of Electricity through Gases. 325 



soon as the spark passed the potential between the electrodes 

 in the dry gas indicated by the voltmeter fell to a fraction of 

 the value it had just before the spark passed ; the arm EF 

 could now be moved back in the direction of diminishing 

 potential until it had often got more than halfway back to D 

 before the sparks ceased. The readings of the voltmeter showed 

 that the spark-potential did not vary greatly during this time 

 from the value it had fallen to immediately after the passage 

 of the first spark. Another point to be noticed was that the 

 first spark did not begin as a small faint spark, but, on the 

 contrary, when it did appear it was exceedingly bright, as if 

 the gas under the large electromotive force had been in an 

 unstable condition, and when, by some chance, it did break 

 down it did so with a rush. When, however, after the spark 

 had been started it was made to disappear by moving the 

 arm EF in the direction of diminishing potential, the spark 

 disappeared gradually and was exceedingly faint just before it 

 ceased to be visible. The sudden fall of potential at the spark- 

 gap which took place just after the passage of the first spark 

 was often fatal to the carbon resistance. The reason of this is 

 that the difference of potential between C and D is constant ; if 

 the revolving arm is in the position EF when the spark passes 

 and the potential suddenly falls, as it does in these experi- 

 ments, to less than half its original value, the loss in potential 

 between F and D must be compensated by a gain in potential 

 between C and F. In consequence of this increase in the 

 potential-difference the current along C and F is increased ; 

 in the actual experiments this increase in the current was so 

 great that the carbon between C and F continually took fire 

 after the first spark passed, and broke the connexion of the 

 battery with the spark-gap. In the later experiments this 

 inconvenience was avoided by inserting a very large resistance 

 between C and the hydrogen-bulb. 



The experiments showed that while the potential at which 

 the sparks stopped when the arm EF was moved in the 

 direction of diminishing potential was nearly constant, and 

 was little influenced by the time the spark had been passing, 

 or by whether the sliding arm had only just been moved far 

 enough to start the sparks or whether it had been carried 

 far past this point ; on the other hand, the potential at 

 which the sparks began was very variable, and changed ap- 

 parently capriciously from time to time. Again, the potential- 

 difference when the sparks stopped was not very different for 

 the damp and dry gases (the voltmeter was not sensitive 

 enough to measure small differences in the electromotive 

 force); on the other hand, the potential-differences when the 



