Time-Lag in the Spark Discharge. 215 



I wish to express my gratitude to Sir Richard Glazebrook, 

 Director of the National Physical Laboratory, and more especially 

 to Mr. C. C. Paterson, in whose department the work was done, 

 for many suggestions and much invaluable advice. 



Introduction. 



IT is a fact familiar to all who have been concerned with 

 spark-ignition systems that the sparking potential of 

 a given gap in a given gas may vary with the frequency 

 of the applied potential. In some gaps at least, sparking 

 will occur only if the peak potential of a rapidly alternating 

 source (such as a magneto) exceeds very greatly the steady 

 potential which would be necessary to start a spark across 

 the gap. In other gaps the peak potential of an alternating- 

 source required to spark the gap is very much more nearly 

 equal to the steady potential sparking, though there is some 

 conflict of evidence whether the power of a source to cause 

 the spark is ever completely independent of the frequency 

 and ever depends simply on the peak potential. It is known 

 further that among the gaps in which the sparking potential 

 varies greatly with the frequency are those in which the 

 electrodes are fine points; but it is generally suspected that 

 the dependence on frequency is not determined entirely by 

 the shape of the electrodes, and that the nature of the 

 material of which they are made and of the gas in which 

 they are placed may also exert some influence. 



The matter has been investigated in some detail by 

 F. W. Peek, who described his results in his book 

 ' Dielectric Phenomena in High Voltage Engineering ' 

 (1915). The ratio of the peak potential of an alternating 

 source which will just spark the gap to the steady sparking 

 potential is called by Peek the "impulse ratio" of the gap. 

 For a " sphere gap," or any gap in which the radius of 

 curvature of the electrodes is large compared with the 

 distance between them, the impulse ratio was found to be 1 

 for all spark potentials in air and some other gases, and for 

 frequencies of the applied potential as high as a million a 

 second. For a " needle " or " point " gap, on the other 

 hand, the impulse ratio is always greater than 1 and in- 

 creases both with the frequency and with the spark potential, 

 if the spark potential is varied by changing the distance 

 between the electrodes. Values as great as 3 or 4 were 

 found. This difference between needle and sphere gaps is 

 associated with another difference which has been described 

 by Peek and his colleagues of the General Electric Co. of 



