( 780 ) 
above the solvent to bring about that an equal number of particles 
enter and leave through the membrane, is called the ‘osmotic tem- 
perature’ of this solution. Now the supposition naturally suggests 
itself that this state is reached when the thermodynamic potential of 
the substance passing through the membrane has again become equal 
in the solution and the solvent. This thesis would undoubtedly be 
correct, if the considerations concerning the kinetic signification of 
the thermodynamic potential which I gave in These Proc. April 1905, 
were perfectly true. There I tried to demonstrate that the physical 
meaning of the thermodynamic potential of an homogeneous phase 
on which no external forces act is nothing but the number of 
molecules which per second reaches a wall, which is placed in the 
middle of this homogeneous phase, when the wall does not attract 
the molecules, and is thick with respect to their sphere of action. I 
expressed this definition by the formula : 
MS CNE Eren Aret nd A cl | 
in which N is the number of molecules in question, and C a con- 
stant. It is clear that what has been said above follows from this 
formula; for the numbers of molecules which leave solution and 
solvent through the membrane, are then entirely determined by the 
thermodynamic potential. 
If now, however, on this ground, we try to determine from equa- 
tion (2), what the ‘osmotic temperature’ will be for a certain solu- 
tion, we come to an absurdity. For, again neglecting the terms with 
we get for the osmotic temperature at constant pressure: 
“0 
n AE WES he en ie SE (4) 
So this would imply that we could determine the total amount of 
(vo ea 
entropy of the solvent through the experimental determination of ——, 
Av 
And this now is an absurdity. For when we reason purely thermo- 
dQ 
dynamically and so define the entropy by dy = ~~ — and we have 
7 
used no other definition of the entropy in deriving equation (4), — 
the entropy has no definite value, but an indefinite, and undefinable 
additive constant occurs in it; only differences of entropy have a 
definite value. *) 
3. It is not difficult to point out the error in the reasoning which 
has led us to this absurdity. We have imposed the condition that 
1) Only by starting from another, kinetic definition of the entropy, we can assign 
a definite value to the entropy in a definite system of units. Cf. These Proc. XIII 
p. 705, seq. 
