Page:Journal of the Optical Society of America, volume 30, number 12.pdf/54

 which seemed to differ in chroma from the standard was then evaluated in terms of it, and a suitable radial vector was drawn to indicate the displacement ($$R$$ form). Subsequently, the masks were removed, and the radii of constant Munsell hue were examined for irregularities in the size of the saturation steps between adjacent colors. After selection of a representative saturation step, each of the other steps could be evaluated in terms of it ($$R'$$ form).

Saturation differences were estimated again, this time from the constant-hue charts. A constant-chroma column was masked off, and the $$R$$ form applied. After each column had been so studied, the masks were removed, and the equality of the chroma steps checked by the $$R'$$ form. Incidentally, most subjects found the applications of the $$R'$$ form to saturation the most difficult part of the entire procedure.

Lightness differences were also estimated from the constant-hue charts. Rows of constant Munsell value were masked off and deviations in lightness of each sample from the average lightness of the row were evaluated by the $$R$$ form, after which the masks were removed, and the size of the lightness difference between successive constant-value rows was estimated by the $$R'$$ procedure. These estimates were recorded numerically as the ratio of the actual size to the average size of value step. It will be noted that two color attributes were estimated from both types of chart, and it frequently happened that two vectors, each for a different attribute, had to be recorded for a single sample. If, for a sample on a constant-value chart, a radial vector had been drawn already to indicate a chroma deviation, the tangential vector to indicate a hue deviation would be started from the end or arrow-head of the radial vector. Thus the resultant of the two vectors would be indicated by the position of the final arrow-head, and would represent the difference between the 1929 Munsell notation of the sample and that estimated by the observer to indicate the color. In case of constant-hue charts an analogous combination of horizontal and vertical vectors would be required to record combined saturation and lightness deviations.

Complications

The unusual difficulties of the judgmental situation justify some account of the special problems reported by the observers.

(A) There was the problem of complex chromatic comparisons forcing, as it does, the abstraction of the attribute to be estimated. Attributive abstraction is peculiarly difficult in cases of strong association; for instance, when saturation is the judged attribute, hue being constant and lightness variable, there is a strong tendency to estimate darker colors as of higher saturation. This relation is often found in everyday life. A familiar related problem is that of comparing the lightness of colors differing in hue. A similar difficulty, especially for unsophisticated subjects, is that of comparing chromatic and achromatic colors in respect to lightness. The untrained observer usually perceives chromatic samples as lighter relative to achromatic samples than does the trained observer. Inspection of data from the constant-hue charts revealed the interesting fact that the vectorial lightness indications for the chromatic samples were rarely downward. The indication was usually that the chromatic sample seemed too light relative to the achromatic sample of the same Munsell value. (B) Observers making estimates of saturation difference by the $$R'$$ method reported greater facility when all colors departed definitely from neutral gray than when one of the colors was perceived as gray. The trouble seemed to be that a first step extending from zero saturation appeared as a qualitative step and so tended to give the impression of an infinite quantitative step. The second step was, on the other hand, a definitely limited quantitative step. Seven of the observers made a written memorandum to the effect that the first step was or seemed definitely too great.

Possibly related to this was a certain change of attitude correlated with presence or absence of gray. When observing saturation differences from gray some observers reported that they “just look for any difference” whereas between two chromatic colors of the same hue they think more definitely in terms of saturation. Such a change of attitude might be expected to occur on the basis of least effort. Looking for a mere difference