ACCURATE MEASUREMENT OF IONIC VELOCITIES, ETC. 453 



is lighter and the other heavier than the solution to be measured, one-half of the 

 form of fig. 1 and one-half like fig. 2 are used in constructing the apparatus. 



In fig. 2 a hole with a collar is shown in the bottom of the thermostat H. The 

 apparatus can be passed through this hole and held in position by a rubber ring, L. 

 This was found to be the best way of supporting this half of the apparatus. Both 

 forms of the apparatus consist essentially of three tubes, an electrode vessel, E or E', 

 a vessel, B or B', containing the indicator solution, and the tubes A or A' in which 

 the boundaries move. The tubes A and A' are made from carefully selected glass 

 quill-tubing of about 4 to 5 millims. diameter, and are accurately calibrated. They 

 are sealed into the wider tubes B or B' in such a manner that they form a shoulder 

 projecting inwards. The lower extremity of the capillary tube K is expanded into 

 a cone, C, around which a piece of parchment paper can be tightly stretched. This 

 parchment paper fits over the projecting shoulder, so that when pressed upon it there 

 is no mixing of the liquid poured into the apparatus through the tube K with the 

 liquid previously placed in A. The tubes K and K' slide easily through holes in 

 rubber corks, so that after an electrolytic margin has appeared in the tubes A or A' 

 the membrane can be withdrawn from the shoulder, and a free passage is allowed for 

 the current. The cones C are provided with holes to allow the passage of the liquid 

 into B and also to permit the free flow of the electric current when the membrane is 

 resting on the projecting shoulder. 



The choice of electrodes and electrode solutions needs careful consideration if 

 accurate results are to be obtained. It is obvious that no gas evolution at the 

 electrodes must occur if a good margin is to be maintained. The indicator solutions 

 most frequently employed were lithium chloride and sodium acetate, and in all our 

 experiments the cathode consisted of lead wire dipping into and completely covered 

 by a paste of lead peroxide, which completely prevented any evolution of hydrogen. 

 The anode consisted of a copper wire dipping into cadmium amalgam, thus con- 

 stituting a completely unpolarisable electrode. A rough calculation showed that the 

 total volume change at the electrodes corresponding to the changes 



PbO 2 +H 2 = Pb + 2H 2 O and Cd+2Cl = 



was negligible. To prevent the hydroxyl ions formed at the cathode from finding 

 their way back into the middle electrolyte, the cathode was surrounded with a 

 solution containing acetic acid. 



The cadmium chloride formed at the anode is to some extent hydrolysed, and the 

 hydrogen ions thus formed travel towards the cathode across the cation margin and 

 decrease the resistance of the solution there, and so also the fall of potential and the 

 velocity of the margin. To prevent this, a little sodium or lithium hydroxide was 

 added to the solution surrounding the cadmium electrode. If these precautions were 

 neglected, slightly false values for the transport numbers were obtained. 



The whole apparatus was held in position by means of a hinged wooden arm, and 



