230 Prof. L. P. Wheeler on the Reflexion of 



That this well-nigh universal though small discrepancy 

 between theory and experiment cannot be entirely assigned 

 to contamination of" the reflecting surface seems to have been 

 definitely settled by the experiments of Lord Rayleigh on 

 very clean water surfaces *. He found that as the surface 

 was made progressively cleaner, the negative ellipticity 

 previously observed became numerically smaller, passed 

 through a zero value, and for the cleanest surfaces obtain- 

 able assumed a very small positive value. On the other 

 hand, Drude, working with the fresh cleavage surfaces of 

 transparent crystals where the chance of surface contam- 

 ination is very remote, found that within the limits of error 

 of observation the ellipticity vanished f. It would thus 

 seem to be established that the theory stands in need of 

 correction in the case of reflexion from liquids, while remaining 

 entirely competent in the case of solids. 



Now the boundary conditions from which the equations 

 of Fresnel are deduced assume an abrupt, discontinuous 

 change in physical properties as one passes through the 

 surface separating the two media. As such physical dis- 

 continuity is, a priori, highly improbable, it is natural to 

 seek the source of the observed discrepancies in a thin 

 transition layer where the two media interpenetrate, and the 

 physical properties change continuously, though rapidly, 

 between the constant values they possess beyond the limits 

 of the layer on either side. On this hypothesis it is natural 

 to expect a thinner transition layer, and hence a closer 

 approximation to the conditions of a discontinuous change, 

 in the case of solids than of liquids. Thus, admitting that 

 such layers must exist in all cases, we must conclude from 

 the experiments mentioned, that it is only in the case of 

 liquids that they attain a sufficient thickness to be appreciable. 



But while the existence of the transition layer seems a 

 fairly certain inference from the phenomena of reflexion, it 

 is equalty certain that the greater part of the divergences of 

 experiment from the predictions of theory are to be ascribed 

 to films of surface contamination due to dirt or polisher. 

 Such films may be exceedingly thin — of the same order of 

 magnitude in fact as the transition layer. In any given case 

 both the layer and the film may be present and the observed 

 ellipticity be due to their combined action. The principal 

 difference in the effect produced by these tw T o causes would 

 be due to the fact that in the case of the transition layer its 

 index of refraction must vary between those of the two 



* Phil". Mag-, ser. 5, vol. xxxiii. 1892. p. 1. 

 t Wied. Ann. xxxvi. 1889, p. 532. 



