74 
PROFESSOR R. THRELFALL ARD MR. J. H. D. BREARLEY 
investigation, if only from the great discrepancies which appear above. Other values 
which have been obtained for the dielectric constant of sulphur will be found in 
Wiedemann’s ‘Lehrbuch,’ vol. 2, p. 25, and onwards. The only determination of the 
constant for sulphur in a special state is that of Boltzmann, but the purity of the 
natural crystal he used was not established. 
Methods of Determining High Specific Resistances. 
(1.) The most direct method is that by means of a galvanometer. This method was 
used by one of us in 1888 (‘ Phil. Mag.,’ 1889, vol. 28). The advantages of the 
method (which will be fully discussed presently) are that it involves a direct deter¬ 
mination of voltage and current, and. can be made practically independent of any 
leakage other than the one under investigation. It is also fairly rapid and very 
flexible. Its drawback is that large voltages are required, and, as generally prac¬ 
tised, the method is wanting in sensibility on the one hand, and is more or less 
dependent on some assumed galvanometric formula on the other. These difficulties we 
have, however, overcome. The galvanometric method was used with success by the 
Brothers Gray, ‘ Proc. Boy. Soc./ 1884, vol. 36, p. 287; Monckman, loc. cit. 
Brooks, ‘ Journ. Soc. Tel. Eng.,’ vol. 9, p. 5 (1881); Ayrton and Perry", £ Proc. Boy. 
Soc.,’ vol. 27, p. 219, 1878, and many others. 
(2.) Leakage Methods .—These are of two classes, and are so well dealt with in 
Professor Gray’s book on ‘ Electric Measurements ’ that very little remains to be said 
here, and that little is only by way of criticism. In the first place, the advantage of 
the method is that it is very sensitive and convenient, and can be carried out without 
special training by anyone possessing a good electrometer. Also, it works well over a 
large range of voltages. The disadvantages for very high resistances are—(1) the 
insulation of the instrument is only made sufficient with great difficulty, (2) the time 
required for an observation is very long, and this is apt to mask very important effects. 
Again, if the calculation of the results be made from the rate of fall of potential with 
time (as was done by J. J. Thomson and Newall, ‘ Proc. Boy. Soc.,’ vol. 42, p. 410, 
1887), a certain want of sensitiveness is to be observed. If, on the other hand, two 
potentials and two times sufficiently far apart for accuracy are observed, an assump¬ 
tion of a law of resistance (generally Ohm’s) requires to be made in order to integrate 
the equation (Gray, ‘ Absolute Measurements,’ &c., p. 404), since, in the case of a 
very high resistance, the fall of potential is a small fraction of the total voltage only. 
This assumption, although philosophically repugnant, probably does not produce any 
very disastrous effects on the subsequent statement of results. An objection of 
greater gravity lies in the fact that very often the capacity of the quadrants of the 
electrometer cannot be neglected, and as these are affected by electric absorption 
(with electrometers constructed with glass insulation), a source of uncertainty is 
introduced. If the potential-time-curve-slope he used to give a value of the equiva- 
