ON OVli KNOWLEDUE OF .SPECTRUM ANALYSIS. 263 



reader is referred for all historical details, as well as for all measurements. 

 Great care must be taken when experimenting with oxygen to exclude all 

 impurities containing carbon ; for the electric spark oxidizing these com- 

 pounds shows the spectrum of carbonic oxide, which is much more bril- 

 liant than the spectrum of oxygen, and may entirely eclipse it. We 

 distinguish four spectra of oxygen. 



The Elementary Line-spectrum of Oxygen. — This is the spectrum 

 which appears at the highest temperature to which we can subject oxygen ; 

 that is, whenever the jar and aii'-break are introduced into the electric 

 circuit. It consists of a great number of lines, especially in the more 

 refrangible part of the spectrum. It has been called elementary line- 

 spectrum to distinguish it fi-om the other line-spectrum, because, accord- 

 ing to one hypothesis, which has been suggested, to explain the variability 

 of spectra, the molecule which gives this spectrum is in a simpler or more 

 elementary state than that which gives the other or so-called compound 

 line-spectrum. We may, however, adopt the nomenclature independently 

 of any hypothesis which may have suggested it. 



The Compound Line-spectrum of Oxygen. — This spectrum appears at 

 lower temperatures than the first. It consists of four lines : one in the 

 red, two in the green, and one in the blue. With the exception of the 

 blue line, all the lines in this spectrum widen very easily, and with an 

 increase of pressure, more easily even than the hydrogen lines. They do 

 not widen out equally on both sides, but more towards the red than 

 towards the violet. This fact is especially noticeable in the more re- 

 frangible of the two green lines. The blue line always remains sharp, 



The Continuous S-pedrum of Oxygen. — This spectrum appears at the 

 lowest temperature at which oxygen is luminous. The wide part of a- 

 Pliicker tube, filled with pure oxygen, generally shines with a faint yellow 

 light, which gives a continuous spectrum. Even at atmospheric pressure 

 this continuous spectrum can be obtained by putting the contact breaker 

 of the induction coil out of adjustment, so that the spark is weakened. 

 According to Becquerel an excess of oxygen in the oxyhydrogen flame 

 produces a yellow colour, which colour very likely is due to this continuous 

 spectrum of oxygen. The contiuuous background which often accom- 

 panies the elementary line-spectrum must not be confounded with thisr 

 spectrum. 



The Spectrum of the Negatire Glov;. — This spectrum, which was first 

 accura,tely described by Wiillner (1872), is always seen in the glow sur- 

 rounding the negative electrode in oxygen. It consists of five bands : 

 three in the red and two in the green. The least refrangible of the red 

 bands is so weak that it easily escapes obser\-ation ; the two other red 

 bands are rather near together, and may be taken for one single band if 

 the dispersion applied is small. The two green bands, which appear of 

 the same brightness throughout, with pretty .sharply defined edges, are 

 resolved into a series of lines, when looked at v.-ith high optical powers. 

 The same no doubt holds of the red bands ; only the resolution has not 

 been effected, owing to the weakness of the light. 



Transformation of Spectra into each other. — The following description 

 of the appearance of a vacuum-tube filled with pure oxygen, as it under- 

 goes gradual exhaustion, will give an idea of the way in which the spectra 

 of oxygen gradually diffuse into each other : — 



'At fii'st the spark has a yellow colour, and the spectrum is perfectly 

 continuous. Almost immediately, however, four lines are seen in the 



