and the Constitution of the Atom. 263 



place from q^ to 1, and, as we saw in the first part of this 

 paper, the change of qu between 1 and 12 has only a 

 comparatively small influence on the frequency. Further 

 we saw that qk=l was always among the values which gave 

 the best agreement with observations. 



Hence we conclude that the determinations of the ring- 

 systems from the Rontg en-ray spectra which were undertaken in 

 the first part of the present paper also apply to this modified 

 form of recombination from secondaries. There is, however, the 

 difference that some of those combinations of q values which 

 were possible when q^ was variable will be much less good 

 when qu has the fixed value 1. Thus the last two com- 

 binations of Table VIII. [q 2 — 2, # 3 = 4) and {q 2 = l, 23 = 4) 

 do not any longer give a satisfactory agreement. 



Comparing the equation (25) with equation (15) of my 

 previous paper, we notice that they are identical if 



k = i + l. 



Thus the modified and the original hypothesis of recom- 

 bination from secondaries will be identical in all those cases in 

 which there is no electronic system between the secondary circle 

 and the broken ring. Consequently, the equation for the 

 frequency of K a will be the same in both cases and equal to 

 that of Debye. As regards L a we also get the same equation 

 for the frequency in the two cases, provided we give up the 

 assumption of two I .-rings with 7 and 8 electrons. From 

 the treatment of the L a -line on the hypothesis of recom- 

 bination between primaries, we conclude that also the 

 modified hypothesis of recombination from secondaries may 

 be consistent with two L-systems, for which q 2 + q% = 7 ; 

 but such an assumption was found to be of no use for the 

 explanation of the /-line, and we shall also for the recom- 

 bination from secondaries assume that there is only one ring 

 with quant-number 2. If so, the equation (25) gives the 

 same frequency for L a as the equation (15) of Paper I., and 



the calculations of ( p J of my previous paper also apply to 



the modified hypothesis. 



With regard to M a there will be principally the same 

 possibilities as in the case of recombination between 

 primaries. Their, may be two M-sy stems or only one. 



In the latter case the M-ring should consist of 12 or 13 

 electrons. 



Now we saw from the treatment of K^ in the first part of 



