A modification of the method of equal-appearing-intervals.

One of the obvious defects of the method of equal-appearing intervals has been the 'end effect,' which results in seriously truncated frequency-distributions for stimuli near the extremes of the list. This, in turn, results in difficulties in the estimations of central tendency and variability. It may have other, less obvious, effects in the form of distortions of the scale so that we do not have, after all, an equal-interval-scale. We propose here a procedure that is aimed at elimination of the 'end effect.' It was applied to a series of lifted-weights, with treatment of the data by the method of successive categories to determine how well equality of intervals was achieved. An incidental result is a test of the applicability of Fechner's law to lifted-weights.