1 K4^ UJ 'it O o u CO ri tr ^ r=\ ^^^ru ^ ° O r-^ > tr _r^ >-— — ■ ■ - 1 i^^^^^H .ijj ISHj^^^^^^^^H ^^^^^1 M^AIN 1/^ ' J y(uuUu^/^^^/UL^ Digitized by the Internet Archive in 2007 with funding from IVIicrosoft Corporation http://www.archive.org/details/coloritsapplicatOOIuckuoft QC^==- kj 0) Qi COLOR AND ITS APPLICATIONS BY M. LUCKIESH DIRECTOR OF APPLIED SCIENCE, NELA RESEARCH LABORATORIES NATIONAL LAMP WORKS OF GENERAL ELECTRIC CO. AUTHOR OF " LIGHT AND SHADE AND THEIR APPLICATIONS," " THE LIGHTING ART," " THE LANGUAGE OF COLOR," "ARTIFICIAL LIGHT, ITS INFLUENCE UPON CIVILIZATION," "LIGHTING THE HOME," ETC. 150 Illustrations — 4 Color Plates — 34 Tables SECOND EDITION ENLARGED NEW YORK VAN NOSTRAND COMPANY EIGHT WARREN STREET I92I COPYRIGHT I915, I921, BY D. VAN NOSTRA ND COMPANY THE PLIMPTON, PRESS NORWOOD -MASS-U-S 'A PREFACE The aim of this book is to present a condensed treatment of the science of color. An attempt has been made to cover as many phases of the subject as possible within the confines of a small volume. During several years of experimental work in the science of color I have been brought into contact with many persons interested in its applications, and the desire has been frequently expressed for a book that treated the science of color as far as possible from the viewpoint of those interested in the many applica- tions of color. These applications are constantly in- creasing in scope and interest. With this viewpoint in mind I have attempted to treat the subject, exercis- ing my judgment in drawing freely from the work of other investigators in order to make the volume as comprehensive as possible. I do not feel that the work comprises a complete treatment, for there are many interesting phases of color science that have been barely touched upon, and some that have been purposely omitted, because of the danger of straying too far afield. It is believed, however, that this treatise will be helpful to those interested in any of the arts involving the science of color. I have referred to my own investigations quite freely, but trust that this will not be attributed to a lack of per- spective. Naturally much of the text involves my own conclusions, but I have aimed to include only those that are supported by experimental data, be- cause only in so far as they are thus supported does ill iv PREFACE the work become authoritative. Many unsolved prob- lems have arisen throughout the text, which em- phasizes the need for more workers in the field. No attempt has been made to present a complete bib- liography of even the recent work in this branch of science; but references have been given freely, which, if followed, will provide a substantial beginning to the almost endless chain of material available. It is a pleasant duty to record my acknowledg- ments to the management of the National Lamp Works of the General Electric Company, whose broad- minded spirit in establishing the Nela Research Laboratory has made this work possible, and to the director of the laboratory and members of the staff, who always have given freely of their time and counsel. SECOND EDITION Some changes have been made in the original text and an extensive chapter has been added. This con- sists of useful data and methods for their use. M. LUCKIESH September, 1920 CONTENTS CHAPTER I Page LIGHT 1 Wave Theory. Electro-magnetic Theory. Radiation and Light Sensa- tion. Temperature and Radiation. Spectra of Ulimiinants. CHAPTER II THE PRODUCTION OF COLOR 23 Refraction. Diffraction. Interference. Polarization. Reflectioiij, Ab-_- sorption, and Transmission. Color of Daylight. Color Sensations Produced by Colorless Stimuli. Fluorescence and Phosphorescence. Useful Filters. CHAPTER III COLOR-MIXTURE 54 Subtractive Method. Additive Method. Juxtapositional Method. Simple Apparatus for Mixing Colors. CHAPTER IV COLOR TERMINOLOGY. 69 Hue, Saturation, and Brightness. Tri-color Method. Color Notation. CHAPTER V THE ANALYSIS OF COLOR 86 .The Spectroscope. The Spectrophotometer. The Monochromatic Colorimeter. The Tri-chromatic Colorimeter. Other Methods. Templates. Reflectometer. Methods of Altering Brightness Non- selectively. CHAPTER VI COLOR AND VISION 116 The Eye. B^^htness Sensibility. Hue Sensibility. Saturation Sensi- bility. Visui Acuity in Lights of Different Colors. Growth and Decay of Color Sens tions. Signaling. Other Uses for Colored Glasses. Vi CONTENTS CHAPTER VII THE EFFECT OF ENVIRONMENT ON COLORS 163 niiimination. After-images. Simultaneous Contrast. Irradiation. CHAPTER VIII THEORIES OF COLOR VISION 181 Young-Helmholtz. * Duplicity.' Hering. Ladd-Franklin. Edrigde- Green. CHAPTER IX COLOR PHOTOMETRY 191 Methods of Color Photometry. Other Means of Eliminating Color Dif- ferences. Direct Comparison and Flicker Methods. Luminosity Curve of the Eye. CHAPTER X COLOR PHOTOGRAPHY 213 Lippmann Process. Wood Diffraction Process. Color Filter Processes. CHAPTER XI COLOR IN LIGHTING 224 Artificial Daylight. Units for Imitating Daylight. Effect of Colored Sur- roundings. Color in Interiors. Color Preference. A Demonstration Booth. CHAPTER XII COLOR EFFECTS FOR THE STAGE AND DISPLAYS 272 Stage. Displays. CHAPTER XIII COLOR PHENOMENA IN PAINTING 282 Visual Phenomena. Lighting. Pigments. CHAPTER XIV COLOR MATCHING , 302 The niuminant. The Examination of Colors. CONTENTS vii CHAPTER XV THE ART OF MOBILE COLOR 312 Color Music. Its Relation to Sound Music. CHAPTER XVI COLORED MEDIA 327 Available Coloring Materials. Dyeing. Gelatine Films. Solvents. Lacquers. Celluloid. Phosphorescent Materials. Miscellaneous Notes. CHAPTER XVn CERTAIN PHYSICAL ASPECTS AND DATA " . . . . 344 Three Types of Colored Media. Pigments. Optical Properties of Pig- ments. Applications of Spectral Analyses of Pigments. Reflection- factors of Pigments. Spectral Analyses of Dye-solutions. Applications of Spectral Analyses of Dyes. Laws Pertaining to Colored Solutions. Dichromatism. Graphical Method for Using Spectral Data. Spectral Analyses of Glasses. Red, Yellow, Green, Blue, and Purple Glasses. Use of Spectral Analyses of Glasses. Influence of Temperature on Transmission of Colored Glasses. Ultraviolet Transmission of Media. Compounds Sensitive to Temperature. Transmission of Light by Fog and Water. Color Temperature of Illuminants. INDEX 407 COLORED PLATES Prismatic Spectrum Frontispiece Diffraction Grating Spectrum " Subtractive method of mixing colors Facing page 54 Additive method of mixing colors " " 54 Showing the effect of environment on the appearance of colors " " 163 Illustrating the effect of the spectral quality of the illuminant Daylight, below; ordinary artificial light, above . . " " 282 LIST OF ILLUSTRATIONS Figure Page 1. Radiation curve of an incandescent solid 8 2. Showing the relation between radiant energy and hght sensation ... 10 3. Showing the effect of temperature on the radiation from an incandes- cent solid (black-body) 12 4. Representative spectra 17 5. Distribution of energy in the visible spectra of various illuminants . . 20 6. Newton's experiment 23 7. Effect of the character of the slit of a spectrograph on the grating spec- trimi of the mercury arc 24 8. Dispersion curves of various optical media 26 9. Young's double-slit experiment illustrating the principle of the diffrac- tion grating 26 10. Diagrammatic illustration of polarized hght 31 11. The Nicol prism for obtaining plane-polarized light 33 12. Analyses of ordinary colors 36 13. Showing the variation in the spectral character of sunlight due to at- mospheric absorption 38 14. Benham disk for producing subjective colors by means of black and white stimuli 39 16. Diagrammatic illustration of the action of the rhodamine fluorescent reflector 44 16. Spectrophotographic analysis of the action of the rhodamine fluores- cent reflector 46 17. Screens for producing lights of the same hue but diflfering in spectral character 48 18. Ultra-violet spectra 60 19. Ultra-violet spectra 61 20. The subtractive method of mixing colors (colored plate) 21. The additive method of mixing colors (colored plate) 22. The color-wheel for showing complementary hues . 69 23. Maxwell disks 62 24. An erratic color-mixing disk 64 25. A simple color-mixer 64 26. A simple color-mixer for transparent or opaque media 66 27. Lambert's color-mixer 66 28. A shadow demonstration of the additive and subtractive methods of color-mixture 66 29. Illustrating a disk for approximating a prismatic spectrum 68 30. Disk * a,' for varying only the satiu-ation of a color. — Disk * b,' for vary- ing only the brightness of a color 71 31. The Maxwell color-triangle 73 32. Spectral complementaries 76 33. A color pyramid 76 34. The double pyramid (after Titchener) 76 36. A demonstration color-triangle 76 LIST OF ILLUSTRATIONS 36. The A. H.Munsell color tree 81 37. Prang's color and brightness scales 82 38. Ruxton's color mixture chart for prmting inks 82 39. A direct-vision prism spectroscope 86 40. A simple grating spectroscope 86 41. The spectrophotometer. 88 42. The Nutting pocket spectrophotometer 88 43. A small portable spectrophotometer for quantitative analysis .... 89 44. The variable sectored disk (after Hyde) 90 45. Scheme for reducing the amoimt of spectrophotometric work in ex- amining transparent colored media 91 46. Abney's spectrophotometric attachment for a spectrometer 93 47. Ives' spectrophotometric attachment for a spectrometer 93 48. Nutting's spectrophotometric attachment for a spectrometer .... 94 49. The Nutting monochromatic colorimeter 95 60. Analysis of two component color-mixtures 99 61. A simple method of converting a spectrometer into a combined mono- chromatic colorimeter, direct comparison photometer, flicker pho- tometer, and spectrophotometer 100 62. Illustrating the principle of the Maxwell ' color box ' 101 63. The F. E. Ives colorimeter 103 64. Koenig's sensation curves 104 65. Tri-color colorimeter measurements 104 66. Arrangement for using color filters before a photometer eyepiece . . 106 67. Arons colorimeter 108 68. Abney's template for carmine 110 69. Adaptation of Abney's scheme for the spectroscopic synthesis of color 111 60. The Nutting refiectometer 113 61. A vertical section of the human eye 116 62. Showing the effect of chromatic aberration in the eye 118 63. A simple achromatic lens 119 64. Limits of the visual field for colored and colorless lights 120 65. Brightness sensibility data. (See Table X) 121 66. Hue sensibility. (Steindler's Eye) 125 67. Hue sensibility, limen, and color scale 126 68. Apparatus for determining visual acuity in monochromatic lights . . 133 69. Visual acuity in monochromatic lights of equal brightness 135 70. Visual acuity in the mercury spectrum, the lines being reduced to equal brightness 136 71. The growth and decay curves for white light sensation. (Broca and Sulzer) 138 72. The growth and decay curves of color sensations 139 73. Showing the maxima attained by flickering Ughts at various frequencies 140 74. Showing the maxima of sensations produced by flickering red light on a steady green field (R), and vice versa (G) 141 75. Showing the relation between brightness and critical frequency for colored stimuli 145 76. Effect of contour of flicker on critical or vanishing-flicker frequency . 147 77. Effect of yellow-green glasses on vision imder a bright sky 155 78. Ultra-violet transmission curves of various glasses 158 79. Effect of the intensity of illumination on the appearance of a pigment 165 80. Illustrating why a purple appears differently under two different illuminants 167 81. Effect of brightness on the duration of the after-image 171 LIST OF ILLUSTRATIONS xi 82. Showing the effect of simultaneous contrast. The V*s are of equal brightness 174 83. Showing induction. Each band, though uniform in brightness, appears brighter at the right-hand edge 176 84. An arrangement for showing the reduction in the contrast effect by separating the two colored objects 176 85. An arrangement for showing the effect of simultaneous contrast and after-images 176 86. Illustrating irradiation 179 87. The evolution of the Ladd-Franklm gray molecule 187 88. The results of four methods of photometry. (Ives) 196 89. Spectral sensibilities of seleniimi and photo-electric cells compared with the spectral sensibility of the eye 200 90. Spectral sensibility of a panchromatic photographic plate 202 91. An accurate color filter for the panchromatic plate considered in Fig. 90. 203 92. Results by flicker and direct comparison photometers, illustrating dif- ferences including the Purkinje effect and a reversed effect . . 206 93. VisibiUty data. (See Table XVI) 209 94. Illustrating the standing waves produced in the Lippmann process. . 216 95. Illustrating the Wood diffraction process 216 96-98. Illustrating three processes of color photography 219 99. Illustrating the limitations of certain processes of color photography . 220 100. Ideal transmission screens for producing artificial daylight 230 101. Showing the loss of light when using the ideal artificial-dayUght screens with the tungsten lamp operating at 7.9 lumens per watt .... 231 102. Showing the loss of light when using the ideal artificial-daylight screens with the timgsten lamp operating at 22 lumens per watt .... 232 103. Showing the spectral analyses of two subjective white lights com- pared with the spectral analysis of noon sunlight 236 104. Showing the additive method of producing artificial daylight .... 236 105. Showing the relative amounts of light of the character of A and B (Fig. 104) necessary to produce artificial daylight by addition . . 237 106. Illustrating the effect of multiple selective reflection of light from a green fabric 248 107. Showing the relative proportions of red, green and blue components in the reflected Ught from a green fabric after various successive reflections 249 108. Screen for altering timgsten light to the same spectral character as carbon incandescent electric light ; c, d, e show the transmission curves of amber glasses of different densities 264 109. Comparison of ideal screen a, Fig. 108, with amber glass 256 110. Showing the preference or rank of a ntmiber of fairly saturated colors . 261 111. Wiring diagram of an experimental and demonstration booth . . . ." 267 112. Showing dimensions and locations of lamps in the demonstration booth 268 113. Illustrating the effect of colored light upon the appearance of six colored papers 273 114. Illustrating the changing of scenery by the use of colored lights . . . 276 115. Illustrating the disappearing effects produced on a specially painted scene by varying the color of the illuminant 276 116. Illustrating a flashing sign produced by properly relating the hue and brightness of the pigments with the color of the illuminant. . . 279 117. Showing the reflection coefficents of fairly saturated colors for day- light and tungsten incandescent electric light. (See Table XV) . . 286 xii LIST OF ILLUSTRATIONS 118. Showing the effect of the illuminant upon the appearance of a colored frieze 288 119. Showing the effect of the spectral character of the illuminant upon the values of a painting 290 120. Effect of distribution of light on the expression of a painting 293 121. Illustrating the optics of picture lighting 294 122. Spectral analyses of pigments 298 123. Spectral analyses of pigments. 298 124. Dlustrating the effect of the amount of the green components in blue and yellow pigments on the amoimt of ' black ' in the mixtures . 299 126. Diagrammatic illustration of the results of mixing blue and green pig- ments containing various amounts of green 300 126. The * Luce ' part for the ' Clavier k lumieres ' in Scriabine's * Pro- metheus * 315 127. Illustrating an instrument for studying the emotive or affective value of colors and color phrases ; Rimington's color code also shown 323 128. A color-mixture instrument for studying the emotive and affective value of colors and color phrases 324 129. Showing the relative positions of the colored lamps in the apparatus diagrammatically shown in Fig. 128 325 130. Michrophotographs of white cotton and silk fabrics against a black background 348 131. Spectral reflection-factors of pigments 352 132. Spectral reflection-factors of pigments 353 133. Spectral luminosities of pigments 360 134. Spectral luminosities of pigments 361 135. A study of a pigment (light chrome yellow) 362 136. Reflection-factors of pigments 368 137. Relative reflection-factors of pigments 369 138. Influence of the illuminant on the appearance of a pigment 370 139. Relation between spectral transmission-factor and depth or concen- tration of a solution of methylengriin 381 140. Relation between spectral luminosity and depth or concentration of a solution of rosazeine 382 141. Complete relation between thickness, wave-length, and transmission- factor for a gold ruby glass 383 142. Spectral transmission-factors of selenium glasses 387 143. Spectral transmission-factors of copper, sulphur, chromiimi, and uranium glasses 388 144. Spectral transmission-factors of gold glasses and combinations with cobalt ' . 389 145. Spectral transmission-factors of carbon glasses and combinations with cobalt glasses 390 146. Spectral transmission-factors of cobalt glasses 391 147. Spectral transmission-factors of iron and of manganese glasses . . . 392 148. Relations between spectral transmission-factor and thickness of a gold glass (23Au) 393 149. Relations between spectral luminosity and thickness of a gold glass (23Au) 394 160. Test of the relation between spectral transmission-factor and thick- ness of a blue-green glass 395 COLOR AND ITS APPLICATIONS CHAPTER I LIGHT 1. The word Light has acquired two meanings; one pertains to sensation and is therefore physiological and psychological in character, while the other refers to the external cause of the sensation and is there- fore physical in nature. As both meanings are used in the study of color, they will be distinguished wher- ever necessary; for example, light rays when imping- ing upon the retina of the eye produce the sensation of light. In order to understand the phenomena of color, a fair knowledge of the physical nature of light must first be acquired. Unfortunately the field to be covered in this book is too extensive to permit of a detailed treatment of this interesting subject; only those phenomena will be discussed that are essential to an understanding of the subsequent chapters. Those wishing to pursue this line of study further can readily do so by consulting the many excellent treatises on the subject. 2. Wave Theory, — The passage of a beam of light from a source (flame, sun, etc.) to a receiver (the earth, the eye, etc.) involves a transfer of energy, and the question arises as to how this transfer takes place. All around us in Nature, energy is contin- ually being transferred from one place to another and whenever such a transfer does occur something is moving. On the ocean, for example, energy is COLOR AND ITS APPLICATIONS transferred by water due to its wave motion. Moun- tain streams are carrying energy, which fortunately can be made to do useful work by means of a water motor, but here the energy is transferred by the onward flow of the water. In air the same two meth- ods of transferring energy are found; currents in the case of winds and waves in the case of sounds. Solids also can be made to transmit energy in these two ways; by currents as in the sand blast and by waves as in the case of sound and other elastic dis- turbances. Since currents and waves are such com- mon methods of transmitting energy, it is quite natural that they should be called upon to explain the transfer in the case of light. Light travels in straight lines, casts (comparatively) sharp shadows, is reflected from a smooth surface as a regular succession of rubber balls would be if thrown against the same surface, and in many other ways acts much like a current of particles would act if projected from the source of light at a high velocity. There is one phenomenon, however, that cannot be explained by the assumption of a current of par- ticles; under certain conditions two rays of light of equal intensity can be sent to the same spot in such a manner that the spot will be dark and not twice as bright as it would be if either ray were present alone. This fact is explained by assuming that light energy is transmitted in the form of wave motion for it is seen that if two equal waves are made to pass in the same direction through any medium, but in such a manner that the crests of one wave coincide with the troughs of the other, the two waves will annul each other everywhere, there will be no resultant wave, no transfer of energy, and hence no light at the spot in question. To this phenomenon was given LIGHT the name * interference,' but the term has been extended to include all the phenomena that may take place when two or more waves travel in the same me- dium at the same time. The foregoing case and all others in which there is destruction of motion are now grouped under the term, destructive interference; in contradistinction to this, there is constructive interference wherever the motion due to all the waves is greater than that due to one. The simplest case of the latter type is that of two equal wave trains traveling in the same direction at the same time but in such a manner that the crests of one coincide with the crests of the other. The two waves reinforce each other and the resultant wave has twice the am- plitude of the original waves. Another very important special case of interference is that to which the term * standing wave' or * stationary wave' has been applied. This occurs whenever two equal wave trains are passing in opposite directions through the same me- dium at the same time. The most common way of obtaining equal waves traveling in opposite directions is by means of reflection at a surface perpendicular to the direction in which one train is traveling. Stand- ing waves can be readily demonstrated by fastening one end of a long rope (preferably so that it hangs vertically) and by shaking the other end, timing the motion of the hand so that it is in unison with the reflected waves. It will be seen that some points of the rope remain at rest and others swing through a large amplitude. The points at rest are called nodes and the part of the string between the nodes is a segment. By varying the speed of the hand or the period of vibration the string can be made to vibrate in one, two, three, or more segments. A brief consideration of wave motion in general COLOR AND ITS APPLICATIONS and a few definitions may not be out of place here. In the first place, it is evident that in any wave motion, the parts of the medium do not travel as far as the wave. They remain each in its own region, each causing adjacent parts to move and in so doing gives up to the adjacent part some of its energy. The motion of the particles differ in various kinds of waves. " In waves in deep water each drop moves in a vertical plane in a circular orbit. In waves in shal- low water the orbit is an elongated ellipse. In media transmitting sound waves the motion of each particle is to and fro in a straight line in the direction in which the wave is traveling. For all waves the wave-length is the distance between any two suc- cessive particles that are moving through the same points in their orbits at the same instant. The ampli- tude is half a particle's path length (the diameter of the orbit). The period is the time taken for the wave to travel one wave-length. The frequency is the reciprocal of the period or the number of waves that pass a given point in a unit of time. If two waves are *in step' so that a crest of one occurs at the same time and place as a crest of the other, the two waves are said to be in phase. If a wave is confined to a surface such as that due to a pebble dropped in a quiet pond of water, the waves will be circular; any circle is a wave front , and the direction in which the wave is traveling at any point is that of the radius drawn to the point and is therefore per- pendicular to the wave front. In the case of light under ideal conditions, the wave will spread out in all directions from a point, so that the wave front will be spherical. The direction of propagation will again be perpendicular to the wave front along the radii of the sphere. LIGHT 5 Light energy or radiant energy passes through a vacuum. The phenomenon of interference has been explained by wave motion. Hence it is assumed that there is something in the vacuum that can move. This something is called the ether and is further assumed to penetrate all matter so that light waves always are ether waves; the properties of the waves may change as the matter imbedded in the ether is changed, but it is the ether and not the matter imbedded in the ether that is responsible for the propagation of the light waves. Some scientists, are not reconciled to this view but fortunately in this treatise we need not enter into the discussion. The adoption of the wave theory necessitates new and somewhat elaborate explanations for such simple phenomena as the rectilinear propagation of light; these have been made by the aid of Huy- ghen's principle, which states that each point on a wave front may be regarded as a new source of dis- turbance, sending out spherical waves, and that at any instant the new wave front will be the envel- ope of all of these secondary wavelets. By the aid of this principle it is at once evident that a light wave in going through a wide slit will pass on in such a manner that the sides of the slit cast a rather sharp shadow, whereas in going through a very narrow slit, comparable in width with a wave-length of light, it will pass on and spread out in all directions thus * turning a corner.' This phenomenon has been termed diffraction. It is helpful to visualize light waves by means of the water wave analogy as has been done in the foregoing, but it is well to guard against being misled by following the analogy too closely. For example water waves diminish in amplitude on account of 6 COLOR AND ITS APPLICATIONS dissipation of energy through molecular friction. In free space the amplitude of light waves (which is a measure of their intensity) has not been observed to decrease; in other words, there is no friction in the transmitting medium. 3. Electro-magnetic Theory, — At first it does not appear that there is any relation between light and electricity, but such a relation was predicted by Max- well in about 1870. This theory assumes light rays to be identical with the electro-magnetic disturbances which are radiated from a body in which electrical oscillations are taking place. Some years later Hertz actually produced these waves and by this discovery the electro-magnetic theory expounded by Maxwell was supplied with the necessary physical foundation. In enunciating this theory it is customary to state that the oscillating electrons in the atoms which constitute a body send forth through space pulses of electro-magnetic energy. The electron at present is supposed to be an atom of electricity. These electric waves emanating from a radiating body whether it be the sun or a red-hot poker have many properties depending upon their wave-length. Al- though all travel at the same velocity, about 3 x lo ^^ centimeters (186000 miles) per second, in free space, they differ somewhat in velocity in the ordinary trans- parent media. In glass, for instance, violet rays travel less rapidly than the red rays. All these rays represent energy and therefore regardless of wave-length have the property of producing heat when absorbed. Some- times the energy is not wholly converted into heat but enters into chemical reactions or is converted into electricity or radiant energy of other wave-lengths than those of the absorbed rays. Some of the rays, especially those of shorter wave-length than the visible LIGHT 7 rays, are very active chemically, affecting a photo- graphic emulsion, destroying bacteria and animal, tissues such as the outer membrane of the eye and causing sunburn; they are also largely respon- sible for exciting phosphorescence and fluorescence. Other rays have varying effects on organisms and play a more or less important part in the growth of plants. Rays within a certain range of wave-lengths, separately and in groups, produce the sensation of light and color. In other words the retina of the eye can be likened to a receiving station in wireless telegraphy which is * tuned' to respond to electro- magnetic rays of a certain (limited) range of wave- lengths. 4. Radiation and Light Sensation. — There are many ways of decomposing radiation into its various component rays. The rainbow is the result of one of Nature's means of dispersing the radiation from the sun into rays of various wave-lengths. The eye sees in the rainbow many colors, the most conspicu- ous being violet, blue, green, yellow, orange and red. These are seen to be of different brightnesses. If the retina were sensitive to rays of an infinite range of wave-lengths the rainbow would appear much wider than it does. That is, colors whose appearance can not be imagined, would be seen in the now in- visible regions beyond the violet and red, because energy corresponding to those wave-lengths is present in sunlight. The distribution of the energy among the different wave-lengths radiated by a hot solid is shown in Fig. 1. This curve is technically known as a radia- tion curve and shows that energy of a great range of wave-lengths is present. Such a continuous spec- trum is characteristic of the radiation from solid 8 COLOR AND ITS APPLICATIONS bodies. On the basis of the electro-magnetic theory of light, it may seem strange that rays of all wave- lengths are produced by the vibrating electrons. This may be pictured sufficiently well for the present purpose by assuming that in a solid body there is considerable damping of the vibrations and that other influences are present which result in the emission WAVE LENGTH Fig. 1. — Radiation curve of an incandescent solid. of no characteristic single ray or series of rays but, instead, of rays of all wave-lengths. The height of the curve at any point above the axis of abscissae or base line represents the relative amount of energy of that particular wave-length present in the total radia- tion. It will be noted that the amounts of energy of various wave-lengths are by no means of the same value. This characteristic of illuminants is of great importance in a study of colors as will become evident later. The region to which the eye is * tuned,' the visible spectrum, lies between V (violet) and R (red) which is exaggerated in relative extent for the sake of clearness. The relation between radiation and light sensa- tion is not simple. The ability of the various rays to produce light sensation is shown roughly by the dotted curve in Fig. 1. The maximum light sensa- tion is produced by rays in the middle of the visible spectrum, namely by those giving rise to a yellow- LIGHT 9 green sensation. Beyond the limits of the visible spectrum, V and i?, it is obvious that an infinite amount of electro-magnetic energy causes no sensa- tion of light. The total range of wave-lengths might be called the energy spectrum of this particular radi- ator. It is obvious that the greater the percentage of the total radiant energy confined to the visible spectrum, the greater is the * luminous efiiciency ' of the radiating body. In the production of light the total range of wave-lengths is of interest, but in the con- sideration of Color, interest is very largely confined to the visible spectrum. As the temperature of an incandescent body is increased, the energy of the shorter wave-lengths increases more rapidly than the energy of the longer wave-lengths. Considering the visible spectrum, the violet and the adjacent rays increase in intensity more rapidly with increase of temperature than the red and its adjacent rays thus causing the light emitted by an incandescent filament, for instance, to become bluer as its temperature is increased. Here it is well to note that when the various visible rays are permitted to impinge separately upon the retina each produces its own color sensation but when all the visible rays simultaneously stimulate the retina, as in the ordinary viewing of colorless objects, an integral sensation is produced. In the case of most common illuminants the integral sensation is an unsaturated yellow, that is, a yellowish white, while the combined sensations aroused by average day- light produce the integral sensation of white light. In the discussion of Fig. 1 it has been seen that energies in the various wave-lengths differ in light- producing effects. To this must be added the fact that the light-producing effect varies with the inten- 10 COLOR AND ITS APPLICATIONS sity. In Fig. 2, A represents the light sensation pro- duced in the author's eye by equal amounts of energy of various wave-lengths as measured with a direct comparison or equality-of-brightness photometer. The photometric field was a circle whose diameter sub- 0.40 0.44 (M Q52 Q56 050 M, WAVE LENGTH 0.64 0.6& Fig. 2. — Showing the relation between radiant energy and light sensation. tended an angle of about four degrees at the eye, and whose brightness was equivalent to that of a white surface illuminated to an intensity of 20 meter candles ; (a meter candle is the illumination received on a surface everywhere one meter distant from a source of one candle, the surface being perpendicular to the straight line from it to the light source). On decreasing the intensity and therefore the brightness of the photometer field to about one two-hundredth of its original value or to an equivalent of 0.1 meter candle on the foregoing basis, the relation between light and radiation become as shown in curve B. It will be noted that the maximum of the luminosity curve (as it is called) has shifted toward the shorter LIGHT 11 wave-lengths. In other words at the low illumina- tion the light-producing effect of visible rays of the shorter wave-lengths has not decreased as much as that of the longer wave-lengths. To illustrate by a simple experiment, suppose a red and a blue sur- face appear of equal brightness at a high illumina- tion. If the intensity of illumination is reduced to a very low value, the blue surface will appear much brighter than the red one. To further complicate matters it is found that even normal eyes differ some- what in spectral sensibility for experiment shows that the luminosity curves for various normal eyes do not exactly coincide (see #56). Curve C is plotted from Koenig's ^ data obtained at a very low illumination, practically at the threshold of vision. This phenome- non of shifting spectral sensibility which was discov- ered by Purkinje, and which bears his name, will be discussed in later chapters, and the quantitative rela- tion between radiation and light sensation will be fur- ther treated in the chapter on color photometry. 5. Temperature and Radiation. — As already stated the effect of raising the temperature of a heated solid body is to increase the luminous effi- ciency and also the relative amount of energy in the rays of shorter wave-lengths. These effects are shown diagrammatically for a solid body in Fig. 3. The numbers on the curves indicate the absolute black-body temperatures. The wave-length is given in terms of ten-thousandths of a centimeter, this unit being usually expressed by the Greek letter, /x. The rays to which the eye is sensitive lie between V and /?, respectively about 0.4/x and 0.7^. The eye in reality is sensitive somewhat beyond these wave-lengths but for practical purposes the amount of light sensa- tion produced by rays beyond these limits is usually 12 COLOR AND ITS APPLICATIONS / "X '450 L \ 1 \ i \ / 12^ V \ / / 700 lOOC^ •^ V -V ■R^ // y r ■ ^ ^ ** z L /^ 700 b^ S 0 2 3 4 5 6 negligible. It is seen that as the temperature rises the maximum of the radiation curve shifts toward the shorter wave-lengths. The maximum of the radi- ation curve of sunlight lies in the visible region. This has brought forth the suggestion that the eye 80 70 >. 60 S 50 ZL U ui 40 °= 20 10 0 0 I Jl, WAVE LENGTH Fig. 3. — Showing the efifect of temperature on the radiation from an incandescent solid (black-body). in its process of evolution has become most sensitive to the rays of such wave-length as are a maximum in the radiation from the sun. As the maximum of the radiation curve shifts toward the shorter wave- lengths, it is seen that a relatively greater proportion of the total energy is found in the visible region between V and R which accounts for the increase in luminous efficiency. One of the tendencies in light production is toward the development of materials and methods which will enable the light source to be operated at higher temperatures in order to appease the ever-present demand for higher luminous effi- ciencies. It is evident that the ideal light source emitting a continuous spectrum would be one that radiated no energy beyond V and R. The area under each curve is proportional to the total amount of energy emitted by the radiator at a certain tem- LIGHT 13 perature and the ratio of the area under that part of any curve included between V and R is propor- tional to the energy that can effect the eye. The ratio of the latter area to the former (for the same curve) is called the * radiant efficiency' of the radiator as a light source. To make the idea of radiant effi- ciency of practical value it must be combined with the relations between luminous sensation and radiation. 6. Spectra of Illuminants. — The spectral distribu- tion of energy in the radiation from different illumi- nants is of great importance in the consideration of color owing to the fact that the appearance of the colored objects depends upon the spectral character of the illuminant under which they are viewed. The variation in the spectral character of illuminants is due to the temperature and composition of the radi- ating body and also to the state in which it exists when radiating luminous energy. A simple means of producing light is that of heating a solid conductor by passing an electric current through it. At first it will emit invisible radiant energy known as infra-red rays. As the temperature is raised it will finally become luminous, at first appearing a dull red. This is evident from an inspection of Fig. 3. If these light rays be studied by means of a spectroscope which disperses the radiation into its component rays, it will be found that deep red rays are the only visible rays present in appreciable amounts. As the temperature is in- creased the appearance of the body passes from red to orange, then to yellow and so on. If the body were sufficiently refractory to withstand higher tem- peratures and remain in solid form, at a certain tem- perature it would appear white and with increasing temperature would assume a bluish white appearance. 14 COLOR AND ITS APPLICATIONS The latter temperatures have never been reached in the production of artificial light. Notwithstanding the fact that all solids produce a continuous spectrum and obey the general laws mentioned, it does not follow that they all emit the same amounts of light per unit area at equal temperatures. Kirchhoff has shown by the theory of exchanges that the emissive and absorptive powers of all bodies at the same tem- perature for rays of a particular wave-length are pro- portional to each other when the radiation is a pure temperature effect. For a particular kind of radiator called a black body or a full radiator, the relation between emission E, the wave-length, X, and the absolute temperature J, has been deduced theoretic- ally. The black body is defined as a body that will absorb all radiation incident upon it and reflect none. When it radiates it emits in each wave-length more energy than any other body at the same tem- perature. The nearest approach to such an ideal radiator is a hollow space enclosed by emitting walls of uniform temperature provided with a small open- ing through which the radiation can escape. The laws deduced theoretically by various investigators do not agree entirely. The one that best fits exper- imental data is Planck's law given in Ex= CA-'(e^-l)-' (1) Another law known as the Wien-Paschen law found to hold for the short-wave region of the visible spec- trum is shown in equation (2). Ex= Ci\-' e ^T (2) In the foregoing equations Ci and C2 are constants. The values for these differ somewhat as determined by various investigators. LIGHT 15 A simple relation between the wave-length of the maximum of the radiation curve and the absolute temperature is derived from (2) and is expressed in equation (3). KT = constant (3) Another interesting relation known as the Stefan- Boltzmann law connects the total radiation, E, from a unit area of the radiator with the absolute temper- ature, T, and is expressed in equation (4). E = CT' (4) These laws are of chief importance in the theory of radiation but are given here as a matter of reference. A gaseous body is found to emit only certain definite rays and the spectrum is said to be a line spectrum. Sometimes the various lines (which are in reality the images of the slit of the spectroscope, (8) are found to be crowded together in such a manner as to give to the spectrum a fluted appear- ance. Such a spectrum is called a banded or fluted spectrum. A further striking fact is the constancy of the appearance of the spectra emitted by elements in gaseous form. For instance the spectrum of sodium is always recognized by the position of the emitted rays in the spectrum — that is, by their wave-length. The visible spectrum of sodium con- sists of a double line (0.5890m and 0.5896^) and whenever this double line is found in a spectrum it is certain that sodium is present in the radiating substance. This constancy of the spectra of the elements forms the basis of spectrum analysis by means of which traces of elements far too small to be weighed by the most sensitive balance are readily detected. By means of the spectroscope helium was discovered on the sun before it was distinctly isolated 16 COLOR AND ITS APPLICATIONS by scientists on earth. The vacuum tube, the arc, the electric spark, and the flame are used in studying the spectra of elements and compounds. Sometimes both a line and a continuous spectrum are emitted by an illuminant. Such a case is found in the ordinary carbon electric arc. The crater of the arc being an incandescent solid, emits all visible rays while the incandescent gas of the arc between the electrodes emits a line spectrum the appearance of which depends upon the surrounding medium and the composition of the carbons. In Fig. 4 are shown several representative spectra photographed by means of a grating spectrograph on a Cramer spectrum plate which is sensitive in varying degrees to all the vis- ible rays. This particular brand of photographic plate is relatively less sensitive to blue-green rays so that on viewing the spectrograms the energy in this region appears to be less prominent than it really is. It is seen that the two gases, mercury and helium, emit line spectra. The arcs emit both continuous and line spectra, the latter as indicated above being emitted by the vapor in the arc itself. The relative prominence of the line spectra depends upon the relative intensities of the radiation from the arc as compared to that from the solid electrodes. For instance the line spectrum is much more prom- inent in the yellow flame arc than in the ordinary carbon arc and as is well known the arc vapor con- tributes a much greater proportion of the light in the former than in the latter illuminant. The line spec- trum of a carbon arc is subject to momentary changes both in character and intensity due to impurities and also to irregularities in the amounts of the chem- icals with which the carbons are impregnated. The injection of various chemicals into the arc as sug- ^//w- I \/j^}f:,iQ Spectrum a. tiercurcj Arc b. Helium c. Iron Arc d. Yellow Flame Arc e. Carbon Arc f. Carbon Arc ^H g. Carbon Arc h. Tungsten Incandescent Lamp I. SKijIight j, Skij light UltroA VBOYOR Violet I Visible Spectrum Fig. 4. — Representative spectra. 17 18 COLOR AND ITS APPLICATIONS gested by the heating of metallic salts in a Bunsen flame, affords a means of varying the color or spectral character of the light from the carbon arc lamps. Spectra / and g were obtained from the same carbon arc within a period of a few seconds. The tungsten filament is seen to emit a continuous spectrum, /i, the dark band being due to the low sensibility of the photographic plate to blue-green rays. Two spectro- grams of light from the sky are shown in i and j in an effort to bring out the dark lines which cross the spectrum. The solar spectrum is of special interest. As already indicated a photograph of the spectrum of sunlight made with a narrow slit, shows practically a continuous band crossed by many fine dark lines (see Plate I). These lines were discovered by Wollaston in 1802 but were later studied with better instruments by Fraunhofer in 1814 who found several hundreds of them. These dark lines in the position of various colors show the absence of the corre- sponding images of the slit of the instrument and therefore the absence of these rays in sunlight. Their absence is attributed to absorption by vapor chiefly in the solar atmosphere. Luminous gases or vapors, as has already been indicated, emit only a limited number of rays, their spectra being dis- continuous in appearance. These vapors when lumi- nous are usually opaque to the particular rays which they emit and therefore the light from the sun is robbed of some of the rays in passing through its atmosphere. The Fraunhofer lines are often used as reference points in examining spectra, although electric discharges through gases in vacuum tubes and the heating of salts in a gas flame furnish convenient means of identification or reference spectra in the LIGHT 19 experimental laboratory, are given in Table I. The chief Fraunhofer lines TABLE I Principal Fraunhofer Lines Line Wave-length Color Source A 0.7594M Red Oxygen in atmosphere a .7185 <( Water vapor in atmosphere B .6867 (( Oxygen in atmosphere C .6563 (( Hydrogen, sun Di .5896 YeUow Sodium, " D2 .5890 (( <( n E .5270 Green Calciiun, " bi .5184 {( Magnesium, ** b2 .5173 ({ (( (( b4 .5168 11 (( ({ F .4861 Blue Hydrogen, " G .4308 Violet Calcium, " H .3969 (( (( (i K .3934 (t It it Other convenient lines obtained by heating salts in a Bunsen flame are — potassium red, 0.7699 and 0.7665 m; lithium red, 0.6708; sodium yellow, 0.5896 and 0.5890; thallium green, 0.5351; magnesium green, 0.5184 ; strontium blue, 0.4607. The mer- cury arc gives a double yellow line, 0.5790 and 0.5764; a bright green line, 0.5461; a faint blue-green line, 0.4916; an intense blue line, 0.4358; and deep violet lines, 0.4078 and 0.4047. Hydrogen at a fairly low pressure in a glass tube through which an electric discharge is passed yields a red line, 0.6563, blue- green, 0.4861, and blue 0.4341. The helium spec- trum obtained from a tube such as the foregoing is rich in useful lines throughout the spectrum, yielding red lines 0.7282, 0.7065, 0.6678; yellow 0.5876; green lines 0.5048, 0.5016, 0.4922; blue lines 0.4713, 20 COLOR AND ITS APPLICATIONS 0.4472, 0.4388; and violet lines 0.4026, 0.3888. Iron, copper, zinc, and cadmium when used as the ter- minals of an electric spark yield many useful lines, many of which are in the ultra-violet. The iron arc and the quartz mercury arc are particularly useful for exploring the ultra-violet region. Three useful cadmium lines are red 0.6438, green 0.5086, and blue 0.4800. 