1065 



by means of an induction coil the nitrogen band spectrum dominates here also. 

 The third spectrogram liowever represents tlie light emission of such an argon 



Fig. 2. 



nitrogen gasmixlure, when it has been exposed to the action of phosphorus (by 

 means of glow discharges) We see, that then the argon line spectrum is observed 

 (in this spectrogram the wavelength scale has not quite the right place). From this 

 we may draw the conclusion, that such glow-discharges in P vapour enable us at 

 the same time to analyze in a simple way mixtures of Ar — N<i. 



Also in the case of the nitrogen fixation by phosphorus we must imagine the 

 gas molecules to be first split up into atoms by the discharges, which atoms can 

 combine afterwards with the phosphorus. 



This is quite in correspondence with what K. J. Strutt has found viz. that 

 "luminous" atomic nitrogen can directly be bound by phoshorus. i) 



This can make clear to us why phosplioriis in a glow-lamp is 

 sucli a strong- means to remove rest gases. Also the binding of water 

 vapour by the phosplioi'us is very well observable in liie glow-lamp. 

 Unless very many precautions are taken during the evacuation of the 

 heated lamp, it blackens much more rapidly witfiout phospiiorue 

 than with this agent. It will be well-known that a trace of water 

 vapour liiglily accelerates the desintegration of the tungsten filament, 

 by which process the glass wall is coloured dark. That really 

 the removal of the water vapour is the cauvse of the im|)rovement 

 is confirmed by the result found when lamps without phosphorus 

 and with water vapour are used immersed in liquid air : no abnormal 

 blackening is then observed. 



^ 2. On the applications of silicates. 



A. Action of the silicates. 



The author has found, that when the filament is covered with 



1) R. J. Strutt. Pioc. of the Roy London Soc. A. 85, 219, (1911). 



69* 



