108 
PROFESSOR R, THRELFALL AND MR. J. H. D. BREARLE7 
In order to be sure that everything was in order when no deflexion was obtained 
on the galvanometer, the wire at P was touched with a piece of paper, which, of 
course, gave a deflexion, and, as additional evidence, the contact at H was brokeu 
and the sulphur plates regarded as a condenser discharged and the spark noted. 
Observation (l).—Galvanometer sensitiveness, I division double deflexion for 
5*5 X 10 _n ampere. Voltage = 313 volts. 
No deflexion could be observed. 
Therefore resistance of sulphur is greater than 
: '^ A n =57 X 10 13 ohms. 
0‘5 x 10 -11 
And the specific resistance is greater than 3'66 X 10 35 C.G.S. 
The sulphur had been exj)Osed during two days to the action of the drying 
material and to the continued application of voltages of from 150 to 300 volts in 
both directions. 
The next point was to measure the specific inductive capacity. We were delayed in 
this by finding, during a series of preliminary experiments, that the smaller subdivisions 
of a standard condenser by Elliott Bros, agreed very badly, and consequently we had 
to make a number of comparisons in order to form an opinion as to the most reliable 
condenser division to use—at that time we had no condenser with standardized 
divisions. 
Finally, we decided that the fault lay, as was to be expected, in the smaller divisions 
of our microfarad ; we were therefore obliged to make use of the larger divisions. 
This necessitated the finding of a way to conveniently bridge the great difference in 
the capacity of the sulphur condenser and that of '2 microfarad, the smallest division 
available. We finally arrived at what we found to be a very satisfactory method. A 
preliminary experiment showed that, using a certain number of Clark cells, we 
obtained, by simple charge and discharge of the condenser through the galvanometer, 
a certain convenient deflexion. The comparison was then made by subdividing the 
voltage of one Clark cell and charging and discharging the selected division of the 
microfarad with it until a nearly equal deflexion was observed. This avoids the 
necessity for determining the logarithmic decrement. The voltage was divided by 
using points of derivation in a circuit of 100,000 ohms, including a Clark cell. 
The capacity of the key and connections was found to be very nearly two hundredths 
(’02) of the combined capacity of the key, condenser, and connections. This correc¬ 
tion is amply accurate enough for the results got by the ballistic method, and is 
rather too large—uncertainty due to large absorption in the standard prevents really 
good comparisons being made. 
There being a difference between the divisions of the microfarad, one of them 
was selected as a provisional standard pending calibration when our standard shall 
arrive. 
