256 Wisconsin Academy of Sciences, Arts and Letters. 
and the same outer vessel filled with pyridine, they neverthe¬ 
less in general failed to indicate identical pressures. All this 
led me to the conclusion that the non-concordance of the pres¬ 
sures measured could not be due to leaks, defects or various 
degrees of u semi-permeability ’ ’ of the membranes. 1 
I confess that I was about at my wits’ end to understand 
why such duplicate experiments made as nearly as possible the 
same in every way should yet yield results that were not more 
concordant. In all, some seventy independent trials had been 
made, in which great care had been used to get the experi¬ 
ments alike, but to no avail. Yet the membranes were not de¬ 
fective. One day after measuring carefully the height of the 
mercury column in one of the experiments which had been 
running for two weeks and in which the pressure had changed 
inappreciably for several days, I happened to brush against the 
apparatus in such a way as to thoroughly jar it without, how¬ 
ever, upsetting it or otherwise modifying it; in any way. Some 
minutes later when I chanced to look at the apparatus, I 
noticed that the level of the mercury, which had been prac¬ 
tically constant, had risen over two centimeters higher. The 
apparatus was then shaken repeatedly from time to time, seiz¬ 
ing it with an insulating cloth so as not to alter the temper¬ 
ature, with the result that each time the pressure rose some- 
i Measurements of the electrical resistance of the membranes, were 
also made which further establish this fact. Placing on each side of 
the membrane of an osmotic cell, such as was emjployed in the pres¬ 
sure measurements, a circular platinum electrode about 2.1> cm in 
diameter, so that the planes of the electrodes were parallel to that 
of the membrane, and that the metallic surfaces were close to the 
membrane without actually touching it, the cell was filled with an 
aqueous saturated solution of NaCl and the same solution was used as 
the outer liquid. The electrodes were thus immersed in the saturated 
NaCl solution and the membrane separated them from each other. 
The electrical resistance between the plates was measured by means 
of the Kohlrausch method; but it was found to be so great that it * 
was not possible to estimate it accurately, being upwards of 70,000 
ohms. Various rubber membranes tested in this manner gave results 
of the samje order of magnitude. When the same experiment was 
performed using a parchment membrane the edges of which protruded 
so that they did not dip into the brine, the resistance measured was 
less than an ohm. Using the same arrangement with a rubber mem¬ 
brane and a normal solution of silver nitrate in pyridine as the liquid 
on each side, the resistance was over 90,000 ohms, and remained the 
same for four days, when the experiment was discontinued. A small 
hole pricked through the membrane with a needle in the latter case 
caused the resistance to drop to 450 ohms. 
