266 PHENOMENA DEPENDENT ON MOLECULAR PATHS 100 



Hausmaniger 1 has remarked that the formulae employed 

 by Wait z are founded on the assumption of the constancy 

 of the coefficient of diffusion, and therefore do not apply 

 to the case when it is variable. For the same reason, the 

 conclusions which Winkelmann has drawn from his 

 observations can be maintained only so far as it is shown 

 by them, too, that the coefficient of diffusion, as my theory 

 had predicted, is variable with the ratio of mixture, and 

 therefore also with its rate of fall. 



Hausmaniger has therefore taken the trouble to re- 

 calculate Waitz's observations by formulae in which the 

 coefficient of diffusion is taken as variable. But this 

 attempt, too, has unfortunately been unable to obtain 

 agreement between theory and observation ; for the mea- 

 surement of the variability of the coefficient, since it is 

 small, will be made too uncertain by the errors of observation 

 for the law of variation to be established with sufficient 

 exactness. 



A comparison of the numbers found by the different 

 observers shows that the uncertainty of the observations is 

 really as large as we say. For the diffusion of carbonic 

 acid into air Loschmidt 2 has found the value of the 

 coefficient, referred to centimetres and seconds, for C. 

 and 760 mm. pressure, 



D = 0-142, 



while von Obermayer 3 gives 



D = 0-135. 



These two numbers show as great a difference as the 

 numbers observed by Waitz, which, according to von 

 Obermayer 's calculation, assume the values 



0-131, 0-133, 0-137, 



when they too are reckoned for the temperature C. and 

 the pressure 760 mm. 



We should also on theoretical grounds have been led to 



1 Wien. Sitzungsber. 1882, Ixxxvi. Abth. 2, p. 1073. 



2 Ibid. 1870, Ixi. Abth. 2, p. 367 ; Ixii. Abth. 2, p. 468. 



3 Ibid. 1880, Ixxxi. Abth. 2, p. 1102. 



