538 Mr. R. T. G-lazebrook on the Application of the 



^Sf+WS-B'S. • • • (65) 



27T 



Hence if u = u sin — (z— Vtf), 



pV 2 -2N /9 'Y-B = 0. 



Hence 



v== N p'+*/Ny 2 +B 2 = B-fN/ 



P P 



••• V = V 0+ ^< (66) 



to the same approximation, if V is the velocity when the 

 medium is at rest. 



A g ain > „s_ P _ Po + p' 



and 



Po Po 



/_A*'-1 



P P< 



... V=V Q + ^~K . . . (67) 



And this is Fresnel's formula, which has been obtained by 

 Boussinesq in a similar manner. 



The consideration of phenomena connected with the rotation 

 of the plane of polarization must be deferred to a future article. 



It remains now to refer to one point of great importance 

 which the theory as it stands will not explain. 



Experiment shows us that in the case of the reflexion of 

 light from transparent media FresnePs tangent-formula does 

 not hold. Some light is reflected near the polarizing angle. 

 According to the theory the tangent-law is true at least when 

 A vanishes. 



Now it is clearly true that there must be a thin layer of the 

 aether near the separating surface of two media, air and glass 

 say, across which the optical density of the aether changes 

 from that of air to that of glass. If this layer be infinitely 

 thin compared with the wave-length, then the transition is 

 practically sudden ; but if the layer has a thickness com- 

 parable with the wave-length, then effects such as are actually 

 observed would occur. And L. Lorenz * has shown that the 

 effects of elliptic polarization observed by Jamin would agree 

 numerically with the results of a theory of gradual transition, 

 if the thickness of the variable film lies between -fa and j-J^ 

 of a wave-length. 



* Pogg. Ann. t. cxiv. p. 238 j Glazebrook, Report on Optics, p. 188. 



