1898.] 



Fall of Potential at the Electrodes, etc. 



23 



The presence however of the convection currents described in 

 § I. interferes with such a measurement. 



Various attempts were made to get rid of the effects of their 

 pressure, but no reliable method was obtained. 



•20 



15 



10 



Fig. 4. 



Still, it can be shown that during conduction there is an in- 

 crease in the force acting upon the electrodes. 



This is most readily done with a modification of the apparatus 



Fig. 5. 



shown in fig. 1, which is represented in fig. 5, where the box PP' 

 which encloses the plates A and B is not drawn. 



