392 
Proceedings of the Royal Society of Edinburgh. [Sess. 
KI, and two or three drops introduced of water containing 3 molecules, 2’75 
molecules, 2*25 molecules, and 2 molecules of potassium iodide respectively. 
The tubes were then sealed and hung in the thermostat, with repeated 
shaking. Crystals quickly separated from the solution containing a few 
drops of 3 molecules and 2’75 molcules, but no crystals could be detected, 
after several days, in the 2 molecule or 2'25 molecule solution. Slight 
formation of crystals was detected after some time in a 2*5 molecule 
solution in another experiment. 
These results, then, show that there is an equilibrium between alcohol 
saturated with KI and water containing about 2 molecules of KI, no 
further precipitation taking place. It seemed important to confirm this 
in another way. 
A series of solutions was therefore prepared of KI in water, and to 
equal quantities of water increasing volumes of alcohol saturated with 
KI were added. The water solutions were chosen so that the amount of 
crystals separated should be small. 
The various mixtures were stirred for several days in a thermostat at 
25° C., and then known portions of the total volume drawn off and 
evaporated. On adding alcohol to water there is of course a contraction ; 
but by adding a known volume of alcohol to a known volume of water 
and evaporating a known portion of the total volume, accurate results are 
obviously obtainable. 
The method adopted was therefore to take a given volume of water, 
then dissolve in it a known weight, of potassium iodide, then make up 
this solution to 100" with the KI saturated alcohol, measuring the volume 
of alcohol used for this purpose independently. 
Then after treatment in the thermostat 10" was withdrawn, evaporated, 
and weighed, and from the total weight, the weight of salt due to the 
alcohol content (which is known from the volumes of water and of 
alcohol originally taken) was subtracted. 
The following curve contains the results, in which the additions of 
saturated KI alcohol to the potassium iodide solutions in water are plotted 
against the KI left in solution in the water content of the mixture. 
It will be seen that in approximating to 2 molecules of water, precipita- 
tion has practically ceased. 
We find, therefore, as the result of these experiments, that if a solution 
of alcohol saturated with KI is brought in contact with a solution of 
water containing about 2 molecules, the diffusion of the alcohol into the 
water will produce no further precipitation of KI, and that such an 
arrangement, if silver silver iodide elements are introduced, is in 
