NOT only are there many techniques for determining and defining scales of subjective magnitude, but scales determined by various methods are not related in simple ways. Under the circumstances attempting to determine the properties of resultant scales poses a vexing problem.As Torgerson (1958, p. 113) has pointed out, validation of a scaling procedure cannot be accomplished in general, but only with respect to a particular attribute. Thus our reliance on a scale, or confidence in a scaling procedure, and our decisions on how it might be applied will depend on accumulated empirical information as to how it functions in different contexts.The matter of individual differences has received little attention, but Stevens (1956) and McGill (1960) consider certain aspects of the problem. Epecially where numerical responses are required one might ask whether observed differences are a function of the manner in which individuals perceive the psychological dimension involved or the manner in which they use numbers in reporting these perceptions. Experiments involving cross-modality procedures are encouraging in that they have shown, at least for several dimensions, that 0's do make consistent judgments whether the responses are made in numerical terms or in terms of some other perceptual dimension (Dudek and Baker, 1957;Stevens, 1959).