506 Mr. J. Larinor on the Theory of 



unpolarized, the intensity of the middle line would be twice 

 that of either of the outer ones. But viewed along the field 

 the middle line would disappear, as the exciting vibration 

 would be end on to the observer and could not therefore send 

 out transverse radiation : the other lines (equally sharp as 

 before) would be circularly polarized, and their directions of 

 polarization would, as Zeeman remarks, determine whether 

 the vibrator involves a positive or a negative electron. 



4. A view has been enunciated that it is only one kind of 

 ions, namely the negative ones, that are mobile and free to 

 vibrate in the atom or molecule, the other kind being fixed to 

 the matter and immobile. On such an hypothesis, if the 

 charges of these negative ions are proportional to their effec- 

 tive masses, for example if they are simple electrons without 

 inertia other than that of the electric charge, the intervals 

 (measured in difference of frequency) between magnetic 

 doublets and the outside lines of magnetic triplets in the 

 spectrum should be the same for all lines. Moreover, they 

 should be the same in different spectra. Thus an hypothesis 

 of that kind can be definitely put to the test. 



5. When there are ions of different kinds describing orbits 

 in the molecule, these exact results no longer hold : but even 

 then we can assert that the difference of frequency between 

 the lines of a magnetic doublet is of the order eH/27rMc 2 , and 

 the order of magnitude of e/M. can be thence derived. Thus 

 Zeeman concludes from his experiments that the effective 

 mass of a revolving ion, supposed to have the full unitary 

 charge or electron, is about 10 -3 of the mass of the atom. 

 This is about the same as Professor J. J. Thomson's estimates 

 of the masses of the electric carriers in the cathode rays. If 

 we took these carriers to be simply electrons, as their con- 

 stancy under various environments tends to indicate, there 

 would thus be about 10 3 electrons in the molecule. 



6. In view of the above considerations, the circumstance 

 that in a magnetic field certain lines, viewed transversely, 

 are divided into sharp triplets with perfect plane polarizations, 

 which has been described by Zeeman and assented to by 

 Lodge and by Michelson"*, is an important clue to the character 

 of the principal oscillations which emit those lines. In an 

 oscillating molecule undisturbed by a magnetic field there 

 must be three types of vibration which all have the period 

 belonging to that line, namely two types which differ only by 

 involving rotations in opposite directions round the axis of 

 that magnetic field and would naturally have the same 



* In Cornu's experiments, Compter Rendus, Oct. 18, the application of 

 the polarizing apparatus seems to have been required to divide the lines. 



