226 Dr. Norman Campbell on 



difference would not be greater than 100 volts. On the 

 other hand, in extreme cases of indetiniteness the difference 

 at the lower pressure might amount to 1500 volts ; at the 

 higher pressure it was never found to exceed 600. 



(2) If the sparking potential was definite, the sparking 

 potential with the magneto was equal to the steady sparking 

 potential. If it was indefinite, it was higher with the 

 magneto*. In this matter, then, hard and soft gaps resemble 

 point and sphere gaps; irregularity is associated with an 

 4 * impulse ratio *' greater than 1. 



(3) The " impulse ratio " was much less at the higher 

 pressure, when the sparking potential was greater ; in fact, 

 at these higher pressures it could never be established cer- 

 tainly that the impulse ratio was really greater than 1. In 

 this matter hard and soft gaps are unlike sphere and point 

 gaps, for in these the impulse ratio increases with the spark 

 potential. 



(4) The cause of " hardness " was connected with some 

 very easily variable surface condition of the plug. It had 

 nothing to do with the geometrical form or material. The 

 hardness was liable to change with the most trivial alterations ; 

 it usually seemed much easier to make a soft plug hard than 

 to make a hard plug soft. 



These observations were supplemented by others on a 

 specially constructed gap in which one electrode was a plane 

 of which the material could be varied, the other a steel 

 sphere. At the same time a very convenient and sensitive 

 method for detecting hardness was introduced, which de- 

 pended on the application of the potential for a very short time 

 only. It was a modification of the device of the " auxiliary 

 gap," which has been applied to making the spark pass 

 across a " leaky " ping, that is one in which the insulation 

 has become covered with a conducting layer. 



The gap under consideration was shunted by a resistance 

 of 10,000 ohms, representing the " leak." If the distance 

 between the electrodes is now set to represent a steady 

 sparking potential of about 5000 volts, a magneto will not 

 produce a spark across the gap. But if, between the gap 

 and the magneto, is inserted a second gap which is not leaky, 

 the passage of a spark across this auxiliary gap will often 

 be accompanied by a spark across the leaky gap. In order 

 that this may happen certain conditions, involving the 

 capacity in parallel with the two gaps and their sparking 



* Some few exceptions to this statement were noted, the sparking 

 potential being less with the magneto. These observations were never 

 explained. 



