1083 
in greatly diluted, non-solidifying solutions'), which, then, ought 
also to be considered as systems of two phases; a view, which to 
us seems to be without any foundation whatsoever. 
6. the forming of a membrane in gels by opposite diffusion of 
salts that give a precipitate’). It is not clear to us, why these preci- 
pitates should only arise in the cavities of the gel. 
Let us first of all examine Bancrort’s explanation, which is 
identical with the one, originally put forth by us, but whieh we have 
rejected for the reasons, we shall presently discuss. It assumes two pha- 
ses in the gel — one with much, the other with little water — 
which are separated by curved surfaces. The equilibrium in the 
vapour decides the composition of the second phase; the water, 
which is taken by the gelatine up in the liquid, forms the first. 
According to our observations, the concentrated phase of agar would 
contain 50°/, agar, whilst 3 to 5°/, would follow from Harpy’s 
pressing experiments. So this does not agree exactly! If we try to 
obtain — as is necessary a more detailed conception of the struc- 
ture of the gel, we have to choose between an open and a closed 
cell structure. Assuming the former, one could only accept BANcROFT’s 
hypothesis, if the surface tension of the diluted phase with regard 
to the concentrated one is as that of mercury with regard to glass. We 
have investigated, whether this is the case by covering glass capillaries - 
on the inner surface with a thin layer of gelatine, agar, celloidin or 
rubber. We found a behaviour as that of water-glass ; only in the case 
of vapour-swollen or dry gelatine we observed a convex meniscus; 
gelatine, swollen in liquid, behaved as the other bodies. An open 
cell structure is, therefore, not consistent with BANCROFT’s explanation. 
Another fact may be mentioned, which also speaks against this 
assumption; a plate of gelatine, dipped half way and vertically 
in water, only swells for the lower part, while the part above the 
water surface presents exactly as gelatine in equilibrium with vapour. 
For if there were an open cell structure, the canals should fill them- 
selves by capillary action. Whether an open or a closed structure is 
obtained, will depend on the question, which phase separates first. If 
this is the most concentrated and consequently the most viscous one, an 
open structure will arise and the water will have a concave surface; 
if, on the contrary, the latter appears first, it will cf course show 
a convex meniscus. If, therefore, we accept BaNncrort’s explanation, 
we are obliged to suppose that the phase with much water separates 
first in all the systems that show the phenomenon ; of course, this 
1) See especially Bacumann, loc. cit. 
*) Becnnotp, Z. phys. Chem. 52, 185. 
