(738) 
The experimentally measurable difference in pressure on either side 
of the membrane must therefore have the same value as the quantity 
defined by us. 
Lorextz *), however, has shown that the assumption made here 
concerning the membrane is by no means necessary. On the contrary ; 
if we assume that the membrane is thick compared with the sphere 
of action, that its substance fills a volume large compared with the 
apertures present and that it feebly attracts the molecules of the 
dissolved substance, whereas these are strongly attracted by the 
solvent — none of which are improbable assumptions — we arrive 
at the result, that none of the dissolved particles reaches the membrane, 
much less exerts a pressure on it; the membrane is then quite 
surrounded by the pure solvent. And that this case is really the 
usual one in nature is made probable by the fact, that it is by no 
means always the smaller molecules which pass the membrane, as 
we assumed above. The membrane seems therefore not to work as 
a molecule-sieve. We are then easily led to suppose that the mem- 
brane does not exert a positive repulsion at all on the non-passing 
substance, but that it only attracts those particles much less strongly 
than the solvent, so that the dissolved particles do not pass through 
the membrane, because they occur but extremely rarely in its neigh- 
bourhood. This view is supported by the fact, that only those 
substances seem to be non-passing which are not easily converted 
to vapour, and so cannot reach the limits of the liquid in virtue of 
their own thermal motion alone. 
However this be, also in this case our conclusion holds good. 
For when the molecules of the dissolved substance do not (or only 
in an intinitely small number) reach the membrane, two planes will 
be found not far from the membrane, A where the molecules of the 
dissolved substance still have their normal density, B where this 
density has diminished to zero. Between B and the membrane we 
find then pure solvent. If we wished to discuss such a layer fully, 
we should, of course, have to give a theory, as VAN DER WAALS 
has given for the transition liquid vapour’), extented to a mixture 
in the way van Expik*) has done. But for our purpose this is 
fortunately not necessary. We need only observe, that the layer 
AB as a whole has now exactly the same influence on the condition 
of motion of the dissolved molecules as the mathematical upper surface 
of the membrane had just now. The layer AB as a whole will now, 
Det 
8) Verh. dezer Ak. (2) 1; Arch. Néerl, 28, 121 and Zsch. phys. Ch. 13, 657. 
5) Diss. Leiden 1898. 
