CHEMISTRY: WASHBURN AND MILLARD 
143 
taining enough raffinose to produce the desired rotation placed between 
a silver anode and a cathode consisting of silver covered with solid silver 
chloride. The apparatus was constructed so as to allow the separation 
of the portions around the electrodes from one another and from three 
middle portions. After the electrolysis the separate portions were care- 
fully analyzed; the salt being determined by precipitation with silver 
nitrate, the rafhnose polarimetrically, and the water by difference. The 
changes in content of the portions were computed on the assumption 
that the raffinose remained stationary during the passage of the cur- 
rent. Inasmuch as solutions containing 10% of raffinose have been pre- 
pared which had a specific conductance of 2 X 10~^ mhos (without 'cor- 
recting' for the conductance of the water) , it is clear that the raffinose 
cannot conduct electricity appreciably either as an electrolyte or as a 
colloid. 
The apparatus and experimental procedure were essentially the same 
as those employed in the previous work; but the concentration of the 
reference substance (raffinose) was reduced from 0.1 to 0.07 formula- 
weight per 1000 grams of water. This reduction in the concentration 
of the reference substance, while theoretically desirable, would tend to 
decrease considerably the accuracy in the measurement of the water 
transference, especially where, as in the case of cesium chloride, this 
transference is very small. This loss of accuracy was, however, com- 
pensated by employing a more sensitive polarimeter with a layer of 
solution one meter in length and substituting the E line (from a quartz 
mercury-vapor lamp) in place of the D line (from a sodium flame) pre- 
viously employed. Without this great refinement in the polarimetric 
measurements the water transference in the case of cesium chloride 
could hardly have been detected, much less measured. The actual ro- 
tation measured in the polarimeter was about 50° and was reproducible 
to about 0.004°. 
A practice run using sodium chloride as the electrolyte confirmed the 
previous work with this salt; thus this run gave 0.82 mols of water and 
0.623 equivalents of chloride-ion transferred per faraday, as against 0.81 
and 0.617 obtained in the previous work. 
Two runs were made with 1.1 molal cesium chloride solutions. One 
of the runs was completely successful; but only the results obtained from 
the cathode portion of the second run were reliable. The number of 
mols of water transferred from anode to cathode per faraday were in 
the first run 0.33 and 0.39, and in the second run 0.28, giving a mean 
result of 0.33 mols per faraday. The values obtained for the number 
