QC 
 
 4-37 
 
 UC-NRLF 
 
 B ^ 
 
 THE ABSORPTION OF LIGHT 
 BY INORGANIC SALTS 
 
 DISSERTATION 
 
 Presented In Partial Fulfillment of the Requirements for 
 
 the Degree of Doctor of Philosophy In the Graduate 
 
 School of the Ohio State University. 
 
 BY 
 
 ENOCH FRANKLIN GEORGE 
 
 The Ohio State University 
 1920 
 
EXCHANGE 
 
THE ABSORPTION OF LIGHT 
 BY INORGANIC SALTS 
 
 DISSERTATION 
 
 Presented In Partial Fulfillment of the Requirements for 
 
 the Degree of Doctor of Philosophy in the Graduate 
 
 School of the Ohio State University. 
 
 BY *., 
 
 ENOCH FRANKLIN GEORGE 
 
 The Ohio State University 
 1920 
 

 
 
 
 &s^-+^ 
 
INTRODUCTION. 
 
 The mechanism of the absorption of light is one of the least 
 understood of physical phenomena. The literature upon the subject 
 is voluminous, but not always illuminating. The matter has been 
 dealt with by theoretical and experimental physicists and chemists. 
 On the experimental side there has, in general, been no attempt to 
 distinguish btween absorption, scattering, and reflection. In dealing 
 with solutions, especially, the difference between the intensity of the 
 incident light and that transmitted has been taken as a measure of 
 the absorption, although the decrease in intensity is due to all three 
 factors. This seems to be unavoidable. Certainly it is not easy to 
 differentiate between them experimentally. But in any theoretical 
 investigation or discussion, some attempt should be made to distinguish 
 between them. 
 
 When an electromagnetic wave, as a light wave, is incident upon 
 a material dielectric, there is set up in the medium, in addition to the 
 displacement current in the ether, a polarization current due to the 
 periodic displacement of the electrons in the atoms or molecules of 
 the medium, and the magnitude of this current will be proportional 
 to the time rate of change of the electric intensity. If the displacement 
 of the electrons is resisted by some force similar to that of friction, 
 energy will be absorbed from the waves and converted into heat, as 
 when a condenser is heated by a rapidly alternating current. If in 
 addition to the bound electrons of a dielectric there are also free 
 electrons, as in a metal, there will be conduction currents propor- 
 tional to the electric intensity. Thus again we will have heat gen- 
 erated at the expense of the radiant energy. 
 
 The chief difficulty with the foregoing explanation of light 
 absorption is, that it does not fit in with observed facts. The reason 
 is obvious. We have neglected the inertia of the electrons. True, 
 the inertia of an electron, when compared with that of familiar things, 
 is small. But when acted upon by a force whose direction changes 
 through 180 a million billion times per second, the inertia becomes 
 very considerable. 
 
 The reduction in the intensity of light in passing through material 
 media is generally thought of as a reasonance effect. Resonance is 
 a phenomenon encountered in every branch of physics. A massive 
 pendulum can be set into violent motion by a tiny force if applied 
 
: .;< /./ /::..: 
 
 /****** 4 k* V * * 
 
 periodically at the proper time that is, if the frequency of the 
 applied force is made exactly equal to that of the natural frequency 
 of oscillation of the pendulum This process of adjusting one fre- 
 quency to that of another is known as "tuning". If one tuning fork 
 is set vibrating, another of exactly the same frequency will take up 
 the vibration and thus absorb some of the energy of the first fork, 
 while a third fork that is not so tuned will not be disturbed. The 
 wireless operator adjusts the capacity and inductance of his sending 
 apparatus until it oscillates with a certain frequency, which is the 
 frequency of the radio waves transmitted through the ether. At the 
 other end the operator must adjust the capacity and inductance of his 
 receiving set until its frequency of oscillation is exactly equal to that 
 of the oncoming radio waves; that is, he must tune his receiving set 
 until its frequency of oscillation is equal to that of the electromagnetic 
 waves propagated through space. When properly tuned it responds to 
 the signals and absorbs energy from the ethereal waves. In light we 
 have many examples of this effect. The characteristic lines of metal- 
 lic vapors may be "reversed", that is, the bright emission lines of a 
 glowing vapor are reversed and become dark absorption lines when 
 the vapor is placed in the path of a more intense light that gives a 
 continuous spectrum. Certain phenomena, however, are not easily 
 explained by resonance. 
 
 Vapors of sodium, potassium, iodine, oxides of nitrogen, etc., 
 show thousands of very fine absorption lines, whose spectral distri- 
 bution follows more or less definite laws, as observed by Deslandres 
 and others, and the emission spectrum is, broadly speaking, the com- 
 plement of the aborption spectrum. It is easy to assume that for 
 each line there is a set of (resonators or oscillators whose frequency 
 is identical with that of the spectral line, and that they remove from 
 the incident radiation, the frequency corresponding to their own. But 
 this explanation is not quite sufficient. Iodine vapor illuminated with 
 monochromatic light emits a series of isolated bright lines, spaced at 
 nearly equal distances throughout the spectrum. Only one frequency 
 can be absorbed, since monochromatic light is used. According to 
 Wood the absorbed energy is then re-distributed among other fre- 
 quencies by some mechanism within the molecule. Another stumbling 
 block in the path of the resonance theory is found in the behavior of 
 Roentgen rays. According to this theory any substance should be 
 
opaque to a radiation whose frequency is the same as that of one 
 of its characteristic emission lines. It is well known, however, that 
 any substance is unusually transparent to X-rays whose frequency is 
 identical with that of one of its own characteristic radiations. Thus 
 the behavior of X-rays is diametrically opposite to that which the 
 phenomenon of resonance would lead us to expect. 
 
 Assuming the validity of the resonance theory, and certainly 
 none better has been advanced, just how does it account for the reduc- 
 tion in the intensity of light? Presumably the vibrating electron 
 extracts from the incident radiation the energy of that frequency 
 corresponding to its own, and re-radiates it in all directions. But this 
 is scattering, and not absorption. If this energy is ire-radiated in all 
 directions there will be one beam in the direction of the incident beam, 
 but, according to Planck, in the re-radiated waves there will be a lag 
 in phase of 180 degrees and consequently destructive interference. 
 There is also a beam in the opposite direction, and since there is no 
 other wave in this direction, there is no destructive interference. Con- 
 sequently we should expect to find a beam traveling backward from 
 the medium. But this is only selective reflection. 
 
 Does an electron absorb energy in quanta? Is this absorption 
 accompanied by a shift to an orbit of greater radius, with its attendent 
 increase in potential energy? Absorption, when strictly defined, 
 means a conversion of a part of the radiant energy into heat energy, 
 v/ith its consequent increase in temperature. But temperature is a 
 function of molecular velocity, that is, of the kinetic energy of the 
 molecules. How dose increase in electronic energy, that is, intra- 
 atomic energy, increase the velocity of molecules? How dose trans- 
 ference of energy between elctrons and molecules take place? Ap- 
 parently it does not take place unless it occurs during molecular 
 collision. Just what the process of transfer is, no one seems to know. 
 According to Schuster and Nagaoki, absorption is probably due to 
 sudden changes in vibration caused by molecular impacts. 
 
 According to most of the current theories of atomic structure 
 the atom consists of a number of electrons equal to the atomic number 
 of the atom, revolving in elliptic orbits about a positive nucleus. The 
 number and position of electronic orbits for a given atom under given 
 conditions is perfectly definite, but the number of orbits may be 
 exceedingly large in comparison with the number of electrons. That 
 is, an electron may shift from one orbit to another, radiating energy 
 
in definite quanta if the shift is inward, and absorbing energy if the 
 shift is outward. The shift of an electron from one orbit to the near- 
 est orbit on either side is characterized by the absorption or emission 
 of one quantum of energy. 
 
 An electron in motion is, by definition, an electric current. A 
 number of electrons revolving in coplanar orbits about a common 
 center is essentially an electromagnet. Two atoms would naturally 
 attract or trepel each other depending upon their orientation. If two 
 or more atoms combine there will certainly be a rearrangement of the 
 magneic lines of force. According to Humphreys, the electromagnetic 
 "force fields" of the individual atoms "condense" with the emission 
 of energy, to form the force fields of the resultant molecule. 
 
 What effect does the union of two or more atoms in the forma- 
 tion of a molecule have upon the characteristic frequencies of the 
 atoms? Perhaps the best answer has been given by Baly. According 
 to him, when two atoms unite to form a molecule, the individual 
 atoms may still vibrate with their characteristic frequencies, but added 
 to these are frequencies which are characteristic of the molecules; 
 these frequencies being the least common multiple of the atomic fre- 
 quencies. In a more complex molecule we would have the fre- 
 quencies characteristic of the individual atoms, frequencies character- 
 istic of atomic groups within the molecule, and frequencies char- 
 acteristic of the molecule. 
 
