568 SCIENCE PROGRESS 



pressed solution. The extent of the guard-ring leak was 

 determined by filling the extension-tubes VV with solution 

 whilst the rest of the apparatus was charged with water both 

 inside and outside the membrane. In the final series of experi- 

 ments the guard-ring leak was reduced, from a rate equivalent 

 to that produced by a pressure of 2 or 3 atmospheres on 

 the solution, to a rate equivalent to only about 0*15 of an 

 atmosphere and therefore almost negligible. 



(b) Determination of the turning-point. The chief operation 

 was to determine the pressure at which water just ceased to be 

 drawn from the porcelain-tube into the solution and com- 

 menced to flow in the opposite direction. After measuring the 

 guard-ring leak at o° the space surrounding the porcelain-tube 

 was emptied, rinsed with solution and filled as quickly as 

 possible. The apparatus was then immersed again in ice and 

 pressure gradually applied by increments of about 10 atmospheres 

 until within 10 per cent, of the equilibrium pressure, when 

 smaller increments were applied at longer intervals until the 

 rate of flow was almost exactly equal to the guard-ring leak. 

 During the process of filling and before the pressure was 

 applied a small quantity of water was drawn through the 

 membrane into the solution, giving rise to a film of slightly 

 diluted solution on the surface of the tube ; by applying 

 pressure gradually in the manner described the excess of water 

 was driven out again and the solution restored to its original 

 concentration without damaging the membrane. Measurements 

 of the rate of flow with pressures a little above and a little 

 below the turning-point were made at intervals of an hour or 

 more until it was clear that a definite and steady value had 

 been reached. 



(c) Solution-leak. At the close of the experiment the 

 apparatus was taken down but in such a way that the porce- 

 lain tube and its contents remained intact. After two days the 

 water in the tube was washed out and the sugar-content deter- 

 mined, this process being repeated until no more sugar could be 

 extracted from the interior of the tube. No attempt was made 

 in the later experiments to apply a correction for the greatly 

 reduced leakage of solution, which seemed to have no regular 

 influence on the " turning-point " : instead, all experiments were 

 rejected except those in which the leakage of sugar was proved 

 to be less than 0*0003 gramme, a stringent test which eliminated 



