Münsterberg asserts that such continual change is disturbing and confusing (p. 150), but it does not appear that he gave the method an extended trial.
Now the difficulty with the method of gradations is, that if we make the size of the steps roughly proportionate to the distance between the terminal tones, so as to cover with the variable approximately a relatively like stretch within each interval, with the same number of steps, we introduce a sort of 'dressur' into our experiments; the observers, unless warned by a marked change in interval, become used to judging 'middle' at a certain point in the course of the variable.
But if the excursions of the variable are not made with about the same number of steps in the several intervals, if for example we take ten steps in one interval to pass over relatively the same ground that in a succeeding interval we cover in twenty steps, we are destroying those conditions of mental equipoise which are indispensable for trustworthy judgments. In short, regularity of gradation leads to mechanically formed judgments, and irregularity of gradation, between different rows of steps, destroys the conditions for forming trustworthy judgments.
Again, it has been customary to give no further thought to the size of the steps, than to see to it that they were 'minimal,' and then perhaps to arrange them so as to make possible a symmetrical series of experiments during the hour of experiment. Indeed it is a curious feature of psycho-physical experimentation in general, that so little attention has been given to the size of the steps in gradation methods, or what amounts to the same thing, the rate of change of the variable. Initial threshholds, and difference threshholds have been determined time after time with no farther definition of the step of the variable than is implied in the vague term 'minimal,' and yet every experimenter has found that in determining difference threshholds for different intensities of any given modality of sensation, he had to make the size of the steps approximately proportional to the intensity of the corresponding terms in order to get consequent results. The size of step of the variable seems also to affect the judgment in the method of mean gradations, and this influence, joined to the stereotyping process spoken of above, was so strong as to make possible, even with so critical and conscientious an observer as Dr. Külpe of the Leipzig laboratory, a regularly recurring series of middle judgments of sound intensities for almost any point in a comparatively wide stretch of the path of the variable, that the experimenter might fix upon.
