Page:Journal of the Optical Society of America, volume 33, number 7.pdf/60

 falls at (V$X$—V$Y$) =0.06; (V$Z$—V$Y$) =0.085; and from 5/ up to 9/ which falls at (V$X$—V$Y$) = —0.10; (V$Z$—V$Y$) = —0.235. Figure 11 shows the appearance of the family of concentric constant chroma loci at the region of the 5/ value level in the Adams’ coordinates. Each chroma locus is seen to be not only of the same shape but also at the same distance from adjacent loci as measured along any radius.

The achievement of these relations not only involved the adoption of the decentering expedient but also some considerable local departures from the data. On the other hand, the specification of chroma is greatly simplified and a much needed basis is provided for its extrapolation beyond the data through much used areas to the theoretical maximum (22). This extrapolation was accomplished as follows: First a listing was made of the I.C.I. (x, y)-figures for the MacAdam limits corresponding to 40 hues at each of nine values. Approximations to these limits are given in Table IV. Then the new chroma loci were extended at equal intervals (in Adams’ space), up to the last even step which would come within the MacAdam limit (Table I).

When the system of ovoids established in Adams’ space was transformed into I.C.I. (x, y)space, the results at the various value levels were as shown in Fig. 12. All the loci are seen to be smoothly transitional at a given value level or from one value to another. The chroma loci in Figs. 1-9 and Table I are adjusted for the differences required by the recommended value scale as compared to the Munsell-Sloan-Godlove value scale.

The spacing on the I.C.I. diagram is drastically affected by the major departures from perceptual uniformity of the I.C.I. system. The marked bunching of the chroma loci in the purple region emphasizes the difficulty which was avoided by smoothing in the more regular space employed.

After charts for 1/ to 9/ value levels had been completed, visual comparisons of chromatic samples were found to agree very well with the new chroma loci. There was, however, a marked exception in the blue and purple-blue region, especially at the lower and intermediate value levels. Although the sample comparisons pointed to real discrepancies between 5.0B and 7.5PB, on either side of this rather narrow range the agreement was good. The most pronounced bulge in the entire sweep of the chroma loci occurred at Constant Chroma Loci