the Luminosity, of Gases. 255 



The only alternative to the above explanations is, so far as 

 I am aware, that indicated by Arrheniiis in his experiments 

 on the electrolytic conduction of the heated vapours of salts 

 (Wied. Ann. xlii. p. 18, 1891). 



Arrhenius supposes that when salt is supplied to a name, 

 sodium hydrate is formed by the influence of steam and that 

 the hydrate then undergoes partial ionic dissociation, the 

 liberation of sodium taking place without the intervention of 

 ordinary chemical action. The process is thus regarded as 

 analogous to the dissociation of electrolytes in dilute aqueous 

 solution. I cannot attempt to summarize the experimental 

 evidence by which Arrhenius seeks to establish his interesting 

 theory. If true, it would certainly remove many difficulties 

 which have been met with in attempting to account for 

 the phenomena of flame-coloration on purely chemical 

 grounds*. 



With regard to Arrhenius's view of the matter, I will only 

 remark at present that the liberated sodium must sooner or 

 later meet with oxygen, and burn. If the yellow light of the 

 flame is due to the vibrations of ionic sodium, it would seem 

 likely that the light emitted during combustion of the sodium, 

 and due presumably to sodium oxide, should be different. 

 Now the light from sodium burning in oxygen gives a con- 

 tinuous spectrum, and it is possible that the faint continuous 

 spectrum which accompanies the line-spectrum of sodium salts 

 heated in a flame may be due to the combustion of the sodium, 

 whilst the line-spectrum is due to the mere vibration of the 

 sodium ions. 



The genesis of the sodium spectrum has recently been made 

 the subject of experiment by Pringsheim (Wied. Ann. xlv. 

 p. 428, 1892). He shows first of all that sodium carbonate 

 and sodium chloride, when vaporized in a hot porcelain tube, 

 give neither absorption nor emission spectra. Metallic sodium 

 heated in the tube filled with nitrogen gives first a coloured 

 vapour showing a banded spectrum, and then a transparent 

 vapour exhibiting the ordinary sodium spectrum. He then 

 raises the question whether this spectrum is dependent on 

 temperature alone, or is caused by the act of chemical union 



* In addition to those already pointed out I may adduce the follow- 

 ing : — When a flame coloured by a bead of sodium salt is viewed against 

 a bright background of D light, it is seen to be surrounded by a black 

 mantle. If this is due to absorption by free sodium, then the element 

 must be able to exist in an atmosphere of steam and carbon dioxide. 

 This is utterly opposed to our ordinary notions of the chemical relations 

 of _ the substances in question, but would present no difficulty from the 

 point of view of Arrhenius. 



