ROTATING A DIELECTRIC IN A MAGNETIC FIELD. 129 



required by means of a potentiometer consisting of 100 resistance coils, each of 

 100 ohms resistance. A known P.D. was applied to the 100 coils in series and the 

 deflection due to the P.D. on one or two of the coils measured. 



(3.) Method of Experimenting and Results. 



The method of performing an experiment was very simple. The motor was started 

 and the water set running through the water-jacket and the rate of revolution 

 measured. The electrometer quadrants and outside coating of the rotating cylinder 

 were then insulated hy pulling up the rod out of the mercury cup. The scale reading 

 of the electrometer was then observed, and, if it remained steady, a current was passed 

 through the solenoid and the resulting deflection noted. The current was then 

 reversed several times and the corresponding deflections measured. The rate of 

 revolution was measured again and then the motor stopped and the sensibility of the 

 electrometer for quantity tested. 



The difficulty in making these observations was that the index of the electrometer 

 often did not remain steady, but wandered about in a more or less irregular manner. 

 The adoption of the water-jacket effected a great improvement in this respect, but 

 the chief point to be attended to in order to get the index steady was found to be 

 the adjustment of the sliding contacts. In the first place it was found necessary to 

 have the surfaces on which the brushes pressed turned true very carefully, and the 

 bearings adjusted, so that there was no shaking when running. The trueness of 

 these surfaces was tested as follows : the end of the lever attached to the rod 

 carrying one of the brushes was observed with a low-power microscope and the 

 cylinder slowly rotated. If the surface was not true or the bearings loose, the lever 

 moved up and down. The Wirings were adjusted and the cylinder repeatedly 

 re-turned until the levers remained steady on rotating slowly. 



Another thing which required careful adjustment was the pressure of the brushes 

 on the moving surfaces. If the pressure was too great the surfaces got hot and a 

 pyroelectric effect was produced which caused a continual drift of the electrometer 

 index. In the earlier experiments the outside surface of the ebonite was coated with 

 graphite to form a conducting coating. But after running a few minutes the graphite 

 got rubbed off and the ebonite under the brush caught fire. A metallic coating 

 therefore had to be used to conduct away the heat generated by the friction. 



If the brushes were not sufficiently tightly pressed down the index also drifted 

 continually, but in the opposite direction to the drift which appeared to be due to 

 heating. The drift, when the brushes were not sufficiently tightly pressed down, 

 was especially rapid when the surfaces were not true or the bearings loose. The 

 cause of this drift is not certain, but it seems probable that it was due to the brush 

 jumping on and off the surface and to particles of matter being torn off at each 

 impact and carrying a charge away with them. This theory was confirmed by 



VOL. cciv. A. s 



