330 Refractive Indices of Liquid Nitrogen and Air. 



to the vertical plane by observing with a cathetometer the 

 reflexion at their surface of a flame which had been adjusted 

 to lie in the same horizontal plane as the centre of the plates. 



The arrangement was tested by the measurement of the 

 refractive indices for the D ray of water, and of alcohol, 

 contained in one of the same double vessels which were 

 subsequently used for liquefied gases. The results were 

 satisfactory, though, owing to the irregular form of the 

 vessels and the striae on their surfaces, the extinction of the 

 light on turning the plates was hardly so sudden as we had 

 expected. The angles were measured by first turning the 

 pair of plates in one direction until the ray under observa- 

 tion was extinguished, and reading the position on the 

 graduated circle of the theodolite, and then turning the pair 

 of plates in the opposite direction until the ray was again 

 extinguished, and again reading the circle. The angle between 

 the two positions was the double of the angle required, 

 namely the angle of incidence for total reflexion of the ray 

 observed. 



Trying the apparatus with liquid oxygen, we found that 

 the extinction of the ray was less sudden than with either 

 water or alcohol. This was no doubt caused by the small 

 bubbles which scattered the light. Nevertheless it was easy 

 to trace the passage of commencing obscurity from the blue 

 to the red end of the spectrum as the plates approached the 

 positions of total reflexion. Hence approximate values could 

 be found for the indices of refraction for the D ray. In this 

 way liquid oxygen gave //,= 1*226, a slightly larger value 

 than that we had found with the prism, which was 1*2236. 



We then tried liquid nitrogen at its boiling-point of 

 — 190° C. at atmospheric pressure. The mean of six readings 

 gave for the D ray /x= 1*2053. The nitrogen was not quite 

 pure, but contained about 5 per cent, of oxygen, notwith- 

 standing that it had been passed before compression through 

 a stack of tubes filled with copper and heated red-hot. We 

 have no reason to think that this small quantity of oxygen 

 had any other effect than to increase the index of refraction 

 by a very small quantity. 



Liquid air was next observed. The mean of ten mea- 

 surements gave for the D ray //,= 1*2062. With liquid 

 air the evaporation of the more volatile nitrogen goes on 

 more rapidly than that of the oxygen, so that the liquid 

 gradually gets stronger in oxygen, until at last there is little 

 but oxygen left. Liquid nitrogen is colourless, and the 

 colour of liquid air is merely that of the oxygen it contains. 



