Page:Journal of the Optical Society of America, volume 30, number 12.pdf/49

 level, the resultant probably would remain an unsuitable scaling unit for the solid as a whole. This is because the size of each unit seems to exhibit some tendency to vary from one part of the color solid to another. The best single scaling unit would seem to be some kind of equal-contrast unit like the National Bureau of Standards unit (36), but that is a consideration for the future. At present, it seems unwise to attempt any changes which would do away with the well-known and useful Munsell notation. This notation is briefly described here for the convenience of any reader who does not happen to be familiar with it. In general the notation correlates with the coordinates of the surface-color solid described in the first paragraph of this paper. Thus, hue is indicated by angle about the neutral axis, value corresponds to distance above the base plane containing the point representing black, and chroma corresponds to distance from the neutral axis. A number and one or two letters indicate location of the given color in the hue circuit which is divided into 100 hue steps. Each of the 10 principal Munsell hues (R, YR, Y, GY, G, BG, B, PB, P, RP) may be accompanied by any number from 1 to 10 to indicate the lesser variations or steps in the circuit. The principal hues are understood to fall at positions indicated by the number 5, and it is conventional to omit this number in the hue notation. Any hue with a designation greater than 5 lies farther along in the above hue series (clockwise direction) while designations less than 5 indicate counter-clockwise departures from the principal hue. Thus, 9R is a yellowish red, for it departs from R (or 5R) by four hue steps in the direction of yellow-red (YR). On the other hand, 2R is a somewhat purplish red because it lies three steps in the counterclockwise direction toward RP. Numbers ranging from 0 to 10 indicate location on the value scale. Thus, value 1/ indicates a near black, 5/ indicates middle gray, and 9/ a near white. The extreme values or ideal limits, 0 and 10, are not realized in practice. Numbers ranging upward toward theoretical limits represent degrees of chroma. Upper limits in practice depend upon the availability of suitable pigments. Zero chroma means an achromatic or gray color in which hue is absent. In the complete notation of a color, the hue designation is given first, followed by a fractional form of which the numerator is the value designation and the denominator is the chroma designation. Thus, for instance, 7R 3/8 specifies a color which is predominantly red in hue with a minor yellow component, somewhat low in value, and only moderately high in chroma.

Incidentally, the spacing of the samples of nearly neutral colors on the constant-value charts of the Book of Color is intentionally wider than that suggested by the psychological color solid; this distortion is a practical expedient to provide sufficient space for the inclusion of low-chroma samples.

As early as 1935, some chromatic smoothing was accomplished by Gage and by VanArsdel. The latter’s analysis was confined to yellowish hues but showed also a connection between the Munsell samples and samples of maximum chromatic efficiency.

In the same year, the 1929 Munsell constant-value samples were plotted in three different colorimetric-coordinate systems, and five years later in a fourth system. Illustrative charts of given (5/ value) data plotted in these several systems of coordinates are presented as. The location corresponding to Munsell neutral five (N 5/) is indicated by the open circle near the center of each chart, and that corresponding to I.C.I. Illuminant C by the small triangle. Of course, the constant-hue loci are oriented differently in each chart. These trials were made with original samples, unadjusted by any visual smoothing, the object being to discover a coordinate system which would effect minimum distortion of the concentric circular spacing proper to constant chroma loci at the various Munsell value levels. Such a system would facilitate the smoothing operation, partly because a circle is easy to draw and partly because departures from a circle are easy to estimate.

The Glenn-Killian determinations for value 5/ are shown in  plotted on the I.C.I. (x, y) diagram. Decentering and elliptoid forms of distortion are obviously present and they persist in varied amounts at the other value levels. The U.C.S. (35) chart shows the spacing, if anything, to be less uniform than in the I.C.I. chart. A great improvement is evidently effected by the use of Judd’s modification of his U.C.S. system. However, the fit was nearly optimum around value 5/ shown in and there is considerable and increasing distortion toward the extreme value levels. The remaining diagram illustrates Adams’ adaptation of the I.C.I. system in which X — Y is plotted against Z— Y. This diagram yields the closest approximation to