Page:Journal of the Optical Society of America, volume 33, number 7.pdf/34

 larities in the spacing of the samples of the Munsell Book of Color. To this end, constant-value and constant-hue charts of the form found in the standard library edition of the Book of Color (25) were systematically examined by 40 observers using the ratio method (27) and totaling some 3,000,000 color judgments. The data were summarized in the form of averaged visual estimates of the correct notations of the hue, value, and chroma of each sample. These estimates were then published in Table II of the preliminary report of 1940 (28).

The present report is concerned with the remaining aim of the subcommittee which was to produce a psychophysical system of surface colors which should be based upon the above data and which should correspond as closely to the ideal psychological color solid as is consonant with practical usefulness. The general method for achieving this aim was: (a) To eliminate minor variations in the averaged visual estimates by drawing through plotted points representing them, smooth curves defining new loci of constant hue and constant chroma; (b) to extrapolate the chroma loci beyond the Munsell samples and out to the theoretical pigment maximum (22); (c) to extrapolate the hue loci as far as feasible, that is, to the 1/ and 9/ value levels; (d) to adjust the value dimension by applying a new formula which eliminates irregularities in the Munsell-Sloan-Godlove function (24); (e) to express in terms of the I.C.I. notation (15) (Illuminant C) enough surface colors corresponding to this recommended psychophysical system to define it adequately; and finally, (f) for the sake of comparison, to redesignate all current samples with the revised Munsell notation.

RECOMMENDATIONS

The charts and tables in this report constitute the definition and standards of the new system. These tables and charts are recommended for general use in determining the Munsell renotation of a given color sample when the I.C.I. (Y and x, y) specification is known, or for determining the I.C.I. specification when the Munsell notation is given. If the chromaticity of the given color is specified in terms of dominant wave-length and purity, the I.C.I. trichromatic coordinates (x, y) must be found first, and then with reflectance given (Y), the Munsell value, hue, and chroma may be obtained.

Figures 1-9 present 40 hue loci corresponding to the full complement of samples in the latest edition of the Book of Color (25) together with the locus for every even chroma. Standard I.C.I. Illuminant C, which approximates 6700°K, has been taken as neutral origin for both the hue and chroma loci. The aim has been to make, as nearly as feasible, both series of loci perceptually equispaced. Of course, the spacing of the hue loci is not even approximately the same as that of the chroma loci, nor for that matter of the value loci. While some data on the subject are available (29), (4), no attempt was made in this study to equate the dimensional scales. It was felt that the greatly altered notation would detract seriously from the utility of the proposed system.

The charts presented are located at the recommended value levels indicated on the figures and presently to be described. It should be noted that Figs. 1-9 do not extend at all points to the theoretical pigment limit. It seemed more useful to include here only the more frequently used portions of color space, and thus obtain the advantage of a correspondingly larger scale. Table I was read from large unpublished charts which extend to the pigment limits at all points and from which the fourth place could be estimated. The Y entries in this table must be multiplied by 100 to obtain Y percentages, as in Table II.

Table II presents the I.C.I. luminous reflectance Y equivalents (percent form) of the recommended Munsell value scale. For convenience in computation the value-step intervals are given to 0.01. It should be noted that the reflectances indicated are not absolute but relative to magnesium oxide; whereas the maximum at value 10/ was formerly 100 percent, it is now 102.57. Use of this relation facilitates results and also avoids the somewhat dubious conversion to absolute scale, by permitting Y determinations with a MgO standard to be converted directly to