1895.] The Measurement of High Potential Difference. 157 



effective. Finally a method was devised by which a proper tempera- 

 ture correction could be easily calculated without necessitating any 

 additional readings, and which has the great advantage that the 

 coach-springs themselves are virtually used as a metallic thermometer 

 to indicate their own temperature, and this method of correction was 

 found to be quite satisfactory. The greatest value of this tempera- 

 ture correction in a large number of subsequent measurements of 

 potential difference was found to be ^ per cent. 



A description is next given of a lengthy series of experiments to 

 determine values for "v." (the ratio of the units) in order to test the 

 a' -curacy of the electrometer when employed according to the authors' 

 method. The results show that the values of " v " in each experiment 

 form a very consistent set among themselves, but that the sets taken 

 on different days do not always agree with one another, on some days 

 the mean value obtained for "v" being about 2 999 x 10 10 , while on 

 other days it was about 2'985 x 10 10 . The explanation of this is that 

 the electrometer was on some days measuring quite correctly in 

 absolute units, in so far as we were able to judge, while on other 

 days its measurements were almost uniformly too high by about 

 •J per cent. 



This result tends to show that in the instrument used by the 

 authors there is some defect, probably of a mechanical nature, which 

 causes an intermittent error of such a nature that, when it occurs, it 

 is almost always approximately the same in magnitude. The authors 

 spent some time trying to detect the cause of this error, which 

 appeared to be inherent, at any rate in this specimen, of the absolute 

 electrometer, but they failed to arrive at a satisfactory explanation. 



By employing the various precautions adopted by the authors, the 

 mean of a number of measurements of a potential difference made 

 with this absolute electrometer, in the way described in the paper, 

 may be relied on as being correct, certainly to within \ per cent, in 

 absolute electrostatic units, if 2*998 x 10 10 is the true value of M v." 

 On the days that the special error, already referred to, did not exist, 

 the measurements were correct to about per cent. Part II con- 

 cludes with a description of some alterations, suggested by the 

 authors' experiments, which will be carried out in the instrument 

 with which these experiments were made. 



In Appendix I are theoretically investigated the most suitable values 

 of the mass to be used for the initial adjustment of the springs, and 

 of the potential difference to which the electrometer jar is to be 

 charged, for the measurement of a given potential difference in the 

 heterostatic method. The results show that, in order that the prob- 

 able error shall be as small as possible, the potential difference of the 

 electrometer jar should be proportional to the potential difference to 

 be measured, and the mass should be proportional to the -§- power of 



