470 Professor Dewar [May 10, 



On letting down the pressure the bands were all weakened; A, 

 though weaker, became more sharply defined at the more refrangible 

 edge. The faint band in the iudifjo A 4470 remained just visible 

 until the pressure fell below 110 atmospheres. At 90 atmospheres 

 A and B were still well seen and sharp, but all the other bands 

 weaker. B remained visible until the pressure fell to 40 atmosplieres. 

 A was then still well seen, the band just above D very faint, and the 

 others almost gone. At 30 atmospheres A was still easily seen, and 

 there was a trace of the band above D. At 25 atmospheres this band 

 had gone, but A remained visible until the j^ressure fell to less than 

 20 atmospheres. Hence an amount of oxygen not greater tlian that 

 contained in a column of air 150 metres long at ordinary pressure, is 

 sufficicDt to produce a visible absorj^tion at A. The quantity of 

 oxygen in the tube at the highest ])ressure we used falls, however, far 

 short of the quantity traversed by the solar rays in passing through 

 the atmosphere when the sun is vertical. 



It will be noted that the bauds, if we except the faint two in the 

 green and indigo respectively, aj^pear to be ^identical with those 

 terrestrial bands in the solar spectrum which Angstrom found to be 

 as strong when the air was dried by intense frost as at otlier times. 

 At least the positions of the maxima agree closely, and that near D 

 shows the same j^cculiarity in having its maximum near the less 

 refrangible end. We did not, however, observe a, which would be 

 fainter than B, and if, like A and B, unresolvable, would be lost iu 

 the diffuse band which covers that region. The bands above 

 numbered 3, 4, 5, 6, agree also with those observed by Olszewski,* 

 to be produced by a layer of liquid oxygen 12 millimetres thick. 

 The point also at which the absorption of the ultra-violet rays begins, 

 agrees with that at which the absorption by ozone begins, as observed 

 by Hartley | ; but the oxygen, as we used it, did not appear to 

 transmit the more refrangible rays beyond 2320, which seem to pass 

 through ozone. Egoroff j found that A remained visible when he 

 looked through 80 metres of atmosphere, but 3 kilogrammes of 

 atmosphere failed to produce a. 



When the pressure in our tube was reduced , a cloud was always 

 formed which rendered the contents of the tube nearly opaque ; the 

 faint light which was then transmitted had always a green tinge. 



It is remarkable that the compounds of oxygen do not show any 

 similar absorptions. Angstrom thought it improbable that oxygen 

 should have a spectrum of such a character, since he failed to 

 obtain an emission spectrum resembling it ; and suggested that the 

 absorptions might be due to carbonic acid gas or to ozone, or possibly 

 to oxygen in the state in which it becomes fluorescent.§ Neither 

 carbonic acid gas nor nitrous oxide, at a pressure of 50 atmospheres 

 in our tube, show any sensible absorption in the visible spectrum; 



* Wied. *Ann.,' xxxiii. p. 570. f ' Jouin. Chem. Soc..' xxxix. p. 57. 



X 'Coraptea Rendu.s,' ci. p. 1144. § 'Spect. Norm.,' p. 41. 



