108 
This is in conflict with the hyperbolical line War for this 
(2 
leads to an expression of the form: - 
Ee e(l—e) 8 
For a course from «=O to «=1 equations of these two types 
can certainly not perfectly accurately agree; it is, however, the 
question in how far they deviate within the region in which the 
observations lie (s—= 0.1" to-# == 0.4). Jf-now for z—015 0:28 
by 
136.2 + 5563.82 
and if we divide the result by the value for «= 0.4, we find: 
0.7342, 0.8223, 0.9110 and 1.0000 
0.3, 0.4 we calculate the value of the expression 
these values do not ascend linearly, but they differ from the purely 
linearly ascending ones: 
0.7336, 0.8223, 0.9110, 0.9997 
everywhere less than 1°/,,, the experimental errors certainly amount- 
ing to a few percentages. So it is clear that the discrepancies which 
exist between a formula of the type (7) and of the type (8), are 
much too small in the considered region to allow of an experimental 
decision. We must conclude that a formula of type (7) represents 
the experimental data as well as a formula of type (8) *). Farther 
reaching conclusions are of course excluded, as we already remarked 
1) Perhaps we may go still further and say that in the general case a formula 
as (7) represents the experimental relations better than (8). For according to the 
latter formula the total heat of mixing W and also the differential heat of mixing 
yr 
must always retain the same sign, while en the other hand for certain values 
of the a’s and b’s a reversal of sign is possible according to formuia (7). And this 
A7 
change of sign of Th’ which can never take place for a hyperbolical formula, 
di 
seems indeed to appear in reality in some cases e.g. for inulin, as appears from 
the subjoined table. 
i W in Cal. 
0 
0.052 
0.095 16.7 
0.116 19.0 
0.223 22.4 
0.293 23.0 
1.05 21.8 
It is also in connection with this deviation of the theoretically required formula 