 The width of an absorption band is, according to Baly, due to 
 subsidiary frequencies. According to Bjerrum (Zeitsch. Electrochem. 
 17, 731, (1911); ibid, 18, 101. (1912); Nernst, Festschrift, p. 90, 
 (1912).) the subsidiary frequencies are due to the rotation of the 
 molecules. If a molecule has a frequency of vibration v, and a 
 frequency of rotation u, then it will exhibit frequencies v plus u and 
 v minus u as well as v. Since the rotational frequencies of different 
 molecules are nturally different, we should expect the absorption 
 bands of any element to change with its chemical composition. 
 
 If an atom consists of a system of coplaner irings of electrons, 
 each of definite mass, revolving in the same sense about a positive 
 nucleus, then the atom, due to its configuration, will have a certain 
 moment of inertia, and on account of the revolution of the electrons, 
 will possess a certain angular momentum. It is easy to see that the 
 atom will function as a gyroscope. Rotation of the atom, in the 
 ordinary sense, will not occur. Molecular and atomic impacts will, in 
 
general, be followed by precessional vibrations, or "wabbles", as 
 when an ordinary gyroscope is struck a sharp blow. The frequencies 
 of these vibrations will be a function of the moment of inertia of 
 the atom, which, in turn, is a function of the number and distribution 
 of the electrons in their orbits. The moment of inertia of a neutral 
 atom will be different from that of an ionized atom, and the moment 
 of inertia of a free atom will be quite different from that of the same 
 atom in chemical combination, because the configuration of the elec- 
 trons will be different. 
 
 If we assume the validity of the principle of equipartition of 
 energy, then from a study of the experimental values of the molecular 
 heats of gases, we must conclude that diatomic gas molecules possess 
 only five degrees of freedom; that is, one degree of freedom is ap- 
 parently missing. There are, of course, three degrees of freedom 
 of translation, which leaves only two to be accounted for. Kruger 
 has interpreted this fact as evidence in favor of his theory of pre- 
 cessional vibrations, which corresponds to two degrees of freedom. 
 These precessional vibrations are, according to Kruger (Annal. der 
 Physik. IV, 50, 346; ibid, 51, 450, 1916), responsible for the 
 subsidiary frequencies. 
 
 According to Baly, the subsidiary frequencies associated with a 
 given true molecular frequency may be represented by the expression 
 M ^nl mA, where M is the particular molecular frequency concern- 
 ed, I is one or the other of the intra-molecular frequencies, A stands 
 for an atomic frequency, while m and n stand for the numbers 
 0, 1,2, 3, 4, etc. The center of the band will be found by making 
 m = and n = 0. This will result in the entire group of subsidiary 
 frequencies which are associated with a given molecular frequency, 
 being divided into sub-groups. Thus for a gas we should expect the 
 absorption curve to consist of a series of isolated peaks arranged in 
 groups, the whole being more or less symmetrically placed with 
 respect to the molecular frequency M. The various peaks will de- 
 crease in height as they irecede from the center of the band. If the 
 envelope of the peaks is drawn, we should have a smooth curve fall- 
 ing off on either side of the frequency M. On account of the crowding 
 of the molecules in a liquid, the vibrations and oscillations are more 
 or less forced, so that absorption does not consist of sharp lines, but 
 is more or less continuous. What we get, then, is a band whose 
 curve is approximately that of the envelope mentioned above. 
 
It is thus evident that chemical action that is, regrouping of 
 atoms may exert a very great influence upon absorption of light if 
 the atoms, either individually or in combination, have characteristic 
 frequencies which fall within the visible spectrum. As an example 
 of this effect, it is well known that the introduction of certain colorless 
 compounds, known as chromophores, result in colored compounds. 
 
 One quantum of energy absorbed at the higher frequency might 
 be re-radiated at the same frequency (resonance radiation), or as a 
 whole number of quanta at the lower frequencies. It seems reason- 
 able to suppose that the reverse process might also take place; that 
 is, several quanta absorbed at the lower frequencies might combine 
 according to the principle of the least common multiple and be given 
 out again as a single quantum at a higher frequency. The true 
 molecular frequency is the convergence frequency of the atomic 
 frequency series. By means of this principle of the combination and 
 resolution of frequencies, Baly has apparently offered a simple ex- 
 planation of fluorescence and phosphorescence. It is apparently 
 similair to the analysis of a wave by means of a Fourier's series. 
 
 According to Lambert's law, which holds strictly within the limit 
 of experimental error for monochromatic light of any wave-length, 
 layers of any homogeneous substance of equal thickness absorb the 
 same fraction of the light incident upon them. In other words, there 
 is an exponential relation between the intensity of the transmitted 
 light and the thickness of the absorbing medium. According to Beer's 
 law the absorption of light by a fluid of definite thickness is directly 
 proportional to the concentration, or, stated another way, the absorp- 
 tion is constant so long as the product of the concentration and thick- 
 ness is kept constant. Beer's law holds approximately in a large num- 
 ber of cases and fails completely in perhaps a larger number. It is 
 well known that the position, intensity and width of the absorption 
 bands of many solutions vary greatly with the concentration as well 
 as with the nature of the solvent, even when the product of con- 
 centration and thickness is kept constant. In other words, the so- 
 called extinction coefficient is not independent of the concentration. 
 And further, if to a solution of a colored salt, say ferric chloride, a 
 solution of a colorless salt, as ammonium chloride, is added, a pro- 
 found change in the absorption takes place. 
 
 Jones has explained these changes in terms of his "Solvate 
 Theory" ; or, rather, he has interpreted these phenomena as one link 
 
in the long chain of evidence in favor of his solvate theory. Accord- 
 ing to this theory when a salt goes into solution, both the ions and 
 the molecules of the salt become "solvated", that is, form loose ag- 
 gregates with the solvent. When copper chloride, for instance, is 
 dissolved in water, both the copper ions and the molecules of copper 
 chloride are attached to large numbers of molecules of water. The 
 frequency of vibration of the resonators of these aggregates is natur- 
 ally different from that of the free molecule of the salt or of the 
 copper ion. The number and size of these aggregates depend upon 
 the concentration, consequently the extinction coefficient is not inde- 
 pendent of the concentration. According to this conception, if a 
 colorless salt, as calcium chloride, is now added, it proceeds to form 
 aggregates with the molecules of the water, and, in general, will rob 
 the copper ions and molecules of some of their water of solvation. 
 The resultant effect, of course, will be a change in absorption. 
 
 Since our knowledge of the many variables, both physical and 
 chemical, which enter into this problem is so slight, and since the 
 investigations of Jones and his co-workers were so extensive, the 
 writer has no desire to seriously question the validity of the solvate 
 theory as applied to absorption phenomena, but should like to venture 
 an alternative explanation which seems more simple and perhaps 
 clearer. It is well-known that the degree of dissociation of an elec- 
 trolyte is a function of the concentration; the more dilute the solu- 
 tion, the more nearly complete is the dissociation. We might well 
 expect, also, that the absorption of light by a salt in the ionic state 
 is, or may be, quite different from what it is in the molecular state; 
 whether the reason for it is that given by Baly, by Bjerrum, by 
 Kruger, or some other. If we cannot accept this assumption, we can 
 see no reason for accepting Jones's assumption that the absorption 
 is influenced by the state of aggregation. On the assumption, then, 
 that the absorption is a function of the dissociation, it is easy to see 
 why Beer's law does not hold. 
 
 If to an aqueous solution of a colored salt, as copper chloride, 
 in a definite state of dissociation, we add a certain amount of sodium 
 chloride, keeping the concentration of the copper chloride, in gram 
 molecules per liter, constant, we should naturally expect that some 
 of the copper ions would be driven back into the molecular state, be- 
 cause of the increase in concentration of the chlorine ions. This 
 statement may be justified by the law of mass action of Guldberg and 
 
10 
 
 Waage, or by the principle of van't Hoff that molecules and ions in 
 solution behave as gases and obey all the gas laws. This seems to 
 explain, at least as well as the solvate theory, the changes in absorp- 
 tion which take place when a colorless salt is added to a colored one 
 in a binary solution. 
 
 It is well known that increase in temperature is generally ac- 
 companied by marked changes in the absorption of light. The ab- 
 sorption band usually broadens and deepens and sometimes shifts to 
 the right. Jones attributes this to the splitting up of the molecular 
 aggregates and the formation of simpler complexes. It seems that 
 this effect could be readily explained by changes in ionization with 
 changes in temperature. Bohr, in order to explain the fact that many 
 lines of the Balmer series of hydrogen are observed in the spectra 
 of the heavenly bodies which are not found in the laboratory, ad- 
 vanced the hypothesis that the electrons do not occupy the outer 
 orbits except when the pressure is vanishingly small ; in other words, 
 the pressure exerted by neighboring molecules restrain the electrons 
 from occupying the outermost orbits. This explanation is accepted 
 as rational by several subsequent writers. Now if the pressure of the 
 other molecules exerts an inward force on the electronic orbits, then, 
 since action and reaction are equal and opposite, it is reasonable to 
 suppose that the electrons exert a pressure tending to keep the mole- 
 cules, and in many cases the atoms of the molecules apart. If the 
 absorption of energy by the electrons is accompanied by shifts to 
 orbits of greater radii, we might well expect that, in general, the ab- 
 sorption of energy by the electrons would tend to increase dissociation 
 in those molecules where the electrons of all the atoms are not revolv- 
 ing in orbits about a single axis. According to this line of reasoning, 
 increase in temperature would probably be accompanied by increased 
 dissociation. Very often, however, other forces may be brought into 
 play which tend to accelerate the reverse process. 
 
