294 Professor Zeeman [March 30, 



spectral lines yield values ranging between 1*4 and 1*8 by 10'. 

 This number is about 1500 times the corresponding number for 

 hydrogen as deduced from the phenomena of electrolysis. 



We must, then, conclude that at least a majority of spectral lines 

 is due to the vibrations of the negative electron. This conclusion is 

 not only valid for incandescent sodium or mercury. All elements, 

 which can give colour to a flame, or which can be evaporated in a 

 spark, show the magnetisation of the spectral lines, and hence in all 

 elements these negative electrons are present. 



Independent experimental evidence for the existence of electrons 

 has been derived from the study of the kathode rays in a vacuum 

 tube. The discontinuous structure of electricity was also proved by 

 other phenomena, and in this way physicists were led by purely 

 experimental methods to the negatively charged cor|3uscle of J. J. 

 Thomson, 1500 times smaller than the hydrogen atom, in fuU 

 accordance with the electron necessitated for the explanation of the 

 magnetisation of the spectral lines. 



All fundamental characteristics of the magnetic resolution of the 

 spectral lines were then explained, and the truth of the explanation 

 proved beyond the possibihty of doubt. More detailed knowledge of 

 the effect has been greatly extended by a whole series of investigators, 

 especially by Becquerel, Cornu, Cotton, Michelson, Konig, Righi, 

 Runge and Paschen, and in this country by Gray, Preston, Lodge, 

 Lord Blythswood and others, and from the theoretical side by 

 Larmor, Fitzgerald, Jeans, and J. J. Thomson. 



Not all spectral lines are tripled, some are split up into quartets, 

 others into sextets. The Hues D^ and T)^ in strong fields are an 

 example. (Figs. 9 and 10.) The whole of such a system of lines is, 

 even in the strongest fields, confined to the space of one-sixth of the 

 distance of the sodium lines. In some cases still more complicated 

 sub-divisions have been observed, especially by Michelson. In such 

 cases the simple electro-magnetic model of a molecule emitting light 

 is insufficient. We shall return to this subject afterwards, and first 

 proceed to a discussion of phenomena accompanying the inverse effect. 



This investigation, which I carried out in Amsterdam together 

 with my pupils, Drs. Hallo and Geest, was suggested by a theoretical 

 investigation by Professor Voigt, of Gottingen. Lorentz's theory 

 relates to one single vibrating particle and can only be applied to 

 substances of very small density, which emit very narrow spectral 

 lines. With greater density, and therefore broader spectral lines, the 

 mutual influence of the molecules must be taken into account. It 

 seems, however, that a theory of emission of a system of reciprocally 

 reacting molecules is rather difficult. In the case of absorption 

 the problem is easier and is considered by Professor Voigt, in his 

 theory of magneto-optical phenomena.* He does not deal with the 



. , • Voigt. Aonalen der Physik, Bd. G7, S. 345, 1899, 



