Charges on Ions in Gases. 205 



with double charges may be produced in the gas are criticised 

 and said to be inconclusive and inconsistent. 



I should like to point out that while admitting that it is 

 difficult to reconcile all the experiments that have been made 

 on this subject, still it appears to me to be obvious that if 

 experimental results are to be accepted at all the balance of 

 evidence is entirely in favour of supposing that such double 

 charges exist, and that the difficulty of reconciling the results 

 of different investigations lies probably in the fact that the 

 exact circumstances under which a large proportion of 

 positive ions with double charges are produced are not yet 

 completely specified. 



The first investigations by which the ratio of the charges 

 on ions in liquids and gases were determined are those 

 involving the comparison of the rate of diffusion with the 

 velocity under an electric force from which a value of N x e 

 was deduced (N being the number of molecules per c.c. in a 

 gas at atmospheric pressure and 15° C, and e the charge on 

 the ion), Nxg being 1'23 x 10 10 when e is the charge on a 

 monovalent ion in a liquid electrolyte (Phil. Trans. A. 1899, 

 vol. cxciii.). It was found that when the mean rate of dif- 

 fusion of positive and negative ions was compared with 

 the mean velocity under an electromotive force the values 

 of N x e for different gases were nearly the same as that for 

 monovalent ions. The mean rates of diffusion and mean 

 velocities under an electromotive force were compared in 

 those determinations because at that time the velocities of 

 the positive and negative ions were not known. Shortly 

 afterwards the velocities of both positive and negative ions 

 generated by Rontgen rays were determined by Zeleiry 

 (Phil. Trans. A. vol. cxcv. 1900), so that the values of N x e 

 for positive and negative ions could be deduced separately 

 and the numbers obtained for positive ions were in seven 

 cases out of eight distinctly larger for positive ions than for 

 negative ions, the average excess being about 12 per cent. 

 Professor Millikan attributes this to experimental error, as 

 variations occur in determinations of N X e in different gases. 

 Such an explanation is not what would in an ordinary way 

 be accepted as the most likely, since errors with different 

 gases would be introduced, unless the gases were equally 

 free from impurities, in determining the rate of diffusion 

 and the velocity under an electric force. 



It is not quite correct to say that in view of my recent 

 experiments I revised my previous conclusions. When I 

 found the value of N x e produced by ions from a point 

 discharge, I drew attention to the fact that the value of N" x e 



