ON THE ELECTRICAL PROPERTIES OP PURE SULPHUR. 
145 
impurity from the plaster of Paris is localized to the edge of the plate, i.e., is about 
half an inch outside the guard-ring, where it can do no harm during the weighing. 
Warped Plates .—One of our plates warped in a very regular manner, so as to 
become concave on one side and convex on the other ; the curvature looked spherical 
and was assumed to be so. 
As we did not know what error this might lead to, and still did not wish to lose 
the plate—on which a good deal of time had been spent—we measured the curvature 
by a spherometer, and Mr. Pollock kindly investigated for us the proper formula. 
As, however, it involved a difficult integration, it was considered advisable to 
calculate the effect on the value of K of some extreme assumption as to the thickness 
of the plate. For this purpose it was supposed that the plate increased in thickness 
by the same amount all over, as at the boundary of the suspended plate. The radius 
of curvature was found to be for the concave side of the plate, 133'234 centims. ; and 
for the convex side, 134T77 centims. ; and the thickness of the plate was "94 centirn, 
The diameter of the plate was taken at 5 inches = 12 70 centims. In a particular 
experiment it was found that the value of K, when the plate was considered as 
having the maximum effective thickness of '94407 centirn. instead of '943 centirn., 
would be 3'6836, instead of 3"6878 when the measured thickness was taken. Since 
this difference is outside the experimental limits of weighing or measuring, and is 
itself probably so far in excess of the actual correction, even if the lines of force are 
appreciably “ refracted,” we decided to ignore the effect of the curvature. 
Method of Experimenting .—The experiment consists in observing the attraction in 
grams weight on the suspended disc, both when the sulphur plate is in and out. 
To begin with, everything must be carefully dusted, to get rid of shreds of all kinds, 
aud the hanging plate counterpoised—it weighs, as has been stated, about 154‘3 grams. 
The current is then turned on and observed to be steady by the voltmeter—it never 
really gets steady, the voltmeter oscillating through about one division, corresponding 
to, say, a ‘‘permanent” variation of half a division or 100 volts in 5000, or say, 
2 per cent.; this would cause the force to vary by 4 per cent.—this is an extreme limit 
to the fluctuations. Weights are added until the pointer of the balance is observed 
just to incline to leave the zero—of course, as soon as it does leave it the force 
diminishes and the balance tilts over. The point having been ascertained as nearly 
as possible, the rider is run down to increase the weight by •005 gram, and if this 
produces a decided tilt, the observation is taken as exact. This means that the 
force can not be measured to less than the weight of 5 milligrams owing to 
fluctuations of potential difference. The forces to be observed vary from about 
'5 gram weight without the sulphur to 2 grams with it, and these forces appear 
in the final result as a ratio which enters to the power of one half. The uncertainties 
of weighing, therefore, amount to about '005 gram in one gram, or, say, ‘5 per 
cent., and affect the value of K to about '25 per cent. 
MDCCCXCVI.-A. 
u 
