Time- Lag in the Spark-Discharge. 217 



gap, or by causing the emission of electrons from the 

 electrodes. It is generally attributed to the fact that a few 

 casual ions must be present in the region of the electric 

 field in order to start the ionization by collision which 

 results finally in the passage of the spark. If this expla- 

 nation is correct, it is to be expected that the lag should be 

 greater in the needle gap than in the sphere gap; for in the 

 former the volume of the region in which there is an intense 

 electric field is much smaller than in the latter : in the former 

 the gradient is concentrated in the immediate neighbourhood 

 of the electrodes ; in the latter it is approximately uniform over 

 the whole distance between the electrodes. In the needle 

 gap therefore, the chance that there will be a casual ion at a 

 place where the electric field is sufficiently great to cause 

 ionization by collision is much smaller than in the sphere gap; 

 and, if we rely on chance to produce an ion in the right 

 place, the time that we may have to wait before that event 

 occurs will be correspondingly longer. 



But if it were lag of this nature which caused Peek's 

 observations, it would be expected that the impulse ratio for 

 sphere gaps should be rather greater than 1, that it should 

 be greater than 1 for all periods of alternation less than a 

 few seconds, and that, for such periods, it should be de- 

 termined by the wave-form rather than the frequency; for 

 the chance of an intense electric field coinciding with the 

 presence of a casual ion in the right place would depend 

 on the fraction of the whole period during which the peak 

 lasted and not on the absolute period. None of these ex- 

 pectations are fulfilled. Again, the time-lag should be 

 extremely irregular, the potential at which the spark passes 

 should be extremely variable and it should sometimes happen 

 that a spark passes at practically the steady sparking potential ; 

 further, the time-lag should be abolished by ionizing the air 

 in the gap. Peek denies that in his experiments initial 

 ionization had any effect upon the spark potential and 

 makes no mention of any irregularity in his measurements 

 of that potential. 



In this last matter, Mr. Peek's experience seems at first 

 sight inconsistent with some facts familiar to all magneto 

 manufacturers. They realize that the point or needle gap 

 is utterly unreliable for measuring the peak potential of a 

 magneto, not only or mainly because its sparking potential 

 depends on the frequency, but because its indications arc 

 so very indefinite; there is always a very large range of 

 magneto potentials over which the gap will sometimes spark, 

 but will not spark at every break. Thus a point gap may 



Phil. Mag. S. 6. Vol. 38." No. 221. Aug. 1919. Q 



