Discharges in Insulators. 441 



the readings were reduced to comparable measure by compa- 

 rison with a long-range electrometer which I had made for 

 the most part according to Thomson's description. 



It was found that 6 of the units in which in the follow- 

 ing the differences of potential are expressed correspond to a 

 potential-difference of about 5 Daniells. I would not, how- 

 ever, lay too much stress upon these data, since the battery 

 at my disposal was too small to enable me to accomplish a 

 more accurate determination. 



Now, if we assume that the Holtz-machine electrode which 

 is connected with the gas-pipes conducts away negative elec- 

 tricity, the positive electricity issuing from the other electrode 

 finds two paths — the first through the rheostat to the gas- 

 pipes, and the second through the discharge-apparatus and the 

 galvanometer, likewise to the gas-pipes. The quantity of 

 electricity which passes through the discharge-apparatus can 

 now be varied within wide limits by altering the resistance of 

 the rheostat. The galvanometer indicates this quantity ; and 

 the electrometer measures the difference of potential between 

 the point and the plate. 



I soon observed that the discharge does not take place with 

 every difference of potential, but that rather a perfectly fixed 

 difference is always necessary in order to induce it. If at 

 the commencement of the experiment the resistance of the rheo- 

 stat has been made nearly = (when of course the deflections 

 of the galvanometer and electrometer are likewise = 0), and if 

 now the resistance be gradually increased, on the electrometer 

 indeed a steady rise of the potential will be observed, but the 

 potential must have reached a certain value before the galva- 

 nometer will show, by a sudden, proportionally great, and 

 constant deflection (if the resistance of the rheostat remains 

 invariable), that the discharge has commenced. When once the 

 discharge is present the resistance of the rheostat and conse- 

 quently the potential can be again diminished, through which 

 the discharge, it is true, steadily decreases, but does not at 

 once sink to zero. Only with a considerably less potential- 

 difference than that with which the discharge commenced does 

 it again entirely cease. 



I found, further, that the commencement of the discharge 

 was dependent on many collateral circumstances, e. g. whether 

 a discharge had taken place a shorter or a longer time pre- 

 viously ; unavoidable dust particles, too, have probably an 

 influence. On the other hand, determinations of the difference 

 of potential at which the discharge ceases, derived from differ- 

 ent experiments separated from one another by considerable 



