62 G. S. HALL AND J. JASTROW : STUDIES OF RHYTHM, I. would the judgment incline to the phis or rn/nn* side. So, too, inserting the variable between two invariable and like intervals greatly facilitated judgment, which between two unlike terms is far less accurate. D. and S. made each twenty judgments when the middle interval was varied ^ of the 4-27 sees, of the extremes, ten times each way with no error. G. S. H. judged ninety times under the same conditions with no error, while J. J. made only twelve errors in ninety judgments. When the variation of the mean was y-i-jy of the same time of the extremes, D. and S. made no errors in ten judgments, J. .1. made three errors in forty judgments, and G. S. H. made two errors in thirty judgments. These latter judgments and the effort to ' hold time ' which they involved were extremely fatiguing, and yet occasionally a judg- ment would be rendered with far less than the usual degree of attentive effort, and such judgments seemed hardly less likely to be correct than the most laboured ones with many muscles in- volved in the repressed but often quite compounded ' time-beats '. Confidence in the power to judge the finer intervals, or in the correctness of a judgment when made, diminished greatly as the differentiation required was hard, and surprise, when a short series was found at the end to be mostly correct, was almost invariable. C. Full and Vacant Intervals. A third set of comparisons was made. It is well known that if a horizontal line be bisected in the middle and one half un- touched and the other half crossed by short regular perpendicular lines, the latter half will seem the longer. It was found that under certain conditions the same illusion held for the time- sense. The intervals are arranged as described in the preceding paragraph, only there are but two of them. Of these the first is set full of cogs which give a corresponding number of click- they pass under the quill. In this case the illusion was invari- able. Full tables were constructed for four individuals. With 10 clicks the following vacant interval to be judged equal to it must be extended to the time of 14 to 18 clicks. 15 clicks seemed equal to the time of from 16 to 19. Preliminary experi- ments upon other individuals indicate that these differences are extreme. If the absolute length of interval is increased beyond from 1 to 3 sees., the illusion is less. It is also less if the cl are very near together. The illusion still holds, but is diminished, if, instead of comparing clicks and a vacant time, more or less frequent series of clicks are compared. In these observations also, the time between the two intervals became quite im- portant. In general the illusion was less if this time was short, but if less than about f of a sec. the illusion again became greater. Indeed in a few cases an indifference-time was found in which little or no illusion took place. This entire illusion, however, is reduced to a minimum, and with some persons vanishes, if the order of the terms be reversed, riz., if the vacant or less-filled interval precedes.
