will be noticed in Appendix II, disappears on
diluting the acid.

CADMIUM SULPHATE (CdSCX).
(Saturated, and dilute 1A H2O solutions. Fig. 85.)

The reflecting power is, of
course, much lower for solu-
tions. Curve b shows that the
reflection band is asymmetri-
cal. The maximum occurs at
9.2 //,. The absorption band at
4..6 /*, is harmonic with this one.
This reflection band was found
by Pfund1 for the solid crystal
at 9.1 /u, — more nearly 9.15^,
as read on the published curve.

o
It

a 3 10

FIG. 85.— Cadmium sulphate.

\\JU

Water has no reflection bands in this region.

NICKEL SULPHATE (NiSC>4+7H2O).

In fig. 86 are given the curves (a) for a saturated solution of NiSO4,
and (b) for a solution diluted to about one-half, while curve c is for a
large, solid crystal which was not highly polished. The curves for the

1 Pfund : Paper presented at the meeting of the Amer. Phys. Soc., April 20-22,
1906.

IO2

INFRA-RED REFLECTION SPECTRA.

saturated solution indicate that the reflection band is complex, with
maxima at 9.15 and 9.5^, the latter being coincident with that of
H2SO4. This is better illustrated in the curve for the dilute solution,
in which the band at 9.1 ju, has quite disappeared.

C 3Z
o

•s

(6

a 9 10 n

FIG. 86.— Nickel sulphate.

6 8 10 II IZJUL,

FIG. 87.— Copper sulphate (a) and (c) ; Ziuc sulphate (*).

SOLUTIONS.

103

The reflection band of the solid crystal does not agree with the one
found by Pfund at 9.05 /*, which is the mean value of the present curve.
In the present curve the maximum is evidently complex. Neither does
the maximum of the solid and that of the solution agree, which is prob-
ably to be expected. In fact, the study of solutions was undertaken to
test this very point.

COPPER SULPHATE (CuSO4+sH*O).

In fig. 87, curve a is for a saturated solution of copper sulphate in
which the maximum of reflection is very sharp at 9.15 p. For a solid
crystal of this material the maximum (curve c) coincides with that of the
solution. The reflection band found
by Porter at 2.3 /* really occurs at 3.3 /*.
The former value is due to an error in
computation.

ZINC SULPHATE (ZnSO*+7H2O).

The reflecting power is somewhat
higher than in the preceding com-
pound. The maximum (curve b, fig.
87) occurs at 9.2 /* for a saturated
solution of this compound.

SODIUM SULPHATE (NaSCX+ioHiO).

The selective reflection of a satu-
rated solution of this compound is
shown in curve a, fig. 88. The maxi-
mum occurs at 9.2 p, while Pfund
found it at 9.02 /*, using the surface of PJG
a mass of crystals.

MERCURY (Hg).

to

HjU

-Sodium sulphate (a) ; Potassium
sulphate.

Several observations were made on the reflecting power of mercury,
but the present arrangement was not well adapted to determine the
absolute reflecting power, and no thorough examination was made.
The chief difficulty is in having the two surfaces at the same level.
Values of reflecting power as high as 85 per cent (purity not known)
were observed, while the computed value is 90 per cent at 12 p.

Earth tremors had but little effect on the surface ; while, if pure dust-
free mercury be used, this is the best substance available to compare the
effect of the polish upon the reflecting power in the region of short
wave-lengths. In fact, the determination of the absolute reflecting
power throughout the spectrum would be an interesting study by itself.

IO4 INFRA-RKD REFLECTION SPECTRA.

POTASSIUM SULPHATE (KaSO4).

Curve b, fig. 88, is for a saturated solution of this compound. There
appear to be two maxima — a small one at 9.1 p, and a much larger one
at 9.4 fji — which disagrees with Pfund, who found a maximum at 8.85
for the reflection from the plane surface of a mass of these crystals.

The results, as a whole, show that the single narrow reflection bands
of several of the sulphates at 9 ^ is complex, and is shifted toward the
long wave-lengths when dissolved in water, while in others this band
remains single, and is not shifted in solution. The reflection of most of
the solids was found by Pfund (loc. cit.). In some cases we agree in
the location of the maxima, while in other cases (solids vs. solutions)
we do not agree. This is not, in any inherent errors, in adjustment of
the present instrument, for at the conclusion of the work the quartz
band at 9.05 /j. was found at its proper place, as shown in fig. 86.
Neither is it due to errors in Pfund's apparatus (although his dispersion
was not so great), for we agree in the position of the maxima of quartz,
glass, Iceland spar, potassium nitrate, copper sulphate, and sulphuric
acid. The solutions were examined on the same day and in the follow-
ing order: H,SO4 and the sulphates of Cd, Ni, Cu, Zn, K, Na. The
disagreement is in Ni, Zn, K, Na, which means that the instrument
could not have gotten out of adjustment during the examination.

The conclusion to be drawn is that several of these sulphates, viz,
Ni, K, Na, and possibly Cd, are dissociated, or that the intra-molecular
condition of the molecule (the "bonding") has become similar to that
of H2SO4. Possibly hydrates have been formed. But what may we say
of the sulphates of Ba, Sr, and Mg, which in their solid (anhydrous)
condition have several bands, some of which lie close to those of H2SO4 ?

The more logical wray of attacking this problem would have been to
examine1 the solid crystals at the same time as the solutions ; but life
is too short for one man to do all this, and the aforesaid line of reason-
ing excludes the possibility of the disagreement being due to instru-
mental errors, which would have to amount to from 4 to 6 min. of arc
to account for the shifting of some of the maxima. In Cu and Cd the
shift, if any, is very slight, although the Cd band is evidently complex.

In the sulphates of Ni, K, and Na a definite shift has been noted, and
is similar to the effect observed in solutions of iodine in CS2, and in
C,H5OH in the visible spectrum. No doubt a solvent may exist, in
which the 9.1 \i band of CuSO4 is also complex.

1 Since writing this the author has examined the sulphates of Ni and Cu in the
solid state and in solution. The maximum of solid CuSO* and of the solution in
water occurs at 9.12^, while in NiSO4 the complex band of the solid at 8.9 to
9.15 ji is shifted to 9.2 to 9.5 ^ in solution.

RESULTS.

105

SUMMARY.

The transmission and reflection spectra of at least 125 elements and
compounds have been examined, many of them to 15 p, by means of a
mirror spectrometer, a rock-salt prism, and a Nichols radiometer. The
aim of the investigation was the study of a series of minerals containing
oxygen and hydrogen in the form of what is known as water of consti-
tution and water of crystallization. The interpretation of the results
are based upon the assumption that if the union of the oxygen and the
hydrogen in the molecules is similar to that of water, then the absorp-
tion spectra of minerals, containing these two elements thus united,
should show the absorption bands of water superposed upon the absorp-
tion spectrum of the other constituents.

On the other hand, minerals containing oxygen and hydrogen as
water of constitution should not show the water bands, except hydrox-
yls, which should show a band at 3 p.

The results show that of some 30 minerals containing "water of
crystallization" there are no important exceptions to the rule that they
should show the bands of water. On the other hand, the one important
exception to the rule that minerals containing ''water of constitution"
should not show water bands is cane sugar. Minerals containing
hydroxyl groups generally have a marked band at 3 /*. Sulphates have
a strong band at 4.55^, and a less constantly recurring band at 9.1 /JL,
due to the SO4 ion. On the other hand, the silicates do not have such
definite bands, which would seem to indicate that the union of the
silicate radical is different in each mineral containing that element. In
Table III are given a list of groups of atoms which have characteristic
absorption bands.

TABLE III.

Compounds having the
following groups.

