220 Dr. Norman Campbell on 



that i£ the time occupied by the rise of potential of the 

 electrodes from zero to the peak is increased, the number 

 of casual ions present when the sparking potential is reached 

 will be fewer and the sparking be made less regular. Some 

 observations in which the wave-form of the potential given 

 by the magneto was changed slightly, either by inserting a 

 high resistance between the magneto and the gap or inserting 

 a capacity in parallel with the gap, seemed in accordance 

 with this expectation. On the other hand, if the rise of 

 potential were made sufficiently slow the opposite effect was 

 obtained, and regular sparking occurred at lower peak 

 potentials than before. Such a change must, of course,, 

 occur as the condition of a steady applied potential is 

 approached. But it should be noted that when the regular 

 sparking potential was decreased in this manner, the potential 

 at which a spark would sometimes pass was also decreased ; 

 while when the regular sparking potential was iucreased by 

 changing the wave-form, this potential was unaltered. 



It appears therefore that the difference between the 

 regular sparking potential and the potential at which a spark 

 will sometimes pass can be accounted for completely by the 

 theory which attributes that difference to the necessity for 

 the presence of casual ions to start the discharge. On the 

 other hand, that theory will not account for a difference 

 between the least peak potential which ever gives a spark 

 and the steady sparking potential. It is of course difficult 

 to be sure of the limiting peak potential of the magneto, 

 below which a spark will never pass, but all evidence pointed 

 to a clear difference between this potential and the sparking 

 potential with a steady source : in particular, consecutive 

 determinations of the limiting potential were very consistent. 

 The conclusion is indicated that there are two kinds of 

 " time-lag " involved : one an irregular lag- connected with 

 the presence of casual ions, the other a regular lag which 

 causes the difference between the limiting peak potential and 

 the stead}' sparking potential, and which may possibly be 

 explained by Peek's theory. The fact that Peek did not 

 observe the irregular lag and the change of this lag with 

 initial ionization may plausibly be attributed to his use of 

 much higher frequencies than any given by a magneto. The 

 regular lag increases with the frequency and, when the 

 frequency is as great as a million a second, may be so great 

 as to obscure entirely the irregular lag which, as has been 

 suggested, should not vary much with the frequency if it 

 exceeds some quite low limit. 



But though the irregular and regular lags seem distinct. 



