340 Prof. Wood and Mr. Kimura on Scattering and 



plane polarized by reflexion from the inner surface of 

 the prismatic plate, or, in other words, the incident 

 polarized light was split into two components, one cf which 

 was wholly transmitted, while the other was in part re- 

 flected, with its vibrations parallel to the reflecting surface. 

 The light thus reflected gave a single image after passage 

 through the double-image analyser. The slit of the spectro- 

 graph was opened somewhat and its length contracted 

 to such a degree that the spectrum-lines of the mercury 

 arc appeared as small squares on the plate. Exposures 

 were made with the bulb at room temperature (tig. 6, 

 Plate VI., upper spectrum), and at a red heat (lower 

 spectrum). In the latter case, owiug to the metallic re- 

 flexion of the 2536 line, both components are reflected with 

 equal intensity, and unite into a plane polarized resultant in 

 azimuth 45°. This is doubly refracted by the analyser, 

 yielding two images of the square for the wave-length in 

 question. It will be noticed that all of the other spectrum- 

 lines (squares) are represented by single images only. 



This experiment does not prove, however, that the reflected 

 light is plane polarized, for we should have a similar appear- 

 ance if it was depolarized or circularly polarized by the 

 reflexion. 



To prove that it is plane polarized we must rotate the 

 double-image prism into such a position that one of the 

 two images disappears. It was assumed that this would 

 occur for a rotation of 45°, and we accordingly turned the 

 prism through an angle of about 40°, and then made a 

 number of successive exposures, turning the prism through 

 a small additional angle each time. One of these exposures 

 showed the second image completely absent, proving that 

 elliptical polarization was not present, or at all events that 

 it was too small to be detected. 



Selective Reflexion and Refractive Index. 



Selective reflexion of another type occurs at the boundary 

 surface separating quartz from dense mercury vapour. This 

 occurs in the case of frequencies slightly higher than that 

 of the 2536 line. The mercury resonators in this case emit 

 no scattered radiation, and there is practically no loss by 

 absorption. In some earlier work *, in studying the reflexion 

 of the light of the iron arc by mercury vapour, it was found 



* R. W. Wood, " Selective Reflexion of Monochromatic Light by 

 Mercury Vapour," Phil. Mag. xviii. p. 187 (1909). 