Show characteristic absorption bands

at :

CH3 or C1T3

•; 4^ 6 86 1^.6—1^.8 and

14 M

NH2. . .

2.96 6. i to 6 15 fi

r2S /' 6.7S 8.68 9.8 II. 8 12. qs

/'-

NO2 .

7.47 q 08 ?

OH .

3.0 //

NCS

4.78 p.

4.55 8.7 9.1 u

The examination includes minerals of which the chemical constitu-
tion is in doubt; for example, talc and serpentine. The former is not
supposed to contain hydroxyl groups, while in the latter such groups
are inferred. The present research supports these views in that the.
transmission curve of talc does not contain an absorption band at 3 M,

IO6 INFRA-RED REFLECTION SPECTRA.

while serpentine contains a large band at 3 p., which is in common with
substances containing hydroxyl groups.

The reflecting power of the metals Zn, Co, Al, Cd, Pd, and Ir is high,
and the observed values at 12 ^ are in close agreement with those com-
puted from their electrical conductivity. It is a remarkable fact that
the region of the first occurrence of selective reflection of the majority
of substances examined (other than silicates) lies between the wave-
lengths 8 to 10 /JL. This is probably to be expected. Drude1 has shown
that the infra-red free vibrations (Eigenschwingungen) depend upon
the ponderable mass of the molecule, and, from this, that one can obtain
some idea of the molecular weight of the substance.

The study of the reflecting power of solids in solution is not suffi-
ciently extended to draw general conclusions, but the data shows that
the method is feasible — that the reflecting power is proportional to the
concentration — and that the maxima may or may not agree with those
of the substance in the solid state, depending probably upon the solvent,
just as is true of the visible spectrum. A notable example is iodine, in
CS2 and in C2H5OH. The experiment also shows that a single band
of a mineral in a solid state may appear as several bands when in solu-
tion. This is an interesting field that deserves further investigation.
The difficulties involved are not greater than for absorption spectra,
while the intensity of the energy is still quite large in this region of the
spectrum. In the visible spectrum it is known that different maxima
appear, depending upon the solvent and upon the addition of acids to
the solvent. But the visible spectrum is so narrow in comparison with
the absorption bands that the infra-red is far better adapted to the study
of this phase of the problem.

As mentioned in the text, the chief difficulty in this research was in
obtaining minerals suitable to illustrate the questions involved. This
has placed the writer under deep obligations to the late Prof. S. L.
Penfield, of Yale University, who donated a number of rare minerals not
obtainable from dealers, and to the officials of the U. S. National
Museum, who also supplied a large number of specimens. The Director
of the Bureau of Standards generously supplied apparatus and material.
I am also under obligations to Dr. J. C. Blake, whose advice on the
mineralogical side of this question was frequently sought.

WASHINGTON, D. C., June 20, /pod.

1 Drude : "Optische Eigenschaften and Elektronentheorie," Ann. d. Phys., 14,
pp. 677 and 936, 1904.

INFRA-RED TRANSMISSION AND REFLECTION SPECTRA.

107

TABLES OF MAXIMA.
TABI^E; V.— MAXIMA OF INFRA-RED ABSORPTION AND REFLECTION BANDS.

o*

d

5

d

d

d

d

d

d

d

d

O

«d

d

d

d

d

d

d

Cl

d

d

d

CI

d

CO

ft

tn

ft

ft

ft

ft

ft

ft

ft

ft

X

a1

ft

ft

ft

+

+

ro

+

ct

+

+

+

-r

.+

* +

iix

ta

V

•o

x

d

tn

3

'S

N

•e

CO

9

tn

"cO

ulandite.
H4CaAl2Si0Oi8

Lbite.
CaAI2Si6Oi6

:assium alum.
K2SO4A12SO44

d

d

+j Cl

._.- CO

>lec:te.
CaAl2Si3Oio

1

emanite.
Ca2B6On

cium chloride
CaCI2

Lassium ferroc
K4Fe(CN)0

ophyllite.
,KCa4(Si03)8+,

weylite.
H4Mj;4(Si04)3

omsenolite.
NaCaAlFe

1>

3

a

V

• *N

o

00

y

a

o

CO

o

^•"E

V

J5

<

01

0

ft

tn

&

£

en

<J

CJ

O

OH

<

q

H

I.9M

i-5

2.9

1.46

1.48

i-47

1-5

1-5

i-5

1-5

1-5

1.48

I 48

1-5

3-0

14

3-2

2.05

4-37

1.98

2.0

i-95

2.O

2.0

2.0

2.O

2 0

2.04

2.0

2.0

57

2.0

4-55

5-0

3.0

3-°

3-°

R.

3-1

31

2.96

2 95

3-15

3°

3.O

6 05

2.7

57

465

5-3

4-23

4 4

4-7

9-1

4.62

4-4

4.8

R.

4-75

4.12

R.

R.

3-0

6.15

R.

5-9

5

4-75

6.05

6.1

6.15

7-3

6-3

5-0

9-15

9.6

4-65

6-55

8.7

6.22

5-3

7-4

9-4

6.2

97

10.5

Refl.

6.65

6.1

R.

10 6

8.6

R.

6.6

9-03

11.05

9-05?

8.49

9-5

9-°3

lo.cS

I2..S

£o

' i ci

vj —

ClW

d
ft

co n
LT=>

ft

9,
ft

d

d

ft

0

£

d
A

d

CI

d
w

d

d

d

ft

O

0

d

ft

0

d

ft

0
$

rt rr

xt

^ to

O

N

00

00

CO

. *^

1

I

!/•)

*t

^-1-

+

';? "

_j_

^

to

~|~

-^-

~v~

+

ismondite. H4(
Al2Si6Oi6

iodite.
MgSO4NaSO4

haumasite. Ca:
C03CaS04-i

ydrotalcite.
Mg3A

d

OH
4J

4-*

"C

CO

ivianite.
Fe3P3Os

Cl

d

O»

/

"3

assiterite.

Fructose. C,

extrose.
C6H120

j

CO

ba

cj

a

altose.
Ci2H22Oi

O

d
d

o

si

_o

o

CO

f

uni arabic.
2C5H100

ochelle salt.
NaKC4H400

0

ffl

H

ft

>

§

O

•«

Q

u

S

~

O

ftj

1.4

i-5

i-5

1-5

15

1.48

1.4

3

3-25

[-5

1.5

1.5

i-5

1-5

1.48

1-5

2.0

2.O

2.05

2

19

2.05

2.O

5-8

5-8

3 °

3-o

2.15

3 to

2.O

2.O

2.1

3-0

3 to

3-0

3 to

3 to

6.6

to

3-4

2.6

3-7

3-o

3-0

3 to 4

4.75

3-5

3.8

38

7-3

3-8

4-75

} o

4-75

33

3-4

4-75

4-7

4-75

475

8.25

4-75

6.0

3-4

6.1

4-75

4-75

5-6

R.

64

9-7

6.0

4-75

6.03

o.o

9-9

R.

6.2

7-1

925

74

9.72

8.1

9 to

10.5

1 08

MAXIMA OK INFRA-RED BANDS.

TABLE V. — MAXIMA OF INFRA-RED ABSORPTION AND REFLECTION BANDS—

Continued.

d

a
+

o

Ci

s
a

te. MnO(OH).

FeO(OH).

A12O(OH)4.

e.
Al(OH)3-f3lIaO.

,A12(OH)2(P04)2.

llite. A1(OH)3.

AIO(OH).

CatBOH)SiO4.

Mg(OH)2.

cement.

C6H8(OH)6.

ydrate.
CC13CH(OH)2.

A

h*

phelite.
!2(Si04)3-f3H2O.

HNaCa2(SiO3)3.

