required to start a Gaseous Explosion. , 737 



second WS = /L 1 W/Y is constant so that W = /t 2 V; in the third 

 part WV 2 is constant, a curve falling slower than the first 

 branch. These agree so well with the observed form of 

 energy-voltage curve that it can be fairly said that the re- 

 quired energy for ignition by alternating current depends 

 almost entirely on the duration of the arc in the gas, sug- 

 gesting that this type of ignition may have a thermal origin. 

 This is supported by the shape of the current-percentage 

 curves of fig. 2 of the previous paper, in which the igniting 

 power is proportional to the square of the current broken. 



4. Shape of Curves. — Since the relation that VIS is con- 

 stant appears to explain in a satisfactory manner the origin 

 of the energy-voltage curves, and since the energy of the 

 break-spark is itself proportional to VI, it follows that the 

 whole sequence of curves is derived from the simple relation 

 that the duration of any least igniting break-spark is inversely 

 proportional to its energy. Though the law is the same 

 whatever the material of the poles, as seen especially by 

 fig. 4, it is clear that ignition is a relation between the 

 nature of the gaseous mixture and the incandescent matter 

 projected into it from the poles, such that the above relation 

 may hold. In continuous current ignition it is the maerial 

 of the poles which is decisive, with alternating currents the 

 nature of the gas; for in the former case the matter is pro- 

 jected as fully as possible from the point of break, in the 

 latter it is constrained to move to and fro between the poles 

 by the alternating gradient. Everything, therefore, favours 

 a thermal origin of self-ignition in the case of low-frequency 

 alternating currents, but something much more intimate in 

 the case of continuous currents. The results obtained by 

 condenser-discharge spark ignition in which the duration is 

 so short that heating from the spark cannot spread far before 

 extinction, show a very close connexion between the ignition 

 and the nature of the gas. It will be shown in a later paper 

 that there are important relations between energy of ignition 

 and the relative numbers of molecules of the reacting gases 

 which are quite masked in travelling explosion waves. 



5. Ignition by Condenser Discharge-sparks. — When, how- 

 ever, ignition is made by condenser sparks at different volt- 

 ages in the same strength of mixture, the form of energy 

 curve is very nearly the mean of the two types of break- 

 spark. For example, in fig. 5 the curve for methane is in 

 shape, though not in magnitude, an average between figs. 1 

 and 3. Carbon monoxide, fig. 6, is nearer to the continuous 

 current type, and hydrogen, fig. 7, still more so. Fig. 8 for 

 coal-gas is given, for since it is composed of about 47 per 



