460 Messrs. V. Cremieu and H. Pender on 



charge the diminution \va.s less. Further, the deflexions 

 ceased to be })roportional to the potential of the fixed sectors 

 for voltages above 2000 volts. Further, the charge going on 

 the sectors became less than it should be according to calcu- 

 lation, and above a certain definite voltage the charging 

 current remained practically constant, no matter how much 

 the voltage was increased. 



These facts made clear the cause of the negative results. 

 Moreover, in 1900, in his first experiments on open currents, 

 Cremieu observed a part of these phenomena. However, as 

 he was eager to complete his experiments on convection, he 

 did not stop to investigate completely these eifects. 



We now undertook a systematic study of the role of the 

 dielectric, but, in consequence of the short time at our disposal, 

 we have not been able to carry this study very far. As of 

 first importance from the point of view of the theory of 

 convection we verified the following points : — 



1. When the continuous disks were covered with caout- 

 chouc, the magnetic effects diminished and presented the 

 dissymmetries of sign observed with the sectored disk. In 

 these experiments the charge on the disks can be determined 

 only by measuring the potential. These measurements showed 

 that there was the same lack of proportionality between the 

 potential and the magnetic effect observed. 



2. A thin sheet of paraffined mica fixed on the surface of 

 the disks produced the same efiects as the caoutchouc. 



3. In Pender^s induction experiment we found that the 

 presence of the mica diminished considerably the effects 

 observed. For example, in one series of measurements we 

 obtained the following deflexions : — 



Disks bare, condensing-plates bare .... 140 mm. 

 Disks covered with mica, condensing-plates bare 100 „ 

 Disks and condensing-plates covered with mica 15 „ 



These facts show clearly the experimental cause of 

 Cremieu^s experiments^, but they do not explain them. In- 

 deed, the role of the dielectric seems difficult to unravel from 

 any a priori considerations. 



From the phenomena observed in 1900 and in our joint 

 experiments it seems legitimate to conclude that, when solid 

 dielectrics are submitted to considerable penetrations of 

 charge, they act as if they suppressed the electrostatic in- 

 fluence between the conductors they separate. Moreover, 

 for each kind of dielectric the penetration of charge and the 

 suppression of electrostatic influence resulting therefrom 

 occur at a certain well-marked critical voltage of the charged 



