175 



below is hidden. But I liave found by experiment that it 

 varies very little throughout the whole distance from the axis 

 of r. This I did by placing over the radium just enough 

 tinfoil to cut off all the a rays. 



In the foregoing will be found, I think, sufficient justifi- 

 cation for the choice of the method of this paper, and for the 

 assumptions made during the calculation of the results. 



In the previous paper I have already given a brief de- 

 scription of the process of an experiment. Some points, 

 however, deserve reconsideration in the light of further ex- 

 perience, and some changes have been found convenient. These 

 are best discussed in relation to an actual experiment : I will 

 take a determination of EI in carbon bisulphide. 



I have found it best to separate experiments whose object 

 is to determine IiI from those whose object is to finJ the stop- 

 ping power of the gas. In the former the chief difficulty lies 

 in overcoming initial recombination. This requires the pres- 

 sure of the gas to be low, and the applied potential to be 

 high. A little leakage of air into the apparatus, which can 

 hardly be avoided under these circumstances, is no serious dis- 

 advantage, since the proportion of air can be found from a 

 knowledge of F, and of the pressure and temperature at the 

 time when R is measured ; and these data are easily obtained. 

 In the latter, any moderate voltage will do, since the range 

 does not depend on potential ; but it is desirable to have as 

 much gas as possible, and no leakage of air during the ex- 

 periment, so that when the bulb containing a sample of the 

 gas is taken away and weiglied in order to find the proportion 

 of the mixture, it may truly represent the condition of things 

 during the earlier part of the experiment. It is best to work 

 at a high temperature, if such is required to fill the chamber 

 with gas which is nearly at atmospheric pressure. 



Carbon bisulphide vapour is well superheated at a tem- 

 perature of 30" and a pressure of 25 cm. The apparatus is, 

 therefore, heated to that temperature ; and RI for air is first 

 measured. 



Fig. 4. 



