The first paper in this series described a number of experiments on a single subject, carried out to investigate an anomaly observed during performance of the Kohs Blocks test. This anomaly consisted of making the model correctly but rotated. (Illustration 1.)

The first hypothesis adopted to explain this observation was that anomalies of perception are functions of disturbed figure-ground relations (Goldstein and Scheerer, 1941). This hypothesis is not precise enough to permit a deduction which would make possible an exact experimental test, i.e., a deduction defining the conditions which would produce a measurable change in the rotation phenomenon. It was therefore necessary to arrive at more definite ideas about figure-ground relations. Such ideas were found in three hypotheses developed by Goldstein and Scheerer (1941) to account for variations in speed of carrying out the Block Design test. The first of these is that the square orientation (e.g., Cards a and b in Illustration 2) of a figure would tend to diminish the frequency of appearance of the anomaly, while the diamond orientation would increase it (e.g., Cards c and d in Illustration 2).