244 



H. K. SCHACHMAN AND R, C. WILLIAMS 



diffusion data are desired for molecidar weight determinations, this method 

 is the only one that is satisfactory. 



The solution and solvent are placed in sej^arate compartments of a cell 

 containing a porous plate through which the solute diffuses. In the early 

 designs the solution was placed in an inverted funnel across which a sintered 

 glass disk was cemented, and this cell was then brought into contact with 

 the solvent which was in a second, open vessel. The solution was above the 

 solvent, and the diffusing molecules after passing through the porous plate 

 led to stirring of the lower liquid by virtue of the increase of the density of 

 the surrounding liquid to a level above that of the pure solvent beneath. 



SOLVENT 



SOLUTION 



Fig. 1. Diagram of closed porous disk diffusion cell. The glass enclosed magnetic 

 stirrers are so constructed with respect to thickness of glass and length of ■n^ire that the 

 magnet in the upper compartment smks in the liquid while that in the lower one floats. 

 The stirrers are caused to rotate by surroundmg the cell with a rotating permanent 

 magnet. The diffusion cell is placed in a constant temperature water bath so that thermal 

 gradients across the cell are avoided. 



Tliis system was affected by uncertainties resulting from mass flow of solu- 

 tion across the disk due to temperature gradients across the cell and to 

 convective flow" of solution tlirough occasional capillaries of large diameter. 

 The possibility of mass flow through the membrane, with the consequent 

 tremendous errors in the diffusion coefficient, has been reduced substanti- 

 ally in a modified version of the Northrup-Anson cell. This was effected by a 

 change from an open or single cell to one having two closed compartments 

 completely filled with licjuid. A sketch of such a cell (Stokes, 1950) which is 



