246 Professor Arthur Smithells on 



This unexpected confirmation of the conclusion suggested 

 by my own experiments increased my belief in its probable 

 correctness, and further evidence in its favour appeared from 

 a consideration of well-known facts of spectrum analysis. 



Having been requested to show the experiments at the British 

 Association at Edinburgh in 1892, 1 brought the matter before 

 Section A, where it roused an animated discussion. Widely 

 different views were expressed. Professor von Helmholtz 

 held the opinion that flame-coloration and the luminosity of 

 gas-flames not containing solid carbon must be attributed to 

 radiation consequent on chemical action ; on the other hand, 

 Sir G. G. Stokes did not consider that there was any con- 

 clusive evidence in favour of this view. 



Object of the Experiments. — The object of what follows in 

 this paper is to ascertain how far the luminous radiation from 

 gaseous flames (including " non-luminous " gas-flames and 

 such flames coloured by metallic salts) may be ascribed to 

 incandescence, and whether any other cause has to be sought 

 for an explanation of the phenomena observed. By incan- 

 descence is meant the evolution of light as a direct consequence 

 of the application of heat. . 



The currently accepted Doctrine.— When a small quantity of 

 common salt or other salt of sodium is introduced into the 

 flame of a Bunsen-burner a yellow flame is produced, which 

 is ordinarily ascribed to the vapour of metallic sodium glowing 

 in consequence of the high temperature of its surroundings. 

 The green-blue flame due to copper chloride under the same 

 circumstances is attributed to the incandescent vapour of 

 copper chloride. With calcium chloride the red flame which 

 is obtained is said to be due to the vapour of calcium chloride 

 and of calcium oxide. These may be taken as typical cases. 



The authority for these views dates back, as is well known, 

 from the earliest days of spectrum analysis. One or two 

 questions connected with them which are at once obvious to 

 the chemist are as a rule passed over in silence. If it be 

 asked, for instance, how does sodium vapour become free in 

 the flame, what answer is to be given ? The usual answer, I 

 believe, has been that the salt is dissociated, notwithstanding 

 that we have no other evidence of the dissociability of sodium 

 chloride at any attainable temperature*. Another explanation 

 that I have seen, and that only once, is given by Mendeleeff 

 in his ' Principles of Chemistry' (vol. i. p. 563). He says that 

 the NaCl is acted upon by steam in the flame to produce NaHO 



* Sir G. G-. Stokes informs me that he has never entertained this view, 

 but it has certainly been widely expressed. 



