OSMOTIC PRESSURES OF SOME CONCENTRATED AQUEOUS SOLUTIONS. 493 
We give this example in detail, as we thought at first that the error introduced 
from this cause might be considerable ; but even in this case it is evident that it is 
quite small, and in our later experiments, where care has been taken not to work far 
from the turning point and the increment of pressure is smaller, the error is quite 
negligible. 
(3) Solution Leah .—The turning point having been satisfactorily determined, the 
osmotic apparatus is taken apart, and the porcelain tube, with the brass tubes LL 
(see fig. 1) untouched, and therefore still containing the water used in the experiment, 
is placed in distilled water. After an interval of two or three days, about 
100 cub. centims. of water are run through the tube and the content of sugar 
analysed. If enough sugar is found, another 100 cub. centims. are run through 
at the end of two more days, and this process is repeated until there is practically 
no more sugar in the tube. It is then remade electrolytically at 0° C., and kept 
in water at that temperature. 
We would like to point out that even with the best membranes there is always, or 
nearly always, an indication of a very small quantity of sugar as having come through 
the membrane,* and we would therefore emphasize the necessity of always testing the 
water inside the tube. 
The analyses were made by means of Fehling’s solution; the equivalent value, in 
terms of sugar, of the resulting copper oxide, together with a standard method of 
working, being obtained by the analysis of dilute solutions of sugar of known 
content. 
In our preliminary paper we stated that, on certain assumptions, a satisfactory 
correction for the solution leak could be applied to the apparent turning point so as 
to give concordant results; at that time the membranes were new, and a fairly large 
amount of sugar came through during the whole time an experiment lasted. We 
now find that with the very much improved membranes—membranes it must be 
remembered which undoubtedly have given way some scores of times, and been 
remade in probably a different layer of the porcelain—there is only a rough 
connection between the amount of solution leak and the lowering of the turning 
point. It is evident that, apart from the difficulties of analysis, with such small 
quantities of sugar no exact connection could be expected ; the extremely minute holes 
in the membrane, indicated by the smallness of the leak, may easily close up or open 
out during the course of the experiment. All attempts, therefore, to correct for the 
“solution leak” were given up, and the value of an experiment was based on the 
amount of sugar found inside the tube ; practically this resolved itself into rejecting, 
for the final computation of the equilibrium pressure, all the experiments done with 
tubes other than those we call X and N, and only accepting these when it had been 
proved that no more sugar than 0'0003 gramme had come through in the experiment. 
The following table, copied from the laboratory notebook, gives examples of the 
* Cf. Table VI., Column VII. 