280 260 240 220 200 180 ^ 160 140 120 < a; 100 60 60 40 20 / / ^1 / 1 1 d/ / ^^ — , N, 1 1 . 1 •'V //-■ 'f / s \ / 1 ■'V ■"f / \ ^ \ s. L^/ ¥ 'y t s \ ^ e^' r '"^ r ^ > ■4 r — ~K —- ^ ^ ^ i 7 > ^ ^ i ?^ ■/ ^ ^ / ^ ^ ^ t'-' ^ "040 0.44 04& 0.52 0.56 060 ^^.WAVEL LENGTH 0.64 0.66 Fig. 6. — Distribution of energy in the visible spectra of various illuminants. In Fig. 5 and Table II are shown the spectral dis- tributions of energy in the visible region for various illuminants. These data have been obtained in the Nela Research Laboratory chiefly by Hyde,^ Ives,^ Cady and the author.^ As is evident at a glance these LIGHT 21 TABLE n Relative Distribution of Energy in the Visible Spectra of Common Uluminants A B C D E F G H I 1 ! 1 j! g 1 it.! flii Sl<i 1 d 0.41m 72. 79. i77. 186. 1.9 3.6 4. 7. 6.6 9.6 16.6 22.5 .43 21.8 .46 84.3 187. 6. 12. 16. 16.7 30. 29. 17.6 .47 91. 180. 10.6 18. 21.9 23.5 38. 37. 26.4 .49 92.6 162. 16.3 25.6 30.3 32.7 47. 45.6 38.3 .61 96. 146. 26.6 34.6 40. 42.6 56.5 55. 51. .53 98. 132. 37.6 47. 62. 54.9 67. 66.5 64. .66 99. 120. 63.2 62. 66.6 68.6 78. 76. 78. .67 100. 108. 74.6 79. 82. 83.4 88. 88. 90. .69 100. 100. 100. 100. 100. 100. 100. 100. 100. .61 100. 93. 130. 123. 118. 117. 111. 113.5 107. .63 98.6 87. 168. 148. 139. 136. 121.5 127. 111. .66 97.1 82. 210. 176. 160. 167. 131. 142. 114. .67 96.6 77. 260. 204. 182. 179. 140. 156. 119. \69 93.6 72.6 320. 234. 205. 202. 147.5 170. 120. curves are plotted in such a manner that the relative energy of wave-length 0.59 m equals 100. This method of plotting gives the relative distribution of energy for approximately the same amounts of total light sensation. Reference to the spectral distribution of energy in the radiation from the two tungsten lamps, operating at 7.9 and 22 lumens per watt respectively, shows the effect of increasing temper- ature upon the relative amounts of rays of shorter wave-length emitted. See Table XXIII. 22 COLOR AND ITS APPLICATIONS REFERENCES CHAPTER I 1. Ges. Abhandlungen zur PhysioL Optik. Leipzig, 1903, p. 144 et seq. 2. Jour. Franklin Inst., 191 o, p. 439. 3. Trans. I.E.S., 1910, p. 189. 4. Elec. World, Sept. 19, 1914; Trans. I.E.S., 1914, p. 839, OTHER REFERENCES Theory of Optics. Preston. Physical Optics. R. W. Wood. Outline of Applied Optics. P. G. Nutting. Lectures on Illuminating Engineering. Johns Hopkins University. CHAPTER II THE PRODUCTION OF COLOR 7. Colors and color phenomena are encountered in nearly every human activity. In fact they are so ever-present that they have become common-place to the average man excepting in those instances where they enter actively into his work. That there is an explanation for every case of color production is inevitable, however, there are some color phe- nomena as yet unexplained satisfactorily. Color is produced in many ways. It is intimately associated with light as has already been seen. When color is produced it is usually the result either of a sub- traction of some of the visible rays from the total radiation emitted by light source or of the dispersion of the visible radiation into its component parts. 8. Refraction. — Sir Isaac Newton in 1666 dis- covered that sunlight consisted of many rays each of which when permitted to impinge separately upon the retina produced the sensation of a distinct color. He accomplished this by prismatic dispersion and proved that no further change resulted from Fig. 6. — Newton's experiment. subsequent dis- persion. Newton's experiment, which is shown dia- grammatically in Fig. 6, was performed approximately in the following manner. A prism was set up in a 23 24 COLOR AND ITS APPLICATIONS darkened room and sunlight admitted through a small hole in the window shade. This provided him with a parallel beam of light. After the beam traversed the prism he found that there was no longer an image of the sun, but a colored band similar to the rainbow, made up in reality of an infinite number of colored images of the slit overlapping each other. On passing a narrow portion of this colored band through a hole in another screen and permitting this to traverse another prism, he found no further change in color, thus prov- ing that monochromatic light can not be further de- composed. The principles involved in this experiment are used today in a great deal of spectroscopic work. Let us examine some of the characteristics of the spectrum somewhat more in detail. It will be found that an increase in the width of the slit produces an increase in the brightness of the spectrum but owing to the facts that the spectrum consists of overlapping images of the slit and that each image of a broad slit is wider than an image of a narrow slit, it is evident that there will be more over- lapping of the images pro- duced with a wide slit. The smaller the amount of over- lapping the purer the spectrum will be. The difference in the appearance of the spectrum of a mercury arc due to a change from a narrow rectan- gular slit to a wide circular opening is shown in Fig. 7; both spectra were photographed on the same grating spectrograph. The material of which a prism is made also has a marked effect on the appearance of the spec- trum. When a ray of light passes through a prism it is V B 6 Y Fig. 7. — Effect of the character of the slit of a spectrograph on the grating spectrum of the mercury arc. THE PRODUCTION OF COLOR 25 of course bent out of its original direction, the amount of bending or the angle of deviation depending upon the angle of the prism and the index of refraction for the particular wave-length of which the ray consists. If for a given prism the angle of deviation remained the same while the wave-length of the light was changed, then the index of refraction would also remain con- stant, and if wave-lengths were plotted against either 1.72 1.70 1.68 L66 g 1.64 Z 1.62 > £ 1.58 UJ ^1.56 1.54 132 \ ■ > \ \ >^ . ^ ^ 2:i^ ^ hi- = V • \ ^sr ^^ L "^ ■ — ^ //'/?/■ ' -- ^ -£f /gy/- 'z '^ ...^ r..- 2HV/ — QM 042 0.74 082 0.50 038 0.66- .yU. WAVE LENOTH Fig. 8. — Dispersion curves of various optical media. angles of deviation or indices of refraction the disper- sion curve would be a straight line. If such a substance actually existed, it would be possible to determine wave-lengths in its spectrum by the use of an ordinary graduated scale. However, such is not the case and further, the dispersion curves of various prisms differ considerably. The dispersion curves of quartz and three kinds of glass are presented in Fig. 8. It is evident from an inspection of these curves that the different wave-lengths are more closely crowded to- gether in the red or long wave-length end of the 26 COLOR AND ITS APPLICATIONS spectrum than in the other. This is shown in Plate I (Frontispiece) where a prismatic spectrum is com- pared with a normal (grating) spectrum in which equal distances represent equal wave-length intervals. For the study of the visible rays transparent glass prisms are satisfactory but for the study of the invis- ible rays other media must be used because glass is not sufficiently transparent for very short or very long waves. Hence investigations of the ultra-violet region are made with optical systems of quartz and those of the infra-red with fluorite or rock salt. Ex- tremely long waves such as are used in wireless telegraphy are studied by means of huge prisms of pitch or paraffin. 9. Diffraction. — Another common device employed in optical instruments for decomposing visible radia- tion into its components is the so-called diffraction grating. A grating is simple in appearance but is difficult to make for it consists of a great many paral- lel lines (sometimes as many as 40,000 and more per inch) scratched usually upon glass or speculum metal. When a grat- ing is placed in the path of a beam of light a spectrum re- sults, which, unlike a prismatic spectrum, has a constant dis- persion and is therefore called Fig. 9.-Young's doubie-sUt ex- ^ normal spectrum. periment illustrating the princi- To explain the actiou of the pie of the diffraction grating. ^jff ^.^ction grating we shall Con- sider only two of the very large number of slits of which the grating consists. In Fig. 9 has been sketched an THE PRODUCTION OF COLOR 27 instantaneous view of a section perpendicular to the two slits. A source of light, S, has sent out spherical monochromatic waves, successive fronts being repre- sented by the full lines. The dotted lines drawn mid- way between the full lines must then represent parts of the disturbance exactly in opposite phase with the wave fronts. If the diagram be considered for a moment as representing a surface of water into which a stone has been dropped at S, then the full lines might represent the crests of the waves and the dotted lines the troughs. One wave front is represented as contain- ing the two slits. Si and S2. It was stated in # 2 that according to Huyghen's principle any point on a wave front may be regarded as a center of disturbance. The wave front originally proceeding from S strikes the screen and cannot pass through excepting for the two small parts of this wave front that strikes the openings Si and S2. These set up the two sys- tems of spherical waves represented on the right hand side of the screen. It has also been stated (# 2) that, whenever two or more systems of waves travel in the same medium at the same time, interference results. Wherever two wave fronts intersect (two crests in the water analogy) there will be an especially large disturbance; wherever two of the dotted lines intersect there will be an especially large disturbance (an especially deep trough in the water analogy). Where a wave front meets a part of the disturbance in opposite phase, that is wherever a full line inter- sects a dotted line, there will be destructive inter- ference and consequently no motion. The heavy full lines have been drawn through those points where there is maximum motion and the heavy dotted lines through those points where there is no motion and therefore no light. Now suppose these 28 COLOR AND ITS APPLICATIONS waves be permitted to impinge upon a screen repre- sented by the right hand edge of the diagram. There will evidently be light on the screen wherever a heavy dotted line terminates. Imagine that the source S has been sending out blue light and is now exchanged for one that emits red light. A very similar diagram will result but on account of the longer wave-length, the ends of the heavy lines will no longer intersect the screen at the points Oi and O2; there will still be light at Oo as is evident from the symmetry of the diagram, but the new points corresponding to Oi and O2 will lie some- what farther from Oo than in the case of blue light. Finally, suppose the source S to emit white light (light of all visible wave-lengths). Then there will be a white spot at Oo and spectra at Oi and O2, and the blue edge of these spectra will be nearest Oo. In other words, in a grating spectrum the long wave- lengths are deviated more than the short ones, which is opposite to the condition that exists in prism spectra. Further, it will be seen that the distances in the spectra, Oi, O2, etc., are proportional to the wave-length; in other words, that a grating spectrum is one of constant dispersion. (See Plate I.) The spectra, Oi, are said to be of the first ordeVj O2 of the second order, and so on. It is also evident that if more heavy lines had been drawn in the dia- gram the positions of the third and higher order spectra would also have been indicated. It will be further noted that the higher the order the greater is the length of the spectrum, and therefore the greater is the dispersion. The colors of some crystals such as the fiery opal, and of insects and feathers are often accounted for by the phenomenon of diffraction. On looking at THE PRODUCTION OF COLOR 29 an arc lamp through a screen door or the meshes of an umbrella top, diffraction spectra are seen. Like- wise on viewing a light source over a thin edge of an opaque object held close to the eye, a colored fringe is seen. These bands are due to the inter- ference of light waves. Since the wave-length of red light is greater than that of violet light, the red light penetrates further into the shadow than the violet light. The first gratings were made by Fraunhofer in 1821, and consisted either of fine wire or of fine rulings on a smoked glass. At present gratings are ruled by means of a diamond on glass and on the bright reflecting surface of speculum metal. Row- land at Johns Hopkins University contributed much toward the production of satisfactory gratings. On his machine as high as 110,000 lines per inch can be ruled, but usually the number does not exceed 15,000 or 20,000. Cheap copies of gratings are obtained by flowing a film of celluloid dissolved in amyl acetate over a grating, afterward stripping this off and mount- ing it between plate glasses. It is evident that in the case of rulings upon an opaque substance such as speculum metal the spectra must be produced by reflection and not by transmission as discussed above. The individual rulings always act as absorbing bodies while the unruled portions either reflect or transmit this light. 10. Interference. — Other color phenomena be- sides that of the diffraction-grating spectrum arise from the process of interference. Thin films of trans- parent substances such as oil on water, soap bubbles, thin sheets of mica, and iridescent crystals, owe their color usually to the interference of light waves. If a slightly convex glass surface be placed upon a 30 COLOR AND ITS APPLICATIONS plane piece of glass, colored bands known as Newton^s rings will be seen. These are due to interference between the light waves reflected from the upper and lower surfaces respectively of the thin film of air of varying thickness between the two pieces of glass. When white light is incident normally at the point of contact the colored bands are circular and concentric with the point of contact. These bands are violet on the inside and red on the outside but at a short distance from the center they begin to overlap and gradually disappear. If monochromatic light is used the bands appear of the color of the light and many bands can be seen. In the case of some crystals such as chlorate of potash and fiery opals, the colors are found to be very pure. The colors of insects and feathers are often accounted for by the phenomena of interference. The process of color photography devised by Lippmann (# 57) is based upon the principle of interference of light waves. 11. Polarization. — Imagine for the sake of sim- plicity a beam of sunlight emerging from an aperture in a window shade. All the different waves that make up this beam are traveling in the same gen- eral direction. As mentioned previously, light waves are transverse; that is, the particles of the medium that transmit the waves move to and fro in a straight line perpendicular to the direction in which the beam is traveling. In the simplest case all of the particles that transmit a wave move so as always to be in one plane. For example in a. Fig. 10, the particles Ay By C, Dy etc., move to and fro along perpendicular paths but the wave travels horizontally. If such a wave could be seen from one end it would appear simply as a short straight vertical line as indicated in b. Sup- pose that the beam of sunlight emerging through THE PRODUCTION OF COLOR 31 the window shade could be examined minutely end on. We would then see something similar to c for the different waves vibrate in all possible different planes. Suppose further that by some means the beam could be transformed so that when seen end on, it would Fig. 10. — Diagrammatic illustration of polarized light. appear like d. Such a beam consisting of waves in parallel planes is said to be a plane-polarized beam, and curiously, the nomenclature adopted states that such a beam in which the waves all move in vertical planes, is polarized in the horizontal plane. It is understood that these graphical diagrams are used for the sake of presenting the subject pictorially and with no claim that they represent actual condi- tions. Nevertheless if the actual conditions were such the results of experiments would readily be explained by the reasoning used here. One simple way of * polarizing' a beam of light is by permitting it to fall upon a plate of glass so that the angle of incidence is 57 degrees as shown in e. The unpolar- ized beam AB will be separated into refracted (BC) 32 COLOR AND ITS APPLICATIONS and the reflected {BD) beams, and the former will be found to consist mainly of waves vibrating in the plane of the paper, as indicated by the cross lines representing the path of the particles, while the reflected beam will be found to consist of waves vibrating perpendicular to the plane of the paper as indicated by the dots representing the paths seen end on. To show that the beam BD really has different properties than 5 C it is only necessary to attempt to reflect each from another piece of glass. If the second piece of glass be placed at C in a position similar to the first, the beam BC will pass through but if it be rotated about C through an angle of 90 degrees keeping its angle with the beam con- stant it will be found that the beam BC will not be transmitted. The treatment for the beam BD is obvious. Another means of polarizing a beam of light is by permitting it to pass through certain crystals, such as tourmaline, Iceland spar and quartz. If this be done it will be found that the incident beam is divided into two beams polarized at right angles to each other, the two having different directions in the crystal, different velocities and different prop- erties in general. One of these beams will always be found to obey the ordinary laws of refraction, for example, that the incident ray, the normal to the surface, and the refracted ray all lie in one plane; the other beam will not obey the foregoing and other simple laws generally. The former is therefore called the ordinary ray and the latter the extraor- dinary ray. Nicol devised a very convenient method of separating the two beams when produced by a crystal of Iceland spar as illustrated in Fig. 11. If a rhomb of spar be cut into two parts along the plane THE PRODUCTION OF COLOR 33 indicated by the diagonal, PP, and if the parts be polished and cemented together with Canada balsam, the paths of the two beams will be as indicated. The Canada balsam has a refractive in- dex intermediate between that of the spar for the ordinary and extraordinary rays. Therefore the ordinary ray, O, being inci- dent upon the balsam layer at an angle greater than the critical angle, is totally reflected while the extraordinary ray is merely slightly refracted by the layer of balsam. That the beam emerging from the Nicol prism is really plane-polarized can be shown by a mirror, as before, or more simply, by a second Nicol prism. If the second prism be rotated through one com- plete revolution, two positions will be found for which the light transmitted by both prisms is practically of the same intensity as that transmitted by one, and two other positions will be found for which no light is transmitted. For intermediate positions the intensity of the light will be less than that transmit- ted by one Nicol. When the maximum amount of light is transmitted the Nicols are said to be * parallel' and when no light is transmitted they are * crossed.' Some substances, such as quartz (cut in a certain manner), sulphate of lime, sugar solution, and tur- pentine, behave in a peculiar manner when placed between crossed Nicols, for though no light passes the second prism before the introduction of the other substance, some light is transmitted when the sub- stance is inserted in the path. If monochromatic light is used the light can be extinguished by rotating either prism to the right or left by an amount depend- Fig. 11. — The Nicol prism for obtaining plane-polar- ized light. 34 COLOR AND ITS APPLICATIONS ing upon the substance, its state of concentration if in solution, and the thickness of the layer intro- duced. Such layers are said to rotate the plane of polarization. If various monochromatic lights of dif- ferent colors are used in succession it will be found that the prism must be rotated different amounts for the different colors in order to bring about a total extinction of light after the introduction of one of the substances. Hence if such a substance be ex- amined in white light, a certain position of the prism will extinguish the blue light but permitting the re- maining colored rays to pass in varying proportions; at another position yellow will be extinguished and the remaining rays permitted to pass in varying pro- portions, and so on. The transmitted light will of course have a different color in each case. As quartz is one of the chief substances having this property, the following table is given to show the magnitude of the rotation in degrees produced by two thicknesses of quartz for light of different wave- lengths. Thickness of quartz Red Orange Yellow Green Blue Violet 1 mm. 18° 136 22° 161 24° 180 30° 218 32° 232 42° 316 Another instance of the production of color by polarization is that of the examination between crossed Nicols of certain other crystals that trans- form plane-polarized light into so-called circularly or elliptically-polarized light. Many minerals occuring in Nature have either this property or that of rotating the plane of polarization, and an entire system has been developed for determining what substances are THE PRODUCTION OF COLOR 35 present in a given rock by noting the color effects produced by a thin section of the rock in a micro- scope equipped with two Nicol prisms. 12. Reflection^ Absorption^ Transmission. — Ordi- nary colors, which are encountered, are produced by selective reflection or transmission. Since most sub- stances do not have the property of reflecting the same proportions of all light rays received they are said to be selective in their reflection. A red fabric has the ability to reflect chiefly the red rays of the visible spectrum; therefore when white light falls upon it only the red rays are reflected while the remaining visible rays are absorbed. By this process of selective reflection the colors of pigments and most of the colors in Nature are produced. The same remarks apply to the production of color by selective transmission. Colors produced by these means are not as pure as the colors of the spectrum. Each of the latter consists practically of a single wave- length and are said to be monochromatic, while the colors ordinarily encountered in Nature are im- pure, consisting of rays of a considerable range of wave-lengths. In Fig. 12 are shown diagrammatically the spectral analyses of five common pigments. For each pig- ment the reflecting power (i.e. the per cent of energy reflected) was determined for all wave-lengths in the visible spectrum. The results plotted against the corresponding wave-lengths are represented by the full lines in the diagram. The dotted curves represent roughly the relative light values and are obtained from the full curves by multiplying the energy values represented by the ordinates by their relative abilities to produce light sensation (#4). A similar discussion applies to the same colors pro- 36 COLOR AND ITS APPLICATIONS duced by transmission. These analyses will be dis- cussed further in Chapter V. (See Figs. 122 and 123.) The character of the surface of a pigment and the density Of the coloring matter influence the appear- ance of a color. If, for instance, a red aniline dye solution be deposited upon a fabric by means of an air brush, the pigment under some conditions will Purple Blu£ Oreeh Yellow Red 040 050 0.60 JJi, WAVELEMGTH Fig. 12. — Analyses of ordinary colors. be deposited in the form of a fluffy powder whose appearance is a deep velvety red. The purity of the red is largely due to multiple reflections, for the light is able to penetrate an appreciable distance into the medium (#64, 75), and at each reflection it becomes purer — that is, more nearly monochromatic. If the solution were applied by means of an ordinary brush, the deposit would not have been so porous and the appearance of the color would not have been such a deep red. Another instance of the effect of mul- tiple reflections on the color of light is found in a gold-lined goblet. All are familiar with the color of gold plating. This, however, becomes much more THE PRODUCTION OF COLOR 37 reddish when inside a goblet, owing to the purifying effect of multiple reflections. The color of transparent media depends upon the depth of the coloring matter. For instance, a hollow glass wedge, when filled with an aqueous solution of ethyl or methyl violet, will appear bluish at the thin end and reddish at the thick end. Cyanine is an- other dye that exhibits dichroism, which is the name applied to the foregoing. The different appearances of silk and woolen fabric dyed in the same solution are due largely to the difference in the character of the surfaces. Many of these instances could be cited here, but the details will be treated in succeed- ing chapters. 13. Color Due to Scattered Light, — That light is changed in color by being scattered by fine par- ticles is a fact observed daily. Tyndall, by precipi- tating clouds of vapor, observed that as the particles increased in size the bluish color of the clouds dis- appeared. Smoke from the end of a cigar appears bluish, yet the smoke exhaled appears whitish; this latter is perhaps caused by an increase in the size of the particles, due to the condensation of moisture. Rayleigh ^ has mathematically treated the problem of scattered light and experimented with a sulphur precipitate. He noted a polarizing effect which is of interest. J. J. Thomson ^ treated the scattering due to small metallic spheres theoretically. Garnett^ con- cluded that colored glasses owe their colors to the presence of microscopic spheres of the metal of the coloring agent. Colloidal solutions appear to act in the same manner. Mie ^ has extended Garnett's theory very considerably; however, the exact explana- tion of the color of glasses is still somewhat disputed. The color of daylight is of special interest because 38 COLOR AND ITS APPLICATIONS it is our most common illuminant. Light from the sky consists chiefly of scattered sunlight in daytime. If it were not for the finely divided particles of matter in our atmosphere, the sky would be quite as dark by day as by night. When the particles are of a size comparable with the wave-length of light rays, they produce considerable scattering of the latter. This scattering of light is selective, the rays of short 100 Q50 ceo M, WAVE LENGTH Fig. 13. — Showing the variation in the spectral character of sunlight due to atmospheric absorption. wave-length being scattered in greater amounts than those of longer wave-length; this accounts for the bluish color of the sky (Fig. 5). Sunlight is termed white light, but in many cases throughout this book the term * white' light is used to indicate the total light from an illuminant emitting all visible rays. Where necessary, the two meanings are differentiated. Direct sunlight, however, undergoes a change in color as the altitude of the sun changes, on account of the greater thickness of air, more or less laden with smoke, dust, vapor, and ice, through which the light must pass at the lower altitudes. This change in color caused by the combined actions of THE PRODUCTION OF COLOR 39 absorption and scattering, is such as to depress the blue, and hence the sun appears redder as it ap- proaches the horizon. All the brilliant colors of sunset are due to the foregoing phenomena. The effect of different masses of air upon the relative amounts of energy in each wave-length which reach a given point on the earth's surface has been deter- mined by Abney; his data are reproduced in Fig. 13. Of course the results obtained will depend upon the purity of the atmosphere. Over a smoky industrial city, the sun, when near the horizon, almost daily appears a fiery red. Often the absorption is so great that the sun disappears from view long before it actually sinks below the horizon. 14. Color Sensations Produced by Colorless Stimuli. — Colors are often visible to a careful observer when not produced by any of the methods already mentioned. If a disk composed of black and white be rotated at the proper rate — moderately slow — colors appear upon the leading and lagging edges of the sectors. In other words when the black and white stimuli precede or follow each other at certain intervals, colors are produced instead of gray. Fechner, in 1838, was perhaps the first to describe these subjective colors, and his name was later applied to the phenomenon by Brticke. Many have stud- ied the problem and there is a general agreement as to the results obtained, though there is no such agreement Fig. 14. — Benham disk for producing^ subjective colors by means of black and white stimuli. 40 COLOR AND ITS APPLICATIONS as to their explanation. Benham,^ in 1894, produced a disk different from those used by preceding investi- gators and made an attempt to solve the problem. One form of his disk, illustrated in Fig. 14, shows the colors in a striking manner when rotated. The phenomena can only be briefly discussed here. In general, when black is followed by white at a mod- erate speed, a sensation of red results, but if white be followed by black, a sensation of blue is experi- enced. By introducing various angular intervals, as is done in the Benham disk, sensations of inter- mediate colors are aroused. On rotating the disk in one direction the blue sensation is aroused in the inner ring and red in the outer; on reversing the disk, the colors aroused are also reversed in their order. The phenomena are interesting and have received a great deal of attention, though, as already stated, there is no general agreement as to their complete explanation. No doubt retinal inertia and the difference in the rates of growth and decay of the color sensations are important factors in the production of these so-called subjective colors. Often, when suddenly moving the eye over black and white surfaces in the field of vision, these colored effects are perceptible. A simple disk for showing the Fechner colors, though not as effective as the Benham disk, is one containing plain black and white sectors. On rotating such a disk at a certain speed it will appear of a greenish hue, but at a somewhat more rapid rate of rotation it appears a reddish hue. Helm- holtz used a white disk upon which was painted a black spiral. Rood ^ used an opaque disk with four open sectors, each of seven degrees. Through this rotating disk he viewed a clouded sky. With a rate of nine revolutions per second the sky appeared a THE PRODUCTION OF COLOR 41 deep crimson hue, except for a small spot in the center of the visual field, which remained constantly yellow. This latter is probably due to retinal dif- ferences. The center of the retina, called the * yellow spot,' is known to exhibit selective absorption. At eleven and one-half revolutions per second the field appeared bluish-green. 15. Fluorescence and Phosphorescence. — Usually the radiant energy absorbed by bodies is trans- formed into heat energy. There are many sub- stances, however, — some solid, some liquid, and some gaseous, — that have the property of absorbing radiant energy of certain wave-lengths and of emitting it again after transforming it into radiant energy of other wave-lengths (nearly always longer than those absorbed). The name fluorescence was derived from fiuor spar (calcium fluoride), which has long been known to possess the property of emitting light rays of a dif- ferent color than that of the rays with which it is illuminated. Strictly speaking, the term applies to the phenomenon only when it ceases immediately after the exciting light is extinguished; in this sense it is applicable to liquids and gases only. In solids the phenomenon usually continues for some time after the exciting light has been shut off. This prolonged emission, which in some cases lasts for hours, is termed phosphorescence. The phenomena of fluorescence and phosphorescence are sometimes classified under the more general term luminescence. Though there are comparatively few substances that exhibit the phenomenon to a marked degree, it is difficult to find materials that do not show the prop- erty slightly. This is readily seen by examining ordinary substances in an intense spectrum of the 42 COLOR AND ITS APPLICATIONS light from a quartz mercury arc produced by means of a quartz optical system. In examining the fluor- escent light it is often convenient to look through a glass of such a color that it will transmit the fluor- escent light rays and absorb the exciting rays. The phenomenon is influenced by temperature, and in most cases the phosphorescence is temporarily in- creased in brightness by the application of heat or red and infrared rays, but the duration of this in- creased brightness is usually brief, as the phos- phorescence is rapidly extinguished by these agencies. The phenomenon is of great interest to the scientist and also in smaller degree to the colorist and light- producer. Fluorescent phenomena play a part in the appearance of certain colors (# 75), and there are possibilities for utilizing the phenomenon for the production of light for practical purposes. Some examples of fluorescence and phosphores- cence should be of interest. Sunlight and the light from carbon, mercury, zinc, iron, and silicon arcs are rich in ultra-violet rays which ordinarily are the most active in exciting fluorescence and phosphores- cence. In these cases it is convenient to remove most of the yellow, orange, red and infrared rays from the exciting beam by means of a dense violet glass and a water cell. Uviol blue glass is especially satis- factory. Water of a few centimeters depth is practi- cally opaque to infrared rays, but when pure is quite transparent to ultra-violet rays. The fluorescence, consisting in general of rays of longer wave-length than the exciting light, can be viewed through a glass of proper color without being confused by the color of the exciting light. Fluorescent materials are valu- able in visually investigating the ultra-violet region of a spectrum; for example, uranium glass is very THE PRODUCTION OF COLOR 43 convenient for focussing a spectrograph for the in- visible ultra-violet rays. Aesculine, which is ordinarily transparent in solu- tion or in a gelatine film, fluoresces a bluish color under strong ultra-violet excitation. It is valuable in photography as a screen for absorbing ultra-violet rays. A solution of fluorescein or uranin is useful in demon- strating the path of light through various optical systems. It is best to prepare first a solution of moderate concentration and add this drop by drop to the tank of clear water. Of course the difference between the refractive index of the water and that of air must be taken into consideration when demon- strating the path of light through a given optical system immersed in water. Kerosene, an alcoholic solution of chlorophyl, anthracene, and many of the organic dyes exhibit the phenomenon of fluorescence. In cases of strong excitation the emission of light rays by some of these substances continues for some time after the exciting light is cut off. Substances that exhibit prolonged phosphorescence are chiefly the alkaline earth sulphides. Balmain's paint, an impure sulphide of calcium, is one of the most active and least expensive. Its phosphorescent light is of a bluish color. Other phosphorescent sulphides emit light of various colors, and by com- bining various ones, nearly any desirable color can be obtained. For demonstration purposes beautiful designs can be made by the use of various phos- phorescent media which emit light of different colors. The chief difficulties which limit the use of phos- phorescent substances are the scarcity of the exciting rays in ordinary light sources and the rapid decay of the intensity of the emitted light after the excitation has been cut* off. 44 COLOR AND ITS APPLICATIONS One instance of a commercial application of fluor- escence as a light source is the adaptation of a rhoda- mine reflector to the mercury vapor arc lamps by Peter Cooper Hewitt. The reflector consists of a white paper base upon which the rhodamine layer is placed. The latter is apparently protected by a trans- parent varnish. On focussing the spectrum from the quartz mercury arc upon it and viewing it through a red glass, it is seen that practically all rays from green to the extreme ultra-violet excite the red fluor- X ?./^ F \ 1 / / / , "n I \ 1 / / \ \ \ ^^ ^ / S|^ 40 0.45 0.50 0.55 0.60 0.65 0.7C ^. V^AWE LENGTH Fig.' 15. — Diagrammatic illustration of the action of the rhodamine fluorescent reflector. escence. That is, a photograph through a red filter of this projected spectrum appears quite the same as an actual spectrogram of the light from the quartz mercury arc. The action of the reflector is dia- grammatically shown in Fig. 15. The heavy vertical lines represent the visible lines of the mercury spec- trum, their lengths being approximately proportional to their energy intensities; curve R represents the reflection curve of the rhodamine dye, and F the fluorescent light. It is seen that a gap in the spec- trum of the light from a mercury arc equipped with this reflector exists in the blue-green region. While this reflector greatly improves the appearance of colored objects illuminated by the mercury arc, the light is still unsatisfactory for accurate color work, owing to the gap mentioned and to the sttong emission THE PRODUCTION OF COLOR 46 lines. In Fig. 16 is given the spectrophotographic analysis, by means of a prism instrument, of the prop- erties of the rhodamine reflector. The slit of the spectrograph was purposely ad- justed somewhat wider than usual, on account of the long exposure (several hours) required to obtain a spectrogram of the fluorescent light, a represents the reflection of the reflector for tungsten light; b, the spec- trum of the tungsten light; [v b gyor c, the mercury spectrum; d, the uifravioieA v/s/bje reflection of the rhodamine re- ^''^i,^'^^^^^:^ flector illuminated by the total rhodamine fluorescent re- light from the mercury arc ; e, the mercury spectrum (shorter exposure); /, the isolated mercury green line produced by a special filter de- scribed later; ^, the fluorescence spectrum excited by the green line, the latter also appearing owing to diffuse reflection from the fluorescent reflector. This re- flector furnishes red rays to the mercury vapor lamp, when so equipped, at the expense of practically all the other rays. The scheme is an ingenious one, and probably paves the way for other practical uses of the phenomenon of fluorescence. In the study and practical use of fluorescent substances the solvent is of importance, owing to the influence upon the intensity of the fluorescence. Knoblauch ^ investigated the subject, obtaining the results given in Table III. The figures ranging from one to eleven indicate the order of the intensity, eleven being the most intense. The table is also of interest in suggesting a variety of solvents for dyes when these may be unknown to the experimenter. 46 COLOR AND ITS APPLICATIONS TABLE m Effect of Solvent upon the Intensity of Fluorescence -o "o 1 i 1 •J3 g •a -S 1 s S S 6 5 2 n *>, >% % 1^ ii S 3 >> O 3 3 3 4 5 1 3 3 1 3 6 5 1 2 4 4 n §• ■§ & Aesculine Anthracene 4 91 6 1 5 1 5 B. Phenylnaphthylamin 5 1 1 « 3 Chrysoltn 91 3 3 3 3 1 Eosine (sodium) 1 9, fi n 4 3 Fluorescein (lithiutn) , , 9, 3 n 4 1 Fluorescene 1 2 6 9 4 4 11 9 3 10 3 PetroleuTTi 3 6 4 6 3 1 B Phenosafranine 1 6 7 4 9, Magdala red 1 Curcumin 1 2 1 3 4 Phenanthrene 2 16. Useful Filters. — Very often in the study of color phenomena monochromatic light is desired, or ultra-violet and infrared regions of the spectrum must be isolated. For these reasons various filters which have been found convenient will be described. For obtaining intense monochromatic light the quartz mercury arc is a valuable source. Filters can be prepared for isolating the various lines. The filters can be made of dyed gelatine and cemented between glass plates (or quartz, if necessary) with Canada balsam, or the dyes can be dissolved in a proper solvent and used in glass or quartz cells. The filters can be tested visually by means of a spectro- scope or photographically with a panchromatic plate and a spectrograph. THE PRODUCTION OF COLOR 47 For isolating the mercury yellow lines, 0.5790 and 0.5764/1, chrysoidine and eosine are satisfactory. For isolating the green mercury line, 0.5461, neodymium ammonium nitrate and either potassium bichromate or eosine form an excellent combination. The former absorbs the yellow lines, and the latter the blue lines. Neptune green S and chrysoidine have been recommended for the purpose, but the author has found the former method more satisfactory. A cell of water will cut off the infrared rays satisfactorily for most cases when this is necessary. For isolating the blue line, 0.4359, cobalt blue glass and aesculine or sulphate of quinine form a satisfactory combination. The lines 0.4047 and 0.4078 At can be practically isolated by a combination of methyl violet and sulphate of quinine in separate solutions. The line 0.3984 is transmitted to a slight extent. The ultra-violet line, 0.3650, can be practically isolated by methyl violet 4/? and nitrosodimenthyl aniline, methyl violet and acid green, or resorcine blue and aniline green. By the use of other line spectra and ruby glass or red dyes, monochromatic red light can be readily obtained. R. W. Wood has used a combination of strong cobalt-blue glass and a strong yellow, such as a satu- rated solution of bichromate of potash, for isolating the infrared rays beyond 0.69 /x. In photographs of a landscape through this filter the sky appears com- paratively black and the foliage white. Only the near infrared rays and no visible rays are transmitted by a solution of iodine in carbon bisulphide in a cell whose sides are composed of dense red glass. Wood has used lenses coated with 48 COLOR AND ITS APPLICATIONS a thin film of chemically deposited silver for ultra- violet photography. Such a film is opaque to all rays excepting a narrow region in the vicinity of 0.32//. The formulae for silvering are readily found in any recipe book. Many of the organic dyes and colored glasses are useful as filters, depending upon the requirements. It is possible by a careful choice of filters and light sources to isolate any region of the spectrum desired. Photographic plates, owing to their diversity in spectral sensitivities, are valuable assets in some problems. Filters are often much more satisfactory in providing monochromatic illumination than the spectroscope, owing to the much greater intensities of radiation obtainable. An experiment which is very useful and educa- tive is the comparison of two yellows of different spectral compositions. A solution of potassium bi- chromate in water transmits yellow rays. If to a por- tion of this solution an aqueous solution of neodymium ammo- nium nitrate be added, the spec- tral yellow rays are no longer transmitted. The two solutions (c and dy Fig. 17) appear yellow in daylight. If not exactly of the same color, they can be readily brought to the same appearance by the use of more or less of the potassium bichromate in ^^^nimts'^of !hl s^a^l one of the solutions or by the hue but differing in spec- - -.^. - r J.1 11 tral character. addition of one of the yellow or orange dyes. It has been said that color depends upon the wave-length of light. However, color can not always be taken as an indication of wave-length. THE PRODUCTION OF COLOR 49 because, as in this case, the two solutions appear of the same color — yellow; yet when examined by means of the spectroscope one (d) is found to trans- mit green, yellow, and red rays, while the other (c) transmits no yellow rays — only the green and the red rays. The latter is said to produce a subjective yel- low. The transmissions of the two solutions are shown in Fig. 17 compared with the spectrum of the mer- cury arc (g). It is seen that the absorption band of solution c falls in the same region as the yellow mer- cury lines, 0.5790m and 0.5764 /x, so that these yellow lines will not be transmitted by it. Therefore, since the yellow solutions are not transparent to the rays of shorter wave-length, the solution containing the neo- dymium ammonium nitrate (c) will, when illumi- nated by the mercury arc, only transmit the green line, 0.5461, as shown in /. On viewing a mercury arc through each of the two solutions this is readily verified; one solution then appears a brilliant green, while the other remains yellow in color. The color of the green mercury line can readily be matched by a combination of colored glasses and dyes. The transmission of this combination filter is shown in e. The two yellow solutions c and d were made to match the yellow sodium lines, 0.5890 and 0.5896, which are shown (unresolved) in b. These lights, of the same color but of different spectral character, obtained with these solutions and properly chosen illuminants, were used in various interesting experi- ments to be discussed later (#37). An exhaustive spectrophotographic treatment of the transmission spectra of filters is outside the scope of this treatise, but a number of spectrograms of useful filters and ordinary glasses are given in Figs. 18 and 19. Uhler and Wood^ have prepared a ^ U/trav/0/et %^ ^ Visible a. Tungsten Incandescent Lamp b. Carbon Arc c. Iron Arc d. Quartz Mercury Arc Green Ola ss (dense) Signal Blue (medium) Oreen(medium) Canary Yellow (light) Orange Gelatine (light) Green '* (light) Blue " (medium) I. Purple " (mediutm) m. Cobalt Blue Glass n. Clear Glass o. Iron Arc (bare) e. f. 9- h I. J- Ultraviolet %:^ I Fig. 18. p. Signal Green (medium) q. Celluloid r. Cobalt ^lue Glass s. Clear Glass t. Canary Yellow (light) u. Quartz Mercury Arc V. Iron Arc visible Ultra-violet spectra, 50 Ultraviolet J, V/6/b/e 'IIIIIIIIIMNt iiiiiinMiRi IlllllilffrtI 1 r a. Quartz Mercury Arc b. Aesculine c. Acid Oreen -t- Ethyl VioJet d^ " e. Methyl Violet-hHitrosodimenthyl aniline f. Acid Green g. Aniline Oreen h. D/dtilled Wafer L Al rubidium J. Xylene Red k. Rosazeine I. Aniline Red m.Acid Green /?. Aniline Oreen 0, Quartz Mercury Arc p. Aniline Yellow q. Fluorescein r, Tartrazine vS. Uranin t. Orange G u. Aurantia V. Quartz Mercury Arc- Ultraviolet %^Visible Fig. 19. — Ultra-violet spectra. 