Brucine.

Cg3

a

rt
bo

a

rt

V

Gothite.

Bauxite.

Turquois
A1P04

Lazulite.
Mg.Fe

Hydrargi

Diaspore

Datolite.

Brucite.

Portland

Mannite.

Chloral h

Prehnite.
I

Hydrone-
HNa2A

Pectolite.

1-5

3-2

3-1

2.9

3-3

3°

3-0

1-9

2.2

2-5

3-o

1.6

1.5

2.9

2.8

2-5

1-95

6.2

57

to

5.1

4-3

5-2

3-o

2.8

2-7

2.1

2.O

4.1

3-4

2.4

9.1

3-6

535

5-9

4.8

3-0

3-4

3-0

5.0

3-95

3-4

475

5.65

67

5-3

5-"

38

3-2

5-95

4-55

4-75

5.^

6-3

R.

6.2

6.7

5-0

3-8

6.8

R.

5'1

7-o

7-6

8.56

R.

7-7

6.1

4.6

7-3

9-4

6.0

9-45?

S.8

8.2

5-1

10.3

6.8

9.8

9-2

5-7

10.8

7.2

10.2

9-5

6.2

7-85

12.8

IO.OI

7-1

8-3

13.85

10.8

8.0

9.0

14.6

II. 2

9.2

97

IO.O

10.8

ii-3

n.6

11.9

12.3

12.7

13-3

t—

O

ob

6

g

ei

o

P°

6

6

5

0

6

0

6

<s

6

M

6

-n

n

w

&

to

'ft

to

&

Chloritoid.
H2uMg,Fe)Ala5

Clinochlore.
2H8Mg6Al2Si

Muscovite mica.
H2KAl3(Si

Biotite mica.
(HK)2(.Vlg,Fe)2Al2Si

Serpentine.
II4(Mg,Fe)3<

Talc. H2Mg3Si

Kpidote.
HCa(AlFe)3Si

CO

n

V

CO

CO

Glauberite.
Na2SO4Ca

Thenardite. Na..,

Kieserite.
MgSO4 + 2l

Celestite. Sr

Anglesite. Pb

Cadmium sulphate.
3CdSO4+8I

Sulphuric acid. H2

2.2

29

2.85

2.8

3-o

5.58

1.6

3-2

2-35

3.1

1-45

3-2

'•9

1-5

3-6

3-2

4-9

3-5

5-0

4-2

5-95

2.1

4.6

455

4-55

1.83

4-55

3-2

2.05

R.

5-6

5-9

5-6

5-6

5-0

7-25

3-0

6.2

5.65

3 -o

6-4

4-5

3 to

7-2

6-3

6 28

5-05

5-95

6.6

R.

43

6.5

6.2

455 R-

4-97

3-4

8.6

7-1

63

6.2

7-3

9-03

50

R.

R.

6-05

8. "2

6-4

4-6

9-55

7.18

7 i

6.6

8.1

9-74

5-3

8-34

8-3

R.

S.76

6-75

R.

10.4

R.

7-2

8-5

5.65

8.9

8-7

8-7 9-05

9.2

11-35

9.2

R.

R.

5-95

9-i

9-05

9-7

93

97

6-7

9.6

10.5

7-4

985

INFRA-RED TRANSMISSION AND REFLECTION SPECTRA. ICX)

TABLE V. — MAXIMA OF INFRA-RED ABSORPTION AND REFLECTION BANDS—

Continued.

. 1

Potassium dichromate.
K2C2O7

Potassium chlorate.

KC10:J,

Apatite. Ca5F(PO4)3

Garnet.
Ca3(Fe,Mg)3Al2(Si04)3

Monazite.
(Ca,La,Di)PO)4

_O

O*

N

<J
.»j

1

JU

• rt

^

Collodium.
C6H702(ON02)3

Ammonium chloride.
NH4C1

Sodium biborate.
Na2Br4O7

O

a

CO

N

V

i-i
M
^

td

A

PC

en

Orthoclase. KAlSi3O8

Amphibole.
CaMg3(Si03)4

Zircon. ZrSiO4

Albite. NaAlSi3Os

2.9

3-2

2.9

1.2

From

2-9

1.6

1.62

1-9

2-7

1-5

2-75

2.85

2-15

2.85

3-2

3-6

4.8

to

3-2

5-9

2.8

2.2

3-o

3-6

2.7

4.7

4.8

3.1

5-7

37

5-24

59

28

to 5. 3

6.9

3-5

2-95

3-7

4 12

to

5-3

6.0

36

6.3

5-4

6.27

6-5

4-3

R.

6.05

4-75

4.78

3-2

5-7

6.8

R.

5-9

6-93

6-9

to a.

10. 1

7-3

R.

5.66

II. 2

63

7-4

8.7

6.65

R.

10 6

7.85

7-5

R.

13.2

R.

R.

9-7

7-7

9.1

1098

87

9-6

7-45

8.8

8.7

IO.O

8.7

9-7

9-1 '

to 10

9-8

9-45

10.5

9-5

10.02

IO.O

10.8

II- 7

12.15

12.7

136

MgSiO3.

<S

4J

O
U

a
<j

4+7H20.

CO

o

O

bfl

S

•8O8{StV

osphate.
NaPO3.

S

O

.S

feldspar.

O

i

0

C)

ffi

s.
+ ioH2O.

O

°§.

"S

2

•4-*

a"

«

P.

-*•
U

.2 rt

« ~n

tsen

O
• m

*d

£•

o ^

a!

t_i -t-i

— a

*j C3

ct

"5

ti

i

-*-t

QJ ***

cd 3

G.2

p<fj

cc N

r— jrt

u;

u

a

01

4J

V

a

a

.£ £

3

3

m

£*

"%Z

a

Enstatit

Ebonite

Potassiu

Calcite.

Ferrous

Magnes:

Microcli

Sodium

Tourma
H6]

Granite

Nickel s

Copper :

in
o

a

N

Sodium

Potassiu

1.82

3-4

3-5

3-45

R.

3 2

R.

3-9

1.25

R.

Solid

Solid.

Solid.

Solid.

Solid.

2.9

5-9

3-93

9-1

3-8

884

5-5

2-75

8-5

R.

R.

R.

R.

R.

69

4 65

465

4.6

95

6.0

R.

8.8

8.9?

91

9-2

9.2

8.85

R.
9.12

8.25
9-1

5-6

7-15

5-7
6.7

R

6-5

9.92

R.

8.0

7-45
8.0

9.22
96

9 '5
Solu-

Solu-
tion.

.Solu-
tion.

Solu-
tion.

99

9.6

6-9

6.8

9-4

925

9-9

tion.

9.12

92

9-°5

10.2

8-5

9-7

10.7

9.2

9-4

12. o , 9 4

10.8

9-5

13.0 10.0

14.3 10.7

n-3

12.2

12.5

13-2

APPENDIX I.

THE TEMPERATURE OF THE MOON.

I.

