of a Magneto or Induction- Coil. 491 



on the assumption that the break was perfect. Unfortunately, 

 however, there are circumstances (connected with the 

 damping of the oscillations) which make an accurate calcu- 

 lation of the oscillations impossible, even on the assumption 

 that the break is perfect. But there seems to be consider- 

 able indirect evidence that if the break is " sparkless '' it is 

 at least very nearly perfect. For it is found that, so long- 

 as no visible flash occurs, the potential develoj:>ed in the 

 secondary is very nearly proportional to the current broken 

 in the primary ; but as soon as the limiting current is over- 

 stepped and the flash occurs, the potential developed falls at 

 least 30 per cent. The transition from one value of the 

 secondary potential to the other is perfectly sharp ; inter- 

 mediate values are never observed and the lower potential 

 only occurs when a flash is seen. Moreover, so long as no 

 flashing occurs the peak potential obtained when a given 

 current in the primary is broken is perfectly definite and 

 consistent at successive trials ; if there is flashing, the 

 values obtained at successive trials are much less regular. 



Such observations seem to show that the occurrence of a 

 flash indicates the occurrence of some action which does not 

 occur at all if there is no flash, and that it does not merely 

 indicate an increase of some process which occurs even 

 when a flash is not seen. 



(6) True sparking at the break. — The observations seem 

 also to show that if we suppress arcing we are not troubled in 

 normal circumstances with the occurrence of a true spark, as 

 suggested by the theory which we considered originally. 

 But in order to make this conclusion more certain, it was 

 thought of interest to inquire whether, in any circumstances, 

 true sparking at the break could be observed. 



Such circumstances could be obtained if, with a current 

 so small that there is no possibility of arcing, the rate of 

 increase of potential can be made greater than the rate of 

 rise of the sparking potential of the gap ; accordingly either 

 the rate of rise of potential when a given primary current 

 is broken must be greatly increased, or the rate of rise of 

 sparking potential greatly reduced. The first is the more 

 convenient alternative to adopt ; the necessary conditions 

 are readily secured by increasing greatly the self-inductance 

 of the circuit in which the current is broken. Observations 

 were made, therefore, when the contact to be broken was 

 inserted in the secondary, instead of in the primary, circuit of 

 he magneto. 

 It was now found that a visible spark occurred at the 



