877 



B. Deterininntlon of the percentage of nitrogen hy means of a 

 baroscope. 



For this purpose use was made of the ditfereuce in the upward 

 pressure produced by the gas-mixture on a glass body (volume about 

 300 cm.') which was suspended from one arm of a balance as 

 compared to an open glass vessel of an equal outer surface on the 

 other arm. 



The apparatus was arranged in such a manner, that by the 

 turning of a properly shutting glass tap the arresting arrangement 

 of the balance could also be put into action in a high vacuum. The 

 pressures were read by means of a cathetometer. 



The following data^) were used for the baroscope-determinations : 

 density of air, free from water-vapour and carbon dioxide, 45° N.L., 

 sea-level, 0° and 76 cm. 0.0012928 



nitrogen (Rayleigh and Leduc) 12514 



argon (Watson) 17809 



C. Results. 



The following table (p. 878) (column 1 — 6) gives a survey of 

 different determinations by methods A and B. 



In deducing the mean (column 6) we have attributed a double 

 weight to the determinations by method A. 



When the mean of the results by method A is compared with 

 those by method B, a systematic deviation will be seen to exist 

 which increases with the percentage of argon in the mixture. As 

 the baroscope had been previously calibrated with other gases (carbon 

 dioxide, nitrogen, aii') with satisfactory results, we were led to 

 conjecture that the atomic weight of argon, respectively its specific 



the nitrogen molecules by the electric discharge, the atoms which are formed 

 combining with the potassium. Strutt, (Proc. Royal Soc. Serie A 85 219 and 

 subsequent volumes) found that the re-combination of the N-atoms to molecules 

 (which do not react with the potassium) is much accelerated by an increase of 

 the pressure. This was the main ground, on which we chose the dimensions of 

 the absorption-vessel large as compared with those of the burette. (The low 

 pressure also facilitates the production of the discharge). The fact explains in 

 particular, why the time of absorption in our analyses increases with the percentage 

 of nitrogen in the mixtures It is also known (Strutt 1. c. comp. also Koenig 

 Zeitschr. f. Electroch. 1915. 1 June), that metallic vapours accelerate the raolisa- 

 tion of nitrogen atoms. (An afterglow on interrupting the discharge was therefore 

 entirely absent in our apparatus), it is of course possible that mercury has this 

 property to a higher degree than potassium, although a different affinity of mer- 

 cury towards N may also play a part here; this might be one reason for the 

 stronger reaction of the potassium vapour 



) Landolt-Boenstein Phys. Ghem. Tabellen. 4tli ed. 