From his Mount Whitney observations1 Langley concluded that the
soil of an airless planet at the moon's distance from the sun would have
a temperature not greatly above — 225° C. His later observations on
the radiation (reflection) from the moon led him to conclude that the
temperature of the sunward surface of the moon is about o° C. This
inference, he mentions, is contradictory to the one drawn from the
Mount Whitney observations, which in themselves he considered exact.
"The most reliable spectrum comparisons with a blackened screen show
an average effective temperature of +45° C., near the time of full
moon."1 He could detect no radiation from the dark moon. For the
bright moon, his spectrum radiation curves rise from a zero value at
6 to 7 fj. to a maximum at 8.6 /j,, with a smaller maximum at lOju.
There are absorption bands, due to atmospheric water, at 6 /x and 9.6 /x.
He found also direct radiation from the sun in this region, but does
not consider it in arguing for a direct radiation from the moon. He
considered that the part of radiation from the moon, which was trans-
mitted by glass, is reflected energy from the sun, whereas the part
absorbed by glass is radiation from the moon. The same assumption
was also made by Lord Rosse, but it is untenable if we admit a selective
reflection of the moon. To the writer the line of reasoning seems false,
in the light of o.ur present knowledge of infra-red radiation, and a better
criterion for judging whether or not the surface of the moon becomes
appreciably warmer under the rays of the sun would seem to be to
reexamine the rapidity of the fall and rise in intensity of the radiation
from the moon during an eclipse. Langley (loc. cit.) made observa-
tions on the eclipse of September 23, 1885, which indicate a sudden and
very rapid falling off in the heat as the eclipse commenced, with some
indications of a rise nearly as rapid after its conclusion. In fig. 89 are
plotted his observed galvanometer deflections during the progress of the
eclipse. The ordinates are galvanometer deflections; the abscissae are
the predicted times in hours to mid-eclipse. The three curves are for
the center, the east, and the west limbs of the moon. The observations
were interrupted by the rapid formation of clouds just at the predicted
time for the moon to leave the umbra. The eye observations showed

1 Mem. Nat. Acad. Sci., 4, p. 197, 1887.
no

TEMPERATURE; OF THE MOON.

in

an anomalous distribution of light, which was thought to be due to
variations in the terrestrial atmosphere at different points along the
great circle at which the solar rays touched the surface. However, the
results as a whole show a sudden change in the radiation (reflection)
from the moon, with change in illumination. The penumbra began to
affect the center of the moon at +lh 44m- The predicted time for the
moon to enter the penumbra was -|-2h 48™. In other words, in the short

time of about 1.5 hours, in
passing from the penumbra
into the umbra, the radiation
from the west limb has fallen
from a maximum to a mini-
mum value. Without actually
computing1 this fall in radia-
tion, which is so sudden as to
indicate a small conductivity
and a very superficial cooling,
one would think that at the
beginning of totality the radia-
tion would still be detectable.
The question of the gain (or
loss) of heat on the surface of
an isolated body like the moon
does not lend itself readily to
a coherent line of reasoning,
because of the fact that we
know so little about the phe-
nomena connected with it. A
thin surface, like a radiometer
vane, placed in a vacuum and
subjected to radiation, acquires
a maximum temperature in a
very short time. The same is
not true of the earth and its
atmosphere, isolated in space. It is difficult to say what one ought to
expect of the moon, but from ordinary experiences one would anticipate
an appreciable lapse of time for the surface to acquire an extreme
change in temperature, after subjecting it to the sun's heat.

At great elevations on the earth, where the atmosphere is rare, we
know that the soil grows colder, although the direct rays become hotter
with rise in height, and one would infer that similar conditions obtain
over the whole lunar surface.

FlG. 89. — Abscissas = time in hours to mid-eclipse
ordinates = galvanometer deflections.

1 See page 115.

ii2 TEMPERATURE; OF THE MOON.

Harrison1 inferred that the surplus heat of the moon will reach its
maximum several days after the day of complete illumination. In com-
menting on this statement Langley (loc. cit.) reminds the reader that
the region of the moon, which has received at first quarter the solar
rays for rather more than four days, after being subjected to the most
intense cold during the moon's long night, has been but very little
warmed in comparison with the surface illuminated at last quarter,
which has been heated during a mean duration of eleven days, so that at
the last quarter the heat of the moon is certainly not less than at the full.

During the eclipse of October 4, 1884, Boeddicker2 showed that the
moon parts with its acquired heat very rapidly, and "it is hard to admit
that this heat from the lunar surface, which the moon has been absorb-
ing during many days of continuous sunshine, is parted with at once,
the whole earthward surface of the planet cooling almost instanta-
neously."3

From alternate exposures on bright and dark spots Langley found
that the radiation from the bright region around Tycho exceeded that
from the dark surface of the Mare Imbrium by 31 per cent, and as a
whole found that the bright parts of the moon radiate about 14 per cent
more than the dark parts. This is evidently due to a larger reflection
of sunlight. "If the same holds good for the invisible solar rays, it is
certain that less of the sun's heat is absorbed by the bright parts, and
hence there is a possibility that the temperature and proper radiation
of the light regions are less than those of dark areas, though this would
not necessarily follow, since many other factors must be included."3

II.

The writer has examined the infra-red reflecting power of a series of
minerals, including quartz, feldspars, amphiboles, and micas, which are
the chief constituents of the rocks of the earth's crust, and has found a
uniformly decreasing reflecting power of 3 to 5 per cent at 4 ^ to 0.3 per
cent at 7/x, followed by bands of selective reflection from 8.5 to IO/A.
In this latter region the mean reflecting power is from 40 to 50 per cent
for silicates, while pure quartz has a reflecting power of 90 per cent.
The reflection curves are given in fig. 90 in a condensed form. In the
lower part of the figure several of Langley's lunar radiation (reflection)
curves are given, in which the ordinates are in arbitrary units. Of
course, the absolute reflecting power of the moon is unknown. Zollner
has shown that, owing to the irregularities of its surface, the full moon

1 Harrison : British Assoc. Report, 1866, II, p. 20.

2 Boeddicker : Nature, 30, p. 589, 1884.

3 Langley: Mem. Nat. Acad. Sci., 3, p. 17, 1884.

TEMPERATURE OF THE MOON. 113

does not reflect as a smooth sphere would do, but very nearly as a flat
disk of like reflecting power and filling the same angle. Such a disk,
if it diffused all the solar energy which falls on it, would send to us
97300 of what the sun does, provided it reflected perfectly in all direc-
tions the total incident solar energy. The moon, however, is not such
a reflector, and in the visible spectrum the solar light is about 500,000
times moonlight. On this scale the curves of the moon given in fig. 90
would not be visible. The depression in the lunar curves at 9.6 p. is

soar

odine KA! Si3O8

. Reflections

(Lang/*

& 7 & 3 iO

FIG. 90.— Reflection curves of various silicates.

due to an atmospheric water band. The region from 5 to "j n likewise
contains water bands. One can readily see from these curves that a
reflecting surface composed of rocks like granite, basalt, diorite, etc.,
which are mixtures of feldspar, hornblende, and mica, would give a
curve similar to that of the moon.

The general assumption is that the crust of the earth and of the moon
have the same constitution, viz, a series of silicates. It follows from
the present research that the earth and the moon may be considered
8-c

114 TEMPERATURE Ol- THE MOON.

selectively reflecting surfaces, with a band of metallic reflection from
8.5 to 10 p.. Hence, in the region of the spectrum from 4 to 7 /A only
3 to 5 per cent of the radiation from the sun will be reflected from the
moon, under the best surface conditions, while in the region from 8 to
10 fji the reflecting power will rise to 50, or even 90 per cent. Under the
worst conditions the balance will still be in favor of the region of
8 to lOfji.

The occurrence of the maxima of the reflection curves and Langley's
observed emission curve of the moon (that part which is transmitted in
the transparent region of the earth's atmosphere between 7 and lOju,)
in this region of the spectrum may, of course, be nothing more than a
mere coincidence, and the author calls attention to it merely as an inter-
esting side issue to the main research. The reflecting power of the
moon for visible rays is onh 5ooVoo full sunlight. Assuming that at 9ju,
the reflecting power is, on an average, 20 times this, the value becomes
25oo7'or 0.00004 Per cent. Assuming the temperature1 of the moon to
be 300° abs., we can get some idea of its emissive power at 9/x, as com-
pared with the sun at a temperature2 of 5500°.

