218 



Dr. Norman Campbell on 



easily be found which will sometimes pass a spark when the 

 magneto peak potential is no greater than 7000 volts, and 

 yet will not pass a spark at every break unless the peak 

 potential is 12,000 volts; the sparking potential of such a gap 

 for steady voltages might be 5000 volts. This indefiniteness 

 of the spark potential is in accordance with the presence of a 

 time-lag due to the necessity for casual ions. Further, it is 

 known that the indefiniteness of the point gap can be 

 very greatly diminished by the device introduced in the 

 arrangement called the " 3-point gap." This arrangement is 

 A and B are the main electrodes connected 



shown in fig. 1. 



Fig-, l. 



d 



B 



To high potential 



To Earth 



to the magneto terminals; C is a pointed rod insulated from 

 both A and B with its point very near A ; B is usually 

 earthed. As the potential of A changes a charge is induced 

 on C, and the difference of potential of A and C is sufficient 

 to cause a spark to pass across the small gap between them. 

 If C is carefully adjusted by trial, a position can be found 

 in which regular sparking (i. e. sparking at every break of 

 the magneto) begins at a peak potential much lower than 

 that at which it begins in the absence of 0; for instance, the 

 regular sparking potential may be reduced from 12,000 to 

 8000 volts. The obvious explanation of the changes is that 

 the ions produced by the " pilot " spark between and A 

 diffuse or are projected into the space between A and B, and 

 reduce the " lag " in just the same way as ions produced by 

 the action of ultra-violet light on the electrodes. 



These experiments then seem consistent with the expla- 

 nation of the lag by casual ionization rather than by Peek's 

 theory ; while Peek's observations seem consistent rather 

 with his own theory, which asserts that the lag is not a 

 variable quantity depending on chance, but the time required 