51 52 COLOR AND ITS APPLICATIONS valuable atlas of absorption spectra, and C. E. K. Mees ^ has prepared a similar treatise dealing with organic dyes. The spectrograms in Fig. 18 and those from / to V in Fig. 19 were photographed on Cramer spectrum plates, which are sensitive, but in varying degrees, to all visible and ultra-violet rays transmitted by quartz. Spectrograms a to h (Fig. 19) inclusive were made on an ordinary plate not appreciably sen- sitive to rays of longer wave-length than 0.48 /x. In Fig. 18, b, c, d, show the spectra of sources rich in ultra-violet rays. The next group, e to o inclusive, shows the transmission of common glasses for the radiation from an iron arc. The last group shows the transmission of some of the same glasses for the radiation from the quartz mercury arc. In Fig. 19 are shown the transmissions of various special screens. These screens are all cemented between two polished glass plates, each one-eighth inch in thickness. The glass is transparent to rays as short as 0.350^, but begins to absorb at this point, becoming practically opaque to ultra-violet rays shorter than 0.300 /x. Some of the first spectrograms illustrate the ability of specially prepared filters to isolate a narrow region of the ultra-violet spectrum at 0.365/>t. For instance, acid-green transmits this ultra-violet line, but also transmits much visible light not shown in the spectrogram on account of the scarcity of the transmitted visible rays in the spectrum of the mercury arc. However, by combining with this screen a visual complimentary dye, such as ethyl violet (a purple), which likewise transmits line 0.365^ but practically none of the visible rays transmitted by the acid-green, a visually opaque screen is produced which transmits rays near 0.365 m quite readily. In much the same manner combination screens are THE PRODUCTION OF COLOR 53 devised for isolating any region of the spectrum. There is no limit to the number of screens that may be combined. The author has at times found it necessary to use as many as five dyes in combination to obtain the desired results. In Fig. 19, p to u inclusive show the rays in the quartz mercury arc radiation transmitted by six different yellow screens between glass plates. These screens appear of about the same color and transparency in daylight. AH are seen to transmit the yellow and green mercury lines, but three of them also transmit ultra-violet rays. Data regarding other media and the relative exposures and transparencies to tungsten light have been pre- sented elsewhere. 1^ The original negatives are. of course more satisfactory than the prints, because some of the fine detail is unavoidably lost in reproduction. These few specimen spectrograms have been inserted, not only on account of the interest in these special cases, but also as a means of giving some idea of the various details to be considered in the examina- tion and production of special screens to those who may not be famiHar with the procedjire. REFERENCES 1. Phil. Mag. 12, p. 81. 2. Recent Researches, p. 47. 3. Trans. Roy. Soc. A, 203, p. 385. 4. Ann. d. Phys. IV, 1908, 25, p. 377. 5. Nature, Nov. 29, 1894, p. 113. 6. Color, p. 194. 7. Ann. d. Phys. 1895, 54, p. 193. 8. Carnegie Inst, of Washington. 9. Atlas of Absorption Spectra. 10. Trans. I. E. S. 9, p. 472. CHAPTER III COLOR-MIXTURE 17. That there is a tremendous variety of colors present in Nature can hardly escape the most in- different observer. A glance at a modern painting reveals the same abundance of tints and shades of color created by the hand of the artist from a few well-chosen fundamental colors. The artist mixes colors in a qualitative manner. He sometimes begins painting with some knowledge of the science of color-mixture, but after all his knowledge of mixing colors is largely qualitative and based upon asso- ciation with his stock of pigments rather than upon a knowledge of quantitative mixture of spectral colors. His success lies largely in a thorough acquaintance with the tools at his disposal, which are his pigments, yet an acquaintance with the science of color is of incalculable value to him, for the experimental results of the scientific study of color-mixture have largely formed the foundation of pure and applied art as well as of modern color theories. \JL8. Subtractive Method. — There are two dis- tinct methods of mixing colors; by addition and by subtraction of light rays. In a sense, color, as we ordinarily encounter it, is produced primarily by sub- traction (#12). That is, a fabric appears colored as a rule because the chemical used in staining it has the property of absorbing certain visible rays and of reflecting (or transmitting) the remaining rays. This subtraction of colored rays from white light 64 COLOR-MIXTURE 55 results in the residual colored light. The integral color of the light absorbed is said to be comple- mentary to the color of the light remaining if the total light in the beginning were white light, say noon •sunlight. Of course in the foregoing case the ab- sorbed color has disappeared, so there is no oppor- tunity to view the complementaries. Any pair of complementary colors can be readily viewed by a comparatively simple apparatus. By means of a prism the spectrum of sunlight is produced at some point in space. A portion of this spectrum can be deflected from the original path by means of a prism of slight angle. The rays in each beam can be com- bined upon adjacent spots of a white surface by means of lenses, with the result that instead of a spot of white light two adjacent spots of colored light are seen. These two colored lights are obviously complementary, for if they are made to overlap they will be found to produce, by addition, a white light. By separating various portions of the spectrum all the pairs of complementary colors are readily pre- sented to view. As will be shown later, white light can be matched by mixing certain pairs of (and also by mixing three or more) spectral colors. This can readily be demonstrated by means of variable slits cut in a cardboard screen and held in front of the spectrum. If the slits have been placed in their proper position in the spectrum and properly adjusted in width, white light will result when the rays from these slits are combined on a white screen by means of a lens. \_The subtractive primary colors have been termed red, yellow, and blue. In reality they would be more exactly described as purple, yellow, and blue-green. They are the complementaries of the additive pri- 56 COLOR AND ITS APPLICATIONS maries, as will be seen later. Some may prefer to use the term *pink' or * magenta' instead of * purple', but the hue is a purple consisting of red and blue. The tri-color processes of printing and color photography are based upon the subtractive principle of mixing colors. The principle of the subtractive method is well demonstrated by Fig. 20 (Plate II). If the three subtractive primaries, purple, yellow, and blue-green, are carefully made by the use of transparent media, water colors or printing inks, and are superposed, the results shown in Fig. 20 are obtained. First let us take a simple case of a yellow pigment on a white surface. The light passes through the colored film and is reflected back through it by the white surface. As the light passes through the yellow pig- ment it is robbed of the violet and blue rays, there- fore the light which reaches the eye is white minus violet and blue rays, and produces a sensation of yellow. In the processes of painting and color print- ing the three disks may be assumed to be micro- scopic in size, each being a minute flake of pigment. If two flakes be superposed, a yellow above a blue- green, a green color is obtained. The yellow flake does not transmit blue rays, therefore the green rays are the only remaining rays that will be transmitted by the blue-green pigment. These will be reflected by the white surface, and will pass again through the blue-green and yellow pigments, undergoing further changes tending to purify them, so that only green rays reach the eye. If the blue-green flake is above the yellow flake, the explanation must be reversed, but with the same result. The blue-green flake trans- mits blue and green rays; however, the yellow flake does not transmit blue rays. Therefore, only the COLOR-MIXTURE 57 green rays will eventually be reflected to the eye. In the same manner the blue of the purple is sub- tracted by the yellow flake, and as purple consists of red and blue rays only, the red rays remain to be reflected to the eye. Therefore, yellow and purple flakes superposed produce red. Likewise the blue- green flake does not transmit red light, so that super- position of blue-green and purple flakes results in blue light being reflected to the eye. It is further seen that the superposition of the three subtrac- tive primaries results in a total extinction of light and black is the result. For instance, where the yellow and purple disks overlap, red results. The blue-green disk does not transmit red rays, so where it overlaps the red disk a total extinction results. Much interesting information may be obtained by carefully studying Fig. 20. Strips of colored gelatine laid over each other in checkerboard fashion present many striking examples of the subtractive method of mixing colors. In ordinary artificial light, screens made of ethyl violet (purple), uranin or aniline yellow, and filter blue-green, are excellent dyes for making the subtractive primaries for demon- strating the foregoing by superposition. Ethyl violet and naphthol green are practically complementary, so that when superposed no light rays are transmitted. 19. Additive Method. — As already indicated, there are two distinct methods of mixing color, — the additive and subtractive, — but close investiga- tion often reveals both processes entering into some part of the production of color. The additive method always tends toward the production of white, whereas the subtractive method tends toward the production of black. The additive primaries are red, green, and blue. Some prefer to use the term * violet' instead 58 COLOR AND ITS APPLICATIONS of * blue.' Blue, however, appears satisfactory and is a safer term than violet, because there are a great many who apply the term violet to purples. Long ago it was demonstrated th^t, by proper mixtures of the three well-chosen primary colors, any color can be matched. This is largely due to the fact that the eye is a synthetic rather than an analytic instrument. In Fig. 21 (Plate II), are illus- trated the principles of color-mixture by the additive method. It is seen that red added to green produces yellow; and further, when blue is added to this com- bination white is produced. In other words, yellow and blue mixed by addition produce white. It is well known, however, that yellow and blue (in reality blue-green) pigments when mixed by the subtractive method, as is done in painting and color printing, produce green. This is a much confused point, but is very simply explained when the character of the procedure of mixture is analyzed. Red and blue when added produce purple; and blue and green produce blue-green. It is to be noted that combina- tions of two of the additive primaries produce the subtractive primaries and vice versa. The additive method can be readily demonstrated by the use of colored lights projected upon a white surface. Prop- erly selected color-screens are necessary, but can be readily made from aniline dyes by carefully mixing them. It is difficult to describe the procedure quan- titatively, but there is no difficulty in producing the proper colors. Owing to the very unsatisfactory state of color terminology, it is impossible to present an accurate and definite list of complementary hues. However, a few complementaries are given in Table IV. COLOR-MIXTURE 59 TABLE IV Complementary Hues Red Blue-green (Cyan blue) Orange-red Green-blue (bluish cyan'i Orange Blue Yellow Blue-violet Yellow-green Violet-purple Green Purple (magenta) Wave-length of Complementary Spectral Hues 0.6562 /A 0.4921 ft .5671 /A .4645 fi .6077 .4897 .5644 .4618 .5853 .4854 .5636 .4330 .5739 .4821 An excellent scheme for showing the comple- mentaries is to arrange the spectrum around the circumference of a circle filling a gap between the ends of the spectrum, violet and red, with a series of purples from bluish purple to reddish purple. This has been called a color wheel, and is diagrammatically shown in Fig. 22. Here yellow and violet are shown as complementary. This may appear inconsistent with the foregoing discussion, but it will be noted that the terms *blue' and * violet' (as well as other color names) are indefinite. The term *blue' will always mean a spectral blue, but when used as a primary color its hue is defi- nite, whether the term stands for blue, violet, or blue-violet. If the complementaries have been correctly applied to the color wheel, a neutral gray should be obtained when it is rapidly rotated. Fig. 22.— The color-wheel for show- ing complementary hues. 60 COLOR AND ITS APPLICATIONS 20. Juxtapositional Method, — If a color be broken up into its component colors and the latter be applied in small dots with the point of a brush, the sensation of the original color will be obtained if it be viewed from a distance at which the eye is un- able to resolve the individual dots and providing the relative areas covered by the various colored dots are correctly balanced. Colors, excepting those en- countered in the spectrum, are usually far from mono- chromatic (#12), (Figs. 122, 123). For instance, a colored fabric which may appear a pure red will be found to reflect rays throughout considerable range of wave-lengths. If these component colors be repre- sented as pure as possible in minute dots of proper relative amounts, the foregoing result is readily obtained. For instance, if one end of a pack of cards be painted red and the other end green, on revers- ing every other card and viewing an end of the pack at a distance of several feet, it will appear yellow in color. The brightness apart from hue will be an average brightness. Many interesting experiments can be performed by ruling alternate fine lines of different colors on paper or on glass. For instance, purple and green lines alternated on paper will, if well chosen, produce an appearance of gray at some distance. Such a method of breaking a composite color into more nearly monochromatic components and applying the latter in the form of minute dots is the foundation of the principle of impressionistic painting. The processes of color photography devised by Joly, Lumiere and others are also based on this principle. 21. Simple Apparatus for Mixing Colors. — There are very elaborate color-mixing instruments on the market for the purpose of demonstrating th^e COLOR-MIXTURE 61 theory and practise of color-mixture. Apparatus that deals with spectral colors is as a rule the most sat- isfactory for accurate study and demonstration. How- ever, inasmuch as the colors ordinarily available in practise are far from monochromatic, that is, far from spectral purity, there is much virtue in the simpler forms of apparatus that can be made at small expense. In fact, for the foregoing reason the results obtained with some of the simpler instruments for demon- stration are more readily interpreted and applicable in practise than those obtained with apparatus dealing with pure spectral colors. Maxwell's disks offer a ready means for mixing colors. A shaft is arranged so as to be revolved at high speed. Colors painted on a disk can thus be mixed by rotating it at a high speed owing to per- sistence of vision. Light sensations do not reach their full value immediately upon application of the stimulus, nor do they decay to zero immediately upon the cessation of the stimulus. An infinite number of mixtures of pigments, including black and white, can be made with such a simple disk. Colored papers cut in circles and slit along one of the radii can thus be overlapped to any degree, and by the use of circles of various sizes a number of mixtures can be produced upon the same disk. This method is not truly an additive one, excepting in the addition of hues. The brightness is the mean of the sepa- rate brightnesses, each weighted by its angular extent. In Fig. 23 are typical color disks for mixing colors to produce grays. In I and III are repre- sented pairs of complementary colors respectively, yellow and blue, and green and purple. The inner circle consists of black and white, which can be varied in angular amounts to produce a neutral gray to 62 COLOR AND ITS APPLICATIONS match the gray produced by the addition of the two hues. In II are represented the three primary colors which when mixed by rotation produce a neutral gray which is readily matched by means of the inner black and white disks. These matches made under one illuminant will not ordinarily remain matches under another illuminant. Much of the early work in the science of color was done by means of rotating disks and even today they are extremely valuable in some investigations. The disks represented in Fig. 23 can be readily made from Zimmerman's Fig. 23. — Maxwell disks. colored papers. These papers are indicated in the catalogue by the letters of the alphabet and are given herewith as used in the disks already described. Yellow is designated as ^, blue as o, green as z, red as by and purple as a. For the black and white sectors any neutral tint papers with dull finish are satis- factory for producing the grays. The additive and subtractive methods as illus- trated in Figs. 20 and 21 (Plate II) can readily be demonstrated in permanent charts. The imported colored papers have been found satisfactory, owing to their comparative purity and unglazed surfaces. For demonstrating the subtractive method by the three overlapping disks the six colors and black are surrounded with a white background. The Zimmer- COLOR-MIXTURE 63 man colors, designated by a, ^, Z, &, z, and o, may be used respectively for purple, yellow, blue-green, red, green, and blue. These six colors, with black and white, are sufficient for the construction of charts for the additive and subtractive methods. For demon- strating the additive method the three disks should be surrounded with black background, but in the case of the subtractive method the background should be white. For demonstrating these two methods of color-mixture with artificial light by means of transparent media, purple and green are readily produced by using gelatines dyed with ethyl violet and naphthol green respectively. When these two colored gelatines are superposed in proper densities of coloring, no light is transmitted. When light is passed through these media in juxtaposition in proper relative amounts and combined on a neutral tint diffusing surface a white light is produced. These two colors afford an excellent example of comple- mentary colors when used with artificial light. In daylight the ethyl violet screen appears deep blue in color, instead of appearing purple as it does in the light from a tungsten incandescent lamp. Other transparent media for further demonstrating these methods are readily selected from the many organic dyes available. Uranin, fluorescein, carmine, patent blue, and filter blue-green are satisfactory. In Fig. 24 the construction of an erratic color- mixing disk is illustrated. To a disk of stiff card- board a sectored disk of cardboard is rigidly fastened by means of a circular rivet. The latter disk has two 60° openings, as shown in a. Another disk, arranged concentric with the other disks and between them, is permitted to slip at will about the rivet as an axis. If the latter disk is prepared as shown in 64 COLOR AND ITS APPLICATIONS b many striking colors are obtained on rotating the combination. Fig. 25 illustrates a simple arrangement for color- mixing. The wheel is similar to that employed in the Simmance-Abady flicker photometer. The periphery Fig. 24. — An erratic color-mixing disk. of this wheel consists of truncated cones pointing in opposite directions. The axes of the cones are eccentrically placed at equal distances on either side of the axis of the wheel and parallel to it. In the angular position shown in the illustration, S s L' Fig. 25. — A simple color-mixer. the eye, looking at the wheel in a direction at right angles to the axis, sees one conical surface illumi- nated by one light, L, and the other by the other light, L\ Colored screens, SS, axe interposed between the wheel and the light sources. By having the lamps movable on a track any combination of brightnesses of two colors from transmitting media can be mixed COLOR-MIXTURE 65 by rotating the wheel rapidly. Pigments may also be applied directly to the wheel. In Fig. 26 is illustrated another simple instrument for mixing the colors from either opaque or trans- parent media. A wooden box is constructed as shown and painted black inside. G is a transparent plate glass and OO are ground opal glasses free from color. In mixing the colors of two transparent media, CC, the lamp, L, is moved to and fro on its track. Thus any propor- tions of the two colors can be mixed. If the colors of two opaque substances are to be mixed, and 00 are removed, the colored objects Fig. 26. — A simple color-mixer for transparent or opaque media. C/ / / / / / / / / / fc/e \ \ \ \ \ \ \ \C Fig. 27. — Lambert's color-mixer. CC are placed at PP, and the lamp is moved to and fro as before. The range of mixtures in the last case is not infinite as in the case of transparent media; however, modifica- tions can readily be made so that — the range is extended. A simple experiment devised by Lambert, though not having the flexibility of the foregoing instruments, 66 COLOR AND ITS APPLICATIONS is of interest owing to its extreme simplicity. It is illustrated in Fig. 27. G is a plate glass and CC are colored objects. The colors are mixed one by re- flection, the other by transmission. By turning the glass and shifting the eye the proportions can be altered considerably. An apparatus of considerable use is a booth con- taining red, green, and blue incandescent lamps con- trolled by rheostats. If the colors are carefully made many interesting experiments can be performed, in- cluding the effect of quality or spectral character of light upon colored objects (#67). Many instructive experiments can be produced by the use of shadows cast by colored lights. One of especial interest is shown in Fig. 28, because it pre- sents the additive prima- ries, their complementaries (the subtractive primaries) and white light produced by the sum of the three primary colors, — red, green, and blue. In the middle of a circle of white diffusing blotting paper stiffened by a board are erected three planes of white diffusing material about eight inches in height. The latter meet at the center of the circle at angles of 120 degrees with each other. At points several feet away, along the three arrows, red, green, and blue lights are placed somewhat above the plane of the circle. These should be small sources and quite powerful, concentrated tungsten filament lamps being quite satisfactory. The experiment is best Fig. 28. — A shadow demonstration of the additive and subtractive methods of color-mixture. COLOR-MIXTURE 67 seen if the plane of the circle is vertical. It will be seen that the nearly rhombic areas on the circle indicated by i?, G, and B each receive light from only one source. These areas will then appear respectively red, green, and blue. The areas on the opposite sides of the circle, 5G, P, and 7, each receive light from only two sources. They appear in colors com- plementary to the above primaries. They also rep- resent the subtractive primaries and the colors which remain after red, green and blue are subtracted respectively from three white lights. The remaining areas of the circle marked W represent the regions which receive light from each of the three sources, with the result that if the colors and intensities of the light sources are correct, and if the sources are sufficiently distant in comparison with the size of the circle, these areas appear a uniform white. This experiment is simple and is very satisfactory for dem- onstration before large audiences. The lights should be controlled by separate switches and rheostats. A rotating disk can be readily colored, so that it will appear, when viewed through a radial slit placed close to it, a fair approximation to the spectrum. The mixing of the colors by rotation obviates the neces- sity of the great care in blending colors in painting a spectrum that is to be viewed when stationary. The colors will not be of spectral purity, owing to the limitations of the pigments, but the disk will be instructive and affords a ready means of producing a spectrum for reproduction by color photography. An approximation to the prismatic spectrum can be readily produced as shown in Fig. 29. The approxi- mation can be made as close as desired by touching up various points with pigments where necessary or by varying the geometric figures. If a circle be cir- 68 COLOR AND ITS APPLICATIONS cumscribed about the inner square and a square in turn be circumscribed about this circle, and so on until four circles and four squares are pro- duced, the skeleton of the figure is ready for coloring. If the inner square be painted black and the succeed- ing hollow squares be painted red, yellow, green, and violet re- spectively and the re- mainder of the outer circle be painted black, a fair approximation to a prismatic spec- The same results can be pro- Fig. 29. — Illustrating a disk for approxima ting a prismatic spectnun. trum is obtained, duced with more difficulty with only three colors — red, green, and blue. If the spectrum pro- duced by rotation is not satisfactory at all points, it can be readily made so by the judicious use of pigments, or, as already stated, by altering the geo- metric figures. The more simple methods of mixing colors have been described in this chapter; however, it will be borne in mind that many of the instruments and methods considered in succeeding chapters are di- rectly or indirectly applicable to color-mixture. REFERENCES Captain W. de W. Abney, Colour Measurements and Mixture, London, 1891. O. N. Rood, Colour, 1904. Chevreul, Harmony and Contrast of Colours, 1839. CHAPTER IV COLOR TERMINOLOGY 22. Hue, Saturation^ and Brightness. — One of the greatest needs in the art and science of color is a standardization of the terms used in describing the quality of colors and an accurate system of color notation. The term * color' in its general sense, is really synonymous to the term 'light.* It is used here by preference because it implies the consideration of the appearance of a surface or material object. The spectrophotometer is the most analytic instru- ment for examining colors (#26). By means of it the amounts of light of all wave-lengths reflected (or transmitted) by a colored medium may be obtained. These data are plotted in the form of curves shown by the dashed lines in Fig. 12. The full line curves represent the reflection (or transmission) coefficients of the pigments for energy for various wave-lengths. If the region under one of the curves indicated by a dashed line be integrated and this area compared with that obtained with the same illuminant for a white diffusing surface of known total reflecting power, the relative brightness of the colored medium under this illumination is obtainable. The domi- nant hue which is discussed below may be usually approximately determined by inspection of the curve, although in many cases it is impossible to estimate the dominant hue in this manner. It is thus seen that although the spectrophotometer is a valuable instru- ment for analyzing colors, there are further require- 69 70 COLOR AND ITS APPLICATIONS ments in color work better met by other instruments (Chapter V). The quality of any color can be accurately de- scribed by determining its hue, saturation or purity, and its brightness. (The latter term is analogous to the term * value' as used by the artist.) In the broadest sense, white, gray, and black are here con- sidered as colors, and a mere change in brightness alone is considered as a change in color. It appears necessary to assume this broad definition of color, inasmuch as brightness is distinctly one of the products of color analysis. Hue is suggested in the name applied to the color. The dominant hues of most colors are accurately represented by spectral colors; however, there are composite colors, — the purples, which consist of red and violet, for which no spectral colors are found to represent their hues. In these cases it is satisfactory to determine the domi- nant hue of the complementary colors. The satu- ration or purity is a measure of the relative amount of white light in the color. In other words, all colors excepting purples can be matched by diluting spectral light of a definite wave-length with white light. The greater the percentage of white light required in the mixtures, the less saturated the colors are said to be. The brightness of a color can be found by com- paring it by means of a photometer with a surface of known brightness. It is well to note that in the analysis of a color its absolute brightness is measured by comparing it with a brightness of known value. Inasmuch as its brightness depends upon the inten- sity of illumination of a given spectral character, its reflection coefiicient for a standard white light should be determined in order to compare it with other colors in this respect. This latter measurement in- COLOR TERMINOLOGY 71 volves all the difficulties of color-photometry treated in Chapter IX. There is much confusion in the application of the terms *tint,' *tone,' * shade,' 'intensity,' etc. Many use these terms wholly unjustifiably. It is true that the final usage is somewhat a matter of choice at the present time, but the terminology adopted here appears to the author to be consistent with other nomen- clature adopted by the physicist, photometrist, and Q b Fig. 30. — Disk *a,' for varying only the saturation of a color. — Disk * b,' for varying only the brightness of a color. lighting expert, and best justified by usage and the dictates of common sense. On diluting a color with white light, tints are obtained; that is, tints are un- saturated colors. By the admixture of black to a color (in effect the same as reducing the intensity of illumination) the brightness is diminished without altering either the hue or the saturation, and various shades are produced. Only the relative brightnesses of shades are usually of interest, although for obtain- ing a basis of notation it may be desirable to deter- mine their absolute values. In a. Fig. 30, is shown a simple means of varying the saturation of a color without altering either the hue or brightness. On a 72 COLOR AND ITS APPLICATIONS circle of colored paper is glued a gray paper of the same brightness for the given illumination and of the form shown by the shaded area. On rotating the disk this gray will be mixed in various angular proportions from 360 deg. to 0 deg. The gray paper, having been selected of the same brightness as the colored paper under the illuminant used in the experi- ment, does not alter the brightness upon mixing the two components by rotation; being non-selective in its reflection it does not alter the hue. Thus various degrees of saturation of the original color are obtained. The brightness can be varied, as shown in b, Fig. 30, without altering either the hue or saturation by fastening to the original circle of colored paper a black paper cut in the same form as the gray paper shown in a. If the paper were perfectly black it is seen that it cannot alter either the hue or satura- tion. As a matter of fact no available black papers are totally non-reflecting, so that some light is added to the color. This can be reduced to a minimum, however, by the use of a hole in a deep velvet-lined box. In this case the black sectors shown in b would be replaced by openings of the same contour in the disk. For convenience of construction the areas occupied by the black and colored papers may be reversed. In this connection it is well to emphasize that ordinary black surfaces are far from totally absorbing. This can readily be demonstrated by a box open at one end lined with black velvet. Over the open end place a black cardboard with an opening in it and it will be seen that the opening will appear very much darker than the black surface surrounding it. The foregoing demonstration may be easily performed by varying the brightness of a colored paper relative to that COLOR TERMINOLOGY 73 of a paper of the same color which surrounds it, by vary- ing the intensity of the illumination of the patch at the same time maintaining the absolute brightness of the surroundings constant. Instruments have been designed for the analysis of color quality into the three component factors, hue, saturation, and brightness. These are treated in #27. 23. Tri-color Method. — It is well known that any color can be matched by combining the three primary colors, red, green, and blue, in proper pro- portions. Many instruments have been devised for this purpose, the most elementary being the Maxwell disks, and the more elaborate and accurate are those employing spectral colors. The results of such a method are expressed mathematically in the equation xR+yG + zB = C, where the values of jc, ^, and z are the fractional parts of the red, green, and blue lights, respectively, that must be combined to match the color, C. This method has limitations because it does not give the results directly in terms of hue, saturation, and bright- ness. Some of these in- struments, however, can readily be adapted to the measurement of the last two factors. In order to plot the values of jc, y, and z, it is necessary to employ tri-linear coordi- nates, there being three variables to be repre- sented. The results are readily represented in the Maxwell color triangle, illustrated in Fig. 31. The green component increases from zero at the base Fig. 31. — The Maxwell color-triangle. 74 COLOR AND ITS APPLICATIONS line, RBj to 100 per cent at G. Likewise the red, Rj and green, G, components increase from zero at the base of the perpendiculars erected from the sides respectively opposite to their apexes. The data are plotted by erecting three perpendiculars proportional to the respective values of /?, G, and 5, starting at points in the opposite sides of the triangle respectively, such that the three perpendiculars intersect at the same point. Purples are found along the base line, RB, varying in the proportions of R and B from /? = 0, B = 100, to R= 100, B = 0. Yellows are found along RG and blue-greens along GB, White, which is usually represented by ^R-\- lG-\- ^B = W, is found at the center of the triangle. The curved line rep- resents the positions of the spectral colors in the color triangle; that is, each point on the curve rep- resents the primary sensation values of a particular spectral color. Some important lines of the spectra of cadmium and mercury are also shown. The less saturated colors are found near the center of the triangle and the more saturated ones near the sides. It is thus seen that spectral colors throughout a large range of wave-lengths arouse the three primary sensations, according to the Young-Helmholtz theory. (See #28, 47.) The primaries of course are found at the angles of the triangle. Complementaries are represented as being on opposite sides of the center of the triangle on a straight line passing through it. The dominant hue of a color is found by drawing a straight line from the center of the triangle through the point representing the color and continuing it until it intersects the curve representing the spectrum. The latter point of intersection represents the domi- nant hue of the color. The tri-color method in- volves the use of an invariable white light, that is COLOR TERMINOLOGY 75 1 noon sunlight or its equivalent. A curve represent- ing spectral complementaries is shown in Fig. 32. Results obtained by this general method with different instruments are likely to vary considerably. This is due in part to vari- ations in the spectral char- acter of the white light standard, and also to the transmission characteris- tics of the color-screens used in these instruments not employing spectral primary colors. The primary sensation values of the screens should be determined and the measurements be given in sensation values (#28). The use of the plane triangle is limited to the plotting of the analyses of colors of equal brightness. In order to include ^Q the brightness factor the figure \J.OKJ ' 0.45 1 An 055 0.60 yd. WAVE LENGTH 0.65 Fig. 32. — spectral complementaries. takes the form of a solid inverted, pyramid, shown in Fig. 33. The various triangular planes parallel to the base represent planes for plotting colors of different bright- nesses. The apex represents black. A line joining the point, W (white), with the apex passes through a complete range of shades of white, that is, of grays. Along the dotted line from x to the apex are a series of colors of constant hue and saturation, but varying in brightness. The color pyramid has been modified in various rays to fit experimental results in- Black Fig. 33. — A color pyramid. 76 COLOR AND ITS APPLICATIONS volving physiological and psychological influences. One of these modifications from Titchener ^ is shown in Fig. 34. At the two poles of this double pyramid are the extremes of white and black; upon the axis connecting the two poles are located the complete range of grays. Around the pe- riphery of the middle plane are located those colors of middle brightness and maximal saturation. Other points in the solid represent other colors of varying hue, satura- tion, and brightness. From the base toward white, tints are found; in the other direction shades are found. This pyramid, it will be noted, has four sides, the four angles representing red, yellow. Fig. 34. — The double pyra- mid. (After Titchener.) green, and blue. Obviously this solid does not directly represent color analyses as obtained by the tri-color method. Its significance will be better understood on referring to the Hering theory of color vision in #49. The tri-color method is discussed further in Chapter V. A simple means ^ of dem- onstrating the Maxwell color triangle in actual colors is illustrated in Fig. 35. A box 6 inches in depth, and whose section forms an equilateral triangle about 18 inches on a side, is made of wood, with its back containing vent holes. A ground flashed-opal glass in Fig. 35. — A demonstration color triangle. the form COLOR TERMINOLOGY 77 of an equilateral triangle somewhat smaller than the section of the box forms the front side. In the three corners of the box are placed respectively, red, green, and blue spherical-bulb, concentrated- filament tungsten lamps. After proper adjustments of the position and color of the lamps, the diffusing glass, which has its roughed side inward, assumes the colors of a color triangle. A close approximation can be approached, depending upon the care exer- cised in adjusting the position of the lamps and the distribution, color, and intensity of the light. Inter- esting demonstrations of retinal fatigue and after- images are readily made with this simple apparatus. For coloring the lamps ordinary colored lacquers are satisfactory if properly mixed to obtain the exact hues. The aniline dyes can be used with satis- faction. These colors are not permanent, but are sufficiently durable for such an apparatus. If the coloring is placed on separate plates of glass, it will remain unfaded for a long time with proper ventila- tion. 24. Color Notation. — The need for a universal color notation is admirably illustrated by Munsell ^ in quoting from a letter by Robert Louis Stevenson, writing from Samoa to a friend in London, as follows : " Perhaps in the same way it might amuse you to send us any pattern of wall paper that might strike you as cheap, pretty and suit- able for a room in a hot and extremely bright climate. It should be borne in mind that our climate can be extremely dark too. Our sitting room is to be in varnished wood. The room I have particu- larly in mind is a sort of bed and sitting room, pretty large, lit on three sides, and the colour in favour of its proprietor at present is a topazy yellow. But then with what colour to relieve it? For a Uttle work-room of my own at the back I should rather like to see some patterns of unglossy — well 1^11 be hanged if I can describe this red — it's not Turkish and it*s not Roman and it's not Indian, but it seems 78 COLOR AND ITS APPLICATIONS to partake of the two last and yet it can't be either of them because it ought to be able to go with vermillion. Ah, what a tangled web we weave — anyway, with what brains you have left, choose me and send me some — many — patterns of this exact shade." Here is a man accustomed to present his thoughts in writing in a clear manner, yet he acknowledges failure in his effort to describe colors and closes his letter with the request, perhaps a bit sarcastic, that he be sent "patterns of this exact shade." Other sciences have exact and practically universally ac- cepted terminology. Music has its well-developed notation, which is definite and descriptive, and quite universal in adoption, but there is no universal scheme of color notation. Colors are named in very inexact, unwieldy, and often totally non-descriptive terms. We have rose, Indian red, Alice blue, pea green, olive green, cerise, taupe, baby blue, Copen- hagen blue, king's blue, royal purple, invisible green, etc. Thus flowers, vegetables, cities, the savage and the royal family, are used to describe colors. Is there a more ridiculous instance of neglect.'* Those who work in color often find themselves helpless in describing colors to others. Surely a color notation based upon color science should be acceptable, even though somewhat empirical. Musical notation is somewhat arbitrary, yet it has met with almost uni- versal adoption. An acceptable color notation must involve the factors which influence the quality of a color, namely hue, saturation, and brightness. An attempt was made by Runge as early as 1810 to build up a color notation by the use of a sphere with red, yellow, and blue, placed around the equator and separated from each other by 120 degrees, with white and black at opposite poles. Perhaps the greatest virtue in this attempt is the fact that it was COLOR TERMINOLOGY 79 one of the early constructive efforts. Chevreul, whose work on the effect of simultaneous contrast of colors in the practical textile industry is well known, constructed a hollow cylinder built up of ten sections perpendicular to the axis. Around the upper section red, yellow, and blue were equally spaced. The lowest cylindrical section was black, and the eight intervening sections were graded from top to bottom by adding increasing amounts of black. Munsell criticises these attempts, on account of the yellow being very light and the blue being very dark, which makes impossible any coherency in the brightness scales of the three colors. Inasmuch as the bright- ness scale of the yellow in the Chevreul Qolor cylinder increases much more rapidly from the bottom toward the top than the brightness scales for blue and red, Munsell suggests that the yellow side of the cylinder be increased in length. This would result in the tilting of the sections more and more as the scale of brightness progressed from the bottom toward the top. Perhaps a general criticism for most of these schemes of color notation is that geometrical figures are chosen and the colors are made to fit. The latter method is perhaps partially justifiable from the standpoint of physical measurements. There is another viewpoint in considering a color notation, and that is from the standpoint of harmony of color. In this treatise we are not so much concerned with the latter viewpoint, but it is of interest to consider a system of color notation devised by Munsell from the standpoint of the use of color in painting. In describing a color by this system the initial of the name of the color indicates the hue, and numerals represent the saturation and brightness. For example R7 represents a color whose hue is red and whose 80 COLOR AND ITS APPLICATIONS saturation and brightness are respectively 7 and 5. The brightness scale is divided into ten parts, and the degrees of saturation shown vary with the bright- ness. For simplicity ten hues are balanced around the equator of the sphere somewhat after the manner shown in Fig. 22. The lower pole of the sphere is black and corresponds to zero on the brightness scale. The upper pole is white and corresponds to brightness 10 on the same scale. On slicing off a portion of the sphere through a plane corresponding to a certain brightness, various degrees of saturation are encoun- tered. The saturation decreases toward the center, the axis of the sphere consisting of a scale of gray S. However, the sphere does not completely satisfy Munsell. He therefore constructs a * color tree' so that varying numbers of steps in saturation can be represented, depending upon the hue and position in the brightness scale. The scheme is built up on the principle of the harmonious use of colors and in this respect departs somewhat from the scope of this book, which treats more with physical mixtures, regardless of the use of colors in harmony. The system is an interesting one and is the result of a noteworthy attempt to be freed from a state of color anarchy. MunselPs color tree is illustrated in simple form in Fig. 36. The base of the tree is black, the top white. In the small model illustrated three bright- ness levels are shown, namely 3, 5, and 7 in the arbitrary brightness (value) scale. The degrees of saturation shown vary with the * brightness level.' At * level 3' in the brightness scale blue is shown to the eighth degree of saturation. By the irregularity in the contour of the planes representing different brightness levels it is seen that the relative number COLOR TERMINOLOGY 81 of degrees of saturation shown for various colors depends upon the brightness level under considera- tion. At brightness level 3, PB (purple-blue) was shown with the highest degree of saturation, namely, 8. At brightness level 5, R ranked first in degree of saturation, its highest being 10. At the brightness level 7, yellow was shown with the highest degree Wh/fe Ye/fov/ Yet low- red Red Red-purple Fig. 36. — The A. H. Munsell color tree. of saturation, namely 8. For a complete discussion of this system, the reader is referred to the original description by Munsell.