From Planck's formula3 for the distribution of energy in the spec-
trum of a "black body" (which, of course, the moon is not)

/ _£L x

-*f *r >

I=c^ \e -l)

we can obtain a ratio of the intensities for the two temperatures
^i = 300° abs. and T2 = 5500° from the formula

where C2 — 14,500, using ), (max.)=9 /*.

This ratio is 0.0016. But the moon, not being a black body, will have
a smaller emissivity.

If its surface were of iron oxide4 its emissivity would be only 0.3
that of a black body at 300°, and, judging from the rapid decadence of
the radiation curve (which may indicate a low conducting material)
during an eclipse (fig. 89). its emissive power may be even less than

1 Very : Astrophys. Jour., 8, 199, 1898 ; also Langley, loc. cit. ; Poynting : Jahrb.
der Radivaktivitat und Elektronik, 2, 42, 1905.

1 Warburg: Verh. Deutsch. Phys. Ges., i, 2, p. 50, 1899; Day and Van Ostrand:
Astrophys. Jour., 19, p. 1, 1904.

'Planck: Vehr. d. Deutsch. Phys. Ges., 2, p. 202, 1900; Ber. d. k. Akad. d. Wiss.,
Berlin, p. 544, 1901.

4 Kayser : Spectroscopy, vol. II, p. 80.

TEMPERATURE OF THE MOON. 115

this, say o.i. The ratio of the emissive power of the moon to that of
the sun will then be 0.00016, which is four times (0.00016-1-0.00004)
the reflected energy of the sun from the moon. On this assumption,
the moon at o° C. would radiate twice as much as it would reflect from
the sun.

This shows that unless there is something radically wrong in the
assumptions made, the above coincidence is fortuitous. This, however,
does not settle the question, for Langley observed also direct radiation
from the sun in this region, and from existing data of the radiation
from the moon we do not know how much of it is selectively reflected
energy from the sun. Computations which require all sorts of assump-
tions will not settle the question; but a bolometric comparison of the
spectrum energy curves of the sun and of the moon, made at high alti-
tudes, will be of the greatest service in clearing up this matter.

Since writing this appendix I have computed the fall of temperature of the
lunar surface, neglecting the conductivity from the interior (which simplified the
computation) and find that for an emissivity of 0.3 of a "black body" (iron oxide)
the temperature would fall from 300° to 273° abs. in 1.2 hours, while for an
emissivity of o.i, the time would be over 3 hours. By allowing for conductivity
these periods would be considerably increased, so that unless the moon's emissiv-
ity is considerably higher, the cooling curve will not be coincident with the eclipse
curve in fig. 89. Assuming that the sky does not radiate to the bolometer, then
the latter would give zero, and finally negative deflections as the temperature falls.
Langley does not mention negative deflections from the moon. He records
negative deflections for his sky-screen curves.

APPENDIX II.

ADDITIONAL DATA.

\s this work goes to press the experiments on selective reflection by
Pfund1 has appeared, and since it contains considerable new data it is
added for the sake of completeness. In fact, sulphuric acid was not
thoroughly examined by the writer when he learned that another was
working in the same field, although the examination of the sulphates
demanded it. The sulphates given in the appended table furnish addi-
tional evidence of the presence of bands due to the SO4 ion. Further-
more, the nitrates have a band at 7.45 p which the writer found in
several nitro derivatives2 of benzine. He was uncertain, however,
whether it was caused by the NO, group or "ion." In the same man-
ner, the 7.4 fj. and 9 ^ reflection bands of nitroso dimethyl aniline coincide
with the bands found by transmission at 7.4 and 8.9 p. The transmission
curve is very low, and probably most of the bands observed are really
due to selective reflection. Again, glycerin has a reflection band at 9.7 /x,
which is one of the characteristic bands of the alcohols as found by
transmission. No doubt the large absorption bands found by the writer
in ethyl succinate from 7 to 8 /*, in cyanine at 6.6 and 13.2^, and in
methyl salicylate at 7 to 9 p,, are really due to selective reflection, but
this material was not at hand when the present reflection work was
done. In fact, it is what one would expect to find for such large bands,
although there are no data to indicate the necessary size of the extinc-
tion coefficient to cause selective reflection. His interesting experiments
showing that a substance in a liquid and in a solid state has the same
absorption (reflection) maxima is further proof of what the writer
found for the absorption bands of several carbohydrates, such as thymol,
paraffin, stearic acid, phenol, and menthol at 3 p., and is, of course, to be
expected so long as the "physical molecule" is unchanged, as announced
long ago by Stenger. Of more importance are his experiments with
sulphuric acid, in which several bands disappear on dilution, which
may help clear up a similar case observed by Kriiss in the visible spec-
trum. His reflection curves for fuming sulphuric acid show maxima at
7.25, 8.6, and 10.35^, while the writer found absorption maxima at 7.4,
8.7, and 10.37 A* f°r sulphur dioxide (SO2). This is an extraordinary
coincidence which, it is true, is apparently not very close for the first

1 Pfund : Astrophys. Jour., 24, p. 19, 1906.
1 Investigations of Infra-Red Spectra, vol. i, p. 86.
116

ADDITIONAL DATA.

two maxima. But, when one considers that the bands are wide and
completely opaque as found by transmission, while they are very weak
as found by reflection (maxima only 7 and 12 per cent), and that they
were made by different observers with different instruments, it is not
straining matters to consider them to be due to the same ion. For
KNO3, Pfund found the maximum at 7.05 /* (on the published curve it
appears to be 7.1 //.), while the writer found the same band at 7.14^, so
that the discrepancy noted above is more likely to be due to a difference
in our calibration curves. Then, too, the maxima found by reflection
and by transmission should differ with the latter lying toward the
shorter wave-lengths. Whether a gas can have selective reflection will
depend mostly upon its extinction coefficient. To examine its reflecting
power in its liquid state will be almost impossible because of the dense
vapor above its surface.

TABLE IV.

Reflection maxima.

NaK tartrate, KNaC8H4O64-4H2O..

Magnesium nitrate, Mg(NO3)2-j-6H2O

Cobalt nitrate, Co(NO3)2+6H2O

Ammonium nitrate, NH4NO3

Calcium nitrate, Ca(NO3)2-f-4H2O

Silver nitrai e, AgNO3

Potassium nitrate, KNO-,

Nickel sulphate, NiSO4+7H.2O

Cobalt sulphate, CoSO4-f7H2O

Copper sulphate, CuSO4-|-5H2O

Cadmium sulphate, CdSO4+4H2O

Ferric sulphate, Fe2(So4)3+9H2O

Sodium, Na2SO4+ioH2O

Potassium, K2SO4

Fuming sulphuric acid and water, H2SO4-f SO3

Nitric acid, HNO3

Glycerin. C3H8O3

Na silicate, Na2SiO3

Nitrosodimethyl-aniline, (CH3)2NC6H4NO

7-45
7-45
745
745
7 45
705
905
9.05

9-15
9.10

9.05

9.02

8.85

(7.25

18.6

7.85
4.80

9 95
7.4

8 6 10.35
96 n-35
10-55
9.70

9.0

It is generally conceded that when a gas is dissolved in a liquid, part
of the gas goes into solution, part will be actually liquefied, while part
may enter into chemical combination with the liquid. Since change of
state does not affect the absorption (reflection) bands, it would appear
feasible to examine the reflecting power of gases by this method. How-
ever, the question of the solubility of gases in different liquids is also
a quite unexplored field, so that the investigator would first have to
search for proper solvents.

APPENDIX III.

