in the Magnetic Field. 169 



B and C may be very close together, or coincide, while the 

 centre pair A are separated by a considerable space, as shown 

 in fig. 6. In this extreme case we are furnished with a 

 triplet in which the centre as it were encloses the sides. 

 But this is no specially new form, being quite continuous 

 with the other types of modification. 



Fig. 5. 

 A 



Fig. 6. 

 A 



B 



BC 



Once the doubling of the centre line (A, figs. 2 or 4) is 

 explained, the other types follow in sequence as expected 

 variations, for the cause which converts A into a doublet 

 may be sufficiently powerful to separate the constituents of A 

 more widely than B is separated from C, and the separation 

 of the constituents of A might be tolerably large, even 

 though the separation of B and C is quite insensible. 

 Thus, if the so-called reversed polarization is shown by any 

 lines, the explanation offers no difficulty once we have ex- 

 plained the quartet, but it is doubtful if the lines indicated in 

 the spectrum of iron by the French and American observers 

 just mentioned show this peculiarity. Iron was one of the 

 first substances which I examined *, because I considered . it 

 might present peculiarities, but I did not observe in it any 

 marked differences from the behaviour of other substances. 

 Several quartets and other slight modifications occur, but the 

 lines referred to by the French and American observers do 

 not, on my photographic plates, exhibit the exact peculiarity 

 attributed to them. The central part corresponding to A in 

 fig. 4 is a doublet without doubt, but the remainder (corre- 

 sponding to the lines B and C, fig. 4) does not appear to be by 

 any means a single line, but looks rather like a triplet of which 



* See Proc. Koyal Society, January 1898. The doublets referred to in 

 this paper turned out on analysis to be quartets. 



