Popular Science Monthly/Volume 72/February 1908/The Relation of Color and Chemical Constitution

From Wikisource
Jump to navigation Jump to search
1538237Popular Science Monthly Volume 72 February 1908 — The Relation of Color and Chemical Constitution1908William Jay Hale




AMONG the many branches of scientific learning whose early development we owe to the analytical mind of Sir Isaac Newton, none can show a more beautiful discovery than that different colored rays of light suffer unequal amounts of refraction or bending when passed through a prism; and that sunlight itself by similar means is resolved into a series of colors, the order of which, beginning at the red and ending with the violet, corresponds with a gradual increase in refraction. It was this that gave us our first spectrum and proved at once the composite nature of white light.

After these discoveries by Newton, a hundred years and more elapsed before Wollaston in 1802 observed the spectrum of a ray of sunlight admitted through a narrow slit in a window-blind. In this simple experiment, which gave a better distribution of the spectrum colors than could be obtained with the pencil-like rays of Newton's time, certain black lines were seen to cross the spectrum parallel to the slit. The investigation of these lines, however, was left to Fraunhofer, who, several years later, with much improved apparatus for collecting the light rays and projecting the same upon a screen, succeeded in definitely establishing the existence of a large number of black lines in the solar spectrum. In other words, the light from the sun was shown to be incomplete by reason of the absorption of certain of its rays, as was indicated by over 700 of these dark lines. To the principal lines, still called by Fraunhofer's name, he assigned letters beginning in the red with A and ending in the violet with H.

That there also existed an invisible portion of the spectrum lying to either side of the visible spectrum, was early pointed out. Sir William Hershel in 1800 observed the great heating effect of that portion beyond the red, while Sir John Herschel in 1840 demonstrated the existence of Fraunhofer lines in this infra-red region. Ritter and Wollaston showed that silver chloride blackened readily in the invisible portion of the spectrum beyond the violet, a fact readily inferred from Scheele's observation in 1777 that silver chloride turned dark more readily in violet than in red light. E. Becquerel, however, in 1842 succeeded in identifying many of the Fraunhofer lines in this ultraviolet region and lettered the principal lines from L to P in continuation of those already lettered by Fraunhofer.

Fraunhofer had noticed that the yellow spectrum line from common salt, when fed into a spirit lamp, was identical in position with the D-line of the solar spectrum. But though the formation of these discontinuous spectra from various salts in a flame was generally known, it was not until 1859 that the presence of the Fraunhofer lines in the solar spectrum was clearly explained by Kirchhoff, who deduced the following law: "The relation between the powers of emission and the powers of absorption for rays of the same wave-length is constant for all bodies at the same temperature." Thus the particles of a substance under the excitement of some outside force are thrown into a state of vibration which is dependent upon the chemical nature of the substance itself. This vibratory motion gives rise to waves in the ether and we have the phenomenon of emission. Again the particles of a substance are most responsive to these same characteristic vibrations and will absorb them whenever present, just as, by analogy, the strings of a piano pick up sound waves of the exact period in which they vibrate when these waves are set in motion by other musical instruments in the neighborhood. Kirchhoff explained the solar spectrum as one produced by a strong white light from an interior sphere passing through a concentric layer of vapors of many substances, each of which absorbs those particular rays that correspond to their own periods of vibration. The light, thus deprived of many definite rays, indicated their absence when its spectrum was cast upon a screen by the appearance of dark lines—the images of the slit through which the light passed—corresponding always to the wave-lengths absorbed. It must not be assumed that these lines of absorption are regions of total darkness. The particles set in vibration by the rays absorbed will naturally give out some light of this same vibration period, but the light emitted is so small in comparison with the rays from the original source which pass through unmolested that the image cast upon the screen will give the appearance of almost total darkness.

Now when a substance is yellow in color we can readily ascertain that the spectrum of the light it reflects is lacking in a number of rays of various wave-lengths. These rays correspond to the complement of the color reflected, and in the case of a yellow substance belong to that magnitude found in the blue end of the spectrum. If no wave-lengths of the visible spectrum had been absorbed, we should have had the continuous spectrum of white light in the light reflected, i.e., the body itself would not be colored. Colored substances, therefore, absorb the rays of their complementary colors and, consequently, when white light is transmitted through them their spectra will indicate the regions of this absorption by dark bands of varying intensity. The absorption spectrum coincides always with the spectrum obtained from the reflected light.

Though Fraunhofer had failed to grasp the true significance of the dark lines in the spectrum he was able to solve another highly important question—that of determining the wave-lengths to which these lines corresponded. From the wave-theory of light it may be readily understood that certain ether particles in the courses of different rays of light (e. g., those of equal amplitude) may receive a strengthening or retardation in their transverse vibrations according as they fall in with the same phase of vibration or out of it. Upon this principle of interference of light as developed by Young, Fraunhofer based his method for studying and measuring the lines of the spectrum. He made what he called a grating by ruling close together a number of parallel lines upon a glass plate. When light is thrown upon this series of equal and equidistant apertures a certain amount of the light will be diffracted to either side of the direct course. Among these diffracted rays as collected by a convex lens may be found several series of bright and dark bands which correspond to the points of augmentation and retardation, respectively, of the ether particles under the influence of light from certain apertures. By simple calculation the first bright band is known to be formed when the light rays from two adjacent apertures differ by exactly one wave-length in their respective courses to this band. A ready means, therefore, is given for measuring the wave-lengths of light rays. When white light is used a number of these bright bands will occur, with the light of shortest wave-length—the violet—nearest the central image and that of the longest wavelength—the red—farthest removed. In other words, we have a spectrum, but one so constructed that a direct means is given for determining the wave-lengths of the various lines it may present. The complete map of the wave-lengths of the lines in the visible solar spectrum was published in 1868 by Angstrom. The wave-lengths were expressed in ten millionths of a millimeter. Since that time they have served as a standard in all similar investigations under the name of the Angström Unit (A.U.). One millimicron (the millionth of a millimeter μμ,) is equal to 10 A.U. The visible spectrum extends from light of about 7,600 A.U. in the red to that of about 3,900 A.U. in the violet. A more satisfactory method of expressing the results of observations in the spectrum is to use the number of waves of any particular ray of light which occur in one centimeter in vacuo, or what is called the oscillation frequency (O.F.). This is the reciprocal of the wave-lengths when reduced to vacuum values. As the reduction makes but little difference in the final value, it is usually customary to make the correction by adding one or two A.U. to the observed wave-lengths. Curves constructed from oscillation frequencies approach more nearly a straight line, and thus are easier to draw.