EMISSION SPECTRUM OF BURNING CARBON BISULPHIDE AND
CORRECTIONS TO THE WORK OF JULIUS.1

It is becoming more and more a recognized fact that the infra-red
spectrum is the seat of great disturbances which can be attributed to
well-known groups of chemical atoms or "ions." The pioneers in this
field of investigation were Angstrom and Julius. Their interest in the
subject dates back to the time when rock salt first became recognized as
a means for producing the heat spectrum. The dispersion of rock salt
was then undetermined beyond 5 /u, and in order to express their emis-
sion and absorption bands in wave-lengths they adopted a tentative
method of extrapolation, which, since then, has been found to be erro-
neous. In the meantime, data on infra-red spectra have continued to
accumulate, which are often in violent disagreement. For example,
CS2 is variously quoted as having an absorption band at 6.7 to 8.4 ju,
while the true value is about 6.8 p.

During the past few years the writer has attempted to determine the
values of the maxima of absorption and emission as accurately as pos-
sible in absolute value of wave-lengths, and thus bring a little harmony
out of this chaos. This means repeating part of the work of others in
order to get a check upon the extrapolation. One of the most interest-
ing pieces of work of this type is that of Julius, who found the emission
spectrum of gases during combustion. It contains a very considerable
amount of careful work, certain parts of which appear to have gone
quite unnoticed by later investigators. It is of no little interest, for it
contains evidence of emission bands farther in the infra-red than subse-
quent work, along other lines, has been able to show. These bands
belong to the acid elements and appear at low temperatures, i. c., they
do not appear in spark (and arc ?) spectra. This is just the opposite
of what the writer found for the basic elements (metals) in which no
emission bands were found beyond 2. p. Perhaps this may eventually
give us some clue to the mechanism of radiation. We have two sharp
distinctions between the acid and the basic elements. The metals have
selective absorption in the short wave-lengths, are opaque to infra-red
rays, and no emission lines have been found beyond 2 /*. In fact, accord-
ing to Pfliiger's work, the maximum of the emitted energy lies in the
short wave-lengths beyond the visible spectrum. On the other hand, the
acid elements have selective absorption bands throughout the spectrum
and have emission lines extending far into the infra-red, which gen-
erally coincide with the marked absorption bands. The question of the

'Julius: Licht- und Warmestrahlung Verbrannter Case., Berlin, 1890.
118

EMISSION SPECTRUM OF BURNING CARBON BISULPHIDE;.

119

coincidence of the emission and absorption bands led the writer to
undertake the present work.

A previous examination of the absorption spectra of gases showed
very marked bands in SO, which were in quite close coincidence with
the emission lines of the products of combustion of carbon disulphid,
as given by Julius. The absorption spectrum1 of CS2 has a very large
absorption band at 6.8^ (opacity from 6.6 to 7.0/11) and smaller bands

80 cm

70

60

<n

c
o

°4j

«

*

Q

30

20

10

5 6 7 Q

FIG. 91.— Emission of burning carbon disulphide.

at 4.6 and 1 1.65 //,, while SO2 has a very large band at 8.7 /j., a somewhat
narrower band at 7.4 p. and several small bands; H2S has a series of small
bands which can hardly be considered in the question of the combustion
products of CS2. In fig. 91 is given the emission curve of burning CS2.
In curve a the maximum at 4.35 ^ is reduced to one-half the scale of
the rest of the curve. In curve b the bands at 6.75 and 7.45 /* have been
magnified 10 times to bring out the sharpness of the maxima. In the

1 See Investigations of Infra-Red Spectra, vol. I.

I2O EMISSION SPECTRUM OF BURNING CARBON DISULPHIDE.

present examination the lower part of the flame was examined. The
lamp consisted of a tin lubricating-oil can, the neck of which was 5 cm.
long, to prevent heating the CS2 and thus avoid an explosion. Julius
(loc. cit.) examined the upper and the central parts of a CS2 flame and
found that for the central region the maximum at 6.75 /x is more intense
than the one at 7.45 /A, while for the upper region the 7.45 p. band is the
more intense. This is to be expected, for the 6.75 p band is evidently
due to hot CS2. In the lower part of the flame the combustion is incom-
plete and hence the hot vapor ought to be more intense than nearer the
top. In the present examination the lower part of the flame was used,
and the ratio of the intensities of these two bands is still greater than
that found by Julius; here the ratio is 34 to 73, while Julius found a
ratio of 16 to 22 for the central part of the flame and a ratio of n to 8
for the top.

The band at 7.4 /A is in common with the absorption band of SO2 and
since Julius found this band also in burning sulphur, it no doubt belongs
to this gas. Then the interesting question arises, "Why is there no
emission band at 8. 7 /A?" The question must be left unanswered. It
is interesting to notice, however, that the 8.7 /* band is found in H,SO4,
and in sulphates, and appears to belong to the SO4 ion, while another
band found in SO2 gas at 10.37^ occurs in fuming H2SO4 and has been
attributed to SO3 by Pfund (loc. cit.).

The bands at 2.7 and 4.35 //, are in common with the emission bands
of CO2 and are no doubt due to the combustion of the lamp-wick.
This is one of the few examples of the emission of a heated gas during
the process of combustion. Water vapor and CO,, are other examples
studied by Rubens and Aschikinass, but in the emission of a flame
these are products of combustion. Kayser in his Spectroscopy very
aptly remarks that the emission band of methane (CH4) at 3.3 ju has
never been observed in a flame. This is a very weak absorption band,2
however, and might not appear, just as was found for the 4.6 p. band
of CS2. The 6.75 fj, band, which coincides with that of CS2, was
ascribed to COS by Julius, who records its maximum in wave-lengths
at 8.48^1. The 7.45 /A band of SO2 was thought to be at 10.01 //,. The
emission spectra of H2O and CO2 have been redetermined by others
and need not be mentioned here. It will be sufficient to add that for
CO, the maximum of emission depends upon the temperature. Whether
the SO, band at 7.45 ^ shifts with change in temperature has not yet
been determined. The only other emission band that needs correction
is that of HC1 at 3.68 /*. A more probable value is 4.05 p, which is close
to the HC1 band, found by Angstrom and Palmer1 at 3.41 /x (cor-
rected = 3. 98).

'Angstrom and Palmer : Ofversight Kongl. Vet. Akad., No. 6, p. 389, 1893.

APPENDIX IV.
A VACUUM RADIOMICROMETER.

Radiometers having vanes weighing less than 10 mg. are easily
affected by earth tremors, so that for general work it is not advisable
to reduce the weight much below this amount. Of course, such a mov-
ing system is not very sensitive for a short period. For such a vane,
with a fiber suspension of such a diameter that the maximum deflection
is reached in 15 to 20 seconds, the sensitiveness is not much greater
than 9 to 12 cm. deflection per square millimeter surface exposed — the
candle and scale each being at a meter's distance. However, a radio-
meter of such a sensitiveness always gives reliable readings and is to
be preferred to a more sensitive instrument which gives larger deflec-
tions, but which on account of the lightness of the vanes is affected by
earth tremors, temperature changes, etc.

On the other hand, in the radiomicrometer of Boys, the period is
governed by the magnetic moment of the moving parts, rather than by
the torsion of the fiber suspension, and in nearly all instruments, thus
far described, is much shorter than that of the type of radiometer just
mentioned. This instrument has received but scant attention since it
was first described by Boys, who had a sensitiveness of only about
i cm. per square millimeter (candle and scale at I in.) for a suspension
weighing 32 mg. and a period of 10 seconds. He used a window to
prevent air currents. Paschen1 attempted to improve the instrument,
but for the best junctions (out of some 50) he could obtain only about
three times the sensitiveness of Boys, for a period of about 40 seconds.
He then turned his attention to the bolometer, with his well-known
success.

