STUDIES OF EHYTHM, I. 57 of a kymograph. Around the entire circumference of these plates notches had been sawed 4 mm. deep and 2 wide at regular intervals of 2 mm. for one and 4 for the other half circumference. A hundred uniform brass slots, stamped out with a die, were made to fit these notches so exactly that they would go in easily with the hand and yet not be thrown out by the revolutions of the plates. These slots could thus be set into the notches to represent any interval or combination of intervals so far as the circumference of the plates would admit. This limit might of course be readily enlarged by increasing the circumference or by constructing two or more pairs of plates each with one uniformly distinct series of notches all the way round. Upon the upright iron beam which supports the shaft of the drum, was fastened a frame to hold large quill tooth-picks which were kept in position by a screw and clamp to play upon the slots as they rotated past. We could find no other substance which produces, when cut down to the proper form, clicks so sharp and distinct, even if the eyes or slots are very close together or the rotations very rapid, while offering so little resistance to the rotation of the drum. The upper part of the annexed cut (A) represents the screen and letters, and the lower (B) the simple aparatus for producing the clicks which we call a rhythmometer and which can be fur- nished by our University mechanic. When such an adjustment had been found that a semi-circumference filled with slots (*) moves under the quill (q) at exactly the same rate, measured by an electric tuning fork (/') of 50 vibrations per sec. on the drum above, as a semi- circumference with no slots, i.e., when the pres- sure of the quill producing the clicks did not retard the drum, and when a mm. scale had been pasted under the points of the fric- tion-wheel and the time-interval between two slots for each of several desired positions of the points determined once for all, observations could be begun. A. Counting. A number of cogs was set up by the operator (following no order of numbers) and one cog was put in as an avert issement at what seemed the most convenient interval of about f of a sec., and the observer sought to count the clicks. The drum was allowed to revolve several times till he had attained a satisfactory degree of certainty, when the record was made and another number set up. In the observations on which the Table on next page is based, the effects of fatigue are in large measure eliminated by beginning each series of observations with a small number of clicks, passing upwards, skipping from four to eight, to a maximum of two or three score clicks and then down again on the same numbers in inverse order and excluding all series which showed any .considerable deviation. In this way from three or four to ten observations on each number (more on the small than on the
