44 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 54 



Including these figures with those of the first series, the mean of 

 all becomes 2. 31023, ±.00008. Correcting these data for the compres- 

 sion of the empty globes, the mean weights become, for oxygen, 2.6276, 

 ±.00004, and for nitrogen, 2.31079, ±.00008. The ratio between them 

 is 16: 14.0704, ±.0005. 



The combination of these determinations is as follows : 



Dumas and Boussingault 14.0657, ± .0113 



Regnault 14.0570, ± .0009 



Von Jolly 14.0805, ± .0003 



Leduc 14.0729, ± .0005 



Rayleigh 14.0704, ± .0005 



General mean 14.0758, ± .00022 



Now, to correct this mean for the argon contained in the nitrogen. 

 Good measurements have shown that normal air contains, by volume, 

 0.937 per cent of argon, and 78.122 of nitrogen. The density of argon, 

 referred to the oxygen standard, is 19.940. Applying these values, the 

 final figure for nitrogen, derived from air, becomes 14. 0052, ±.00022, a 

 result which is in harmony with others to be considered presently. 



In Eayleigh's investigation of the density of nitrogen it was found 

 that nitrogen from chemical sources was lighter than that extracted from 

 the atmosphere. This led to the discovery of argon, to which reference 

 has already been made. In two of his memoirs ^ Eayleigh has given 

 determinations of the density of " chemical nitrogen " obtained from 

 nitrous oxide, nitric oxide, ammonium nitrite, urea and magnesium 

 nitride, and the gas from all these sources is precisely the same. His 

 weights, given now as one series, and representing the same volume as 

 those previously cited, are as follows: 



2.30143 From nitric oxide 



2.29890 



2.29816 



2.30182 



2.29809 From nitrous oxide 



2.29940 



2.30074 



2.30054 



2.29849 From ammonium nitrite 

 2.29889 



2.29870 



2.29850 From urea 



» 2.29918 From magnesium nitride 



Mean, 2.29949, ± .00024 



Proc. Roy. Soc, 55, 340, 1894, and 57, 266, 1895. 