The bolometer, however, with a delicate galvanometer requires an
elaborate installation, and needs so much attention that the investiga-
tor's time is occupied principally with the care of the instrument, which
should be a secondary matter in any work.

A great many radiometric investigations do not require the highest
attainable sensitiveness (e. g., the author's present work), and in such
cases a convenient accessory to a laboratory is a radiometer or a radio-
micrometer. The thermopile, of course, also has its place, but here heat
conduction, local e. m f.'s, etc., must be contended with.

'Paschen: Ann. der Phys. (3), 48, p. 275, 1893.

121

122 VACUUM RADIOMICROMETER.

The present study of the radiomicrometer was undertaken in con-
nection with the question of the feasibility of combining it with the
radiometer, thus developing more energy in the suspended system.
The chief difficulty encountered in this combination is to obtain systems
which are free from magnetic material. The diamagnetism also plays
an important part in this form of thermo-j unction, so that the radio-
meter effect is almost obliterated by the magnetic effect of the coil.

In the radiomicrometer proper the Bi-Sb junction hangs vertically
at the axis of the system and hence is not so seriously affected by its
diamagnetism, and if the magnet is not too strong a fairly sensitive
instrument is obtainable. Since a window improves its steadiness, the
whole might as well be in a vacuum, which will eliminate heat conduc-
tion, as is well known for thermopiles. By so doing the sensitiveness
of a certain system, to be described presently, was increased by at
least 70 per cent. By using No. 40 copper wire and bits of Bi and Sb,
soldered with Wood's alloy, it is possible to reduce the weight to 10 mg.,
which, as already mentioned, is a convenient weight for radiometers.
The period of such a system will be much less than that of the radi-
ometer of similar weight and time required for maximum deflection.

Before describing the present instrument it will not be out of place
to mention some facts concerning tests of sensitiveness of instruments.
In previously described instruments no mention is made of the shield-
ing of the thermo-j unctions from radiation other than that which falls
directly upon the vane, so that the test of sensitiveness is not neces-
sarily a fair one, since reflection from the walls has an enormous effect
in increasing the sensitiveness. In fact the indicated sensitiveness may
be twice the true value.

In the present instrument (fig. 92), which is a design (similar to
that of Boys) for a radiometer, radiomicrometer, or a combination of
the two, the receiving surface is directly behind the window, which is
covered with an adjustable slit the length of the vane. In this manner
no radiation reaches the walls of the instrument, which are black.

The Bi for one part of the junction was obtained by melting the metal
between glass or iron plates, which were then pressed together. These
thin plates were then cut into strips from 0.15 to 0.2 mm. wide. The Sb
part of the junction was split from a large, well-crystallized piece of the
metal. These pieces could not be obtained quite so thin as the bismuth.
The Bi and Sb pieces were then soldered with Wood's alloy, in the form
of a V,to a thin sheet of copper. This is not so good as to have the pieces
more nearly parallel, since in the latter case the diamagnetic moment is
reduced. The dimensions of the Bi and Sb pieces were about 3.5 by
0.2 by o.i mm. The length of the copper loop of No. 40 wire was 4.5

VACUUM RADIOMICROMETER.

123

cm., and about 8 mm. in diameter. The area of the thin copper vane
was 3 sq. mm. It was blackened by covering it with a little Canada
balsam, upon which was dusted fine copper oxide. The weight of the
complete loop was less than 10 mg. The magnet was taken from a
Weston ammeter. It was used without pole pieces, since the latter

FIG. 92. — Radiomicrometer, about two-thirds actual size.

caused too much damping. The top was covered with the bulb of a
thistle-tube, the mouth of which was ground flat. The heavy metal
base of the radiomicrometer is of Swedish iron. The narrow vertical
tube is of brass. All the joints are made with Kohtinsky cement, except
the one at the top and the one containing the support of the rock-
salt window. The top seal may be made with mercury. A torsion head

124 VACUUM RADIOMICROMETER.

is provided to bring the mirror in any desired position. The loop for
the radiometer-radiomicrometer is shown in fig. 92, a. The Bi is about
2 by 0.15 by o.i mm., soldered to a thin sheet of copper 8 by 0.6 by
o.i mm. In one case a sheet of mica of the same area was used and
extended below the copper sheet, which was then only 2 mm. long.
This loop also weighed less than 10 ing. As a radiometer the time
required for a maximum deflection was 25 seconds, and its sensitiveness
was about 3 to 4 cm. per square millimeter area of vane for a Geryk
pump vacuum, which was not high enough for maximum radiometer
sensitiveness. This same loop used as a radiomicrometer had a sensi-
tiveness of about 5 cm. deflection per square millimeter area of vane,
while its maximum deflection was reached in 8 seconds. The combina-
tion of these two was no better than the radiomicrometer, simply
because of the fact that the periods were different and the magnetic
moment of the radiomicrometer obliterated the radiometer effect.
Another loop twice as heavy had a half period (time of maximum
steady deflection) of 5 seconds when used either as a radiometer or
radiomicrometer. The actual sensitiveness was much lower than in the
preceding loops. In this case the radiometer contributed1 only 15 per
cent to the total deflection, while the sensitiveness of the radiomicro-
meter was increased 50 per cent by placing it in a vacuum.

In fig. 92, b, is shown the loop for the radiomicrometer. The dimen-
sions of the Bi-Sb junction are indicated above. Its half-period in air
was 20 to 25 seconds, and the deflections in centimeters per square milli-
meter for meter, candle, and scale were 3.6 cm. On exhausting the
instrument with a Geryk pump the half-period dropped to 12 to 14 sec-
onds, while the. deflection increased to 5.5 and 6 cm. per square milli-
meter of exposed area of vane, and as the pumping progressed the
sensitiveness continued to rise. The result as a whole indicates that the
weight of the moving coil can be reduced to at least one-third of those
previously described and that its sensitiveness can be increased by at
least 70 per cent by placing the loop in a vacuum. Since a window is
used to increase the steadiness, this is no drawback, while in comparison
with the type of radiometer having a heavy vane, as already described,
it can be given a shorter period.

The purpose in describing this form of instrument is not so much
to show the sensitiveness of the present one as to indicate directions in

1 The method of testing is as follows : The radiomicrometer effect is first found
in the air. It is then exhausted and the deflection again noticed. The magnet is
then removed and the deflection obtained is due to the radiometer effect. The
magnet must of course be placed so that the radiomicrometer deflection is in the
same direction as that of the radiometer.

VACUUM RADIOMICROMETER.

which further improvements are possible. On account of the difficulties
in preparing a non-magnetic loop the final word has not been said on
the subject of combining the radiometer and the radiomicrometer. At
the present writing the best combination was only about 20 per cent
more sensitive than the use of the instrument as a radiomicrometer
only.

For detecting electric waves the system can be made still lighter,
while the heating arrangement can be brought close to the junction.

In conclusion, the fact may be emphasized that in the present test of
sensitiveness only that radiation which passed through a slit and fell
directly upon the vane was measured. If this precaution had not been
taken, the results would have been much higher and would have been as
misleading as those obtained by several other investigators.

The sensitiveness of the bolometer varies as the square root of the
exposed surface. The writer has found experimentally that the sensi-
tiveness of a radiometer is proportional to the area of the exposed vane,
and hence is justified in expressing his results for unit (sq. mm.) area.
The use of a candle as a standard is questionable, but since the sensitive-
ness of the different vanes differs by a factor of from 2 to 25 the candle
is a convenient source for comparison.

APPENDIX V.

NOTE ON BLOWING FINE QUARTZ FIBERS.

