TEMPER COLORS1 BY C . W. MASON
Ever since the time of Newton, when the nature of the interference colors produced by thin films was first recognized, it has been generally assumed that the so-called ‘(temper colors” of metals were examples of this phenomenon.2 This view is widely held in works on Metallurgy and the change of color as tempering continues is considered to be due to the increasing thickness of the oxide film formed. The influence of temperature and of time on the film production and on the reactions of tempering in the metal itself are comparable, so that temper colors serve as a more reliable guide to tempering than could information as to temperature alone. The above explanation has been put in question by Mallock3, who says: “I had thought, and I believe it has generally been assumed that the colours of tempering were instances of the ordinary interference colours of thin plates, but the following simple experiment seems to prove conclusively that this cannot be the true explanation. If the colours were due to a film of appropriate thickness, a reduction of that thickness ought to change the colour; the blue should change to green, orange to yellow, and so on. I found however, that if the tempered steel surface was gently polished, until the clean surface of the metals was reached, there was no change of colour during the process, the blue remaining blue, and the yellow yellow, until the whole of the colour was removed. The intensity of the colour decreased as the film became thinner, but the character remained the same.” This observation, unquestionably a correct one, has raised the whole problem again, and later workers have accepted Mallock’s conclusion without sufficient examination of its justifiability, so that we have a number of writers ascribing the colors to selective reflection, diffraction, or resonance phenomena because they cannot be due to thin films. Before ruling out this explanation, however, it is desirable to survey the evidence for it rather carefully, and the present paper was written with this in mind. Previous study of various types of structural colors4 involved the development of criteria by which the different kinds might be distinguished from each other, and these served as a basis for the study here presented. The most powerful argument against the non-structural basis of temper colors is the fact that the same sequence of color may be observed, wholly or in part, from a number of different metals acted upon by a number of ‘This paper is a necessary corollary of experiments supported by a grant from the Heckscher Foundation for the Advanckment of Research, established by August Heckscher a t Cornel1 University. I n the seventeenth century metal powders were given various colors by heat treatment, for use in decoration. Beckmann: “History of Inventions,” 2, 160, (1814). Proc. Roy. SOC.94A, 561 (1918). Mason: Jour. Phys. Chem. 27, 201, 401 (1923).
I234
C. W. MASON
different gases to form films of widely different chemical nature. Silver iodide or sulphide, copper or iron oxides, in fact, any case where a metal is slowly acted upon by some substance capable of forming an appreciably transparent and coherent product on its surface, will exhibit, to a greater or less degree, the same sequence of colors as the action proceeds. The colors formed are, in order, as follows: pale yellow, straw yellow, red orange, purplish red, indigo, blue, green, yellow, orange, purplish red, blue, green, purplish red, etc. with perfect gradation of each into the next. While the colors which develop first are very similar in the different cases, the range of the series is limited by various factors, so that the colors produced by more extended action upon various metals do not correspond strictly to each other. The general likeness in the series of colors produced by surface action is so marked, however, that the identical nature of those obtained on different metals would be self-evident a t once, were it not for this dissimilarity under further action. The probable explanation of this will be discussed later. Another very characteristic property of temper colors is the change in hue which they exhibit with changing angle of incidence. This is best observed when the color is uniform over a considerable area, or where some definite spot on the surface can be watched without the observer’s attention shifting with the color. Naturally, there is a marked decrease in saturation as the angle of incidence increases; but, accompanying this, the hue itself changes, in a manner that is definite for a given color, whatever may be its chemical nature, and the color produced, to put it briefly, is one immediately preceding the original color, in the series given above. I n other words, the color observed at grazing incidence is one which corresponds to that obtained by somewhat less chemical action on the metal, seen a t vertical incidence. For example, purplish red will change to orange a t grazing incidence, orange to pale yellow, etc. The character of the color sequence, and the change with angle of incidence, of course, are the foundation of the thin-film theory of the origin of temper colors. I n terms of this theory, the observed facts may be described thus: As chemical action proceede, increasing thickness of the film formed gives rise to a series of interference colors (Newton’s series) increasing in order with the extent of the action. As the angle of incidence is increased, these colors change to ones preceding them in the series, that is, to colors of thinner films. The concordance, as far as it has been carried, is adequate, and the only serious discrepancy is that stated above by Mallock. The thickness of the color film on metals has not been measured directly, but soap films showing similar colors are of the order of magnitude of 0.25p, for purple to deep magenta, (of the first order)’, which corresponds to a common temper color of steel, and if the index of refraction of the film itself (perhaps 3.+) is taken into account, this would indicate an oxide film only about 0.1p thick. It is hardly t o be expected, therefore, that “polishing” with abrasive particles of 1
Boys: “Soap Bubbles,” 149 (1912).
