ORIENTATION AND BACKGROUNDThe object of this experiment was to proceed a step further, along the line followed by Thompson (i), in investigating age differences in preference for rectangular proportions. While his investigation was limited to the study of constant length (but varying size), ours provided also for analysis of the influence of the size factor, He had four chronological-age groups-preschool, third grade, sixth grade, and college-rank rectangles on the basis of aesthetic preference. The rectangles, 12 in all, were of black cardboard and uniform length of 2f in., ranging in width-length ratio from .25 to .75, with a duplicate at .40, by ratio steps of .05. Among other things his data revealed certain characteristic differences between the preferences of the two school-age groups-the third and sixth grade-and the college group. (The preschool group showed no consistent preference for any of the rectangles.) In terms of median preference ranks the school-age groups-except for a slight inversion between the two narrowest rectangles-showed an increasing preference for the wider rectangles. The college group, on the other hand, and this is in line with the general findings of other investigators working with adults showed an increasing preference for rectangles only up to width-length ratio of .55, though with continued high preference for ratios .60 and .65 (in the region of the 'golden section'), followed by a marked dropping off in preference for the higher ratios.It occurred to us that inasmuch as the size, i.e., area, of the rectangles increased with the width, the Ss might have been influenced in their judgments by size, as well as by proportion, especially as they had been instructed simply to pick the rectangle which they liked best, next best, etc. The present experiment, which employed two series of rectangles-one with length, and the other with area, held constant-was designed to investigate this possibility.
PROCEDUREMaterials, presentation and directions.-The materials consisted of 12 black cardboard rectangles, a white card to display them on, and a covered box to deposit them in. The rect-* The significances are only approximations, as the standard errors of the medians on which they are based were computed from ranked data.
