Regular Reflexion of Light by an Absorbing Gas. 389 



time before the appearance of the reflected wave, which is 

 additional evidence against such a view. We had hoped to 

 find that, as the density increased, an increasing percentage 

 of polarization would be found in the scattered radiation, but 

 such does not seem to be the case. 



The complete absence of polarization appears to be rather 

 remarkable, and not easy to explain. A rapid rotation of 

 the molecule would probably act as a depolarizing factor; 

 and if the vibrations of the resonator were constrained to 

 take place along fixed lines or planes in the molecule, these 

 would be oriented in all possible positions, and we should 

 expect, from the most elementary considerations, less than 

 50 per cent of polarization, even with the incident light 

 plane polarized. This was found to be the case with the 

 resonance spectra of sodium and iodine vapour ; but in; 

 the present case, where we are apparently dealing with a 

 vibration of much simpler type, there is no trace of polari- 

 zation whatever. 



If we are to regard the metallically reflected wave as the 

 resultant effect of vibrations emitted by a closely packed 

 system of resonators in perfect synchronism, we must show 

 how the polarization results, for the reflected wave is plane 

 polarized as we shall now see. 



In the case of the reflexion of light from metal surfaces, if 

 the plane of polarization is parallel or perpendicular to the 

 plane of incidence, the reflected light is also plane polarized. 

 If, however, the incident light is polarized in an azimuth of 

 45°, the reflected light is usually more or less elliptically 

 polarized, due to the phase difference between the two 

 reflected components. 



We. examined all three cases, but need mention in detail 

 only the one in which the incident light is polarized in 

 azimuth 45°. The light of the water-cooled arc was passed 

 through a small Foucault prism arranged to transmit vibra- 

 tions inclined at 45° to the vertical. It was then reflected at 

 the polarizing angle from the inner surface of the prismatic 

 plate of the bulb into a quartz spectrograph. Between t lie 

 slit and the collimating lens, and close to the latter, we 

 mounted the double-image prism already referred to, arranged 

 so as to transmit horizontal and vertical vibrations, the images- 

 formed by it lying one above the other. It may be worthy 

 of mention that in working with a quartz spectrograph it is 

 important to analyse the polarized light before it enters the 

 lenses, on account of the natural rotation of the latter. 



The rays which entered the spectroscope were of course- 





