It is obvious that the same disturbing influences would be active in reaching distance comparisons in all cases of regular gradation of the variable with successive stimuli, whatever the modality or quality of sensation. But to all these arguments the first part of Dr. Münsterberg's results appears a flat contradiction: his experiments not only show a close agreement in the position of the estimated middle of the tone interval at the arithmetical mean of the rates of vibration of the terminal tones, for the shorter intervals, but a remarkable accuracy in judging this point; thus for the interval 256-512 (1:2) the tone 384 appeared to observer B to halve the interval 100 per cent, and to observer A 90 per cent of the times it was sounded. If this is surprising, the close agreement with Lorenz's results is simply astonishing. It must be clearly understood that Dr. Münsterberg gets results which agree closely with Lorenz's and which are definite and regular, by using a method which Lorenz discarded, because he could get no definite results whatever from it. It is evident that the chief question at issue here is not one of results, but of method. None of the results can lay claim to validity until this contradiction in method is investigated and explained. But we are only told that the method of irregular gradations to which Lorenz was forced, appears of dubious value (p. 159), and then, without further discussion, the method with which Lorenz could get no results is put in use to get results agreeing with Lorenz's, which, in turn, were gained by a method condemned by our author (p. 159), and then, on the strength of this curious agreement, we are asked to accept generalizations from experiments on intervals not investigated by Lorenz.
As the matter stands, we do not feel that the opinions advanced above in regard to the invalidity of the method of regular gradations are negatived by those experiments of Dr. Münsterberg, nor do we consider that his elaborations on the method, e.g., the substitution of two pairs of tones for three tones, and the intercalation of tones in an interval of comparison, have anything more than a suggestive value. The problem of the comparison of tone-intervals is not to be solved in an off-hand way even by a man of Dr. Münsterberg's psychological insight; the first step towards definite solution must consist in a thorough-going investigation of the method to be employed.
The volume contains nine other studies : on Association (five short sections dealing with special questions); chain-reactions; investigations of Memory; the influence of the content of a time-
