Discharge of Electricity. 391 



parts and the earth. A wire covered with gutta percha con- 

 nected the insulated parts with the spherical ball A, which was 

 supported on a varnished glass stem surrounded at the base 

 by pumice-stones soaked in sulphuric acid. The equal and 

 similarly insulated ball B was connected with the insulated 

 half-ring of the Thomson's reflecting electrometer. The long- 

 range absolute electrometer represented at the corner of the 

 table was obtained on loan from Sir William Thomson, to 

 reduce the results to absolute measure. 



Thus an essential feature of the method, due to Professor 

 Tait, is the introduction of the two insulated balls A and B, 

 by the separation of which to a greater or less distance the 

 effect of a charge upon the electrometer could be reduced to 

 any convenient amount. By this means we have been able to 

 measure very high potentials. When the potential of A is 

 raised by working the machine, the potential of B is raised 

 simultaneously ; and this goes on until a discharge takes place 

 between the electrodes inside the receiver. We found that it 

 was possible, by breaking the contact between the conductors 

 of the Holtz machine before beginning to turn the wheel, and 

 by turning slowly and uniformly, to cause the wire-image to 

 move along the scale continuously, and to be at rest at the 

 instant of the discharge. After the passage of the discharge 

 the image fell back to a position very near to the one it started 

 from ; a^nd the coincidence was improved by discharging the 

 couple of Leyden jars, which were generally attached to the 

 conductors of the Holtz machine, for the purpose of making 

 the charging more gradual. We noted the latter position of 

 the index, and took the difference as proportional to the differ- 

 ence of potentials of the electrodes at the instant of discharge; 

 for we found, by putting a charge on the ball A, dividing the 

 charge by means of an equal ball, and so on, that the differ- 

 ences referred to were proportional to the amount of charge 

 on A. Any escape of electricity from the charged conductor 

 was at once made evident by the behaviour of the wire-image; 

 it was this escape which put a limit to our measurements of 

 potential, and not the range of the electrometer arrangement. 

 In general we took three readings for each entry, so that we 

 might be able, by taking their mean, to eliminate any error 

 due to small variations in the mode of passing of the spark. 

 A notion of the amount of this error may be obtained by in- 

 specting the readings given below (p. 398). The method 

 admits of so much rapidity, that we have taken a series of 

 observations involving forty discharges in the course of an 

 hour. 



2F 2 



