Part II. — Magnetic Perturbations of the Spectral Lines. 



11 



separated, while B and C are not sensibly separated, or coincide, we see that the 

 case cited by the French and American authors, if it exists, presents no further 

 difficulty, being merely an extreme case of the ordinary quartet form shown in 

 fig. 4. Thus, in fig. 5 we have a quartet in which the separation of the central 

 part A is nearly as great as that of the side lines B and C, while in fig. 6 the 

 case is shown in which the separation of the components of A exceeds that of the 

 side lines B and C\ and, in fig. 7, the extreme case is shown in which B and C 

 coincide, or are not visibly separated, while A is divided into two widely-separated 

 parts. This, then, is the so-called "reversed polarization" type, which, if it 

 exists, forms merely the end link in a continuous chain, and offers no special 

 theoretical difficulties. Once the ordinary quartet is explained, all the foregoing 

 forms become explained, and follow in sequence as expected variations. For the 

 cause which converts A into a doublet may be sufficiently active to separate 



jb c 



Fig. 6. 



BO 



Fig. 7 (not observed?). 



the constituents of A by an amount which may be either greater or less than 

 the separation of the side lines B and C. 



But it is very doubtful if the extreme type (fig. 7) has yet been observed ; for 

 although the present writer made an early study of the spectrum of iron, which 

 is the spectrum in which some of the lines are said to show the so-called 

 phenomenon of " reversed polarization," yet in no case has he been able to detect 

 the peculiarity described above. In fact, as I have stated elsewhere,* the 

 spectrum of iron presents no specially new types of effect, and on my photo- 

 graphic plates the linos referred to by the French and American observers as 

 showing reversed polarization do not show B and C as coincident lines, but show 

 B and C as two broad, weak lines overlapping at their inner edges, so that a dark 

 rib runs down the centre, giving BC the appearance of a dark central line (on 

 the negative) winged by two broader and fainter bands. It is thus likely that 

 these lines are really quartets in which -4 is a doublet and B and C are broad and 



* Proc. Eoy. Soc, January, 1898. The doublets referred to in this paper turned out, on further 

 resolution, to be quartets. « 



