176 Dr. T. Preston on Radiation Phenomena 



therefore, that perturbations of this kind are sufficient to 

 account for all the observed phenomena, and, further, that 

 perturbations of this kind are almost certain to be in 

 operation throughout some, at least, of the ionic motions. 



The existence of all these variations of the normal triplet 

 type are therefore of great interest, not only in showing that 

 the perfect uniformity required for the production of the 

 normal triplet is not maintained, as we should expect, in all 

 cases, but also as an experimental demonstration that the 

 causes supposed by Dr. Stoney, in 1891, to be operative in 

 producing doublets and satellites in the natural spectra of 

 gases may be really the true causes by which they are produced. 



Nevertheless Dr. Stoney's explanation of the natural doublets 

 is opposed by a serious difficulty in the fact that the two lines 

 of a given doublet, say the two D lines of sodium, behave in 

 different ways, as if they arose from different sources rather 

 than from the perturbation of the same source. For, in 

 addition to the differences previously known to exist, there is 

 the difference of behaviour in the magnetic field. Thus T) x 

 is a wide-middled quartet in which the distance between the 

 central lines A (fig. 4) is nearly as great as the distance 

 between the side lines B and C, while D 2 shows as a sextet of 

 uniformly spaced lines. 



In a similar manner individual members of the natural 

 triplets which occur in the natural spectra of the zinc, 

 cadmium, magnesium, &c. group behave differently. Thus 

 if we denote the members of one of the natural triplets 

 by the symbols T 1? T 2 , T 3 , in ascending order of re Iran - 

 gibility (for example the triplet 5086, 4800, 4678 of 

 cadmium, or the triplet 4811, 4722, 4680 of zinc, or the 

 green b triplet of magnesium), we find that T 3 in all cases, in 

 the magnetic field, shows as a pure triplet, or suffers accord- 

 ing to the foregoing merely precessional perturbation. On 

 the other hand, T 2 shows in each case as a quartet while T 1 is 

 a diffuse triplet in which each of the members may prove to 

 be complex on further resolution. This would seem to point 

 to an essential difference in the characters of the lines T 1? T 2 , T 3 , 

 as if they sprang from different origins rather than immedi- 

 ately from the same. It is also of great interest to note that, 

 so far as my observations yet show, these natural triplets 

 behave differently according as they belong to Kayser and 

 Runge's first subsidiary series or to the second subsidiary 

 series. Thus if the triplet T 1? T 2 , T 3 , belongs to the first 

 subsidiary series, then the magnetic effect decreases from T 2 

 to T 3 , while if it belongs to the second subsidiary series, the 

 magnetic effect increases from T x to T 3 . Examples of this 