^ Numerous scales have been devised involving, either separately or combined, the factors hue, saturation, and brightness. All of these assist in bringing order out of chaos, but they constitute only the first steps toward a comprehensive system of color notation. The hues are usually expressed by names of spectral colors and purple, but the bright- ness is seldom more definite than is found in such expressions as W, HL^ L, LL, Af, HD^ Z>, ZZ>, and 82 COLOR AND ITS APPLICATIONS By which represent white, high light, light, low light, medium, high dark, dark, low dark, and black re- spectively. Examples of such charts (devoid of color) are shown in Fig. 37. These were taken from Book VI of Prang's text-book of art education. Such NtUTRAL VALUe CHART -B CHART - C Fig. 37. — Prang*s color and brightness scales. charts should pave the way toward a final scientific color notation. Another system of color notation is shown in Fig. 38. This is used by Ruxton in the mixture of Fig. 38. — Ruxton's color mixture chart for printing inks. printing inks. The chart is printed in colors, there being 144 colors, varying in hue, saturation, and bright- ness. The terminology is somewhat different than used in this text. The 144 colors are obtained from six fundamental colors, namely red, orange, yellow, green, blue, and purple. These six colors are de- scribed as spectral colors, so it is likely that purple COLOR TERMINOLOGY 83 is the name applied to a color meant to be violet. The starting point in obtaining the total array of colors is in the bottom row of Section 3. The larger rectangles represent the six fundamental colors, which are the purest or most saturated on the chart. The fundamental red is marked 820. The small square areas represent the intermediate hues and are ob- tained by mixing the fundamentals on either side. Red-orange for instance is obtained by mixing red and orange (820 and 840). The three horizontal rows above this row of twelve colors are made by adding white to the colors of the bottom row. Thus in the top row are found the least saturated colors in Section 3. Two degrees of saturation lie between the top and bottom rows. Thus in Section 3 there are 48 colors, six fundamental colors increased to twelve by mixing adjacent fundamentals (red and violet are mixed, producing 810) and these twelve colors decreased in saturation in three steps by the addition of white. Sections 1 and 2 are produced by adding black to the corresponding colors in Section 3, thus reducing their brightness. In Section 1 are the colors of lowest brightness. These are named *hues' but in a different sense than that in which the term *hue' is employed here. They could be termed * Values' with better consistency. The *bi- hues' (bi-values) in Section 2 are obtained by mixing one part by weight of the colors in Section 3 to one part of the corresponding color in Section 1. Thus it is seen that in a more correct sense there are twelve hues represented on the chart (bottom row in Section 3). With the hue and brightness constant saturation is found to be present in four degrees (moving vertically in Section 3). With the hue and saturation constant, the brightness is found 84 COLOR AND ITS APPLICATIONS to be present in three values (moving from left to right, Sections 3, 2, 1, so-called * colors,' *bi-hues' and * hues'). Besides these there are 72 other colors in which brightness and saturation appear in six combinations for each of the 12 hues. In other words, in a broad sense there are present 144 colors made up of twelve hues by varying the brightness and saturation. Six of the twelve hues are made by mixture of the adjacent hues in the bottom row of large rectangles in Section 3. Each rectangle being numbered, the chart systematizes the mixture of printing inks. Such progress is commendable and highly desirable, even though empirical. There are many other methods, but these few have been cited to show the lack of standardization of color notation and to illustrate that a system however empirical is just as desirable for the de- scription of color as a system is for music notation. There is much yet to be done before a system of color notation is devised which will be universally adopted. First there should be some definite terms adopted descriptive of the factors influencing the quantity of a color, namely * hue,' * saturation,' ' brightness.' The term *hue' is used in a more definite sense than the terms applied to the two other factors. For satura- tion the terms * chroma,' * purity,' 'intensity,' and others are being used. For brightness the terms * luminos- ity,' * value,' * hues' or *bi-hues,' and others are being used. Purples are often called violets or reds. These are examples of usage from which general confusion arises. The problem of color terminology does not defy solution. As a matter of fact all the quantities involved in a scientific system of notation are readily measurable. Hue, saturation, and bright- ness are easily determined. The available hues. 1 COLOR TERMINOLOGY 86 with the exception of purple, are invariable, consisting of the spectral hues. Scales of brightness (value) can be divided into any given number of parts and named in some consistent manner. The use of the terms *high light,' *low light,' * medium,' *low dark,' etc., is perhaps satisfactory, but the brightnesses that they represent should be standardized in absolute measurements in order to produce a universal scale of relative brightnesses. In fact all the terms re- quired in a satisfactory and scientific system of color notation can be measured for their absolute values. This would reduce the systems to one basis. Such a universal system must certainly be adopted even- tually, and those interested in color should put forth effort to hasten the day. REFERENCES 1. Primer of Psychology, 1899, p. 41. 2. Psych. Rev. 20, May, 1913. 3. A Color Notation. OTHER REFERENCES Sir William Abney, Color Mixture and Measurement. p. N. Rood, Textbook on Color. J. G. Hagen, Various Scales for Color-Estimates, Astrophys. Jour. 1911, 34, p. 261. k. Zindler, Color Pyramid, Zeit. f . Psych. 1899, 20, p. 225. R. Ridgeway, Color Scales. CHAPTER V ANALYSIS OF COLOR 25. The Spectroscope, — As already indicated, colors can be analyzed in various ways. The method adopted in a given case will naturally depend upon data desired. The spectroscope affords a simple means of examining colored light, but the results of the visual inspection are only qualitative. There are various designs of spectroscopes available, all based upon either the principle of the prism or of the diffraction grating (#8 and 9). An ordinary prism spectroscope can be converted into a direct-vision instrument by combining two prisms made from different kinds of glass, so that dispersion is obtained for a certain ray without deviation. Crown and flint glasses differ in refractive index (Fig. 8), hence if prisms of each of these two glasses be made of proper refractive angles and combined so that their j separate deviations prac- I /\f/\ (a I tically annul each other, I / V N \i I say, for the sodium line, Fig. 39. — A direct-vision prism spec- dispersion is produced *^°^^°P^- without deviation for this ^ ray. Such a simple spec- troscope is shown in! Fig. 1 39. A simple diffraction Fig. 40. — A simple grating spectro- grating SpOCtrOSCOpe Can ^^°^®* be readily made, as shown in Fig. 40. A replica of a diffraction grating (#9) is placed between two pieces of plate glass at G. By I ANALYSIS OF COLOR 87 placing a lens at L the instrument is considerably shortened, so that it can readily be made of a pocket size. These, the simplest forms of spectroscopes, are only useful for rough qualitative analysis of the spectral character of light emitted by light sources or transmitted or reflected by colored media. A convenient form of spectroscope for qualitative analy- sis is the comparison spectroscope. This contains two or three distinct optical systems, so that two or three spectra may be viewed in juxtaposition. Such an instrument might be considered as roughly quan- titative in its analyses, owing to the opportunity of estimating relative intensities of a given light ray in the two or three spectra. More elaborate spectrometers will not be con- sidered here, for the function of the spectrometer is for qualitative analysis. However, in this respect such instruments are of considerable value in color work. Photographic accessories are readily attached in place of the eyepiece. If an absorption wedge be placed before the slit, so that its transmission varies along the length of the slit, the spectrograms will roughly indicate the relative spectral distribution of energy, providing proper filters are used to allow for the variation in plate sensibility. Sometimes it is advantageous to compensate for the unequal spectral distribution of energy in the illuminant, especially in the examination of colored media. 26. The Spectrophotometer, — This instrument consists in principle of two spectroscopes, arranged so that the intensity of rays of the same wave-length in the two spectra can be photometrically compared. The results obtained are quantitative. A diagram- matic sketch of the optical system of a spectro- photometer is shown in Fig. 41. Light enters the 88 COLOR AND ITS APPLICATIONS 0-eE instrument from two sources at the slits S and S\ respectively. At L is a Lummer-Brodhun photometer cube so constructed that through a part of the* field light rays are trans- mitted directly from S' to the prism P and from the remainder of the field light rays from S are reflected toward the prism. After being dis- persed by the prism the colored rays pass on to the eye placed at T. The wave-length of these rays depends upon the angular position of the prism which can be rotated. The photom- eter field is similar to that viewed in an ordinary Lummer-Brodhun photometer. A small direct-vision comparison spectroscope is of considerable use in color work. Such an instru- ment designed by Nutting ^ contains a pair of Nicol prisms, NNy for altering the brightness of one of the spectra, as shown in Fig. 42. A right-angled Fig. 41. — The spectro- photometer. a^MI )( i Fig. 42. — The Nutting pocket spectrophotometer. prism, R, reflects light into the slit from one of the sources. The instrument, which is called a pocket spectrophotometer, in itself is merely for qualitative analysis. It can be set up permanently and used for quantitative measurements. However, in order to make such an instrument portable and compact, yet available for obtaining qualitative data, the author devised the attachments shown in Fig. 43. An attachment, i4, containing a miniature tungsten lamp ANALYSIS OF COLOR 89 which illuminates a ground flashed-opal glass, can be removed from its present position if desired and attached at B. When the comparison source in A Fig. 43. — A small portable spectrophotometer for quantitative analysis. is in the position shown the unknown source is placed at B. The sleeve, O, was placed on the instrument to support A. S controls the slit widths. A mi- crometer screw with a graduated scale and drum is attached at M to the slide containing the observing slit, C, which is moved across the image of the spectrum. The drum is calibrated in terms of wave- lengths and the scale of the revolving Nicol prism in terms of transmission of light. The current through the lamp is obtained from a battery controlled by a rheostat and is measured by means of a small ammeter. The range of the instrument can be extended by varying the current through the lamp. This photometric field is not as satisfactory as might be desired, for it consists of two narrow bands juxta- posed at their ends. The instrument is very small, 90 COLOR AND ITS APPLICATIONS less than 8 inches long, is mounted on a tripod, and is really portable. There are many designs of spectrophotometers, but all have the same object. It is necessary to be able to vary the luminous intensity of one-half of the photometer field. This is done by varying the position of one of the light sources, by the use of Nicol prisms, by a neutral tint absorbing wedge, or by sectored disks. A convenient means is the variable sectored disk developed by Hyde,^ a diagram of which is shown in Fig. 44. The disk is mounted upon a motor-driven shaft and arranged to be moved horizontally in its plane along the line CD in front of the slit S, by means of a micrometer screw. The transmission is nearly proportional to the lateral displacement. By means of the spectrophotometer, results can be obtained directly in terms of relative energy such as are plotted in Fig. 5 (Table II) and Figs. 122, 123. In this case the various rays in the unknown spectrum are compared directly with the corresponding rays in a spectrum of known distribution of energy. As has been previously stated the spectrophotometer is an analytical instrument, and by its use the spectral character of the light reflected or transmitted by colored media is readily obtained. An example of its use and a practical means of greatly reducing the number of readings is given below. During the Fig. 44. — The variable sectored disk. (After Hyde.) ANALYSIS OF COLOR 91 development of a glass which could be used with the tungsten lamp to produce artificial daylight ^ the procedure involved the examination of many glasses containing various proportions of coloring ingredients. A glass which proved unsatisfactory at the thickness at hand might be found satisfactory at another thick- ness. Therefore it was necessary to grind and polish the samples as they came from the glass factory into many different thicknesses or in the form of a wedge. 1.00 0.80 0.70 0.60 0.50 0.40 0.50 o OZO V) r 0.08 < 0O7 jl 005 004 005 0.02 6 i ^-0 0.466 0A85 0.508 0.52 f 0.536 '0.544 0.552 0.570 aseo 0.612 0.65Q 0.668 12 3 4 6 6 7 THICKNESS OFGLASS IM MILLIMLTERS Fig. 45. — Scheme for reducing the amount of spectrophotometric work in exam- ining transparent colored media. This necessitated making a set of spectrophotometric readings for a considerable range of thicknesses. By utilizing the law relating transmission and thickness (density of coloring matter) of the glass, namely / = he~^'^y a simple method was devised, h represents the original intensity of light of a certain wave-length and /, its intensity after traversing a thickness d of the colored glass and e, the extinction coefficient. By considering the reflection from the two surfaces of the glass a relation was deduced in the form of 92 COLOR AND ITS APPLICATIONS. Log T = log 0.92 -\-kd (where T is the transmission, d the thickness and k a constant) which is sufficiently accurate for ordinary purposes in the spectrophoto- metric analysis of the transmission characteristics of colored glasses and other media. The term *log 0.92' can be eliminated by obtaining the transmis- sion of the colored glass in terms of a clear glass if so desired. This method necessitates an analysis of only one thickness, for, on plotting these data on logarithmic paper, as shown in Fig. 45, the data for various other thicknesses (even thicker than the sample) are readily obtained. Proof of the accuracy of this method is shown by the fact that the circles which represent data obtained on the same sample of glass at five different thicknesses lie close to the straight lines indicated by the mathematical relation expressed above. See Chapter XVII. The spectrophotometric examination of colored media is valuable inasmuch as the eye not being analytic, other methods fail to reveal the true spectral character of the light emitted by the colored medium. This was demonstrated by the three yellows in Fig. 17 which appeared of the same hue (and practically of the same saturation), but differed greatly in spectral character. A spectrophotometer is an elaborate and expensive instrument, therefore where the need for such an instrument is not great enough to warrant its pur- chase, an ordinary spectrometer with modifications can be made to serve the purpose. There are various ways of converting ordinary spectrometers into in- struments satisfactory for spectrophotometric work. A double bilateral slit and a combination prism for transmitting and reflecting respectively two juxta- posed beams of light from the different sources into ANALYSIS OF COLOR 93 the collimator is a ready means of converting a spectrometer into a spectrophotometer. However, the comparison field which consists of narrow lines juxta- posed endwise is not very sr — -^zi— q_s ^ Fig. 46. — Abney*s spectrophotometric attachment for a spectrometer. satisfactory. Abney used the scheme illustrated in Fig. 46 in his early studies in color. Two slits, SS, were placed in a plane at right angles to the col- limator. One slit was be- low the other, so that their respective images could be reflected toward the coUimating lens by the two right- angled prisms which were placed one below the other. This arrangement no doubt yielded a photometric field which was not divided by an invisible line, as is desirable for high sensibility. ^^ I HZ^ -3 a- b Fig. 47. — Ives* spectrophotometric attachment for a spectrometer. A more satisfactory method is illustrated in Fig. 47. This arrangement, used by Ives,'' was designed chiefly to avoid the errors due to instruments having two collimators becoming asymmetrical and also to avoid errors due to scattered light. At 1 in Fig. 94 COLOR AND ITS APPLICATIONS 47 is placed a combination of two right-angle prisms cemented together. The face b is entirely silvered and face a silvered halfway up. A lens at 2 forms an image in the field of the telescope tube at 3 which is observed by means of an ocular lens at 4. The two light sources are placed at 6 and 7 respectively. A large monochromatic field is obtained which is equally affected by scattered light if any is present. Further- more colored glasses can be judiciously used to elimi- nate scattered light if necessary. A further improvement of the foregoing attach- ment, which was added by Nutting, ^ is illustrated in Fig. 48. The attachment consists of two reflecting prisms. Pi and P2, two Nicol prisms, Ni and N2, and a lens arranged as shown in the figure. The whole can be attached to the slit of any spectrometer. The essential factor is that a real image of the photo- metric field (the common surface of the two reflect- ing prisms) is thrown on the slit by an achromatic lens and is thus brought into the plane of the slit. The two beams of light to be compared, one passing through a portion of the photometric surface and the other being reflected by the other portion which is silvered, are brought to a brightness balance for any wave-length by rotating the Nicol prism M. High sensibility is claimed by Nutting for an instru- ment of this type. 27. The Monochromatic Colorimeter. — Colorim- 1 Ma Fig. 48. — Nutting's spectrophotomet- ric attachment for a spectrometer. eters vary in design, depending upon the data to be ANALYSIS OF COLOR 95 obtained. In some industrial processes tintometers are employed which determine the color of substances in terms of arbitrary standards. Such instruments are colorimeters, but give no quantitative analyses of the colors. Their purpose is largely to keep the product within certain limits as to color, but they perhaps serve the purpose in many of these cases as satisfactorily as a more complex instrumenti Of the instruments that analyze colors into the three terms *hue,' * satura- tion' and * brightness,' the Nutting^ colorimeter, being of the latest type, has been chosen for description. The optical system of this instrument, which has been called a monochromatic colorimeter, is shown in Fig. 49. Light entering the slit of collimator, Aj which is movable, traverses the prism and is dispersed by prism P into its spectral components, thus furnish- ing the measurement of hue of the unknown light which enters through the slit of collimator C and is reflected by a portion of ''^^•'^•"^^^^^fS?'''''^''^''^ the diagonal surface in L, which is a Lummer-Brodhun photometer cube. White light enters the slit of collimator B and is reflected by the prism face and joins a portion of the beam from A. The eye placed at the ocular slit in D sees an ordinary photometric field, the two parts of which can be matched in hue, saturation, and brightness. The hue is matched by varying the angular position of A and the saturation by varying the amount of white light added. The amounts of light entering the slits can be varied by changing the slit widths, by rotating sectors, or by rotating one of a pair of 96 COLOR AND ITS APPLICATIONS Nicol prisms placed just inside the slits. In analyz- ing a purple, for which no spectral match in hue exists, a spectral color is mixed with the unknown, the remaining procedure being obvious. The later instruments have been altered somewhat in con- struction, but the principle remains the same. The accuracy with which the dominant hue is obtainable is claimed to be about .001 to .002jLt except in the extreme regions of the spectrum, for very unsaturated colors and dark shades. Data obtained by Nutting « are given in Table V. TABLE V Materials Hue Per cent white 0.571m 48 .586 56 .680 9 .697 65 .695 70 .680 45 .580 28 .472 90 .511 56 .582 65 .597 70 .575 60 .583 61 .591 64 Reflection coefficient Sulphur Cork Dandelion Tobacco leaf (medium) Chocolate Butter, light Butter, dark Navy blue (U.S.) Paris green Manila paper Copper Brass, light Brass, dark Gold, medium 0.80 .14 .05 .64 .019 .386 .57 .23 .32 .26 .21 Data obtained by Abney ^ in the analysis of the color of glasses and pigments are presented in Table VI. In Table VII are given some data on the color of illuminants obtained by L. A. Jones ^ with the mono- chromatic colorimeter. ANALYSIS OF COLOR 97 TABLE VI Glasses Ruby Canary Bottle-green Signal-green Cobalt Pigments Vermillion Emerald-green French ultramarine blue . Brown paper Orange Chrome-yellow Blue-green Eosine dye Cobalt-blue Hue Dominant hue 0.622/z .686 .661 .4926 .610 .4676 Saturation Per cent white 26 31 32 61 42 Brightness Transmis- sion coef- ficient 0.131 .820 .106 .069 .194 .038 Dominant hue Per cent white 0.610/x .622 .472 .694 .6916 .6836 .6006 .640 2.6 69 61 60 4 26 42.6 72 66.6 Reflection coefficient 0.148 .227 .044 .26 .626 .777 .148 .447 .146 TABLE Vn Source Sunlight Average clear sky Standard candle Hefner lamp Pentane lamp Tungsten glow lamp, 1.26 w. p. c Carbon glow lamp, 3.8 w. p. c Nernst glower, 1.5 w. p. c Nitrogen-filled tungsten lamp, 1.00 w. p. m. h. c. Nitrogen-filled tungsten lamp, 0.6 w. p. m. h. c. Nitrogen-filled tungsten lamp, 0.36 w. p. m. h. c. Mercury vapor arc Heliiun tube Neon tube Crater of carbon arc at 1.8 amperes Crater of carbon arc at 3.2 amperes Crater of carbon arc at 6.0 amperes Acetylene flame (flat) 98 COLOR AND ITS APPLICATIONS In colorimetric work a standard white light is necessary. Jones used noon sunlight, which he found to be constant in color from 9 A.M. to 3 P.M., the obser- vations extending over several weeks. This light was reflected into the instrument from a magnesium car- bonate block. Many interesting studies in color-mixture can be made with such a colorimeter. An example is found in the work of L. A. Jones ^ in the analysis of mixtures of two component colors. Filters were chosen in several cases practically complementary in color. These filters were in the form of sectors of a circle and of equal angular extent. An opaque sector equal in size to one of the filters was varied in position over the sectors, so that they could be left open in any desired proportions. Lights passing through these filters were mixed in a complete range of ratios and the resultant mixtures were examined by means of a monochromatic colorimeter for hue and saturation or per cent white. For example, we will choose one of the pairs of filters, a red and blue- green of dominant hues 0.624jLt and 0.497/x respec- tively. The saturation or purity of the colors are indicated by the per cent of white light (noon sun- light) that each transmitted, these being for the red and blue-green filters respectively 3.3% and 28%. The transmission coefiicients of the two filters were respectively 24% and 16%. The data obtained in analyzing various mixtures of the two colored lights are shown in Fig. 50. It is seen that practically only two hues are obtained in a complete range of mix- tures and these are the dominant hues of the re- spective colored lights. The dominant hue of the mixtures changes abruptly from the hue of one of the colored lights to that of the other at the point ANALYSIS OF COLOR 99 near where the mixture contains the maximum amount of white light, or in other words where the two lights are nearest to being complementary. The per cent white reaches a maximum of 95% (indi- cating that the colored lights are here practically com- plementary) when the blue-green filter was open about 62% and the red filter about 38%. A con- clusion, among others drawn by Jones from this in- 100 90 80 70 60 50 40 50 20 10 / '-N, / N hv / / -"«i^ \ / — N \ y \ > y\ \ V / \, / y / I. o^ ■ 0.640 0620 ^ 0.600 ^ 0.580 5 z: 0.560 E o 0.540 .. 0.520 0500 0 10 20 30 40 50 60 70 80 90 100 COMPOSITION OF mixture: IN PER CENT. OF BLU£-GREEN FILTER Fig. 60. — Analysis of two-component color-mixtures. vestigation, is that it is not possible with two filters that are complementary or nearly so to produce mix- tures that show appreciable color of more than two dominant hues, these hues being the dominant hues of the two components of the mixture. He points out other possibilities for this colorimeter in investiga- tions in color-mixture. The author has used an arrangement diagram- matically shown in Fig. 51 for the study of various problems, chiefly that of the influence of saturation of color in heterochromatic photometry. This ar- rangement has all the essentials of a colorimeter for 100 COLOR AND ITS APPLICATIONS analyzing colors into hue, saturation, and brightness. Light from a source L emitting light of a continuous spectrum enters the collimator of a Hilger spectro- scope and is dispersed by the prism. A standard white light illuminates the non-selective ground opal glass, O, an image of which is reflected into the objective telescope from the prism face as shown. A white sectored disk, Z>, which is smoked with magnesium oxide by holding near a burning mag- nesium ribbon, is placed so that it bisects the field. 5>>..(|: / / / /PI u f- D D^ Fig. 51. — A simple method of converting a spectrometer into a combined mono- chromatic colorimeter, direct comparison photometer, flicker photometer, and spectrophotometer. F^ vertically. If the edges of the sectors are beveled and well sharpened, the dividing line can be made to disappear almost completely. The light from the unknown, JJy is reflected from the disk. By varying the intensity of the various lights the desired meas- urements can be made. The hue is determined by the position of the wave-length drum; the amount of white light can be measured by comparing with a standard at C/ or by a previous calibration. The brightness can be measured either by the direct comparison or the flicker method of photometry (# 55). The sectored disk provided with a motor drive is in reality a Whitman-disk flicker photometer. As al- ready stated the arrangement was originally devised for another investigation; however, it readily serves ANALYSIS OF COLOR 101 the requirements of a monochromatic colorimeter. Transparent colored media can be illuminated by a standard white light placed at U. Likewise opaque colored media can be placed on the disk D and illuminated by a white light. 28. The Tri-color Method. — It is well known that any color can be matched in hue by mixing the three primary colors, red, green, and blue, in proper proportions. The Young-Helmholtz theory of color vision is largely based on this experimental fact (#47). Koenig found, by a rather complex method, the rela- tive amounts of the three primary color sensations aroused by the various spectral colors and determined KO^^x Fig. 52. — Illustrating the principle of the Maxwell ' color box.' the so-called sensation curves of the eye (Fig. 54). Maxwell was one of the first to obtain quantitative data in matching colors by a mixture of three primary spectral colors. His apparatus, known as the * color box,' though somewhat more complex, involved the fundamental principle shown in Fig. 52, and is based upon the fact that an optical path is * reversible.' For example if a spectrum is formed at A by means of a collimator and a prism by light entering the slit, S, of the collimator and traversing a prism, we can obtain a patch of light of any color by placing slits, /?, G, and 5, in the spectrum and combining the light from these on a distant screen. Conversely, if the latter slits be illuminated with white light on looking into the collimator slit, S, the prism face will appear of a color which is the result of the mixture of the colors of the slits which, in the first case, were 102 COLOR AND ITS APPLICATIONS combined by means of a lens into one colored patch on the distant screen. Hence, instead of forming a spectrum at Ay and producing colored light by mix- tures of jR, G, and 5, by slits placed at these points and combining the three colored lights, Maxwell adopted the reverse process. He illuminated the three slits by sunlight reflected from a white diffus- ing surface placed in front of them, and on looking into the slit, S, he saw the prism face appear in colors corresponding to the positions and proportions of /?, G, and B. This composite color he compared with the original white light or any colored light. The Maxwell color box was actually constructed in a different manner, but the principles involved are the same as indicated above. By means of this instru- ment he obtained many color equations of the form xR + yG + 25 = C. By a similar method Koenig obtained data which resulted in the production of the so-called sensation curves of the eye. Abney has employed the method in a great deal of work in color analysis, including the study of color vision, the analy- sis of pigment colors, and of the color of illuminants. A colorimeter based upon the tri-color method of analysis was developed by F. E. Ives.^^ Instead of spectral colors, red, green, and blue colored filters are employed in this instrument, which is illustrated in Fig. 53. By means of this instrument colors are analyzed in terms of the colors of the filters /?, G, and B, These can be reduced to sensation values as shown later. Z) is a variable slit which is illumi- nated by light of the color to be analyzed and A is an optical mixing wheel consisting of twelve convex lenses arranged to rotate. By means of this wheel the various amounts of the red, green, and blue com- ponents are mixed to match the light from D. F ANALYSIS OF COLOR 103 is the field lens and C a prism or small angle which divides the photometric field by a sharp line in the middle. H is the eyepiece, 7 is a hinged front carrying the objective lens, K^ and prismatic lens L. These are unnecessary for some work and can be replaced by a non-selective ground opal glass. The procedure in making observations with this ^\^^m^/y/yy////////////////^^^^^^ ^//<r/^//////^//^//////^^^^^ Fig. 53. — The F. E. Ives colorimeter. instrument is obvious. If the colorimeter readings which are obtained from the position of the levers which control the slit widths of /?, G, and 5, be reduced to sensation values they become much more valuable. H. E. Ives ^^ has done this in analyzing the color of illuminants, by using the sensation curves obtained by Kbenig and modified by Exner. These are shown in Fig. 54. They are based upon experi- mental data which has afiforded strong confirmation of the Young-Helmholtz theory of color vision which assumes three fundamental color sensations are responsible by different degrees of excitation for 104 COLOR AND ITS APPLICATIONS the perception of all colors (#47). It is noted that each of the three supposed primary color sensations r \ Vue z: o r- / \ ^-. ^Oreen ^ / \ /' ^ s. s/5 / \ / Red^-^ ^ S. UJ > / \ 1 / / V \ ?, / \ y / \ \ LlJ a: / \ / 1 \ \ / / \\ / \ V \ N / ^^ >^ — ■ "'"»«>, "^ 0.59 042 046 045 Q5I 0.54 0.57 0.00 0.65 0.66 0.68 0.72 Fig. 64. — Koenig's sensation curves. is not excited by a limited portion of the spectrum. In fact, spectral rays in general are supposed to Fig. 55. — Tri-color colorimeter measurements. excite the sensations in relatively different degrees, depending upon the wave-length. After a prelimi- nary investigation Ives concludes that these curves are a much nearer approach to the truth than those obtained by Abney. In reducing the colorimeter ANALYSIS OF* COLOR 105 readings to sensation values it was necessary to obtain the red, green, and blue sensation values of the colorimeter screens. Spectrophotometric analy- sis of the screens combined with the data in Fig. 54 yield the primary sensation values of the screens which are obtained in relative values by integrating the areas under the sensation curves for the three screens and reducing the colorimeter readings accord- ingly. Each of the three colorimeter readings repre- sents a mixture of the three primary sensations, depending upon the primary sensation values of the colorimeter screens. The procedure is simple but more details, if desired, can be obtained from the original paper. The primary sensation values of various illuminants compared with average daylight as determined by Ives are found in Table VIII, some of which are plotted in Fig. 55. The data represent TABLE Vm Color of niuminants by Tri-chromatic Colorimeter (See Fig. 56) Source Sensation values Red Green Blue 1. Black body 5000° abs 2. Blue sky (S) Blue sky (C) 3. Overcast sky 4. Afternoon sun 5. Hefner lamp 6. Carbon incandescent lamp, 3.1 w. p. m. h. c. . . 7. Acetylene 8. Tungsten incandescent lamp, 1.25 w. p. m. h. c. 9. Nemst 10. Welsbach, I % cerimn 11. Welsbach, | % ceriiun 12. Welsbach, 1^ % cerium 13. D. C. Arc 14. Mercury arc 15. Yellow Flame arc 16. Moore carbon-dioxide tube 33.3 26.8 32.0 34.6 37.7 54.3 51.1 48.6 48.3 49.2 42.5 45.4 47.2 41.0 29.0 62.0 31.3 33.3 27.2 32.0 33.9 37.3 39.5 40.5 40.8 40.8 40.7 40.8 42.0 41.8 36.3 30.3 37.5 31.0 33.3 46.0 35.8 31.6 25.0 6.2 6.4 10.6 10.9 11.1 16.7 12.6 11.0 22.7 40.7 10.6 37.7 106 COLOR AND ITS APPLICATIONS the means of the values determined by two methods, namely colorimeter readings and likewise spectro- photometric data reduced to sensation values. (The primary color sensation values of the spectral colors and principal lines of the cadmium and mercury spectra are plotted in Fig. 31.) The dotted line represents the color of a black body (or an incandes- cent solid emitting radiation non-selectively) for tem- peratures between 3000 and 7000 degrees absolute (C). Most of the artificial illuminants lie along this curve. Those radiating selectively in the visible spectrum, such as the yellow flame arc and Wels- bach mantle, do not lie upon it. Ives concludes that the spectral distribution of energy in noon sunlight which reaches the earth's surface is quite similar to that of the black body at 5000 degrees absolute (C) as computed from radiation laws (#6). Another instrument for tri-color analysis which is extremely simple is illustrated in Fig. 56. This Green method, which has Red __• Qj^^ been applied by many in various color investi- Fig. 56. — Arrangement for using color filters before a photometer eyepiece. gations, has been used by Bloch.i2 ^ ^jgj^ ^q^- taining four circular apertures, three being respectively covered by red, green, and blue screens, is pivoted so that the various screens can be brought before the ocular aperture in a photometer head. Pho- tometric balances are made while viewing the field through the various filters separately and the re- sults are plotted on rectangular coordinates, the ratio of red to green intensities being plotted against the ratio of blue to green intensities. Bloch presents ANALYSIS OF COLOR 107 plats containing his color analysis of many illuminants and the spectrophotometric analyses of the filters are also shown. Such results can hardly be consid- ered more than approximately comparative and of limited usefulness. In general, data concerning the color of illuminants or of colored media obtained by the tri-color method of analysis are limited in usefulness, owing to the fact that the method is not sufficiently analytical. The usefulness of such a method is broader than the tintometer with its arbitrary standards of color, but the spectropho- tometer and monochromatic colorimeter as a rule yield more useful data, the former being quite analytical for spectral examination and the latter rendering data in terms of the specific qualities of a color, namely, hue, saturation, and brightness. 29. Other Methods of Color Analysis. — Many instruments have been devised for color analysis based on principles differing from the foregoing. A number of colorimeters employing colored solu- tions have been used, the measurements usually being made in terms of the depth of liquids of cer- tain concentrations. Purple and green solutions have been used by Fabry for eliminating color difference in photometry. In a sense such a procedure is a colorimetric method if it is desired to use it as such. The Kirchoff-Bunsen and Stammer colorimeters em- ploy colored solutions for the measurement of color. Leo Axons ^^ has devised a colorimeter based upon the rotation of the plane of polarization by quartz plates (#11) which have been cut perpendicu- larly to their crystallographic axes. This instrument is illustrated in /, Fig. 57. White light from a dif- fusely reflecting porcelain disk is reflected into the instrument through a circular hole, 5, and is rendered 108 COLOR AND ITS APPLICATIONS plane-polarized by the Nicol prism, P. A quartz plate at Q rotates the plane of polarization of various rays through various angles depending upon the wave-length. The beam then passes through another Nicol prism, A^ thence through the central portion of the Lummer-Brodhun photometer cube, W, and to the eye beyond R, The eye sees a circular patch of light of a certain color depending upon the thick- ness of the quartz plate and the relative angular posi- tions of the Nicol prisms. This colored patch is \/K7/I7m B lQ /t7 qQ fjry ws g) U B' 0ZZ7J jy Fig. 57. — Arons colorimeter. matched in color with the light entering the side tube, N. The latter beam is controlled in intensity by the two Nicol prisms. Pi and P2, and is reflected by the totally reflecting prism, Z>, to the photometer cube, which in turn reflects the light to the eye in a beam concentric with the first beam. The colored media are placed in front of tube Ny and are pref- erably illuminated by the same source that illumi- nates the porcelain disk in front of B. Mixed colors are obtained, the Nicol, A^ subtracting certain rays depending upon its angular position leaving the remaining light colored instead of white. Six quartz plates are provided, of thicknesses 0.25, 0.5, 1.0, 2, 4, and 8 millimeters respectively, which are mounted in brass plates. These plates have two identical ANALYSIS OF COLOR 109 holes, one covered with the quartz plate, the other unobstructed. These are arranged to slide in or out of the instrument at or near Q, By sliding any of the brass plates to the side any number of quartz plates can be arranged one after another and thus the total thickness of quartz in the path of the beam from B can be adjusted in steps of 0.25 mm. to a total thickness of 15.75 mm. A still greater variety of colors can be obtained by using two sets of Nicol prisms and quartz plates in series. Therefore the tube in / can be removed at the plane xx^ and tube // connected at the end yy. B' takes the place of B and lens V the place of L, Any thickness of quartz plates at Q' can be inserted; however, only a single plate is employed by Arons, this one being 3.75 mm. in thickness. In case transparent colored media are to be examined, a white diffusely reflect- ing porcelain disk similar to the other one is used in front of the tube N, The two disks should receive the same intensity of illumination from the same source. If opaque colors are to be examined for reflection, these are placed on the porcelain disk, and the observer sees an outer ring through R of the color of the unknown. This is matched in color and brightness by adjusting the thickness of quartz and the angular position of the Nicol prisms until the inner circle appears of the same color and brightness as the outer ring. The measurements are recorded in terms of the thickness of quartz, the angle between A and P and the angle between Pi and P^. and also between P' and P if the tube II is in use. 30. Templates. — Much of the early investigation in color was done with the rotating disks (Fig. 23) and it is quite natural that modifications of these would be made. Abney devised an ingenious method 110 COLOR AND ITS APPLICATIONS Fig. 58. — Abney's template for car- mine. for showing the effect upon the color of the integral light of various spectral energy distributions and also of showing that a certain determined spectrophoto- metric curve was in reality the analysis of an integral color. On determining the relative amounts of light of various wave-lengths reflected by a pigment these, instead of being plotted on rectangular coordinates as shown by the dotted lines in Fig. 12, were plotted in a speciar manner on a disk. Along a portion, VRy of a radius of the circle in Fig. 58, a wave-length scale is laid off. The relative amounts of light of different wave-lengths reflected from the pigment as determined by means of a spectrophotometer are laid off on circumferences of circles concentric with the center of the disk starting at a certain point of VR corresponding to the wave-length. The cardboard is now cut out along the boundary line, the template in Fig. 58 being Abney's template for car- mine. If this disk be carefully adjusted in the plane of a spectrum formed in space so that various wave- lengths along VR coincide with corresponding wave- lengths in the spectrum and the disk be rotated, on combining the colored rays passing through the rotating aperture upon a white screen by means of a lens the color of the integral light reflected from car- mine is seen. This patch will be exactly like the original color in appearance providing the optical parts of the instruments are non-selective and the same light is used in producing the spectrum as was used in illuminating the pigment when the spectrophotometric observations were made. Of course the irrational dispersion of the prism must be properly allowed for and the spectrum must be narrow. ANALYSIS OF COLOR 111 Instead of rotating the template before an actual spectrum Abney used the principle adopted by Max- well in his * color box' (Fig. 52), thus rotating the disk before a long narrow slit illuminated by the total light from the illuminant. The integral color was viewed through the eyepiece of the spectrometer. Abney made a number of these templates represent- ing pigments, illuminants, and the luminosity curve of the eye. Elevation Enp View Fig. 59. — Adaptation of Abney's scheme for the spectroscopic synthesis of color. Recently Ives and Brady ^^ applied Abney's prin- ciple to the alteration of the light from a 4 w.p.m.h.c. carbon lamp to that of * average daylight' and also to that from the blue sky. A Hilger constant-devia- tion spectrometer was used, as shown in Fig. 59. The regular camera attachment B was placed in the position ordinarily occupied by the collimator, the latter being placed in the position of the objective telescope. The slit at S is long and narrow and is illuminated by light from the carbon incandescent lamp reflected from a white surface at F, the prin- ciple being the same as just presented in the description of Abney's work on templates. These templates were computed on the assumption that the 112 COLOR AND ITS APPLICATIONS relative spectral energy distribution in the spectrum of the carbon incandescent lamp operating at 4 watts per mean horizontal candle is that derived from the Wien equation (equation 2, #6) for a black body at a temperature of 2080 deg. absolute (C), and that of white light corresponding to a temperature of 5000 deg. absolute. The templates for converting the carbon light into blue sky light were made from relative spectrophotometric measurements. The disk in position is shown in the three views of the appa- ratus taken from the work cited above. The advan- tage of using the templates before a slit illuminated by white light is that a much greater amount of light is available than in the case of using it before a spectrum and recombining the transmitted light by means of a lens. A comparison field can be arranged by reflecting the light L into the instrument as shown. Abney cut a template corresponding to the luminosity curve of the eye which is of interest, but owing to the work of various modern investi- gators this has been more accurately established. The template scheme can be applied by using disks in which openings are cut corresponding to the lumi- nosity curve of the eye and replacing the surface at F before the objective slit by a straight incandes- cent filament; thus the transmissions of absorbing media can be determined by pure energy measure- ments. 