48 Dr. Schroder van der Kolk on the Velocity of Sound. 



same^ but the probable error would be somewhat smaller if each 

 observation had first been reduced to the same temperature and 

 pressure. But, on the other hand, this statement of the proba- 

 ble error does not include the errors of the constants employed 

 in the reduction, nor a possible error in the rate of the clocks. 

 The former, however, are taken from Regnault's determinations, 

 and are known within much less than 2 oVo^ °f the truth, and 

 the latter cause could not well occasion any larger error than this. 



Still, though it is not quite certain that tbe true value lies 

 within the limits indicated by the probable error, I am of opi- 

 nion that the value we have found is certainly accurate within 

 Yq-qq) and hence it can be asserted with confidence that the 

 true value lies between 3331 and 332*44. 



There can be no question of controlling this value by theory, 



c 

 for the value of - is not known with anything like the same 



c i 

 accuracy. 



Conversely, however, we can calculate the value of this func- 

 tion from the formula s=\/ — =-^ . - • 



V b c, 



s = 332-77 



H = 0-760, 



$ = 13-59593, 



g = 9-8096, 



b = 0-001293187. 



Hence we get 



-=1-4128 ±0-0008, 



taking account of the probable error on the velocity of sound. 

 If we again assume roW ^° ^ e ^ e extreme nm ^ °f the error 



on the velocity of sound, we find that the value of —must in 



c i 

 any case lie between 1*4104 and 1*4152. 



Strictly speaking, this value must still undergo a slight cor- 



c 



rection, for we have here assumed that the value of — is the same 



for air and for aqueous vapour. The quantity of the latter at 



the time of these observations was about T J T of that of the 



c 

 air. The value of — for aqueous vapour is not known ; but if 



we take the value 1*2 which Dulong found for easily condensi- 

 ble gases, such as sulphurous acid, we get a correction of 0*0026, 



