﻿478 Dr. Norman Campbell on 



less. But the divergence of the calculated mean from the 

 observed is probably due to deficiencies in the theory. In 

 § 8 of the previous paper it was pointed out that a different 

 value for P would be obtained according as the conditions 

 used in determining the constant of integration were those 

 at ,r=0 or those at x=b. The latter have been used ; if the 

 former had been used the calculated values of n Y would have 

 been still smaller ; they would have been from 35 per cent, 

 to 40 per cent, of the observed values, and it was this dis- 

 crepancy which first threw doubts upon the accepted calcu- 

 lations. Now it is almost certain that the error in the 

 theory caused by taking the conditions at x = b is in the same 

 direction, though smaller in amount, than that caused by 

 taking those at x=0; and it is therefore to be expected that 

 the calculated values of n 1 should turn out to be slightly too 

 small. 



I do not think that anyone will quarrel with the conclusion 

 that Table XXIII. provides as complete a proof as we could 

 possibly hope to attain of the substantial accuracy of the 

 physical assumptions underlying the theory on which the 

 interpretation is based. 



16. (3) Accepting this conclusion, we may inquire what 

 light it throws on the various physical problems under dis- 

 cussion. 



First, there is the question of which of the two theories 

 discussed in the previous paper, as to the nature of the 

 negative ion after it has made a few collisions, is correct. 

 The results which have been recorded were obtained on the 

 first theory that the electron remains an electron, and does 

 not become a complex ion. How far the results would agree 

 with the second theory cannot be told completely without 

 detailed calculation, but I think a little consideration will 

 show that it is very unlikely that they would agree at all. 

 The values obtained for v and V wdl differ according to the 

 theory adopted, but it is to be noticed that, if the field is so 

 great that every collision results in ionization, the two theories 

 become identical. Adopting the figures for these quantities 

 which we have obtained, we find that, for the largest value 

 of y, the average number of ions made by collision of an 

 electron starting from the electrode is, on the first theorv, 

 between 1*3 and 2'5 for the pressures used ; while, on the 

 second theory, the number is from 0'6 to 0"7. On the other 

 hand, for the lowest value of ?/, the same number on the 

 first theory ranges between 03 and 1*0, on the second theorv 

 it is 0*027. If the same values of v and V are applied to 

 both theories for the large value of y, the ionization predicted 