 According to Jones, however, the above conclusions are wrong. 
 
 In discussing this phase of the subject he says : "Rise in tempera- 
 ture not only does not increase the number of ions present, but, as is 
 well known, diminishes dissociation." It is doubtful, however, if this 
 sweeping statement can be justified. Walker says, for instance: 
 "Since the heat of dissociation into ions is sometimes positive, some- 
 times negative, a rise of temperature may in some cases be accomp- 
 
11 
 
 anied by increased dissociation, in others by diminished dissociation." 
 This seems to be in accordance with Le Chatelier's theorem. 
 
 It seems quite probable, however, that some of the changes 
 which take place are far too complex to be explained either by the 
 solvate theory of Jones or by the changes in ionization. There are 
 probably secondary reactions which play an important role. For 
 instance, Baly (Phil. Mag. 27, p 632, 1914) found that in certain 
 reactions the solutions passed through intermediate phases which 
 could be traced by means of changes in the absorption bands. A 
 great many electrolytes, on going into solution, react chemically with 
 water. For instance, when ferric chloride goes into aqueous solution, 
 one or more of the chlorine atoms may be replaced by OH radicals 
 from the water with the formation of a basic ferric salt or the ferric 
 hydroxide and hydrochloric acid. The ferric hydroxide is brick red 
 in color. A colorless salt, as aluminum chloride, when added to the 
 above solution, may have a catalytic effect in bringing about these 
 reactions. Furthermore, the effect would undoubtedly be increased 
 by heating, as hydrolysis is promoted by rise of temperature. 
 
 Jones's work in the visible and ultra-violet was done by the 
 method of spectro-photography. This method, while possessing many 
 advantages, has the disadvantage that the results are qualitative only. 
 Changes in the width and position of an absorption band can be 
 measured, and some notion of the relative intensities of different 
 bands can be obtained from the blackness of the photographs, but no 
 information concerning the shape of the absorption curve is possible. 
 
 The plan of the present work was to repeat some of the work 
 of Jones, making use of the method of spectro-photometry, thus 
 making the results quantitative, and to extend the work to other 
 salts. In particular it was desired to study the effect of those color- 
 less salts which do not form hydrates, upon the absorption of colored 
 salts, and the effect of one colored salt upon another. It was also 
 planned to study the effect of temperature on solutions of certain 
 salts and mixtures of different salts. In addition, it was desired to 
 measure the absorption of certain organic compounds. No attempt 
 has been made to investigate Beer's law. 
 
12 
 
 APPARATUS AND METHOD. 
 
 The method of the present work was that of spectro-photometry. 
 The results are quantitative, although the method does not adapt 
 itself to a high degre of precision. The instrument consists of a 
 Hilger spectrometer with a Nutting photometer box. A sketch of the 
 apparatus, showing in detail the optical parts of the photometer box, 
 is given in plate 1 . The light source is at S. A 40 watt tungsten lamp 
 was used. A Nernst glower or other line source might have been 
 better, but the lamp used gave very good results. In the earlier part 
 of the work direct current from the storage battery was used for 
 illumination, but later on ordinary A. C. power was used. The ap- 
 paratus functioned about as well in the latter case as in the former. 
 Two beams of light, rendered parallel by the wedge-shaped prisms, 
 pass through the photometer box. E is a glass cell about 1 cm. thick, 
 2 cms. wide by 2J/2 cms. deep which contained the solution whose 
 absorption was to be measured. D is an exactly similar cell which in 
 general contained either distilled water or an aqueous solution of a 
 colorless salt. 
 
 In order to obtain even fair results, certain precautions were 
 found to be necessary. The illumination must be neither too faint nor 
 too brilliant. At either extremity of the spectrum the illumination 
 was too feeble for good results. Where the illumination was too in- 
 tense, it was cut down by means of a variable resistance. The best 
 results were obtained with a rather faintly illuminated field. The eye 
 should be in good condition, rested, and not recently exposed to a 
 strong light. It is useless to try to take readings immediately after a 
 walk in the bright sunlight. As far as possible all light was eliminated 
 except that passing through the telescope. The heating effect from 
 the lamp was largely eliminated by placing a large cell full of water 
 in front of the cells D and E. This cell was broad and deep and 
 about 3cms. thick. Considerable care was taken to keep the solution 
 free from undissolved particles in suspension. If, on close inspection, 
 there was the slightest indication of murkiness, the solution was 
 filtered. In the case of the salts of chromium, the filtration was re- 
 peated many times. The wave-length readings on the drum was 
 occasionally checked against a known wave-length, as the prism P 
 sometimes became slightly displaced. 
 
 To obtain the absorption of a simple solution the following 
 
13 
 
 method was used: a definite weighed quantity of the salt was dis- 
 solved in distilled water and then water added to make a definite 
 volume of the solution. This same concentration was always used 
 when dealing with the same colored salt. The cell E was filled with 
 the colored solution. The cell D was filled with distilled water. The 
 interval between the readings ordinarily vairied from 50 to 200 A. U. 
 The number of readings for any given wave length varied from three 
 to eight. 
 
 To measure the absorption of a mixed solution of a colored salt 
 with a colorless one, the following plan was used: the two salts, 
 after having been weighed, were mixed together. Sufficient distilled 
 water was added to dissolve both salts. Then water was added in 
 sufficient quantity to make the desired volume of solution. The cell 
 E was filled with this solution. An aqueous solution of the colorless 
 salt was then made by dissolving the same quantity of the colorless 
 salt in the same volume of solution. This solution was placed in the 
 Cell D. 
 
 In the case of two clored salts, as for instance cobalt chloride 
 and copper chloride, the procedure was as follows : the solution of the 
 two salts, each having the same concentration as when its absorption 
 was determined alone (that is, the same number of molecules per 
 unit volume), was placed in the cell E. In the cell D was placed an 
 aquous solution of copper chloride of the same concentration as that 
 in the ceil E. The absorption of the copper chloride in the cell D 
 should balance that of the same salt in the cell E. The readings should 
 then indicate only the absorption of the cobalt chloride plus whatever 
 change was brought about by the influence of the one salt upon 
 the other. 
 
 When dealing with a mixture of three colored salts, still another 
 plan was followed. The solution of the three salts, each having the 
 same concentration as before, was placed in the cell E. In D was 
 placed distilled water. The absorption was measured throughout the 
 visible spectrum and the curve plotted. But in this case the extinction 
 coefficient was not determined, the extinction coefficient of a mixture 
 apparently having no meaning. The curve was plotted in instrument 
 readings; that is, in terms of the logarithm of the reciprocal of the 
 square of the cosin of the angle of rotation of the analyser. The 
 comparison curve was obtained by adding together the absorption 
 curves of the three salts as determined separately. The difference 
 
14 
 
 between these two curves shows the changes due to the influence of the 
 different salts on each other. Two sets of readings are given in 
 Table I, copied directly from the original records. In the first 
 column are recorded wave-lengths in millimicrons and in the second 
 the corresponding readings of the instrument. It may be observed 
 that where the solution is nearly transparent, and therefore the read- 
 ings very small, the percentage of error is liable to be quite large. 
 
 CONCENTRATIONS. 
 
 In this series of experiments the same concentration was always 
 used with the same colored salt. In preparing the solution, the salt 
 was always weighed to within one tenth of one per cent. Consider- 
 able care was taken to free the salt from excess moisture. The con- 
 centration was determined in gram molecules per liter, and are as 
 follows: Copper chloride, .2742; copper nitrate, .1645; copper 
 sulphate, .1720; nickel chloride, .342; nickel nitrate, .2286; nickel 
 sulphate, .3202 ; cobalt chloride, . 1 1 39 ; cobalt nitrate, . 1 600 ; cobalt 
 sulphate, .2025; ferric chloride, .1560; ferric nitrate, .085; 
 chromium chloride, .03396; chromium nitrate, .01067; uranyl 
 chloride, .727. The concentration of the colorless salts was not kept 
 strictly constant throughout, and will be taken up in the discussion 
 of individual cases. 
 
 MIXTURE OF COPPER CHLORIDE WITH OTHER CHLORIDES. 
 
