875 



case, are not more accurate than to about 0.2 7o- Fortunatelj- we 

 were able to carry out the analysis more accurately by a cliemical 

 method, which enabled us to attain an accuracy of 0.1 "/„. ') We 

 shall begin by a description of the latter method. 



A. Determination of the percentage of nitrogen in Ar — N mixtures 

 by means of glow discharges in potassium vapour. 



It has been long known ^), that in electric discharges through gases 

 such as nitrogen, hydrogen etc. a cliemical reaction may occur, 

 especially between the material of the cathode and the raritied gas. 



Mey ') pointed out, that this provided a means of liberating rare 

 gases from admixtures. This method was further developed by 

 Gehlhof ^), who succeeded by means of glow discharges through 

 potassium vapour in preparing speclroscopically pure rare gases 

 comparatively rapidly — although not in large quantities. 



In order to adopt this method to a quantitatixe analysis of .4r — .V 

 mixtures the following apparatus was constructed by us. 



A definite quantity of tlie gas-sample which is collected above 



1) The readings miglit have been further refined by the use of a cathetometer. 

 but we did not adopt this method, as an accuracy of Ü,1"'q was sufficient for 

 our purpose. 



3^ Comp. G. Salet, Pogg. Ann. (158) 332, 1876. L Zehnder, Wied. Ann. 

 (52) 56, 1894. 



3) Mey, Ann. d. Phys. 11 127 (1903). Gehlhoff and Rottgardt, Verb. d. D. 

 phys. Ges. 12 411 (1910). 



■') Gehlhoff, Verb. D. phys. Ges. 13 271 (1911). 



