Page:Journal of the Optical Society of America, volume 30, number 12.pdf/53

 ena of contrast and end-effect (,, , p. 546; , , , p. 567) due to field factors aforementioned. The manner of using the masks is indicated later.

The $$R$$ form of the ratio method here involved the direct comparison of several samples with respect to a given attribute while the $$R'$$ or difference-ratio form called for the comparison of the attributive differences between samples. Thus, in the one case the observer asked himself, “How much does each of these colors differ in this attribute from my chosen standard?” And in the other case, “How much do the attributive intervals between these various pairs of adjacent colors differ from my standard interval?” In both cases, the estimates of the departures from the standards were recorded in vectorial notation. The record forms used by the observers are shown in and are similar in shape and pattern to the color charts themselves.

The observers were instructed to represent their estimates of the kinds and degrees of departure by drawing vectors of appropriate directions and proportionate lengths. Outlines of the prescribed procedure for estimating hue, saturation and lightness differences are as follows.

Hue differences were estimated from the constant-value charts. The series of samples in a given radius of constant Munsell hue was isolated by masking the remainder of the chart, and the series was then examined for variations in hue. If such were observed, the effort was made to estimate a representative or average hue and indicate departures therefrom by tangential vector notation ($$R'$$ form). The same procedure was followed in turn for each radius of constant Munsell hue on the chart. Next, the hue spacings of the hue radii considered as wholes were examined, and if the hue differences seemed unequal, a tangential vector of length proportional to the estimated adjustment required was affixed to the end of the radius (R’ form).

Saturation differences, like hue differences, were estimated from the constant-value charts. The samples in a given constant-chroma ring were masked off, and the subject looked them over in an effort to select a sample of representative chroma to serve as standard. Any sample