TEMPER COLORS
1235
order of magnitude of the thickness of the film itself, or more probably, several times this, should do anything except plow off the film without any progressive and uniform thinning, any more than in “rubbing down” a varnished surface with coarse sand paper. If the polishing were of the nature of surface flow, uniform reduction of the thickness of a film of brittle and imperfectly adherent oxide is not very probable. Mallock has not shown that it is possible to carry out his polishing experiment successfully in the case of a color that is admittedly due to interference. Chemical attack would be much more likely to prove something, one way or the other. When a brilliant blue temper color on highly polished steel is covered with a very dilute solution of nitric acid, the color does not disappear all a t once, but instead passes through a sequence of colors, the exact reverse of that which is observed when the film is forming. It is possible to stop this action at any point by washing off the acid, and comparison of treated and untreated color films leaves no question as to the above facts; the blue may be changed to reddish purple, or, by longer exposure to the acid, to straw yellow. On reheating such a specimen, the original colors may be wholly or partially restored, passing through the sequence in the reverse direction. The above behavior may be duplicated with oxidized copper, the same lowering of the order of the color being observed. Nitric acid is not suitable, but dilute hydrochloric acid (concentration determined by trial) attacks the oxide slowly and thins down the film evenly. Study of the colors of thin films with polarized light shows that at angles of incidence greater than the “polarizing angle” their color, for vibrations in the plane of incidence, is complementary to that for vibrations normal to this plane. This phenomenon is readily observable in the case of the “temper colors,” and points definitely to their identity with thin film colors, particularly since all hues of the series show this behavior, a fact hardly to be expected from a group of substances showing selective reflection. Incidentally, accurate measurement of the “polarizing angle” ought to furnish a means for determination of the approximate index of refraction of the film (n= tan. angle of incidence), and possibly for its identification. This angle is large, and indicates a very high index of refraction for the surface film. Mallock’ studied the behavior of temper colors with polarized light: “The colours of tempering are best seen by polarized light, and their intensity is greatest a t the angle of maximum polarisation. When eo observed the blue changes, as the angle of incidence increases, through reddish-purple to a dark orange and finally to a straw yellow. The yellow and orange parts, on the other hand, change but little, becoming rather more intense a t the angle of maximum polarisation; but the whole surface, which has been coloured by the tempering, assumes a nearly uniform yellow when the incidence is large. Thus the blue moves toward the red end of the spectrum, while the orange does the opposite, and although similar changes occur in the case of the higher order of Newton’s rings, the thickness of plate required ~~
1
Proo. Roy. SOC. 94 A, 561 (1918).
