498 Potentials required to produce Discharges in Gases. 



would be much greater when the valves were closed all the 

 time than when they were open during the passage of the 

 current, as in the former case there would be no opportunity 

 for diffusion. But as a matter of fact the readings indicate 

 the opposite. As has already been pointed out, the readings 

 are not of great value quantitatively, but one might reason- 

 ably expect that on the whole those taken with the valves 

 closed would average higher than those taken with them 

 open, while actually they are considerably lower. 



Another possibility is the existence of >ome sort of electric 

 fatigue in the gas after the passage of the current. At first 

 thought it would seem that the existence of such an effect 

 could not have escaped the observation of so many previous 

 observers. But the conditions of previous observations have 

 been such that the gas between the electrodes has been in 

 free communication with that in the rest of the apparatus at 

 all times, and under those conditions in the present experi- 

 ments no change in the sparking potential is observed. On 

 this supposition, however, the effect should be at least as 

 great when the valves are closed during the time when the 

 current is run as when they are closed only at the end of 

 that time, though not necessarily greater, since the effect 

 might be produced as rapidly as iresh gas could diftu.-e into 

 the chamber. This supposition, therefore, is open to the 

 same objection on the ground of quantitative results as is 

 that mentioned in the last paragraph, but to a less extent. 



It would have been well if a few more readings had been 

 taken at certain points, but the apparatus was taken down 

 immediately after the present readings had been completed, 

 in order to make room for other apparatus, and the results 

 were not considered in detail until later, so that the desira- 

 bility of further readings was not at the time observed. As 

 it is, the conclusions arrived at are : — 



1. Bunning the current under the conditions of the ex- 

 periments produces a rise in the sparking potential, and the 

 cause of the change lies in the gas. 



2. As to the main object of the experiments, the investi- 

 gation of the suggested heating of the gas and expulsion 

 from between the electrodes, with the resulting effect on the 

 sparking potential, the effects obtained are distinctly incon- 

 sistent with this hypothesis. The explanation of the relation 

 between the maintenance and sparking potential curves below 

 the critical pressure is evidently to be sought on different 

 grounds. 



3. One possible explanation of the effects obtained is that 

 air is driven out from the electrodes by the passage of the 



