99 



383 



Iransition (02 >ül) will be pos.sil)k'. In the presence of a magnetic field it will split 

 up into a normal Lorentz triplet, where the undisplaced parallel component corres- 

 ponds to the transition (02; 1 -*01; ll and each of the outer components to the tran- 

 sition (02; 2 ^01; 1). The values ol' /?'- and /?"- corresponding to the former transition 

 are easily seen to be equal to •'■ i and respectively, while those corresponding to 

 the latter transition are both equal to 1. 



Ill this connection it may be of interest to notice that, for all the line structure 

 components considered, the value of s(R^) Perp. is larger than that of s{R-) Par., 

 and that in the case of the components (04 -» 03), (03^02) and especially (02 ^01) 

 this seems to be due mainly to the large values of R - and /?"- (/?'- = /?"-' = 1) cor- 

 responding to transitions between tw^o stationary states in both of which the orbit of 

 the electron is circular. This is in agreement with the analogous facts mentioned in 

 the discussion of the theory of the Stark effect and of the fine structure, which 

 seemed to indicate the general result that the estimate of the intensities of 

 spectral lines by means of the values of the a m-p lit u des of the corres- 

 ponding harmonic vibrations in the states implied in the transitions 

 assumes, in the region of small ;j's, an exaggerate character as soon 

 as, owing to the singular character of the motion in these states, the 

 values of these amplitudes become either especially large (e. g. transition 

 from circular orbit to circular orbit) or especially small (e.g. transition from 

 non-circular orbit to circular orbit). 



An interesting remark may further be made in connection with the Zeeman 

 effect of the component (21^*02) in Ha. This component will, under the inlluence 

 of the magnetic field, be split up into an undisplaced parallel component corresjiond- 

 ing to (21 ; 1 02; 1) and two perpendicular components corresponding to (21 ; 1 02; 2) 

 the intensity of each of which is equal to half the intensity of the undisplaced 

 component. The values of /?'- and R"'- corresponding to (21; 1 -*02; 1) are, however, 

 both e(jual to zero, so that we are able to conclude by purely theoretical argument 

 that the a-priori probability for a transition, for which the ampli- 

 tudes of the harmonic vibrations o 1 corresponding frequency occur- 

 ring in the motion in the initial state and in the final state are both 

 equal to zero, will not necessarily be equal to zero. In the discussion ol 

 the Stark effect and of the influence on the fine structure due to a small electric 

 field we have already met with analogous transitions, and just as in those cases, 

 we have in the i)resent case that the amplitude of the corresponding harmonic 

 vibration is different from zero in the mechanically possible states lying between 

 the initial state aiui final state, and which here are characterised by (2/,, 2 — /,; 1 ), 

 where < / 1. 



Especially when considering transitions of the type just discussed the (jueslioii 

 arises whether the estimate of the intensities of the components in which a spectral 

 line is split up would not he essentially improved by comparing these intensifies 

 with some kind of mean value of the square of the corresponding amplitude taken 



49* 