 With cerium chloride. The concentration of the cerium chloride 
 was .60. The curves are shown on plate 2-e. In all these curves the 
 abscissa represents wave-length in milli-microns and in nearly all of 
 them the ordinate represents absorption in terms of the extinction 
 coefficient. The various sets of curves are displaced the one above 
 the others at convenient distances so that while the origin of 
 abscissas is the same, the origin of the ordinates is different for each 
 set. Unless otherwise specified the dotted line represents the ab- 
 sorption of the colored salt alone, while the full-line curve represents 
 that of the mixture. In this case no very great effct is shown, although 
 there is increased absorption in the red. 
 
 With zinc chloride, plate 2-d. Concentration of the zinc chlor- 
 ide= 1 .50. Increased absorption in the ired. 
 
 With sodium chloride, plate 2-c. Concentration of the sodium 
 
15 
 
 chloride=2.8. Increased absorption in the violet, decrease from 
 wave-length 530 to 615, and marked increase in the red. The equa- 
 tion of the curve has been determined as shown in the graph. 
 
 With potassium chloride, plate 2-b. Concentration of potas- 
 sium chloride =2. 2. Marked increase in absorption in the red. The 
 curve is found to follow an exponential law, as shown in the graph. 
 
 With ammonium chloride, plate 2-a. Concentration of am- 
 mounium chloride = 3.1. The mixture shows very great absorption 
 in the violet, although the copper chloride is transparent in that region 
 of the spectrum. There is decrease in the green and yellow and in- 
 crease in the red. The equation has been determined and found to 
 be expotential. 
 
 With ferric chloride, plate 6-a. No very pronounced influence. 
 
 COBALT CHLORIDE WITH OTHER CHLORIDES. 
 
 With sodium chloride, plate 3-e. Concentration of sodium 
 chloride = 2.8. Very considerable increase in the green. 
 
 With potassium chloride, plate 3-d. Concentration of KC1 = 2.2. 
 Very greatly increased absorption in the green. 
 
 With cerium chloride, plate 3-c. Concentration of the cerium 
 chloride = .60. Increased absorption throughout the spectrum. 
 
 Cobalt chloride with zinc chloride, plate 3-b. Concentration of 
 zinc chloride = 1.5. Considerable increase in the green; increase 
 throughout except in the yellow. 
 
 With ammonium chloride, plate 3-a. Concentration of am- 
 monium chloride = 2.8. Very great increase in absorption in the 
 green. It may be noticed that ammonium chloride, which does not 
 form hydrates, produces a greater change in the absorption of cobalt 
 chloride than either zinc chloride, sodium chloride, or lithium chloride, 
 all of which form hydrates. 
 
 With lithium chloride, plate 16-b. Concentration of lithium 
 chloride = 3.9. Considerable absorption except in the violet and blue. 
 
 With ferric chloride, plate 4-a. Marked increase in absorption 
 in the blue and considerable increase in the region of the long wave 
 lengths. 
 
 With chromium chloride, plate 4-b. Broadening of the principle 
 absorption band in the blue and green, and increased absorption in 
 the red. 
 
16 
 
 With nickel chloride, plate 4-c. Increased absorption through- 
 out, but especially at the red end of the spectrum. 
 
 With copper chloride, plate 4-d. Decrease in the yellow and 
 increase in the red. 
 
 NICKEL CHLODIDE WITH OTHER CHLORIDES. 
 
 With sodium chloride, plate 5-d. Concentration of sodium 
 chloride = 2.75. Decrease in absorption throughout the visible 
 spectrum. 
 
 With potassium chloride, plate 5-c. Concentration of potassium 
 chloride = 2.00. Marked decrease in absorption for the middle of 
 the spectrum. 
 
 With cerium chloride, plate 5-b. Concentration of cerium 
 chloride = .60. Marked increase in transparency in the blue, green 
 and orange. 
 
 With lithium chloride, plate 5-e. Concentration of LiCl 4.0. 
 There is a very great increase in absorption throughout the spectrum. 
 
 With copper chloride, plate 6-c. . There is a rather large in- 
 crease in absorption in the gnreen. 
 
 With ferric chloride, plate 6-b. Enormous increase in the violet. 
 
 AQUEOUS SOLUTIONS OF FERRIC CHLORIDE 
 WITH OTHER CHLORIDES. 
 
 With lithium chloride, plate 7-c. Concentration of LiCl = : 1 .8. 
 Greatly increased absorption in the region of the short wave lengths. 
 
 With calcium chloride, plate 7-b. Concentration of calcium 
 chloride = 1 .00. Great increase in absorption in the violet and blue. 
 
 With aluminum chloride, plate 7-a. Concentration of aluminum 
 chloride = 0.340. This is the most striking example of color change 
 produced by the addition of a colorless salt to a colored one that has 
 been met with in this series of experiments. The change in color is 
 from that of a clear amber to a brick red. Undoubtedly there is 
 formation of ferric hydroxide. 
 
 With ammonium chloride, plate 8-a. Concentration of the am- 
 monium chloride = 2.80. There is tremendous increase in absorption 
 in the violet and blue, with considerable decrease in the green. At 
 wave-length 480 the increases is more than 200%. This curve is 
 found to be exponential, as shown in the graph. 
 
17 
 
 With zinc chloride, plate 8-b. Concentration of zinc chloride 
 = 1 .30. Enormous increase in absorption in the violet, and decrease 
 in the green. The curve is exponential in form. 
 
 With sodium chloride, plate 9-c. Concentration of NaCl = 2.60. 
 There is enormous increase in absorption for the shorter wave lengths. 
 
 With potassium chloride, plate 9-b. Concentration of KC1= : 2.00. 
 Great increase in absorption in the violet and blue. 
 
 With cerium chloride, plate 9-a. Concentration of cerium chloride 
 = 0.60. There is very great increase in absorption in the violet. 
 
 MIXTURES OF CHROMIUM CHLORIDE WITH 
 OTHER CHLORIDES. 
 
 With sodium chloride, plate 10-f. Concentration of NaCl=3.00. 
 There is increased absorption throughout the spectrum. 
 
 With potassium chloride, plate 10-e. Concentration of KC1 == 
 2.50. There is increased absorption throughout the spectrum. 
 
 With aluminium chloride, plate 10-d. Concentration of the 
 aluminium chloride = 0.30. There is increase in the absorption up to 
 wave-length 590, beyond which there is decrease. 
 
 With calcium chloride, plate 1 0-c. Concentration of the calcium 
 chloride = 1 .50. There is scarcely any change in absorption. This 
 fact seems to be widely at variance with the solvate theory, for 
 calcium chloride, which has an enormous affinity for water, should 
 produce a great change in the absorption. 
 
 With zinc chloride, plate 10-b. Concentration of zinc chloride 
 = 1 .60. No very decided change in absorption. 
 
 With ammonium chloride, plate 10-a. Concentration of am- 
 monium chloride 2.8. Decreased absorption throughout. 
 
 With copper chloride, plate 1 1 -d. There is increase in absorp- 
 tion for the shorter wave-lengths. 
 
 With nickel chloride, plate 1 1 -c. There is increase in absorp- 
 tion in the violet and decrease from wave-length 540 to 660. 
 
 With ferric chloride, plate 1 1 -b. There is increase in absorption 
 throughout. 
 
 With cerium chloride, plate 1 1-a. There is scarcely any change 
 in absorption. 
 
18 
 
 URYANYL CHLORIDE WITH OTHER CHLORIDES. 
 
 With cerium chloride, plate 1 2-e. The curves show considerable 
 increase in the violet. 
 
 With potassium chloride, plate 1 2-d. Concentration of KC1=2.30. 
 There is very greatly increased absorption in the violet and blue. 
 
 With aluminium chloride, plate 12-c. Concentration of alum- 
 inum chloride = 0.30. There is great increase in absorption for the 
 shorter wave lengths. 
 
 With zinc chloride, plate 12-b. Concentration of zinc chloride 
 1.60. There is decrease in absorption throughout. 
 
 With sodium chloride, plate 1 2-a. Concentration of NaCl=3.00. 
 The curve shows very great increase in absorption for the shorter 
 wave lengths. 
 
 With copper chloride, plate 13-d. There is very little change 
 except in the red where there is considerable increase in absorption. 
 
 With nickel chloride, plate 13-c. There is no very decided 
 change in absorption. 
 
 With cobalt chloride, plate 13-b. There seems to be a slight 
 increase in absorption for most of the visible spectrum. 
 
 With calcium chloride, plate 13-a. Concentration of calcium 
 chloride = 1.50. There is a very lairge and general increase in 
 absorption throughout the spectrum. 
 
 MIXTURES OF THREE COLORED CHLORIDES. 
 
 Copper chloride, cobalt chloride and chromium chloride, plate 
 14-c. There is increase in absorption from 440 to 550, decrease from 
 550 to 670 and increase the Test of the way. 
 
 Copper chloride, nickel chloride, and chromium chloride, plate 
 1 4-b. Just as in the preceding case there is increased absorption from 
 440 to 550, decreased absorption from 550 to 650 and an increase 
 from 650 to the end of the visible spectrum. 
 