A few of the best known lines may be given in order to show the relation in values:

Symbol. Color. Wave-length. (λ) 1/λ O. F. Frequency per Sec.
D Yellow 0.0005893 mm. 1,696  510 trillion.
F Blue 0.0004862 mm. 2,000 618"
H Violet 0.0003970 mm. 2,520 760"

Our best modern instruments for work in the infra-red region depend entirely on the heating effect. So sensitive indeed is the bolometer, as devised by Langley, that a difference in temperature of one five-hundred-thousandth of one degree F. can be determined and, by its use in this region of the spectrum, rays of wave-length 100,000 A.U. have been detected. The study of the ultra-violet region of the spectrum depends upon the sensitiveness of silver compounds and accordingly on photographical measurements. As glass was found to absorb the rays of shorter wave-lengths than 3,300 A.U., quartz lenses and prisms must be used. Quartz absorbs rays of a wave-length less than 1,850 A.U., but fluorite may be used for rays down to a wave-length of 1,000 A.U. Air itself exerts a powerful absorption for rays of a wave-length of 2,000 A.U. and under, hence for these finer observations the apparatus must be exhausted and observations made in a vacuum. In the photographical determinations of this region the greatest care must be taken in preparing the silver bromide plates. No gelatine can be used upon the plates as it is found to exert a strong absorption for the shortest rays. The sensitive plates are usually made by precipitating silver bromide in a solution over a glass plate and allowing the precipitate to settle slowly upon this plate. When these sensitive films are colored, the plate becomes more sensitive to the rays which the dye absorbs.

The principle established by Kirchhoff was applied with intense vigor to the study of all the lines of the solar spectrum. The introduction of various substances into a flame was found to give spectra of many colored lines, but these were always definite for each and every substance examined. The lines in these discontinuous spectra were seen in many cases to have their exact counterpart in certain of the Fraunhofer lines and consequently the existence of the particular element producing them may be assumed in the solar atmosphere. In this manner, the chemical composition of the sun's atmosphere has been determined, and even new elementary substances discovered therein by the selection of certain lines or groups of lines, unaccounted for by any element previously studied. The line-spectra of many elements are readily obtainable at low temperatures, but for iron and similar metals a far higher temperature is required, as, for example, that of the arc. For gases a strongly induced electric current (one of high tension) is necessary. In the arc spectrum of iron over 2,000 lines are observed, whereas the spectrum obtained at the flame temperature consists of only a few bands and lines. At the hottest portion of the spark the iron spectrum shows the same lines as in its arc spectrum, but in addition a number of "enhanced lines," as Lockyer has chosen to call them. By reason of these latter, Lockyer assumes that the atom of iron (as well as of other elements) may consist of more elementary constituents at extremely high temperatures, and, if the cooler vapors could be removed from this hottest zone, the enhanced lines might stand alone for the elementary form of iron—as a proto-iron. Such conditions are said to be obtained in sun spots and our hottest stars. Whether the extremely high temperature alone is sufficient to produce the enhanced lines, or whether their origin lies in the enormously rapid changes of electric stress, can not be answered at present. In either case, there seems to be no doubt but that the atom of an element consists of yet smaller particles, which, with rise of the disintegrating forces, show a marked increase in their activities, and, owing to the similarity existing between these particles, give spectral lines of greater and greater simplicity.

When a group of lines in a spectrum has oscillation frequencies that obey a single formula we call this group a series. The simplest elements usually give three series, each of which consists of lines in doublets or triplets. The action of a strong magnetic field upon the series of an element's spectrum tends to decompose the series; each line is resolved into two or three lines (doublets or triplets) according as the light is viewed along or across the magnetic lines of force. This is called the Zeeman effect. Of the three components of motion of the particles, that one which lies in the direction of the lines of force with vibrations backwards and forwards can emit no light except when viewed at right angles to these lines of force. The other two motions at right angles to the lines of force suffer a retardation and acceleration, respectively, with the result that their oscillation frequencies are similarly affected and consequently two separate lines will be developed. These may be observed by themselves when the light is viewed along the lines of force or in conjunction with the original line—with position between these two—when the light is viewed across the magnetic field. The electro-magnetic composition of the atom therefore seems to be corroborated by these results.

As an analogy to this gradual disintegration of the atom under the great stress brought to bear upon it, and. further to show how the more complex molecules behave under the influence of temperature, we have only to examine the spectrum of a compound. Whatever compound is admitted into a flame, the characteristic spectrum of the molecule first makes its appearance. This consists not of lines, but of bands of varying widths. On further increase of temperature the decomposition of the compound molecule is attained, and the bands gradually give way to the characteristic lines of the elements concerned. With numerous compounds, for example, the metallic chlorides, this temperature is exceedingly low. Since the presence of spectral lines is undoubtedly to be accounted for by the vibrations within the atoms, we may well have recourse to the modern conception of the atom as advanced by J. J. Thomson. Here the atom is considered as made up of a central mass carrying a positive charge. Surrounding it are numerous electrons of a negative charge, the number of which increases directly with the atomic weights of the elements concerned. The electrons are undoubtedly arranged in some systematic order and may, as Nagaoki imagines, follow parallel courses closely analogous to the rings of Saturn. A disturbance of any one group or belt of electrons will undoubtedly produce a disturbance in yet other groups and, according to the amount of disturbance, the definite vibratory motions established will set up vibratory motions in the ether, later to be detected in the spectrum. From this hypothesis the spectrum of an element of high atomic weight might be expected to contain more lines than one of low atomic weight. Such, however, need not necessarily follow. If we take the case of radium, uranium, etc., we may imagine the electrons in its atom to be grouped closely together in only a few courses or belts. In fact this very hypothesis may well account for the discharge of electrons from such highly condensed arrangements and give rise to radioactivity.

From this modern standpoint the molecule is regarded as a combination between atoms as effected by the loss or gain of one or more electrons from one to the other, developing what is commonly termed "bonds of affinity" and corresponding in number to the valence of the particular atoms concerned. These may be more correctly construed as Faraday tubes of force.[1]

With these ideas in mind the banded spectra of compounds may be accounted for by disturbances induced between the atoms, as well as by small electronic vibrations set up in the atoms themselves and due to the perturbances of the Faraday tubes of force. The vibrations resulting from this composite arrangement of vibratory centers may be sufficient to extend over a considerable area of wave-lengths and thus produce a band. As the temperature increases these band spectra, always obtained with compounds, pass over gradually into the line spectra of the constituent elements concerned. There follows, then, with increase of temperature or electric stress, as has already been noted, a gradual disintegration of the series of lines into simpler arrangements, caused probably by reason of a similarity existing between the ultimate constituents of our elements. This explanation is made more plausible from a study of the Zeeman magnetic effect upon similarly charged particles.