1236
C. W. MASON
is far greater than that which produced the colours of tempering. It would seem probable, therefore, that the latter colours are due to some form of selective opacity depending on damped molecular periods comparable with the wave period,l rather than on a structure comparable with the wavelength.” As regards the change of color with increasing angle of incidence, Mallock’s own report corresponds perfectly to the behavior of a film of a thickness corresponding to about blue of the second order. It is not a question, however, of the “end of the spectrum” toward which the colors move. People have been confused by this in the case of iridescent feathers. Thin-film colors change to those corresponding to thinner films as the angle of incidence is increased, and this is exactly what Mallock observed. As he says, the blue changes “through reddish purple to a dark orange and finally to a straw yellow,” precisely as does a thin-film color of the same hue. This, of course, is the reverse of the sequence of color changes which takes place when the film is forming (growing thicker). His statement regarding the thickness of film necessary is open to question, for the colors he describes apparently are in the second and upper first order, and for these only a very thin film is necessary (probably less than 0.3~). If Mallock’s studies had been made with ordinary unpolarized light, the difference between the behavior of temper colors and that of substances shbwing selective metallic reflection (surface color) would have been more apparent, for while thin film colors change markedly with unpolarized light as the angle of incidence is increased, surface colors from substances showing selective reflection (magenta, malachite green, crystal violet, etc.) exhibit little if any change in hue under these conditions. Any theory of selective reflection as the cause of the colors necessitates postulating several differently colored oxides of each metal, formed always in the same sequence and, in some instances, of two or more oxides showing the same selective reflection. Apparently there is no direct evidence to support such a series of compounds. Beilby2, in describing the temper colors of steel, ascribes the straw yellow which first appears to a thin film of Fe203; a somewhat thicker film is red, the color of thin films of hematite; the brilliant blue which next appears is considered to be the eurface color of the Fe2O3,which is now so thick as to be opaque, and to show no body color. However, no massive oxide of iron exhibits a. blue surface color, and the identical appearance of thin films of colorless substances (e.g., AgBr) is evidence that the series observed does not require such postulates for its explanation. Then too, presence of two whole “orders” of color can hardly be explained on the basis of anything other than thin films. I n the case of some of the This must be true of all pigment colours, but something more is required to explain the dependence of the colours of tempering on the angle of incidence] a feature which is strongly marked also in the case of many of the aniline colours when examined as dry films by reflected light. (Mallock.) “Aggregation and Flow of Solids,” 62. (1921.)
TEMPER COLORS
I237
more readily oxidized metals, copper, for example, three “orders” or more may be observed. The resemblance to the well-known sequence of thin-film colors is so striking that it is natural that the great body of opinion on this subject inclines to this explanation, and it is unfortunate that Mallock’s experiment, inconclusive as it is, should have been allowed to outweigh such obvious resemblances’. Another experiment which adds a little to the thin-film explanation, depends on the fact that when exposed thin films are brought in contact with a, medium of index of refraction near their own, the intensity of their colors is markedly decreased, and vanishes if the index of the film and of the external medium are identical. On covering a series of “temper colors” on steel with a liquid of index of refraction greater than 1.8, the diminution of intensity of reflection is very evident, as compared with that of the uncovered film. Care must of course be taken to eliminate effects due to reflection from the surface of the liquid itself. Since any oxide which may exist probably has an index of refraction a t least as great as that of hematite (ca. 3.0)~ complete disappearance of the color is not to be expected; but the behavior observed corresponds very well to that of a color-producing thin film. Where the index of refraction of the film is more nearly matched, the effect is very striking; burnished silver becomes coated with a film of AgBr on exposure to bromine vapor, and colors corresponding to the first three complete oiders and part of the fourth may be recognized. Covering the surface with a liquid of about I .8 index of refraction causes almost complete disappearance of the colors. [nAgBr = 2.21 On the other hand, selective reflection is ordinarily accompanied by anomalous dispersion, and substances exhibiting it show marked change in color with change in index of refraction of the surrounding medium. This is very strikingly exemplified by the solid dyestuffs, but there is no indication of it in the case of any of the temper colors. The change of color with angle, exhibited by films on metal surfaces, is not as marked as that observed with oil films on water; this is on account of the high index of refraction of the metallic oxides, sulphides, halides, etc., in question. Even when the light is grazing, the angle in the film is small (less than the critical angle), and, as a consequence, the retardation is less than if the light traveled a path more inclined to the normal. Newton2 recognized this general property and its application t o films on metalB: “I have sometimes observed, that the colours which arise on polished steel by heating it, or on bell-metal, and some other metalline substances, when melted and poured on the ground, where they may cool in the open air, have, like the colours of water-bubbles, been a little changed by viewing them at divers obliquities, and particularly that a deep blue, or violet, when view’d very obliquely, hath been changed to a deep red. But the changes of these colours are not so great and sensible as of those made by water. For the
+
~~
-
Even Wood [Phil. Mag. ( 6 ) 38,98 (1919)Jseems to have been put off the track by the “polishing” experiment. “Opticks,” 194 (1721).
