dT y=-b  v 2 
dp RT a (5) Ti 
Poe 
and 
‚dp a AFD Ly 
AT. we mete tala +o) 
and 
Tr. Ere feb 
T dp AN a Er dj 
p dT pv v 
So we have at 7: 
fe—1 a 
: nee ai N 
De ten P+ DP skr 
ke 
F Te Tr” » a fe—1 1 
“or — |=| — | we write ==} ‚ hence 
PAT 7 > RT oe ewe : 
Tiz 
ED eer es es ape 
oRT vm s l+um 
If we compare this value with that which we obtained on the 
supposition of constant a, we see that it is 
(14u) m* 
times smaller. This expression can be either greater than 1 or 
smaller than 1. For u=1 it hecomes 2m; and so for all values of 
m above 4 the value which we had wished greater, becomes smaller 
on the contrary. 
Nor could the supposition that 6 is a function of the temperature 
a . 
tend to make aa larger. In this case we should find: 
UL 
a % fel vr db 
vRT wv s (ozb)? \dT i. 
The explanation of this apparently paradoxical result must be 
found in this that if @ had been a temperature function, fx —1 
would also have been much greater, and we may accept this as an 
indirect proof that the quantity a is no function of the temperature. 
So we have to look elsewhere for the cause of the fact that 
equation («) does not hold, and to put the question whether the 
existence of quasi association can account for the found differences. 
1 
m 
