Page:Journal of the Optical Society of America, volume 30, number 12.pdf/87

 percent of the 2B type) is so close to that of the former unmixed dye that, when the two curves are drawn by the fountain pen of the Hardy-G. E. instrument, they overlap too much to be separated. Making computations on the ICI system of colorimetric specification, the 10 percent addition is found to cause a change of about 0.0014, 0.0004 and 0.0017 in the x, y and Y values, respectively. These values were obtained when all errors due to weighing and hygroscopicity of samples and to making up solutions were eliminated; some other errors were eliminated by reading directly on the instrument counter at selected wavelengths. If these differences are considered as errors due to a single impurity of known character, and the indicated precautions are taken, then the performance of the instrument, using solutions, becomes comparable to that of the eye, using dyed skeins. But these conditions are ones which cannot be used in routine standardization. In particular, the number and nature of the impurities are often unknown and variable; also, it is inconvenient to eliminate errors by measurements on the standard every time the corresponding lot comes up, or by reading on the counter. The time element, as well as serious difficulties of “levelness” and incompleteness of “exhaust,” makes measurements on skeins or pieces wholly impracticable; but measurements of this sort will be reported. It is our experience that the majority of cases fall between the two extremes discussed; but enough fall near the “similar-curve” extreme to make standardization “for shade” (chromaticity) frequently very unsatisfactory when compared to standardization by “dye testing.”

22. Noticeability of Color Difference in Daylight., Eastman Kodak Company.

The probable error of two component additive visual color matching has been adopted as the most reproducible criterion of the noticeability of color differences. The probable errors of matching a series of colors are proportional to the corresponding just noticeable differences, within the rather great uncertainties of the latter. An apparatus has been employed in which the color of each half of a two-degree circular field can be varied corresponding to the points along any of a very great number of straight lines in the chromaticity diagram. The luminance of each half of the test field remains constant for all of the additive mixtures of the light from any two of a set of over one hundred color filters, all of which have equal luminous transmittances for the quality of light incident upon them in the instrument. A surrounding field of forty degrees diameter can be uniformly illuminated to any desired luminance and chromaticity. Over 18,000 color matches have been made by one observer, with 15 millilamberts luminance in the test field and 7.5 millilamberts of daylight quality (ICI illuminant C) in the surrounding field. The probable errors of determination of position along specified straight lines in the chromaticity diagram can be shown as functions of the positions on those lines corresponding to the chromaticities of the colors matched. The probable errors of purity determinations for representative dominant wave-lengths and their complementaries (including purples) reveal a direct relation between the noticeabilities of purity differences for nearly neutral complementary chromaticities. The noticeabilities of chromaticity differences along straight lines close to the spectrum locus and the boundary of the purples are very simple functions of distance along those lines. A curve for the probable error of wave-length matching under conditions of automatically constant luminance of the spectrum has been deduced from the observed data and shows only two minima, at about 486 m&mu; and 582 m&mu;. The probable errors of matching white (illuminant C) with mixtures of complementary colors can be represented by points on an ellipse around the point representing white in the chromaticity diagram. Similar ellipses represent the observed probable errors of all kinds of two-color mixtures matching eighteen other chromaticities well distributed over the chromaticity diagram. The major characteristics of all of the curves and loci representing the results for the principal observer have been confirmed by less extensive series of color matches by other observers.

23. ., Lamp Development Laboratory, General Electric Company, Nela Park.

The Maxwell triangle usual in colorimetry is predicated on a symmetrical relationship among the three components of the Young-Helmholtz theory. There is no reason to treat them symmetrically; the Munsell color system is one of cylindrical coordinates about the axis of value, the 1931 ICI system attaches all the luminosity to the Y component, and observations of chromaticity differences are made after equating brightness. By plotting X and Z values for colors of the same luminous reflectance (albedo, Munsell value) it is found that a suitably chosen ratio of scales (threefold greater for X) gives a nearly uniform radial and circumferential spacing of the Munsell colors, centered about the neutral color of equal value. For measurements made under ICI standard illuminant C (as were those of Glenn and Killian) normalization by division by (respectively) X and Z for standard illuminant C gives X$e$. and Z$e1$, and a scale-factor of approximately. Plotting Z$e$-Y against X$e$-Y brings the neutral color for each value of luminous reflectance to the origin of coordinates. Approximate register of colors of equal hue and chroma, and unequal value, is obtained by converting X$e1$, Y$1$, and Z$e$ to V$x$, V$x1$ V$y1$ using the Munsell-Sloan-Godlove value function, and plotting V$z$-V$y$ against V$x$-V$y$. These subtractions are the mathematical representation of inhibitory nervous connections (synapses) in the retina, according to the author’s theory of color vision.

Bibliography: E. Q. Adams and P. W. Cobb, J. Exp. Psych. 5, 39 (1922). E. Q. Adams, J. Opt. Soc. Am. 6, 932 (1922); Psych. Rev. 30, 56 (1923). D. B. Judd, J. Opt. Soc. Am. 23, 359 (1933). ‘A. E. O.Munsell, L. L. Sloan and I. H. Godlove, J. Opt. Soc. Am. 23, 394, 419 (1933), D. Nickerson, ‘‘Use of I.C.I. tristimulus values in disk colorimetry,” U. S. Department of Agriculture (1938). W. D. Wright, Nature 146, 155 (1940).