and the Observed Velocity of Sound in Air and Gases. 31 



Experiment. 



Calculation. 



By weight. 

 O =2301 

 N =76-99 



100-00 



By weight. 

 =23-595 

 N =76-405 



100000 



The discordance in this case being greater than in the former, 

 suggests that possibly in both instances the discrepancy may be 

 attributable to errors in the determinations of the densities of the 

 gases. The annexed Table will test the validity of this sugges- 

 tion. The densities used are — 1st, the mean values as given by 

 the experiments of Regnault ; 2nd, the maximum values given 

 by the same experimenter; and 3rd, the values given by the 

 experiments of Dumas and Boussingault. 



Calculated percentage of oxygen and nitrogen. 



Experiments of Regnault. 



Experiments of Dumas. 



£=1-105612, 



d 1 =0-97 1346. d= 1-105636, d'=0-9715395. 



d= 1-1057, <f =0-972. 



Volume. 



O =21341 



N =78-659 



100-000 



Weight. 

 O =23595 

 N =76-405 



Volume. 



0= 21-224 



N= 78-776 



Weight. 

 O =23-466 

 N =76-534 



100000 



Volume. 

 O =20 942 

 N =79-058 



Weight. 

 O =23-156 

 N =76-844 



100000 



100-000 



100000 



100-000 





Experiment. 

 O =20-96 

 N =79 04 



Calculation. 



O =23-174 

 N =76-775 



99-949 



Experiment. 



O =20-96 

 N =7904 



100-00 



Calculation. 



O =23174 

 N =76-790 



99-964 



Experiment. 

 O =2096 

 N =79-04 



10000 



Calculation. 



O =23175 



N =76-827 



10000 



100002 





Calculation. 

 O =20812 



N =79-261 



Experiment. 

 O =2301 

 N =76-99 



10000 



Calculation. Experiment. 



O =20-812 ! O =2301 



, N =79245 N =76-99 



Calculation. 



O =20-810 

 N =79-208 



100-018 



1 



Experiment. 



O =2301 

 N =76-99 



10000 



100073 



100057 



100-00 



Taking the results obtained by the volumetric method as the 

 test, it seems, by a comparison of the numbers in the foregoing 

 Table, that the densities given by the experiments of Dumas and 

 Boussingault furnish results more nearly accordant with experi- 

 ment than those calculated from the densities afforded by the 

 more recent experiments of Regnault. The methods of the latter 

 experimentalist are so far superior to those of his predecessors, 

 as to preclude the assumption that the densities given by the 





