Sttn7nzary.-The technique of the Rod-and-fmme Test as used by most psychologists is criticized for difficulties in control of head position, starting position effects, control readings, and the effect of instructions. Differences in results reported with Witkin's method and with other methods are discussed.Culver (1967, p. 377) remarked that "the Rod-and-frame Test has evolved over the last decade into one of psychology's most frequently used measures of individual differences." This has occurred, it appears, in spice of considerable criticism of Witkin's interpretations about relations between perceptual and personaliry factors (Zigler, 1963; Holtzmann, 1955).The present paper is not concerned with the status of the Rod-and-frame Test (RFT) as a quick personality measurement. Instead, attention is directed to the problems of che RFT as an experimental method p e~ J e and to the methodological errors made by chose employing it for research in both perception and personality.The original RFT, as performed by Witkin and his co-workers (Witkin, 1948; Witkin & Asch, 1948b), involved setting a lumino~is rod to vertical or horizontal. The rod was si~rrounded by a luminous square frame and was observed in a dark room. Ss performed the task in three positions: sitting erect, tilted to the left in a cilting chair, and tilted to the right. More recent work (e.g., Vaughc, 1965) has tended to employ only the body-erect position.This tendency may be due to fear of contamination of results by the A and E effects (changes in apparent vertical contingent upon tilting of the head), or it may simply arise from the infrequency of tilting chairs. In any case the present paper will attend primarily to methodology for the body-erect position. All commencs may be generalized to the body-tilted conditions, however.The basic procedure used by Witkin for the RFT was the following. S was seated in a dark room and shown a luminous square, tilted to left or right, within which was a luminous line. The line was positioned at "near-vertical" or "near-horizontal" (Asch & Wickin, 1948a, 1948b, or ac "near-vertical" only (Witkin & Asch, 1948b), or in varying positions of tilt (Witkin, 1948).S was instructed to tell E how the line should be adjusted in order to appear vertical (i.e., parallel with the walls or a flagpole). Clockwise and counterclockwise movements of the rod were alternated in some snidies (Asch & Witkin, 1948a). The errors made by S were apparently then analyzed without regard to their direction. No control readings were taken; error scores were calculated with respect to objective vertical rather than to the individual's subjective vertical.