Maxwell's Theory of Light. 287 



or, by equation (5), 



rr d 2 c d ( d , d \d\ d / dr , 7 dr , dr\ 



_d^c_ ,d?c_,di_ c /q\ 



. " c^ 2 + df + ^ 2? ' * ' { } 



with similar equations for a and 5. 



So long as the dielectric moves as a rigid body with uniform 

 velocity and without rotation, the equations for a, b, c are ex- 

 actly the same as those for /, g, h, and are of the type 



^K W <fy 2 ^ 2 / ~ ^ 2 v *» * q dy r dz K dt ' 



The boundary conditions at the surface of separation of two 

 media are : — 



1. That both the electric and magnetic displacement resolved 

 along the normal to the surface of separation should be the 

 same in both media. 



2. That both the electric and magnetic forces resolved 

 parallel to the surface of separation should be the same on both 

 media. 



If the medium is at rest, the equations reduce to the form 



/*KW % 2 dz 2 J~~dt 2 ' 



which corresponds to a wave-motion propagated with a velo- 

 city -— . In the media we are concerned with, fi differs 

 V fjuK 



from unity only by a very small fraction ; so that the velocity 



of propagation is very nearly — 7= . Now, if the specific induc- 



\/ K 



tive capacity of a dielectric depended upon the number of 

 times in a second the electric displacement was reversed, the 

 velocity of the different coloured rays would be different ; 

 i. e. there would be dispersion. There seems to be evidence 

 that the specific inductive capacity does depend on the rate at 

 which the electric displacements are reversed ; for Schiller and 

 Gordon, reversing the displacements about 2000 times in a 

 second, found the specific inductive capacity of glass was much 

 smaller than when it was found in the usual way (Electrical 

 Eesearches of the Hon. H. Cavendish, Appendix). To ex- 

 plain dispersion, the specific inductive capacity ought to in- 

 crease with the rapidity of reversal ; and as in light the dis- 

 placement is reversed billions of times in a second, it is no 

 argument against its doing so that the specific inductive capa- 



X2 



