of Metal Particles for Li</ht- Waves. 265 



>vith the ohsorvations which I have made with the silver fihiis, 

 the ab^orption-baml inovino- towards the red when the plane 

 of polarization is so oriented that the electric vector has no 

 component normal to the surface. 



With sodimn and potassium films I have obtained just the 

 opposite eftect, the absorption-band moving in the other 

 direction. These results appeared at first to be very confusing, 

 until I finally discovered what I believe to be the explanation. 



1 have studied the behaviour of the films at oblique inci- 

 dence with polarized light with considerable care, and have 

 obtained results which appear to be concordant among them- 

 selves. 



If the plane of incidence is horizontal and the electric 

 vector is parallel to this plane, i. e. with a component per- 

 pendicular to the surface^ we find for example a heavy 

 absorption-band in the green. As the plane of polarization is 

 rotated through a right angle the band in some cases shifts 

 its position in the spectrum (sometimes in one direction and 

 sometimes in the other) and sometimes simply fades away 

 entirely without undergoing any shift. On examining the 

 spectrum with greater care, however, I found that in general 

 other bands are present. If the band in the green moves 

 towards the red, a band comes into the spectrum from the 

 red end to meet it, and another band which was originally in 

 the violet moves out of the spectrum in the direction of the 

 ultra-violet. It is difficult to say at present whether the 

 appearance and disappearance of these tw^o fainter bands is 

 due to their motion along the spectrum^ or to the fact that 

 one fades away and the other springs into existence, as in 

 the case above cited. This difficulty is due to the fact that 

 a part of the band lies outside of the visible spectrum, and a 

 fading away can easily be mistaken for a drift. I am of 

 the opinion that in some cases we are dealing with one 

 phenomenon, and in other cases wdth the other. 



I have examined several films in which a band in the yellow 

 and one in the blue distinctly approached each other when the 

 plane of polarization was rotated through a right angle. This 

 explains why it is that a single band near the centre of 

 the spectrum is sometimes seen to move in one direction and 

 sometimes in the other. 



In the one case it was doubtless the right-hand band of 

 a pair which approached (the other lying in the invisible 

 portion of the spectrum), in the other case it was the left- 

 hand band of an approaching pair. Some typical cases of the 

 changes which occur when the plane of polarization is rotated 



