30 Prof. J. Le Conte on the Discrepancy between the Computed 



tuents of the air as has been done in the foregoing computation. 

 In other words, the volume of oxygen is determined by detona- 

 tion with hydrogen ; the nitrogen is estimated by the difference ; 

 in both cases the assumption being that no alteration of volume 

 takes place in the separation of the two gases. Of course, in 

 order to compare these results with those of calculation, the 

 proportion of oxygen and nitrogen is estimated for air which has 

 been deprived of other constituents. The following Table will 

 place this in a clear light : — 



By Lewy*, at New- 

 Granada. 



By Bunsenf, at 



Marburg. 



By Regnault J, at 

 Paris. 



By calculation. 



By volume. 

 Nitrogen = 21-023 

 Oxygen = 78977 



100000 



By volume. 



0= 20 964 

 N=5 79-036 



100-000 



By volume. 



0= 20 93 

 N = 79-07 



10000 



By volume. 

 0= 21-341 



N= 78-659 



100000 



The difference between these numbers and those deduced 

 from the density of the constituent gases (varying from 0*32 to 

 0*41 per cent, of the volume) seems too great to be accounted for 

 by the errors of the chemical analyses. For it must be recol- 

 lected that M. Lewy regards the improved volumetric method of 

 MM. Regnault and Reiset (that employed in the above analyses) 

 as reliable within less than tttu o"u °f tne volume, or 0*01 per 

 cent.§ It is likewise too great to be ascribed to the known 

 variations in the composition of the atmosphere ; the extreme fluc- 

 tuations in this respect scarcely ever amounting to 0*10 per cent, 

 of the volume. 



On the other hand, let us look at the results of the analysis of 

 the air by the method of weights. This method, although not 

 claiming such extreme accuracy as the preceding, is entirely 

 independent of any hypothesis in regard to the volume of the 

 gases composing the mixture, and is capable of giving both the 

 oxygen and nitrogen directly by weighing. The mean results 

 of the experiments of MM. Dumas and Boussingault at Paris 

 furnish us with the following comparison : — 



* Comptes Rendus, September 29, 1851 . Also in Phil. Mag. S. 4. vol. ii. 

 p. 500 etseq. (1851). 



t Bunsen's ' Gasometry,' translated by Roscoe, p. 78. London, 1857. 



J Regnault's Cours Elementaire deChimie, 5th edit. vol. i. p. 147. Paris, 

 1859. 



§ Comptes Rendus, September 29, 1851. Also Phil. Mag. S. 4. vol. ii. 

 p. 501 (1851). 



