ilw Maanetic Effect of Electric Convection. 4(53 



the oalvanoinoter. but of the opposite sign to that previously 

 taken, and this etfeet continued even for several hours there- 

 after \vith no appreciable diminution. These phenomena 

 ■Nvere extremely irre^iular, the size of the sui)plementary 

 currents seemino- to de])end upon the previous state of the 

 mica and the duration of the experiments. 



At the sectors Si the same phenomena were ol^served as in 

 the preceding case. 



3. Moring Sectors and Fi'.ved Sectors covered ivitli Mica. — 

 As long as the potential of the fixed sectors S was below 

 2000 volts no anomalies were observed. Above 2000 volts 

 the charo-e-current became considerablv 2'reater than the 

 discharge-current. This dissymmetry attained even as much 

 as 30 per cent. In the neighbourhood of 4500 volts the 

 proportionality Ijetween voltage and charge-current ceased, 

 as in the preceding case. There were also supplementary 

 currents after the charoino- sectors were earthed. These 

 could attain as much as 50 per cent. o£ the value of the 

 currents when the sectors were charged. But there was 

 this peculiarity, that for the moving sectors charged positively 

 they were in the opposite direction to the normal currents, 

 while for the moving sectors charged negatively they were 

 in the same direction as the normal currents. The sectors Si 

 took up no charge by influence. Even more so than in the 

 former case did all these irregularities depend upon the 

 previous state of the dielectric and the duration of the 

 experiments. 



All this shows how complex the phenomena are, and that 

 only a prolonged study can fully explain them. This much 

 seems certain, that there is a penetration of charge in another 

 sense from that usually understood by the expression. That 

 which seems to prove this conclusively is that, in the third 

 case above, the supplementary currents are obtained only 

 when the two brushes A and C both make contact with the 

 sectors. If either of the brushes is suppressed, there is no 

 longer a supplementary current. If the supplementary cur- 

 rents are due merely to the ordinarily considered penetration 

 of charge, this charge would gradually flow out through the 

 brush left in contact, and therefore produce a current in the 

 galvanometer, without the presence of the second brush 

 being necessary. It seems to us that the mechanism of the 

 supplementary currents is analogous to an electrophorus, the 

 mica on the sectors being in some way electrolysed. 



It is essential, in any case, from the point of view of the 

 accepted theories, to know exactly why the dielectric dimi- 

 nishes or suppresses the magnetic effect, and what is the 



