

Time-Lag in the Spark Discharge. 225 



This very obvious view seems to explain qualitatively the 

 Avhole matter. It shows the connexion between the regular 

 and irregular lags, for the occurrence of both is due to the 

 fact that the region occupied by the field in which ionization 

 by collision can occur is much smaller in the needle gap than 

 in the sphere gap. It explains why, during the period which 

 leads up to the spark, the current flowing from one electrode 

 does not all arrive at the other, and it shows also that the 

 discharge during this period need not be the same as the 

 fully developed corona. The matter needs, of course, much 

 fuller investigation, and there are many lines of experi- 

 mental research which might be followed up. But since it is 

 unlikely that I shall pursue the subject further myself, it 

 has seemed well to record these fragmentary observations. 



Hard and soft gaps. 



It is not only in point gaps that there is a time-lag and 

 that the sparking potential depends on the frequency; the 

 same dependence is sometimes found in sphere gaps and 

 others in which the sparking distance is not large compared 

 with the radius of curvature of the electrodes. Sparking- 

 plugs are often found which are "hard, 1 " and will not pass 

 the spark from a magneto, although to all appearance they 

 are precisely the same as " soft " plugs which will pass the 

 spark, and although they have the same steady sparking 

 potential as such soft plugs. At first sight this difference 

 between hard and soft plugs seems essentially similar to that 

 between point and sphere gaps. 



A large amount of information on the subject was obtained 

 in the course of routine tests on numerous types of sparking- 

 plug. In these tests the sparking potential 'o£ the plug was 

 measured, both with a steady potential and with a magneto, 

 in air at atmospheric pressure and also in air at a pressure 

 of about 5 atmospheres. (By the sparking potential with a 

 magneto is always meant the peak potential which will 

 cause a spark.) The following facts were established: — 



(1) Plugs could be divided into two classes, according as 

 their sparking potentials were definite or indefinite. When 

 the sparking potential was definite the difference between 

 the potential which would sometimes give a spark and that 

 which would always give a spark was scarcely greater than 

 could be read on the voltmeter. At atmospheric pressure, 

 when the sparking potential was about 2000 volts, the 

 difference would not exceed 50 volts ; at the higher pressure, 

 when the sparking potential was about iHXX) volts, the 



