( 705 ) 
absolute zero? but that he should have asked: How can 1 get to 
know anything about those constants of entropy ? 
Or in other words: The regularities discovered by Nernst do not 
prove anything about the absolute zero, much less about the equality 
OA 7 
of { — |) and — ) in that point, but they show that in the cases 
(or). Gr), | 
examined by Nernst there must exist a certain relation between the 
constants of entropy of the reacting substances and other quantities 
characteristic of these substances, in this sense that the difference of 
the sum of the constants of entrupy before and after the reaction 
can be calculated from quantities which are characteristic of the 
pure, unmixed substances. 
6. Now in what way can we account for the fact that such a 
relation appears to exist, and of what nature are the quantities 
characteristic of the pure substances which occur in it? It is clear 
that we cannot restrict ourselves to purely thermodynamic considera- 
tions to answer these questions. For the way in which the entropy 
; dQ eae 
is defined in thermodynamics, M= naturally implies that the 
constant of entropy is an indefinite and arbitrary quantity. Only 
when we succeed in giving another, integral definition of the entropy 
it is conceivable that we should get a better insight into this constant. 
So we have to take recourse to the kinetic methods of BOrTZMANN 
and GiBBs, by the aid of which integral definitions have really been 
drawn up for the entropy so that it is possible entirely to determine 
the quantities defined by them, at ieast for the rarefied gas state in 
a definite system of unities. The entropy for a monatomic gas defined 
according to the method of GiBBs has been calculated by one of us 
(Q) in his thesis for the doctorate p. 56: 
5) 3 
n= ont 5 nlog2a64m + log v. 
It is now natural to suppose that the relation discovered by Nurnst 
means that the entropy-constants thus defined are the quantities which 
determine the value of the constant member of the equation of 
equilibrium. But on further consideration it appears that this supposition 
is untenable, at least for so far as it concerns these entropy constants 
as wholes. Not only because Gipss’s definition, even though one takes 
the dye, ), leads to another constant than BorirzMann’s H contains, 
which, moreover, need only be equal to the thermodynamic entropy 
1) Cf. Postma. These Proc. Oct. 1908, p. 311. 
