ELECTROMAGNETIC UNIT OF ELECTRICITY TO THE ELECTROSTATIC UNIT 59 L 
Correction for a small difference of j^otenticd, hetiveen tlte guard ring and the 
middle cylinder. —The arrangement we used necessitates the existence of a small 
diiference of potential between the cylinder and the guard ring, a slight modification 
of the preceding investigation will enable us to find the correction for this. If we 
put 
+ n^ = 
V t — a 
- lOD- - 
TT ^ t -{■ a 
+ 
gv 
TT 
log {t-\- a), 
then the potential over FGH = 0 , that over EDO = SV, and over AB = V. 
The breadth of the strip which must be added to compensate for the electricity on 
the portion QGH due to the difference of potential SV between CD and GH is 
A SV , (i'q + n) 
- — log - 
F * (G + 
Now, if QG is large, 
G(.^ = A log ^ , 
where e is the base of the Napierian logarithms. Hence, since t^ — a = sin a, the 
breadth of the additional strip is 
h SV [QU , 
; V |x - '“g 
4 sill a] 
but A = c ‘Iff and sin a — 
two guard rings is 
cjh approximately, hence the breadtli of the strip for the 
loc 
4c 
TT 
(0 
In our experiments hjc = 6 ' 6 , QG = 1 , so that the correction amounts to a strip 
whose breadth is about 
^ 5 y • 
The value of SV/V depended on the speed, the usual value was about -xy- 3 . In this 
case the breadth of the strip would be about 3 ^- of a centimetre, and, since the length 
of the cylinder was about 60 cm., the correction amounts to about 1 part in 1800. 
To test the accuracy of these corrections, determinations of the capacity of the 
condenser were made when the top guard ring was separated from the middle 
cylinder ( 1 ) by pieces of ebonite ‘504 cm. thick, ( 2 ) by pieces '067 cm. thick. The 
capacity of the cylinders with the thick ebonite was greater than that wuth the thin 
by about 11 parts in 2760. According to the results we have obtained for these 
