IN STUDYING QUESTIONS OF CHEMICAL MECHANICS. 371 
abscissa is e or , does not render them less incompatible 
E 
Pee 
with the more complex supposition which would transfer them 
successively, by abrupt intermittences, to different right lines 
having for abscissze = as in the third possible case of the phase 
of non-saturation, which we have discussed in § 27. 
58. Since, therefore, among the three mechanically supposable 
modes of constitution of non-saturated solutions which we have 
established, there are two which we find incompatible with the 
law of variation assigned to the function [a] by experiment, in 
so far as they suppose that the whole mass of acid, or a portion 
of this mass, forms with water fixed groups, which a subsequent 
addition of this liquid does no further modify, it is requisite, 
excluding these, to admit the third, which consists in the fact, 
that the acid forms, with the water presented to it, groups per- 
petually variable, which only become fixed by an infinite addi- 
tion of water. Now, this is precisely the same consequence to 
which we were led in discussing the physical law, according to 
which the densities of the solutions here considered are modi- 
fied, without term or limit, by fresh quantities of water subse- 
quently added. 
59. Thus, from these two independent classes of facts, it 
equally results that, in liquid systems composed of tartaric acid 
and water in different quantities, at a constant temperature, the 
water does not combine specially with certain molecules of the 
acid, forming with them complete and saturated chemical groups, 
but is divided among them all uniformly, so as to form groups 
always incomplete, which become the actual constituent elements 
of the total mass. The specific deviating action of these groups 
is therefore expressed in each solution by the function a proper 
to media of uniform constitution, as we have seen in the first 
case of non-saturation in § 27, which is here realized. Now, 
the physical and rectilinear law 
[2] = (A) + (B)e, 
which represents the variations of the function [«], allows us 
a 
to calculate generally those of the function Te 
In fact, since we 
have 
