622
SIDNEY M. NEWHALL
significant because, as pointed out by Judd,[1] amounts to the direct application of the definitions of the attributes of color-perception. Until other methods are proved comparable, only the ratio method is strictly applicable. Then, too, precision psychophysical threshold methods of the traditional sort (20, p. 23) were not considered feasible, partly because of the great length of time required and partly because the inescapable preliminary problem of the equality of supraliminal magnitudes representing equal numbers of just noticeable differences still remains to be solved (5, p. 286; 20, p. 143).
Fig. 4. Examples of Munsell constant-value charts and constant-hue charts mounted on white, gray, and black grounds; shown (A) without and (B) with several types of masks used to facilitate visual estimates by exposing simultaneously only samples designed to differ in a single attribute.
Procedural details
The observers were instructed to use daylight or its equivalent, to illuminate the samples at 45°, and to view them along the perpendicular as recommended in 1931 by the International Commission on Illumination (33). They were also requested not to fixate a given sample for long but to keep the fixation relatively mobile; for preadaptation is an important factor (29, 37, 42, 55, 71, p. 544). They made visual estimates of hue, lightness, and saturation on the Munsell constant-value and constant-hue charts. All of these charts were mounted and viewed on three different, approximately neutral grounds of relatively high, medium, and low reflectance, respectively; see Fig. 4. The I.C.I. tristimulus specifications (33, 38) of these white, gray, and black (matt cardboard) surfaces were determined under artificial daylight illumination approximating 6500°K.[2] The daylight apparent reflectance relative to magnesium oxide (Y) and the trilinear coordinates (x,y) were, respectively, as follows: white, Y=0.852, x=0.3204, y=0.3292; gray, Y=0.253, x =0.3140, y=0.3244; black, Y=0.040, x=0.307, y=0.312. Masks of the same materials, Fig. 4, were made and used to expose separately the various rows, columns, and arcs of samples designed to yield colors constant in two attributes and regularly varying in the third. These grounds and masks were of some service in controlling the psychological phenom-