THE AUSOrPTTOT^ A^'D EMISSrOW OF AIK. 15 



electrodes obtained from aiiotlier source, hut there was no dian-re in the s|n'c(ra. 

 The bands afterwards, as before, were the regular accompaniments ..f ihc photd- 

 graidis of the spectra of hydrogen and of oxygen. Asa last atlempL in liiat di- 

 rection, I shall use a form of (leissler tube wliieli is nndei' construction and whicii 

 is so arranged that, in the cleaning during exhaustion, a far more euerijetic heatiiu,' 

 can be attained than was possible for the tubes pi'eviously em[)loyed. 



It ought not to go unmentioned that all the tubes were observed with a spectro- 

 scope both during the cleaning and during the photogra[ihing of the s{)ectrum with- 

 out any impnirities making themselves noticeable. This shows how little the visible 

 spectrum can serve as critei'ion of puiity, in filling a tube. Tiie most refrangible 

 ultra-violet is far better adapted to the purpose, and according to my ob.servations, 

 the most sensitive spectral reactions belong to the region of the shortest waves. 



Production of the ingredients of air and the spectrographic results. — I pass 

 now to the I'esults directly relating to my investigation. Since these results are 

 not complete, notwithstanding my numerous exiteriraents, I can in this preliminary 

 communication only give the most important of them, and not even those without 

 reserve. Foi- it is clear that, as long as the question of the origin of the supposed 

 bauds of oxide of carbon is open, a not inconsiderable measure of uncertainty at- 

 taches to them all. Besides this, substantially nothing is known of the part of 

 the spectra of gases to which my observations i-elate, so that at no point did any 

 information come to me from the scientific literatni'e. Finally it ought not to be 

 forgotten how difficult the analysis of the facts relating to spectra in Geissler tubes 

 has, at all times, been found, and how little we leally know about the subject to- 

 day, notwithstanding the fact that scarcely any otlier part of specti'oscopy has oc- 

 cupied so much acumen and gift for obsei'vation as the exploration of the spectra 

 of electrical discharge-tubes. 



I begin with the spectrum of 



Nitrogen. — It was pi-eparecl in two ways : first, by passing air over potassium 

 pyrogallate; and, secondly, from potassium nitrate and ammonium chloride. It 

 was dried with phosphorus pentoxide, over which the gas remained standing for 

 some time befoi'e it was passed into the tube. Nitrogen proved itself very trans- 

 parent, even beyond 162 /'/'; yet it absorbed particular wave-lengths very ener- 

 getically. Its emission-specti'um consists of a number of groups of bands of 

 moderate i)hotographic power shading off toward the shorter wave-lengths and 

 ending below (that is on the less refi'angible side of) 185 ;</'■ The observation of 

 this specti-um is uncommonly interfered with by the bands of carbon monoxide 

 overlying it. Beyond 185 /'/', I have been unable to prove any introgen bands. 



"aT^yz/e??..— Electrolytically formed oxygen was exclusively employed. The 

 carbon dioxide which was invariably present in small amount at the beginning of 

 the evolution was eliminated by caustic potash ; the aqueous vapor, by phosphorus 

 pentoxide or sulphuric acid. In i)assing to the absorption-tube, the gas was 

 strongly heated in ordei' to destroy the ozone. Oxygen absorbs the rays in the 

 neigh°borhood of 185 /'A' in a series of clearly i-esolved groups of lines, fourteen in 

 number. These groups, which are of baud-like form, constantly approach nearer one 



