Page:Journal of the Optical Society of America, volume 30, number 12.pdf/73

 from this origin are 100 lighter lines of average dominant wave-length. Twenty of these indicate approximately the average dominant wavelengths of the 20 Munsell hues, while the remaining 80 are spaced arbitrarily at equal angles between them. Wave-length is shown for reference only, the relation between hue and dominant wave-length not being simple. differs from the others in that it is not based directly on data but simply represents at the 9/ value level the results of the intervalue smoothing.

Discussion

The iso-chroma contours at all value levels evince several distinctive characteristics. They are of a definitely elliptoid form with the major axis approximately coinciding with the Y to PB hue plane. There is an obvious broadening out of the contours in the first (10GY to R) as compared with the third (10P to B) quadrant, a tendency which is especially marked for higher chroma. Sections of contours lying in the first quadrant (10GY to R) show the greatest separation while those in the second quadrant (R to 10P) lie closest together. These sweeping departures from uniformly spaced concentric circles, may be ascribed reasonably to the I.C.I. system itself. They constitute a preliminary indication of the nature and magnitude of the general departures of the I.C.I. (x, y)-diagram from a uniform-chromaticity-scale system. This interpretation is admissible on the ground that our table of visual estimates has revealed no such irregularities in the Munsell system. On the other hand, the fact that the /2 chroma contour exhibits a tendency to a wider separation from the neutral point than the chroma contours do from each other, may represent a real irregularity in the Munsell spacing. As already. mentioned, a number of observers had specially reported that the /0 to /2 chroma interval seemed too great. Other data on the relation of chroma to purity may be found elsewhere.

The experience gained in this preliminary smoothing of the chroma estimates from the constant-value charts should, for the final report, prove useful in the tridimensional smoothing of hue, value, and chroma from both the constantvalue and constant-hue charts. Visual data for tridimensional smoothing are now available in above.

This paper reports to date the work of the Optical Society of America subcommittee on the spacing of the Munsell colors. The ideal of this study is to produce a psychophysical specification of a system of surface colors corresponding as closely as possible to the true psychological color solid, but all along the general attitude of the subcommittee has been simply that any substantial improvement in spacing which might prove possible would be well worth while. The work of the subcommittee consists in revaluation by visual observations of the regular 400 Munsell samples of 1929, in order that they may conform more closely to the regular contours of the ideal solid. All samples were viewed against white, gray, and black grounds, independently.

A table is presented containing the averaged visual estimates of the numerous subjects who participated in the revaluation procedure. The unsmoothed corrections in the spacings are evident from a direct comparison of these data with the 1929 Munsell notation. The approximate validity of the latter is in general confirmed by the smallness and the distribution of the deviations.

Background reflectance is found to exert upon the lightness estimates a significant and systematic influence which is in the expected direction of simultaneous lightness contrast. There is evident from inspection no comparable general influence of background on either hue or saturation.

The plan is to smooth these averaged visual estimates in the I.C.I. colorimetric coordinate system for presentation in the final report. The smoothing operation is here illustrated by preliminary results for chroma, based on incomplete visual estimates with the constant-value charts.

The writer is particularly indebted to the staff of the color measurements laboratory of the Agricultural Marketing Service,, for substantial aid in the preparation of tables and figures.