of the Ionization Potential of Helium. 569 



with anything excepting where it is specifically stated 

 otherwise. As might be expected, the main difficulty was 

 in getting the occluded gas out of the cylinder A and in 

 properly ageing the filament. The occluded gas was finally 

 eliminated by running the pump steadily while maintaining 

 the entire apparatus (excepting the seals I, I 1? and E) at a 

 bright red heat for two hours by the use of a blowpipe. 

 Spurious effects due to the filament surface were got rid of 

 by glowing the filament out at a very high temperature 

 during several hours. The only impurity which appeared 

 subsequent to this treatment was mercury, which could, 

 of course, be kept down only by the constant use of 

 liquid air. 



In addition to the points mentioned above, there is a 

 difference between this apparatus and the arrangements 

 used by Franck and Hertz and by Pawlow which might 

 have been of considerable importance. These authors made 

 use of a device by which a large number of electrons were 

 subjected in a space A to an accelerating field, after which 

 they passed into a retarding field in a space B where impact 

 ionization occurred. The amount of impact ionization was 

 measured by the number of positive ions collected by an 

 electrode bounding space B, all of the negative electrons 

 being returned by the field into the space A. Such an 

 arrangement is very sensitive for detecting the potentials at 

 which ionization sets in, but it has the disadvantage that it 

 affords no opportunity for an accurate comparison between 

 the extra current due to impact ionization and the electron 

 current which is the cause of it. For this reason the results 

 of such experiments may always be open to the objection 

 that the impact ionization results observed are possibly due 

 either to the presence of small amounts of some more easily 

 ionizable impurity, or to the occurrence of multiple impacts 

 under low voltages, which, although giving rise to some 

 ionization, may only form an insignificant part of the impact 

 ionization in the same gas when higher voltages are avail- 

 able. The present arrangement, on the other hand, measures 

 directly both the primary electron current and the extra 

 current it gives rise to by impacts. 



A series of observations made before the last traces of 

 occluded gases had been expelled is of interest in showing 

 the marked effect of minute quantities of impurities. In 

 fig. 2 we have the steady deflexion of the electrometer after 

 a 30-second charge plotted against the negative driving- 

 potential from F to A. The heating current and the 

 capacity across the quadrants of the electrometer are shown 