Even under the ordinarily obtainable conditions of the laboratory a great similarity may be noted between the series of lines in the spectrum of one element and the series of all other elements belonging to the same family. Thus with a gradual increase in atomic weights there occurs a corresponding gradual shifting of the series toward the red end of the spectrum. Increase in atomic complexity is ever seen to have a marked effect upon the vibratory motion of the simplest particles such that vibratory frequence is retarded. Among compounds, as well as with the elementary substances, this influence of mass is clearly shown in their spectra. Owing to the great tendency among most compounds to undergo ready decomposition when heated an examination of their spectra is restricted to the absorption spectra alone. The relations for absorption spectra having already been noted, it need hardly be further stated that the absorption bands in the spectra of compounds indicate at once the color of the compounds themselves and, what is most important of all, anything that can be brought to bear upon the interpretation of these bands and their positions should give us an insight into the cause of color as existent among compounds generally. In the examination of absorption spectra of compounds, the best results are obtained when the substances can be dissolved in some solvent which exerts but little or no absorption action for light. Among the best examples of such solvents are water, methyl alcohol (wood-spirit), and ethyl alcohol, none of which will absorb rays of a wave-length over 2,000 A.U. The absorption spectrum of a compound dissolved in a medium of this nature is identical with its absorption spectrum observed in the free state.

Among the first to obtain any positive results whatsoever in the examination of the absorption spectra of compounds was W. N. Hartley. He studied the solutions of metallic nitrates and found that the absorption in these cases was slightly modified with increase in atomic weight of the metal present, and concluded, therefore, that that portion now termed the nitrate ion—or negatively charged portion of a nitrate when dissociated by a solvent—has no effect upon the band. Not, however, until 1879, when Hartley and Huntington turned their attention to the study of absorption bands in the ultra-violet regions of the spectrum, could any hypothesis of a definite nature be formulated as regards the relation of these bands to chemical constitution. Their method of observation, which has been in use up to the present time, depended entirely upon obtaining a series of photographs of the spark spectrum as viewed through layers of a solution at varying concentrations. Hartley and Huntington used a cadmium alloy, which they found to give a great number of lines, but in recent work the arc spectrum of iron has been adopted. This latter gives, as already noted, a vast number of lines extending throughout the visible and invisible spectrum in a more or less equally distributed manner. The presence of an absorption band is detected by the absence of lines in the photograph, hence the advantage of their great number and equal distribution. By successive photographs accompanying an increase in dilution the greatest degree of absorption, and thus the head of any particular band, may be observed.

The oscillation frequencies at the edges of these bands were determined up to the point of complete transmission following the increase of dilution. The figures obtained were plotted as abscissæ against the corresponding volumes in which definite amounts of the substances were dissolved. The curved lines drawn through these points, called by Hartley the "curves of molecular vibrations," were found to be closely related to the chemical constitution of the compound studied. More recently a better method of plotting results consists in photographing through varying thicknesses of a solution of known strength and then diluting the solution ten times, repeating the observations, again diluting ten times, and so on till the point of complete transmission is reached. The relative thicknesses are now expressed in millimeters of those thicknesses that would be required of the last or most diluted solution, and these values plotted in the form of logarithms as the ordinates over against the oscillation frequencies as abscissæ. Curves thus plotted show at once the same relative shift with the ordinates as is made with the thicknesses examined. The persistence of a band, or change of concentration through which a band asserts itself, is well illustrated by this curve. In this point—the persistence—we have a characteristic function of the bands which connects them closely with chemical properties.

The compounds studied have been entirely within the realm of organic chemistry. In this class we meet with the most pronounced and, at the same time, the most easily varied tints. A study of these variations in color in the closely related organic compounds has, up to the present, occupied the entire attention of investigators, among whom, after Hartley and Huntington, are to be named Baly, Desch and Stewart.

The absorption spectra in the ultra-violet region may be classified under two broad types; the first type is that in which only a general absorption is present; the second is that in which distinct absorption bands occur, a type usually defined as one of selective absorption. To the first class belongs, broadly speaking, the majority of the aliphatic or open chain compounds; to the second belongs the majority of the aromatic compounds or those of ring structure. Among the first observations made it was discovered that an acid and its alkyl ester (esters formed by such, groups as methyl, CH3, ethyl, C2H5, etc.) gave identical absorption bands; a fact that pointed conclusively to the identity in molecular constitution existing between the two compounds. But among the most interesting cases bearing upon the relation of these absorption bands to chemical constitution stand the two substances acetyl acetone and ethyl aceto-acetate. We assume that each of these compounds can exist in either of two forms—one in which an oxygen atom is doubly linked to a carbon atom which bears in turn two carbon radicals and thus forms a so-called ketone; the other, where this same oxygen atom is singly linked to the carbon atom in question and has its second affinity absorbed in a hydrogen atom, thus forming a so-called hydroxyl derivative, or one usually designated by the term enolic. The two forms may be graphically represented thus:

Such compounds are described as tautomeric, i. e., they contain a labile atom, hydrogen, which in its wandering or change of position brings into existence two distinct modifications of a compound without altering its general structure. As this change is not complete at any one instant and may vary with change of conditions, we have a condition of equilibrium always existing between the two forms. In the compounds just cited the labile hydrogen atom may be replaced by the atom of a metal and thus give what are called metallic derivatives, which from chemical evidence are supposed to exist entirely in the enolic form. Upon examination of the absorption spectra of these compounds, acetyl acetone itself, as well as its aluminium derivative, was found to give similarly banded absorption, but with that of the aluminium salt showing a greater persistence. Now ethyl aceto-acetate gives only a slight general absorption without trace of a band. Its aluminium derivative, however, gives a banded spectrum which bears a great similarity to the spectrum of acetyl acetone. Therefore, if the metallic salts are enolic, as chemical evidence strongly favors, the free ethyl aceto-acetate certainly must be ketonic.

