OF ARTS AND SCIENCES. 195 



The amplitude of the vibration was fixed to give about one tenth of 

 a second to the charging, one tenth of a second between the charge 

 and discharge, and one tenth of a second to the discharge into each 

 other, at last leaving the condenser plates in communication thiough 

 the galvanometer. 



The proper insulation of the wires at the key gave some trouble, but 

 was at last secured by imbedding the wires in ijai'atfine at all points of 

 contact with the key, and placing the drops of mercury at the bottom of 

 deep grooves in a paraffiue block, separating those in the same groove 

 from each other by a transverse groove of greater depth. In fairly dry 

 weather, this arrangement gave satisfactory results. This key could 

 be used to charge and discharge a single condenser ; in fact, in the later 

 experiments it was found better to use but one condenser, regulating 

 its size to make the deflections manageable. 



With the apparatus thus arranged and tested for connections and 

 insulation, the labor of experiment was comparatively light. The con- 

 densers were charged and discharged several times, the electro-magnet 

 meanwhile being inactive. The dynamo was then set in motion, the 

 magnet excited, and the condensers charged and discharged as be- 

 fore. Usually this process was repeated several times before making 

 any change in the apparatus, as will be seen in the tables which follow. 

 Repeated tests were made to determine whether the dynamo or the 

 electro-magnet by their direct action would affect the scale reading, but 

 no motion of the index line could be detected when the machine was 

 started and stopped by an assistant. This test was repeated, with the 

 battery, condenser, and galvanometer circuits joined, in every combina- 

 tion assumed by them in the course of the experiment. 



The magnet circuit was always closed. The charging battery was 

 disconnected from the condensers ^^ of a second before the condensers 

 were put upon the galvanometer. 



In these experiments no attempt has been made to obtain accurate 

 quantitative results, and as no care was taken to adjust the galvanome- 

 ter to the same degree of sensitivcMiess, the readings in different series 

 of experiments are not comparable. 



The directive power of the earth upon the needle was very nearly 

 neutralized by magnet3 placed near it, so that the torsion of the sus- 

 pending fibre, which is constantly changing both as to direction and 

 amount, comes to have a larger influence than is desirable upon the 

 results. It is fortunate that the influence of a progressive change of 

 this sort may be largely eliminated by a suitable order of experi- 

 mentation. 