 Copper chloride, ferric chloride, and sodium chloride, plate 1 4-a. 
 There is an enormous increase in absorption at both ends of the 
 spectrum with no change whatever in the center. An exponential 
 curve was determined which fits the experimental curve almost 
 prfectly. 
 
 Copper chloride, cobalt chloride, and ferric chloride, plate 1 5-a. 
 
19 
 
 There is considerable increase in absorption at both ends of the spec- 
 trum, with a rather marked decrease from 560 to 640. 
 
 Copper chloride, chromium chloride, and ferric chloride, plate 
 1 5-b. There is very great increase in the violet, considerable increase 
 in the blue, decrease in the yellow, and increase in the red. 
 
 Cobalt chloride, nickel chloride and chromium chloride, plate 
 15-c. There is considerable decrease in absorption between wave- 
 lengths 540 and 640. 
 
 Chromium chloride, cobalt chloride, and ferric chloride, place 
 16-a. There is rather marked increase in absorption for the shorter 
 wave lengths. 
 
 MIXTURES OF COPPER NITRATE WITH OTHER NITRATES. 
 
 With aluminum nitrate, plate 1 7-c. Concentration of aluminum 
 nitrate was .44. The mixed solution is more transparent than that 
 of copper nitrate for practically all of the visible spectrum. 
 
 With calcium nitrate, plate 17-b. The concentration of calcium 
 nitrace was 0.70. The mixed solution is more transparent than the 
 solution of copper nitrate from wave-length 540 to 640, while for 
 longer wave-lengths it is more opaque. 
 
 With sodium nitrate, plate 17-a. The influence of the sodium 
 nitrate manifests itself as an increase in transparency between wave- 
 length 540 and 640. 
 
 With lithium nitrate, plate 17-d. Concentration of lithium 
 nitrate = 2.40. The change consists in an increase in absorption 
 from wave-length 560 to the red end of the visible spectrum. 
 
 With ammonium nitrate, plate 1 7-e. Concentration of the am- 
 monium nitrate = 1.85. There is increased absorption in the region 
 of the longer wave-lengths. 
 
 COBALT NITRATE WITH OTHER NITRATES. 
 
 With ammonium nitrate, plate 18-a. Concentration of ammon- 
 ium nitrate = 1.85. The influence of the ammonium nitrate mani- 
 fests itself as an increase in transparency throughout the visible spec- 
 trum, the greatest change being at about wave-length 600. 
 
 With sodium nitrate, plate 18-b. Concentration of sodium 
 nitrate = 1.90. There is decrease in absorption from wave-length 
 480 to the red end of the spectrum. 
 
20 
 
 With calcium nitrate, plate 18-c. Concentration of calcium 
 nitrate = 1.50. It will be seen that there is a general increase in 
 transparency throughout the spectrum. 
 
 With aluminum nitrate, plate 18-d. Concentration of aluminum 
 nitrate = .440. There is a slight decrease in absorption throughout 
 the spectrum. 
 
 With lithium nitrate, plate 19-a. Concentration of lithium 
 nitrate == 2.400. There is some increase in absorption in the violet, 
 blue, green, and red. 
 
 With copper nitrate, plate 21 -a. There is greatly increased 
 transparency from wave-length 570 to the red end of the spectrum. 
 
 MIXTURES OF NICKEL NITRATE WITH OTHER NITRATES. 
 
 With ammonium nitrate, plate 20-a. Concentration of am- 
 monium nitrate 1.850. It will be observed that there is a re- 
 markable increase in transparency throughout the spectrum. The 
 mixed solution is transparent as water to wave-length 580. 
 
 With calcium nitrate, plate 20-b. Concentration of calcium 
 nitrate = .700. There is a general decrease in absorption except for 
 a small region in the green. 
 
 With aluminum nitrate, plate 20-c. Concentration of aluminum 
 nitrate = .440. There is a considerable increase in absorption in the 
 green and decrease in the red. 
 
 With lithium nitrate, plate 20-d. Concentration of lithium 
 nitrate = 2.40. The only striking change is an increase in trans- 
 parency from wave-length 550 to 640. 
 
 With sodium nitrate, plate 20-e. Concentration of sodium nitrate 
 = 1.90. Increased transparency in the yellow and red. 
 
 With cobalt nitrate, plate 21-b. There is an enormous increase 
 in transparency for almost the entire length of the spectrum. 
 
 With copper nitrate, plate 21-c. The curves show decrease in 
 absorption in the blue, increase in the green, decrease in the yellow 
 and enormous decrease in the red. 
 
 MIXTURES OF FERRIC NITRATE WITH OTHER NITRATES. 
 
 With calcium nitrate, plate 22-a. Concentration of calcium 
 nitrate = .700. There is marked increase in absorption in the region 
 of the shorter wave-lengths. 
 
21 
 
 With lithium nitrate, plate 22-b. Concentration of lithium 
 nitrate = 2.40. There is very little change in absorption except from 
 wave-length 520 to 610, where there is increased transparency, 
 
 With sodium nitrate, plate 22-c. Concentration of sodium nitrate 
 1 .90. There is considerable increase in transparency in the middle 
 of the spectrum and a general but smaller increase for the longer 
 wave-lengths. 
 
 With aluminium nitrate, plate 22-d. Concentration of aluminum 
 nitrate = .440. Slightly increased transparency throughout. 
 
 With ammonium nitrate, plate 22-e. Concentration of am- 
 monium nitrate = 1.850. Very great increase in transparency 
 throughout the visible spectrum. 
 
 With cobalt nitrate, plate 23-a. There is enormous increase in 
 transparency especially in the middle of the spectrum. 
 
 With nickel nitrate., plate 23-b. There is a very great increase 
 in transparency at about wave-length 600, and a smaller general 
 increase throughout. 
 
 With copper nitrate, plate 23-c. There is very great increase 
 in transparency from wave-length 540 to 660, and a smaller increase 
 in the region of the shorter wave-lengths. 
 
 MIXTURES OF CHROMIUM NITRATE WITH OTHER NITRATES. 
 
 With aluminium nitrate, plate 24-a. Concentration of aluminum 
 nitrate = 0.440. There is very great decrease in absorption in the 
 violet, the decrease growing gradually smaller as the wave-length 
 increases. 
 
 With calcium nitrate, plate 24-b. Concentration of calcium 
 nitrate = 0.500. There is considerable increase in transparency in 
 the region of the shorter wave-lengths. 
 
 With lithium nitrate, plate 24-c. Concentration of lithium ni- 
 trate 2.400. There is great decrease in absorption in the violet, 
 blue, and green. 
 
 With sodium nitrate, plate 24-d. Concentration of sodium ni- 
 trate = 2.300. The surves show decrease in absorption in the region 
 of the shorter wave-lengths. 
 
 With cobalt nitrate, plate 25-a. There is a general and very 
 great decrease in absorption, the greatest effect being found in the 
 region of the shorter wave-lengths. 
 
22 
 
 With nickel nitrate, plate 25-b. The decrease in absorption is 
 very pronounced throughout the visible spectrum. 
 
 With copper nitrate, plate 25-c. Very great increase in trans- 
 parency is shown throughout the spectrum. 
 
 DIDYMIUM NITRATE WITH ALUMINIUM NITRATE. 
 
 Two and 728-1000 grams of didymium nitrate were dissolved in 
 sufficient distilled water to make 7.5 c. c. of solution. The concentra- 
 tion of the aluminium nitrate was 0.400. The absorption was meas- 
 ured and the curve plotted as shown in plate 36-b. The aluminium 
 nitrate seems to produce no appreciable effect on the absorption of 
 the didymium nitrate. The curve of the mixture practically coincides 
 with that of the simple salt. It will be noticed that there are many 
 sharp bands, the most prominent having their maxima at about wave- 
 lengths 445,423, 578.5, and 744. The extinction coefficient was 
 not plotted. The ordinates are plotted in terms of the readings of the 
 instrument. 
 
 MIXTURES OF THREE COLORED NITRATES. 
 
 Copper nitrate, nickel nitrate, and ferric nitrate, plate 26-a. 
 There is a small decrease in absorption in the violet, blue, and green, 
 and a very pronounced decrease for the remainder of the spectrum. 
 
 Copper nitrate, cobalt nitrate, and ferric nitrate, plate 26-b. 
 There is greatly increased transparency throughout, the greatest change 
 coming in the middle of the spectrum. 
 
 Nickel nitrate, cobalt nitrate, and ferric nitrate, plate 26-c. 
 Again there is a very great decrease in absorption throughout the spec- 
 trum, with the greatest decrease coming at about wave-length 580. 
 
 MIXTURES OF COPPER SULPHATE WITH OTHER SULPHATES. 
 
 With aluminium sulphate, plate 27-a. The absorption curve 
 shows very little deviation from that of the copper sulphate alone. 
 Concentration of aluminium sulphate = 0.270. 
 
 With nickel sulphate, plate 27-b. There is considerable decrease 
 in absorption for the longer wave-lengths. 
 
