296 



SCIENCE. 



[N. S. Vol. XII. No. 295. 



vised to account for these variations, which 

 will be discussed later. 



Some attempts to classify the spectral 

 lines according to the character and extent 

 of their magnetic separation have met with 

 partial success. Preston* found that in 

 the spectra of magnesium, cadmium and 

 zinc corresponding lines of the homologous 

 groups of three at the head of Kayser and 

 Runge's second subordinate series act in 

 exactly the same way in the magnetic 

 field. That is, the most refrangible line in 

 each group becomes a sharp triplet with the 

 value 10 (relatively speaking) for the ratio 

 A''//Ji, the middle line a sextuplet with the 

 value 11.5, and the least refrangible a 

 rather diffuse triplet with the value 18. 



A magnetic effect has been noticed on 

 some of the air-lines, f but, with the possi- 

 ble exception of nitrogen peroxide, no effect 

 has been observed on band-spectra, either 

 by emission or absorption methods. 



On the continent of Europe most of the 

 work has been done with absorption spec- 

 tra, particularly with that of the sodium 

 flame ; and in this field several most im- 

 portant discoveries have been made. 



Egoroff and Georgiewsky J noticed that 

 the light from a sodium flame in a mag- 

 netic fleld is partially polarized as a whole, 

 i. 6., without being dispersed. Lorentz§ 

 showed that this phenomenon can be ex- 

 plained by absorption even when the field 

 is uniform. 



Righi 1 1 and Cotton % have shown how 

 the Zeeman effect may be demonstrated 

 without a spectroscope by passing a plane- 

 polarized beam of white light through a 

 magnetized sodium flame or absorbing gas. 



* PMl. 3Iag. , vol. 47, p. 165. 



t Becquerel and Deslandres, Compies Bendua, vol. 

 127, p. 18. 



X Comptes Bendus, vol. 124, pp. 748, 949. 

 I Proc. Boy. Amst. Acad. Sci., vol. 6, p. 193. 

 il Oompius Bendus, vol. 127, p. 216. 

 1[ Compius Bendus, vol. 125, p. 865. 



By this method, which is very sensitive, 

 nitrogen peroxide was shown to be subject 

 to magnetic separation. 



Macaluso and Corbino * discovered that 

 a magnetized sodium plane rotates the 

 plane of polarization to a very great extent 

 for light whose wave-length is nearly that of 

 one of the D-lines. Very close to the ab- 

 sorption-lines the rotation amounts to as 

 much as 315 degrees. The immediate de- 

 pendence of this phenomenon upon the 

 Zeeman effect is shown in a very beautiful 

 way in Cotton's little book ' Le Phenomene 

 de Zeeman,' although I believe the more 

 general principle that magnetic rotation of 

 the plane of polarization is dependent upon 

 the optical dispersion of the medium com- 

 bined with a sort of generalized Zeeman 

 effect, is due to Fitzgerald. f 



An analogous effect of the magnetized 

 flame upon light passed through it acro&s 

 the lines of force was discovered independ- 

 ently by Voigt X and Cotton §. They found 

 that the flame acts like a uniaxial crystal ; 

 that is, it introduces a phase- difference be- 

 tween waves polarized parallel and perpen- 

 dicular to the lines of force. This phase- 

 difference increases very rapidly as the 

 wave-length approaches that of one of the 

 -D-lines. The explanation of this is also 

 given in Cotton's book. 



When we review the experimental facts 

 concerning the effect of magnetism upon 

 light we find many things inconsistent with 

 the elementary theory first given by 

 Lorentz. The equation which he obtained 

 indicated that all spectral lines should be- 

 come triplets under the influence of the 

 field, and that the separation should vary as 

 the square of the wave-length and as the 

 strength of the fleld. On the contrary we 

 find a considerable number of lines which 



* Comptus Bendus, vol. 127, p. 548. 

 tProc. Boy. Soc, vol. 63, p. 31. 

 t Wied. Annal. No. 2, 1899, p. 345. 

 ? Compius Bendus, vol. 128, p. 294. 



