214 ESSAYS BIOGRAPHICAL AND CHEMICAL 



spectrum disappear and blue lines become visible ; hence 

 the colour of the discharge changes from red to blue. 

 And other gases exhibit the same kind of change, though 

 not generally in so striking a manner. Further, it has 

 been shown by my colleague, Dr. Collie, that an element 

 may develop a new and strong line in its spectrum 

 on being mixed with a certain gas, which it does not 

 exhibit if another gas be substituted. All these questions 

 are very obscure, and have not as yet been investigated ; 

 only the fringe of the subject has been touched. As the 

 aurora, without doubt, is visible on different occasions at 

 very different altitudes, it is more than possible that the 

 spectra will differ. The appearance of red auroras would 

 imply a spectrum in which red lines predominate ; but I 

 am not aware of any observation having been made of the 

 spectrum of a red aurora. From the similarity of colour 

 it might well be conjectured that the red tint is due to 

 the discharge occurring through an atmosphere compara- 

 tively rich in neon. 



Assuming for the moment the identity of the line of 

 wave-length 5570 with that of krypton, two questions at 

 once suggest themselves. First, why should this line be 

 so remarkably brilliant when krypton is present in the 

 atmosphere in comparatively very minute amount ? 

 What are the relative intensities of the spectra of krypton 

 and of other gases under similar circumstances ? And 

 second, is there any process which will tend to increase 

 the relative amount of krypton in the upper regions of 

 the atmosphere ? I have attempted to answer both these 

 questions. 



Some years ago, in conjunction with Professor Collie, 



experiments were made on the visibility of the spectrum 



of one gas in presence of another with which it was 



diluted. 1 The results are given in the following table: 



1 Proc. Roy. Soc., lix. 257. 