In order to investigate this matter more closely the two ethyl derivatives of ethyl aceto-acetate were examined. These compounds made by entirely different processes have different properties and correspond in constitution to the two distinct forms, ketonic and enolic, of the free ethyl aceto-acetate. They may be graphically represented as follows:

Upon examination of their absorption spectra, it was observed that the enolic compound exerts only a general absorption without a band, while the ketonic compound was practically free from absorption. This is exactly what might be anticipated from the results of Hartley, who had already shown that no difference exists between the absorption spectrum of the compound and that of its alkyl (here ethyl) derivative. Even a mixture of these two alkyl derivatives fails to show the presence of absorption bands in the spectrum. We may conclude, therefore, that an absorption band is not to be attributed to either the one or the other form of a tautomeric substance, but rather to the changing of one form into the other—a dynamical isomerism. If this intramolecular transformation is the source of the disturbance which produces the absorption bands, then an acceleration of this transformation should show itself in the increased persistence of the band, while retardation of the same should diminish this persistence. For some time it has been known that alkalies exert a marked positive influence upon the velocity of tautomeric changes and, as may be naturally inferred, acids retard this change. On the addition of a small amount of sodium hydroxide to a solution of ethyl aceto-acetate, the form of the absorption-curve changes at once and a band appears. On the further addition of alkali, the depth of this curve, that is the persistence, increases until it reaches a maximum corresponding to the presence of a large excess of alkali. The absorption-curve of the aluminium derivative of this ester has not the persistence of that of the sodium salt when the sodium hydroxide is present in excess of one molecular equivalent. With the addition of hydrochloric acid a retarded action is developed and even the absorption curve for the free ester is seen to diminish slightly in its persistence when an excess of acid is present, indicating, therefore, that the free ester is not entirely ketonic, but is in equilibrium with a very small quantity of its enolic modification. Spectroscopic evidence points out that the persistence of the absorption bands over concentration changes is directly proportional to the number of molecules in the state of oscillation or, in other words, is a measure of the dynamical isomerism between substances in equilibrium with each other.

From these considerations it is evident that this dynamical isomerism must be closely connected with some peculiar vibration or free period synchronous with the oscillation frequency of the light rays absorbed. The oscillation frequency, however, is about the same for all the substances just examined and lies between the limits 3,600 and 3,800, no matter what the labile atom may be. We are, therefore, forced to the conclusion that the absorption bands can not be due directly to this oscillating labile atom, or, in other words, the vibration frequency of this atom is not identical with the oscillation frequency of the light absorbed. This inference is strongly substantiated by physical evidence which represents atomic motion as far less than that of this magnitude of light rays. There remains then but one final solution of this question, or the conclusion that the absorption band is due directly to the change of the linking which accompanies the wandering of the labile atom. In the case of the keto-enol tautomerism just discussed, we may represent the change graphically as follows:

At some stage in its transformation the hydrogen atom must have wandered to the half-way point of its journey and have been linked definitely to neither the carbon nor the oxygen atoms. We should then have the phase in which the carbon and oxygen atoms change linking.

By the examination of countless numbers of organic compounds it is found that absorption bands in the ultra-violet region of the spectrum are shown only by compounds exhibiting some form of tautomerism, whether this be due to the keto-enolic type or to a periodic type, like that present in ring compounds. The oscillation frequency of the light absorbed in all cases of keto-enol tautomerism is about the same, but with an increase in the mass of the molecule as brought about by the displacement of one atom by a second atom or group of atoms of greater relative weight, a decrease in the oscillation frequency is observed. This displacement, however, is only small and does not interfere with the general assumption that there is present some condition common to the whole group upon which the absorption directly depends.

From the standpoint of modern theories regarding the structure of the molecule, there must arise in this tautomeric change a constant making and breaking of the Faraday tubes of force. This means a constant disturbance of the electron systems and, consequently, similar vibrational disturbances in the ether. From Hewitt's studies in fluorescence these electronic disturbances, due to dynamic isomerism, appear to be of the same order as light waves and, consequently, by absorption of their wave-lengths white light should show the absorption spectra we have noted. The sodium and aluminium derivatives of ethyl aceto-acetate may be described as equilibrium mixtures of the enolic and ketonic forms. The fact that the sodium salt is so easily hydrolyzed in an aqueous solution need not enter into the discussion of the absorption spectra. The evidence in all these cases goes to show that the metallic ion still exerts its influence and does not lead an altogether separate existence from that of the negative ion. Accordingly we may regard the Faraday tubes of force as stretched, but not necessarily broken, by the action of the solvent. On this basis, an ionizing solvent is to be considered as one that will bring about this lengthening of the Faraday tubes. Among the best examples, we may cite water, liquid sulphur dioxide, and liquid ammonia, or those substances which possess in reality a certain amount of "residual affinity"—affinities that may yet be exerted. Tautomeric changes in solution receive their interpretation, then, in the lengthening of the Faraday tubes of force to that point where the labile atom comes so far under the influence of a neighboring atom that a break occurs, which in turn gives rise to the oscillatory disturbances already discussed. With tautomeric compounds in which the labile atom has been replaced by an alkyl group there is absence of tautomerism due to the non-formation of alkyl ions, in which case it is seen that water and alcohol have not sufficient power to lengthen the Faraday tubes of force. The persistence of an absorption band may be defined now as the measure of the atoms in this transitorial state or the measure of the extent to which the labile atoms are separated from the molecule proper. Wherever the tautomeric compounds display the phenomenon of fluorescence a second free period of vibration is present. The latter must depend upon the former since a compound does not fluoresce except when exposed to light rays of the frequency of the first free period. Recently it has been demonstrated that a fluorescent substance gives two bands in its absorption spectrum, one for each of these periods of vibration. The band from the incident light is well marked, but the band from the fluorescent light is so faint that it can be made fairly visible only when the light of the first free period is strengthened; a fact that substantiates the dependence of the second free period upon the first.

