ON STANDARDS FOR USE IN ELECTRICAL MEASUREMENTS. £7 
Condenser 
compared with —_ 1 Il. 
Standard 
05 A+a—Al 2°3 —73 
—a -3 —2°6 
i) A+a—Al 2°2 -9 
ae —3 —3 
‘1 , A+a—Al 2'2 -—7 
—a, —2°2 —3°5 
5 A+a—Al 33 —4 
—a —3:3 —25 
pt A+a—A! 4 —3:2 
—a -5 —57 
If we take the comparisons with condenser I. first, it appears that 
throughout A—A! is small. For the ‘05 and ‘1 microfarad it may be 
about —°‘5 division, while a is about 3 divisions: for the ‘5 microfarad, 
a is rather larger, being about 3°3, and A—A! is zero, while for the 
1 microfarad a the absorption effect is distinctly larger, being 5 divisions, 
and A—A! is about —1. All this is, of course, quite consistent with the 
fact that condenser I. and the mica condensers insulate well while there 
is absorption by the mica. 
When, however, we come to the condenser II. the results are quite 
different. While the absorption effects are comparable, as of course they 
ought to be, with those obtained in the comparison with I., the leakage 
_ effects are very large. 
The values of A\—A! in order are as follows: —9, —12, —10°5, —6°5, 
—8. Now, we know that the mica condenser shows very little leak effect ; 
the above leaks are therefore almost entirely in the air condenser II. If 
we suppose the total leak to be proportional to the time, then for the 5 
second charges used in the experiments the corresponding values of y in 
the corrections to be introduced for leakage will be one-sixth of the above, 
and thus we get the following results :— 
| Correction for the Correction for the 
Condensers | Value of y | Leak to Capacity | Condensers | Value of y | Leak to Capacity 
in Microfarads in Microfarads 
16 | ‘00007 5 iL “0003 
2 | ‘00016 | 16 1 ‘0007 
It is clear that the corrections are in all cases small, being not much 
over 1 in 1,000, but they serve to illustrate the method. The above cor- 
With a view to testing the method in a case in which a leak only 
existed without absorption, a number of comparisons of I. and II. 
were made. 
In these experiments the resistance with I. was 296,240. The 
resistances with II., and the deflections due to the leak obtained by 
breaking the battery and then making the galvanometer, are given below, 
together with the ratio of the two capacities corrected for the leak. 
