222 Dr. Norman Campbell on 



to a source o£ steady high potential through a large resist- 

 ance (10 megohms or more), a discontinuous stream of 

 sparks passes across the gap ; Q, the quantity of electricity 

 passing in each of these sparks, is the product of V, the 

 sparking potential of the gap. and (J, the capacity in parallel 

 with the gap. In these observations there was no time-lag,, 

 and the actual sparking potential was the same as the steady 

 sparking potential ; but if the two are not identical, then, so 

 long as the discharge still consists of successive individual 

 sparks, V in the relation Q = CV is doubtless the actual 

 sparking potential, that is, the potential between the elec- 

 trodes when the spark starts; for the time that the individual 

 spark lasts is so short that, with the highest frequencies inves- 

 tigated here, the change of potential during its passage is too 

 small to be appreciable. The experimental devices described 

 in the paper to which reference is made enable Q and C to 

 be measured ; accordingly Y can be determined. 



This method of determining the actual spark potential was 

 applied to a gap between needle-points, the steady spark 

 potential being varied between 3000 and 6500 volts by variation 

 of the distance. The steady high potential, to which the gap 

 was connected for a time long enough to allow one spark to 

 pass, was of course always greater than the steady spark 

 potential, and could be increased to 12,000 volts. It was 

 found that, owing to the irregular lag, the spark did not 

 always pass even when the greatest potential reached ex- 

 ceeded considerably the steady spark potential, and, owing to 

 the regular lag, never passed unless this potential was some- 

 what greater than the steady spark potential. On the other 

 hand, if the spark did pass, Y turned out always to be the 

 steady spark potential, and to be the same as it was if there 

 were substituted for the point gap a sphere gap, free from 

 time-lag, with the same steady spark potential. 



This coincidence of the actual sparking potential with the 

 steady sparking potential was found for all the conditions 

 investigated. No theory can be suggested which explains it 

 fully, but it receives a partial explanation on Peek's theory of 

 the lag. On that theory the period of the time-lag is occupied 

 in the development of the corona. Now when the corona is 

 formed, the current through the gap is not passing between 

 the electrodes only, but from the high potential electrode to 

 all the surrounding earthed conductors of which the low 

 potential electrode forms a small part. The quantity of 

 electricity" passing during this stage will not be included 

 in Q, which is the quantity received by the low potential 

 electrode. Accordingly, if the discharge starts in the form 