31. The Nutting Reflectometer. — In the study of color it is sometimes desirable to ascertain the reflection coefficients of colored media. This can be done if the object is diffusely reflecting by means of an ordinary brightness photometer, although the uncertainties of color photometry will be present in any case. However, Nutting ^^ has devised a simple instru- ANALYSIS OF COLOR 113 ment shown in Fig. 60 that is very useful for deter- mining the reflection coefficients of any colored media for light incident from all possible directions simul- taneously. Two crown glass prisms of 21 deg. angle are fastened over the two apertures in the end of a Koenig-Martens polariza- tion photometer and the lat- ter is inserted into a metal ring which is nickel-plated and polished inside. The light enters the apertures of the instrument along the dotted lines shown and is divided into two plane-polar- ized beams by a Wollaston prism. These beams can be balanced in intensity by rotating the Nicol prism. Fig. 60. The The Nutting reflectom- eter. surface whose re- flection coefficient is desired is placed on one side of the ring completely covering it and this is illuminated by a non-selective ground opal glass on the other side of the ring. The instrument is placed upon a wooden frame for convenience. The light is reflected back and forth between two planes of infinite' extent made practically so by the polished ring. Simple theory shows that the ratio of the brightness of the unknown to that of the ground opal glass is a direct measure of the reflection coeffi- cient of the former for the character of the illumina- tion it receives. Certain precautions must be taken into consideration as explained by Nutting. 32. Methods of Altering Brightness of Colors Non-selectively, — It is often desirable to alter the 114 COLOR AND ITS APPLICATIONS brightness of colored lights without altering them spectrally. A simple means is found in varying the distance of the light source and computing the rela- tive intensities from the * inverse square law.' How- ever, sometimes this is inconvenient. Sectored disks are often resorted to with satisfactory results. These are now being used in photometry to a great extent, the variable sectored disk devised by Hyde (Fig. 44) being especially convenient and reliable for spectro- photometry. The Brodhun variable sector is another device very often applicable. In this instrument a beam of light is rotated and is controlled in intensity by a variable stationary sector. Plate glass varied in its angular position with respect to the axis of the beam affords a means of obtaining a slight range of brightnesses, although non-selective glass is rarely found. Wire mesh and grids thoroughly blackened are satisfactory in some problems. Neutral tint wedges have been used, but it is difficult to obtain strictly non-selective smoke glass. Ives and Luck- iesh 1^ studied the transmission characteristics of half-tone gratings (black lines on clear glass) and found them to be satisfactory if properly used. Pho- tographic screens are found to serve some purposes, but they must always be calibrated in position owing to their tendency to diffusely reflect light. These are a few methods which have proved helpful in the proper place. REFERENCES 1. Bul. Bur. Stds. 1906, 2, p. 317. 2. Astrophys. Jour. 1912, 25, p. 239. 3. Trans. L E. S. 1914, p. 853. 4. Phys. Rev. 1910, 30, p. 446. 5. Bul. Bur. Stds. 7, p. 234. 6. Bul. Bur. Stds. 1913, 9, No. 187. ANALYSIS OF COLOR 115 7. Color Mixture and Measurement, p. 165. 8. Trans. I. E. S. 1914, 9, p. 687. 9. Phys. Rev. N. S. 1914, 4, p. 454. 10. Jour. Franklin Inst. July, Dec. 1907. 11. Trans. I. E. S. 1910, p. 189. 12. Electrotech. Zeit. 1913, 46, p. 1306, 13. L'Industrie Elec. July 25, 1911. 14. Jour. Franklin Inst., 178, p. 89. 15. Trans. I. E. S. 1912, 7, p. 412. 16. Phys. Rev. 1911, 32, p. 522. CHAPTER VI COLOR AND VISION 33. The Eye. — Color vision is not essential, because achromatic vision serves the totally color- blind person well. However, the ability to perceive colors extends the usefulness of the sense of sight very much. It not only adds greatly to our pleasure but is utilized in many ways. The eye can be considered optically as a rather simple instru- ment, as indicated by the photograph of the middle vertical section of a human eye shown in Fig. 61. It is seen that the refracting media consist of the cornea, aqueous humor, lens, and vitre- ous humor. The retina, which consists of the optic nerve spread out over the interior of the eyeball, is a very thin membrane, and can be seen in the illustration partially de- '1ke^;;»1;:\f:^"^'°°°' tached from the wall. The radii of curvature, thickness, and refractive index of the various eye media as determined by Helmholtz ^ are given in Table IX. The normal eye, while being a wonderfully adapt- able instrument, is not free from errors, owing to the fact that it is optically quite simple. The chief error of interest here is its lack of achromatism. If an image of an object illuminated by light having a continuous spectrum be produced by a simple lens 116 COLOR AND VISION 117 TABLE IX Optical Constants of the Eye Index of refraction of the humors and cornea Index of refraction of the crystalline lens Effective index of refraction of lens surrounded by humors Radius of outer stirface of cornea Radius of first lens surface Radius of second lens surface Thickness of cornea Thickness of crystalline lens Distance of first lens surface from cornea Distance of second lens surface from cornea Distant 1.3365 1.4371 1.0753 7.8 mm. 10.0 6.0 0.4 3.6 .3.6 7.2 Near vision (15 cm.) 7.8 mm. 6.0 5.6 0.4 4.0 3.2 7.2 it will be found to have a red, blue, or purple fringe. This is readily understood from Fig. 62, which repre- sents a schematic eye in which only the simple lens is considered. Owing to the difference in the refrac- tive index of a medium for rays of different wave- length, such a result as is exaggerated in Fig. 62 will obtain. The refractive index being greater for rays of shorter wave-length, the blue rays will be deviated or refracted more than the yellow rays, and the latter more than the red rays. Naturally the eye focuses for the brightest rays, which in ordinary light are the yellow-green or yellow rays. There- fore, the blue and red rays will be out of focus, with the result that the image of the point, P, will be surrounded by a purple fringe. This is of im- portance in vision, as will be shown later. The lack of achromatism of the eye can be demonstrated very simply. On viewing, by reflected light, the concentric circles shown at the right of Fig. 62 held close to the eye they appear colored. A very striking experi- ment is found in focusing a line spectrum — that of mercury will suflBice — upon a ground glass. On 118 COLOR AND ITS APPLICATIONS viewing it at a normal distance (14 inches), the yellow and green lines will appear sharply focused, but the blue and violet lines will appear hazy and quite out of focus. On bringing the eye closer the latter lines will begin to appear clearer, and finally, when the eye is within about six inches of them, they will still appear clear-cut, while it will be quite impossible to accommodate the eye sufficiently to focus 'the yellow and green lines. In other words the eye is near- sighted (myopic) for blue rays and far-sighted (hyper- opic) for red rays. On viewing a narrow continuous spectrum at some distance the blue end appears to Fig. 62. — Showing the effect of chromatic aberration in the eye. flare out. Another simple demonstration is found in viewing an illuminated slit through a dense cobalt glass which transmits extreme red and violet rays. On accommodating the eye for a point behind the slit a red image with a violet halo is seen. On accommo- dating for a point in front of the slit a violet image with a red halo is seen. This defect plays a promi- nent, though usually unnoticed, part in vision. A lens can be made practically achromatic by combining a convergent lens of crown glass with a divergent lens of flint glass. The former is more strongly conver- gent for blue than for red rays, while the latter is more strongly divergent for blue than for red lights. It is thus possible to bring the red and blue rays in coincidence at a focus. Inasmuch as it is only pos- sible to bring two rays exactly into coincidence by COLOR AND VISION 119 a two-piece lens, such a lens is not truly achromatic, though practically so for most purposes. By com- bining more lenses the approach to true achromatism is brought as near as desired. A simple achromatic lens is illustrated in Fig. 63. The retina has been found to vary in its sensibility to colors. The central region is sometimes known as the yellow spot, be- cause it apparently absorbs the Fig.63.-A simple achromatic violet and blue rays to a greater degree than other rays. The effect of the yellow spot is often seen in viewing colors one after another, and it is quite noticeable at twilight illumination. It appears of somewhat irregular outline in after- images. Studies of the various zones of the retina as to their sensibility to various colors yield results in general similar to those shown in Fig. 64. The center of the fovea corresponds to the center of the circle. The solid line shows the boundary for the perception of light. The visual field for one eye extends outward about 90 deg. from the normal optical axis of the eye, inward about 60 deg., down- ward 70 deg., and upward 50 deg. The dashed line represents the extreme limits where blue can be perceived as such and the remaining two lines repre- sent respectively the limits for red and green per- ception. These facts must be reconciled with any satisfactory theory of vision. It might be noted here that each eye has a blind spot — the point of entrance of the optic nerve — which is totally insensitive to light. The retina, which consists of the optic nerve spread out, is covered with a mass of microscopic * rods' and 120 COLOR AND ITS APPLICATIONS 'cones' (#48) projecting outward toward the lining of the eyeball which play an important part in theories of vision. 34. Brightness Sensibility. — The sensibility of the retina to brightness differences is greatest over a wide range of intensities, falling off at extremely Co/or/ess Blue .. Red - Oreen Fig. 64. — Limits of the visual field for colored and colorless lights. low and extremely high brightnesses. With decreas- ing intensities the sensibility diminishes more rap- idly for rays of longer wave-length than for those of shorter wave-length. Kbenig and Brodhun ^ have done excellent work in this field as well as in many other fields pertaining to vision. They determined the least perceptible brightness increment for lights COLOR AND VISION 121 of various colors including white, for brightnesses of a neutral tint surface (* white') illuminated to various intensities from 1,000,000 meter-candles to nearly the ^ 1-0 UJ z: X 0.8 is 04 ujz: § 0.2 2: w \ "A ^ \ ■ 0 \ 1 ■ "«^ V ^ X ^ N ^^ -4 -3 -2-10 1 2 5 LOGARITHM OF ILLUMIMATIOM.! Fig. 65. — Brightness sensibility data. (See Table X.) threshold of vision, using an artificial pupil of 1 sq. mm. area. They started at 600 meter-candles and extended the illumination above and below by the various steps indicated in the accompanying table. The data for Kbenig's eye after modification by Nuttings are shown in Fig. 65 and Table X. Koenig and Brodhun did not include the increment (bB) in the total brightness (J5) in calculating the values bB/B. Nutting recomputed the data with the thresh- old value included. It is seen that the increment of brightness difference just perceptible, increases as the brightness decreases and more rapidly for the rays of longer wave-length. At high illuminations the minimal perceptible increment is about the same (1.6%) for all colors, including white. For the ordi- nary range of brightnesses 55/5, is constant, which fact is known as Fechner's law, and the constant is called Fechner's coefficient. 122 COLOR AND ITS APPLICATIONS TABLE X Data of Koenig and Brodhun on Brightness Sensibility as Recalculated by Nutting Wave- 1 length = = 0.670m 0.605m 0.575m 0.505m 0.470m 0.430m Bo = = 0.060 0.0056 0.0029 0.00017 0.00012 0.00012 Meter SB Candles B 200,000 0.0425 100,000 0.0241 0.0325 50,000 0.0210 0.0255 0.0260 20,000 0.0160 0.0183 0.0206 0.0195 10,000 0.0156 0.0163 0.0179 0.0181 5,000 0.0176 0.0158 0.0166 0.0160 2,000 0.0165 0.0180 0.0180 0.0176 0.0180 1,000 0.0169 0.0198 0.0185 0.0184 0.0167 0.0178 500 0.0202 0.0235 0.0180 0.0194 0.0184 0.0214 200 0.0220 0.0225 0.0225 0.0220 0.0215 0.0246 100 0.0292 0.0278 0.0269 0.0244 0.0225 0.0246 60 0.0376 0.0378 0.0320 0.0252 0.0250 0.0272 20 0.0445 0.0460 0.0385 0.0296 0.0320 0.0345 10 0.0655 0.0610 0.0582 0.0362 0.0372 0.0396 5 0.0918 0.103 0.0888 0.0488 0.0464 0.0494 2 0.1710 0.167 0.136 0.0656 0.0716 0.0600 1 0.258 0.212 0.170 0.0804 0.0881 0.0740 0.5 0.376 0.276 0.208 0.0910 0.096 0.0966 0.2 0.332 0.268 0.110 0.127 0.116 0.10 0.396 0.133 0.138 0.137 0.05 0.183 0.185 0.154 0.02 0.251 0.209 0.223 0.01 0.271 0.189 0.249 0.005 0.325 0.300 0.312 0.002 0.369 The value of the minimal perceptible increment depends largely upon the method of making the measurements. Usually the brightness of one of the two parts of the photometric field is varied until it appears just perceptibly brighter or darker than the comparison field. This procedure yields values of COLOR AND VISION 123 the least perceptible increment comparable with the foregoing value. In precision photometry the accu- racy is often as high as 0.1 per cent; however, another factor enters into such procedure. The brightness of one part of the field is varied between certain limits at which it is respectively distinctly brighter and darker than the comparison field, and these limits are gradually brought nearer together until finally an attempt is made to estimate the middle point. This cannot be considered a measure of brightness sensibility. However, P. W. Cobb has employed a method which is of considerable interest here inas- much as he obtains values for the minimal percep- tible increment for white light smaller than 0.5 per cent. In these experiments the test field was exposed to the view of the observer for a brief, but constant, period, after which his judgment was recorded. One side of the field appeared either brighter or darker, or no difference in brightness was distinguishable. This procedure was repeated for a range of aspects of the test field varying from that in which one side appeared distinctly darker for a number of succes- sive exposures to that in which it appeared definitely brighter. Obviously, by progressing in small steps between these two limits (presenting these various aspects in haphazard order) there were several near equality where the judgment was uncertain. After reducing the data by a special method Cobb con- cludes that the minimal perceptible increment is much smaller than that obtained by Koenig and Brodhun. The data of Koenig and Brodhun has been ex- tended by Nutting by computation to the point where 8B/B = 1; that is, to the threshold value. This computation is very interesting, though perhaps not 124 COLOR AND ITS APPLICATIONS entirely free from criticism. Bq in Table X repre- sents the threshold value of brightness measured as a fraction of the standard high brightness. Brightness Bj is proportional to illumination, /, and inasmuch as it is a brightness that is perceived the symbol B is used. 35. Hue Sensibility. — Notwithstanding the fact that the visible spectrum is generally considered to exhibit only six or seven colors, four of which, red, yellow, green, and blue are strikingly distinctive, there are theoretically present an infinite number of hues. The number of distinct hues that a person is able to distinguish depends upon the manner in which the experiment is conducted. Edridge-Green ^ states that he has 'never met with a man who could see more than 29 monochromatic patches in the spectrum.' Rayleigh,^ who is able to detect the dif- ference in hue of the sodium D lines (0.5890m and 0.5896/x), could distinguish only 17 hues on Green's apparatus, and claims this is due to the method of comparing the patches. In Green's apparatus the principle is that of two opaque screens held over a spectrum and slightly separated from each other. One is then moved until the hue at its edge appears different from that at the edge of the other. With an apparatus employing the principle of the Maxwell color box Rayleigh was able to distinguish many more hues. By the use of spectral apparatus as high as 128 distinctly different spectral hues have been seen. It is not difficult to obtain by the use of dyed media a series of 25 distinct spectral hues. Ridgeway,^ by beginning with papers dyed to represent six spectral hues and adding various intermediate hues, obtained 36 distinct hues. The data on hue sensi- bility vary considerably, which perhaps is due to COLOR AND VISION 125 variations in the refinement and nature of the experi- mental methods employed. Some excellent data have been obtained by Steind- ler 7 on hue sensibility for twelve subjects. The posi- tions of the maxima differed somewhat for the various ^ / 00 /^ / 1 5 f \ / \ / \ / \ X tiJ / \ ) \ ^ / / \ UJ f \ / V. r \ 1 V y \ \ 040 0.44 046 052 Q56 060 ^. "HML LENGTH Fig. 66. — Hue sensibility. 0.64 (Steindler's Eye.) observers. The hue sensibility curve for Steindler's eye is shown in Fig. 66 and the mean positions of the maxima and minima of the hue sensibility curves for the twelve observers and the wave-length limen of *just perceptible difference' are given in Table XI. Nutting 8 has used the mean results obtained by Steindler in deriving a natural scale of color. These mean results, including Nutting's color scale, are plotted in Fig. 67. The hue sensibility curve, S, was plotted by connecting the mean positions of the minima and 126 COLOR AND ITS APPLICATIONS TABLE XI Steindler's Data on Hue Sensibility (The mean for twelve eyes) Position Perceptible limen First rnaximuTn 0.455/i 0.534 0.621 0.440 0.492 0.581 0.635 0.0293m 0.0334 Second TnayiTniiTn ... Third maxitrmm 0.0375 First minimiiTn , 0.0247 Second rninitntim 0.0136 Third rniniTnutn 0.0139 Fourth TninitTiutTi 0.0300 maxima for the twelve observers with smooth curves. The limen (least perceptible difference in terms of m) curve, L, is plotted in the same manner. For the 24 20 16 U ^N - C^^ _.-J~^: V V ^ IXT, X^ %^^'^^^T^-^ ^■'u^^ \ ixi: t \ ^^A ^t A 7kj~ z^^- ks/ VV .. /- ^^^^; ^^ 0.044 0.040 -^ 0.036 Sv 0.032 t 0.02.8 ^ 0.024 5 0.020 ^ 0.01 & < 0.012 g 0.008 3 0.00.4 .0 0.42 0.4& 0.50 0.54 0.5a 0.62 0J&&" ,My, WAVE LENGTH Fig. 67. — Hue sensibility, limen, and color scale. details of the procedure adopted in obtaining the color curve, C, the reader is referred to the original paper. A difference of one unit in the color scale represents a difference in color that is just easily perceptible. It will be noted that the color curve COLOR AND VISION 127 indicates there are 22 of these colors *just easily perceptible' within the spectral limits shown. 36. Saturation Sensibility. — The data on the sensibility of the eye to changes in saturation are not very extensive or definite. Nutting ^ states that with his monochromatic colorimeter the probable error in the *per cent white' observations on a nearly spectral matte orange pigment was about ten per cent. L. A. Jones ^° claims an accuracy of the order of three per cent for the *per cent white' readings (Table VII) for this monochromatic colorimeter of improved type. The accuracy of course will vary with the hue, brightness, and degree of saturation of the colors. H. Aubert ^^ determined the smallest sector of color that would be just apparent on a rotat- ing white disk to be 2 or 3 degrees — less than one per cent. With black and gray disks he found that even smaller sectors were recognized. His experi- ments on the differential limen of color sensitivity indicated that on a black background the stimulus- increments for orange, blue, and red were respectively 0.95, 1.54, and 1.67 per cent in order to produce a just noticeable increase in saturation. Geissler 12 studied the problem whether the number and sizes of the colored stimulus-increments corre- sponding to just noticeable saturation differences would lend themselves to a measure of saturation. The problem was attacked from two extremes; one by gradually reducing a maximally saturated pigment color, and the other by introducing more and more color into a colorless stimulus. He employed the rotating double color disk with the Zimmerman col- ored and gray papers illuminated with an artificial daylight devised by Ives and Luckiesh. In the first method he used red beginning with maximal sat- 128 COLOR AND ITS APPLICATIONS uration — 360 degrees of red — for both the inner and outer concentric components of the double disk and gradually added small amounts of gray (of the same brightness as the red as measured with a flicker photometer) to the inner or smaller disk until it appeared just perceptibly less saturated than the outer or larger disk. This procedure was then re- versed, the outer disk being decreased in saturation until the change was just perceptible as compared with the inner disk whose saturation was kept con- stant. This was done for seven different degrees of saturation, ranging from 360° of red to 110° of red plus 250° of gray of the same brightness as measured by the flicker photometer. His results indicate that the stimulus-increments corresponding to just notice- able saturation-differences are approximately con- stant (about 4° of gray) at such different stages of saturation as 325° red plus 35° gray, 230° red plus 130° gray, and 110° red plus 250° gray. Geissler states that *it seems fair to assume that the incre- ment-values would have remained constant at the intervening stages and perhaps also at a stage not far removed from the absolute color-limen,' which latter averaged for the four observers with the red paper about 1.2°. That is, a sector of 1.2° of red when mixed with 358.8° gray causes a just percep- tible appearance of color. It appears from the fore- going that the estimated number of least perceptible differences in saturation of the red pigment under the conditions of the experiment is about 100. Another group of experiments was made with nine observers using red, yellow, green, and blue colored papers and their corresponding grays. These measurements were made for each eye separately and for binocular vision. Geissler places no great COLOR AND VISION 129 emphasis upon the absolute values of the results because of the lack of sufficient observers and the incompleteness of the investigation at present. How- ever, it is of interest to give the mean results for the nine observers. The averages for binocular vision were, as a rule, lower than for monocular vision. The results for all observers for monocular and binoc- ular vision gave as the mean limenal values of color saturation for red, yellow, green, and blue respec- tively, 2.23°, 5.81°, 7.19°, and 2.99°. That is, these values represent the smallest increments required to distinguish between * color and no color.' The comparison was made between a gray disk and a concentric disk of the same gray in which the color was introduced. The brightnesses were previously equated by means of a flicker photometer. The colored papers differed from each other in brightness and saturation, which appeared to have an influ- ence on the values of just perceptible saturation-dif- ference. Since the green requires a limen three times as great as that of red it appears to Geissler that it is reasonable to assume that its saturation is only one-third as great as the red and about one- half that of the blue. These figures agree approxi- mately with a number of estimates of saturations made by some of the observers, but in the absence of sufficient data little emphasis is given to this point. Experiments with a practically color-blind subject indicated that his limenal values were extremely high, being 37°, 18°, 140°, and 8.25° respectively for the red, yellow, green, and blue papers. No analysis of his defect was made. There appears to be a need for a further explora- tion in this interesting field. 37. Visual Acuity in Lights of Different Colors. — 130 eOLOR AND ITS APPLICATIONS As has already been shown the eye is not achromatic; that is, rays differing in wave-length do not come to a focus at the same point, with the result that the image of an object illuminated by light of extended spectral character is not sharply defined upon the retina. Louis Bell ^^ compared the acuity of the eye or its ability to distinguish fine detail in tungsten and mercury arc lights and obtained results indi- cating an advantage for the latter illuminant. This he attributed to the more nearly monochromatic light emitted by the mercury arc. It will be remem- bered (Fig. 4) that the preponderance of visible rays is confined to a rather narrow wave-length range in the yellow and green regions of the mercury spectrum. The author ^^ verified these results and extended the investigation to lights of the same color but dif- fering in spectral character. By using the lights whose spectra are shown in Fig. 17, no difficulties of color photometry were encountered. Screens 6, c, dj used with a vacuum tungsten lamp operating at 7.9 lumens per watt yielded lights of the same yellow color but of different spectral character. Like- wise screens e and /yielded two green lights, one purely monochromatic (mercury green line), and the other a green of extended spectral character. The data, except in case 4, Table XII, were not obtained as usual by using fine detail at the limit of discrimi- nation but instead, in terms of equal * readability' of a page of type, which proved after some practise to be a rather definite criterion. Some such method should be applicable to many practical investiga- tions in lighting, for it renders results in terms of a criterion which, although apparently indefinite, is found to be quite definite and one which renders results full of significance. The results for the COLOR AND VISION 131 TABLE Xn Relative Illumination for Equal Readability Case Source Screen Color Approx. foot candles Relative illimiination 1 Mercury arc Tungsten lamp f e green line green 2.0 1.00 1.75 2 Tungsten lamp Ttmgsten lamp d c yellow yellow 4.0 • 1.00 1.33 3 Sodium lines Timgsten lamp none c yellow lines yellow 0.5 1.00 1.66 4 Mercury arc Tungsten lamp f e green line green 0.6 1.00 5.10 author's eye are shown in Table XII and are given in terms of the relative illumination required for equal readability of a page of type. In case 4 an acuity object proposed by H. E. Ives ^^ and devel- oped by P. W. Cobb was used. Here the criterion was the ability to perceive fine lines at the limit of discrimination. Other observers obtained results of a similar nature with the same apparatus. No stress is laid upon the accuracy of the absolute values, but it is conclusively evident that monochromatic light is superior for discriminating fine detail. Later it was shown, 1^ as was expected from the foregoing, that monochromatic light was superior to daylight for discriminating fine detail. In this case the Ives acuity object was viewed against a white magnesium oxide surface which was illuminated to an intensity of 10 meter candles (approximately one foot candle). The visual acuity on the Snellen scale was found to be 1.28 and 1.11 respectively for daylight, and mono- 132 COLOR AND ITS APPLICATIONS chromatic green light of equal intensities and results for tungsten light and daylight were practically iden- tical. Another experiment showed that for visual acuity of 1.28 on the Snellen scale the intensity of illumination with daylight or tungsten light was nearly three times that required for the same visual acuity with monochromatic green light. As the brightness of the background was increased it ap- peared that the difference in visual acuity under a given illumination of tungsten light and monochro- matic light decreased. The superior defining power of monochromatic light having been demonstrated, it is of interest to learn if there is any difference in the defining power of monochromatic lights of different colors. Dow ^^ measured visual acuity in light of different colors using electric lamps screened with colored media and arrived at the conclusion that the blue-green region of the spectrum showed greater defining power. Ashe ^^ used red, green, blue and clear glasses with incandescent lamps and found visual acuity least for the red and increasing in the order green, blue and clear glass for the same illumina- tion; however, the data were too incomplete to war- rant any definite conclusions. Loeser ^^ used red, green, and white papers on which black characters were printed. The papers were brought to equal brightness and visual acuity was determined by noting the greatest distance at which the observer could distinguish the details on the papers. He found acuity greater for green light than for red light, and also that the characters on the white card could be distinguished at nearly as great a distance as those on the green card. A serious defect in this method is the fact that, the distances not being constant, the COLOR AND VISION 133 change required in the accommodation of the eye complicates the results. Uhthoff ^o determined visual acuity in monochromatic lights of different wave- lengths, but gives no data on the relative brightnesses of the colored lights. A serious defect in most of the above work is the fact that the lights were neither monochromatic nor did their spectra extend over equal ranges of wave-lengths. The same criticism is applicable to the work of Rice,^! who performed an extensive investigation of the problem. In order to determine visual acuity in mono- chromatic lights of different colors at ordinary bright- ^fC eld ■0 n j^ Q> -%^ \gnmrTv/ Fig. 68. — Apparatus for determining visual acuity in monochromatic lights. nesses, the author 22 devised the apparatus shown diagrammatically in Fig. 68. The lines of the acuity object/^ c, having a highly illuminated ground glass background, cf, were focused crosswise on the slit of a Hilger wave-length spectrometer by the lens, /. On looking into the eyepiece these lines were viewed against a background whose color depended upon the position of the prism, the wave-length being indicated on the drum, n. On the pointer in the eyepiece was mounted a minute piece of magnesium sulphate, mmj at an angle leaning away from the eye at the top. This was illuminated by means of the frosted tung- 134 COLOR AND ITS APPLICATIONS sten lamp, y, the light being reflected downward by the mirror, o. Slides hh controlled the width of the photometric field, and an artificial pupil, &, was placed in front of the eyepiece. The drum, k^ controlled by means of a belt the size of the lines of the test object which was read from drum e. The photo- metric balance was made, in the case of each mono- chromatic light used, by balancing it against the white surface mm^ the lines of the acuity object at the time being too small to be visible. A feature of this acuity object which is essential for such a use is that the average brightness of the object is constant regardless of the width of the lines. Of course in making the photometric balance the un- certainties of color photometry are present, but these are not of much importance in this investigation, because visual acuity changes very slowly with change in brightness of the object at the illumination used; therefore, a large error in the photometric meas- urements would cause but a slight error in the visual acuity measurements. The brightness of the photo- metric field as seen by the eye through the artificial pupil was equivalent to the brightness of a white surface illuminated to an intensity, of 4.2 foot candles. After the photometric balance was made by varying the current through the large lamp illuminating the test object, the lamp, 7, was extinguished and a series of acuity settings were made by varying the size of the lines. The results obtained are shown in Fig. 69. Curves a, c, represent extreme series made by the author showing the fluctuation in the ability of the eye to distinguish fine details, and h is the mean curve of a great many observations. Curves d and e represent single series of observations (ten read- ings at each point) made by two other observers. COLOR AND VISION 135 In every case the observer was permitted to focus the instrument. These data indicate an advantage in the defining power of monochromatic yellow light over other monochromatic lights of equal brightness. In order to extend the observations into the violet end of the spectrum, the test object was illuminated by means of a mercury arc. The mean results for each of two observers are shown in Fig. 70, for three mercury lines. Curve E was combined with curve b in Fig. 69 (obtained by the same observer) which 0-40 0.44 0.4a 0.52 Q56 0.60 0.64 0.65 Fig. 69. — Visual acuity in monochromatic lights of equal brightness. extended the latter as indicated. This investiga- tion indicates that monochromatic lights differ in their defining power and that yellow monochromatic light is superior to others in this respect. It was also found that for a given change in brightness of the test object the change in visual acuity was least for yellow monochromatic light than for light of any other spectral hue. A striking experiment illustrating the effect of spectral character of light on visual acuity is given below. The test-object was viewed through an ethyl violet screen (purple under the illumination from a tungsten lamp) and visual acuity settings were made. After obtaining the mean of a series of observations 1.4 ^^ ^ _r^ — ■ \s. s. ^'l ^ .^^. ^-^ \ "*-*- \ ^ a b I 2 t-^ — x^' y \ k N d JO .-" \ -^. e 136 COLOR AND ITS APPLICATIONS a yellow screen was also placed before the eye. This screen absorbed the blue and violet rays transmitted by the purple screen, thus reducing the illumination at least 50 per cent. Notwithstanding this reduction in illumination visual acuity noticeably increased. In place of the yellow screen was now substituted a blue screen which absorbed the red rays transmitted 1.4 I.-5 1.2 3.1.1 < < I 1.0 > 1-^ J .^ ■ ^' -<, X / ^ y^ y ^ / y 1^ ^ x^ y /' ,^ ^ y ■ • y y V / V ^' ::S: 1 V ' ■% -J ^ AD- 1 II 042 0.46 0.50 0.54 0.58 Fig. 70.' — Visual acuity in the mercury spectrum, the lines being reduced to equal brightness. by the purple screen, the resulting light being blue. Again visual acuity increased, notwithstanding the reduction in brightness. This experiment strikingly demonstrates the influences of chromatic aberration and spectral character of light on the ability to dis- tinguish fine detail. It is interesting to note some results on the legi- bility of colored advertisements. Le Courrier du Livre ^3 reported the legibility of various combinations COLOR AND VISION 137 for reading at a considerable distance, the most leg- ible print being black on a yellow background. The order of merit was found to be as follows: 1. Black on yellow 8. White on red 2. Green on white 9. White on green 3. Red on white 10. White on black 4. Blue on white 11. Red on yellow 5. White on blue 12. Green on red 6. Black on white 13. Red on green 7. Yellow on black It is noteworthy that in this list the customary black-on-white combination is sixth on the list. These results are interesting, although perhaps not final, owing to the many variables that enter such- a problem. 38. Growth and Decay of Color Sensations. — Many investigators have studied the problem of the effect of time of exposure and intensity of the stim- ulus on the growth and decay of luminous sensations. It has been noted (# 14) that colors are seen on rotat- ing, at a proper speed, aTdisk composed of black and white sectors. It appears that this is due, in part at least, to the difference in the rate of growth and decay of the various color sensations excited by white light. Of the work in this field, that of Broca and Sulzer24 is especially comprehensive. They com- pare the brightness of a white screen illuminated by a light of short duration with that due to a standard steady light. Some of their results which are plotted in Fig. 71 show that, excepting for lights of low inten- sity, the luminous sensation * overshoots' its final value; that is, the maximum luminous sensation is passed a comparatively short time after the begin- ning of the exposure and that the luminous sensa- tion reaches a steady value less than the maximum 138 COLOR AND ITS APPLICATIONS only after the elapse of an appreciable fraction of a second (depending more or less upon the intensity). The numbers on the curves indicate the final steady value of the various stimuli. Their data obtained with colored light, plotted in Fig. 72, indicates that under the stimulation of blue rays the luminous 80 75 y 70 o 5 65 O i 60 lij 13 50 jz S 45 I 40 35 i 50 u. <^ 25 , ,. i \ 1 \. \ \ r N — 122^ _ / ^ /PfT / r ^_ ^ \5Li J26 ^ y^ J^^J ux /x -m ^ . /6.2 Lux ^ 1 ""^^ 005 0.1 0J5 0.20 SECONDS 0.25 0.3 Q35 Fig. 71. — The growth and decay curves for white light sensation. (Broca and Sulzer.) sensation overshoots very much more than in the case of red or green light, the latter showing the least overshooting. In studying the growth and decay of color sensa- tions in connection with the flicker photometer ^^ some data of interest here were obtained. Red and blue- green lights, practically complementary, were matched by the ordinary direct comparison method of photom- COLOR AND VISION 139 etry. These were then separately flickered against darkness by means of a rotating disk with equal open and closed sectors. The maximum brightness of the flickering light was compared with a steady brightness of the same color for a large range of flicker frequencies. The data is shown in Fig. 73, 0.25 UJ g 100 UJ tl 0 J— ■ m - / /^ Gn U SI -05 S^ u r 25 26 ' ^~ . 0.05 0.1 0.15 StCONDS 0.2 0.25 Fig. 72. — The growth and decay curves of color sensations. the initials R and G representing the red and blue- green lights and the subscripts, high and lower in- tensities. The intensities used were those ordinarily considered satisfactory in photometry as is indicated by the frequency in cycles per second required to cause flicker to disappear. It will be noted that the colored lights were alternated against darkness, the steady values of the colored lights (sectors open) as 140 COLOR AND ITS APPLICATIONS determined by the direct comparison method being represented by unity on the relative brightness scale. The flicker of Gl, /?l, C^h? and i?H completely dis- appeared at frequencies corresponding respectively to Aj B, C, and D. Next red and blue-green brightnesses equivalent to the foregoing were placed so that on one side of 1.4 1.2 1.0 0.8 a: 0.6 0.4 \ Y^M \ 0, \ y ■ L Vi K '\ V V v^. ^^\ 1 ^^ s. •^ V ^^> -> ^^^* t B 1 "D" 1 2 4 6 8 10 12 14 16 18 20 22 24 2& FLICKER FREQUENCY (CYCLES PER SECOND) Fig. 73. — Showing the maxima attained by flickering lights at various frequencies. the photometer field a red light flickered on a steady blue-green field and vice versa on the other side. This was done by means of identical sectored disks (180° opening) placed one on each side of the photometer. On one side a disk intercepted the blue-green light and on the other the red light was intercepted. On increasing the speed of rotation of the disks (which were fastened to the same shaft) the side on which blue-green light flickered upon a steady red field became quiescent long before the flicker COLOR AND VISION 141 disappeared on the other side. At all times when flicker was visible the side upon which red flickered on a steady blue-green field appeared to attain higher maximum values of brightness and to be more agi- tated. The brightnesses on either side were later 24 '\ / / \ 22 / 1 / UJ 1 1 5 1.8 1 1 \ CD 1 1 \ r gl.6 >< 1 \ 1 1 \ uJ 1.4 1 1 \ 1/ \ \ f \ \ \ 1 \ \ \ 1.0 1 1 > \ \ S \ X N^ Q6 N s \ ^-%. ,^ ^"^'t^'*^- 5 12 16 20 FLICKER FREQUEIiCY 24 2& Fig. 74. — Showing the maxima of sensations produced by flickering red light on a steady green field (R), and vice versa (G). measured separately against a steady white light (there being little color difference excepting at low speeds) throughout a wide range of frequencies. These results are shown in Fig. 74, R and G indi- cating that red and blue-green were respectively the flickering components. The steady value reached at 142 COLOR AND ITS APPLICATIONS a high frequency is 0.75, unity being taken as the steady value at zero speed with the sectors open. The latter intercept only one of the two components which make up the brightness on either side; there- fore, the sectors being of 50% transmission, the final value at a high frequency of alternation is 0.75 of the original steady value with sectors open. Of course these experiments involve the measurement of the brightness of sxxrfaces differing in color, but it is this problem that was involved in the study. All steady brightnesses were chosen equal as measured by an ordinary direct-comparison photometer. While these effects of different rates of growth and decay of color sensations are operative when there is an apparent flicker, evidence points to the disappearance of such influence upon the brightness of a mixture of colored light by alternately presenting the colored stimuli when the rate of alternation is so high that flicker has disappeared. For instance the foregoing red and blue-green lights were mixed by alternating them by means of a sectored disk (50% opening) and also by directly superposing the steady lights. The former mixture was found to be just one-half as bright as the latter, within the slight possible errors of the experiment. There was no color difference present in this experiment so the photometric data is correct to within one per cent. Other evidence of the same kind was obtained by comparing two yellow lights of the same hue, but differing in spectral character, by means of both the flicker and direct comparison methods of photometry. Identical results were obtained by the two methods. These results were also confirmed by comparing tungsten light by the two methods with a light of the same hue consisting of red and blue-green lights. (See # 55.) J COLOR AND VISION 143 Talbot 2^ long ago expounded the law that a sectored disk rotating at high speed transmitted light in direct proportion to the angular openings of the sectors. This law has been stated by Helmholtz^^ as follows: *If any part of the retina is excited with intermittent light, recurring periodically and regularly in the same way, and if the period is sufficiently short, a continuous impression will result which is the same as that which would result if the total light re- ceived during each period were uniformly distributed throughout the whole period." Plateau,^^ Kleiner, ^^ Weideman and Messerschmidt,^^ Ferry,^! Lummer and Brodhun,32 Aubert,^^ Hyde,^^ and others have inves- tigated the problem, and have generally agreed that the law holds for white light. Fick concluded that it holds only at moderate intensities and Ferry veri- fied the law for white light but found discrepancies when one side of the photometer field was bluish as compared to the other side. Hyde, after a thorough investigation of the problem, concluded that the law holds within the accuracy of the work (about 0.3%) for the range of sectors used by him, namely from 288° to 10° in opening. He further concluded that the law held for red, green, and blue lights within the accuracy of precision photometric apparatus, and found that when a color difference existed on the two sides of the photometer field no appreciable deviation from the law was observed. The author has had many opportunities to test the law for colored lights and found no deviations within the accuracy of the experimental work, which was usually well within one per cent. The sectored disk, there- fore, affords a means of altering the intensity of colored light in definitely measurable amounts. Lights of very short duration are perceptible if 144 COLOR AND ITS APPLICATIONS intense enough. For instance, a lightning flash as short as one-millionth of a second is visible and by rotating mirrors flashes of light as short as one eight- millionth of a second have been perceived. Blondel and Rey ^^ studied the perception of lights of short duration at their range limits. Bloch ^^ had pre- viously contended that the excitation necessary for the production of the minimum sensation was per- ceptibly constant and proportional to the product of the brightness and the duration. Charpentier ^^ veri- fied the law within certain limits. Blondel and Rey conclude that Bloch's law can be applied only to intense lights of very short duration. After a very extended investigation they deduce a simple law, (B — Bo) t = aBoy where Bo is the minimum per- ceptible brightness of the field, t the duration of the stimulus in seconds, and a is a constant of time equal to 0.21 second. They show by simple integration one can deduce from the law of the flashes which are not uniform, their range and the intensity of the equivalent constant light from the point of view of range, fhdt a + U-ti where 4 represents the photometric intensities of the luminous points measured in a horizontal section of the beam and referred to unit distance. They conclude by taking into consideration the curves of sensation of Broca and Sulzer ^^ *that the maximum utilization of a source of light must demand short flashes without its being necessary to take any notice of an inferior limit of the period of the signals, except in the case of telegraphic signals. It more- COLOR AND VISION 145 over suffices that the period of the flash, ^2-^1, should become a negligible quantity in the presence of the constant a, in order that a maximum efficiency may be assured.' On alternating a given brightness with darkness by means of a sectored disk with 50% openings, a violent flicker is evident at low speeds; however, there is a certain minimum frequency, called the critical frequency, at