I 238
C . W. MASON
scoria or vitrified part of the metal, which most metals when heated or melted do conthually protrude, and send out to their surface, and which by covering the metals in tho form of a thin glassy skin, causes these colours, is much denser than water; and I find that the change made by the obliquation of the eyes is least in colours of the densest thin substances.” Mallock suggests that the colors are due to diffraction or at least he points out that in some instances the diffraction systems that he was studying gave a sequence of colors very close to that of the colors of tempered steel, and Hinshelwoodl states definitely that the colours are due to diffraction. “That the colours shown by oxidized copper and by tempered steel are due to diffraction was pointed out by Mallock, who found that a film of given tint could be ground away while retaining its colour, which must therefore be dependent not upon its thickness, asit would be if interference were the cause, but upon its minute structure.” Corroborative evidence is cited by Raman.2 Any theory of diffraction as the cause of temper colors is open to the cogent objection that if this is the case the colors ought to disappear in uniform light, and ought t o be observed, not in the direct line of reflection, but to the side of this. Neither is the case. Hinshelwood3 gives some experimental etudies of the oxidation of copper, which are of interest in the light of the preceding pages: “When the surface of a metal is exposed to the action of a gas with which it reacts chemically, brilliant colour phenomena are frequently produced. I n Rome cases the colours are recognised as diffraction colours, produced by the scattering of light in the surface film, and not as simple interference effects. When this is found, the film must have a more or less complex structure, fine-grained compared with the wave-length of light, but of a coarse-grained granular nature compared with molecular magnitudes. The work of Beilby has drawn attention from other points of view to the complex structure which the surface layers of metals, or thin films of metals, may assume. “It seems t o be of considerable importance to correlate the chemical activity of such surface films with their structure, in view of the bearing this correlation may have on the problems of heterogenous catalysis. A first attempt in this direction is made in this paper, the reaction investigated being the oxidation and reduction of a copper-copper oxide film on the surface of metallic copper. “The general nature of the phenomena observed is as follows: When bright, rolled copper foil is exposed to the action of oxygen gas at low pressure, and at 200°-3000C., a film of oxide forms on the surface, and, as oxidation proceeds, the velocity of oxygen absorption diminishes considerably, owing to the difficulty of penetration of the oxide film.The surface assumes various tints, all of which, however, are quite faint, finally becoming black. I t s metallic lustre is retained throughout. After reduction, a second oxidation is brought about very much more readily than the first, and the surface ‘Proc. Roy. Soo. 102A, 322 (1923). *Nature. 109, 105 (1922). a Proc. Roy. SOC. 102 A, 318 (1923).
