266 H. E. J. G. du Bois on Kerr's 



appears suited. These names are formed analogously to that 

 of another magneto-optic quantity, viz. " Verdet's constant," 

 which is quite generally accepted ; they recall the names of 

 the discoverers of the corresponding phenomena, and are not 

 limited to any particular language. I therefore believe that 

 no objection can be made against their introduction, which 

 appears desirable in order to avoid confusion among three 

 essentially different constants of nature. 



§ 25. Conclusion. — To my mind the experiments discussed 

 leave no doubt but that the peculiar phenomena connected 

 with reflexion from magnets solely depend upon the magneti- 

 zation existing immediately behind the mirror ; and I believe 

 that no theory ought to ignore this fact. They also supply an 

 experimental proof, which, however, is hardly required, for 

 the assumption that at least part of the incident radiation 

 penetrates below the surface and is there acted upon by mag- 

 netism, to be reflected out again afterwards. For, supposing 

 the ray's path to lie entirely outside the metal, the action 

 could only depend upon the magnetic condition of the air, 

 which, however, is not the case. 



I have been able to show, on a former occasion, that mag- 

 netization also is the quantity upon which the effect on trans- 

 mission of radiations through magnets depends ; this of course 

 is intimately related to Kerr's phenomenon. The latter is 

 doubtless the more complicated of the two phenomena, and 

 differs from the former in sign and in the character of its dis- 

 persion*. I believe the following simple kinematical ex- 

 planation maybe given for this difference; in doing so the 

 dynamics of both phenomena are left aside as a remoter 

 question. 



§ 26. On normal transmission through a magnetic film the 

 resulting difference of phase between the two opposite cir- 

 cularly polarized rays depends on their different velocity of 

 propagation only ; for the geometrical path is the same for 

 both, viz. the film's thickness. On normal reflexion, however, 

 the "optical path"f in the metal ( = retardation of phase) is 

 proportional for each ray to the quotient of the geometrical 

 path by the velocity of propagation. Now the latter as well 

 as the geometrical path (in consequence of the different depth 



* The rotation is known to be positive on transmission through Co, 

 Ni, and Fe ; its dispersion is anomalous, but according to Dr. Lobach 

 {Dissert., Berlin, 1890) it shows neither maxima nor minima. 



f "Optical path "= geometrical path x index of refraction j a con- 

 venient expression used by some authors. 



