Though previous studies had suggested a link between the rate of concept learning and instructional formats in which instances of concepts were held constant or grouped together, a 1964 study by Richard Anderson questioned the conclusion, suggesting that previous studies had been confounded by other irrelevant attributes that changed from trial to trial. The study was replicated to test certain anomalies. Via video screen, 50 undergraduate volunteers were delivered stimuli consisting of seven attributes: number, color, form, shading, vertical or horizontal bar, border, and field orientation. Subjects were given five tasks involving the learning of one pair of concepts. Five treatments differed in terms of the sequence of stimuli used to present the training. Results showed that: (1) when adjacent trials contained the same relevant cue, learning was maximized; and (2) the effects of alternate and constant series are conditional upon the number of stimulus attributes that change from trial to trial. (EMH)