372 TRANSACTIONS OF SECTION B. 



occui's near the open or closed end of the tube. In the first case, the flame 

 proceeded for some distance down the tube at a practically uniform and fairly 

 slow velocity, corresponding to the true rate of propagation ' by conduction ' ; 

 this period of uniform movement is succeeded by an irregular oscillatory period, 

 in which the flame swings backwards and forwards with increasing amplitudes, 

 finally either dying oui altogether or giving rise to detonation. With certain 

 oxygen mixtures the initial period of uniform slow velocity was shorter and 

 appeared to be abruptly succeeded by detonation, without the intervention of 

 any oscillatory period. When, however, such mixtures were ignited near the 

 closed end of a horizontal tube, the forward movement of the flame was con- 

 tinuously accelerated from the beginning, under the influence of reflected com- 

 pression waves, until detonation was set up. Such, in general, was the sequence 

 of the phenomena observed by these distinguished French investigators. 



They proceeded to determine experimentally the velocities of the uniform 

 slow movement of the flame in the case of a number of air and combustible gas- 

 mixtures, and plotting their results (in cm. per sec.) as ordinates against 

 percentages of inflammable gas as abscissse. they obtained ' curves ' which were 

 in each case formed of two inclined straight lines converging upwards to a 

 point which represented the composition and flame-velocity of the most explosive 

 mix'ture. And they concluded that the points at which the downward produc- 

 tion of the two lines met the zero velocity line would define the upper and 

 lower limits of inflammability for the particular series of gas-air mixtures. 

 Thus the curve they obtained for methane-air mixtures (fig. 1) showed a 

 maximum velocity of 61 cm. per second for a mixture containing about 

 12'2 per cent, of methane, with lower and upper limits corresponding to 5'6 

 and 16"7 per cent, of methane, respectively. 



An exact knowledge of the velocities of flame-propagation during this initial 

 period of uniform slow movement, as well as of the limits of inflammability for 

 mixtures of various combustible gases and air, is very important from a prac- 

 tical point of view. Makers of appara'tus for burning explosive mixtures of gas 

 and air want to know the speed of flame-propagation through such mixtures, not 

 only at ordinary temperatures and pressures, but also when the mixtures are 

 heated and used at higher pressures. Also it would be important to know 

 whether or not in the case of a complex mixture of various combustible gases 

 and air, where complete composition can be determined by analysis (as, for 

 example, coal-gas and air), the velocity of flame-propagation can be calculated 

 from the known velocitiies for its single components. Unfortunately, although 

 more than thirty years have elapsed since Mallard and Le Chatelier's work was 

 published, the necessary data are still wan-tdng to answer such questions, and 

 anyone who will systematically tackle the problem and carefully work it out in 

 detail will be doing a real service to the gas-using industries. I am hoping 

 shortly to make a beginning with such an investigation in my new Department 

 at the Imperial College, London, but the successful and rapid progress of such 

 work will involve considei-able financial outlay as well as organisation and expert 

 direction. Who will help us with the necessary funds? 



An accurate knowledge of the behaviour of methane-air mixtures imder 

 known variations of conditions is of prime importance from the point of view 

 of the safety of coal-mines, and it is rightly occupying the attention of my 

 friend and former collaborator, Dr. R. V. Wheeler, at the Home Office Experi- 

 mental Station at Eskmeals. And from papers which he has already published, 

 as well as from some unpublished results which he has very kindly permitted 

 me to refer to in this Address, it is now possible to correct certain errors in 

 Mallard and Le Chatelier's results, and to arrive at a clearer view of the 

 phenomena as a whole. 



In the first place, it would appear that the initial ' uniform movement ' of 

 flame an a gaseous explosion, or, in other words, propagation of the flame from 

 layer to layer by conduction only (as defined by Le Chatelier), is a limited 

 phenomenon, and is only obtained in tubes of somewhat small diameter, wide 

 enough, however, to prevent appreciable cooling of the flame, but narrow 

 enough to suppress the influence of convection currents. Moreover, ignition 

 must be either at, or within one or two centimetres of, the end of the tube, or 



