1895.] 



On the Question of Dielectric Hysteresis. 



471 



terminals. The charge that the condenser possesses at any time can 

 be ascertained by discharging it (by means of the switch) through a 

 ballistic galvanometer. 



Since the methods for testing for hysteresis consist in putting the 

 thing tested through cyclic series of states, it is essential that any 

 change of state involved in the measuring operations should itself 

 form part of a cycle. This was accomplished by making the cycles 

 in the manner indicated in Fig. 2, which has been drawn as it might 

 appear if considerable hysteresis were present. 



FIG. 2. 



Potential difference. 



Commencing at the point R in the cycle (i.e. with no difference of 

 potential between the terminals), the potential difference is gradually 

 increased until the point P on the curve is reached ; the condenser is 

 then discharged ; the portion of the curve PU is traced during the 

 " instantaneous " discharge, and the galvanometer indicates the loss 

 of charge PT, which includes whatever viscous flow takes place 

 during the time of throw of needle (2| seconds). While the galva- 

 nometer is still in connection with the condenser a further viscous 

 flow takes place, until, if the cyclic state has been set up, the start- 

 ing point B is arrived at. Before allowing the switch to again 

 connect the condenser to B (Fig. 1) the rheostat is unwound so as 

 to bring B back to A ; there is then no potential difference between B 

 and A, and the condenser receives no charge when the switch 

 completes connection. After this connection has been made the 

 potential difference is again gradually increased, but to a higher 

 value than before to S (say) ; it is gradually diminished till it has 

 the same value as at P ; we thus arrive at a point Q ; the condenser 

 is then discharged through the galvanometer; the throw of the 

 needle measures the loss of charge QT, which includes the same 