which the flicker just disappears. The critical frequency depends upon the intensity of illumination or brightness of the observed field and increases with the brightness. Porter ^s has found i LOGARITHM OF BRIGHItlESS Fig. 75. — Showing the relation between brightness and critical frequency for colored stimuli. that the relationship, / = a log / + 6, holds for white light where / is the critical frequency, /, the illumi- nation, and a and b are constants; that is, there is a straight line relation between the critical frequency and the logarithm of illumination. The constant, a, has two values, one for brightnesses above those resulting from illuminations on a white surface greater than about 0.25 meter candles and one below. It is thought by adherents to the von Kries * duplicity 146 COLOR AND ITS APPLICATIONS theory' (#48) that this point of abrupt change in slope corresponds to the change from cone to rod vision. Haycraft^^ has studied the critical frequency for spectral lights, but the results are complicated, be- cause the intensity of the various rays was not con- stant throughout the spectrum. Ives^'^ studied the relation between critical frequency and brightness for various spectral rays and obtained results which he expresses diagrammatically as shown in Fig. 75, the logarithm of brightness being plotted against critical frequency. It is noted that the *red' curve shows no change in direction at low intensities. The blue curve changes from a diagonal to a horizontal straight line; that is, at low illuminations the critical fre- quency becomes constant for blue light of various feeble intensities. Intermediate curves represent spec- tral colors between red and blue. It is significant to note that the slopes of the curves are different for the higher illuminations, the *blue' slope being steeper than the *red' slope, which indicates that the Purkinje phenomenon is operative. The author ^^ has shown that the critical frequency depends upon the wave form of the stimulus or the contour of flicker. Some of the data for white light is shown in Fig. 76. In cases a, &, c, the maximum, minimum, and mean cyclic illumination were re- spectively the same. A difference in critical fre- quency was obtained throughout a wide range of illumination, the critical frequency being higher the greater the period of darkness in a given cycle. This also appeared to hold for colored lights, but no extensive study has yet been made. 39. Signaling. — The chief requisite of a colored light for signaling purposes is high intensity, because its range depends largely upon this factor. This COLOR AND VISION 147 precludes the use of very pure colors owing to low intensities obtainable in practise, and for this reason signal glasses are a compromise between saturation of color and transparency. As is seen by the redness of the setting sun, red rays are less absorbed by smoke and dust in the atmosphere than the blue rays, therefore, a red signal should have a greater range than a blue signal through a smoke and dust laden 20 50 40 50 RELATIVE MEAN BRIGHTNESS (LOGARITHMIC SCALE) Fig. 76. — Effect of contotir of flicker on critical or vanishing-flicker frequency. atmosphere. Churchill ^^ quotes results obtained by the Geheimrat Koerter of the German Lighthouse Board on the selective absorption of artificial fog. The results indicated that red rays were absorbed to a greater degree (about 20% more) than blue rays. These results indicate that the selective absorption of clouds plays no part in the shifting of the color of the setting sun toward red. On the other hand the German Government in about 1900 found that an installa- tion of arc lamps in a lighthouse on the island of 148 COLOR AND ITS APPLICATIONS Heligoland had a lesser range in a fog than a kero- sene light of only one-hundredth the candle-power. The former contains a predominance of blue rays in comparison with the latter, which from the foregoing tests in artificial fog apparently suffered less absorp- tion. The German Lighthouse Board of Hamburg in 1894 carried out an extensive series of tests on the range of signal lights with the following results as presented by Churchill, where R represents the range in miles and / the candle-power. For white light in clear weather R = 1.53a/I For white light in rainy weather R = 1.09\/l For green light in clear weather R = 1.6 v^ I M. Busstyn ^^ estimates the range of red light as R = l.SVi. The spectral character of the illuminant of course has an important influence on the color of the signal glass. It is interesting to note that on a certain modern battleship a lighting system of blue lamps has been installed for use at night when in action. The reason given for installing blue lights is that they are in- visible to the enemy. No information was obtain- able as to whether the short range is due to the faintness of the blue lights or to a supposed lower range for blue than for yellow light of equal intensity. The Railway Signal Association (1908) after exten- sive tests arrived at the conclusions expressed in Table XIII for the effective range of the principal signal colors under average weather conditions. The colored glasses are assumed to be used with the customary semaphore lamp and lens. Paterson and Dudding ^^ have performed some interesting experiments on the visibility of point COLOR AND VISION 149 TART.E Xm Color Effective range (miles) Approx. transmission coef. of glass in service Red 3 to 3.6 1 to 1.5 2.6 to 3.0 0.6 to 0.76 0.6 to 0.75 2 to 2.5 0.20 Yellow .35 Green .17 Blue .03 Purple .03 Lunar-white .16 sources made by placing plates containing minute apertures before a wax flame. While most of their work was done indoors at distances as great as 650 feet, an experiment on the visibility of the light from tungsten and carbon incandescent lamps over ranges which extended more than a mile showed no differ- ence in the carrying power of these lights on a clear night. They established the theorem that the visi- bility of a point source is proportional to the candle- power of the source and to the inverse square of the distance. They also found that the visibility is independent of the intrinsic brightness for sources subtending less than two minutes of arc. In this connection it might be noted that they assumed a point source to be one whose linear dimensions sub- tend an angle at the eye less than the resolving power of the eye, i.e. about 30 seconds of arc for a mean wave-length of 0.5/>i and pupillary aperture of 4.5 mm. They found that the visibilities of red and green lights in clear air were closely proportional to the inverse square of the distance. On slightly illuminating the field surrounding the point source there was a loss in visibility of about 10% for a red light, 15% for a white light, and 18% for a green light, all being of equal candle-power. Their method 150 COLOR AND ITS APPLICATIONS of determining the candle-power of the red and green lights in the latter experiment was to determine the visibility of the unknown in terms of the visibility of a white light of known candle-power assum- ing the visibility porportional to the candle-power and the inverse square of the distance. Their unit of visibility was equivalent to the visibility of a point source of one candle-power at 1000 meters distance, and they state that the lowest visibility considered desirable was 0.12 of this unit. Their results for white light agree fairly well with those obtained by the Deutsche Seewarte of 1894, as is shown in Table XIV. TABLE XIV Range (Deutsche Seewarte) Candle power re- quired in clear weather (Paterson & Dudding) Computed from results of experiment 1 sea-mile (1855 meters) 2 sea-miles (3710 meters) 5 sea-miles (9275 meters) 0.47 1.9 11.8 0.41 1.6 10.0 By using artificial pupils they found little evidence of any influence of the spherical aberration of the eye on the visibility of point sources. They showed by using positive lenses that a green light equivalent to a point source was greatly dimmed relatively to a red light of similar dimension, which they attribute to the chromatic aberration of the eye. They con- clude that unless an observer has sufficient accom- modation available to focus properly a green light at infinity the latter will appear dimmed in proportion to the amount this image is out of focus. This is not so likely to occur with red light, because images of this color do not require as much accommodation COLOR AND VISION 151 in order to focus them on the retina. A purple roundel at some distance sometimes appears red in the center with blue diverging from it, which is attrib- uted to chromatic aberration of the signal lens. It is hardly necessary to state the importance of tests for color-blindness of eyes engaged in dis- criminating colored signals. Numerous test methods have been devised, but one that has been used very much is the Holmgren test, conducted with colored skeins of wool. In the choice of signal colors the four most dis- tinctive are red, yellow, green, and blue. To these can be added white (clear) and purple. The blue and purple owing to their low intensity are suitable only for short-range signals. Results by Churchill on the reaction times, or the intervals re- quired to distinguish and name signals of different colors, were in general in the following order. Red was recognized and named in the shortest time, green ranked next, then yellow, and lastly white, which required the longest interval for recognition. The relative length of the time interval varied with different subjects, but the order given was found to be generally the case. Of course the spectral char- acter of the illuminant has an important influence on the color of the signal glass. 40. Other Uses for Colored Glasses. — It is well known that dust and smoke (and very likely fog) scatter the visible rays of short wave-length more than those of longer wave-length. For this reason it is contended by some that, if the blue and violet rays are subtracted from white light, the remaining light (yellow in color) will enable an observer to see further than the total light. It is of interest to inquire further into the matter. In the case of the search- 152 COLOR AND ITS APPLICATIONS light, if the operator wishes to see a distant object in a fog he is required to look through an illuminated veil caused by scattered light, which results in de- creasing the ability to distinguish the object. If it is true that the blue and violet rays are scattered more by the fog particles, the luminous veil would become more annoying for lights containing relatively greater amounts of blue rays. Rough quantitative tests by the author, employing auto lamps with para- bolic reflectors, indicated that with yellow light, which was less intense than the total light by the amount of light absorbed by the yellow screen, objects in a fog could be seen more clearly than with the total light. There is another view-point, namely that of the person who desires to distinguish a light signal at a considerable distance. Here again an illumi- nated veil relatively near the light source if visible is likely to decrease the visibility of the signal light. This point, however, requires investigation. Based on the foregoing principle many patents have been obtained for methods of eliminating the violet rays. Colored glasses, gold-plated reflectors, fluorescent glass reflectors, etc., have been employed, but all for the same object. A noteworthy problem of projection arises with the carbon arc search-light. In order to obtain a beam of parallel light by means of silvered reflectors the area that emits light must be small and be located at the focus of a parabolic re- flector. With high-amperage arcs an appreciable portion of the light is emitted by the arc flame of relatively large area as compared with the crater of the positive carbon which is located at the focus of the parabolic reflector. The light from the arc flame has a decided violet tinge as compared with the light from the crater, and furthermore, being out of the focus COLOR AND VISION 153 of the parabolic reflector, its light is not * paralleled,' but escapes in a cone of relatively large angle. By using a yellowish glass in the aperture of the search- light, this light of a bluish tinge is greatly reduced in comparison with the reduction of the yellower light from the crater, thus decreasing the possible annoy- ance due to the * luminous veil.' The same result would be obtained if the * look-out' wore yellow glasses. In the case of a very powerful searchlight such a glass in the aperture probably would be broken by the rise in temperature due to its absorption of radiant energy. If it were inconvenient for the look- outs to wear the yellow glasses before their eyes there would be some virtue in the gold-plated re- flector which would reduce the amount of blue and violet light in the reflected light. However, in all such cases there is a cone of light which escapes directly without being altered by selective reflection. In the application of the foregoing principle to auto headlights or to any projectors employing electric incandescent lamps, any possible objectionable effect of the excessive scattering of violet and blue rays can be overcome by incorporating the yellow glass directly in the lamp bulb, by applying a yellow color- ing to the exterior of the bulb, by inserting a yellow glass in the aperture of the reflector, or by wearing yellowish glasses before the eyes. An interesting case is found in a fluorescent glass reflector (silvered on the back surface) which absorbs most of the violet and blue rays. One of the claims advanced for this reflector is that it utilizes the ultra-violet, violet, and blue rays of the incandescent lamp by taking advantage of the fluorescent property of ura- nium glass which converts these rays into yellow- green light; however, these rays constitute a very 154 COLOR AND ITS APPLICATIONS small proportion of the total visible rays in the light from electric incandescent lamps ordinarily used for such purposes. Furthermore, the fluorescent yellow- green light produced by these rays is not * directed' by the parabolic reflector, because this light is emitted in all directions. On looking at such a reflector it appears of a yellowish green tint, but close examina- tion shows that the yellow-green fluorescent light is emitted by the glass in all directions. Therefore, no practical gain in intensity of the directed beam results from the conversion of the ultra-violet, violet, and blue rays into yellow-green light, because the latter is diffusely emitted and the effect of such a fluorescent glass amounts to little more than elimi- nating most of the violet and blue rays from the radiation that is intercepted by the reflector. In this case there is also a cone of unaltered light, equal in solid angle to that subtended by the aperture of the reflector, which escapes * undirected.' This scheme appears to have little value, inasmuch as a yellow glass in the aperture of the reflector would accomplish the purpose in a more satisfactory and simple manner. Amber, yellow, and greenish yellow glasses have been used successfully for eliminating glare from the blue sky. Riflemen have found such glasses of extreme value in range shooting and a number of sportsman's glasses are available in the market. In the case of amber or greenish yellow glasses the improved condition of seeing is perhaps largely due to the reduction of glare from the blue sky, but also in part to an increased defining power due to the elimination of blue and violet rays and a relative reduction of the extreme ^ed rays (#37). An illus- tration of the effect of greenish-yellow glasses ^^ in increasing the ease of distinguishing detail is shown COLOR AND VISION 155 in Fig. 77. An acuity object ^^ (#37) was set up in the shade of a building on a clear day and light reached the object from at least one-half of the open sky. No direct light from the sun reached the eye, test object, or immediate surroundings. The author who made the observations wore no visor to shield the eyes. Only a slight sensation of discomfort was apparent before beginning the test; however, as soon as acuity observations were begun the glare became very evident and rapidly grew painful. Five readings 5 6 9 12 15 16 TIMECMIISUTES) Fig. 77. — Effect of yellow-green glasses on vision under a bright sky. were made first through clear correcting glasses (rep- resented by the black dots) and as quickly as pos- sible the clear glasses were replaced by yellow-green glasses of about 50 per cent transmission for the total light and five acuity readings were taken (repre- sented by crosses). A decided decrease in discom- fort was experienced when wearing the yellow-green glasses and, as will be noted, visual acuity is higher in this case, notwithstanding the decrease in illumination was fully 50 per cent. These glasses were again replaced by clear glasses and five acuity readings were made. This procedure was continued as indicated in Fig. 77. The interval of time required 156 COLOR AND ITS APPLICATIONS to make five readings including the change of glasses was the same in each case (being three minutes), but the actual time of making the individual readings was not noted; therefore, they are plotted at equal intervals. While the above conditions are rather complex and involve problems worthy of much careful investigation, the experiment answered the intended purpose in bringing forth several points: (1) Glare conditions are not always apparent when the eyes are not engaged in serious work such as reading or distinguishing fine detail. However, bad lighting conditions are readily recognized when the eyes are called upon to do such work. (2) There is a rapid falling off of visual acuity when the conditions of glare are severe. (3) Such a harmless appearing light source as a wide expanse of sky can produce a very severe condition of glare. The intrinsic brightness is very low as compared with artificial sources, but the quantity of light is high and the image of the sky is spread over a large portion of the retina. Its annoyance can be decreased by the use of colored glasses, which absorb much of the blue light. (4) There was an apparent recuperation of the eye during the periods that the yellow-green glasses were worn. (5) Notwithstanding the effect of glare (when clear glasses were worn) in reducing visual acuity the values of the latter when the colored glasses were worn remained considerably higher. (6) This experiment emphasizes the necessity of prolonging acuity readings over a considerable period if acuity is to be a criterion of the satisfactoriness of illumination conditions. Some of the increase in visual acuity when the yellow-green glasses were being worn can be accounted for by the nearer ap- proach to monochromatism of the light that passed COLOR AND VISION 157 through them. However, conditions indicated that the advantage was due very largely to a reduction in the glare from the sky because the glasses absorbed much of the blue and violet light. Other interesting conclusions can be drawn, but the illustration has already fulfilled its object in bringing forth the fact that glare conditions are very complex and that cog- nizance of glare often depends upon the character of the activities in which the eyes are engaged. Glasses for protecting the eyes from visible, ultra- violet, or invisible rays are coming into prominence. In considering only the visible rays, colored glasses may be combined after the principle of the subtractive method of mixing colors (#18, Fig. 20, Plate II). A superabundance of violet, blue, or green rays can be reduced by the use of red glass. That is, a colored glass will greatly reduce rays roughly complementary in color. Spectrophotometric analyses, affording data such as are shown in Fig. 12, are quite necessary for intelligently combining glasses for protecting the eyes. Spectrophotographic analysis is a convenient means of studying the transmission characteristics of glasses in the ultra-violet region and radiometric methods are applicable to the infrared region. Ordi- nary glass is sufficiently protective against moderate amounts of ultra-violet energy. In ordinary lens thicknesses it is transparent to about 0.360^, from which point it begins to absorb ultra-violet rays, be- coming practically opaque to rays of shorter wave- length than 0.300)U. Some green, yellow, orange, and red glasses are totally opaque to all ultra-violet rays, but this cannot be ascertained by a mere visual inspection. However, the ultra-violet transmission can be roughly determined by means of a quartz spectrograph and an iron arc or a quartz mercury arc. 158 COLOR AND ITS APPLICATIONS On focusing the spectrum of the radiation emitted by the quartz arc on a fluorescent material such as uranium glass, the various ultra-violet lines will be seen owing to their production of fluorescence. On inserting a specimen of glass before the slit of the spectrograph the region of absorption will be readily perceived by the disappearance or decrease in bright- ness of various fluorescent lines. The transmission 100 90 80 5 70 ai 60 .50 /- C/ear lead Class Z-Ho.O Smoke 3- Amethyst 4- Light Amber Ultraviolet ■'050 034 5-N0.7 5moke 6- Ho. 6 Smoke 7- Medium A mber d-Euptios d-Akopos Visible 1 / }> z:z. — - ■^ / / y j^ — . >-^ — "~~ ~ ^' ^■' "" / I /, ^^ y > < ^ / h 7 s / / / ^ ^ *< -^ / 1 1 1 / / > \j ■^^ A ^ ^ / / 1 f / / ^ V. ^ 1 / 1 / / / / / /i V \ ^--, ^. .-' — — X «^ — - ~7 ^-^ y /' V y V ^ --- ..: •" :^ y ^ y 034 Q36 038 0.40 0.42 Jl, WAVE LENGTH 044 (Me 048 0.50 Fig. 78. — Ultra-violet transmission curves of various glasses. of glasses in the ultra-violet region ^^ has been de- termined by using a wide slit, the spectrum of the quartz mercury arc, and a combined photographic and photometric method. For qualitative analysis an iron arc is a satisfactory source rich in ultra-violet rays. Some specimen transmission curves for various optical glasses, employed for protecting the eyes from ultra-violet energy, are shown in Fig. 78, the ultra-violet region being represented to the left of the heavy vertical line at 0.40/x. It is interesting COLOR AND VISION 159 to note the difference in ultra-violet absorption of the two samples of * smoke ' glass. Of course the absorp- tion depends upon the thickness of the specimen and its density of color. All the glasses excepting three specimens, 3, 5, and 6, transmitted 50% or more of the total light from a tungsten lamp. Spectrophoto- graphic analyses of various glasses are shown in Fig. 18. In general it is no doubt advisable to use glasses as free from color as possible and yet providing protection if they are to be worn for long periods. Yellow-green glasses when otherwise filling the re- quirements appear to distort colors (more commonly encountered) less than medium amber. A striking instance was found in the lap-welding department of a steel mill, where the operators judge tempera- ture visually. They became confused when wear- ing amber glasses, but found no difiiculty in using yellow-green glasses. This brings to mind the fact that through a yellow-green glass transmitting only a limited region of the spectrum the relation of bright- ness and temperature appears practically the same as to the unobstructed eye when the luminous sub- stance radiates light approximately the same as a black or *gray' body. Years ago Crova suggested a method of photometry involving this principle (# 54). Schanz and Stockhausen, Voege, Crookes,. Parsons, and others have studied the subject of protecting the eye from harmful rays. Crookes ^^ concludes with his associates that the relatively great amounts of infrared energy emitted by molten glass is responsible for glass-blowers' cataract, although this conclusion is questioned by some. He has made an exhaustive study of the manufacture of glasses for eye-protection and has published the valuable results. 160 COLOR AND ITS APPLICATIONS Colored glasses are often used for bringing out certain colored portions of an object in more striking contrast with the surroundings. For instance, if a black-line drawing be made on blue-lined coordinate paper and viewed through a dense blue glass, the blue lines practically disappear. If the drawing be photographed through this glass the coordinate lines will not appear on the negative. In the same manner if blue and red appear upon the same background, one or the other can be made practically to disappear by using a colored screen of exactly the same color. Of course the degree of change in contrast will depend upon the purity of the colors and the care exercised in choosing the colored screen. In using field glasses distant vision can be im- proved sometimes by the use of a light yellow screen which eliminates the blue haze from the visual image. In this connection it is well to note also that blue rays are normally out of focus at the retina. The author has experimented with colored screens for use with field glasses for detecting colored objects at a distance by altering their contrast with the sur- roundings by the use of colored screens. For in- stance, a khaki uniform (yellow-orange in color) can be made to appear either lighter or darker than the green foliage surrounding it by respectively using a yellow-orange screen or one of a complementary hue. For instance if the ratio of the brightness of a piece of khaki cloth to that of a certain green leaf be taken as unity under daylight illumination, through an ordinary orange filter this ratio became 1.5 and through a blue-green filter, 0.7. With care the con- trast can be made practically a maximum. In the case of objects more striking in color the problem is not as difiicult. Whether or not the reduction of I COLOR AND VISION 161 brightness more than offsets the advantage of in- creased contrast in distinguishing distant objects can be solved by actual trial. The point is mentioned here to illustrate the possibilities in the use of colored glasses as an aid to vision. REFERENCES 1. Physiol. Optik. 1896, p. 140. 2. Sitz. d. Berliner Akad. 1888, p. 917. 3. Bui. Bur. Stds. 1907, p. '59. 4. Lancet, Oct. 2, 1909. 5. Proc. Roy. Soc. A, 84, p. 464. 6. Color Scales. 7. Wein. Sitz. 1906, II a, 115, p. 1. 8. Bui. Bur. Stds. 6, p. 89. 9. Bui. Bur. Stds. 9, p. 59. 10. Trans. I. E. S. 1914, 9, p. 700. 11. Physiol, d. Netzhaut, Breslau, 1865, p. 138. 12. Amer. Jour. Psych. 1913, 24, p. 171. 13. Elec. World, 1911, 57, p. 1163. 14. Elec. World, 1911, 58, p. 450. 15. Elec. World, 1910, 55, p. 939. 16. Elec. World, Dec. 6, 1913. 17. Lon. Ilium. Engr. 2, p. 233. 18. Elec. World, Feb. 25, 1909. 19. Graefe Arch. f. Ophth. 69, p. 479. 20. Graefe Arch. f. Ophth. 26, p. 40. 21. Columbia. Cont. to Phil, and Psych. 20, No. 2. 22. Elec. World, 1911, 58, p. 1252; Trans. I. E. S. 1912, p. 135. 23. Sci. Amer. Sup. Feb. 2, 1913. 24. Jour, de Physiol, et de Path. Gen. No. 4, July, 1902; Comp. Rend. 2, 1903, p. 977, p. 1046. 25. Phys. Rev. 1914, p. 1; Elec. World, May 16, 1914. 26. Phil. Mag. 1834, 5, p. 327. 27. Physiol. Optik. II, p. 483. 28. Pogg. Ann. d. Phys. 1835, 35, p. 457. 29. Pfluger's Archiv. 1878, 18, p. 542. 30. Wied. Ann. 1888, 34, p. 465. 31. Phys. Rev. 1895, 1, p. 338. 162 COLOR AND ITS APPLICATIONS 32. Zeit. Inst. 1896, 16, p. 299. 33. PhysioL der Netzhaut, p. 351. 34. BuL Bur. Stds. 1905, 2, p. 1. 35. Acad. Sc. Paris, July 3, 1911; Trans. I. E. S. 1912, 7, p. 625. 36. Comp. Rend. Soc. Biol. 1885, 2, p. 485. 37. Comp. Rend. Soc. Biol. 1887, 2, p. 5. 38. Proc. Roy. Soc. 1902, 79, p. 313. 39. Jour, of PhysioL 21, p. 126. 40. Phil. Mag. 1912, p. 352. 41. Meeting Ry. Signal Assn. 1905. 42. Ann. d . Hydrographie, 1886. 43. Proc. Phys. Soc. London, 1913, 24, p. 379. 44. Elec. World, Dec. 6, 1913. 45. Elec. World, Jan. 15, 1912; Trans. I. E. S. 1914, p. 472. 46. Proc. Roy. Soc. London, A, 214, p. 1. ( f CHAPTER VII EFFECT OF ENVIRONMENT ON THE APPEARANCE OF COLORS 41. Colors have been largely treated in other chapters as if they were invariable in appearance. However, the study and application of the science of color is rendered very complex owing to the fact that the appearance of a color is so modified by its environment. The intensity, spectral character, and distribution of the light illuminating it, the adaptation of the retina for light and color, the duration of the stimulus and the character of the stimulus preceding the one under consideration, the surroundings, the size and position of the retinal image, the surface character of the colored medium, all affect the appear- ance of a given color. Thus an analysis that holds for a color in a certain environment does not in general hold for the identical colored object, viewed under other conditions. The size of a colored image and its position and duration on the retina affects its appearance, owing to the variation of sensitivity of the various retinal zones. MacDougal^ found that with small colored areas (squares from 1 to 16 sq. cm. in area viewed from a distance of one meter) the larger areas ap- peared more saturated than the smaller. He found the saturating effect of increasing the area greatest for violet and decreasing in the order, blue, green, yellow, orange, red. He even concludes that a color field is not fully saturated until it extends over the 163 164 COLOR AND ITS APPLICATIONS whole field of vision. This can hardly be true, for an observer in a room with neutral tint surround- ings illuminated with pure red light is not conscious of a saturated red color. Similarly if a white paper on a black velvet ground be illuminated by a moder- ately intense red light it will appear quite unsaturated owing to the lack of anything with which to contrast it in color. The loss in saturation appears to progress with time, no doubt largely due to * adaptation.' Whether or not this adaptation is psychological or physiological there is a lack of agreement. However, the effect of area is of importance, although there is much work to be done in this field before definite conclusions can be drawn. Another experiment of importance in viewing colors which is connected with the rate of growth of color sensations and, perhaps, to a slight degree, with chromatic aberration, is found in viewing a red piece of paper on a blue-green ground held at an arm's length under a moderate illumination. If the paper be moved back and forth without relaxing fixation at a point in the plane in which the card is moved, the red patch will appear to shake like jelly and will appear not to be in the same plane as the blue-green paper. Thus there are numerous visual phenomena associated with the appearance of colors. 42. Illumination. — It has already been shown that the maximum spectral sensibility of the eye shifts toward the shorter wave-lengths at low inten- sities (Purkinje effect #4, Fig. 2). Therefore colors ordinarily encountered appear to shift in hue under low illumination. For example, a green pigment appears to assume a more bluish hue as the illumi- nation is greatly decreased. On referring to Fig. 2 it is seen that the relative values of luminous sensa- EFFECT OF ENVIRONMENT ON COLORS 165 tion produced by equal amounts of radiant energy depend upon the wave-length. A colored pigment has the ability to reflect certain proportions of the rays of various wave-lengths. The latter is a purely physi- cal operation which remains invariable regardless of the intensity of illumination. However, the relative physiologic effect of the different rays change so that the maximum luminosity is produced by energy of a shorter wave-length at low intensities than at high illumination. By multiplying the reflection coefficients ^ V ^ \^ o /, f \ v\ z: / / \\ / / 2: \\ 3 1 / / > / V ^ A f tij y / * a: -<i^ ^^ •' \ i^ 0.42 04& 0.50 0.54 058 0.62 J^O, WAVE LENGTH Fig. 79. — Effect of the intensity of illiunination on the appearance of a pigment. of a pigment for various rays by the luminosities of the corresponding rays at high and low intensities an idea of the shift of the dominant hue is obtained. This was done for a green pigment by using the luminosity curves in Fig. 2 for high {H) and low (L) illumination. The results plotted with equal maxima are shown in Fig. 79. Colors appear more saturated at low than at high intensities of illumination. In fact, intense illumina- tion causes colors to appear very much less saturated. For instance, a deep red object illuminated by direct sunlight is painted orange-red by the artist. The employment of this illusion is successful in conveying 166 COLOR AND ITS APPLICATIONS to the observer the idea of intense illumination. Simi- larly colors appear more saturated when exposed only for a very short interval of time. . Quality or spectral character of light affects the appearance of colored objects very much. Except in very special cases a red fabric for example appears red because it has the ability to reflect chiefly the red rays (# 12, Fig. 12). Such a fabric must appear black when viewed under an illuminant which con- tains no red rays. This is the case under the light from the mercury arc, which contains practically no visible rays longer than 0.579^ (yellow). It is a fundamental principle that, excepting in special cases, a colored fabric cannot appear the same under two different illuminants. Therefore two colors that ap- pear alike under one illuminant will not match when viewed under another illuminant, unless the colors in each case* show the same spectral character by spec- trophotometric analysis. In other words, because the eye is not capable of analyzing a color spectrally, it is possible to produce colors which appear the same but whose spectral compositions differ. Such a match will not in general remain a match under another illuminant differing in spectral character. In Fig. 80 the effect of the illuminant upon the appearance of a colored pigment (purple) is shown diagrammatically. The relative luminosities (dotted lines) produced by the relative amounts of energy (full lines) of cor- responding wave-lengths are shown for daylight in the illustration on the left. The result as shown by the dotted curve is to give to the pigment the appear- ance of a blue-purple. However, when this same fabric is illuminated by ordinary artificial light of continuous spectral character, the excessive amounts of energy of the longer wave-lengths and the defi- EFFECT OF ENVIRONMENT ON COLORS 167 ciency in short-wave energy as compared with day- light alter the spectral character of the light reflected (or transmitted) by the fabric as shown in the dotted curve on the right. The appearance under the artificial light is red-purple. It is difficult to distin- guish a blue fabric as blue under ordinary artificial light owing to the scarcity of blue rays in most of the artificial illuminants. Of course a truly mono- chromatic pigment (if such existed) would not be HooN Sunlight ^, ORDinARY Artificial Light / V B G Y 0 R V B G Y 0 R Blue -Purple Red-Purple Fig. 80. — Illustrating why a purple appears differently under two different illuminants. changed in hue under various illuminants but would be altered in brightness. In the special case where no energy existed in the illuminant of the wave- length corresponding to that reflected by the mono- chromatic pigment, the latter would appear black. However, no monochromatic colors are found in practise, but if pigments that were practically mono- chromatic existed very generally, a greater intensity of illumination would very often be required than at present, because such colors would reflect very little light. Pigments ordinarily encountered re- flect considerable light, owing to the fact that they 168 COLOR AND ITS APPLICATIONS reflect energy throughout an appreciable range of wave-lengths. The spectral character of an illuminant not only influences the hue but also affects the brightness or 'value' of a pigment. Practically the whole series of Zimmerman papers were measured for their rela- tive brightness with a reflectometer (Fig. 60) under illuminations respectively from an overcast sky and from a vacuum tungsten lamp operating at 7.9 lumens per watt. The data are given in Table XV, the cata- TABLE XV Effect of Spectral Character of Light on the Brightness of Colored Papers Color of paper Reflection Coefficient Paper Overcast sky. Tungsten 1.25 w. p. m. h. c. R (Tungsten) ^^*"r (Skylight) Red-purple 0.16 .14 .21 .19 .38 .60 .60 .67 .46 .49 .32 .23 .13 .14 .30 0.23 .22 .31 .24 .48 .66 .65 .70 .42 .45 .24 .17 .09 .12 .25 1.44 b Deep red 1.57 Red 1.48 d Red 1.22 Orange 1.26 f g h Orange-yellow Yellow 1.10 1.08 Greenish yellow Yellow-ereen . . 1.04 0.91 i Dull green 0.92 q Saturated green Blue 0.75 0.74 Q Deep blue 0.69 P m Blue-purple 0.86 Gray-blue 0.83 logue designations of the papers being found in the first column. As would be suspected, the papers which have the ability to reflect the rays of longer wave-length predominantly appear relatively brighter under the artificial light. In other words their reflec- tion coefficients are greater for the tungsten light EFFECT OF ENVIRONMENT ON COLORS 169 than for daylight. Those colors having the ability to reflect the rays of shorter wave-length predomi- nantly, have relatively greater reflection coefficients for daylight; that is, they appear relatively brighter under daylight illumination. (See Figs. 113 and 117.) Thus it is seen that the spectral character of the illuminant has a great influence on the appearance of a colored object, inasmuch as it influences both the hue and brightness (value) very much. Owing to the surface character of colored media the distribution of light is of some importance in the consideration of the appearance of colors. Few pig- ments are applied in such a manner as to be per- fectly diffusing, therefore some light is specularly reflected without having penetrated the pigment. This light is unchanged by selective absorption and dilutes the light that is colored by penetrating the pigment and being selectively reflected. That is, when the light is distributed in such a manner that an appreciable amount is specularly reflected into the eye of the observer the color appears less satu- rated. In the extreme case of high specular reflection the pigment appears the same as a gray. A striking illustration of the effect of distribution of light is found in the case of the so-called changeable silks. Such fabrics have a nap, and when the fibers end in the direction toward the light the latter penetrates the fabric and is deeply colored by multiple selective reflections. The light that comes from other direc- tions is more or less specularly reflected, thus under- going less change by Selective absorption, with the result that various portions of the surface appear differently. Adding to the foregoing another property of aniline dyes and the colors of changeable silks are accounted for. For instance, a dye which in solu- 170 COLOR AND ITS APPLICATIONS tion appears pink or purple in color is often found to reflect green light predominantly in the solid state. Thus the specularly reflected light in the case of the changeable silk is sometimes roughly complementary to the light that penetrates the fabric and is returned after multiple reflections which in effect correspond to traversing a certain depth of an aqueous solution of the dye. A color will often appear different by reflection than when examined * over-hand' by look- ing through the fibers by glancing along the surface at a grazing angle (#75). 43. After-images, — It has been seen that the retinal excitation requires appreciable time to decay after the stimulus has been removed. If the filament of an incandescent lamp be viewed for an instant and the eye be then closed, an image brighter than the surroundings will persist for some time. This has been called a positive after-image. Soon, depending upon the intensity of the stimulus, the image will reach a stage of decay when it appears darker than the surroundings. If the closed eyelid be illumi- nated the visual field will appear brighter than in the case where the eyelid is shielded from the light by the hand placed gently against it. In the former case the after-image will remain * positive' a shorter time than in the case of the darker surroundings. The same will be found when viewing the after- image against various white or gray backgrounds with the eyelid open. The effect of the brightness of and exposure to the stimulus on the duration of the after-image is shown in Fig. 81. In this experi- ment the actual duration was somewhat longer than indicated, the criterion being the time aftlff exposure that was required for the after-image to decay to a certain definite, though faint, appearance- The after- EFFECT OF ENVIRONMENT ON COLORS 171 images were found to go through a certain cycle of brightness and hue changes which were not recorded. The stimulus was a bare tungsten filament varied in brightness by a variable sectored disk which was rotated at a high speed. The brightness is given in terms of candles per square inch. The after-images were observed against a faint background produced by the illumination of the closed eyelid by a small amount of stray light in the room. The changes in lOOt ^ 90 E 80 C 70 UJ I 60 g 50 < 40 o B 50 10 ^ ——- w w __ w / ^ 7 ^--'^' — f--^^ — ?00- ->^ £i^K — - If ^^ 100 " • < — —"" 1 15 f \ 1 J EXPOSURE (SECONDS) Fig. 81. — Effect of brightness on the duration of the after-image. hue were not recorded. Positive after-images, ob- tained by fixating, for a few seconds, a white paper illuminated by sunlight, can be seen for a brief period, but they rapidly decay to a brightness lower than that of ordinary surroundings. In their decay they pass through a series of hues, namely blue- green, indigo, violet-pink, dark orange, etc., which are more or less definite. Helmholtz explains the colored after-images obtained in the above manner by assuming different rates of decay of the three hypothetical color sensations which are the basis of 172 COLOR AND ITS APPLICATIONS the Young-Helmholtz theory of color vision (#47). The negative after-image is explained by Helmholtz as being caused by retinal fatigue, due to the origi- nal bright image of the white object. On stimu- lating the whole retina with white light the portion previously fatigued does not respond in the same degree as the unfatigued portions. It is difficult, however, to reconcile all the facts gleaned from studies of after-images with this fatigue hypothesis. Hering explains these phenomena by assuming that the retina is not fatigued, but that a metabolic change is aroused which is opposite in character to that pro- duced by the original excitation (# 49). After-images are also produced by fixating colored objects. For instance, if the object shown in Fig. 85 be fixated for a few seconds and the eye be turned toward a white surface, a pink after-image will be seen where the green had previously stimulated the retina. After- images viewed in this manner will usually appear approximately complementary in hue to the original stimulus. A striking illustration of approximately complementary after-images can be performed with the apparatus shown in Fig. 35. Here we have a large variety of colors, the corners of the triangle appearing red, green, and blue. After fixating the color triangle for a few seconds, if the lights be turned off and white light be permitted to illuminate the white opal-glass surface, approximately complemen- tary after-images are seen. The experiment is strik- ing, owing to the many colors present. In order to produce the complementary after-images in a striking manner, it is necessary to stimulate the retina with white light. If the experiment be performed in a dark room after the lights in the colored triangle are extinguished, the ordinary after-images will be per- EFFECT OF ENVIRONMENT ON COLORS 173 ceived, depending upon the color of the stimulus, its intensity, and other factors. After-images play quite an important part in vision, especially in viewing paintings and many other colored objects. For in- stance, if a blue skyline be viewed in juxtaposition with a green landscape, the natural shifting of the eye, even when attempting moderately to gaze steadily at the picture, will cause a shifting of the image of this dividing line upon the retina, with the result that the pinkish after-image due to the green stimulus (and likewise that due to the blue stimulus) will, in shifting above and below the horizon line, produce a vivid effect. Such phenomena often greatly add to the *life' of a painting. After steadily fixating a colored object for some time the color appears to become less saturated, and often there is an apparent change in hue. If a small piece of black paper on a larger background of red be fixated for a few seconds and the black paper be suddenly removed without disturbing the fixation, in its place will be seen a red spot more luminous than the sur?:oundings and of a more saturated reddish appearance. These experiments can be successfully performed with the Zimmerman colored papers. Successive contrast further complicates the appear- ance of colors. After stimulating the retina with red light, if the eye be suddenly fixated upon a green color the latter will appear more intense or saturated in color for a moment than if the eye had not been previously stimulated by the red light. On alternating these colors by means of a rotating disk at a low speed very brilliant effects are seen. Such successive contrasts are of importance in the study and application of color science. For instance, in permitting the eye to rove over a painting or bril- 174 COLOR AND ITS APPLICATIONS liantly colored rug the appearance of the various individual colors are influenced by the previous retinal stimulation. The phenomenon of after-images has been only briefly touched upon here. The many details connected with them serve to show how intri- cate is the reaction of the retina to light. 