and the Constitution of the Atom. 245 



In my previous paper it was shown that in a number of 

 oases the X-ray spectra might be explained by recombination 

 from secondary circles. If in our general equations (5) we 

 put k = i + l and q. x = l we get the equation 15 of Paper I. 

 corresponding to recombination from a secondary circle. 



Thus we know that in a number of cases g { =1 gives 

 an approximate solution of (11) ; but then we see from 

 Table I. that in any such case there will bo another value 

 01 9i + i which ought to give an equally good solution. 



We also see that the values of <j)(q) -q inside the interval 

 of possible values of q show comparatively small variations, 

 and, if q = l is an approximate solution, any value of q 

 between 1 and 12, say, would give a fairly good agreement 

 with observations. 



As we cannot, at any rate from the present scheme, claim 

 or obtain a perfect agreement between observed and calcu- 

 lated values we cannot be sure that the value of q. +l which 

 gives the best numerical agreement is the true one. In other 

 words, even if our present, hypothesis should prove to be 

 right, we cannot with any claim of accuracy determine the 

 number of electrons of a ring only from a radiation process 

 in which this ring is the one from which the electron starts 

 recombination, thus e. g. we cannot by means of the K« line 

 determine exactly the number of electrons in the L-ring. 



As it appears from my previous paper, this is no longer 

 the case \\ hen we assume the energy to be maintained, in 

 that case the K«-line gave quite definite values for the 

 number of electrons of both the K- and the L-rings. 



On the assumption of conservation of momentum, however, 

 the number of electrons of each ring-system must be deter- 

 mined from a line which is produced, when the ring in 

 question is the one to which recombination takes place, or 

 the number of electrons must be determined by the value 

 of q i which gives a possible value of the right term of 

 equation (11 h). 



Now \~^)<l>{g i )--qi is the variable part of the term to 



the right, and for small values of n { the ratio will differ 

 considerably from unity, and the expression to the right of 

 116 will vary fairly rapidly with q. Thus, if there is a 

 value of </. which gives an approximate solution of (11 b) 

 there will be no other value which satisfies it. When ?ii 



becomes fairly large, however, -^- approaches unity and 

 also the right term will show a similar variation as <f>{q) — q- 



