216 Dr. Norman Campbell on 



America. It is that whereas in the sphere gap the sparking 

 potential is independent of the humidity of the air if the 

 pressure is unchanged, the sparking potential of the needle 

 gap varies considerably with the humidity, even when ihe 

 frequency is so low (say 500 a second) that the impulse ratio 

 is indistinguishable from 1. 



Mr. Peek explains his observations on impulse ratio by 

 supposing that there is a " time-lag n in the passage of the 

 spark. According to his theory, a finite time must elapse 

 after the potential reaches the steady sparking potential 

 before the spark will pass. If a potential greater than the 

 steady sparking potential is imposed, the time-lag though 

 less than before is still finite : it diminishes with increase of 

 the excess of the applied potential above the spark potential. 



It is obvious that such a theory will explain the variation 

 of the sparking potential with the frequency, and indeed it 

 may seem that it is the only theory that can possibly do so.. 

 For if the spark gap can distinguish between different 

 frequencies, there must be involved in the passage of the 

 spark some process requiring a finite time against which the 

 frequency can be measured ; and it is difficult to see where 

 this process can occur except between the attainment of the 

 requisite potential and the passage of the spark. But it 

 remains open to inquiry by what the time-lag is determined, 

 whether it is constant in successive discharges, and why it is 

 so very different for a sphere gap and a needle gap. Peek 

 supposes that the period of the finite time-lag is occupied by 

 the development of a "corona " or brush discharge, which is 

 a necessary forerunner of the true spark ; in the sphere gap,, 

 so long as the sparking distance is not large compared with 

 the radius of the spheres, the corona does not develop. But 

 ho does not explain (indeed it is not his purpose to do so) 

 why the corona is a necessary forerunner of the spark in a 

 needle gap but not in a sphere gap. 



The conception of a time-lag in the spark discharge is,. 

 of course, not new. It is discussed at some length and the 

 experimental evidence for it reviewed at the opening of 

 Chap. XV. of J. J. Thomson's k Conduction of Electricity 

 through Gases." Thus it is well known that if the potential 

 across any gap is raised slowly, the spark will not usually 

 start until a potential is reached considerably greater than 

 that required to maintain the spark once started, and that a 

 spark will sometimes pass after a steady potential has been 

 applied for a time amounting to several minutes, although 

 it did not pass when the potential was first applied. This* 

 " lag " in the spark is abolished by ionizing the air in the 



