Experiments on Residual Ionization. 407 



then it is possible that about 5 ions per c.c. per second are 

 produced in the gas by the collisions of thermal agitation. 

 Now it is a well-known fact that when a radiation produces 

 ions in a gas, the total number of ions made is the same so 

 long as the radiation is totally absorbed in the gas ; also the 

 effect of recombination will be least, and therefore the 

 current will be greatest, when the gas is at such a pressure 

 as just to absorb the rays. Hence it is possible that for a 

 certain range of pressures the current will increase as the 

 pressure decreases, on account of the effects of recombination. 

 The conditions of this experiment are particularly favourable 

 for a large recombination effect, since the direction of the 

 rays is perpendicular to the direction of the electric field, 

 and also since carbon dioxide is a heavy gas. Yet in spit" 

 of these arguments it seems improbable that the effects ov 

 recombination could cause a rise from 7 to 9 ions per c.c. per 

 second, or an increase of about 33 per cent, in the current, 

 as is the case here. These results would therefore seem to 

 show that the residual ionization cannot be wholly due to the 

 collisions of thermal agitation. They cannot, however, be 

 said to exclude the possibility that a smaller number of ions 

 than 5 per c.c. per second may be due to these collisions. 



Variation of Ionization with Temperature. — The gases used 

 were carbon dioxide and acetylene, and the range of tempe- 

 rature was from 18° to 100° C. Great difficulty was ex- 

 perienced in getting reliable sets of readings; for as each set 

 required a period of from six to eight hours, it was quite 

 possible that the leak of the electrometer might vary during 

 this time. As a determination of the leak required that the 

 ionization-chamber be exhausted, and as also it was found 

 that in order to obtain consistent readings the gas had to be 

 allowed to stand in the chamber for some time before com- 

 mencing readings, it was only possible to obtain one leak 

 reading for each set of temperature readings. All the sets 

 of readings, however, agree in showing that the ionization is 

 practically unchanged from 18° to about 80° C. For tempe- 

 ratures from 80° to 100° some sets of readings show a very 

 marked increase in the current, while in others this increase 

 is very small. The two chief sources of possible error would 

 appear to be thermo-electric currents, and the driving off of 

 minute quantities of radioactive emanation from the walls 

 of the ionization-chamber as its temperature was raised. 

 These errors were guarded against by eliminating possible 

 thermo-junctions, and by exhausting the ionization-chamber 

 while it was heated to 100°. To test for i\\a presence of 

 thermo-electricity the junction of the brass electrode and the 



