78 VISCOSITY OF GASES 189 



sideration in which for simplicity equal mean values, both 

 of the speed and the free path, were ascribed to all the mole- 

 cules instead of values which really vary from molecule to 

 molecule. A more exact calculation of the internal friction of 

 gases, carried out in the fourth Mathematical Appendix 

 ( 47* and 48*) in accordance with the rules of the calculus 

 of probability, shows that by the simpler calculation the 

 numerical factor of the formula has become affected by a 

 slight error. It is more exact to put 



77 = 0-30967 pLtl 



for the coefficient of viscosity, where fl is the arithmetical 

 mean of the speeds on Maxwell's theory, and 



is the mean value of the free path on the same theory 

 ( 68). 



Although atmospheric air is a mixture of molecules of 

 different kinds which have unequal speeds and traverse 

 unequal paths, yet we may apply the theoretical formula to 

 air, if not with exactness, still with approximate admissi- 

 bility, to calculate the mean free path of its heterogeneous 

 molecules. My transpiration experiments l gave, for the 

 coefficient of viscosity of air at C. values from 



77 = 0-000168 to T? = 0-000174, 



which, as mentioned already in 75, are expressed in terms 

 of the centimetre, the density of water, and the second as 

 units. Puluj, 2 from his transpiration experiments, found 



77 = 0-000179 ; 



von Obermayer, 3 also by transpiration experiments by 

 two different methods, 



77 = 0-000171 and 0-000168 ; 



1 Fourth and fifth memoirs on the viscosity of gases. Pogg. Ann. 1873, 

 cxlviii. pp. 37, 203. 



2 Wiener Sitzungsber. Abth. 2, 1874, Ixix. p. 287 ; Ixx. p. 243. 



3 Carl's Rep. f. Experimentalphysik, 1876, xii. p. 13 ; Wiener Sitzungsber. 

 Abth. 2, 1875, Ixxi. p. 281 ; 1876, Ixxiii. p. 433. 



