Four groups of twenty isopods (Porcellio scober L.) were given 4 0 right turns and four groups were given 2 0 right turns in square mazes 1, 1.5, 2.5, and 4 cm. on a sidc. Another four groups were run equivalent distances without turns, and one group was given no treatment at all. Ss from all groups then received one trial in a T-maze. Left curns in the T-maze increased as a function of number of prior turns and decreased with distance (time) between turns. These results wet? obtained in a situation where "centrifugal swing" and "forward-going tendency" hypotheses provide no useful predictions. In comparing these results and procedures with those of other investigations, it is apparent that is rypically a weak determiner of choice-point behavior in isopods, but under special circumstances can be the principal source of variance.When an animal runs a short distance down an alley of a maze, curns to the right, then approaches a T. the probability of a left nirn is generally higher than for a right rurn. At least three differenc hypotheses concerning chis behavior have been presented. Schneirla (1929) argued thac the animal would swing out against the farther wall after the nlrn, and as long as the "centrifugal swing" acted, would nirn in the same direction. Dashiell and Bayroff (1931) suggested that the animals established a "forward-going tendency" during the start alley run, and the later nlrn was a correction. Hull's ( 1943) concept of reactive inhibition wich which the first right :urn results in a red~iction of the probability of its occurrence at the T , has also been used to account for these observations.A recent increase ill incerest in this problem has raised questions concerning the adequacy of these interpretations when applied to invertebrate maze behavior (Akre). Analysis of these studies suggests thac all three hypotheses have some utility; however, maze problems may be devised thac severely tax the utility of any of the three.This report considers a maze problem where the turning behavior is prcdictable using the In hypothesis, but neither the "centrifugal swing" nor "forward-going tendency" hypothesis yields usef111 predictions. The hypothesis yields predictions of an increase in left-rurn tendency with an increase in the number of prior right turns, and with a decrease in the distance (time) between prior right turns, and chance T-maze performance when the organisms m n an equivalent distance but make no prior curns. 'L. B. S. at Palo Alto VA Hospital. J. E. M. now at California State College at Fullerton.