536 Proceedings of Royal Society of Edinburgh. [sess. 
In each of the above experiments, electrolysis was continued all 
night, so that it may he assumed that the nickel had been com- 
pletely removed again when the readings were taken next morning. 
In other experiments, greater quantities of nickel, up to 20-30 m.g., 
were added, without producing any further appreciable change. 
These results seem to indicate that when the potential begins to 
rise in an ordinary determination, only two or three milligrams of 
nickel remain to he deposited. The time necessary to effect this is 
out of all proportion, as may be seen by referring to experiments 
I.-IY. The current employed to eliminate these last traces is 
very greatly in excess of that theoretically necessary, and it is 
therefore very questionable whether any advantage is secured by 
increasing the current towards the end of an electrolysis, as is 
sometimes recommended ; the increased escape of gas would cer- 
tainly have the effect of stirring up the liquid more, but probably 
a much better result would he obtained by diminishing the current 
and agitating the liquid. 
When small quantities of nickel have to he determined, the pro- 
portion of current wasted becomes greater, and there is therefore 
greater risk of the deposit being affected by secondary actions ; and 
it is precisely in such cases where a slight error, due to incomplete 
deposition on the one hand, or to impurity of deposit on the other, 
has to be specially guarded against. In the Report already referred 
to, it was therefore recommended to employ small vessels and small 
volumes of solution — other quantities being in proportion — when 
determining small quantities of nickel, as better results are then 
obtained, though the benefit is not nearly so marked in the case of 
nickel as in that of cobalt. The standard current density then 
means a much diminished current strength, but, owing to the much 
smaller volume of liquid from which the nickel has to be deposited, 
deposition was shown to be completed at least as soon as in dupli- 
cate determinations on the larger scale. By making use of the 
voltmeter readings as indicator, it has now been shown by a 
number of experiments that the advantage is distinctly in favour 
of the smaller vessel. For these experiments wide and somewhat 
shallow platinum crucibles were employed, with an anode of coiled 
platinum wire. The volume of solution was about 18 c.c., giving 
about 25 sq. cm. of cathode surface, which was one-fourth of that 
