572 Dr. C. B. Bazzoni : Experimental Determination 



no trace of mercury or of any other impurity. The electron 

 currents were also reduced by lowering the temperature of 

 the filament until no capacity was necessary across the 

 electrometer. Here saturation is complete at 10 volts, and 

 ionization is seen to set in at 19'5 volts. Curve C was taken 

 after the liquid air had evaporated, and a trace of mercury 

 was present. The mercury, which presumably ionizes under 

 4*9 volt impacts, makes it impossible to get complete satu- 

 ration, but, nevertheless, the amount present is too small to 

 obscure the setting in of the helium ionization, which is 

 again seen to start at 19*5 volts. 



In order to deduce the correct ionization potential from 

 these data, it is necessary to consider two disturbing factors — 

 first, the drop of potential along the filament and, second, 

 the initial velocity with which the electrons leave the wire. 

 The first factor might be important because the filament is 

 cool at the ends, so that no electrons are liberated exactly 

 where the voltmeter is connected. The total drop along the 

 filament with the currents used was found to be *75 volt. 

 Allowing one-tenth of the length of the filament at each end 

 as the part where the cooling effect would be appreciable — 

 which, judging from a visual examination, was an ample 

 allowance — the corresponding correction to be applied to 

 the voltmeter readings is seen to be O'l volt. The distri- 

 bution of velocities amongst the emitted electrons was 

 studied by applying positive potentials to the filament, and 

 it was found that the maximum velocity of emission at the 

 temperatures used was one volt. It is seen, therefore, that 

 when the voltmeter read 19*5 volts there were some electrons 

 coming from the negative end of the filament with velocities 

 corresponding to 20*4 volts. The results are consequently 

 in general accordance with those of Franck and Hertz. If 

 the true value is different from 20*4 volts, it is more likely 

 to be somewhat below this value than above it, since very 

 few of the electrons have initial velocities approaching one 

 volt. The experimental arrangement, however, with a large 

 anode completely surrounding the filament, and with the 

 greater part of the drop of potential close to the filament, is 

 a sensitive one, and ought to detect the ionization even when 

 the number of electrons impinging with the necessary energy 

 is small. 



The phenomena of ionization in gases like helium and 

 mercury vapour, involving, as they apparently do, elastic 

 collisions below a certain velocity, and totally inelastic ones 

 above that velocity, are worthy of close study. The hypo- 

 thesis of a critical velocity for a totally inelastic impact was 



