114 L. C. CRAIG, W. KÖNIGSBERG, A. STRACHER, T. P. KING 

 Both acetylated membranes readily pass amino acids. The further possi- 

 bihties here do not require emphasis. 



It would appear that the considerable data we have now accumulated all 

 support the view that membrane diffusion with cellophane can be studied 

 in such a way that the size and (or) shape of the molecule is the predominant 

 property being measured. Perhaps, in conclusion, a speculation in regard 

 to the basis for the membrane selectivity noted is in order. 



Regardless of how thin a membrane may be, it can be regarded as offer- 

 ing a stationary structure with interstices or passages through which the 

 molecules must diffuse. These passages will not all be of the same size. 

 Thus a cross section might appear to be a mosaic of different sized holes 

 as shown in Fig. 7. Some of the holes will be smaller than the molecular 



Fig. 7. Schematic drawing of a hypothetical cross section of a membrane. 



dimensions of the solutes trying to diffuse through them and thus a small 

 solute would have a larger total cross-sectional free area through which to 

 diffuse than a larger one would have. This differential could be considerably 

 enhanced as the point of total exclusion of passage is approached. If this 

 simple mechanistic theory is correct, the selectivity of a membrane could 

 be a function of the size distribution of the passages through which the 

 solutes must diffuse. 



The simple theory briefly outhned is not new. It has been considered 

 before in the extensive studies dealing with the passage of solutes through 

 membranes. No attempt will be made here to discuss this literature. A recent 

 excellent review^^ dealing with capillary permeability covers the subject, 

 including present information about the pore size of certain membranes, 

 the relation of this to the rate of passage of solutes with known effective 

 diameters, etc. The factors controlling restricted diffusion^* have been treated 

 in a mathematical way. 



In spite of these studies, Spandau^^'^^ has presented data to show that 

 dialysis with many smaller solutes can be performed in such a way that the 

 rate seems to correlate better with the inverse of the square root of the mole- 

 cular weight than it does with molecular volume. Our experience with the 



