7 
: F ON STANDARDS FOR USE IN ELECTRICAL MEASUREMENTS. 105 
placed by calcium chloride, and after this the leak in I. rose to about 
1 per cent. per minute or ten times its former value, while that in II. 
was from 3 to 4 per cent. of the charge. With the calcium chloride inside 
the leak was never reduced to less than about ‘8 per cent. per minute. 
In August, the condensers having been closed since June with calcium 
chloride, there was a leak in I. of about 3 per cent. per minute, while in 
the same time IT. lost about 8 per cent. of its charge. 
On August 14, immediately after this test, the calcium chloride was 
replaced by sulphuric acid, and the leak was quickly reduced to about 
1 per cent. per minute for I. For II. no improvement showed itself at 
once. The next day the leak in I. was about ‘4 per cent. per minute; 
that in II. had not been greatly reduced. On August 16 the ebonite 
was therefore well cleaned, and air was blown through the tubes of II. 
and the whole closed for about two hours; the leak had then fallen to 
about 2 per cent. per minute. By August 18 the leaks were still more 
reduced, that in I. being ‘2 per cent. per minute, while that in IT. was 
*6 per cent, per minute. 
By the afternoon of this day, the upper parts of the condensers having 
_ been open to the air of the laboratory for some six hours during other 
tests, the leaks had appreciably increased, but they had fallen again the 
next day when the condensers were left closed during the night. 
Thus, during the observations in August, with the exception of those 
on August 14, the condenser I. was losing its charge at the rate of about 
soo part per one minute, while the leakage in II. was some five or six 
times as great, being about +1, part of the charge per one minute. 
As will be seen later, several mica condensers were compared with I. 
and IT.; the leaks in them were all small, and did not exceed 51, per 
minute. 
We come now to the experiments for determining the capacities of 
the two condensers. Of these, three independent series were made, viz. 
in December 1889, May and June 1890, and August 1890. 
Tke method already referred to was used. Fig. 3 gives a diagram 
TG. Oo. Fig. 4. 
eee method ; in fig. 4 the connections actually employed are shown. 
With the notation employed ‘ Phil. Mag.,’ August 1884, we have, if ¢ 
be the capacity of the condenser, » the number of times it is charged per 
one second, 
