Page:Journal of the Optical Society of America, volume 30, number 12.pdf/57

 zontal dimension represents hue while the perpendicular dimension represents chroma. The dimensions of a similar form used with the constant-hue charts represent value and chroma. On these forms each observer’s estimate of the deviations of the particular color in respect to two of the attributes in question have been recorded by inserting his initials in the appropriate place. A total of 800 record sheets were required to summarize the color. judgments, roughly three million in number.

The summarization forms are so similar that only the form for use with the data from the constant-value charts need be described in detail. Each vertical third contains data secured from the observations against a different background. In any third, entries falling in the central vertical column represent judgments that the Munsell hue notation is correct, while entries in parallel columns to the left or right represent, respectively, estimates of hue deviations in the counter-clockwise or clockwise direction around the hue circle. The recording unit of estimate is one Munsell hue-step. Similarly, the middle horizontal column represents the judgment that the Munsell chroma notation is correct, while the parallel columns above and below include respective estimates of “too strong” and “too weak.” The recorder’s unit of estimate is here 0.4 chroma-step. Of course, entries falling in the central rectangle, formed by the intersection of the central columns, represent judgments of correctness for both attributes. Preliminary estimates (encircled initials) by the $$R$$ method, and final or check estimates (unencircled initials) by the $$R'$$ method, were made on each of the three grounds. Consequently the initials of a given observer usually appear several times on the summary sheet for a given color. Inspection of Fig. 6 shows at once that the particular color sample, G 5/6, was judged on the average to have a chroma higher than /6 and hue corresponding closely to the Munsell notation, regardless of background.

The summarization process varied somewhat with the attribute. In case of hue, it will be recalled, the final instructions called for an unmasked inspection of the general spacing of the several hue radii in the constant-value charts, and the assignment of tangential vectors when adjacent hue radii, considered as wholes, seemed too close together or too far apart. The hue-shifts represented by such general radial adjustments were individually added to the results of the preliminary hue adjustments of the colors concerned. Incidentally, it may be noted, that only a minority of the subjects made any general adjustments.

In case of value, an analogous type of treatment was required. The instructions had called for an unmasked inspection of the spacings of the successive value rows in the constant-hue charts, and the indication of departures from perceptual uniformity by direct numerical estimates in terms of the standard value-step unit. The given observer estimated the eight spacings between values 1/ and 9/, and the results were added to his preliminary value estimates. If the