32 On the Velocity of Sound. 



earlier experiments are to be preferred. Hence we are led to 

 look for an explanation of the discrepancy between experiment 

 and calculation in the direction of a possible change of volume 

 taking place during the mixture of the two gases. Assuming 

 the values of the mean densities given by Regnault's experi- 

 ments, a change of volume to the amount of only 0*0512 per 

 cent, would account for the whole discrepancy ! For my original 

 equations now become 



dxV-f-d'xV' = 99*9488xl, 



and 



V + V'=100. 



Hence 



Tr 99-9488-100 xd' 2-8142 on ftfl , 



V= jp^p = 0^3426 = 20 ' 96 P er Cent 



Similarly, 



V'= =79-04 



100-00 



Or, otherwise, estimating the percentage in relation to the mix- 

 ture, a diminution of 0*0527 per cent, in volume would be suffi- 

 cient : thus 



dxY + d' XV' = 100x1, 

 and 



V + V =100*0527. 



Hence 



Tr 100- 100-0527 xd ! 2*8142 _. na 



V = d^ji = 003426 = 2 °' 96 per Cent 



Similarly, 



V' = = 79*04 



100-00 

 Consequently, if this view is correct, there should be a diminu- 

 tion of volume when oxygen and nitrogen are mixed in the pro- 

 portions above indicated, to the extent of somewhat more than 

 0*05 per cent. A direct experimental determination of the cor- 

 rectness of this explanation is certainly within the limits of 

 accuracy afforded by the refinements of modern methods*. Its 

 verification would fortify the view in relation to the mutual action 

 of the gaseous constituents of the atmosphere, towards which we 

 are led by acoustic phenomena. 



Columbia, South Carolina, 

 July 8, 1861. 



* The memorable experiment which Berthollet executed in the cellars 

 of the Observatory at Paris is usually cited as a direct proof that no change 

 of volume takes place by the mixture of hydrogen and carbonic acid. As a 



