and Dispersion of Liquid Oxygen. 269 



silvered all over except a narrow vertical strip about 4 millim. 

 wide, which was left un silvered to allow of the passage of 

 light. This vessel was used, exactly as the cylindrical vessel 

 had been used in the former experiments, as a lens by which 

 an image of a source of light was thrown on to the slit of a 

 spectroscope. The pair of glass plates, separated by a thin 

 stratum of air and fixed to a rod which was the prolongation 

 of the vertical axis of a theodolite, were arranged at about 

 the centre of the globe. The oxygen in the globe was very 

 tranquil, and the silvering cut off all light which did not pass 

 nearly centrally through the globe. The result was that the 

 Light of the rays observed was cut off, when the glass plates 

 were turned through the proper angle, much more sharply 

 than before, and the measures are so much the more trust- 

 worthy. 



We found the spark between cadmium electrodes a con- 

 venient source of light, both because the rays are bright, and 

 because they are dispersed through a considerable range in 

 the visible spectrum, and it was possible to watch their 

 extinction one after another as the glass plates were slowly 

 turned. Even with this arrangement, the extinction of the 

 rays when liquid oxygen was in the globe was not quite so 

 sharp as when the experiment was made with alcohol. This 

 was probably due to the scattered light from the bubbles 

 in the oxygen, and was more troublesome in regard to the 

 brightest rays. 



We obtained as the mean of several observations for the 

 blue ray of cadmium, X4416, //,= 1*2249 ; for the red ray, 

 X6438, yu,= . 1*2211 ; for the green ray of thallium, X535, 

 fi = 1*2219. Also by using a flame we obtained for the red 

 ray of lithium, X6705, //.= 1*2210, and for the yellow rays of 

 sodium, \ 5892, //,= 1*2214. 



The last figure is less than we had found in 1892 by the 

 prism method, which was 1*2236, and still less than that found 

 in 1893, which was 1*226. It is also less than that recently 

 found by Olszewski and Witrowski (Bull, de V Acad, de Cra- 

 covie, July 1894, p. 246), which was between 1*2222 and 

 1*2235. The values we have now found for the refractive 

 indices corresponding to the red ray of lithium and the green 

 ray of thallium are also less than those found by Olszewski and 

 Witrowski, which were about 1*2213 and 1*2235 respectively. 

 We think, however, that our measures for the red and blue 

 rays of cadmium are better than those made for the thallium 

 and sodium rays. 



These give for the mean green nearly yLtl'222, and, taking the 

 density of oxygen at its boiling-point as 1*137, the refraction- 