As the origin of absorption bands in the ultra-violet spectrum have received an explanation in the change of linking brought about by the shifting of a labile atom, so clearly represented in the examples of keto-enol tautomerism, we may rightly expect to find absorption bands in the spectra of other compounds in which some change of linking is exhibited. No more beautiful example can be found than that of the compound known as benzol, where six carbon atoms, unchangeable in their order, are bound together in a single ring. To each carbon is attached an atom of hydrogen, but as carbon is usually considered quadrivalent, a fourth seemingly unused bond of affinity remains free to each of these carbon atoms. This affinity may be considered as the residual affinity. To this substance Kekulé has assigned the structure illustrated by the graphical formula:

The introduction of three double linkings between the alternate pairs of carbon atoms satisfies the demands for quadrivalence in these atoms. But Kekulé clearly called attention to the fact that a sort of equilibrium existed between all the carbon atoms, such that the presence of one of the three double linkings between any two adjacent carbon atoms, when both were involved in the formation of a derivative, would not necessarily change the properties of the derivative from that one formed when two adjacent carbon atoms were united by only a single bond. Now the Kekulé formula, and in fact all the older formula? assigned to this compound, represent only particular phases in the motions of the molecule. The space-formula proposed by Collie,[2] in which the atoms are represented as in a state of continual vibration, serves well for the basis of our modern conception. Upon examination of the absorption spectrum of benzol we note the presence of seven distinct bands all quite similar and closely situated with reference to each other, appearing between the oscillation frequencies 3,725 and 4,200. These bands are in that part of the ultra-violet spectrum where the absorption bands due to keto-enol tautomerism displayed themselves. At once the idea of a similar make-and-break of linkings between the carbon atoms suggested itself, and, in exact accordance with this hypothesis, the seven distinct bands may find here their cause of formation.

In keto-enol (aliphatic) tautomerism an even number of carbon atoms is always involved in the make-and-break of linkings. Accordingly with the benzol molecule we may assume that two, four or six carbon atoms may enter into this phase at one time. If the carbon atoms are numbered consecutively from 1 to 6, we should have in order the following conditions which represent the change of linkings between certain numbered carbon atoms: (1 and 2), (1 and 3), (1 and 4), (1 and 2, with 3 and 4), (1 and 2, with 3 and 5), (1 and 2, with 4 and 6), (1 and 2, with 3 and 4, with 5 and 6). At the outset we shall suppose the benzol ring to be elastic and capable of undergoing various pulsations, such as may be illustrated by the accompanying figures:

The clotted lines show the linkings developed from the free affinities when the ring is pulsating between the two forms (a) and (b). The centric formula for benzol (Baeyer's), as shown in (c), may be, therefore, an intermediate form for all the possible forms. The free or residual affinity possessed by each carbon atom asserts itself under the various conditions which can be brought into existence by these pulsations, with the effect that the several linkings produced must involve always a pair of carbon atoms and then in turn during the second stage of the pulsation must suffer a break and consequently give rise to some particular one of the seven possible phases, with its characteristic absorption band of course depending upon the carbon atoms in question. Altogether, when the entire ring is free to pulsate in every direction, there will arise seven absorption bands which represent the seven possible combinations of linking-change.

The derivatives of benzol may be expected to show some variation in type and manner of pulsation from that of the parent ring, but whatever changes occur the effect upon the characteristic absorption spectrum of the original molecule will always indicate the exact nature of each change. In this connection it will be well to consider a few of the more important derivatives, which, as is generally known, are primarily formed by the replacement of one or more of the hydrogen atoms by an equivalent atom or group of atoms—a process called substitution. The alkyl radicals (methyl, ethyl, etc.) stand as a type of the neutral groups and consequently, when they are present, little change in the spectrum of the original substance should be observed. The spectrum of toluol, C6H5 • CH3, ethyl benzol, C6H5C2H5, etc., are almost identical, but only the first two absorption bands of the original benzol spectrum are well marked, the remaining bands having fused more or less into one broad band. With aniline, C6H5NH2, where the basic unsaturated amido-group (NH2) has replaced the hydrogen atom, we get only a broad absorption band caused, no doubt, by the residual affinity of the nitrogen atom which binds or holds all the free affinities of the benzol ring. Upon the addition of an excess of hydrochloric acid to aniline, we obtain the saturated compound known as aniline hydrochlorate, C6H5 • NH3CI, the nitrogen having passed from the trivalent to the quinquivalent state. This compound, as might be anticipated, gives an absorption spectrum resembling very closely that of the mono-alkyl derivatives of benzol just mentioned.

The absorption band of phenol, C6H5OH, differs from that of the mono-alkyl derivatives in that one pronounced band has replaced the two prominent bands in the spectra of the latter. In the case of anisol, C6H5 • OCH3, the methyl derivative of phenol, known as an ether, the two prominent bands are again in appearance. In other words, the substituting group methoxyl (OCH3) partakes more of the nature of a saturated alkyl group, whereas the hydroxyl group (OH) acts very differently. By a close examination of the two bands from anisol and the one from phenol we see that the transmitted portion, or that portion which serves to divide the one band into two, lies between the oscillation frequencies, 3,640-3,655. This is exactly the region where the absorption bands due to keto-enol tautomerism make their head. In other words, the presence of just such dynamical isomerism as may be caused by the wandering of the labile hydrogen atom of phenol will account for this absorption band and its position in overlying the regular bands due to phenolic structure, as shown in the case of anisol, etc. That a condition of dynamical isomerism is really present in a free phenol is further proved by the shifting of the absorption band to the left upon the addition of sodium hydroxide to its solutions; a result always observed in keto-enol tautomerism. Upon the bands formed by anisol the addition of alkali has no effect. On the other hand, the addition of hydrochloric acid to a phenol retards this tautomerism and when large excess of the acid is used the transmitted portion of the spectrum or that which is due to the free benzol nucleus begins again to make its appearance. The spectrum observed in the case of nitrobenzol, C6H5 • NO2, and other derivatives, where the substituent possesses marked residual affinity (due here to the oxygen atoms) shows only a general absorption. This condition, therefore, is brought about when the active residual affinity of the new groups restrains or locks up the free affinities of the benzol ring and thus retards its internal motions.

As with the mono-derivatives of benzol, so also with the disubstituted derivatives, the general rule holds true; wherever the substituents are groups well saturated, they will exert scarcely any retarding action upon the pulsations of the original molecule. The disubstituted derivatives are classified as ortho, meta and para, according as the groups are adjacent, once removed, or twice removed (diametrically across the ring) respectively, from each other. The para compounds give a spectrum more closely resembling that of the parent substance, benzol, and hence may be said to be the more symmetrical arrangement, or that which accords best with the even or symmetrical pulsations of the benzol molecule. With the ortho-and meta-compounds we may say that the unsymmetrical loading of the ring operates against the even pulsations and tends to produce irregular vibrations which give rise to less distinctive absorption bands.