 With cobalt sulphate, plate 27-c. There is increase in absorp- 
 tion in the violet and blue, and decrease beyond wave-length 560. 
 
 With postasshim sulphate, plate 27-d. Concentration of potas- 
 
23 
 
 sium sulphate = 0.5 1 0. There is no noticeable change in absorption 
 whatever. 
 
 With ammonium sulphate, plate 27-e. Concentration of am- 
 tionmonium sulphate = 1 .500. There is no very pronounced change 
 in absorption. 
 
 MIXTURES OF COBALT SULPHATE WITH OTHER SULPHATES. 
 
 With aluminium sulphate, plate 28-a. . .Concentration of alu- 
 minium sulphate = 0.270. The curves show some increase in absorp- 
 tion in the violet and blue and some decrease in the red. 
 
 With ammonium sulphate, plate 28-b. Concentration of am- 
 monium sulphate = 1.250. There is slight increase in the violet 
 and blue. 
 
 With nickel sulphate, plate 28-c. There is a slight increase for 
 the shorter wave-lengths and decrease for the longer. 
 
 With potassium sulphate, plate 28-d. Concentration of potas- 
 sium sulphate = 0.510. There is a general increase in absorption 
 throughout the spectrum. 
 
 THE EFFECT OF TEMPERATURE UPON ABSORPTION. 
 
 Far the investigation of the effect of temperature upon absorp- 
 tion, an electric furnace of inside dimensions 1 .9 x 8.6 x 5.7 cms. was 
 constructed. It was made of brass and covered with two or three 
 layers of asbestos paper over which was wound about 20 feet of man- 
 ganin wire having a total resistance of about 60 ohms. This in turn 
 was covered with a paste made of magnesium oxide and sodium sili- 
 cate, so that the total thickness of covering was about one centimeter. 
 In order to allow the light to pass through, two holes of 1 .8 cm. diam- 
 eter were bored through each side. A wooden lid, boiled in beeswax 
 and containing a hole for the insertion of a mercury thermometer was 
 made for the furnace. 
 
 About this time the glass cells used as containers for the solutions 
 went to pieces. Some of the solutions attacked the cement of the 
 cells, thus setting up strains which caused the glass to break. A gnreat 
 number of cements were tried and found useless. A brass cell was 
 then constructed and gold-plated on the inside with glass plates 
 fastened on the sides by mechanical means. The hot solutions at- 
 tacked the brass through the gold, probably for the reason that the 
 
24 
 
 gold plating was not thick enough or not uniformly distributed. 
 Finally the inside of the cell was coated with de Khotinsky wax. This 
 resisted the action of the aqueous solution perfectly and also proved to 
 be a good cement for fastening the parts of the cell together. Some 
 of the readings are shown in tables III and IV. These values are 
 copied from the book just as they were recorded at the time of tak- 
 ing the observations. In the curves the absorption at the higher tem- 
 perature is always represented by the full line curve, while that at 
 room temperature is shown by the dotted curve. 
 
 MEASUREMENT OF TEMPERATURE EFFECT UPON 
 VARIOUS SOLUTIONS 
 
 Copper chloride, plate 34-d. Readings were taken at 23.5 
 C, 46.7, and 66.0 C. The increase in absorption extends from wave- 
 length 500 to the red end of the spectrum. 
 
 Copper chloride with ammonium chloride, plate 29-a. The higher 
 temperature was 67.0 C. There is tremendous increase in absorption 
 in the violet and blue, although neither the copper chloride nor the 
 ammonium chloride shows any temperature effect in that region. 
 There is also very great increase in the red. 
 
 The equations of these curves have been determined as shown in 
 the graph. For the full line curve, between wave-lengths 460 and 
 530 the b of the formula has the value -.05128 and the k of the 
 formula 5.956. Between 530 and 610, b=.0296 and k=.1479. 
 Between 610 and the ored end of the spectrum, b = .02257 and 
 k = 1.706. For the dotted line, from wave-length 440 to 500, 
 b = -.04582 and k = 5.83. From 560 to the red end of the spec- 
 trum, b = .01946 and k = .4678. 
 
 Copper chloride with sodium chloride, plate 29-b. The higher 
 temperature was at 58.5 C. The mixed solution shows tremendous 
 increase in absorption in the violet, although neither of the simple 
 solutions of which the mixture is composed shows any temperature 
 effect in that iregion. There is increased absorption throughout the 
 spectrum. The equations of these curves have been determined as 
 shown in the graph. For the full line curve between the wave-lengths 
 440 and 520 the b of the formula = -.04552 and k = 7. 1 3. From 
 wave-length 520 to the red end of the spectrum, b = .02256 and 
 k=.2144. 
 
 Copper chloride with potassium chlroide, plate 29-c. The 
 
25 
 
 higher temperature was 69.6 C. There is a remarkable increase in 
 the violet and a smaller increase throughout the spectrum. The 
 equation of the curves have been determined as shown in the graphs. 
 
 Cobalt chloride, plate 30-a. The higher temperature was 68.5 C. 
 There is great increase in absorption throughout the spectrum. 
 
 Ferric chloride, plate 32-c. Temperatures were 23.5 and 
 5 1 .0 C. There is very great increase in absorption for the shorter 
 wave-lengths. 
 
 Ferric chloride with ammonium chloride, plate 33-a. Tempera- 
 tures were 23.5 and 49.0 C. There is very great increase in ab- 
 sorption for nearly the entire length of the spectrum. 
 
 Ferric chloride with potassium chloride, plate 33-b. Tempera- 
 tures were 23.5 and 52.0 C. The curves show tremendous increase 
 in absorption for the entire spectrum. 
 
 Ferric chloride with sodium chloride, plate 33-c. Temperatures 
 were 23.5 and 48.0 C. Tremendous increase in absorption for the 
 shorter wave-lengths. 
 
 Nickel chloride. The temperatures were 23.5 and 69.7 C. 
 There was no change in the absorption whatever and the curve was 
 not plotted. 
 
 Nickel chloride wtih sodium chloride, plate 34-a. The tempera- 
 tures were 23.5 and 68.0 C. In the violet and blue the increase in 
 absorption is 400 to 600%. There is great increase throughout the 
 spectrum. 
 
 Chromium chloride. The temperatures were 23.5 and 67.1 C. 
 No change in absorption could be detected and the curve was not 
 plotted. 
 
 Chromium chloride with potassium chloride, plate 35-c. Tem- 
 peratures were 23.5 and 73.0 C. There is a rather large and gen- 
 eral increase in absorption throughout the spectrum, although the 
 chromium chloride alone shows no increase and potassium chloride is 
 transparent. 
 
 Neodymium chloride, plate 36-a. Temperatures were 23.5 
 and 73.0 C. It will be seen that there is very great increase in ab- 
 sorption in the violet and blue, and again in the yellow and red. For 
 the middle of the spectrum there is no apparent change. 
 
 Ferric nitrate, plate 31 -a. Temperatures were 23.5 and 58.2 
 C. There is very great increase in absorption in the region of the 
 shorter wave-lengths. 
 
26 
 
 Ferric nitrate with sodium nitrate, plate 31-b. Temperatures 
 were 24.0 and 61.6 C. The increase in absorption is very great, 
 being about 90% at wave-length 460. 
 
 Ferric nitrate with ammonium nitrate, plate 3 1 -c. The tempera- 
 tures were 23.2 and 57.3 C. The increase in absorption is very 
 great. 
 
 Ferric nitrate with calcium nitrate, plate 32-a. Temperatures 
 were 27.3 and 59.3 C. There is very great increase in absorption 
 in the region of the shorter wave-lengths. 
 
 Ferric nitrate with aluminium nitrate, plate 32-b. Temperatures 
 were 23.5 and 57.7 C. The curves show greatly increased absorp- 
 tion for the shorter wave-lengths, the increase at 440 being 60%. 
 
 Copper nitrate. The temperatures were 23.5 and 70.0 C. 
 No change in absorption could be detected. 
 
 Copper nitrate with aluminium nitrate. Up to 70 C no change 
 in absorption could be detected. 
 
 Copper nitrate with ammonium nitrate, plate 34-b. Tempera 
 tures were 23.5 and 69.0 C. There is very little if any change in 
 absorption. 
 
 Copper nitrate with sodium nitrate, plate 34-c. Temperatures 
 were 23.5 and 66.3 C. There is considerable increase in absorp- 
 tion for the region of the longer wave-lengths. 
 
 Cobalt nitrate. Up to 71 . C the absorption seemed to be the same 
 as at room temperature. 
 
 Cobalt nitrate with sodium nitrate, plate 30-b. The tempera- 
 tures were 23.5 and 74.4 C. There is very great increase in ab- 
 sorption throughout the visible spectrum, although neither cobalt ni- 
 thate nor sodium nitrate show any temperature effect. 
 
 Cobalt sulphate, plate 30-c. The temperatures were 23.5, 
 47.0, and 70.5 C. As the temperature rises there is a steady in- 
 crease in absorption throughout the spectrum. 
 
