OF VITAL PHENOMENA 115 



become more permeable to anions or less permeable to cations 

 — perhaps both. 



Selective permeability of membranes to ions has often been 

 noticed in electrical transference experiments. Hittorff separated 

 the solution between the electrodes into segments by means of 

 ox gut membranes, in order to prevent backward diffusion or 

 convection currents. He supposed the membrane did not affect 

 the results (transference numbers), but it is now known that 

 the effect is small in some cases, very large in others. Bein 

 (1899) found that earthenware membranes were the safest to 

 use. Fish bladder membranes caused but small errors with neu- 

 tral salts of alkali metals, but large errors with CuS0 4 solutions. 

 The S0 4 ions passed the membrane, but the Cu ions combined 

 with the substance of the membrane, and hence were held back. 



The action of gelatine membranes in holding back H ions in 

 a similar manner has been noticed by Girard and by Cybulski 

 (1903) and Cybulski and Dunin-Borokowski (1909). 



Where the membrane clearly forms a separate phase the elec- 

 tric phenomena have been called by Nernst and Riesenfeld 

 (1902) phase boundary forces. If the speed of the ions in two 

 phases is different, the addition of the electrolyte to either phase 

 will cause a difference of potential at the phase boundary, which 

 is different from that which occurs in a similar concentration 

 gradient in either phase alone. The limiting case is that in which 

 the speed of either ion becomes zero (in the membrane). Such 

 is the case with a metallic membrane, which may be considered 

 impermeable to any ion except the cation of the same metal. 

 Although the atom and its electron part company in passing 

 through the membrane, and the ion as such does not pass through, 

 the effect is the same, since an ion is given off from one side 

 at the instant that an identical ion is deposited on the opposite 

 side of the membrane. It seems proper, therefore, to speak of 

 the permeability of a metallic membrane to an ion, when the 

 mechanism of the passage of the ion or its components through 

 the membrane is of no significance to the point of discussion. 

 When an electrolyte dissociates into two ions to only one of 

 which the membrane is permeable, the emf may be calculated 

 by Nernst's formula for electrode potentials. The emf is 59 mv 



