OSMOTIC PRESSURE 337 



a task fraught with immense difficulties, on account of the difficulty 

 of preparing ideally pure proteins. The original investigations 

 of Graham (35) appeared to indicate that colloids in general exert 

 a high osmotic pressure. Subsequent investigators, however, 

 attributed these results to an admixture of crystalloids. Starling 

 endeavored to measure directly the osmotic pressure of the pro- 

 teins in blood serum by using for his osmometer a membrane 

 permeable to salts but impermeable to proteins (117) and this 

 method has since then been employed in all accurate work upon 

 the subject, since, as Reid has pointed out (92) it is the only method 

 of procedure which is applicable to the problem. We have no 

 assurance that any given protein preparation is totally free from 

 (not necessarily conductive) impurities which may influence the 

 direct measurement of osmotic pressure; it is, therefore, essential 

 to employ a membrane which is permeable to such impurities and 

 thus, if time be allowed for the system to come to equilibrium, 

 differentiates between protein and non-protein constituents of the 

 solution investigated. For this purpose Reid employs a mem- 

 brane of vegetable parchment, which, as he has shown, is per- 

 meable even to nucleic acid although it is impermeable to the 

 proteins employed by him in his investigations. By extremely 

 prolonged purification Reid has succeeded in obtaining prepara- 

 tions of egg-albumin which exhibit no measurable osmotic pressure 

 when examined by this method. In subsequent investigations 

 (93), however, he obtained osmotic pressures, due to dissolved 

 haemoglobin, of perfectly constant value and such as to indicate 

 a molecular weight of about 48,000. Barcroft and Hill (4) have, 

 however, demonstrated by thermodynamical methods that in 

 solutions containing haemoglobin prepared by less prolonged 

 dialysis the molecular weight of this substance is about 16,669. 

 Roaf (96) employing the differential osmotic method just de- 

 scribed, finds that the molecular weight of haemoglobin, dissolved 

 in distilled water, is about 32,000, while in sodium carbonate 

 solutions it is 16,000. These results appear to confirm the view 

 (Cf. also Barcroft and Hill loc. cit.) that non-ionic protein is 

 polymerized and so exerts a considerably smaller osmotic pressure 

 than ionic protein. 



Benj. Moore and Roaf and collaborators (70) (71) (72) (95) (97) 

 and R. S. Lillie (63) have made the extremely interesting discovery 

 that the osmotic pressure which is exerted by proteins (deter- 