TEMPER COLORS
1239
assumes a new series of tints, brighter than the first, and even qualitatively distinct from them. Successive oxidations and reductions take place more and more readily until, ultimately, a limiting rate appears to be attained, which may be some hundreds of times greater than the original rate. As the surface becomes more and more active chemically the diffraction colours simultaneously increase in brilliancy. Ultimately a permanent colour sequence, with tints of great brilliancy, is established, which may be traversed time after time. The surface loses its metallic lustre, and when in the fully activated condition the copper is salmon-pink in colour. “During the activation process the structure of the surface film seems to change in the following way: At first the surface layers of the copper foil are compact as a result of its mechanical treatment; during the successive stages of activation by alternate oxidation and reduction, the copper atoms in the surface film are able, under the influence of surface tension, to aggregate themselves more and more completely into small discrete units in what Beilby calls ‘open formation,’ and the film assumes a granular structure freely permeable to oxygen. From the experiments to be described it is possible in principle to determine the size of these small granules, and a.1though there are difficulties in the way of finding accurate values, it emerges clearly that the order of magnitude of their diameter is only a small fraction of ~ p .The view has been expressed that metallic grains as small as this assume a spherical f0rm.l. But Sir George Beilby has kindly pointed out to me that the evidence of microphotographs shows the film to be lenticular in structure. In the last section of this paper the magnitude of the granules is calculated, for the sake of simplicity, on the assumption that they are spherical. The radius thus determined will be their mean or ‘effective’ radius. “Each small unit of copper will be oxidised independently, and the extent to which it is converted into oxide determines the colour of the diffracted light. The amounts of oxygen absorbed, corresponding to various wellmarked steps in the colour sequence, has been determined, and may, incidentally, be of interest in connection with the optics of the phenomenon. “There is a curious difference between the phenomena observed during reduction and those observed during oxidation, which suggests a n interesting mechanism for the gradual activation of the film.” “In the oxidation of copper, the brilliancy of the colour increases pro rata with the chemical activity, the granular film which most effectively scatters light presenting the largest surface to the action of the oxygen. When fresh copper is oxidised the colours are faint and the sequence is not constant, but is usually silvery or steely, pale straw, violet, black. The permanent colour sequence is purple, blue, green, very light green, (almost yellow), purple, blue, black. These are very brilliant, except the second appearance of purple and blue; they appear uniformly and are quite constant. in hue from time to time. . . . , 1
Maxwell Garnett: Phil. Trans. 205 A, 279 (1906).
I 240
C. W. MASON
“The oxidation of an active film must be imagined to consist in the independent oxidation of each of the small granules composing it, a layer of oxide being formed on the surface of each and extending inwards. The size and composition of each thus changes, and consequently the colour of the film alters. Thus, when the granules consist of approximately one-third oxide and two-thirds copper, the colour of the scattered light is bright blue. Within the boundary of each granular unit of copper or of copper oxide there is no reason t o doubt that the density of the material is normal, the low mean density of the film being due to the spacing of the granules. “The effects observed during reduction differ in a remarkable manner from those observed during oxidation. No diffraction colours appear when a black film is reduced, the colour of the film a t any stage of reduction being simply a shade of brown produced by the combination of the red colour of the copper and the black of the oxide.” Hinshelwood’s experiments were duplicated roughly by oxidizing burnished copper in air and reducing it in illuminating gas, this being repeated several times. The color sequence produced by the oxidations was reversed by the reduction, and development of the matte character of the surface was very apparent. This was retained after reduction, and resembled unburnished electrolytic copper. The microscopic appearance of the surface corresponded with this; a finely reticulated or granular structure was apparent, seemingly existing not as bosses, but rather as microscopic fissures in the surface. On repeating the oxidation and reduction a considerable number of times the oxide film became so brittle that it flaked off. This behavior on a coarse scale is consistent with the development of a reticulated microscopic structure on alternate reduction and oxidation. Beilbyl discusses at some length the development of porous or lenticular character in surface films on metals, and evidently some such structure plays a part in the optical behavior of the material, as well as in allowing the oxidation to proceed instead of “passivation” taking place after the first formation of the oxide film. The color sequence which Hinshelwood reports (“purple, blue, green, very light green, purple, blue, black”), corresponds to the series of thin-film colors of the second and lower third orders; the first order (yellow, red) is obscured by the color of the copper, and the upper third by the opacity of the oxide formed. The increased brilliancy which he observed is due no doubt to the matte surface developed on the copper by the oxidation and reduction, since the same apparently enhanced depth of color is observable on other metals which are only coarsely polished and then tarnished. Such matte character enables the color to be observed through a relatively wide angle of vision instead of a t the angle of reflection only, and also decreases the “diluting” effect of the light reflected through the film from the metal surface beneath it. Both of these effects would serve to increase the apparent brilliancy as ordinarily observed. At the same time the specular metallic lustre is mark‘’Aggregationand Flow of Solids,” (1921),
