1897.] on the Source of Light in Flames. 373 



marks the period when the oxygen required for combustion is got 

 partly from the air mixed with the vapour before it leaves the burner 

 and partly from the air outside, one cone corresponding to each part 

 of the supply. From analyses of the interconal gases, it appears that 

 large quantities of carbon monoxide and hydrogen are generated in 

 the inner cone, and that these are the gases which burn in the outer 

 cone. The evidence that the formation of carbon monoxide is the first 

 step in the combustion of carbon has been greatly strengthened by 

 the experiments of Prof. H. B. Dixon, and is at variance with no 

 important facts. 



The source of the light in a blue-burning hydrocarbon flame has 

 been the subject of most elaborate investigation and of prolonged 

 controversy. The spectrum of this light was one of the first to be 

 carefully described, and is often called the Swan spectrum, from 

 the fact that it was first accurately mapped by Swan in 1856. It 

 is seen in the blue part at the base of all ordinary hydrocarbon 

 flames and in the inner cone, but not in the outer cone of flames fed 

 with air in the manner of the Bunsen burner. In so far as the charac- 

 teristic ijroduct of these jiarts of flumes has been found to be carbonic 

 oxide, it would be natural to attribute tlie Swan spectrum to this gas. 

 This view, however, has never been adopted. The Swan spectrum 

 has been attributed either to carbon itself or to a hydrocarbon (acety- 

 lene), and the whole discussion and investigation of the subject has 

 centred round these alternatives. The neglect to consider the likeli- 

 hood of carbon monoxide being the source has arisen from a disregard 

 of the occurrence of this gas in flames, and from a belief that it has 

 another distinct spectrum. At the same time the difficulty presented 

 by the other explanations has been fully realised, and it is admitted 

 that the support of either demands somewhat strained hypotheses. 



The question of the origin of the Swan SiDcctrum is too large and 

 complicated to be fully discussed here. It will suffice to j)oint out 

 that if the formation of carbon monoxide is the first act in the oxida- 

 tion of a hydrocarbon two results would follow : (1) it would hardly 

 be supposed that carbon vapour existed free even momentarily in the 

 flame; (2) that the preponderating product with which was associated 

 the energy of the chemical change should contribute mainly to the 

 emission of light. The chief difficulty opposed to the view that 

 carbon monoxide is really the source of the Swan spectrum appears 

 to lie in the fact that this gas may be made to yield a different spec- 

 trum by the electric discharge. A full consideration of the evidence 

 bearing on the subject leads to the view, first, that this spectrum is 

 not undoubtedly due to carbon monoxide, and secondly, that it may 

 be due to carbon dioxide. 



The evidence derived from the study of flames, and much other 

 evidence, is favourable to the view that carbon monoxide is the source 

 of the Swan spectrum, and if this be the case, the chemi-luminescence 

 of a hydrocarbon flame like that of a flame of the hydrides of j^hos- 

 phorus, silica and antimony, would be attributed to the act of oxidation. 



Vol. XV. (No. 91.) 2 c 



