CHEMISTRY: /. WEB 
445 
brane as if its particles were positively charged. On the basis of this 
statement the initial expulsion of water by solutions of alkalies in con- 
centrations above M/256 should be due to the fact that above a con- 
centration of M/256 the repelling action of the cation upon the posi- 
tively charged water particle increases more raipdly with increasing 
concentration of the elctrolyte than the attractive action of OH ions. 
When the cation is bivalent, e.g., when we use Ca(0H)2 or Ba(0H)2 
as alkahes, the repelHng action of the Ca and Ba ion with their two 
charges is so much stronger than the weak attractive action of the OH 
ions that water diffuses more rapidly from solution into pure water than 
pure water can diffuse into the solution, and the level of the water rises 
(during the first twenty minutes or more) in the pure solvent and falls 
in the solution. It harmonizes with our suggestion that the negative 
osmosis does not exist in the case of alkahes with monovalent cation 
like NaOH or KOH. 
If the solution is an acid, water diffuses through the membrane as if 
its particles were negatively charged, being attracted by the H ion and 
repelled by the anion of the acid. In concentrations of acid of M/256 
to about M/8 the solution expels during the first twenty minutes water 
into the pure solvent and the more the higher the concentration. This 
negative osmosis should be due to the fact that within this range the 
repelling action of the anion of the acid upon negatively charged water 
particles increases more rapidly with the concentration than the attrac- 
tive action of hydrogen ion. 
This suggestion is supported by the fact that the rate in which the 
acid solution expels water increases for the strong acids with the valency 
of the anion. Thus the amount of negative osmosis is small or negligible 
in the case of HCl or HNO3, is considerable in the case of H2SO4 and 
oxalic acid, and still greater in the case of H3PO4. The phenomenon is 
not noticeable in the slightly dissociated acetic acid. 
5. The definition of osmotic pressure as given in the introduction of 
this paper suggests that in the case of strictly semipermeable membranes 
the permanent osmotic pressure of the solution should show a simple 
relation to the difference in the initial rate of the diffusion of water in the 
two opposite directions through a membrane. Our experiments have 
shown that the initial rate of diffusion of water through a collodion mem- 
brane is determined for lower concentrations of electrolytes to a large ex- 
tent by the electrostatic effects of the ions present and we should expect 
that these electrostatic effects would also influence the permanent osmotic 
pressure of solutions of electrolytes. Collodion membranes which are 
