Time-Lag in the Spark Discharge. 223 



of a corona at a potential considerably higher than the steady 

 spark potential and continues for a finite period until it is 

 ultimately transformed into a spark ; and if, during this 

 period, the potential difference between the electrodes falls 

 continuously: then the measured value of V, which is the 

 potential at which the spark as distinct from the corona 

 starts, will be lower than the potential at which the corona 

 starts, and lower than the potential required to cause the 

 spark to pass ultimately. Now, for a reason to be given 

 immediately, it is probable that in these observations the 

 potential between the electrodes did fall continuously after 

 the discharge began, and it is therefore possible to explain 

 why V is always less than the potential required to start 

 the discharge. On the other hand, it is not obvious why 

 it should always coincide with the steady sparking potential ; 

 that is, why the development of the corona and the preparation 

 for the true spark should never be complete until the potential 

 has fallen to the steady sparking potential. Nevertheless the 

 observations do provide some support for the theory that the 

 process which occupies the period of the time-lag is one in 

 which current is flowing to other places than the low potential 

 electrode. 



However, it is not probable that this process consists in the 

 fall development of the ordinary corona. The period during 

 which the potential was applied to the spark gap was about 

 *005 sec. and very nearly the same in all these experiments. 

 It was divided into two parts, in the first of which the 

 potential rose from zero to the maximum, in the second of 

 which it remained constant. It was found that the frequency 

 with which the spark passed decreased with the time occupied 

 by the rise of potential between the electrodes from zero to 

 its maximum value. This observation accords with that de- 

 scribed on p. 220, and the explanation given there is probably 

 correct. But it was found also that, if the potential was 

 raised very slowly, then, even if it were kept on indefinitely, 

 no spark passed, but a visible corona was formed on the high 

 potential electrode. Owing to the presence of a very high 

 resistance between the gap and the source of potential, the 

 current conveyed by the corona reduced the potential between 

 the electrodes so much that, even when the potential would 

 have risen to 12,000 if no current had been flowing, the 

 potential did not actually rise above 4000 volts, the steady 

 sparking potential of the gap. (It is for this reason that it is 

 probable that the potential always fellas soon as the discharge 

 began at all.) It appeared then that if the full corona did 

 develop, the spark would never pass owing to the drop in 



