very small Intervals of Time. 345 



intervals adjusted mechanically and those calculated by means 

 of electrical leakage is very close, and that the formula 



t=fr log e — J is not only applicable for slow discharges through 



very high resistances, but equally so for quick discharges 

 through low resistances. 



These two series of test-experiments I think show conclu- 

 sively that this system affords results which, through a wide 

 range of time-intervals, are comparable with each other — and, 

 further, that within a small percentage these calculated inter- 

 vals are correct. 



In the method in which two circuits are broken, the dischar- 

 ging resistance is kept in circuit whilst the accumulator is being 

 charged, on the assumption that the battery-resistance is so 

 small in comparison with it that the potential of its poles is not 

 thereby altered. This is theoretically evident; and the following 

 observations of the instantaneous discharge when the battery- 

 poles were connected by resistances of various values show 

 that, until the discharging wire is reduced in resistance much 

 below that which is required to be used in the method, no ap- 

 preciable error is introduced. 



Resistance between -rv -, 



poles of battery. Discharge. 



Infinite 277 



100000 ohms ... 277 



10000 „ , . . 277 



1000 „ ... 276 



It is also assumed in both ways of working, that the time 

 which elapses between the interruption of the discharging- 

 wire and the closing of the galvanometer-circuit is insufficient 

 to allow of leakage through the dielectric of the accumulator. 

 The following experiment was made for the purpose of ascer- 

 taining the actual leakage of the accumulator for various pe- 

 riods of insulation : — 



Instantaneous discharge . =275 



Discharge after 1 minute . =275 



„ 15 minutes . =272 



„ 30 „ . =269 



It is evident, therefore, that, were even fifteen minutes 

 allowed to elapse before taking t^ie reading, the error through 

 this delay would be only about 1 per cent. There is nothing, 

 however, to prevent the reading being taken within a second 

 or two. 



