OF ARTS AND SCIENCES. 153 



of the pendulum, but we have found no trouble from leakage in these 

 experiments. The action in cases (2) and (3) will be sufficiently 

 evident from the above description. 



The weight of the pendulum rod being considerable, we found that 

 it was somewhat bent by its own weight when the pendulum bob was 

 supported by the latch, and when released a vibration was set up in 

 the rod on its own account, which caused a rapid vibration of the 

 disks in addition to the motion due to the pendulum. The result of 

 this was to produce an uncertainty in the beginning and end of con- 

 tact, depending upon the relative phase of the two vibrations at the 

 beginning and end of the arc of contact ; the amount of this uncer- 

 tainty was less than 0.0005 sec, and was greatly lessened by adding 

 steel wire stays to the rod when it became desirable to use the pen- 

 dulum for times approaching 0.001 sec. in magnitude. 



A similar effect, probably attributable to the same cause, was 

 noticed as producing a considerable error in a much larger pendulum 

 of similar construction, when used for times approaching 0.001 sec, 

 although the rod was of wood 1£ inches square in section, the bob 

 weighing 40 lb. and the time of swing being 1.8 sec, with an initial 

 displacement of 26°. 



We have used a more simple apparatus, shown in Fig. 3, in cases 

 where it was desired to charge a condenser for a definite interval 

 of time, the absolute value of which is required to be known with a 

 higher degree of accuracy. With the pendulum already described, 

 the time is liable to considerable error when determined by an arc 

 of contact nearly equal to the half arc of swing ; that is, when the 

 time of contact is in the neighborhood of 0.5 sec. 



Where different times are not to be experimented upon, a greater 

 degree of accuracy may be attained if the contact existing through 

 the upward swing be maintained through the downward swing as 

 well, beginning and ceasing nearly at the lowest point of the swing. 

 The time of contact is then equal to the time of vibration, affected 

 by a very small correction, which depends upon the departure of the 

 beginning and end of contact from the lowest point, and which may 

 be determined with considerable accuracy. 



The pendulum bob is of iron, and is held displaced from its position 

 of rest by one pole of a horseshoe magnet b, in contact with it at its 

 middle point; this magnet is attached to the plate c, which turns on 

 the screw d, so that the magnet maybe moved in the plane of vi- 

 bration of the pendulum by a cord leading to the observer al the 

 galvanometer. The pin e acts as a stop limiting this motion of the 



