1907-8.] Electromotive Force of Iodine Concentration Cells. 393 
from the alcohol to the water solution, and the main source of the energy 
of the current is probably the different solution pressures of iodine in water 
and alcohol respectively. In both cases, namely, the iodine and the KI, 
they are being transferred from a solution in which they dissolve with 
difficulty to a solution in which they dissolve easily. 
Errors due to Traces of Water. 
The main source of error in taking the E.M.F.s of alcohol cells is probably 
due to the presence of and absorption of water, and it was with a view to 
determining the probable amount of these errors that the experiments with 
mixtures of alcohol and water and the readings of alcohol against water 
were originally undertaken. The absolute alcohol used was that prepared 
by Kahlbaum, and is supposed to contain about '2 per cent, of water. 
Alcohol is, however, very hygroscopic, and though precautions were taken 
to expose the alcohol as little as possible to air while the solutions were 
being made up, it is highly probable that some water was absorbed during 
transference from one vessel to another, and also that the water absorbed 
was not necessarily of equal amount. 
If, in the first place, we assume the water absorbed at both ends of the 
cell to be of equal amount, then the E.M.F.s obtained for water-alcohol 
admixtures give a very fair idea of the maximum error probable from this 
source. If these values are plotted as a curve, and the curve continued, it 
is evident that the presence of 1 per cent, of water on both sides would 
lower the E.M.F. about '0002 volts, and therefore this source of error may be 
neglected. If, however, the amount of water absorbed at the two ends was 
different, the errors would be more serious. The addition of 10 per cent, of 
water on the strong iodine side, the strength of the solution remaining the 
same, caused a lowering of E.M.F. of '02 volts, agreeing closely with the 
results obtained and already quoted for alcohol against water, thus making 
a possible error for an excess of 1 per cent, of water on one side of about '002 
volts. Such a source of error, however, would not be constant, but would 
appear sometimes on the one side and sometimes on the other, and would 
therefore not affect the mean of several readings. This probably accounts 
for the observed differences in readings for independently prepared solutions, 
which are greater than those for water cells. For these reasons the proba- 
bility is that the readings given for alcohol against water are probably from 
'001 to '002 volts too low. 
In conclusion, I wish to thank Dr Denison and Mr King for the 
assistance they have given me in this research. 
( Issued separately July 22, 1908.) 
