( 706 ) 
with the exception of an additive constant; what we said appears 
in a much more conclusive way from this that the values of these 
constants depend on the system of units used, and that in such a way 
that with change of these unities new additive constants appear. If 
e.g: the unit of time is multiplied by ¢,, the constant is increased 
by n log ¢;*). On the other hand the values determined by NERNsT can, 
evidently, not change by a change of the unit of time *); so these 
values can never be the whole of the entropy constant, however 
kinetically defined. So we have to divide the constants of entropy 
of the reacting substances into two parts, one part which remains 
constant during the reaction, and which finally disappears from the 
equation of equilibrium, and one part which changes during the 
reaction. Or to take a concrete example, the constant of entropy of 
a definite quantity of hydrogen must consist of a part that refers to 
the atoms MH, and which remains constant for the same number of 
atoms in whatever state they may be before and after the reaction, 
and of a part which is determined by the configuration in which 
these atoms are found, and which therefore varies during the reaction. 
The former part must depend on the unit of time in the way 
indicated by, GriBBs or BOLTZMANN ; but being constant in every chemical 
reaction, it will necessarily disappear from the difference of entropy 
before and after the reaction. The latter part, on the other hand, 
determines the equation of equilibrium, and so can reversely, be 
found from it. The values found by Nernst, must give us information 
about this part. What will be the physical meaning of it? The only 
indication which thermodynamics can give about it, is implied 
in the observation in note *). For when the number of molecules 
changes, log A becomes dependent on the unit of volume in this 
sense that an additive constant rv loy c‚ appears, in which c‚ is the 
ratio between the old and the new unit, and rv the modification of 
the number of molecular quantities by the reaction. So the difference 
of entropy must not depend on the unit of volume for a reaction 
with constant number of molecules, and for another reaction it 
must be increased by an additive constant on modification of the 
unit of volume. So it lies at hand to suppose that the quantities charac- 
teristic of the difference of entropy must be logarithms of volumes 
1) Cf. Gress, Principles in statistical mechanics, p. 19. Bottzmann’s H depends 
on the units in another way, but it is also increased by an additive constant if 
the unit of time is increased and the other units remain constant. 
2) As Ink can depend on the unit of volume, when namely the number of 
molecules changes during the reaction, the values found by Nernst can be subjected 
to a change when we change the unit of volume. 