TEMPER COLORS
I 241
edly lessened, and the resemblance to pigment coloring of the surface is heightened. These factors also govern the appearance of any matte metal surface, not covered by a color film. The differences between electrolytic and burnished copper or silver are striking enough to emphasize their significance in the case of the colors observed when the metals are tarnished. It is pcssible to overlay such a matte metal surface by a thin film of varnish, prepared by allowing a drop to spread and harden on still water. Here we have a film which unquestionably owes its color to interference, yet when it is brought into contact with the metal the appearance is identical with that produced by tarnishing. Similar films on burnished metal show very much less vivid colors. On the whole, the brilliancy of the colors sometimes observed does not appear to indicate any different origin from those of burnished metal surfaces. Raman’ seems to have been troubled by this appearance, and in his paper mentions another property of the colore which throws light on their physical nature, and is closely related to some of Hinshelwood’s ideas on the structure of the surface film which causes them: “The well-known and characteristic tints that appear on the surface of a tarnishable metal when it is heated in contact with air have been usually regarded as interference colours due to the formation of a thin film of oxide on the surface of the metal. The correctness of this explanation has recently been questioned2 and rightly so, as a continuous film on a strongly reflecting surface cannot on optical principles be expected to exhibit such vivid colours as those observed. I have recently made some observations which shed a new light on this subject. It is found that the missing colours complementary to the tints seen by reflected Jight appear as light scaltered or diffracted from the surface of the metal. I n other words, if a plate of blue-tempered steel be held in a beam of light and viewed in such a direction that the regularly reflected light does not reach the eye, the metal shows a straw-yellow color and not the usual blue. It will be understood that the scattered light, being distributed over a large solid angle, appears much feebler than the regularly reflected colour, and in order to observe the effect satisfactorily the metal should have a smooth polished surface before being heated up. Scratches and irregularities show the ordinary colour of the film, and not the complementary tint. The most attractive effects are those exhibited by a heated copper plate, both on account of the vividness of the colours and on account of the ease with which the surface can be given a satisfactory polish. “It is clear from the observations mentioned above that the colours under discussion are in the nature of difraction effects arising from a film which is not continuous, but has a close-grained structure. Interesting effects are observed when the surface of the illuminated plate is viewed through a nicol, the colour and intensity of the scattered, as well as of the regularly reflected, beams varying as the nicol is rotated about its axis. The most striking Nature, 109, 105 (1922).
* Proc. Roy. SOC.94 A, 561 (1918).
1242
C. W. MASON
effect is obtained when the direction of observation is nearly parallel to the surface of the plate. The scattered light in this case is nearly completely polarised, and the color of the regularly reflected light changes nearly to its complementary when the nicol is turned through goo. The phenomena strongly recall to mind the observations of R. W. Woodl on the colours of a frilled collodion film on a silvered surface, which have been discussed by the late Lord Rayleigh2,and it seems probable that the explanation of the phenomena will ultimately be found to be somewhat similar in the two cases.” This letter moved the editor of Nature to append the folIowing note:
“Mr. Mallock has shown that the colour of the oxide film is an intrinsic property of the material of which it is composed, and the material retains this property as it is gradually ground down from its original thickness to the vanishing point. Sir George Beilby’s observations have confirmed this, and have further shown that the film is an aggregate in open formation through which oxygen molecules can penetrate to the metallic surface. For each temperature above the temperature range the thickness of the film is determined by the porosity of the aggregate to the oxygen molecules at that temperature. Direct experimental observations have shown the part played by time of heating at any given temperature. For example, a t 27soC, a deep purple was reached in ten minutes, and this changed to blue from the margin inwards during a further period of twenty minutes. It was thus shown that the watch-spring blue, which could immediately be produced by a temperature of 3oo0C, should also be produced by heating a t 275OC for thirty minutes. Sir George Beilby’s view is that the intrinsic colours of the films which are produced at different temperatures result from changes in molecular aggregation in relatively open formation of a similar nature to thoee which have been shown to occur in thin metal films, e.g. gold. This is referred to in his recently published volume entitled “Aggregation and Flow of Solids”, sections 3 and IO.” Although, as Raman points out, the “temper colors” of metals are similar in some respects to those of “frilled films,” it must be borne in mind that this does not preclude their explanation in terms of ordinary thin films. The distinctive feature of frilled films is their power of scattering light of complementary color to that which they reflect directly. It is admitted that the color of the light reflected directly from such a film depends on the thickness of the film in a manner identical with that of a non-frilled thin film. The fact that the light scattered by frilled films is complementary in color to that directly reflected by them renders it also dependent on the thickness of the film, and points to its being related to the transmission color, which is also complementary to the directly reflected color, though less saturated. Probably the “frilling” serves, in some way not definitely understood, to scatter the colored light which would be transmitted by a film not “Physical Optics,” 172 (1914). *Phil. Mag. ( 6 ) 34, 423 (1917).
