342 Dr. B. Hasselberg on the 



appearance in Wiillner's tube is completely explained by the 

 small pressure of the gas, since under these circumstances the 

 complete exclusion of oxygen is attended with considerable dif- 

 ficulty. The true spectrum of acetylene, on the other hand, is, 



according to the investigations of Angstrom and Thalen and 

 Liveing and Dewar, an altogether different one, and coincident 

 with the well-known flame-spectrum of gases containing car- 

 bon observed. In the spectrum under discussion there is no 

 trace of these bands, unless we identify the band 6192 with 

 the hydrocarbon band 6187, which appears to me somewhat 

 doubtful, since the much stronger bands at 5633, 5164, and 

 4736 are absent. 



These experiments do not therefore prove directly, in the 

 way intended by Wiillner, that the second spectrum observed 

 in hydrogen tubes really belongs to hydrogen and not to 

 acetylene, though they do so indirectly; since, according to 

 Wiillner's observations, this spectrum gradually predominates 

 during the observation over that of carbonic oxide, until finally 

 it takes its place almost completely. This observation agrees 

 completely with Berthelot's experiments, according to which 

 the continued action of the electric discharge gradually con- 

 verts acetylene into the stable mixture of hydrogen and acety- 

 lene, the percentage composition of which in the present case 

 is expressed by the numbers 97 and 3. It is exactly this ulti- 

 mate great preponderance of hydrogen which shows that the 

 spectrum in question belongs to this gas and not to acetylene; 

 and its gradual production sufficiently explains the gradual 

 fading of the carbonic-oxide bands. 



From this account of the older investigations I pass now to 

 the description of my own experiments. They were designed, 

 in the first instance, to determine the origin of the spectrum, if 

 possible, more certainly than had already been done, and then, 

 further, to investigate as fully as possible the different con- 

 ditions under which they are produced. It is not necessary 

 to insist upon the close connexion of these two problems. In 

 fact it is exactly the complete identity of the spectral pheno- 

 mena of the gas prepared in different ways, taken together 

 with the above criticism of previous investigations, which has 

 convinced me that this spectrum must be ascribed to hydrogen 

 itself. But how little these investigations have contributed 

 to the more accurate knowledge of the spectrum appears from 

 the fact that, with the exception of a few isolated measure- 

 ments by Seabroke* and Vogelf, no actual measurements 

 were made. The drawing given by Wiillner a few years ago 



* Monthly Notices, vol. xxxii. p. 63. 

 t Pogg. Ann. vol. cxlvi. p. 569. 



