1902.] 



Theory of the Capillary Electrometer. 



223 



dilute sulphuric acid, the surface tension of which causes it to fill the 

 trough to its upper end. The acid is contained in a U-tube, the bend 

 of which is filled with mercury. Two platinum wires, dipping respec- 

 tively into the mercury in the U-tube and that in the capillary, serve 

 to connect the electrometer with the experimental circuit. A few 

 weeks after it was set up, it was accidentally left, charged to about 

 0-025 volt, with the circuit open from 1 p.m. on Saturday to 10.30 a.m. 

 on Monday. The image of the meniscus was still, as it had been left 

 45J hours before, between two reference lines on the screen corre- 

 sponding to a difference of potential of 0-0008 volt, so that this repre- 

 sents the maximum loss, and as the zero-position had not altered, at 

 least 97 per cent, of the original charge still remained in the electro- 

 meter. And as the circuit included no less than seven keys and 

 switches and about 23 metres of wire, it is probable that the instru- 

 ment was not responsible for all the leakage. Inasmuch as it might be 

 objected that the mercury had stuck in the tube and so maintained its 

 position before closing the circuit, I waved an electrified ebonite tube 

 to and fro near one of the terminals. This causes the meniscus to 

 move up and down by induction exactly as it would cause the gold 

 leaves of a charged electroscope to diverge more widely or collapse. 

 But just as the gold leaves would remain divergent after the final 

 removal of the ebonite at the same angle as when they were first 

 charged, so the meniscus when the ebonite is taken away returns to 

 the position it had assumed in virtue of the original charge put into it, 

 always supposing that no sparks have been allowed to pass. This 

 method is extremely useful in guarding against false readings due to a 

 sticky tube. 



I have verified this result several times since by observations ex- 

 tending over 5 or 6 hours, but I have not cared to risk leaving the 

 key open during my absence from the room. 



After a lapse of nearly four years the insulation resistance is 

 naturally less, but it varies greatly with the weather. The easiest way 

 of making comparative measurements is to observe the time required 

 for any charge to fall to half its initial difference of potential. This 

 I shall refer to briefly as the " time of half -discharge," and it answers 

 to the "time constant" of a condenser, but is simpler to use in 

 practice, as it is easier to divide a number by 2 than by 2-71828, 

 and also to observe deflections consisting of a whole number of scale 

 divisions. 



On March 27th of this year, the time of half-discharge was 

 13 minutes. On April 14th, just before rain, the weather having 

 suddenly become warmer, the time of half-clischarge was only 

 65 seconds. This is an exceptionally low value. A gold-leaf electro- 

 scope in the same room could not be made to retain a charge for 

 10 seconds. 



