ELECTROMAGN'ETIO UNIT OP ELECTRICITY TO THE ELECTROSTATIC UNIT. 611 
The other end of the armature carrie.s a fork in which a small disc can rotate. When 
the magnet is not excited the disc presses against a roller covered with printer’s ink. 
When the magnet is excited the disc presses on the paper tape as it passes underneath 
it. The disc is kept inked bj revolving the inking roller. 
The general disposition of the apparatus will be seen from the figure. 
As the tape is drawn along we get two series of marks, one due to the commutator, 
the other to the clock. By comparing the position of the marks made by the 
commutator with those made by the clock we get a very accurate method of timing 
the commutator. We can find the time at which say 15 commutator contacts are 
made, and then, after the lapse of 5 or 10 minutes, find the time at which another 
15 commutator contacts are made. Subtracting the one set of times from the others 
we get 15 intervals which should, of course, be equal. Taking the average we get a 
very accurate value for the time occupied by (say) GOO revolutions of the cog wheel. 
The paper tape was at first drawn along by a Morse receiver, but as this did not 
act very uniformly it was replaced by a small winding arrangement driven by the 
same motor as the commutator. This ran perfectly unifin'inly. 
The method of experimenting was as follows : when the observing forks had been 
running for a few minutes, so that everything had got steady, the beats between the 
standard fork and the auxiliary fork were counted. The forks were so arranged that 
20 beats occurred in about 65 seconds. The motor was then started and regulated 
so that the commutator could be kept at balf the speed of the fork, i.e., about 
32 revolutions per second. 
The recording apparatus was then started and allowed to run for 5 or 10 minutes, 
the commutator being kept at a constant speed by observing the stroboscopic pattern 
through the fork. The apparatus was then stopped and the beats again counted. 
Our reason for adopting this method was that we had originally intended to 
measure the speed of the commutator while we were making the electrical observa- 
lions. This was, however, found to be too laborious, and had to be given up. The 
Laboratory clock was compared with the clock belonging to the Cambridge Philo¬ 
sophical Society, which is regularly rated from the Observatory. 
The last three observations, when the apparatus was working very satisfactorily 
give for the rate of the standard fork at 16° C. 
December 19 (tape running for 6 minutes) . . . 128H081 
February 14 ( ,, ,, 10 ,, ) . . . 128‘0909 
„ 15 „ „ „ „ ... 128-m6 
Mean.128 HO 45 
Thus according to our observations the fork is slightly slov/er than when used by 
Lord Rayleigh. This is what might be expected from the secular jftening of the 
steel. 
4 1 2 
