November 23, 1^93] 



NATURE 



87 



mixture may vary considerably and still retain its explosive pro- 

 perties. There i<, of course, a certain mixture which presents 

 the greatest expl sive power; a further quantity of the com- 

 bustible gas or of ihe air will diminish the explosibility, but not 

 entirely destroy it till a large excess is used. With hydrogen, 

 for example, two and a half times the volume of air (which 

 contains exactly the oxygen requisite to combine with the hydro- 

 gen and produce water) is the right quantity for the maximum 

 explosive effect, but we still get explosion when we have 

 much mere than two and a half times as much air as hydrogen, or 

 u-hen, on the other hand, we have much less. In one case 

 there will be oxygen left uncombined, in the other case hydrogen. 

 T dwell upon this in order that we may be prepared to find 

 the same thing in flames, in order that we may not be sur 

 prised to find combustion taking place in mixtures where 

 either gas or air is in excess of the quantity actually 

 required for the purpose of chemical combination. Bear- 

 ing this in mind, let us revert to the experiment that I 

 have just shown. It consists, you remember, in mixing air 

 with gas before burning it, to such an extent that the flame 

 strikes down the tube. On a close examination we find that 

 this is not quite a correct statement, for when I regulate the 

 air with nicety you see that it is only part of the flame that strikes 

 down the tube. There remains all the while at the top of the 

 tube ano her part of the flame which is not mobile. With a 

 little care I can adjust the proportion of air and gas so that the 

 part of the flame which is mobile shall move up and down the 

 tube like a piston. All the while you see the pale steady flame 

 at the top of the tube. When in this critical condition a little 

 more air determines the descent of the movable part of the flame, 

 1. little less sends it to the top. 



Let us now turn to the explanation of this phenomenon. It 

 i-; clear, in the first place, that coal-gas and air form an explosive 

 mixture long before there is enough air to burn all the gas. For 

 it is only part of the flame that descends the tube, and there is 

 jnough gas passing through this part to form a second flame as 

 lon as it reaches the outside air at the top of the tube. There 

 s, as a matter of fact, only about two-thirds as much air enter- 

 ing the tube at the bottom as would be necessary to burn the 

 whole quantity of ga<. We see, in the next place, that the ex- 

 plosibility varies greatly according to the proportions of gas and 

 air. For what is the cause of the descending flame? It is 

 simply that we have an explosive mixture in process of inflam- 

 mation. The inflammation is tending downwards ; opposed to 

 it is the movement of the explosive mixture upwards. If the 

 upward movement of the unburned mixture is more rapid than 

 the downward tendency of the inflammation, the flame cannot 

 descend. W.e can only make it descend by making the down- 

 ward tendency greater. This we do by adding more air, and 

 making the mixture more explosive. We see that we can 

 balance these two opposite velocities with the greatest nicely by 

 a careful adjustment of the proportions of the explosive 

 mixture. 



In order to ascertain what proportion of gas is being burnt in 

 this movable flame, and what is the chemical character of the 

 products there formed, it is necessary to keep the two parts of 

 the flame separate, and to take out some of the gases from the 

 intervening space. 



This is very easily done. The flame descends, we have seen, 

 i because its rate of inflammation is greater than the rate of 

 I ascent of :he combustible mixture. If now we can make this 

 rate of ascent more rapid at one part of the tube than it is any- 

 where else, we may expect to stop the descent of the flame at 

 I that point and keep it there. We can do this simply by ch' king 

 \ the passage, for just as a river must flow rapidly where its banks 

 I are close, so must the stream of gas rush more rapidly where 

 the tube is choked than either below or above, where there is 

 i a wide passage. If, then, I replace the plain glass tube by one 

 I that has a constriction in one part, and if I cause the inner cone 

 j of the flame to descend as before, it stops, as you see, at the 

 I constriction, and will remain there any length of time. Its rate 

 • of descent is greater than the rate of ascent of the gas where 

 \ the tube is wide, but not 50 great as that where it is narrowt-d 

 I by the constriction. We have now got the two cones of flame 

 [ widely separated. In this state of things we can, if we choose, 

 1 draw off the gases from the space between the two cones by 

 I putting in a bent glass tube and aspirating. We could then 

 : analyse these gases and see what has happened in the first 

 cone. (Fig. i, A.) 



I will now show you another method in which the two cones 



NO. 1256, VOL. 49] 



can be separated. It is based on the same principles as the one 

 just used. I have here a iwo-cmed flame burning at the top 

 of a glass tube. I shdl let the air supply t)e liberal, but not 

 quite sufficient to cause the descent of the inner cone. The 

 rate of ascent of the gas is now just a trifle greater than the rate 

 of descent of the flame. If now I retard the rate of ascent of 

 the gas, the balance will be disturbed and the inner cone will 

 descend. I can easily do this by laying an obstacle alotip, the 

 stream of gas, for at the en^l of it there will he no more current 

 than you would find over the stern of a boat anchored in mid 

 stream. I take this obstacle, then, in the form of a glass rod 

 fixed centrally along the current of gas ; I push it up until it 

 touches the tip of the inner cone, and then pull it down again. 

 Vou observe v\ hat has happened. The cone has followed the 

 rod into the tube, and remains attached to it. You will notice, 

 too, that the cone is inverted. That is easily understood. » It 

 is only at the tip of the rod that the current is slowed down ; 

 there only is the rate of ascent of the stream less than the rate of 

 inflammation. The tendency in every other part of the stream 

 is for the cone to go to the top ; hence the inversion. (Fig. i, v..) 



h\ 



n 



^ 



h 



n 



^y 



'^^.— fc 



iA 



A BCD 



Fig. I. — ^Methods of separating tfie two cones of an air coal-gas flame. 



We can get a still more convenient apparatus by a modification 

 of the first method. Instead of choking the bore of the single 

 lube by a constriction, we may use two tubes of different 

 diameter, one sliding wiihin the other This apparatus is shown 

 in Fig. I, C ; a is the wider tube, 6 the narrower one. The two 

 tubes are connected by an india-rubber collar (c), and kept steady 

 by the brass guide (d). The outer tube can be slid up and 

 down the inner one as desired. If we place this apparatus over 

 a Bunsen burner and turn on the gas, we shall have a tolerably 

 rapid upward current in the inner tube, but as roon as the gas 

 emerges into the wider one its velocity will of course diminish. 

 The consequence is that if we now light the gas and gradually 

 increase the air supply, the inner cone will descend until it 

 reaches the orifice of the narrower tube ; but at that point, 

 meeting with the rapid stream, its p-ogress is arrested, and it 

 remains perched on the end of the tube. By sliding the tubes 

 we can thus separate the cones any desired distance, or we can 

 bring their orifices level and restore the original flame. Lastly, 

 we can reverse the experiment, for we can begin with a two- 

 coned flame burning at the protruding end of the nairower tube, 

 and by sliding up the wider tube detach the outer cone and carry 

 it upwards. (Fig. I, D.) 



Having now learnt the relation of flame to explosion, having 



