Experimental studies of rhythm and time: II. The preferred length of interval (tempo)
The question as to what particular tempo, or length of interval between auditory impressions, is felt to be the most pleasing, or the most favorable for rhythmizing sounds, has been investigated by various methods, and conclusions have been reached which have not always been in complete accord. It seemed to the writer that the problem ought to be investigated afresh by means of different methods. In the first investigation recorded here a method of paired comparison of the beats of a metronome
... ats of a metronome (Verdin's make) was employed. The notches on the pendulum of the instrument were so arranged that the following speeds could be obtained by properly sliding the pendulum weight: 40, 200 and 208. A few times, as the rates of the clicks became nearly equal, the weight was moved half way between the notches, so as to halve the difference between the two rates. The half-way point was not determined precisely, but the resulting errors are probably negligible. •An earlier article in this series on 'Qualitative Limens or Grades of Rhythm, and the Difference Limens in the Perception of Time,' appeared in the PSYCHOLOGICAL REVIEW for March, 1911. The remaining articles will detail the results of experiments on the estimation of the mid-points' between pairs of different tempos by two methods, and the grouping of metronome clicks into the maximum number of speed categories Some of the topics discussed here will come up again in these later articles. The metronome experiments described in this article were made in 1905; the observations of the rhythmical responses in the theaters were made from time to time during 1905 and 1906. 2O2 l In 1906 measurements were made by the fork of the rates 80, 96, 104,1x2, 124, 126 and 144 (beats per minute). Each interval was found -to be too slow by the amount indicated in parenthesis: .75 sec. (by .025 sec); .625 (.021); .577 (.032); .535 (.026); .484 (.023); .476 (.033); and .416 (.25). By consulting the table it will be noticed that die correction used for .577 was .009 sec .smaller than the empirical result, «nd X113 too small for .416 sec. These errors are immaterial for the purpose in hand. *In the 1906 measurements the 'break' records were likewise shorter by about the same amounts, except for die speeds (corrected readings) .507 and .775 sec.