44. Simultaneous Contrast. — A detailed exami- nation of the mutual effect of two visual excitations is difficult, although the fundamental principles are Fig. 82. — Showing the eflfect of simultaneous contrast. The V*s are of equal brightness. not difficult to demonstrate. On viewing a gray pattern on a black background it appears brighter than when viewed upon a light background. The illus- tration shown in Fig. 82 was originally made by cut- ting the two figures in the form of a V from the same gray paper. On placing them as shown — one on a black and one on a white ground — the one on the white ground appears darker than the other. The effect is so persistent that a much darker gray can be placed on the black background and yet it will appear brighter than the one on the white ground. EFFECT OF ENVIRONMENT ON COLORS 175 In fact it is practically impossible to make both appear alike by decreasing the brightness of the gray V on the dark ground. If several gray papers of different shades be placed as shown in Fig. 83, the edge of a lighter gray strip that is adjacent to a darker one will appear brighter than the other edge of the lighter gray strip. Such a specimen can be obtained by juxtaposing gray papers of different shades or by exposing a photographic plate in a very weak light by pulling out the slide of the plate holder a half- inch at a time at regular intervals. A print from a Fig. 83. — Showing induction. Each band, though uniform in brightness, appears brighter at the right-hand edge. successful negative will afford an excellent specimen for showing this phenomenon of induction. On rotat- ing such a disk shown in 6, Fig. 30, this phenomenon of induction can be demonstrated before a large audience. A striking demonstration of brightness contrast can be performed by viewing a gray paper through a hole in a white unilluminated screen. It appears very bright in contrast to the dark surround- ings. However, on illuminating the white surround- ings it is possible to make the former bright spot appear very dark by contrast. The demonstrations of color contrast are very numerous. M. E. Chev- reul,2 who directed the dyeing laboratory of the famous Gobelins many years ago, carried out very extensive researches on the effect of simultaneous contrast of colors as used in the textile industry. The 176 COLOR AND ITS APPLICATIONS R G record of his experiments is monumentaL Many have investigated the problem with the view of throw- ing light on color-vision theories, but many of the details garnered by the vast number of investigators remain unsatisfactorily explained. The intensity of the contrast effect diminishes rapidly on passing away from the point of maximum contrast. In Fig. 84 when two colors such as red and green are juxtaposed they appear accentuated in satu- ration and deeper in hue. In the case of these two Fig. 84. — An arrangement for COlorS they appear tO mOVO showing the reduction in the further apart in hue. When contrast effect by separating the ^ two colored objects. the two colors are separated as shown above, the con- trast effect practically disappears. If a disk of green be placed on a larger disk of red the contrast is very effective but if the smaller disk is outlined by a black circle the effect is reduced. If a gray figure, as in Fig. 85, be placed upon a green background, the gray figure will appear of a pink hue. The contrast hue induced in this manner is approxi- mately, though in general not pig. gS.-An arrangement exactly, complementary to the for showing the effect of exciting color. If a sheet of thin white tissue paper be placed over the arrangement shown in Fig. 85, the hue induced in the gray paper will be considerably strengthened. Green G roLj simultaneous contrast and after-images. EFFECT OF ENVIRONMENT ON COLORS 177 Colored shadows were noticed by such great colorists as Leonardo da Vinci. These are illus- trated by casting the shadow of a pencil on white paper by light entering a window; only black and white contrast is seen. However, if from another direction light from an incandescent lamp be per- mitted to fall on the paper, another shadow is pro- duced. If the two shadows are of approximately the same brightness, the contrast colors of the shad- ows are very striking. The white ground outside the shadows is receiving the mixed light from the two illuminants, while the shadow cast by daylight receives light from only the incandescent lamp and appears yellow. The other shadow receiving only daylight appears blue. Shadows in a landscape appear blue because they receive light from the sky, and they often appear more vivid owing to contrast. Hering devised a most striking demonstration of binocular contrast. Red and blue glasses were placed in front of the two eyes respectively. The glasses sloped away from the eyes from the nasal to the temporal side. This permitted a control of satura- tion by introducing a white image from the sides by reflection. A black stripe on a white ground is doubled by increasing or decreasing the ocular diver- gence. The observed ground appears spotted, alter- nately blue and red, and sometimes a purplish white, which is due to * retinal rivalry.' The stripe seen through the red glass appears green and through the blue glass appears yellow. Briicke, Helmholtz, and others contend that the contrast effects are not of a physiological nature, but rather * errors of judgment'; that is, through the in- fluence of an adjacent color our * standard white' undergoes a change which alters our judgment. In 178 COLOR AND ITS APPLICATIONS other words they claim the effect is of a psychological nature. These arguments have been repeatedly attacked and not without a considerable degree of success by Hering and others. For example, Mayer » devised methods for showing contrast color phe- nomena on surfaces large enough so that the colors could be matched by means of rotating color disks and thus he obtained quantitative measurements. He found that the subjective contrast colors were perceptible when viewed through a small opening for exposures as short as 0.001 second. They were also perceptible with instantaneous illumination from an electric spark, the duration of illumination in this case being of the order of one ten-millionth of a second. He concluded from this experiment that fluctuation of judgment was an untenable hypothesis for explaining subjective color contrast owing to the extremely short period of time of exposure. On the other hand, Edridge-Green ^ contends 4hat all our estimations of color are only relative and formed in association with memory and the definite objec- tive light which falls upon the eye. In many of the most striking contrast experiments the color which causes the false interpretation is not perceived at all; for instance, if a sheet of pale green paper be taken for white, a piece of gray paper upon it appears rose colored, but appears colorless when it is recog- nized that the paper is pale green and not white.' Thus the controversy continues. Many contra- dictory experimental data and opinions could be cited. Contrast may be due to unconscious eye-movements, to incipient retinal fatigue, to fluctuation or error of judgment, or to some other cause. Nevertheless there is no agreement as to the true explanation at the present time. EFFECT OF ENVIRONMENT ON COLORS 179 In Plate III are provided a number of arrange- ments which show the effects of simultaneous con- trast — brightness and hue contrasts — and various mixtures of these. Some of these illustrate restful and 'lively' combinations of color. There will be found much of interest in this illustration upon care- fully observing the various combinations alone and in comparison with adjacent ones. The four smaller squares in each row are identical in hue and bright- ness, which can be readily proved by the use of a Fig. 86. — Illustrating irradiation. mask. If this plate be covered with a white tissue paper some of the color contrasts are very striking. 45. Irradiation. — This name is applied to the phenomenon of apparent increase in size of objects as they are increased in brightness. For instance, the crescent of the new moon appears larger than the remainder of the disk. A filament of an incandes- cent lamp appears, to increase in diameter as its temperature is raised from a dull red to its normal operating temperature. This effect has been attrib- uted by some to a spreading of the retinal image on account of a stimulation of nerves outside its actual geometric boundary. Others attribute the effect to 180 COLOR AND ITS APPLICATIONS the aberrations in the optical system of the eye. In Fig. 86 the inner white square appears larger than the inner black square under high illumination, yet both are identical in size. The phenomena of simul- taneous brightness contrast is also evident, the white square amid black surroundings appearing brighter than the larger white square. Such effects are also perceptible with colored objects, as will be seen in Plate III. REFERENCES 1. Amer. Jour, of Psych. 13, p. 481. 2. The Principles of Harmony and Contrast of Colours, 1860. 3. Amer. Jour, of Sci., July, 1893. 4. Proc. Roy. Soc. B, 1913, 86, p. 110. OTHER REFERENCES Tschermak, Ueber das Verhaltnis von Gegenfarbe, Kompen- sationsfarbe und Kontrastfarbe, Phliig. Arch. 1907, 117, p. 204. F. Klein, Nachbilder, Uebersicht und Nomenklatur, Englemann's Arch. f. Physiol. 1908, Sup. Bd. p. 219. G. J. Burch, Proc. Roy. Soc. 1900, 66, p. 204. An excellent bibliography of the work on simultaneous contrast is given by A. Tschermak, Ueber Kontrast und Irradiation, Ergeb- nisse d. Physiol. 1903, p. 726. General references are Helmholtz, Handbuch d. Physiol. Optik and NagePs Handbuch. CHAPTER VIII THEORIES OF COLOR VISION 46. Recorded writings, centuries before the be- ginning of the Christian era, contain speculations on the visual process. Alcmaeon, Empedocles, Aristotle, Democritus, Anaxagoras, Plato, and Diogenes are among the early writers and philosophers who pre- sented views on the nature of light and colors and on the process of vision. However, their specula- tions — which can hardly be considered otherwise, owing to lack of experimental data — are of little value since the modern development of the sciences. Color vision is largely physiological and psychologi- cal. The process of vision involves the physical cause, the physiological retinal process, and the psy- chological elements in the experience of sensations. As the knowledge of the three sciences involved in the process of color vision developed, theories of color vision became more intricate. In fact the vari- ous theories which are given credence at the present time are found on strict analysis to include in vary- ing degrees the physiologic process of vision, color vision, and the nature of perception. A theory of color vision must include all the foregoing factors, yet the dominating influence of one of these is usually perceptible in a given theory. In this chapter it is proposed to set forth briefly the latest theories which pertain to the subject of color vision. 47. Young-Helmholtz Theory. — Thomas Young is credited with the conception of the three-color 181 182 COLOR AND ITS APPLICATIONS theory, but it seriously lacked experimental founda- tion until after the epoch-making work of Helm- holtz/ and since that time it has become known as the Young-Helmholtz theory. The hypothesis is that color sensations depend upon the action of three independent physiological processes involving three substances or sets of nerves. This theory approaches the matter chiefly from the side of physics; that is, the facts of color-mixture are used in building up the theory. There is no anatomical evidence that the three substances or sets of nerves are present. The primary sensation curves shown in Fig. 53 were determined by Koenig by being built up from experi- mental data; these have been proposed as repre- senting the three independent processes. They are plotted so as to enclose equal areas on the assumption that the sensation of white results from the stimula- tion of equal amounts of the three primary sensations. It is noted that spectral hues involve more than one of the primary sensations. This theory explains the main facts of color vision, although many details uncovered by experimenters have iiot yet been reconciled with it to the entire satisfaction of many scientists. After-images are explained by assuming fatigue of one or more of the processes in varying degrees. For instance after fatiguing the eye to green light a white surface appears an unsaturated purple — pink. Many of the observed facts in the study of after-images are only approximately conciliable with this theory. The problem of simultaneous contrast offered no diffi- culties to Helmholtz, because he assumed that the phenomenon is the result of * false judgment.' While it may be purely psychological, it appears probable to some that it is actually physiological in nature. THEORIES OF COLOR VISION 183 one part of the retina being influenced by stimulation of another region. Color-blindness is explained by assuming that one or more of the three processes are absent, the remaining process (or processes), if necessary, being assumed to be * redistributed' to some extent. This theory has some advantages in explaining the cases of red and of green blindness by assuming the absence of the corresponding process and if necessary a slight modification of the other two. It fails to explain total color-blindness, however. When it is attempted to reconcile this theory with all the observed facts, one finds a highly complex state of affairs. Such a discussion is outside the scope of this chapter, therefore only the main theories and facts will be presented. Extended discussions will be found in the treatises referred to at the end of this chapter. The Young-Helmholtz theory satis- factorily explains the observed facts of color-mixture, but the chief objection to the hypothesis as it exists at the present time, is that it fails to explain many other facts, such as those of contrast. 48. 'Duplicity' Theory.— This theory, which at- tempts to differentiate colorless and color vision, is chiefly associated with the name of Von Kries. It is based upon anatomical evidence of the existence of * rods' and * cones' in the retina. The former are assumed to be responsible for achromatic sensations and the latter for both achromatic and chromatic sen- sations. The rod action is supposed to be largely responsible for light sensation at twilight illumina- tion and is in general more responsive to rays of shorter wave-length. The cones, however, are sup- posed only to act under stimuli of brightnesses repre- sented by the range above twilight illumination and not to be greatly increased in sensitiveness by dark 184 COLOR AND ITS APPLICATIONS adaptation. Examination of the retina shows that the cones alone exist in the very center of the retina, the fovea centralis, and rods appear just outside of this and predominate in the outer zones. The chief observed facts that this theory explains fairly satis- factorily (perhaps because it was chiefly built up from these facts) are (1) colorless vision over the whole retina in dim light, for instance in moonlight, (2) the decreased sensitivity of the fovea in twilight, (3) the shift in the maximum of the luminosity curve of the eye (Purkinje effect) at low illumination, (4) the absence of such a shift for foveal vision, (5) no achromatic threshold is found for any light for foveal vision, (6) no achromatic threshold for red light for any region of the retina, and (7) colorless vision over the whole retina in the case of the totally color blind. Some of the experiments with color-blind eyes fur- ther support the theory. For instance the luminosity curve for a totally color-blind eye at ordinary illumi- nations is similar to that for a normal eye for twi- light vision. There are also evidences of diminished foveal sensibility, abnormally good vision in twi- light, and decreased ability to fixate small objects with color-blind eyes. Further support is found in the presence of rods almost exclusively in the retinae of such nocturnal animals as the owl and bat. The supporting evidence in general is represented by more dependable and convincing data than in the case of any theory of color- vision. The * duplicity theory' does not attempt to explain color- vision, but is of interest here because of the attempt to separate vision into chromatic and achromatic processes. 49. The Hering Theory. — The principal foun- dation of this theory ^ consists of facts such as those of contrast, and the apparent simplicity of black, THEORIES OF COLOR VISION 186 white, and yellow as well as red, green, and blue. Hering assumes there are six fundamental sensa- tions coupled in pairs, namely, white and black, red and green, yellow and blue. In order to account for these six fundamental sensations he assumes the presence somewhere in the retinocerebral apparatus of three distinct substances. Each substance is capable of building up (anabolism) or of breaking down (katabolism) under the influence of radiant energy or its effects. The building up of the black- white substance causes a sensation of blackness, and the breaking-down of this substance, a sensation of whiteness. Likewise anabolism of the red-green substance is connected with the sensation of green and katabolism with red sensation. Similarly, the building up of the third substance produces blue, and the breaking down is associated with yellow sensation. For example, red rays cause a breaking down of the red-green substance, with the result that red sensation is experienced. It is claimed by many that this theory has an advantage over the Young- Helmholtz theory, because it deals more directly with the sensations of color. The theory has many en- thusiastic supporters and is fully as favored in this respect as its most formidable rival. A favorite argument in support of it is the observed fact that yellow appears to be a primary color because there is no simultaneous suggestion of both red and green in a yellow made by mixing these two colors (Fig. 17). Many observed facts concerning after-images agree with the theory. For example, if the eye be stimulated by blue rays, anabolism will take place in the yellow-blue substance and an accumulation of the substance results. If now yellow rays are per- mitted to stimulate the same area of retina, the break- 186 COLOR AND ITS APPLICATIONS ing down of the yellow-blue substance proceeds at a greater rate and the sensation is greatly augmented. Conversely yellow decreases the amount of substance and increases the rate of anabolism under the sub- sequent stimulation of blue rays. Positive after- images are explained by assuming a continuation of the anabolic (or katabolic) change for a brief period owing to chemical inertia. All the general phenomena of after-images are explained satisfactorily, but as in the case of the Young-Helmholtz theory, details are troublesome. Some of the data on color-blindness readily support the theory, but the latter must be modified in order to explain other data. Bonders ^ and others conclude that the Young-Helmholtz and Hering theories, having been formulated from dif- ferent points of view, have arrived at different con- clusions and that both are in part correct. This is a rather safe conclusion, but nevertheless an important one, inasmuch as they are both thus stamped with the partial approval of scientists highly familiar with the subject. 50. Ladd- Franklin Theory. — In this theory ^ the rods and cones are used. Colorless sensations white, gray, and black, are assumed to be caused by a primi- tive photo-chemical substance which is composed of many 'gray' molecules. These exist in their primi- tive state only in the rods, but upon dissociation they cause the colorless sensation. In the cones the gray molecules undergo development and for some reason only a portion of the molecule becomes dissociated by rays of a given wave-length or color. The evolution of the gray molecule is assumed to take place in three stages diagrammatically shown in Fig. 87. In the first stage the gray molecule exists, but is so constructed that it is disintegrated by light of THEORIES OF COLOR VISION 187 all colors, thus producing a white or a gray sensation. In the second stage the molecule has become more complex and contains two groupings. The disso- ciation of one of the latter 0 STAGE 1 (z^^z^^ stage: 2 0 (V^0— 0 STAGE 3 causes a yellow sensation and the other, blue. Their simultaneous dissociation causes a sensation of white or gray. Molecules are assumed to exist in this stage in the outer zone of the retina, where neither red nor green can be per- ceived as such. In the third stage the yellow grouping is divided into two new combinations, the dissociation of one giving rise to a red sensation, the other producing a green sensation. If the red and green are dissociated sim- ultaneously, yellow sen- sation results, while all three (red, green, and blue) together produce gray. There is much of interest in this theory, and it appears to explain many observed facts satisfactorily. 51. Eldridge- Green Theory. — Boll discovered a substance diffused in the retina which has been named visual purple. This discovery gave rise to hopes that a photochemical theory of vision would explain the observed facts, inasmuch as the visual purple was found to be sensitive to light. However, after the elaborate work of Kiihne the visual purple lost much of its significance in this respect. If an 0 Fig. 87. — The evolution of the Ladd- Franklin gray molecule. 188 COLOR AND ITS APPLICATIONS eye which has been unexposed to light for some time be cut out in a room illuminated by means of a dim red light, on removing the retina it appears a purple color under ordinary light. The color fades rapidly on exposure to ordinary intensities of illumination, passing through red and orange to yellow, finally disappearing. The yellow appearance is supposed to be due to the formation of another pigment, the visual yellow. The appearance of the preceding stages is thought to be due to mixtures of the visual purple and visual yellow in various proportions. It apparently has been established that normally the visual purple is confined to the outer portions of the rods. It is extracted readily by a watery solution of bile salts. Spectroscopic examination of this solu- tion shows it to have a maximum absorption for yellow-green rays and a minimum for red rays and is bleached by the rays in about the proportion that it absorbs them. The visual purple is so sensitive to light that pictures of very bright objects have been seen in purple and white on retinae of the eyes of animals. Such experiments have been performed by exposing an eye extracted from an animal which has been kept in darkness for some time. Many attempts have been made to weave the visual purple into a theory of vision. Edridge-Green ^ has done so, as briefly outlined below. He assumes *that the cones of the retina are insensitive to light, but sensitive to the changes in the visual purple. Light falling upon the retina liberates the visual purple from the rods, and it is diffused into the fovea and other parts of the rod and cone layer of the retina. The decomposition of the visual purple by light chemically stimulates the ends of the cones (probably through the electricity which is produced) THEORIES OF COLOR VISION 189 and a visual impulse is set up which is conveyed through the optic nerve to the brain.' He further assumes that 4he visual impulses caused by the different rays of light differ in character just as the rays of light differ in wave-length. Then in the impulse itself we have the physiological basis of the sensation of light, and in the quality of the impulse the physiological basis of the sensation of color.' He also assumes 4hat the quality of the impulse is perceived by a special perceptive center in the brain within the power of perceiving differences possessed by that center or portion of that center. According to this view the rods are not concerned with trans- mitting visual impulses, but only with the visual purple and its diffusion.! On this theory he attempts to explain all the observed facts encountered. He concludes the paper, from which the foregoing is quoted, by stating that *I am not aware of any fact which does not support the theory.' Needless to say, however, there are those who entertain a dif- ferent opinion on this last and other points. REFERENCES 1. Handbuch der Physiologischen Optik. 2. Grundziige der Lehre vom Lichtsinn, Leipzig, 1905, p. 41. 3. Archif. f. Opth. 1881, 1, p. 55; 1884, I, p. 15. 4. Zeit. f. Psych, u. Physiol, der Sinnesorgane, 1892. 5. Lancet, Oct. 2, 1909. OTHER REFERENCES Ebbinghaus, Theorie des Farbensehens, Zeit. f. Psych, u. Physiol, der Sinnesorgane, 1893. A. Koenig, Ges. Abhandlungen, Leipzig, 1903. M. Greenwood, Jr., Physiology of the Special Senses. W. Nagel, Handbuch der Physiologie des Menschen, 1905. 190 COLOR AND ITS APPLICATIONS W. Wundt, Grundzuge der Physiologischen Psychologie, Leipzig, 1911. H. Aubert, Grundziige der Physiologischen Optik, Leipzig, 1876. Captain W. de W. Abney, Colour Vision, London, 1895. F. W. Edridge-Green, Colour Blindness and Colour Perception, London, 1909. W. Nicati, Physiologie Oculaire, Paris, 1909. J. H. Parsons, Colour Vision, New York, 1915. 1 CHAPTER IX COLOR PHOTOMETRY * 52. The relation between radiation of various wave-lengths and luminous sensation has long been the subject of investigation; but, notwithstanding the extensive data obtained, there is no general agree- ment as to a method that yields correct results. Much of the early data is practically useless at the present time, owing to the lack of control of various influential factors, due to the absence of definite knowledge regarding their ability to influence the judgment of brightness. This data of course has served well in lighting the pathway of investigation. From foregoing chapters it has been seen that the size of the photometric field, owing to the variation of retinal sensibility to colored light, is of importance in color photometry. Due to the Purkinje phenom- enon the brightness at which measurements are made also affects the results. In this connection it should be noted that the brightness of the photometric field as seen by the eye is sometimes greatly reduced by absorption of light in the optical path and by a small ocular aperture or artificial pupil of the instrument. Other factors, such as the adaptation of the eye and the character of the surrounding field, are influential. Most important is the method, for no two methods yield exactly the same results. It is well to remem- ber that the brightness of a colored area is so influ- enced by its environment that its determination, in comparison with a standard in an isolated photometric 191 192 COLOR AND ITS APPLICATIONS field, is not in general a measure of its brightness as it appears in another environment. Therefore the photometry of colored surfaces yields measure- ments of the brightness in terms of the particular standard used and the results depend upon the hue, saturation, and brightness of the comparison field, the surroundings, the condition of the eye, and the photometric method used. 53. Primary Methods of Photometry. — A method of photometry should ordinarily have for its object the measurement of the illuminating value of the illuminant with respect to its ability to make objects visible by reflected or transmitted light. The method of visual acuity has been proposed and used by some for the measurement of illumination. Obviously such a method determines the defining power of the illumi- nant or of the light reflected or transmitted by an object. The criterion of such a method is usually the discrimination of fine detail or the adjustment of the illumination so that the detail appears to be equally legible as compared with a standard. In general this method is quite insensitive, and the results are greatly dependent upon fatigue and the state of adaptation of the eye. However, there is another complication, that of the spectral character of light. The experiments of the author described in #37 showed that a reduction in the amount of illumination or brightness of the acuity test object, when accompanied by certain changes in the spectral character of the illuminant, sometimes results in an increase in visual acuity. From these experiments it is seen that the method of visual acuity cannot be depended upon to determine the relative illumi- nating values of illuminants or of lights altered in spectral character by reflection from colored surfaces. COLOR PHOTOMETRY 193 The visual acuity method is valuable in many cases, but it must be understood that a large amount of our seeing does not include the perception of fine detail at the limits of discrimination, but only requires the recognition of relatively large surfaces through dif- ferences in color and brightness. Even reading under ordinary conditions does not involve visual acuity at the limit of discrimination, for the illumination is usually far above that necessary to distinguish the type, and it has been shown that in reading the eyes recognize characters in groups, travel by jumps, and come to rest only a few times during their progress across a page. At one time the critical frequency method was looked upon as a possible solution of the problem of color photometry. It was shown in #38 (Figs. 75 and 76) that if a brightness be alternated against darkness there is a certain minimum frequency of alternation, called the critical frequency, at which flicker just disappears. In general, it has been found that the critical frequency varies directly as the logarithm of the illumination or brightness of the test surface. Thus plotting these two factors yields a straight line, the slope of which is different for lights of different colors. The slope of this straight line changes abruptly at a very low illumination (thought by some to be the point at which the cones just cease to be sensitive to light) for lights of all colors with the ex- ception of red. By this method two surfaces were assumed to be equally bright when their critical or vanishing-flicker frequencies were equal. The method has proved too insensitive for practical use and too susceptible to various physiological factors, such as fatigue and adaptation. The ordinary direct comparison or equality-of- 194 COLOR AND ITS APPLICATIONS brightness method is claimed by many to involve the only true criterion for the measurement of brightness. Others claim that its shortcomings have disqualified it for use in the photometry of lights of different colors and have accepted the flicker method. The latter method, which is in high favor for color photometry, has not been proved to measure the true brightness of a colored surface — if there be such a thing. Nev- ertheless, the uncertainties in the measurements by the direct comparison method has brought this or- dinary method into disfavor with many photometri- cians for the photometry of lights differing in color. The flicker method involves the alternation of two brightnesses — the standard and the unknown. A match is made with this instrument by altering both the brightness of the field due to one of the sources and the frequency of alternation. The two bright- nesses are considered equal when the frequency of alternation is such that a slight change of either illumination produces a just perceptible flicker. When there is no color difference between the two bright- nesses being compared, the theoretical frequency should be zero, but owing to imperfections in the photometric apparatus this condition is never obtained. The flicker photometer is based upon the fact that color difference is eliminated by mixing the two brightnessies by persistence of vision, the color flicker apparently disappearing before the brightness flicker. Numerous instruments have been devised for this purpose, but all involve this fundamental principle. No extended comparative study of flicker photometers has been made, although it is possible that instru- ments differing in design might yield different results. For instance, in many instruments the stimulus changes abruptly from one color to another, but in COLOR PHOTOMETRY 196 some the stimuli dissolve into each other. Whether or not such instruments yield different results is a question to be solved by further investigation. To summarize, the methods of visual acuity and critical frequency are impracticable owing to their 10 A Q y \ / / \ O / • \ V / \ 00 -, i N uJ •7- / / CXI / GO UJ K / \ y 5 [ \ < -J A 1 \ UJ 4 01 / \ •X y f \ / V / ^ ,9. -*. »^^ ~^ c y -::; >^ ^ *■•«., -^ <sV \ 1 ,'i. ^ ^ .-o' € '^■^ •^.. ^ ^ I '^ <^ <»«^. ^ *."«a n '^ > Nv. S2I 1 0.50 0.52 054 0.56 05& 0.60 0.62 0.64 0.66 Fig. 88. — The results of four methods of photometry (Ives). extreme insensitiveness and the influence of eye fa- tigue and adaptation. The influence of the spectral character of light further complicates the visual acuity method. The direct comparison method, though claimed by many to yield measurements of *true' brightness is unpopular, owing to the imcertainties in the measurements. The flicker method, hoTyever, owing to its elimination of color difference and high 196 COLOR AND ITS APPLICATIONS sensibility, had won many ardent supporters even before extensive investigations of the method had been made. In Fig. 88 are shown data obtained by Ives ^ with the four methods. In each case the standard was the total light from a tungsten lamp. Spectral colors were compared with this white standard. Curve V was obtained by the visual acuity method; Dy by the direct comparison; F, by the flicker; and C, by the critical frequency method. If the four methods gave identical results, the curves would coin- cide. The general shapes and positions of the max- ima are similar, but the areas under the curves are very different. The enormous area under curve V is in accord with the previous work of BelP and of Luckiesh,^ which showed that acuity was much better in monochromatic light than in light of ex- tended spectral character. 54. Secondary Methods of Color Photometry. — Various schemes have been proposed and developed for eliminating color difference in heterochromatic photometry, such as the use of colored filters, and physical and chemical photometers used with filters that properly weigh the energy of various wave- lengths according to their light-producing effects. Among the latter possibilities are the radiometer, thermopile, selenium cell, photo-electric cell, and photographic plate. Obviously, in order to reduce measurements to absolute values, the transmission coefficients of the colored filters must be determined by some acceptable method. Likewise, determina- tions of the relation between radiation of various wave-lengths and the corresponding luminous sensa- tions and of the sensibility of the instruments to energy of various wave-lengths must be made before COLOR PHOTOMETRY 197 the results obtained with the selenium cell, the radiom- eter, filters, etc., are useful in measuring brightness. Crova ^ suggested as a method of comparing lights possessing continuous spectra, but differing in color, the determination of their intensities at one wave- length, 0.582/z. The lights to be compared in this manner must not differ much in spectral energy dis- tribution from the black body. Rayleigh,^ Nernst,^ Fery and Cheneveau,^ Lucas,^ Rasch,^ and others have made various applications and modification of Crova's original proposal. The filter used by Crova consisted of an aqueous solution of anhydrous ferric chloride (22.321 grams) and crystallized nickelous chloride (27.191 grams), the total volume being 100 c.c. at 15° C. A thickness of 7 mm. of this solu- tion was used which transmits energy from 0.63^ to 0.534a( with a maximum of transmission at 0.582^, the wave-length which Crova found to be satisfactory for carrying out his proposed scheme. Ives ^° tested Crova's method by comparing the luminous intensities of a tantalum and a carbon incandescent lamp ^t various wave-lengths. He found that the wave-length for such a comparison lies between 0.56/x and O.SSju, depending on the range of temperature. The latter wave-length was found to hold best of all within the limits of temperature represented by ordinary incan- descent lamps of that time. Twelve years ago Fabry ^^ recommended the use of two or more col- ored solutions for eliminating color difference, having fijrst calibrated these solutions for thickness and trans- mission by an acceptable method. Aniline dyes were not used, because of the need for definite and re- producible solutions. By using two solutions, A and 5, he was able to match the Carcel lamp with almost any illuminant. The solutions were made as follows: 198 COLOR AND ITS APPLICATIONS A. Crystallized copper sulphate 1 gram Commercial ammonia (density 0.93) 100 c. c» Water sufficient to make one liter. B. Potassium iodide 3 grams Iodine 1 gram Water sufficient to make one liter. Ives and Kingsbury ^^ have recently investigated the problem of obtaining suitable solutions that would eliminate color difference after the manner pro- posed by Fabry. They developed a yellow solution containing 100 grams of cobalt ammonium sulphate, 0.733 grams of potassium dichromate, 10 c.c. of 1.05 sp. gr. nitric acid, and distilled water to make one liter at 20 deg. centigrade. The method of preparation is considered very important and is pre- sented in detail in the original paper. Of course a given depth or concentration of the solution has a different transmission for illuminants of different spectral character. The transmission values were determined by means of a flicker photometer by averaging the results obtained by specially selected observers. The transmission of the solution was found to vary considerably for different temperatures and the character and cleanliness of the glass sides of the containing cell were found to be of consid- erable importance. It was found possible to eliminate color difference in comparing many illuminants with the carbon lamp standard by placing the solution on either one side or the other of the photometer. Many have used aniline dyes and colored glasses. In practical photometry the use of colored glasses appears to be satisfactory for a large amount of work. The carbon lamp operating at about 4 w.p.m.h.c. is the present standard of luminous intensity. Properly COLOR PHOTOMETRY 199 tinted bluish glasses used with this standard will eliminate the color difference when comparing tung- sten lamps with it. The transmissions of the tinted glasses can be obtained by averaging the determina- tions of a large number of observers, using the direct comparison method. Such a procedure is being used successfully in several laboratories for the above work where the color difference is not excessive. However, it is not a solution of the general problem of color photometry. Houston 13 in 1911 proposed the use of a filter composed of two solutions — copper sulphate and potassium dichromate — for closely approximating in transmission the luminosity curve of the eye, this filter to be used with an energy-measuring instru- ment. Koenig's visibility data were used as a basis for developing the solution. A proper solution would transmit rays of various wave-lengths in the propor- tions corresponding to the relative light-producing values of the various rays. It is necessary to cut off both the infra-red and ultra-violet rays and to re- duce the visible rays in just the correct relative pro- portions so that an energy-measuring instrument (bolometer, thermopile, or radiometer) will record data proportional to the luminous intensity. A dis- advantage of such instruments is found in their extreme sensitiveness to outside disturbances. For instance, the galvanometer used in the procedure must be of a liigh order of sensibility and therefore must be set up where it will be free from mechanical and magnetic disturbances. Karrer,!^ recently follow- ing Houston's lead, similarly employed the visibility data obtained by Ives. By using three solutions he was able to produce a screen whose transmission curve closely approached this luminosity curve of the 200 COLOR AND ITS APPLICATIONS eye. The solutions were made by dissolving (1) 57.519 grams of cupric chloride, (2) 1.219 grams of potassium bichromate, and (3) 9.220 grams of ferric chloride, each in one liter of water. A triple cell was used, each compartment being 1 cm. thick. The selenium cell has been used for stellar pho- tometry and for other special work, owing to its change in resistence on being illuminated. However, it has not yet found a place in color photometry, be- cause at present it is too erratic and undependable. 1 1 6elehium I BtptoElecfric / \ >• / \ H- A / \ =i / S. ^' .^tye \ CD 1 \ / \ 1 \ V5 1 \ / \ T. tiJ / X \ / \ UJ > h / \ / \ \ X / \ § / > / V"~ / \ K / ^ \ / 040 0.46- 0.50 0.55 0.60 0.65 070 0.75 Fig. 89. — spectral sensibilities of selenium and photo-electric cells compared with the spectral sensibility of the eye. Its sensibility to energy of various wave-lengths ap- pears to depend upon the method of making the cell, and is in general far different from that of the eye. A sensibility curve is shown in Fig. 89, compared with the luminosity curve of the eye. The maximum change in resistance is usually due to energy of the longer visible wave-lengths. Obviously a filter that properly weighs the energy of various wave-lengths according to its light value and to the spectral sen- sibility of the cell, must be used for the photometry of illuminants of extended spectral character. The photo-electric cell has been used in special cases of scientific investigation for detecting the COLOR PHOTOMETRY 201 presence of radiant energy. Surfaces of potassium, zinc, and other elements and compounds in vacuo exhibit the property of emitting electrons when illumi- nated. The maximum effect is usually found in the short-wave visible region, as illustrated by a sensi- bility curve of a photo-electric cell, shown in Fig. 89. As in the case of the selenium cell, the photo-electric cell is too erratic at the present time to be adopted as a means of photometering lights of different colors. The strengths of the electronic currents measured by means of a sensitive electrometer or galvanometer afford a measure of the relative intensities of the illumination of a given spectral character when the characteristics of the cell are shown; that is, when the relation between the intensity of illumination and the photo-electric effect is known. Lights differing in spectral character cannot be compared by means of the photo-electric cell unless a correcting filter is used after the manner necessary with the selenium cell. The photographic plate affords another possible method for the photometry of lights of different color, but its general adoption is discouraged, owing to lack of uniformity of the emulsion both as to thickness and sensibility. Some of the difficulty could be obviated by using plates made of plate glass. The panchromatic plates must be used, because the ordi- nary plate is not appreciably sensitive to rays of longer wave-length than 0.48/x, the maximum of sensibility being in the extreme violet region of the spectrum. The relative sensibility of a certain com- mercial panchromatic plate, for equal amounts of energy of various wave-lengths, is shown in Fig. 90 compared with the spectral sensibility of the eye. In order to make the plate record the values of col- 202 COLOR AND ITS APPLICATIONS ored brightnesses as determined with a flicker pho- tometer, an accurate filter was made which consisted of aesculine, tartrazine, rhodamine, naphthal green, and glass three-eighths of an inch thick. How nearly this filter performs its intended purpose is shown in Fig. 91 by the circles in comparison with the lumi- nosity curve of the eye which is represented by the full line curve. This filter was used with the pan- 100 90 80 ^ 70 S 60 §50 ^40 / ,/. \ / i \ f -4^ 1 1 1 \ V, r V / \^ \ — V \ L / \^ / 1^ \ ^ \ / A-TJ 1 I \ \ i 1 1 1 r \ \ s — V ij \ \ N \ / \ I / ' N \ .^^' V. ^^^_ %A0 044 045 Q52 056 060 0.64 0.66 UC, WAVE LEMGTH Fig. 90. — spectral sensibility of a panchromatic photographic plate. chromatic plate considered above, for which it was made by Ives and Luckiesh^^ for various photometric problems. In using the photographic plate for pho- tometric purposes it must be remembered that, in general, the product of intensity of illumination and time of exposure is not a constant for equal photo- graphic effect. The relation between exposure and intensity of illumination for a constant photographic effect as discovered by Schwartzchild is IP = i T^ where / and i are the larger and smaller intensities and T and t are the larger and smaller periods of COLOR PHOTOMETRY 203 exposure. The value of p varies with different plates, generally lying between 0.75 and unity. The manner of development, the temperature, and other obvious factors influence the results so that the photographic method becomes unattractive except for special prob- lems. As already stated, the use of these so-called physical or chemical photometers, while obviating color difference in practise, does not preclude the necessity of establishing the relation between lumi- nous sensation and radiation of various wave-lengths by an acceptable method of color photometry. Vi -K ^ '-xs vO O UJ 7Z f O p \ 1 \ CQ / \ UJ > , J f N V ^ y / S S^ ^ -^ ^ 1 k 0.40 0.44 048 0.72 Fig. 91. 0.52 056 0.60 0.64 0.68 An accurate color filter for the panchromatic plate considered in Fig. 90. 