Jantjaey 28, 1921] 



SCIENCE 



81 



solution increases directly witli the radius of 

 tke ion I > Br > CI; where I has the greatest 

 and CI the smallest attractive action. 



This might be intelligible if the action of 

 the ions on the particles of water were electro- 

 static, since in this case the action of the 

 anion depends on the negative charge in its 

 outermost shell of electrons and the electro- 

 static effect should be the greater the farther 

 the shell is removed from the positive nucleus 

 of the ion; while the electrostatic effect of the 

 cation is due to the positive charge of the 

 nucleus and this should be the greater the 

 smaller the distance between nucleus and the 

 outermost layer of electrons, i.e., the closer the 

 positive nucleus can approach the water par- 

 ticles or the membrane particles on which the 

 ion is to act. 



rv 



We have alluded to the fact that collodion 

 membranes are not strictly semipermeable 

 and that crystalline solutes diffuse out from 

 the collodion flasks in our experiments. It 

 might be argued that the differences in the 

 flow of water measured in the preceding 

 chapter are due to differences in the rate of 

 diffusion of electrolytes from the side of the 

 solution to the side of pure water through the 

 collodion membrane. This assiunption is, 

 however, not tenable since it can be shown 

 that the diffusion of the solutes into the pure 

 water through the collodion membrane seems 

 to follow Fick's diffusion law according to 

 which the rate of diffusion of a solute is 

 directly proportional to its concentration and 

 this seems to hold equally in the case of 

 electrolytes and non-electrolytes. The specific 

 influence of solutions of electrolytes on the 

 rate of diffusion of water from pure water 

 through collodion membranes into solutions 

 can therefore not be due to any difference in 

 the rate of diffusion of electrolytes and non- 

 electrolytes through the membrane into the 

 pure water, but must be ascribed to a differ- 

 ence in the behavior of water towards these 

 two types of solutes. 



V 



We have thus far mentioned only the in- 

 fluence of electrolytes on the rate of diffusion 



of positively charged particles of water. 

 Perrin found in his experiments on electrical 

 endosmose that in certain cases the water 

 migrated to the positive electrode, namely 

 when the solution had an acid reaction, while 

 it migrated to the negative electrode when 

 the solution had an alkaline reaction. 'No 

 such reversal in the sign of electrification of 

 water can be produced in the case of pure 

 collodion membranes, since in this case the 

 water is always positively charged no matter 

 whether the solution is acid, neutral, or 

 alkaline. When, however, we deposit a film 

 of a protein on the inside (or on both sides) 

 of the collodion membrane the latter becomes 

 amphoteric. Wlien the solution is sufficiently 

 acid, the water migrates through the mem- 

 brane as if its particles were negatively 

 charged, while when the hydrogen ion con- 

 centration is lower, i.e., when the solution is 

 only very faintly acid or neutral or alkaline, 

 the water particles move through the protein 

 film of the membrane as if they were posi- 

 tively charged. 



When we separate an acid solution of a salt 

 by a collodion membrane possessing a protein 

 film, from a solution of a pure acid of the 

 same hydrogen ion concentration as that of 

 the salt solution, the hydrogen ion concentra- 

 tion being equal to or above 10"* N, the 

 water migrates through the pores of the mem- 

 brane as if its particles were negatively 

 charged and as if they were " attracted " by 

 the cation and " repelled " by the anion of the 

 electrolyte in solution with a force increasing 

 with the valency of the ion. In this case, 

 water is " attracted " more powerfully by salts 

 with trivalent cation, e.g., AICI3 or LaOlg, 

 than by salts with bivalent cation e.g., MgCl^ 

 or CaCl, ; and it is " attracted " more power- 

 fully by the latter than by salts with mono- 

 valent cation, e.g., NaCl or KCl; while nega- 

 tively charged water is not " attracted " by 

 salts with bivalent or trivalent anions, e.g., 

 Na,SO^ or ISTa* oxalate or ISTa^FeCCN^, etc. 



In the case of salts with monatomic and 

 monovalent cations the " attraction of " the 

 salt for negatively charged water seems to 

 increase inversely with the radius of the 



