1. Collodion bags coated with gelatin on the inside were filled with a M/256 solution of neutral salt (e.g., NaCl, CaCl2, CeCl3, or Na2SO4) made up in various concentrations of HNO3 (varying from N/50,000 to N/100). Each collodion bag was put into an HNO3 solution of the same concentration as that inside the bag but containing no salt. In this case water diffuses from the outside solution (containing no salt) into the inside solution (containing the salt) with a relative initial velocity which can be expressed by the following rules: (a) Water diffuses into the salt solution as if the particles of water were negatively charged and as if they were attracted by the cation and repelled by the anion of the salt with a force increasing with the valency of the ion. (b) The initial rate of the diffusion of water is a minimum at the hydrogen ion concentration of about N/50,000 HCl (pH 4.7, which is the point at which gelatin is not ionized), rises with increasing hydrogen ion concentration until it reaches a maximum and then diminishes again with a further rise in the initial hydrogen ion concentration.
2. The potential differences between the salt solution and the outside solution (originally free from salt) were measured after the diffusion had been going on for 1 hour; and when these values were plotted as ordinates over the original pH as abscissae, the curves obtained were found to be similar to the osmotic rate curves. This confirms the view expressed by Girard) Bernstein, Bartell, and Freundlich that these cases of anomalous osmosis are in reality cases of electrical endosmose where the driving force is a P.D. between the opposite sides of the membrane.
3. The question arose as to the origin of these P. D. and it was found that the P.D. has apparently a double origin. Certain features of the P.D. curve, such as the rise and fall with varying pH, seem to be the consequence of a Donnan equilibrium which leads to some of the free HNO3 being forced from the solution containing salt into the outside solution containing no (or less) salt. This difference of the concentration of HNO3, on the opposite sides of the membrane leads to a P.D. which in conformity with Nernst's theory of concentration cells should be equal to 58 x (pH inside minus pH outside) millivolts at 18°C. The curves of the values of (pH inside minus pH outside) when plotted as ordinates over the original pH as abscissae lead to curves resembling those for the P. D. in regard to location of minimum and maximum.
4. A second source of the P.D. seems to be diffusion potentials, which exist even if no membranes are present and which seem to be responsible for the fact that the rate of diffusion of negatively charged water into the salt solution increases with the valency of the cation and diminishes with the valency of the anion of the salt.
5. The experiments suggest the possibility that the establishment of a Donnan equilibrium between membrane and solution is one of the factors determining the Helmholtzian electrical double layer, at least in the conditions of our experiments.