of Light at Certain Metal- Liquid Surfaces. 87 



of the light employed, its effect can be exjn-essed as a correc- 

 tion term to the ordinary Fresnel equations. The form of this 

 correction term shows that when the reflection takes place from 

 the surface of a substance of greater index than that in which 

 the light is incident, we should expect a positive ellipticity at 

 the polarizing angle for any film which has an index lying 

 between those of the two media, and a negative ellipticity for 

 a film with a greater index than that of either medium. Thus 

 a real transition layer can theoretically produce only a positive 

 ellipticity ; while a negative ellipticity must always be attribu- 

 ted to films of surface contamination with indices greater than 

 that of the reflecting substance. Of course all of an observed 

 positive ellipticity may not be due to a transition layer, for the 

 matter contaminating the surface may have an index intermedi- 

 ate between those of the two media ; but a negative ellipticity 

 is theoretically a certain criterion of a film of contamination. 

 Lord Rayleigh's experiments mentioned above are thus in strik- 

 ing accord with the theory. 



The same general considerations hold in the case of the 

 reflection from metals. Here, however, since metals naturally 

 produce an elliptic (or a circular) polarization at all angles of 

 incidence, the effect of the transition layer or film of con- 

 tamination will be to change the ellipticity and the phase differ- 

 ence of the two components of the vibration from the values 

 they would have if the film were not present. The theory, 

 based on the same assumption as in the case of the reflection 

 from transparent substances,* shows that, providing the index 

 of refraction of the film or layer is greater than unity, the ellip- 

 ticity should be slightly increased and the phase difference 

 markedly decreased by the presence of such a film or layer. 

 Expressed in terms of the principal incidence and azimuth 

 instead of the phase difference and ellipticity, the theory pre- 

 dicts that the effect of the film should be to decrease the prin- 

 cipal incidence markedly and to increase the principal azimuth 

 but a very little. Unlike the result for the reflection from 

 transparent substances the sign of the effect is independent of 

 the magnitude of the index of refraction of the reflecting 

 medium unless that should be less than unity. 



We have here then no such criterion as before to enable us 

 to distinguish between films of surface contamination and a 

 real transition layer. A possible exception to this statement is 

 the case of those metals which have an index of refraction less 

 than unity. That some metals, notably gold, silver, and cop- 

 per, do possess such small values of the index seems certain, 

 since the same result is yielded by the reflection and the direct 

 transmission methods. With these substances it might be 



*Wied. Ann., xxxvi, p. 865, 1889. 



