874 



c. Oxygen was completely removed with yellow phosphorus. 



d. Carbon monoxide. The gas was tested for this by means of 

 IjOj'): it was found not to contain more than Vju %• 



e. Hydroijen. In testing for this gas we used the metiiod gi\en 

 by Phillips'). The gas contained less (han 0.01 % of hydrogen. 

 We may add, that the gases were always condensed before they 

 were used in the measurements and tiiat the liquefied gases were 

 then made to boil under reduced pi-essure; the vapour that was 

 drawn off must have contained the last traces of hydiogen present 

 and tlie small admixture of neon must also have been for the 

 greater part removed in this way. 



Finally we may give the following data as providing a measure 

 of the purity of the gases. 



A. Oxygen. The gas was analysed by means of copper 

 (immersed in an ammoniacal solution of ammonium carbonate), 

 later on with sodium liydrosulphite. It was found to contain more 

 than 99.9 7o of oxygen. 



B. Nitrogen. Observations were made with nitrogen, obtained 

 from air by removal of oxygen, which corresponded completely to 

 those made with chemically prepared nitrogen, taking into account 

 the percentage of argon in air nitrogen. 



This correspondence, with such widely different methods of 

 preparation, may give us additional confidence that our gases were 

 satisfactorily fi'ee from impurities. 



C. Argon. This substance was tested for absence of nitrogen 

 indirectly by means of the determination of the pressure at the 

 beginning and the end of condensation, but also more directly by 

 means of glow-discharges in potassium vapour (eomp. 3) in which 

 no diminution of volume could be detected. (Comp. also the determ- 

 ination of specific gravity § 3c). 



3. Methods of analysis. It follows from the above that the only 

 gas besides argon which could be present in the mixtures which 

 were intended for the measurements was nitrogen. This fact made 

 it possible to determine the pei'centage of nitrogen by means of a 

 baroscope. It appeared, however, that the sensiti\ity of the available 

 balance was not so high as we should have wished, in consequence 

 of which these determinations, at least in the most unfavourable 



1) Gomp. Dennis, Gas analysis (1913) p. 231 and 235. 

 '-) Am. Ghem. J. 16 259 (1894). 