 Copper sulphate, plate 35-a. Temperatures were 23.5 and 
 69.0 C. There is very great increase in absorption in the red. 
 
 Copper sulphate with potassium sulphate, plate 35-b. The tem- 
 peratures were 23.5 and 71.0 C. There is increased transparency 
 at about wave-length 580 and increased absorption in the red. While 
 taking observations on this solution it was discovered that the addition 
 of sulphuric acid completely destroyed the temperature effect. 
 
27 
 
 ABSORPTION OF LIGHT BY ALCOHOUC SOLUTIONS OF 
 ORGANIC COMPOUNDS 
 
 The absorption of three compounds: azobenzene, benzene azo 
 ortho cresol acetate, and benzene azo phenol benzoate was measured.* 
 These compounds are insoluble in water, but soluble in most organic 
 solvents. Absolute alcohol was used as solvent. This introduced a 
 new difficulty, for the alcohol attacked the wax with which the cells 
 were lined. Fortunately a glass cell was found that answered the 
 purpose very well. It was thought, however, that this cell would not 
 stand a very high temperature, so no attempt was made to measure 
 the temperature effect. 
 
 Azobenzene, plate 37-a. The concentration was .01841. At 
 this concentration the solution is a deep red. The solution is practi- 
 cally opaque between wave-lengths 440 and 480. From 570 to the 
 red end of the spectrum it is quite transparent. The equation of this 
 curve was determined as shown in the graph. Unlike all the other 
 curves whose equations have been determined it is not exponential but 
 a power curve. 
 
 Benzene azo ortho cresol acetate, plate 37-b. The concentra- 
 tion was .005961 . At this concentration the solution is a beautiful red. 
 From the curve it will be noticed that there is very great absorption 
 in the violet and blue. From wave-length 560 to the red end of the 
 spectrum it is quite transparents. 
 
 Benzene azo phenol benzoate, plate 37-c. The concentration was 
 .009536. The color of the solution is that of a beautiful red. The 
 absorption curve is almost exactly the type of curve found for the 
 preceding compound, but the extinction coefficient is not so great. 
 There is, as before, very great absorption in the violet. From wave- 
 length 540 to 720 it is almost entirely transparent. 
 
 RATIO OF ABSORPTION OF MIXTURE TO SUM OF 
 ABSORPTIONS OF COMPONENT PARTS. 
 
 A critical study of the various cases discussed on the preceding 
 pages and shown graphically on the sheets that follow, brings to light 
 a very curious fact. The ratio of the absorption of the mixture to 
 the sum of the absorptions of the component parts has a minimum 
 
 "Kindly furnished by Doctor Boord, of the Chemistry Department. 
 
28 
 
 somewhere between wave-lengths 510 and 610 milli-microns, the 
 average position of this minimum being about 580. Or, expressed 
 another way, if the ordinate of the full line curve is divided by the or- 
 dinate of the dotted curve, and if the ratio thus determined is plotted 
 as a function of the wave-length, the resulting curve shows a minimum 
 near the middle of the visible spectrum. Some of the most striking 
 examples are shown on plate 38. This tendency is more or less ap- 
 parent in about 60 out of a total of 1 03 sets of curves taken at iroom 
 temperature, so it cannot be attributed to coincidence. Furthermore, 
 it is believed that this tendency would manifest itself in a greater num- 
 ber of cass were it not for the fact that a great many solutions are so 
 nearly transparent for one or the other half of the spectrum that no 
 accurate measurements are possible over that region. 
 
 On first thought it seems rather remarkable that the result of 
 any kind of mixing of salts should be to increase the absorption at 
 the ends of the spectrum at the expense of the middle. After a little 
 [reflection, however, a possible explanation suggsts itslf. All of the 
 colored salts studied in these series of experiments, with the exception 
 of the cobalt and chromium salts, have their main absorption bands (in 
 so far as the visible spectrum is concerned) at one or the other end of 
 the spectrum. Moreover, the cobalt salts have their maximum ab- 
 sorption to the left of this minimum region under discussion, and 
 chromium nitrate has its principle absorption in the violet and blue. 
 Besides, chromium chloride shows but little change in absorption due 
 to the addition of colorless salts, as compared with similar changes 
 in most of the other colored salts under investigation. Eliminating 
 chromium chloride, then, from the present discussion, we see that 
 most of the absorpiton comes near the ends of the spectrum. Now 
 if we may assume that the main absorption bands are due to the 
 molecules and not to the ions, and if the mixing of salts tends to drive 
 the ions back into the molecular state, we should expect to increase 
 the absorption in the region of the main absorption bands and de- 
 crease it elsewhere. And this is just what happens in the large ma- 
 jority of cases. The most striking examples are shown on plate 8. 
 
 EQUATIONS OF CURVES. 
 
 The equations of a number of curves were determined, and with 
 one exception, found to be simple exponential equations. For some 
 of these curves the points computed from the equation fall directly on 
 
29 
 
 top of the experimental curve. In all cases the greatest deviations 
 occur in those regions where the absorption is slight and the conse- 
 quent experimental error very large. This remairkably close agree- 
 ment between mathematical and experimental curves is quite surpris- 
 ing for two reasons : first, that the photometric readings had actually 
 been taken with sufficient accuracy to give such regular curves ; and, 
 second, that the absorption should obey a simple exponential law so 
 closely. The interpretation of this fact is not at present clear to the 
 writer, but he hopes to be able to make some report on the subject 
 later. 
 
 ANOMALOUS TEMPERATURE EFFECTS. 
 
 Neither cobalt nitrate nor sodium nitrate shows any appreciable 
 change in absorption with increase of temperature, yet the mixed so- 
 lution shows an exceedingly large temperature effect. The same 
 thing is true of nickel chloride and sodium chloride, of chromium 
 chloride and potassium chloride, and of cobalt sulphate and potassium 
 sulphate. These examples suggest that the colorless salt plays the 
 part of a catalytic agent in accelerating secondary chemical reactions, 
 or in accelerating or retarding the dissocciation of the colored salt. 
 It was also found that the addition of sulphuric acid to the mixture of 
 copper sulphate and potassium sulphate completely destroyed the 
 temperature effect. It would seem that this could be easily explained 
 on the assumption that the excess of the hydrogen ions from the dis- 
 sociated sulphuric acid prevented the OH radicals from the dissoci- 
 ated water from uniting with the other salts with the formation of col- 
 ored hydroxides. 
 
 CONCLUSIONS. 
 
 ( 1 ) These experiments show that the change in absorption pro- 
 duced by the mixing of sulphates with other sulphates seem to be small 
 in comparison with corresponding changes in nitrates, and still smaller 
 in comparison with the changes obtained in the mixtures of chlorides. 
 
 (2) The influence of temperature on the absorption of solutions 
 of mixed salts seem to be greatest in chlorides. From this fact, to- 
 gether with the preceding, it seems that Ostwald's idea that the col- 
 ored ion is alone effective in the absorption of light if not well founded, 
 
30 
 
 as Jones has already pointed out. For evidently the radical has con- 
 siderable influence. 
 
 (3 When the ratio of the absorption of the mixture to the sum of 
 the absorptions of the component parts is plotted as a function of the 
 wave-length, the resulting curve generally shows a minimum near the 
 middle of the spectrum. An attempt has been made to explain this 
 on the two following assumptions : first, that the principal absorption 
 bands are due to the molecules of the salt and not the ions ; and sec- 
 ond, that the effect of mixing different salts which contain a common 
 radical is to drive some of the ions back into the molecular state, thus 
 increasing the concentration of the molecules of the salt in the solu- 
 tion. This would naturally increase the intensity of the principal 
 bands and decrease the absorption elsewhere. 
 
 (4) Where the increase in absorption is excessively large and is 
 general throughout the spectrum, it is believed that part of this in- 
 crease is due to secondary chemical reactions. The colorless salt in 
 that case probably plays the part of a catalytic agent in accelerating 
 these secondary reactions. 
 
 (5) Equations have been determined for a good many of the 
 curves and found to be exponential in form. 
 
 (6) It is believed that the application of Jones's "Solvate Theory" 
 to the effects (recorded in this paper is undesirable and unnecessary. 
 The absorption of light by any colored electrolyte is undoubtedly a 
 function of its dissociation, and there seems to be little reason to doubt 
 that the degree of dissociation is changed upon the addition of another 
 salt containing one radical common to that of the salt in solution. 
 
 (7) No attempt has been made to differentiate experimentally 
 between scattering, reflection, and true absorption. In the changes 
 recorded on the previous pages and shown graphically by the curves 
 that follow, all three of these phenomena undoubtedly play their part. 
 
 In conclusion I wish to express my thanks to Profsesor Alpheus 
 W. Smith, at whose suggestion the above problem was undertaken, 
 and whose helpful suggestion and kindly council has been a constant 
 inspiration throughout the investigation. 
 Physics Research Laboratory, 
 
 Ohio State University. 
 
