226 



NA rURE 



[January 5, 1899 



the broadened line now appears as a doublet), does not 

 absolutely prove that the broadened line is a triplet with 

 its components overlapped. It merely determines that 

 the broadened line may be a triplet, and that the theory 

 which anticipates the tripling may be correct. In order 

 to place this matter beyond all doubt, it is necessary to 

 so increase the strent;th of the magnetic field that the 

 ■components of the triplet (if they exist) shall be completely 

 separated from one another : and when this ' is done, it 

 is found that the tripling exists, but it is also found that 

 many divergencies from the uniform expectation of 

 theory pure tripling) exist. Thus, as pointed out above, 

 many lines under the influence of the magnetic field 

 show as quartets, or sextets, or octets, or other modified 

 form of the normal triplets. In the examination of these 

 ■cases the double image prism forms a very valuable 

 adjunct, as all the light polarised in one plane goes to 

 form one image, while all the light polarised in the per- 

 pendicular plane forms the other image. The appearance 

 presented in the field of view of the spectroscope by 

 •different typesof lines,underthese circumstances, is shown 

 m Fig, 3. In this figure the lines of the upper row' are 

 formed by one image from the double image prism — that 

 is to say, by the light vibrating horizontally, and corre- 

 .spond to the central members of the normal triplets ; 

 while the bottom row consists of light vibrating vertically, 

 ■;ind represents the side lines of the normal triplets. Thus 

 .at .\ a' we have the normal triplet, as expected by theory, 

 with the central line. A, polarised in one plane, while the 



B 



D 



X 



B' 



two side lines, a', are polarised in the perpendicular plane. 

 This type exists in the case of by far the greater number 

 of spectral lines, and may be regarded as the general or 

 normal type, if for no other reason than the frequency 

 with which it occurs. The second type, shown at r. «', is 

 a quartet in which, instead of a single middle line, we 

 have two middle lines close together at h, with the two 

 side lines at b' as before. This type of quartet occurs in 

 the blue cadmium line 4S00, and in the blue zinc line 

 4722. \\. c c' another species of quartet is shown ; in 

 this there are two middle lines also, but the separation of 

 these is almost as wide as that of the side lines, so that 

 the appearance presented to the eye when the double 

 image prism is not used is that of two fine doublets, 

 •r.ither than the quartet appearance of the type v. n'. 

 This third type, cc', occurs in the case of the sodium 

 line l)„ the greenish-blue line of barium 4934, and many 

 others. The fourth type, im', is a sextet of fine uni- 

 formly spaced lines, two of which correspond to each 

 component of the normal triplet, ihat is, the central 

 component is a doublet, and each of the side com- 

 ponents is also a doublet. This type is represented by 

 the line D.j of sodium. The fifth type is shown at K e', 

 where the central constituent is a doublet, and each of 



' This was effecleii by the wriler in October i8<)7, and triplcis and 

 . luarlcls were then pholographed directly without the aid of a nicol or any 

 (j}larisini;.ipparatus whatever. (See letter to Nati'ke, dated November 

 10, lS9-. vol Ivii. p. 173.) These phot.jgrajjhs were shown at the November 

 meeting of the Dublin Ifniversily E.vpenmental Science Association, at 

 the December 1897 meeting of the Koyal Dublin Society, and at the 

 January 1898 meeting of the Royal Society of l.ond<in ; but it waji not 

 until .\pril 1S93 that they were reproduced in the ritilosflf'hkal MuL'a-hie 

 (•,lh <-.ries, vol. xlv p, 335, plate xxiii.). 



NO. 1523, VOL. 59] 



the side components is a triplet. The distance between 

 the components of the central doublet in this case is 

 about the same as that between the central members of 

 the side triplets. This type is represented in the yellow 

 line of barium 5850. All the variations so far noted 

 may be embraced in the general statement that each 

 line of the normal triplet a \' may itself become a 

 doublet or a triplet. 



The question now of greatest importance is whether 

 these various types of modification by the magnetic 

 field are consistent with the theoretical explanations of 

 the phenomena put forward by Larnior, Lorentz, and 

 others ? Naturally one must endeavour to reconcile 

 facts and theory. If this reconciliation has not yet been 

 effected, we must not hastily conclude that the theory is 

 wrong, or even that it requires to be modified or patched 

 up ; and it was with this feeling that 1 put forward Phil. 

 Mag.^ ser. 5, vol. xlv. p. 325, April i8i;8i the idea that 

 these various modifications might be due to reversal — 

 that is, to absorption in the outer parts of the spark or 

 other source of light. Thus P. (Fig. 3) might arise from 

 A by reversal of the middle line, and so also might cc' 

 and DL>' be produced, and even ee' might be intelligible 

 from this hypothesis if we supposed double reversal I'.'i 

 to occur in the side components of the triplet .\.\', and 

 a wide absorption band to occur in the middle line (sup- 

 posed much broader than the others). But (as 1 stated 

 when putting forward this view^ the appearance pre- 

 sented to tlie eye is not that of ordinary reversal, so 

 that appearances are against the 

 T" ^ supposition that the modifications 



I ,. are due to absorption in the vapour 

 b. H. surrounding the source of light. But 

 still it is to be remembered that the 

 magnetic field exerts a considerable 

 influence on the source of lighi. 

 ]"''^* and might alter considerably the 

 appearance of an ordinary reversal. 

 However, in order to test this matter, 

 y ^ I observed many lines, which de- 



viate from the normal triplet type, 

 in a magnetic field of gradualh- 

 increasing strength. The object of 

 this was to determine if the separation of the lines form- 

 ing the upper row in Fig. 3 (say, the doublet K or c) 

 depended on the strength of the magnetic field. Thus, 

 if the components of the doublet 11 remain fixed while 

 the distance between the side lines i;' continues to 

 increase as the magnetic field increases in strength, then 

 we might conclude that reversal is not only a possible 

 explanation by very probably the true explanation. But 

 the components of the central parts h, c, l>, do not 

 remain fixed as the magnetic field increases in strength. 

 On the contrary, the distance between the two lines \\ 

 increases as the strength of the field increases ; indeed, 

 as far as rough observations go, the distance between 

 the components of n or c, like the distance between the 

 side lines 11' or c', is proportional to the strength of the 

 magnetic field. .Similar remarks apply to the types i>li', 

 ee', &c. When the field increases in strength, the line? 

 forming n separate from each other, and so also do the 

 doublets 11', and the lines forming each component of 

 the latter also separate, so that the sextet remains a 

 system of equally spaced lines. On the other hand, 

 when the field is reduced in strength the various lines 

 close up till li, c, D, E each appears as a single line 

 with n', c', i>' as narrow doublets— in fact, the normal 

 triplet type is approached in appearance as the field is 

 reduced. 



It appears, therefore, that the explanation of the various 

 modifications of the normal triplet typi- cannot be satis- 

 factorily explained by reversal, and consequently these 

 divergencies must be referred to the action of the mag- 

 netic field on the vibrating structure which emits the 



