Hydrogen and the Spectrum of Aqueous Vapour. 345 



whether they arise from impurities conveyed into the tube 

 from the pump. They do not appear in nitrogen and in 

 rarefied air; and I am therefore inclined to attribute them to 

 hydrogen. The palladium, apparently, supplies hydrogen 

 more or less continuously to the tube to repair losses from 

 the process of electrical dissociation ; and the tube with the 

 palladium adjunct seems more reliable than the form of tube 

 which is filled with hydrogen through drying apparatus 

 which necessarily cannot be subjected to heat, and through 

 which rarefied air is conveyed with the hydrogen into the 

 spectrum-tube. 



The study of the effect of powerful electrical discharges on 

 hydrogen led me to endeavour to find the lines discovered 

 by Professor E. C. Pickering in the star t Puppis. Since 

 the wave-lengths of these lines satisfy a modification of 

 Buhner's formula. Professor Pickering attributes them to 

 hydrogen. The wave-lengths of these lines are comprised in 

 the spectral region extending from about 4200 to 3700. I 

 have plotted them as short lines contiguous to the normal 

 solar spectrum in fig. 7. The long lines correspond 

 approximately to the most intense lines or bands in the 

 spectrum of hydrogen produced in the tube provided with 

 the palladium adjunct. The hydrogen spectrum, when 

 regarded as a whole on the scale of small dispersion I have 

 employed, seems to be made up of lines spaced according to 

 a certain order : very much as if two sets of lines spaced 

 according to a certain arrangement should be superposed — 

 the fingers of one hand shifted over those of the other. 



The hydrogen lines are more or less intensified bands or 

 dark accumulation of lines almost obscured by the spectra of the 

 compounds of nitrogen and aqueous vapour, if a strong current 

 is employed in a tube provided with aluminium electrodes. 

 It is possible that certain hydrogen bands may be narrowed 

 and rendered lines, in a spectium of small dispersion such as 

 astrophysicists are compelled to employ, by an electrical 

 dissociation of water-vapour in the presence of an excess of 

 oxygen, and that the new lines in Puppis may be evidence 

 of the presence of this vapour in the star's atmosphere. To 

 test this theory, I filled a tube with oxygen and submitted it 

 to a powerful condenser-discharge. The resulting spectrum 

 was of the general type obtained by these strong discharges 

 in hydrogen, nitrogen, and rarefied air. It is shown in fig. 6. 

 If we compare this spectrum with fig. 3, in which the broad 

 bands only represent hydrogen lines in the gas at atmospheric 

 pressure ; with fig. 4, the spectrum in rarefied air ; with the 

 lower one in fig. 5, the spectrum obtained in pure nitrogen: 



