Effect of Electrical Convection. 



207 



effect was the same under these conditions as when the 

 surfaces of the disks were divided into a number of separate 

 sectors, and charged at the periphery by a point. To see if 

 the quantitative effect was also the same, the following experi- 

 ment was performed. The conducting surfaces of both the 

 disks and the condensing-plates were each divided into six 

 sectors by radial scratches j mm. wide. The internal edges 

 of the condensing-plates were connected by a ring of tinfoil 

 1 mm. wide. The sectors of the disks were connected only 

 by the charging-rings on the ebonite cores. If, then, the 

 effect observed was in any way due to conduction-currents 

 in the metal surfaces, this arrangement should certainly 

 greatly alter its magnitude. Observation, however, showed 

 that this was by no means the case. In Table VJ. are given 

 the results of four series of measurements. The values of v 

 calculated from these lie well within the range of accuracy of 

 the experiment. 



Table VI. 



No. 



B. 



I. 



II. 



S. 



V. 



C. 



N. 



2D. 



v. 



18 



2894 



+ 



- 



159 



1199 



16-9 



900 



58-7 



2-89 



19 



j» 



- 



+ 



218 



1360 



171 



81-4 



820 



2-96 



20 



j> 



+ 



+ 



218 



1245 



17-2 



769 



76-7 



2-75 



21 



»> 



- 



- 



195 



1400 



153 



106-3 



80-3 



3 23 

















Mean 



. 2-96 



The above results therefore show beyond any doubt that 

 electric convection does produce magnetic action, or, more 

 exactly, that, within the errors of observation, the current 

 induced in a coil by reversing a convection-current in its 

 vicinity is equal to the current induced in this coil by re- 

 versing a conduction-current of the same strength as the 

 convection-current in a circuit coinciding with the path of 

 the convection-current. 



The conditions of the experiment above described are still 

 capable of considerable improvement. A number of alter- 

 ations are to be made next winter, and it is hoped that much 

 more consistent results will then be obtained. Cremieu's 

 recent experiments will also be repeated, with the view of 

 seeing if the explanations above offered will bear experimental 

 test. There also remain a number of questions in regard to 



