440 



Earl of Berkeley and Mr. E. G. J. Hartley. [Apr. 21, 



the effect of the leak was gradually to saturate with solution the 

 surface of the exposed portion of the tube, and thus abstract water 

 from the inside through the membrane. 



We call this leak the guard-ring leak, and it was hoped that the 

 following method gave us a means of estimating it. 



Guard-Ring Leak Correction.— The holes GG (see fig. 2), were bored, 

 and a stream of the solution whose osmotic pressure was to be 

 determined, was directed through them ; and to insure that the 

 surface was thoroughly saturated, the whole apparatus was placed in 

 a bath of the same solution. The inside of the tube together with the 

 capillary having been filled with water, and the temperature of the 

 bath and apparatus having become constant, the rate of fall of the 

 level in the capillary was noted. It should be mentioned that the space 

 enclosed by B was also filled with water, and C was replaced by an 

 open glass tube. The level of the water in this open tube was kept at 

 the same height as that in the capillary, so that any small change in 

 the temperature of B did not alter the rate of fall of the level in the 

 capillary. 



The guard-ring leak correction was found to be the larger the more 

 concentrated the solution, and it also varied slightly with the height 

 of the water in the capillary. 



This correction was of a magnitude such that it raised the " turning 

 point," with reference to the pressure, by an amount equal to from 

 5 to 10 per cent, of the osmotic pressure. 



The Solution-Leak Correction. — We have hitherto been unable to make 

 semi-permeable membranes completely impervious to sugar ; on testing 

 the water from the inside of the tube a trace was always indicated. 

 It was, therefore, considered advisable to determine the amount of 

 sugar which had come through in each experiment. 



At the end of the experiment the inside of the tube was washed out 

 without taking down the apparatus, and its content of sugar analysed 

 by means of Fehling's solution. On the assumption that the resulting 

 quantity of sugar indicated that a corresponding amount of solution 

 had come through the membrane, and on the further assumption that 

 the rate at which this solution came through was proportional both to 

 the time during which it had been subjected to pressure and to the 

 amount of that pressure, the displacement of the "turning point" was 

 calculated. 



It was found that, with a good membrane, the solution leak correc- 

 tion was of a magnitude such that it lowered the " turning point," with 

 reference to the pressure, by an amount equal to from 2 to 3 per cent, 

 of the osmotic pressure. 



In this connection it may be pointed out that a check on the 

 algebraical sum of these two corrections may be obtained from the 

 experiments themselves. For by beginning an experiment at a 



