ON THE ELECTRICAL PROPERTIES OF PURE SULPHUR, 7 5 
lent resistance for a definite potential, when neither the capacity of the quadrants 
nor the leakage of the electrometer can be neglected, then the formula of calculation 
becomes arithmetically very objectionable. This is by no means a slight disadvantage 
if many results have to be reduced. The capacity of the resistance cell also has to be 
known with fair accuracy, which, at all events, involves more work. The gravest 
disadvantage, in our experience, lies in the variability of the rate of leakage of the 
quadrants. If the experiment last for an hour, as is often necessary, then undis- 
coverable changes may have been taking place in the insulation resistances of the 
electrometer. If a large condenser be employed in order to reduce the magnitude of 
the uncertainties arising from want of accurate knowledge of the other capacities 
involved, we have to face the difficulty of making a proper allowance for its probably 
very variable insulation resistance. With our high resistances this difficulty proved 
insuperable. Taking everything into account, we do not consider this method 
favourable for the determination of absolute resistances, particularly when the word 
resistance has no meaning except with respect to an instantaneous voltage. 
In the second class of experiments by the leakage method, the charge which leaks 
is taken up by an auxiliary condenser. This method has been used by Bouty 
(‘ C. B.,’ vol. 110, p. 1362) with fair success, on standard condensers, for the purpose 
of investigating the phenomena of residual charge. For this purpose, and when the 
capacities of the different parts of the apparatus are known, the method is suitable ; 
but, as a method for measuring absolute resistances, it merely adds to the uncertain¬ 
ties mentioned above—those due to the uncertainty of capacity and insulation of the 
receiving condenser. In our case these methods were out of the question, as we had 
to deal with resistances greater than the insulation resistance of our mica condensers. 
In the application of this method made by Foussereau (‘ C. B.,’ vol. 95, p. 216) matters 
were complicated by the use of a Lippmann electrometer, the poles of which can surely 
only insulate fairly well for small voltages, and which presumably has a not neces¬ 
sarily vanishing equivalent capacity. However, we have not had much experience 
with the Lippmann instrument, so our criticism must not be taken for more than it 
is worth. 
A quite different method has been employed by H. Roller (‘ Sitzungsber. Wien. 
Akad.,’ vol. 98, p. 201). This method consists in placing the substance undergoing 
investigation, in series with a high resistance constructed either of a tube of sulphate 
of zinc solution, with zinc electrodes, or of a quantity of saturated solution of iodine 
in carbon bisulphide with platinum electrodes. The latter resistances may be con¬ 
structed (with proper precautions) as high as 10 u Siemens’ units. A divided circuit is 
used to give any desired voltage to the circuit containing the artificial resistance in 
series with the resistance to be measured. The fall of potential between one terminal 
of the cell or derivation circuit and the resistance cell, and between the same terminal 
and the artificial resistance, allow (proper connections being made) of a comparison of 
the resistance in question with that of the artificial resistance. Roller v r as looking 
l 2 
