and Dispersion of Liquid Oxygen. 271 



bands increases as the square of the density of the oxygen, 

 and has recently confirmed this (Comptes JRendus, June 17, 

 1895) by observations of the solar spectrum in the desert of 

 Sahara. This Jaw, as we have already pointed out (Proc. 

 Hoy. Soc. vol. xlvi. p. 228), seems to indicate that these par- 

 ticular bands are due either to complex molecules generated 

 by condensation, or to the encounters between molecules of 

 ordinary mass which are more frequent when their free path 

 is diminished. 



To answer M. Cornu's question, we obtained liquid air 

 directly from the atmosphere by the cold due to the rapid 

 evaporation of liquid oxygen under reduced pressure. The 

 method and apparatus have already been described by Pro*- 

 fessor Dewar (Proc. Roy. Inst. 20th January, 1893). 



The absorption of this liquid air in a thickness of 1*9 centim. 

 was then directly compared with that of liquid oxygen in a 

 thickness of 0'4 centim. The light which had passed through 

 the latter was, by means of a reflecting prism, brought into 

 the field of view of the spectroscope at the same time with 

 that which had passed through the liquid air. The positions 

 of the lamps were then adjusted so that the brightness of the 

 spectra in those parts where there were no absorption-bands 

 was equal in the two spectra. Under these circumstances, it 

 was seen that the absorption-bands were very much more 

 strongly developed by 0*4 centim. of liquid oxygen than by 

 five times that thickness of liquid air. 



The vessels containing the liquids being open, the liquid air 

 gradually evaporated; and as the boiling-point of nitrogen is 

 lower than that of oxygen, the former evaporated more quickly, 

 and the residual liquid contained a larger and larger percentage 

 of oxygen. As this happened the absorption-bands became 

 more intense, until they exceeded in strength those of the 

 thinner stratum of liquid oxygen. 



Another sample of air liquefied as before was rapidly mixed 

 with an equal volume of liquid oxygen, and the absorption of 

 this liquid compared, as before, with that of liquid oxygen. It 

 was seen that the absorption of 2 4 centim. of the mixture 

 was much greater than that of 0*4 centim. of liquid oxygen. 

 The density of the oxygen in the mixture was in fact three 

 times that of the oxygen in pure liquid air, and by Janssen's 

 law the absorption should have been increased ninefold. Our 

 observations accord with this so far as they go. Also these 

 observations agree well with the theory of the continuity 

 between the gaseous and liquid states. It must be borne in 

 mind that air boils at a lower temperature than oxygen, so that 

 the two liquids compared were not at the same temperature 