In aliphatic, as well as in aromatic, compounds, we often observe that a certain amount of residual affinity lurking in the oxygen atoms can exert a strong influence upon an entire group of atoms. One of the simplest and most reactive combinations in which oxygen may be found is that known as the carbonyl group (CO), which, when occurring between two carbon radicals, constitutes a ketone as we have already noted. The simplest ketone is acetone, CH3—CO—CH3. The additive capacity of this carbonyl group for various reagents is well known, but this capacity very often decreases in power with an increase of the molecular aggregation in the near vicinity. For example, the additive capacity of the carbonyl group in the compound methyl-ethyl ketone, CH3—CO—C2H5, is usually less than that in acetone. These and similar facts have been explained upon the hypothesis of "steric hindrance" for lack of a better phrase. Though at times this hypothesis may best explain some of the intricate problems, still it hardly dare be supposed that the paths of intra-molecular vibration of the atoms is other than large in comparison with the size of the atoms themselves; consequently, slight increase in the mass of the substituents should have no appreciable effect upon the activity of a neighboring group. Oftentimes it was found that very large substituents increased the additive capacity of a carbonyl group. Thus when one of the hydrogen atoms of acetone is replaced by a carbethoxyl group (COOC2H5), a group formed by the replacement with ethyl of the hydrogen atom in the regular organic acid group, carboxyl (COOH), we get a great increase in the activity of the original carbon group. The compound so formed would have the formula, CH3—CO—CH2—COOC2H5, i. e., ethyl aceto-acetate—the very same compound as. was studied with reference to keto-enol tautomerism. An explanation of the increased activity in this case from the standpoint of dynamical isomerism which may be present seems to be most adequate. The oxygen atom exists temporarily in the enolic (OH) stage and the hydrogen atom, at the moment of departing, must leave the oxygen atom and consequently the carbonyl group nascent, i. e., in an exceedingly active form, similar here, no doubt, to the state acquired by ionization in solution. Again the hydrogen atom itself at the moment of separation would be most susceptible also to chemical action.

In order to get an idea of the relation of this carbonyl group to the carboxyl group, one of the simplest compounds which exhibits this arrangement was studied. The example taken was the ethyl ester of pyruvic acid, CH3—CO—COOC2H5. Here there was observed an absorption band lying much nearer the red end of the spectrum than that obtained in the case of ethyl aceto-acetate. The band had a head at about the oscillation frequency 3,100, whereas the band of the latter had a head near the oscillation frequency 3,700. As the molecule is lighter than that of ethyl aceto-acetate, we should, from previous observations, expect the band to be shifted farther from the red; the opposite, however, is true and the only explanation that seems possible is that the band is the result of a new kind of vibratory motion arising between two carbonyl groups when in close proximity to each other. In order to substantiate these conclusions other derivatives containing two carbonyl groups were studied. But as the carbonyl group in carboxyl has not all the characteristics of a true carbonyl group, attention was turned to the compound diacetyl, CH3 — CO — CO — CH3. Here the absorption band occurs at the oscillation frequency 2,400 (wave-length 4,170 A.U.), which is in the visible blue region of the spectrum, and hence this absorption of colored rays must result in the compound itself taking on the complementary color—that of yellow. In the same way it can be shown that glyoxal, OHC — CHO, gives an absorption band in the visible blue region, and consequently its distinct yellow color may be explained. Oxalic acid, HOOC • COOH, however, with a hydroxyl group next to each carbonyl group and therefore analogous to pyruvic ester, gives no band in the visible spectrum, and is, therefore, colorless.

Upon the theory that a change in linking produces the absorption bands, the only possible explanation would be indicated as follows:

The make-and-break contact between the oxygen atoms would give marked activity to these atoms. Such a process other than tautomerism, where a wandering of a labile atom is suggested, has been named by Stewart and Baly[3] "isorropesis" (equipose), and differs from the former in that the head of its absorption band lies much nearer the red end of the spectrum or almost in the visible violet region. With the diketone known as benzil, C6H5 — CO — CO — C6H5, an absorption band with head at the oscillation frequency 3,900 was noticed in solutions of small concentrations. This would seem to indicate the presence of a certain amount of oscillation due to the benzol nucleus. The residual affinities of the two carbonyl groups are undoubtedly fixed, to some extent, by the free affinities of the benzol molecule, but even so there may be present a small amount of isorropesis between the two carbonyl groups. That such is really the case is demonstrated in solutions of greater concentration by the presence of a very shallow absorption band with head at the oscillation frequency 2,650. Its shallowness, however, argues for only a slight isorropesis; indeed the color of benzil, which is but faintly yellow, may be made to disappear entirely upon dilution of its solutions, a fact confirming the spectroscopic evidence of oscillations due to benzol structure in solutions of small concentration. The presence of isorropesis, here brought about by two carbonyl groups in juxtaposition and indicated, as we have seen, by vibratory frequencies in the ether of longer wave-lengths than those due to keto-enol tautomerism, stands out at once as the source of color in chemical compounds. Again we note that the more pronounced this isorropesis the more active chemically are the groups undergoing the disturbance. The additive capacity of benzil is markedly less than that of diacetyl.

Among the compounds which furnish us with examples of this nature, or substances in which two carbonyl groups can come under the influence of each other, we may mention the most important of all, that of para-benzoquinone. This quinone is a derivative of benzol in which two oxygen atoms are located in the para-position to each other. In order to satisfy the bivalence of the oxygen atoms, the para carbon atoms are regarded as having their free affinities absorbed in these oxygen atoms, leaving the remaining carbon atoms to arrange their free affinities in two pairs of double linkings (according to the Kekulé hypothesis). It may be, however, that the second affinity of each oxygen atom will assert itself in a linking between these two atoms and therefore leave the characteristic benzol nucleus undisturbed. The two forms may be graphically shown as follows:

Now it has been proved chemically that para-benzoquinone can exist in each of these two forms. We have then just such an example of making-and-breaking as has been indicated in isorropesis, but in addition a change in the manner of linkings in the molecule accompanies this process. The absorption spectrum of para-quinone gives a band with its head at the oscillation frequency 2,150, one almost identical with that obtained from other ring compounds, e. g., camphorquinone, where two carbonyl groups are adjacent. From a study of the pulsations of the benzol ring the para-positions have been shown to be closely related, and in fact so well brought under the influence of each other as to be considered as practically adjacent. Since the position of the absorption band is in the blue region color must, of course, be present in this substance. Simple quinones usually show a yellow to an orange color. This color is undoubtedly due to isorropesis and whenever we have this class of substances—known as quinones— as the base of various derivatives, we have a right to look for this same influence between the unsaturated groups, or that condition which gives rise to isorropesis and hence to color. A great portion of the coloring products known have just this sort or structure and the origin of their colors, therefore, may receive the interpretation indicated.

