1201 



compression. When we pnt in this « :^ (perfectly hard, incom- 

 pressible spheres), we obtain : 



V 



which in this case we should also have obtained immediately by 

 supposing Q constant under the integral sign (see the first Paper). 

 For then the integral had become (the indices and b refer to the 

 interval of collision; in Communication I (>„ was written instead of p^): 



Po 



2 8n(6,,)^ j (()—(>,) rf() = — 8 ?i (6,,)^ (9„ - QbY, 



and as 8 (r^ — r^y is = e [q^ — q^Y = RT' -. N\ Vi, becomes as above. 

 We wrote RT' and not RT. This is owing to this that the mean 

 relative normal velocity Uq, with which a molecule strikes tigainst 

 another, is increased by a certain amount on account of the traversing 

 of the sphere of attraction, and this in such a way that 



in which S is given by (y) in § 9. But ^l^im^ being =;;: RT : N, also 



RT' — RT=S, (11) 



so that we get : 



RT~ '^ RT~^ 

 hence 



n = RT <^ X e ''"^ 



V 



through which f{b) becomes ^ e^^l RT, in agreement with (5). 



XIII. Objections to the Function e-^^'-. Another 

 Factor of Distribution. 



For f {a) and f{b) we find therefore resp. the expressions 



m=e ' =1+ ^4H^K 



(12) 



And these two functions of the temperature are ?iot the same, 

 as the experimental results (see the foregoing communication) have 

 taught. The difference is such that even a very considerable error 