T A B L B I. 
 
 COO P P * R CHLORIDE. 
 
 I 
 
 Read- 
 ing. 
 
 a 
 
 fad- 
 ing. 
 
 a 
 
 Read- 
 ing. 
 
 A 
 
 Read 
 
 ing. 
 
 A 
 
 Read' 
 
 ing. 
 
 A 
 
 Red- 
 ing. 
 
 a 
 
 Read- 
 ing. 
 
 440 
 
 00 
 00 
 00 
 
 480 
 
 00 
 00 
 00 
 
 520 
 
 00 
 00 
 00 
 
 560 
 
 .17 
 .11 
 .04 
 .13 
 
 600 
 
 .344 
 .326 
 
 *3 
 
 640 
 
 .685 
 
 :& 
 
 .670 
 
 680 
 
 1.33 
 1.33 
 1.33 
 1.34 
 
 fll 
 
 *34l 
 
 tf 4 
 
 1.33 
 
 450 
 
 00 
 00 
 00 
 
 490 
 
 00 
 00 
 00 
 
 530 
 
 00 
 00 
 00 
 
 570 
 
 19C 
 
 .177 
 .212 
 
 .191 
 
 610 
 
 .407 
 .405 
 .427 
 .400 
 
 650 
 
 .810 
 .800 
 
 .795 
 .790 
 .825 
 .790 
 
 690 
 
 153 
 
 1.57 
 1.55 
 1.56 
 
 19J 
 
 .410 
 
 1.55 
 
 .802 
 
 460 
 
 00 
 00 
 00 
 
 500 
 
 00 
 00 
 00 
 
 540 
 
 .02 
 
 .00 
 .00 
 
 580 
 
 .29C 
 .24J 
 .260 
 .203 
 
 620 
 
 .492 
 
 .475 
 .475 
 .484 
 
 660 
 
 :$ 
 
 975 
 ,940 
 
 1 
 700 
 
 .77 
 1.77 
 1.82 
 1.72 
 
 .01 
 
 ,2(^ 
 
 
 48? 
 
 .9*4 
 
 1.77 
 
 470 
 
 00 
 00 
 00 
 
 5io 
 
 00 
 00 
 00 
 
 550 
 
 .04 
 .09 
 .02 
 .04 
 
 590 
 
 .303 
 
 : 
 
 630 
 
 .595 
 .545 
 
 *z* 
 
 .580 
 
 670 
 
 1.11 
 
 1.13 
 1.19 
 1.10 
 
 710 
 
 2.02 
 2.02 
 1.90 
 
 1-96 
 
 .05 
 
 .2J 
 
 
 .574 
 
 
 1*12. 
 
 !,2!L 
 
 CO 
 
 TABLE II. 
 
 ITH SODTTTV fTHT 
 
 ^RTT 
 
 Read 
 
 ead 
 
 ead 
 
 lac! - 
 
 ing. 
 
 Read- 
 ing. 
 
 480 
 
 500 
 
 560 
 
 580 
 
 00 
 00 
 00 
 
 Ti5 
 .07 
 09 
 .10 
 
 600 
 
 620 
 
 _ 
 .47 
 
 .49 
 
 66 
 
 680 
 
 1.92 
 
TABLE III 
 
 
 ;opp 
 
 R CHLQSIPJ 
 
 1 WITH AKMONIUV CHLORHIB AT 23.5C 
 
 A 
 
 Read- 
 ing. 
 
 * 
 
 Read 
 
 ing. 
 
 "X 
 
 Read 
 ing. 
 
 X 
 
 Rea< 
 ing 
 
 ~A 
 
 Read 
 ing. 
 
 "A 
 
 Read 
 ing. 
 
 ' * 
 
 Read- 
 ing. 
 
 450 
 
 .86 
 .86 
 .87 
 92 
 
 480 
 
 .13 
 .16 
 
 .15 
 
 - 1 ? 
 
 520 
 
 00 
 00 
 00 
 
 560 
 
 .01 
 .01 
 .01 
 
 600 
 
 .20 
 .21 
 .21 
 
 .20 
 .20 
 
 640 
 
 .53 
 
 .55 
 52 
 
 8 
 
 680 
 
 1.29 
 1.25 
 1.29 
 
 1.25 
 
 .01 
 
 .15 
 
 .$* 
 
 1.3* 
 
 .20 
 
 .89 
 
 460 
 
 '.58 
 .56 
 .57 
 
 500 
 
 00 
 00 
 00 
 
 540 
 
 00 
 00 
 00 
 
 58o 
 
 .05 
 .05 
 .06 
 .08 
 .06 
 
 620 
 
 .32 
 .33 
 .31 
 .32 
 
 660 
 
 .80 
 .82 
 .82 
 .80 
 
 690 
 
 1.49 
 1.54 
 1.52 
 
 1.51 
 
 .57 
 
 .32 
 
 .81 
 
 1.515 
 
 .06 
 
 T A B L B IV. 
 
 COPPER C2 
 
 rr ORITB wi 
 
 TR AMMONIUM CHT.ORU 
 
 B AT 70.0 C 
 
 X 
 
 Read- 
 ing. 
 
 A 
 
 Read 
 
 ing. 
 
 X 
 
 Read- 
 ing. 
 
 ^ 
 
 Read 
 ing. 
 
 X 
 
 Rea4 
 ing. 
 
 X 
 
 Read- 
 ing. 
 
 
 460 
 
 1.55 
 1.70 
 1.64 
 
 5:8 
 -fcS 
 
 500 
 
 P* 
 
 .24 
 .24 
 .24 
 
 .20 
 .20 
 
 540 
 1 
 
 .05 
 .05 
 .05 
 
 580 
 
 .18 
 .18 
 .19 
 a ? 
 
 620 
 
 & 
 
 59 
 .59 
 
 66( 
 
 1.42 
 
 i: ^ 
 
 1.46 
 
 
 Ob 
 
 .185 
 
 .57 
 
 i.*5r 
 
 .24 
 
 480 
 
 :Fe 
 
 .59 
 .60 
 
 520 
 
 .11 
 
 ? 
 .06 
 
 f05 
 
 560 
 
 .09 
 .11 
 .11 
 .09 
 
 600 
 
 ^2 
 .32 
 .33 
 .33 
 
 640 
 
 .89 
 .91 
 .91 
 .92 
 
 67C 
 
 1.79 
 1.76 
 1.79 
 1.7? 
 
 
 .585 
 
 .05 
 .06 
 
 .10 
 
 
 .325 
 
 .$1 
 
 
 1.78 
 
 .07 
 
ABSOB 
 
 AQUEOUS 
 
 Drnpur r <^ frxar 
 
 o & (pgrinherrtolj Curve 
 
 E (peri rttentoM Curve 
 
 o Experimental Curve CuCl t 
 
 o Mixtur^af CuCl 2 -r ZnC^ 
 
 . - bsorption of CuCI 2 
 
 Q-M ixture; of 
 
 -- Absorption of CuCI 2 
 
7OO 
 
7 
 
 + 
 
 ^A 
 
 \ 
 
 oAfixfofie 
 
 -& 
 
 /3 
 
 c--'V 
 
 -4 
 

700 
 
a-h 
 
 6/6 
 
 1 
 
 A3- 
 
 \ 
 
 \ 
 
 so/ution 
 
 222 
 
70O 
 

AUTOBIOGRAPHY. 
 
 I, Enoch Franklin George, was born near Petersburg, West Vir- 
 ginia, March 25, 1878. I received practically all of my secondary 
 education at West Virginia Wesleyan College, and my undergraduate 
 education at Valparaiso University, Chicago University, and West Vir- 
 ginia University. From Valparaiso University I obtained the Degree 
 of Bachelor of Science in 1 907, and from West Virginia University the 
 Degree of Bachelor of Arts in 1914 and the Degree of Master of Arts 
 in 1916. While in residence at West Virginia University I acted in 
 the capacity of Assistant to Dr. C. W. Waggoner during the year 
 1914-1915. The summer of 19141 spent as graduate student at the 
 University of Illinois. The year 1915-1916 I held a graduate as- 
 sistantship, and the years 1916-1917 and 1919-1920 a University 
 fellowship at the Ohio State University, from which school I received 
 the Degree of Doctor of Philosophy in 1 920. 
 
THIS BOOK IS DUE ON THE LAST DATE 
 STAMPED BELOW 
 
 AN INITIAL FINE OF 25 CENTS 
 
 WILL BE ASSESSED FOR FAILURE TO RETURN 
 THIS BOOK ON THE DATE DUE. THE PENALTY 
 WILL INCREASE TO SO CENTS ON THE FOURTH 
 DAY AND TO $1.OO ON THE SEVENTH DAY 
 OVERDUE. 
 
 KAY 10 
 
 LI) Jl-H)(),*/-7.'39(402s) 
 
YD 0228 
 
 484820 
 
 UNIVERSITY OF CALIFORNIA LIBRARY