TEMPER COLORS
1243
backed by a reflecting surface, and a t t’he same time eliminates the loss in saturation due to the reflection from the metal surface itself. The effect of “frilling,” whatever its real explanation may be, is manifest in the scattered light, and is viewed from a point considerably out of line of direct reflection best when the illuminating beam is normal to the surface and the line of vision is grazing. Under such conditions the “temper colors” exhibit more or less scattering of a color complementary to that which they reflect, but this scattering is not nearly as marked as in the case of the collodion films of Wood, and under ordinary conditions plays no part in the appearance which is observed on a piece of metal. Any change in a thin film which will increase its opacity somewhat without decreasing the reflection from the outer surface will result in increased brilliancy of color1 and the frilling no doubt accomplishes this to some extent. The naturally strong absorption of -metal oxides is also a n important factor and helps t o account for the vivid appearances observed, which Raman says “a continuous film on a strongly reflecting surface cannot on optical principles be expected to exhibit.” The effect of absorption in the film itself is exemplified by the case of asphaltum films on white paper. Such films, thin enough to give interference colors, are distinctly brownish by transmitted light, and the colors seen against a white background are distinctly more .brilliant than those of gold-size films, which appear colorless by transmitted light. In a sense, this is as if the asphaltum film were serving as its own dark background, with the enhancement of color that accompanies such an arrangement. A similar appearance is exhibited by “burned out” or old incandescent lamp bulbs in which the metal filament has been volatilized and deposited on the glass. This deposit probably consists of the metal itself, in a very fine-grained and uniform layer. Its opacity is very marked, yet the colors of the deposit are very striking, and resemble closely the temper colors of steel. Where the deposit is considerably thicker it appears almost opaque, and of a steely metallic appearance by reflected light. Here again we have a strongly absorbing substance showing colors in thin films rather more brilliant than in the case of colorless substances. On the whole, it appears that the objections raised to the thin-film theory of temper colors may be rather readily reconciled with this view, while any conclusive evidence against the theory, or in favor of other explanations, seems to be lacking; on the other hand, these latter are open to serious question. The conclusions of this paper are as follows:I.
Temper and similar colors of metals are due to interference of light, reflected from the two surfaces of a more or less transparent film of a compound of the metal formed on its surface, in a manner corresponding precisely to Newton’s rings. Wood: “Physical Optics,”
171 (1911).
I244 2.
C. W. MASON
Any reticulated or granular structure of the film adds t o the vividness of the color.
3. A moderate degree of opacity of the film enhances the color. 4.
The color of the material of the film itself obscures the interference colors in the case of thicker films.
5 . Properties similar to those of “frilled films” are probably due to the fine structure of the film, but these are of minor importance in the appearances observable under ordinary conditions. 6.
Diffraction or selective reflection are inadequate as explanations of the nature of the colors, and inconsistent with their behavior. Cornell University.