55. Direct Comparison and Flicker Methods. — Only two primary methods for the photometry of lights differing in color are worthy of consideration, namely the direct comparison and flicker methods, the other two being ruled out of consideration for reasons already given. These two methods have been compared by many observers, but much of the work is so incomplete that it yields little data for a thorough comparison. It is desirable that the method finally acceptable for photometering lights of different colors should measure light-value with the same order of definiteness as other physical measurements are obtained. Dow 16 compared these two methods by using 204 COLOR AND ITS APPLICATIONS colored lights produced by means of red and green glasses at different intensities and with different field sizes. He found with the direct comparison method that the ratio of the red to the green bright- ness decreased with decreasing illumination, the decrease being rapid below 0.3 meter candle, — the well-known Purkinje phenomenon. With the flicker method this decrease was slight. With a small pho- tometer field the change in the ratio of red to green was considerably less. In general Dow's results indicate that the flicker method is less influenced by the size of the field or by a change in the illumina- tion than the direct comparison method. P. S. Millar ^^ compared mercury vapor arcs with incandescent lamps over a wide range of illumina- tions. The Purkinje phenomenon was in evidence in the direct comparison measurements but absent in the results obtained with a flicker photometer. In other words, with the former method the apparent brightness of the side of the photometer field illumi- nated by light from the mercury arc did not decrease as rapidly as the brightness of the other side illumi- nated by light from an incandescent lamp, as the illumination decreased. Stuhr ^^ compared the four methods — namely, visual acuity, critical frequency, direct comparison, and flicker. He found the critical frequency and flicker methods to yield identical results, but these differed from the results by the other two methods. Various physiological factors, such as field size and illumination, were not considered. Ives ^^ carried out an extensive series of investi- gations which represent the most elaborate and thorough work yet done on the problem. He con- cluded that the flicker method is more sensitive than the direct comparison method and that the results COLOR PHOTOMETRY 205 are more reproducible. He discovered that the flicker method exhibited a * reversed Purkinje effect' and found, as other investigators had, that the two methods yielded different results in general, but con- cludes that the flicker method yields, under certain specified conditions, a measure of true brightness. Much evidence obtained throughout these investiga- tions and some obtained by the author and others point favorably to the flicker method as the best method of photometry. However, notwithstanding the extensive investigations, some take the stand that the case has neither been decided against the direct comparison method nor in favor of the flicker method. This conclusion is perhaps justifiable. However, considering the unsatisfactoriness of the former method, there is considerable virtue in the adoption of the latter method with its many satisfactory fea- tures in default of a method which has been definitely proved to yield the desired measurements. Ives in his early papers did not emphasize the differences in the results obtained by the two methods. His results were plotted in the form of luminosity curves of the eye, so that without careful inspection, the results by the two methods, under certain conditions of high illumination and small field size, do not appear to differ greatly. In order to determine the magnitude of these outstanding differences the au- thor 20 carried out an investigation, a portion of the results {L) being plotted in Fig. 92. Red and blue- green lights were used. The ratio of the intensity of the red to that of the blue-green light is plotted for a wide range of illuminations. The illumination values are those obtained with the flicker photometer and a standard tungsten lamp, but are not corrected for the absorption of the photometer, which, owing to 206 COLOR AND ITS APPLICATIONS a complex optical path, was considerable, or for reduction due to the small artificial pupil. It is seen that the flicker method exhibits a reversed Purkinje effect and the direct comparison method the true Purkinje effect, and further that the ratio of the red to the blue-green brightness obtained by the direct comparison method is only about 62 per cent of that obtained by the flicker method for r.5 1.4 h ao 1.3 1.2 I.I 1.0 0.9 0.8 0.7 0.6 0.5 OA 0.3 0,2 0.1 0, ^^=; 5^ ~iz: (Flicker) Observer K (Flicker') Observer L o,tQUALiTY Of Brightness) Observer 0 I 2 3 4 5 6 7 8 9 10 II 12 15 14 15 METERCAttDLESCOn PHOTOMETER SCREEM) Fig. 92. — Resixlts by flicker and direct comparison photometers, illustrating differences including the Purkinje effect and a reversed effect. a large range of illuminations. The results as to the reversed Purkinje effect were verified in general by another observer {K). It is seen that he did not obtain the same results as the author, even with the flicker photometer, by about 13 per cent. A similar difference, though in general not as great, is found in Ives' data when the lights of the correspond- ing dominant hues (0.64/x and 0.52^) are compared, even though in his measurements the spectral colors were always balanced against a white light. The same extreme difference in the results by the two COLOR PHOTOMETRY 207 methods was confirmed by the writer using the same glasses a year later. The field size in the foregoing experiments was rather large — about ten degrees — but a large difference persists even with smaller fields, though not to such an extent. Morris- Airey ^i suggested that the differences between the two methods might be due to the dif- ferent rates of rise of the sensation with different colors. The author ^2 studied this factor and showed that the maximum of a flickering red light was con- siderably greater than that of the blue-green light for a large range of flicker frequencies when the brightnesses of the two lights were those obtained by a direct comparison balance. This in itself did not prove which, if either, is the correct method. However, another experiment was performed which is perhaps as convincing as any yet performed in indicating that the flicker photometer when properly used is inappreciably influenced by the different rates of growth and decay of color sensations. Lights differing greatly in spectral character, but alike in hue, were compared by the two methods and identical results were obtained. Two yellow lights were ob- tained by means of filters of aqueous solutions of potassium dichromate used before a tungsten lamp. In one of the solutions was dissolved some neodym- ium ammonium nitrate, which absorbed all the spec- tral yellow. The two lights now nearly matched in hue and were readily brought to an exact c6lor-match by altering the concentration and by adding a little ordinary yellow or orange dye to one solution. The spectral characters of these two lights are shown in c and dj Fig. 17. Two 'white' lights were also com- pared, one consisting of the total light from a tung- sten lamp, the other being made up of narrow regions 208 COLOR AND ITS APPLICATIONS of the spectrum, respectively in the red and blue- green. In both cases no difference in the ratio of the intensities of the two lights of the same color was detected in the results by the two methods, the accuracy being well within one per cent. It was also shown that red and blue-green lights add, whether by direct superposition or by alternately flickering them as in the flicker photometer, when the flicker is not more than barely apparent. Further investigations may show that the flicker photometer is influenced by the different rates of growth and decay of color sensations, but the fore- going experiments indicate that such influence is slight. Ferree 23 has attacked this problem and has reported some interesting preliminary results. The flicker method possesses many desirable character- istics, yet at present it can hardly be accepted as yielding *true' measurements of brightness unless the difference in the results, obtained by this method and by the direct comparison method, be ignored. Where color differences are large — just where such a method as the flicker method is most desired — the results by the two methods vary most widely. 56. Luminosity Curve of the Eye. — Ives found that the spectral luminosity curve obtained with the di- rect comparison photometer by the * cascade' method (involving small steps of slight hue difference) agrees at high illumination for a small field with the curve obtained by the flicker photometer. He also found that the latter method fulfilled certain funda- mental axioms, namely, that the sum of several indi- vidual brightnesses of different hue must equal the brightness of the whole and that if each of two bright- nesses of different hue equal a third brightness, they must be equal to each other, while the direct com- COLOR PHOTOMETRY 209 parison method did not. While these experiments point with favor to the flicker method, it is true that a method can fulfill these requirements and yet not yield measurements of *true' brightness. However, it appears at the present time that the balance of experimental data is strongly in favor of the flicker photometer. For this reason the relation between radiation of various wave-lengths and their physi- ologic effect in producing luminous sensation as 1.0 0.9 08 5 0.3 LiJ O.I 4 r\ h i '// ^ K 1 i 1 \ Ij \ k K/ \ A \ / / V\\ / J Y ^ ^ *^ \ k 040 0.44 048 0.52 036 0.60 Jl, WAVELENGTH 0.64 O.60 Fig. 93. — VisibiUty data. (See Table XVI.) obtained with the flicker photometer is of consid- erable interest. Ives ^^ determined the luminosity curves of 18 observers which he has published in comparison with the mean curve. Later he deter- mined the luminosity curves for 25 more observers, which mean, he states, agrees well with that of the previous eighteen. Nutting ^^ has recently obtained such data with 21 observers. The apparatuses used by both Nutting and Ives were practically the same as shown in Fig, 51, the source W being eliminated. 210 COLOR AND ITS APPLICATIONS The data as presented by Nutting are shown in Fig. 93 and Table XVI compared with Koenig's^e origi- TABLE XVI The VisibiUty of Radiation (See #92) Wave length (/x) Nutting mean visibility Ives mean Koenig mean Computed from Nutting*s formula 0.400 0.002 0.410 0.003 0.420 0.008 0.430 0.012 0.440 0.023 0.029 1 0.450 0.038 0.047 1 0.158 0.460 0.066 0.073 1 0.201 0.470 0.105 0.107 1 0.250 0.480 0.157 0.154 0.302 0.135 0.490 0.227 0.236 0.370 0.232 0.600 0.330 0.363 0.476 0.358 0.510 0.477 0.696 0.670 0.614 0.520 0.671 0.794 0.830 0.675 0.530 0.835 0.912 0.960 0.824 0.540 0.944 0.977 0.996 0.933 0.550 0.995 1.000 0.990 0.994 0.560 0.993 0.990 0.945 0.993 0.570 0.944 0.948 0.875 0.939 0.580 0.851 0.875 0.780 0.839 0.590 0.735 0.763 0.680 0.717 0.600 0.605 0.635 0.585 0.586 0.610 0.468 0.509 0.492 0.456 0.620 0.342 0.387 0.396 0.343 0.630 0.247 0.272 0.300 0.236 0.640 0.151 0.176 0.210 0.158 0.650 0.094 0.104 0.128 0.108 0.660 0.051 0.068 1 0.070 0.072 0.670 0.028 0.044 1 0.032 0.680 0.012 0.026 1 0.690 0.007 0.700 0.002 1 Extrapolated. nal data which was obtained by the direct comparison method. Nutting has extended the observations well into the red and violet regions of the spectrum by COLOR PHOTOMETRY 211 using sources emitting line spectra. Koenig's data are shown in curve Ky Ives' data in curve /, and Nutting's data in curve N. Nutting developed a formula of the form V= V^R^ e^^"~^^ from which the values given in Table XVI and represented in Fig. 91 by the circles have been computed. F^ in the formula represents the maximum light-producing ef- fect, R = ^, a = 181, and V the visibility or rela- tive light-producing value of energy of any wave- length, X. The maximum sensibility is at X^ax = 0.555m. The computed values are found to coincide practically with Nutting's mean luminosity curve be- tween wave-lengths, 0A8fjL and 0.65^. REFERENCES 1. Phil. Mag. 1912, 24, p. 847. 2. Elec. World, 1911, 58, p. 637. 3. Elec. World, 1911, 58, p. 450, p. 1252. 4. Comp. Rend. 93, p. 512. 5. Phil. Mag. June, 1885. 6. Phys. Zeit. 1906, 7, p. 380. 7. Bui. Soc. Inst. Elec. 1909, p. 655. 8. Phys. Zeit. 1906, 6, p. 19. 9. Ann. d. Phys. 1904, 14, p. 193. 10. Phys. Rev. 1911, 32, p. 316. 11. Comp. Rend. Nov. 1913; Trans. I. E. S. 1913, 8, p. 302. 12. Trans. I. E. S. 1914, 9, p. 795. 13. Proc. Roy. Soc. A, 1911, p. 275. 14. Lighting Jour. Feb. 1915; Phys. Rev. 1915, 5, p. 189. 15. Trans. I. E. S. 1912, p. 90; Elec. World, 1912, 60, p. 153. 16. Phil. Mag. 1910, 19, p. 58. 17. Trans. I. E. S. 1909, 4, p. 769. 18. Kiel, Phil. Diss. Vol. 19, 1908, p. 50. 19. Phil. Mag. 1912, 24, p. 149, p. 170. 20. Elec. World, Mar. 1913, p. 620. 21. Electrician (Lon.), Aug. 20, 1909, p. 758. 22. Phys. Rev. N. S. 1914, 4, p. 11. 212 COLOR AND ITS APPLICATIONS 23. Before. I. E. S. 1914. 24. PhiL Mag. 1912, 24, p. 853. 25. Trans. I. E. S. 1914, 9, p. 633. 26. Ges. Abhandlimgen. OTHER REFERENCES On the Photo-electric Cell: H. Dember, Beiblatter, 1913, No. 16; p. 1044. Nichols and Merritt, Phys. Rev. 1912, 34, p. 475. F. K. Richtmeyer, Trans. I. E. S. 1913, p. 459; Phys. Rev. July, 1915. H. E. Ives, Phys. Rev. N. S. 1914, 3, p. 68, p. 396. On the Selenium Cell : Seig and Brown, Phys. Rev. N. S. 1914, 4, p. 48, p. 85; 5, p. 65, p. 167. F. Townsend, Sci. Abs. A, 7, 2869. A. H. Pfund, Phys. Rev. 1912, 34, p. 370; Light. Jour. 1913, p. 128. T. Torda, Electrician, 1906, 56, p. 1042; Sci. Abs. 9, 771. Joel Stebbins, Astrophys. Jour. 1908, 27, p. 183. On Color Photometry: E. P. Hyde and W. E. Forsythe, The Visibility of the Red End of the Spectrum, Phys. Rev. July, 1915; Astrophys. Jour. Sept. 1915. Irwin J. Priest, A Proposed Method etc., Phys. Rev. July, 1915. E. F. Kingsbury, A Flicker Photometer Attachment for a Lummer-Brodhun Photometer, Jour. Frank. Inst. Aug. 1915, p. 215. H. E. Ives and E. F. Kingsbury, Flicker Photometer Measure- ments on a Monochromatic Green Solution, Phys. Rev. 1915, 5, p. 230. CHAPTER X COLOR PHOTOGRAPHY 57. At the present time no processes of color photography have been developed which employ the simple principle of fixing the colors of Nature directly upon the photographic plate by chemical means. O. Wiener \ discusses the use of body colors which would assume the colors corresponding to the rays of light by chemical modification. Carey Lea^ ob- tained a form of silver photochloride which assumed different colors on exposure to various rays, but no means was found for fixing them. Most of the com- mercial methods employ colored media which repro- duce colors by one of the common methods of color- mixture. In the first place the emulsion must be sensitive to all visible rays, and preferably the plate should be sensitive to light rays, in closely the same manner as the eye. There are no commercial plates endowed with the latter characteristic, so panchro- matic plates are usually used with an approximate color filter. Ray filters of the accuracy approaching that illustrated in Fig. 91 are rare, but for accurately photographing colored objects in their true values of light and shade, carefully made filters must be used with panchromatic plates, because the latter differ greatly in spectral sensibility from the eye (Fig. 90). In other words, a plate must be rendered of the same relative sensibility to the various visible rays as the eye by the use of sensitizing dyes and ray filters. 213 214 COLOR AND ITS APPLICATIONS Fortunately ordinary photography does not require such a high degree of accuracy. About a century ago Seebeck discovered that silver chloride becomes tinted by exposure to light with an accompanying chemical action. It is also possible by properly selecting luminescent salts to produce a series of tints after exposure which are very effective. Such colors cannot be fixed, and therefore are of little practical interest. The devel- opment of color photography has been confined largely to two methods. In one the phenomenon of interference of light waves is utilized to reproduce colors directly, while the other method is based upon the principles of color-mixture — both additive and subtractive (#18, #19). In the latter method artificial color-screens are used. Sometimes these are of minute size, as will be shown later. 58. Lippmann Process. — The method employ- ing interference of light waves is originally due to Becquerel,^ but Lippmann's name is usually asso- ciated with the process, owing to the improvements which he devised after extensive investigation. ^ Zenker ^ in 1868 explained the colors sometimes ex- hibited by spectrograms made on silver chloride plates as due to the interference of light waves reflected from layers of metallic silver which are originally produced by stationary light waves. Among those who have investigated the process are Wiener,^ Neuhaus,^ Valenta,^ Lehmann,^ and Ives.^'^ In the Lippmann process the sensitive film is backed by a film of clean mercury which acts as a reflector. As light which has passed through the thin film strikes the layer of mercury it is reflected back on its path, and owing to the disappearance of energy at certain points through interference, the Red COLOR PHOTOGRAPHY 215 silver compound is acted upon only in layers — at the antinodes. The phenomenon is diagrammatically shown in Fig. 94. The silver compound, instead of being acted upon throughout the thickness of the film, is largely reduced in thin laminae the dis- v^ioooooox ^re-g-/? tance between which is one-half a |^|oooooooo<^/y5> wave-length of the light producing them. EspeciaUy fine-grain plates "itl^ii^rre^pfo! must be used in order to produce duced in the Lippmann the very minute structure. The p^®^®^^- emulsion must be sensitive to all colors which are so made by the use of certain sensitizers. This discovery is due to H. W. Vogel, in 1873, who found that silver bromide by treatment with certain sensitizing dyes, such as eosine and cyanine, was rendered sensitive to rays of longer wave-length than when untreated. If the plate after exposure in the Lippmann process be developed and illuminated by white light, from various parts of the film only colored light escapes to the eye and a photograph in colors is seen. It is easy to account for the reproduction of pure spectral colors, but the general theory has been the subject of much discussion too extensive to dwell upon here. 59. Wood Diffraction Process. — This method, invented by R. W. Wood in 1899, depends upon the phenomenon of interference, though in a different manner. It depends upon the principle that all colors may be matched in hue by mixtures of three primary colors, red, green, and blue, each consisting of a narrow band of the visible spectrum. These spectral primaries lie near the regions of the spectrum cor- responding respectively to 0.65m, 0.52/z, and 0.45/1. This process utilizes diffraction gratings for the pro- 216 COLOR AND ITS APPLICATIONS duction of the primary colors. If a point source of light or an illuminated slit be viewed through a dif- fraction grating (#9), not only will an image of the source be seen, but displaced on either side a series of spectra will be seen. The displacement of the spectra from the line joining the eye and light source will depend upon the number of lines per inch in the grating; the fewer lines per inch the less is the dis- placement. The primary spectral colors are produced as shown in Fig. 95. If a source of light S, a lens I, and a grating G, be arranged as shown, an image of the source will be seen on a screen at /. With a Fig. 95. — Illustrating the Wood diffraction process. fine grating a spectrum of the source will be seen on the screen extending between aa. With a coarse grating a spectrum of the source will be formed between cc and a medium grating will produce a spectrum at bb. If the three gratings have different rulings, the eye at E will see the lens face illuminated by a monochromatic color depending upon the grating interposed at G. If all three spectra be produced simultaneously in proper intensities, the eye at E would see the lens face illuminated by white light providing gratings of proper rulings are chosen. Such a scheme was used by Wood for viewing the photo- graphs. The latter, which appear colorless, really consist of images of the object made on plates of bichromated gelatine through three properly chosen COLOR PHOTOGRAPHY 217 gratings. In making the photographs one of the gratings was placed in contact with the bichromated gelatine film and the image of the object was pro- jected upon the sensitive film through the grating. This grating was then replaced by another and the procedure repeated. It was then repeated a third time with the remaining grating, but usually with another sensitive plate. On superposing the two exposed plates and viewing by a proper combination of lens and light source the picture was seen in colors. Copies can be made by contact printing. It was found, however, that while satisfactory pictures could be made there was no certainty about obtain- ing them. This was later found to be due to super- posing the three grating exposures. Ives^^ improved the process by printing the grating pictures through a very coarse grating placed at right angles to the lines of the three gratings. The coarse grating had opaque lines twice the width of the transparent strips. After making an exposure through one of the gratings the coarse auxiliary grating was moved in a direction perpendicular to its lines a distance equal to the width of one of its open slits and an exposure was made through the second grating. This procedure was repeated for the third grating. The lines of the coarse auxiliary grating were as in the so-called Joly process to be discussed later, so narrow as to be just unresolved by the eye — about 200 to the inch. This process yielded satisfactory results. Ives further simplified the process by making one grating answer the purpose of the original three by using the finest grating — 3600 lines per inch — and rotating it in its plane respectively 21.5 and 42 degrees for the other two exposures. Thorp, unknown to Ives, had previously suggested the use of one grating for a 218 COLOR AND ITS APPLICATIONS similar purpose. When using three gratings, one with 2400 lines per inch furnished the red component, one with 3000 the green, and one with 3600 the blue. F. E. Ives worked out a viewing apparatus involving important improvements over Wood's original scheme. 60. Color Filter Processes. — If an object be separately photographed on three panchromatic plates respectively through properly chosen red, green, and blue filters (which collectively transmit all visible rays) and these three photographs be separately projected upon a white screen by means of three projection lanterns equipped with the foregoing colored filters, three separate * monochromatic' photographs will be seen. If, however, the three colored images — red, green, and blue — be superposed in exact coincidence, a picture in natural colors will be seen. The principle is that of adding colors as shown in Fig. 21. F. E. Ives, who was a pioneer in this field, developed an apparatus for viewing the three-colored photographs simultaneously and also the so-called chromoscope for tri-color projection of photographs made in this manner. Charles Cros independently developed a similar method. In 1868 Louis Ducos du Hauron described a process for three-color photog- raphy (since known as the Joly process) which in- volved the ruling of red, green, and blue lines of transparent dyes on a transparent screen. The lines were too fine to be distinguished by the eye. The procedure involves the juxtapositional method of color-mixture, a principle long used in the textile industry and in painting. If a photograph be made through such a screen and a positive made there- from, the latter will appear in colors when viewed through the original screen when properly superposed. The screen is diagrammatically shown largely mag- COLOR PHOTOGRAPHY 219 nified in Fig. 96. (In Figs. 96, 97, and 98 red, green, and blue are represented respectively by the hori- zontal lines and the diagonal lines running in direc- tions perpendicular to each other.) There have been numerous variations of this scheme commercialized. The Paget screen is illustrated in Fig. 97. An inter- esting development is the Lumiere process. Minute grains of dyed starch are used in a thin layer over a sensitive emulsion. Three batches of transparent starch grains are -dyed respectively orange-red, green. du Hauron Paget Lumiere FIG.% FIO. 97 FIG. 96. ^ Green Red Blue Illustrating three processes of color photography. and blue. These are mixed in such proportions as to give a mixture of neutral color and are spread on the plate in a single layer. A portion of the plate greatly magnified is shown diagrammatically in Fig. 98. The light passes through the minute color filters of dyed starch before striking the plate. The plate is developed in the ordinary manner, and by chemical means the negative is converted into a positive. The reversal may take place in a bath of potassium per- manganate acidified with sulphuric acid and is later developed again in the same developer as used in the first development. After drying, the plate is varnished and the color photograph is ready for rrvT\r 220 COLOR AND ITS APPLICATIONS viewing. The process is a very ingenious one and reproduces natural colors quite satisfactorily. A de- ficiency of the process of no great importance in most work is shown in Fig. 99. It is also of interest in showing the inability of 040 OiO m 070 the eye to analyze colors. .^.WAVELENGTH The approximate trans- Fig. 99.-iUustrating the limitations of missiou of the three dves certain processes of color photography. "^ are diagrammatic ally shown. It is evident on photographing the solar spectrum that the dyes used are somewhat too monochromatic, because the colored spectrogram which consists of red, green, and blue bands shows gaps between the blue and green, and also between the green and red where little color is visible. For instance in a spectrogram of the mercury spectrum the yellow lines at 0.578/^ appear an orange-red. This defect is evident in a greater or less degree in the foregoing processes, depending upon the spectral character of the colored dyes. However, owing to the fact that colors ordinarily en- countered are far from monochromatic, this deficiency is unimportant and practically negligible in ordinary color photography. This defect is encountered with regret when one desires to reproduce spectra for demonstration purposes. For the latter purpose the methods employing the three colored transparencies about to be described are satisfactory. The fore- going methods are based upon the additive and juxta- positional processes of color-mixture. The processes using the minute color filters shown in Figs. 96, 97, and 98 have a disadvantage in loss of light. For instance, if the process be analyzed it will be seen that a red object will be recorded upon the photo- COLOR PHOTOGRAPHY 221 graph in general in the proportion of one red patch to two black patches. That is, no red light will be transmitted by the minute blue and green filters, so in the final photograph these will appear as black spots. This is a decided disadvantage in the making of colored lantern slides for projection unless an exceedingly powerful arc lamp is available. A num- ber of processes employing subtractive method (#18) have been developed. Sanger Shepherd developed a method wherein three differently colored films such as are indicated in Fig. 20 are superposed in a single transparency. F. E. Ives was also a pioneer in this field. The process is identical in principle with the tri-color printing process in use at the present time, with the exception that in the latter case a black- white record is sometimes used with the three color records. Three negatives are made respectively through red, green, and blue filters from which posi- tives are made on special thin transparent films of celluloid coated with gelatine sensitized by immersion in a solution of bichromate of potash. The trans- parencies are each dyed a color complementary to that of the taking filter, the red record being colored blue-green (cyan-blue); the green record, purple (magenta); and the blue record, yellow. These transparencies are free from opaque silver deposit, the gradation being from a maximal transparency to the deepest color of the dye on each film. On super- posing them the natural colors are produced by the subtractive method, as will be readily understood from an inspection of Fig. 20. F. E. Ives devised a method after this principle whereby the three plates were exposed simultaneously with one lens. Shep- herd first employed a repeating plate holder so that the three plates were successively exposed through 222 COLOR AND ITS APPLICATIONS the proper filters. A few years ago a process of producing moving pictures in colors known as Kinema- color was launched. In order to simplify the matter only two colors are used, namely a blue-green and an orange-red. The different colored images are alter- nately thrown on the screen at the usual rate. It is obvious that the use of three colors would render the problem exceedingly complex. Such a two- color method cannot reproduce all colors with fidelity, but the results are quite satisfactory considering the simplification that is obtained. Recently another scheme, employing only two colors, has been devel- oped, known as the Kodachrome process. By means of a repeating back two plates are successively ex- posed through red and green filters respectively. These are developed in the ordinary manner and after being washed they are bleached and fixed, at this stage appearing transparent. They are next given a final washing in a weak aqueous solution of ammonia and dried. Finally the plates are dyed, the one made through the red filter being dyed a bluish-green and the one made through the green filter an orange-red. In general the processes employing dyed trans- parencies superposed yield more brilliant color records, but are obviously more dependent upon the skill of the photographer. In much work the processes employing the juxtapositional method of color-mixture are more satisfactory owing to the simplicity, notwith- standing the less brilliant results. Of the latter methods those employing the ruled screens are some- what more flexible; however, the adjustment of the viewing screens requires some patience. It is thus seen that at the present time the prob- lem of color photography has been solved by rather COLOR PHOTOGRAPHY 223 indirect methods involving color-mixture. Most of the methods will be completely understood on refer- ring to Chapter III. REFERENCES 1. Wiedmaim*s Ann. 1895, p. 335. 2. Amer. Jour. Sci. 1887, p. 349. 3. Ann. d. Chimie et Phys. 1848, p. 451. 4. Comp. Rend. 114, p. 961; 111, p. 575. 5. Lehrbuch der Photochrome, 1868. 6. Ann. d. Phys. 1899, 69, p. 488. 7. Des Farbenphotographie nach Lippmann*s Verfahren, 1898. 8. Die Photographie in naturlichen Farben, 1894. 9. Beitrage zur Theorie und Praxis der director Farben-photo- graphie, 1906. 10. Astrophys. Jour. 1905, 27, p. 325. 11. Jour. Franklin Inst. June, 1906. OTHER REFERENCES Louis Ducos du Hauron, Les Colours en Photographie, 1868. R. Child Bailey, Photography in Colours, 1900. E. Konig, Natural Color Photography, 1906. E. Konig, Beiblatter Ann. d. Phys. 1909, p. 1027. J. A. Starcke, Sci. Amer. Sup. Mar. 9, 1913, p. 158. A. Byk, Phys. Zeit. Nov. 22, 1909, p. 921. G. E. Brown, Photo Miniature, No. 128, 1913. G. L. Johnson, Photography in Colours, 1914, CHAPTER XI COLOR IN LIGHTING 61. Lighting is of great importance, because it is essential to our most important and educative sense — vision — and color is intimately associated with lighting and vision. Color in lighting is rapidly grow- ing in interest in the science and art of illumination. The recent increase in the luminous efficiency of light sources and the rapid strides in the development of the art of lighting are largely responsible for the growing interest in color and quality of light. Much is yet to be learned regarding the physiological and psychological effects of color, and the laws for its proper use are hazy and not well understood. How- ever, equipped with a full knowledge of the physics of color, an aesthetic taste and a comprehensive view of what is known and unknown regarding the physio- logical and psychological influence of color, a person is capable of utilizing many of the possibilities of color in lighting. The illuminant plays a very im- portant part in the appearance of colors, as has been seen in Chapter VII. The spectral character of the illuminant influences the hue and relative brightness of colors, and the intensity influences the hue and apparent saturation. At low intensities the hue shifts toward the shorter wave-lengths and at high intensities there is an apparent decrease of satura- tion. The distribution of the light affects the appear- ance of colors, owing to the character of these surfaces. All of these factors are of importance in 224 COLOR IN LIGHTING 225 considering the proper illuminant for accurate color work in the dye-rooms of textile and paper mills, in the mixing of pigments for color printing and for painting, for the matching of colors, and in many other places. The spectral character of illuminants is of importance (#37) in the discrimination of fine detail, for it has been seen that monochromatic light is superior in defining power to light of any other spectral character. There are many important problems as yet un- solved which involve color in its application to light- ing. There are practically no data on the influence of color on eye fatigue, although it is known that colors are of influence psychologically. There is a prevalent idea that the kerosene lamp is *easy on the eyes,' owing to its yellowish color. However, the low intrinsic brightness of the kerosene flame as compared with more modern illuminants is a fact worthy of consideration. When it is further noted that there is no general objection to daylight on account of its color — and it is far whiter or more bluish than ordinary illuminants — it must be ad- mitted that the virtue of the kerosene lamp based upon its color is on a rather shaky foundation. It is likely that the eye having evolved under daylight is better adapted to it than to any other illuminant and that the nearer an artificial illuminant approaches daylight in spectral character the more likely is it to be satisfactory physiologically. Misuse of common illuminants is perhaps responsible for eye-fatigue to a greater extent than any spectral characteristics. One cannot look directly at the sun and state con- scientiously that daylight is ideal. It has been found that visual acuity is better in monochromatic light than in daylight (#37), and it may appear from this 226 COLOR AND ITS APPLICATIONS that daylight is not ideaL However, these experi- ments were carried out at ordinary intensities con- sidered satisfactory in artificial lighting, and daylight intensities are ordinarily very much greater, which means that, for the discrimination of ordinary details, the intensity is many times the minimal amount re- quired, so that the limit of defining power is seldom reached. For years many have held that the eyes are less fatigued when reading from yellow paper than from white paper. In a biography of Joaquin Miller we read that *he wrote on yellow paper with a pencil because white paper hurt his eyes.' Bab- bage many years ago strongly advocated the use of yellowish paper in reference books, such as logarithm tables, where the eyes are severely taxed. Javel later advocated the same procedure, claiming that eye-strain was decreased, owing to a decrease in contrast. Many are of the same opinion although, as already stated, quantitative data relating directly to the problem are lacking. After reading from white paper the eyes seem to welcome a change to yellow paper, but this may be due to a decrease in contrast, owing to a lower reflection coefficient of the yellow paper than that of the white paper. However, meas- urements show only a slight difference in the bright- ness of pale yellow copy paper as compared with white, especially under ordinary artificial light. There is no doubt that a yellow or yellow-green light of less extended spectral character than daylight or ordinary artificial light is of superior defining power, due to the reduction of the effects of chromatic aberration in the eye. This fact may partly account for the contention that yellow paper is * easier on the eyes.' It is difficult to focus blue light at a normal reading distance, and impossible to do this at the same time COLOR IN LIGHTING 227 keeping the most luminous rays in focus, therefore, the elimination of the blue rays by means of yellow paper may actually increase the definition. However, reading does not ordinarily involve the discrimination of fine detail, but instead the recognition of groups of characters. Furthermore, the eye is found to pro- gress across a page in a series of jumps, being sta- tionary only a few times per line. It has been found that there is practically no difference in visual acuity when the detail is viewed against a white ground and a ground consisting of yellow copy paper when both receive the same intensity of illumination, that is the same density of light flux. Colored surroundings, such as foliage, brick walls, the wall coverings of the room, etc., alter the spectral character of light before it arrives at the useful plane. Such effects must be considered in any lighting prob- lem requiring a light of a certain spectral quality and are also of importance from the aesthetic viewpoint. Many uses of illuminants of different color and colored media are found in the problems of lighting. ^62. The Production of Artificial Daylight. —The arts having developed largely under daylight illumi- nation, the daylight appearance of colors is naturally considered as standard. With the production of artificial light man became less dependent upon day- light; nevertheless, owing to the impracticability and perhaps impossibility of a dual criterion of color, there has always been a demand for artificial daylight. The efforts in the production of artificial light have been directed toward the production of light of day- light spectral quality. The principal reason, no doubt, is that such a procedure in our most important method of producing light (by high temperature radiation) at the present time tends toward an ever-increasing 228 COLOR AND ITS APPLICATIONS luminous efficiency. Nevertheless each increment in the steady approach toward daylight has been loudly acclaimed by reason of the better * color-value ' of the illuminant. However, there is a method which has been applied whereby light of a daylight character can be obtained by excluding from an illuminant con- taining all the rays found in daylight, those portions which are present in excessive amounts. Such a subtractive method is wasteful of light, but is made practicable by the recent increase in the luminous efficiency of illuminants. However, it is well to remember that efficiency in lighting as in any other case is *the ratio of satisfactoriness to cost and not the reciprocal of the cost' In order to produce artificial daylight it is neces- sary to determine the spectral character of natural daylight. First it is well to distinguish between sun- light and skylight. The latter is scattered sunlight, but owing to the relatively greater scattering of the rays of short wave-length (#13) skylight is more bluish in color than sunlight. Daylight varies tre- mendously with time and place, although north blue skylight and clear noon sunlight, when unaltered by reflection from immediate surroundings, are fairly constant in color. However, the modification due to selective absorption of the particles in the atmos- phere and selective reflection from foliage, buildings, etc., make daylight rather indefinite in spectral char- acter. E. L. Nichols ^ has published interesting accounts of his investigations on the spectral char- acter of daylight under different conditions of weather, cloudiness, location, and time of day. He found among other things unmistakable evidence of the coloring added to daylight by reflection from green foliage by noting the characteristic absorption spec- COLOR IN LIGHTING 229 trum of chlorophyl (a substance in green foliage) present in observations made on land in the summer time. This effect was absent on the sea. Koettgen,^ Nichols and Franklin,^ Crova/ Vogel,^ Ives,^ and others have studied the spectral character of day- light. The data on noon sunlight and skylight plotted in Fig. 5 is a weighed mean of the results of the foregoing investigators as presented by Ives. The distribution of energy in the visible spectrum of clear noon sunlight as it reaches the earth corre- sponds closely to that of a black body at 5000 deg. absolute (C). A number of investigators, including Dufton and Gardner,^ Mees,^ Pirani, Ives,^ Hussey/" and Luckiesh^i have devised colored screens for producing artificial daylight by altering the light from an artificial source emitting a continuous spectrum. In order to demon- strate the procedure and illustrate the advantage of first choosing a light as close to daylight as possible, the productiion of daylight screens for two tungsten lamps of different luminous efficiencies as considered by Luckiesh and Cady ^^ will be presented. The visible spectrum of the light from a tungsten lamp being continuous, it has all the rays present that are found in daylight. The difference in their spec- tral characters is due to the difference in the relative amounts of the various rays present. First let us consider the production of light of noon sunlight quality from a vacuum tungsten incandescent lamp operating at 7.9 lumens per watt (1.25 w.p.m.h.c.). It is found sufficiently accurate to consider no rays of shorter wave-length than 0.42^. An ideal screen for altering the tungsten light to a noon sunlight quality will therefore transmit all the rays of wave-length 0.42)U. It will partially absorb rays of longer wave- 230 COLOR AND ITS APPLICATIONS length in increasing proportions from 0.42^ toward the long-wave end of the spectrum. The reduction of the intensity of the rays of various wave-lengths is readily computed from the ratios of the amounts of these rays present in noon sunlight to the amounts of the corresponding rays present in the tungsten light under consideration. The resultant transmis- sion curve of a colored screen for thus altering the 100 90 Q 60 O tt: 70 ixi ■^ . 60 ^ 30 20 10 V(7 ^^ ^""^^ 040 "0.44 046 Q5Z 03& 0.00 0.64 0.66 0.7? Fig. 100. — Ideal transmission screens for producing artificial daylight. tungsten light (7.9 lumens per watt) to noon sunlight quality is shown in &, Fig. 100. The ideal transmis- sion curve of a colored screen for producing artificial noon sunlight by means of a nitrogen-filled tungsten lamp operating at 22 lumens per watt (0.5 w.p.m.h.c.) is shown in c. In order to produce artificial north skylight it is seen in Fig. 5 that the visible rays of long wave-length must be reduced by relatively greater amounts than in producing artificial noon sunlight. The ideal transmission curve for producing artificial north skylight by means of the tungsten COLOR IN LIGHTING 231 lamp operating at 7.9 lumens per watt is shown in a. The ideal transmission curve for producing artificial north skylight with the gas-filled tungsten lamp oper- ating at 22 lumens per watt coincides closely with b. That is, a screen which produces artificial noon sun- light with the older type of tungsten lamp operating at 7.9 lumens per watt will produce artificial skylight when used with the gas-filled tungsten lamp operat- 100 90 u 80 40 20 JO 0.40 044 048 052 056 0.60 0.64 0.68 0.72 Fig. 101. — Showing the loss of light when using the ideal artificial-daylight screens with the tungsten lamp operating at 7.9 lumens per watt. ing at 22 lumens per watt. This fact has been taken advantage of by the author in developing daylight units. These curves show the increased daylight efficiency of the tungsten lamps operating at higher luminous efficiencies. This is further illustrated in Figs. 101 and 102. In the former Ei represents the luminosity curve of the eye for light from a tungsten lamp operating at 7.9 lumens per watt, that is, the relative light values of the rays of various wave- lengths. This curve may be found directly or by mul- 232 COLOR AND ITS APPLICATIONS tiplying the mean luminosity curve of the eye (Fig. 93) for equal amounts of energy of all wave-lengths by the amounts of energy of various wave-lengths in the spectrum of the light under consideration. In this case it is the 7.9 lumens per watt tungsten lamp whose spectral energy distribution is found in Fig. 5. On multiplying curve Ei by the transmission values of curve a, Fig. 100, curve a' is obtained. The 100 90 >40 §30 UJ. or 20 10 / \ / f \ / \ 1 \ / \ 1 / \ / \ / c' V / /. b' 'N \, \ ^ ^ y 5^ "^ ^ 040 044 04& Q52 036 0.60 064 0.6d 07? Fig. 102. — Showing the loss of light when using the ideal artificial-daylight screens with the tungsten lamp operating at 22 lumens per watt. areas under curve Ei and a' are proportional to total luminous sensations, and the ratio of the area of a' to that of El represents the skylight efficiency of the 7.9 lumens per watt tungsten lamp as based upon the foregoing computations. The reduction in lumi- nous intensity when screen b is used with the source under consideration is found on comparing b^ with Eij in Fig. 101, and the ratio of the areas represents the sunlight efficiency of the 7.9 lumens per watt tungsten lamp. The corresponding data for screens COLOR IN LIGHTING 233 b and c used with the 22 lumens per watt gas-filled tungsten lamp are shown in Fig. 102, where E. repre- sents the luminosity curve of the eye for this tung- sten light. Screen b produces skylight and reduces the luminous intensity an amount represented by the difference between the area of &' and E2 in Fig. 102. Screen c produces noon sunlight with an efficiency represented by the ratio of the area of c' to that of E2. The daylight efficiencies for the two lamps con- sidered in the foregoing were found by determining the relative areas. For the 7.9 lumens per watt tungsten lamp (vacuum type) the noon sunlight efficiency is 14% and the skylight efficiency 4%. However, for the 22 lumens per watt tungsten lamp (nitrogen-filled type) the corresponding values are considerably higher, being 25% and 13% respectively. It has been found in actual practise that the consideration of 0.42/i as the starting point for the computations just described conduces to a higher accuracy than necessary in most cases, therefore beginning with a screen of 100% transmission at OAd/jL the daylight efficiencies are very considerably increased. Under these circumstances for the 7.9 lumens per watt lamp the sunlight and skylight efficiencies are respectively 18% and 9% and for the 22 lumens per watt lamp 33% and 19%. It is thus seen that very accurate artificial noon sunlight can be obtained with an ideal colored trans- mission screen with the 22 lumens per watt lamp at an efficiency of 25% or at 5.5 lumens per watt. This is a higher efficiency than that of the ordinary carbon incandescent lamp operating normally at the present time. Artificial daylight sufficiently accurate for nearly all