and the Constitution of the Atom. 273 



Thus we come to the following conclusion : — 



Whether ice assume recombination from a normal or deformed 

 primary system or recombination from a secondary one, we 

 have always to assume that the angular momentum of the 

 electrons left behind in the atom remains unchanged. And, 

 further, we have to assume that the changes of energy of the 

 systems situated between the broken ring and the ring of 

 departure enters into the energy quantum of radiation which is 

 emitted as the result of the recombination. 



When we assume conservation of momentum the three 

 assumptions I., II., III. will all give a very close agreement 

 between calculated and observed values, and all three as- 

 sumptions lead to the same values for the number of 

 electrons in the various ring-systems. This is due to the 

 fact that the energy changes which accompany the recombi- 

 nation are very little affected by a change of the number of 

 electrons in the system of departure. 



If we were merely regarding the numerical agreement for 

 each line separately, it would be very difficult indeed to 

 decide in favour of any of the three possibilities (I., II., 

 III.) a. All of them explain the principal (a) lines as well 

 as the secondary (f3) lines almost equally well when each line 

 is seen separately ; but still we found that the assumptions 

 I. and II. meet with considerable difficulties. 



First of all we found that the assumptions I. a and II. a 

 in the case of K^ led to a characteristic form of the curve 

 i/=/(N) which was not indicated by the experimental values. 

 At any rate the assumptions I. a and II. a are not consistent 

 with the assumption of only one M-ring. The assumption 

 Ilia., however, gave the right curvature. 



The strongest argument in favour of the assumption Ilia, 

 we got from considerations with regard to the deviation 

 from Kossel's relation. The assumption of recombination 

 from secondaries gave just the right values of the deviation 



"'(iiAki-ai 



both as regards sign and absolute magnitude. On the other 

 hand, the assumption I. a gave d = 0, and II. a led to values 

 of: d which were only a small fraction of those given by 

 experiments. 



Now we must remember that our theory — so far it has 

 been carried — does not give an exact agreement between 

 observed and calculated values. Thus our calculations will 

 require some small corrections, probably due to the fact that 

 our expression for the energy is not quite correct. But if 



