On the Absorption-spectrum of Oxygen. 287 



of the tube, and thereby loss of light, so far as it was due to 

 the distance of the source, was reduced to a minimum. 



Ordinary oxygen was let into the tube from an iron bottle 

 until the pressure reached 85 atmospheres, and on viewing an 

 arc light through the tube the following absorptions were 

 visible : — 



(1) A very dark band sharply defined on its more refran- 

 gible side, gradually fading out on its less refrangible side, 

 and divided into two parts by a streak of light, occupying the 

 position of A of the solar spectrum. 



(2) A much weaker but precisely similar band in the 

 position of B of the solar spectrum. 



(3) A dark band very diffuse on both edges, extending 

 from about X 6360 to X 6225, with a maximum intensity at 

 about X 6305. 



(4) A still darker band a little above D, beginning with a 

 diffuse edge at about X5810, rapidly coming to a maximum 

 intensity at about X5785, and then gradually fading on the 

 more refrangible side, and disappearing at about X5675. 



(5) A faint narrow band in the green at about X 5350. 



(6) A strong band in the blue, diffuse on both sides, ex- 

 tending from about A, 4795 to X4750. 



When photographs were taken of the ultra-violet part of 

 the spectrum of the arc, and of the iron spark, the gas ap- 

 peared to be quite transparent for violet and ultra-violet rays 

 up to about X 2745. From that point the light gradually 

 diminished, and beyond X 2664 appeared to be wholly absorbed. 



The pressure of the oxygen in the tube was then increased 

 to 140 atmospheres. This had the effect of increasing sensibly 

 the darkness of all the bands above described ; but brought 

 out no new bands, except a faint band in the indigo at about 

 X4470. In the ultra-violet the absorption appeared to be 

 complete for all rays beyond about X 2704. 



The foregoing observations were made with a spectroscope 

 of small dispersion. We next brought to bear on the spectrum 

 a large instrument with one of Rowland's gratings. Even 

 with the high dispersion of this instrument the bands at A 

 could not be resolved into lines ; they remained two diffuse 

 bands ; though the red potassium-lines, which were produced by 

 sprinkling the electrode of the arc with a potassium-salt, were 

 sharply defined and widely separated. None of the other 

 bands were resolvable into lines. This we attribute to the 

 density of the gas, by which the lines are expanded so as to 

 obliterate the interspaces ; and this supposition is confirmed 

 by the observation of Angstrom, that the band in the solar 

 spectrum which appears to be identical with that observed by 