For many years the color theory proposed by Witt has been the basis of all chemical investigations in this domain. Here it is supposed that the color of an organic compound depends upon the presence of an atomic group known as a chromophore, such, for example, as the nitro group (NO2), etc., and the introduction of more and more of these groups into a compound produces a gradual increase in depth of color. The various radicals with their respective color-giving groups are known as chromogens; upon union of these with other radicals of an acidic or basic character, we arrive at the conditions for coloring products or dye-stuffs. Now the carbonyl group alone does not appear as a pronounced chromosphere, but when two carbonyl groups, as in the ortho-or para-position in the benzol ring, are present one of the best of chromophores is developed. From such results as these Armstrong was led to believe that the particular linkings present in the benzol ring when two carbonyl groups were para to each other might account for the pronounced color reaction shown by these compounds. He characterized this type of structure, wherein the para carbon atoms have double linkings with the oxygen atoms as "quinonoid" (quinoid), in contradistinction to that of the alternate double linkings in a benzol ring or "benzenoid" (benzoid) . Eventually, he came to the conclusion that color in an organic compound depends upon the presence of this quinoid arrangement. From the chemical standpoint Armstrong had advanced upon solid ground. The real insight, however, into the relative value of one arrangement over that of another, as to their respective powers of light absorption and consequently of color production, must rest upon spectroscopic evidence. For example, we have seen that double linkings in themselves do not possess any power for light absorption. Their mere presence, therefore, can not account for color in a chemical compound, but if by their presence some form of oscillation is produced, we may expect the establishment of definite vibrations in the ether, which will be possible of detection in the spectrum. In compounds of the quinoid type the conditions are precisely those that will produce vibrations in the ether corresponding in wave-length to portions of the visible spectrum, consequently the appearance of color. In compounds of the benzoid type alone the oscillations correspond to vibrations of such frequencies that they fall in the ultra-violet region of the spectrum, and hence such compounds will be free from color. The oscillations which exist whenever the quinoid type of compounds is concerned, and which distinguish this type from that of the benzoid, must be due to the oxygen atoms in the para-position. Now the activity of these oxygen atoms is to be attributed to the residual affinity which each is known to possess, and hence by the assertion of this affinity when in close proximity to each other, followed quickly by a break in the same, we arrive at the condi- tion known as isorropesis, upon which form of oscillation the color depends. It is necessary in this process that the active groups under- going isorropesis should be adjacent. The pulsations of the benzol ring readily furnish the means by which the two para-atoms are suc- cessively brought under the influence of each other, and hence their positions will approach more nearly to that of adjacent atoms, a point that was confirmed by the similarity in the absorption spectra between para-benzo-quinone and compounds where the two carbonyl groups were actually adjacent. The study of ortho-quinones falls in the same category as the para-quinones and may be explained in a similar manner. Meta-quinones, however, can exist, but momentarily on the hypothesis of the benzol pulsations and hence are unstable.

Isorropesis, as has just been indicated, occurs between adjacent atoms possessing residual affinity. It is also to be remembered that some disturbing force must be brought to bear upon these atoms, for otherwise no make-and-break and consequently no oscillation can take place. In the simplest case studied, that of diacetyl (CH3 — CO — CO — CH3), the disturbing influence rests undoubtedly with the hydrogen atoms of the methyl groups which from their electro-positive nature exert a strong attraction for the electro-negative atoms of oxygen. This constitutes a sort of keto-enol tautomerism, the presence of which should certainly be accounted for in the appearance of the absorption-curve; indeed, the slight extension of the absorption-curve of this compound near the oscillation frequency 3,800 corresponds exactly to the location of a band due to keto-enol tautomerism. The cause of isorropesis in a compound rests, then, upon the disturbances of the residual affinities of the two atoms in juxtaposition. In the examples already cited, those of pyruvic acid and oxalic acid, the hydroxyl group is next to the active carbonyl group. The slight posi- tive nature of the hydrogen atom in this capacity will diminish its disturbing effect upon the second oxygen atom, or that of the carbonyl group, and consequently only the slightest amount of isorropesis will be possible. With the alkyl ester of pyruvic acid the conditions will favor a slight isorropesis, as we have seen, but with oxalic acid there should be none at all. In quinones the residual affinities of the benzol ring constitute the disturbing factors. The hydrogen atoms of the benzol molecule may also exert some disturbance. In general, we may say that the amount of isorropesis must rest upon the disturbing in- fluences which can be brought to bear upon the active groups showing residual affinity, or those susceptible of this new kind of oscillation.

Isorropesis need not always be confined to residual affinities between two oxygen atoms. Other unsaturated atoms may show similar reactions. Thus the nitro-anilines, H2N—C6H4—NO2, give an absorption-curve similar to that of para-benzoquinone. Here the residual affinities of the nitrogen atoms are disturbed by the motions of the benzol molecule. And, in addition, the unsaturated oxygen atoms of the nitro-group are disturbed by the hydrogen atoms of the amido-group (NH2). These facts, together with the position of the absorption band, point unmistakably to isorropesis, and the two distinct forms thus in equilibrium may be represented by the following figures:

A solution of nitro-aniline in hydrochloric acid gives a colorless solution showing no trace of absorption band to indicate isorropesis. The structure of the hydrochlorate, therefore, is purely benzoïd in type and enters not into the quinoïd form by reason of the saturation of both nitrogen atoms. In the case of the nitrophenols similar reasoning may be followed. The absorption band of para-nitrophenol, O2N—C6H4—OH, in neutral alcoholic solution, is identical with that of para-nitroanisol, O2N—C6H4—OCH3, the methyl ether of this phenol. Consequently their structures may be assumed to be identical. When the phenol, however, is converted into the sodium salt its absorption-curve alters and a band similar to the band of nitro-aniline appears in the visible blue region of the spectrum; in other words isorropesis has been brought about with the natural consequence—the appearance of color in the compound. The hydrogen atom of the free phenolic group (OH) is seen to be but slightly affected by the residual affinities of the oxygen atoms of the nitro-group; the more electro-positive sodium atom, however, shows a greater activity and may be drawn over to one of the oxygen atoms of the nitro-group, and thus a quinoïd type of linking established. In the equilibrium between

these two forms we may unquestionably look for the conditions which underlie the formation of color in the salts of nitro-phenols. The colorless free nitrophenols present only the regular form of vibration known to the benzoïd structure and hence can give no oscillation of a frequency low enough to produce color. In the salts of these phenols, however, the quinoïd type is developed and, though always in equilibrium with a certain amount of the benzoïd type, isorropesis will be present to an extent dependent upon the degree of unsaturation of the atoms, and indicated by the appearance of color. In the case of meta compounds a measure of the persistence of their absorption bands indicates a smaller amount of isorropesis and consequently they will be less colored than the ortho- and para-derivatives. In all of these investigations care must be used in the selection of a proper solvent. Since water is known to possess a large amount of residual affinity its action upon the ethers of nitro-phenols will be quite apparent. Alcohol serves the purpose here because it is well known to exert little or no ionizing action upon ethers and esters. In general, the new free period of oscillation—isorropesis—may be represented by the equilibrium:

These are conditions which accord entirely with certain known chemical facts.

