the Magnetic Efed of Electric Convection. 453 



it makes at least one complete turn during the time required 

 for a complete oscillation of the system, the latter will take 

 a mean position of equilibrium which will not change when 

 the velocity is augmented. We therefore [)roceeded as follows. 

 The disk was given a slow rotation, the equilibrium position 

 of the system noted, the disk then charged, and the equilibrium 

 position again noted. Thanks to the liquid resistance in the 

 charging circuit, the change in the equilibrium position was 

 too small, if any, to notice. Then the speed of rotation was 

 increased to its maximum, maintained at such for some time, 

 and the position of equilibrium again noted ; finally, the 

 velocity was reduced to its first value and another reading 

 taken. 



In this way we assured ourselves of the following : 



There was a deflexion of the system when the velocity 

 was increased, in the direction demanded by the theory of 

 electric convection. 



The deflexion was permanent. 



It accorded quantitatively with the calculated value to 

 within 10 or 20 per cent. 



From all the foregoing results, corroborated by those 

 previously obtained by Pender, we can conclude that : 



A charged disk, having a continuous metallic surface, 

 turning in its own plane between two fixed condensing 

 plates, parallel to this plane, produces a magnetic held in the 

 direction and of the order required by the theory of electric 

 convection. 



There now remained two questions for us to solve : 



1. Are the magnetic efi'ects thus obtained due to an actual 

 entrainenient of the charge by the moving metallic surfaces, 

 or can they be attributed in some way to conduction-currents, 

 open or closed, produced by the relative motion of the disk 

 and the condensing plates ? 



2. What cause concealed from Oremieu the effects observed 

 by Pender in his previous experiments and also observed by 

 us in common in the experiments just described? 



Experiments on the Entrainement of the Charge by Continuous 

 Metallic Surfaces. — To answer the first question, we first 

 undertook an experiment the idea of which is due toHelmholtz, 

 and the first attempt at realization to Rowland. Consider a 

 plane ring, continuous and conducting, represented sche- 

 matically (fig. 3) by the circle NRMP. This ring can turn 

 in the direction of the arrow under a fixed condensing-plate, 

 represented by the arc SSi; covering a portion of the moving 

 ring. Let the arc covered by this condensing-plate be Ijn 

 of the whole circumference, let p be the distance between the 

 planes of the two plates NRMP and SSj, 7 the linear velocity 