The bow-and-arrow method of shooting quartz fibers invented by
Boys, as well as the method of heating a quartz bead to incandescence
in an oxyhydrogen flame and drawing it out into a fine fiber, seems
familiar to nearly everyone. Nevertheless, in making rapidly the finest
obtainable fibers, such as are necessary for radiometer and for light
galvanometer suspensions, there is one step in the latter method which
seems a novelty to so many investigators with whom the writer has
recently come in contact, that a description, with diagrams, is given
here, though there can not be anything new in the process, which seems
to be handed down by tradition rather than by recorded history.

Fortunately we can now buy rods of vitreous quartz, and are no
longer obliged to build up rods from bits of pulverized quartz crystals
by holding them in the hottest part of the oxyhydrogen flame. By the

Ox

//////////A *~~^

'////////,

'-'•
1

\

.

1

o

c

01
*\

O
o

r*

•x

Hyd

FIG. 93.

following method the manipulator can blow enough fibers in a few
minutes to last him many months, provided he stores them in a box
away from draughts and flies.

To be brief, the oxyhydrogen or oxy-illuminating gas flame is regu-
lated to a length of 5 to 8 cm. A shallow box (5 by 50 by 100 cm.)
with a lid, lined with black canton flannel or simply a black cloth, is
placed vertically at a distance of about 35 cm. from the flame, which
is horizontal. The manipulator, facing the cloth, holds the rod of
vitreous quartz (i to 2 mm. diameter) in the hottest part of the flame,
as shown in fig. 93, which is a horizontal view. When the rod begins
to melt it is drawn apart in a horizontal direction, the one end toward,
the other end away from the cloth. At the point of rupture a fine fiber
will be drawn out by the flame, and the heated gases will drive it against
the cloth. The two melted ends are then stroked past each other as
rapidly as the manipulator desires, and each time a number of fibers
will be drawn out. The fibers are so thin that only the heavier incan-
descent ones will be seen in the flame and hot gases, but on examining
126

BLOWING QUARTZ FIBERS.

the cloth in diffuse sunlight at a larger angle of reflection it will be
found covered with fibers of various sizes and lengths, the heavier
ones, of course, coming from the cooler parts of the flame. Colored
spectacles will be found serviceable in this work.

The handling of such fibers requires some dexterity, and in acquiring
it the manipulator generally develops some system of his own. The
writer has had the best success by employing two flexible glass rods,
of about 2 and 8 cm. length respectively, with a bit of shellac solution
on one end of each. The quartz fiber is secured from the cloth by
means of the longer glass rod, while the shorter one is permitted to
hang free from the other end of the fiber. Holding the radiometer or
galvanometer suspension in the right hand and the long glass rod in the
left, the shellacked end of the suspension is brought against the quartz
fiber, close to the long glass rod, which is then turned several times
around the suspension. A sudden pull severs the rod from the sus-
pension, leaving the quartz fiber, with the short glass rod at the other
end, hanging free. A hot wire is brought near the suspension (now held
vertically) to dry the shellac, care being taken that the shellac does
not become too hot, otherwise the fiber will drop off. The metal sus-
pension head having previously been laid at one end of a wooden tem-
plate, which is used to get the exact length of the fiber, the quartz
fiber is now laid across this head and the suspension is laid on the
template. The short glass rod hangs over the end of the template and
draws the quartz fiber taut. The fiber is pressed against the suspension
head (previously shellacked), the hot wire is applied, and the task is
completed. The manipulator then raises the wooden template into
the vertical position, and, if the lint has previously been burned off, the
suspension will swing free from it. By proceeding in this manner,
which takes less time than required to write this description, no tension
greater than the weight of the suspension is brought upon the fiber
and rarely is one broken in mounting. A glass tube, of a diameter
to hold the metal suspension head, mounted in a wooden block, is
serviceable in storing such a suspension and in carrying it from the
mounting table to the instrument.

The diameter of such fibers, which of course make excellent cross-
hairs, has not been determined, and it will be sufficient to add that they
look blue in dim light. The finest fibers appear crinkled, as one
removes them from the cloth. The bow-and-arrow process will no doubt
produce just as fine fibers, but the present method is better adapted for
speed. As already mentioned, the method is not new, but to the
writer's knowledge no full description of the actual manipulations
involved has been published heretofore, and if this attempt will be of
service to any one interested in the subject its mission will be fulfilled.

INDEX TO MINERALS.

Page.

Adularia 65,85

Albite 66,85

Alum 24, 79

Aluminum 97, 99

Amber 68

Ammonium chloride. 59

Amphibole 63, 86

Analcite 25

Anglesite 54

Anhydrite 18, 79

Apatite 58,87

Apophyllite 27,84

Azurite 43

Barite 53,78

Barley sugar 35

Bauxite 41

Biotite mica 48,83

Blodite 28

Brookite 67

Brticine 39

Brucite 43

Cadmium 97, 99

sulphate 55, 101

Calcite 69,81

Calcium chloride.... 26

Cane sugar 35

Caramel 35

Carbodundum 94

Cassiterite 32, 70

Celestite 54,78

Chloral hydrate 45

Chloritoid 47

Clinochlore 47

Cobalt 97,99

Colemanite 25, 85

Collodium 59

Copper sulphate 103

Datolite 43,87

Deweylite 27, 84

Diaspore 42, 79

Ebonite 66

Enstatite 66, 81

Epidote 51

Feldspar 63

128

Page.
Ferrous sulphate.... 79

d-Fructose 33

Fluorite 68

Galena 93

Garnet 58

Glass 65,87

Glauberite 54,78

Gismondite 28

rf-Glucose 34

Gothite 41

Granite 89

Graphite 94

Gum arabic 38

Heulandite 22

Hydrargillite 42

Hydronephelite 46

Hydrotalcite 30, 87

Iridium 97

Kieserite 54, 78

Lactose 37

Lazulite 42

Magnesite 70, 82

Malic acid 62

Maltose 87

Manganite 41

Mannite 44

Mellite 31

Mercur}r 103

Meta phosphoric acid 52

Microcline 82

Monazite 58

Muscovite mica. . . .48, 83

Natrolite 24,87

Nickel sulphate 101

Oligoclase 65

Opal 21

Orthoclase 63,65,85

Ortho phosphoric acid 52

Palladium 97

Para rubber 66

Peninite 47

Pectolite 46,89

Phloroglucin 61

Porcelain (glazed) . . 89
Portland cement 44

Page.

Potassium alum 24

chlorate . . 57
d i c h r o-

mate... 57
ferro cya-
nide.. . . 26
nitrate. .68,81
sulphate.. 104

Prehnite 46

Pyrite 92

Quartz 21,80

Raffinose 38

Rochelle salt 38

Rock salt 69

Rutile 67

Scolecite 25

vSelenite 18, 77

Serpentine 50, 85

Silver chloride 63

nitrate 62

Sodium biborate . . .62, 86
metaphos-

phate . . . . 52, 84
sulphate .... 103

Sphalerite 63,93

Stibnite 66,91

Stilbite 24

Sulphur 93

Sulphuric acid 55, 100, 1 14

Talc 51,82

Tartaric acid 61

Thaumasite 28

Thenardite 54

Thomsenolite 27

Tin 97,99

Tourmaline 48,84

Turquoise 42

Varicite 30, 90

Vivianite 31

Water 17

Wavellite 30

Willemite 59, 90

Zinc 97,99

sulphate 103

Zircon 65

INVESTIGATIONS OF INFRA-RED SPECTRA

Part III — INFRA-RED TRANSMISSION SPECTRA
Part IV — INFRA-RED REFLECTION SPECTRA

BY

WILLIAM W. COBLENTZ

WASHINGTON, D. C.

Published by the Carnegie Institution of Washington
December, 1906

MBL WHOI LIBRARY

UH IflHU L