12G 
mi. B. D. STKELE ON THE MEASUBEMENT OF IONIC VELOCITIES 
MgCL, with MgO, suspended in the solution. 
KCl, containing- dissolved K^COo. 
AgNOg, with silver anode. 
Ah(S 04 ,) 3 , with AL(OH)q suspended in solution. 
Anion Indicators. 
NaAc, with HAc to prevent formation of OH ioiis. 
KF, containing a small quantity of HAc. 
K,CrO„ with KXhqO^. 
The current wms obtained from a battery of thirty storage cells, and the total 
voltage used in different experiments varied from 2U to 70 volts. 
Tlie tenqjerature of the experiments -was that of the room ; it wms constant to O'^'o 
throughout any experiment, but varied in difierent experiments between 14° and 10°. 
The Accuracy of the ILeasurenients. 
The a]j})aratus having been prepared as already descril^ed, shortly after the current 
is started, the boundaries having advanced into the tube ; their rate of motion is 
measured hj means of a cathetometer, a small gas jet when placed behind the tube 
Ijeing seen to Ije cut Ijy a dark line at the point -where the t-wo solutions are in 
contact. Each tube lias etched on it a horizontal line, wdiich serves as a fixed jDoint 
from wdiich measurements are made. At frequent intervals the time, the current, 
and the distance moved over are measured ; the ratio of the distances moved over by 
V , . A' . 
the anion and the cation boundaries gives directly from wdiich once 
obtained. 
The cathetometer employed -was capaljle of giving readings correct to the tenth 
of a millimetre, ljut from various causes the accuracy with which the position of the 
margin can be read is about one-third of this. The difierence in the values of p 
obtained in the various readings of the same experiment amounts in some cases to 
2 per cent., but tlie error here arises in nearly every case from the fact that in the 
earlier measurements, wdiere the distance moved over by the boundary is often less than 
0‘5 centim., a , small error in reading the cathetometer has a very much greater 
influence than later wdien tlie distance is more, hence it is found ahvays that the 
last few readings agree mucli Ijetter among themselves tlian do the earlier ones. 
All the readings from an experiment are therefore averaged in the following 
manner :— 
Each value for p is nmltiiilied hy the distance moved over liy one of the 
boundaries since the beginning, tlie numbers thus obtained are added together and 
divided by the sum of the niulti})licands. When averaged in this manner, it is 
