Everyone is familiar with those studies conducted during the 1950swhich promised an answer to such tantalizing questions as, 'Who are better judges of people, men or women? Psychologists or laymen? Leaders or followers?' Everyone knows too that satisfactory answers were never given because the investigators ran into almost insuperable difficulties of measurement which were nearly all peculiar to the problem they were trying to solve. Several reviewers pointed out the snags at the time (Cronbach, 1958;Bronfenbrenner, 1958;Cline, 1964). Subjects in 'accuracy' experiments, they said, use scales differently from each other, apply their own stereotypes, assume others are similar to themselves, or employ peculiar response sets. According to these reviewers, the 'naive' experimenter confounds all these factors in his score of accuracy and the results are meaningless. But how does one avoid 'naivete'? No one came up with a practicable answer and virtually all progress in the area has been at a standstill ever since.This article hopes to demonstrate that although these criticisms by Cronbach and others may have been valid they ought not to have presented insuperable difficulties to investigators. The basis of this statement is simple. Cronbach and others, by placing emphasis on the means by which an experimenter must arrive at a 'pure' score of interpersonal accuracy, clouded an issue which is logically antecedent. How meaningful did these naively analysed test procedures make the laboratory task of judging others in the first place? I n general, these laboratory studies of the accuracy of interpersonal perception used to (and recent ones still do) set subjects one of two tasks. There is the 'guess the quantity of a trait' task and the 'guess what he would write' task. I n the 'guess the quantity of a trait' task, subjects are given a list of personality traits like extroversion, intelligence, neuroticism, etc., and they are required to state the quantities in which these are possessed by certain stimulus persons. Often a comparison is then made between J. Theory SOC. Behaviour I, 2 . Printed in Great Britain