In compounds of the benzol structure the cause of color begins with the particular vibrations of the molecule itself. These oscillations, however, as has been seen, are synchronous with light waves of a very high frequency and give rise to absorption bands in the ultra-violet region only. When some other influences can be brought to bear upon these movements, as, for example, the introduction of a potential ketoenol tautomerism, isorropesis is established and the oscillations, which are now of a less frequency, may be low enough to show the beginning of color. When the retardation of these oscillation frequencies is continued, as, for example, by the introduction of heavier atomic complexes for the simpler and lighter hydrogen atoms, the absorption curves due to oscillation will gradually be made to travel toward the red end of the spectrum and the color, naturally, will travel into the blue. A very well-known example of this is the increase in depth of the blue color possessed by certain dyes which accompanies an increase in the number of methyl groups introduced into the molecule. The introduction of a chromophore group, one of a more or less unsaturated nature, may in this light be considered as among the best to push back the oscillation frequency. But with reference to the powerful effect these chromophore groups have upon the retardation of the pulsations of the molecule and the consequent establishment of a new type of linking always in equilibrium with that of the original nucleus, the interpretation of their influence seems best explained in the production of an entirely new, free period of vibration—isorropesis—within the molecule itself. Now the term isorropesis is used to define the oscilla- tion that takes place between the residual affinities of atoms in juxta- position. The idea, however, may have already presented itself that in the case of the benzol structure the presence of keto-enol tautomerism with its particular period of vibration together with the oscillations occurring in the benzol nucleus, might, by a mutual combination of these two periods, give a period of greater wave-length and thus coincide with light rays in the visible region of the spectrum. If this were true, then these two conditions just stated might be looked upon as potential color systems. The actual presence of the conditions for isorropesis in the aliphatic series argues most strongly for the same sort of oscillation in the aromatic series wherever circumstances are favorable for its existence. In fact it seems highly probable that its presence alone will account for all the color-formations in the aromatic series. Other vibratory centers may exist and in fact do exist in the various compounds, but their presence only influences the amount of isorropesis that can take place and does not altogether destroy this particular form of oscillation.

In the quinoïd type of compounds the actual existence of the two distinct modifications which underlie isorropesis has already been shown. The change of one of these forms into the other and vice versa necessitates a change in manner of linking throughout the molecule which accompanies the oscillation in question. The fact that no one arrangement of atoms, no matter what their method of linking, can be made to show an absorption band is sufficient in itself to argue for the make-and-break in the two forms of the quinone as the cause of the color that exists among members of this class. For many years the quinoid linking has been supposed to be the source of color in compounds of quinone formation. It was not until recently, how- ever, that Gomberg has been able to prove conclusively that the quinoid type of linking actually exists in colored compounds of this nature. Not alone the presence of the quinoïd type, but also the benzoïd type has been shown to be present. In fact he has been able to interpret the conditions which determine the equilibrium always existent between these two forms and thus has succeeded in establishing by purely chemical means the amount of quinoid formation and con- sequently of isorropesis possible among aromatic derivatives. The spectroscopic evidence, therefore, on the existence of just such a type of oscillation as may be present in equilibria of this nature is corrob- orated. Upon the amount of isorropesis shown — a factor always dependent upon the relative unsaturated condition of the atoms coming into juxtaposition — we arrive at the depth of color in any given equi- librium. The presence of other groups may augment or retard the influence of these unsaturated atoms undergoing isorropesis, and con- sequently the corresponding variations in the oscillation frequency will be indicated by similar variation in the nature of the color. In many ring compounds where no such arrangement of linkings, as in the benzoïd and quinoïd classes, is seen, almost no evidence of color can be found. Furfuran, pyrol, camphor and many others of a constitution exhibiting double linkings show only general absorption in their spectra. But whenever the benzoïd type is present, no matter whether the ring be composed entirely of carbon atoms or not, the conditions for isorropesis are at once favored so soon as unsaturated atoms or groups can be introduced. These groups by their unsaturated condition give rise to new linkings and then in turn undergo the make-and-break characteristic of substances showing selective absorption.

Indeed we come to the conclusion that isorropesis is the cause of color in the aromatic series as well as in the aliphatic series. In both series the two modifications which must always be in statu nascendi have actually been shown to exist. The change of linking, therefore, that must accompany the transformation of one into the other is certainly to be considered as the source of the oscillations which give rise to vibrations in the ether of a free period corresponding to those in the visible region of the spectrum, and hence the development of color in the substance. The application of these ideas to the interpretation of color among inorganic compounds is yet to be made. There seems, however, no doubt but that, where residual affinity exists, there may arise some form of oscillation, caused by the make-and-break of these induced linkings as brought about by the molecular movements, which will record itself in definite vibratory motions of the ether and consequently, if these vibrations are of low enough frequency, will indicate color in the compound. Not until something of a more definite nature is known as regards the true spatial arrangement of the atoms in these compounds, can anything of positive value be postulated concerning the disturbances which certain atoms may bring to bear upon other atoms or groups of atoms in the molecule. Consequently the periods of oscillation that correspond to many of our well known colored salts have received no explanation in terms of those periods so definitely established among the carbon compounds; periods which through spectroscopic evidence have been made to reveal so much concerning the internal vibrations of the atoms in the molecule and of the disturbances within the atoms themselves.

  1. The lines of force binding two atoms and constituting an electrical field between these charged atoms is conveniently regarded as made up of tubes of force, each with its positive electrical charge at one end, the beginning of this tube, and its negative and equal electrical charge at the other end or termination of the tube. Each Faraday tube, therefore, encloses a charge of electricity of unit value or that denoted by one single electron and consequently an atom that is univalent must enter into combination by means of one Faraday tube of force, one that is bivalent by two such tubes, etc. The positive atoms are those formed by the loss of electrons and the negative atoms are those which can take up these same electrons.
  2. Chem. Soc. Trans., 71, 1013, 1897.
  3. Chem. Soc. Trans., 89, 498, 1906.