Page:Journal of the Optical Society of America, volume 33, number 7.pdf/65

 The value adjustments were made on the neutral or near-neutral samples with the assumption that a constant value-reflectance (V to Y) relation holds for all colors regardless of chroma.

The heavy curve in Fig. 14 represents the recommended value function, while the other curves have been at some time or for some reason considered indicative of the relation between reflectance and value. It should be noted that the Munsell-Sloan-Godlove scale, which is represented by the finer line, is not far from the curve of the recommended value scale. A double adjustment was made, the nature of which deserves some explanation.

The recent National Bureau of Standards (21) measurements of the neutral samples indicate reflectances higher by 6 percent or more than the averages of Munsell-Sloan-Godlove. This considerable discrepancy may be due in part to the use of different reflectance standards but is not vet completely explained. Investigation has revealed that it was not due to failure by the painter of the samples to reproduce to a narrow tolerance the specifications furnished him. The indicated double adjustment consisted in (a) increasing the reflectances reported by MunsellSloan-Godlove by 6 percent for values 2/ to 8/, inclusive, and (b) making the reflectance scale relative to magnesium oxide taken as 97.5 percent (26). This adjustment amounts to multiplying the Munsell-Sloan-Godlove reflectances from 0/ to 9/ by 1.0871 and then smoothing to 10/ which is set at 102.56 percent in order that value 10/ be equivalent to 100 percent absolute reflectance. On such a scale Y values, in terms of MgO at 100 percent, may be read directly. Despite these adjustments and its superficially smooth appearance a curve drawn through these points was found still to contain a number of inflections. By trial and error, a quintic parabola was found which fits closely the adjusted MunsellSloan-Godlove reflectances and has but one trivial inflection. This equation is:

Ry = 1.2219 V—0.23111 V$2$+0.23951 V$3$ — 0.021009 V$4$+0.0008404 V$5$.

This formula was employed in computing the various reflectances in Table II which presents I.C.I. Y (or percent reflectance relative to magnesium oxide) as a function of value.

A psychophysical system of surface colors has been developed from the extended observations by numerous observers of the Munsell Book of Color samples. This system is aimed toward the double ideal of practicability and perceptual uniformity. The necessity of considerable reliance on color judgment, the scattered data, and extrapolation make it clear that this system is to be regarded only as an approximation to the ideal.

The visual conditions appropriate to the employment of the system are evident from the procedure followed in taking the basic data. The observers made their color comparisons by looking at groups of related samples on the Munsell type of constant hue and constant value charts. Thus, at the time of an observation, the observer’s eyes were adapted to the chart as a whole rather than to, for example, I.C.I. Iluminant C. This is the usual situation in the visual specification of a color sample by reference to a standard chart, a type of situation which has been discussed by Helmholtz (11), Helson (12), and Judd (18).

During the experimental investigation of the subcommittee, no simultaneous comparisons were made of samples on backgrounds of different reflectance. White, gray, and black grounds were used at different times, but a single ground was always used in observations of a given set of samples. This uniformity of background is, also, a fairly usual situation for color comparisons. The separate use of the several achromatic grounds explains the lack of influence of ground evident in the hue and saturation estimates of our observers. The striking effects of ground under other viewing conditions (18), (12), (13), were absent here, presumably because of the relativity of the judgments and the tendency toward constancy from adaptation (19). Lightness, of course, was significantly affected by ground in our work, for the comparison of lightnesses always included the achromatic ground itself. Even here most of the effect was due to the black ground while the white and the gray were in fair agreement.

The question whether to make the system right for trained or untrained observers was raised in the preliminary report. The possibility of significant differences in the dimensional relations reported by visual color experts as compared with others seemed worthy of consideration. It