850 THE COLLIERY GUARDIAN. Apbil 25, 1913. close to the gallery through a 3 in. main. The holder, containing 1,500 ft., was charged with a pentane air mixture maintained under 12 in. water pressure. The pentane air mixture could be made more uniform in composition than coal gas, and its analysis could be more readily performed. Experiments with Mixtures of Pentane and Air. Experi- ment number. Pentane in mixture. Per cent. Velocity of current. Feet per minute. 1 1*51* 610 2 1*59 610 3 1'48 610 4 1*40 1,000 5 l*92f 1,000 6 l'82f 680 7 195 680 8 126 680 9 0 84 280 10 1*40 1,000 11 167 1,000 12 — 300 13 134 300 14 1*31 300 * Note.—The “ lower limit mixture ” of pentane and air contains, according to our experiments, about 1*35 per cent, of pentane. The most explosive mixture contains about 2'5 per cent. f Sample taken whilst magnesia was in suspension. These experiments made it evident that in our appa- ratus the addition of incombustible dust did not confer any increase of inflammability, either on mixtures below the lower limit of inflammation or on explosive mix- tures travelling at various rates. The initial temperature of the current appeared to make little difference, for the same phenomena were observed when the temperature of the gallery was 42 degs. Fahr, (in December) as when the gallery was artificially heated between 90 and 100 degrees Fahr. Moreover, since other experiments had shown that a flame is not propagated more quickly through gas mixtures in presence of these incom- bustible dusts, but, on the contrary, its rate is retarded by them, it did not seem possible to reconcile our observations with those of Abel or to accept his explanations. In our difficulty we sought help from Dr. W. Kellner (formerly assistant-chemist of the WTar Department), who had assisted Sir F. Abel in his experiments. We asked Dr. Kellner whether he could supply a sketch of the apparatus, and, in particular, whether he could tell us how the gas was introduced into the gallery, and how its mixture with the air was effected. Dr. Kellner has very kindly complied with our request, and has given us permission to publish his description of the apparatus. It is proper to state that Dr. Kellner says : “ I have no means of refreshing my memory ; ” but he adds: “ As to the details of construction my memory is quite clear, as the apparatus was built and fitted to my design.” Dr. Kellner writes :— “ Near the inlet end, on the top of the gallery, was placed a drum which contained the dust, and which, on being slowly revolved, allowed the dust to fall in dense clouds into the gallery, where it was carried along by the air current. The drum consisted of two wooden discs about 2 ft. in diameter mounted on a spindle, about 6 in. apart, the space between them along the circumference being covered with fine wire gauze, thus forming a sort of sieve. No dust 'would fall through while the drum was at rest. Description. A well-defined flare, between 2 and 3 feet long, spread forward from the lamp-flame towards the open end of the gallery. After the current had been maintained constant for 1£ minutes, without any marked alteration in the flare, calcined magnesia was admitted and passed as a cloud through the lamp-flame and the flare. The flare became more luminous, but no other change could be detected during a minute's run. (A sample of the gas- mixture taken whilst the magnesia was in suspension showed 1*48 per cent, of pentane.) Stonedust was used instead of magnesia. As before, the flare became more luminous (with a red glow), when the cloud of stonedust reached it. No spreading or elongation of the flare could be detected. (A sample taken whilst the stonedust was in suspension showed 1’59 per cent, of pentane.) A smaller quantity of stonedust was used. The result was similar to experiment 2. (A sample taken whilst the stonedust was in suspension showed 1’46 per cent, of pentane.) Calcined magnesia was used. A " cap ” but no “ flare ” was visible stretching from the elongated lamp-flame. This cap became luminous on introducing the duet- cloud. (A sample taken whilst the magnesia was in suspension showed 1*40 per cent, of pentane.) The gas was turned on until the mixture was highly explosive, and a large “ flare " burnt from the lamp and reached the open end of the gallery. A small quantity of magnesia was then introduced forming a faint cloud in the gallery. No alteration in the flare could be observed; no striking back occurred. An experiment similar to the last. The “ flare ” began to show when the mixture contained 1*56 per cent, of pentane. The gas was gradually increased until a well- defined flame extended nearly to the open end of the gallery. A fine cloud of magnesia was introduced with no visible effect except a slight increase in luminosity. The gallery was heated from the fan-drift onwards, so that the current on reaching the lamp-flame was between 96 degs. and 100 degs. F. The gas was turned on until the flare came out of the open end of the gallery in a roaring flame. A fine cloud of magnesia was introduced and had no visible effect. The flare did not strike back. An experiment to compare the effect of a coaldust cloud with a magnesia cloud. The gas-mixture gave a well-marked “ cap ” on the lamp-flame. When coaldust was introduced the flame struck back as soon as the fine cloud reached the lamp, travelling the whole length of the gallery towards the fan and rushing forwards out of the open end. The mixture gave a well-defined “ cap ” on the lamp flame. On passing a very fine cloud of magnesia the cap became luminous, but no “ flare” was produced. (A sample taken whilst the magnesia was in suspension showed 0'83 per cent, of pentane.) The mixture gave a well-defined " cap.” The cap became luminous when magnesia was introduced. (A sample taken whilst the magnesia was in suspension showed 1*40 per cent, of pentane.) The gallery was heated before the current was passed. The mixture gave a “ flare.” No change could be detected on introducing magnesia. (A sample taken whilst the magnesia was in suspension showed 1*63 per cent, of pentane.) The gas was gradually turned on until the “ flare ” was on the point of striking back. A sample collected at this time showed 1*69 per cent, of pentane. The gas was then adjusted so that the percentage of pentane was 1*57 (as shown by analysis). A strong “ flare ” was produced. No change was observed on introducing magnesia. The gallery was heated so that the current on reaching the lamp-flame was at 90 degs. Fahr. A well-defined “ cap ” was formed. A very fine cloud of magnesia was then introduced, but no change could be observed beyond an increased luminosity of the cap. A well-defined cap was formed. Fine coaldust was now introduced instead of magnesia. When the coaldust cloud reached the lamp a brilliant flame spread forwards and backwards filling the whole gallery. The enclosed sketch may help to explain:— Fig. 1. — a “ The modus operands was as follows :—A current of air of definite velocity, measured by means of an anemometer, was produced by adjusting the steam jet of the blower, the boiler pressure being carefully kept constant. The amount of air passing a given point in a given time was then fixed. “ The gas was contained in a gasholder, near by under definite pressure, varied according to circum- stances, and passed on its way to the gallery through a large stop-cock which could be readily set to deliver a definite volume in a given time. These adjust men tB were made while the inlet pipe b was closed by the lid d. “ When all was ready, the lid (d) was raised and the gas entered the gallery. “You will see we made no special provision for mixing the gases, but trusted that they would get sufficiently mixed by their long passage before reaching the lamp or other means of ignition in the gallery.”* * This description of the apparatus agrees closely with the diagram of the gallery used by Abel for testing safety lamps published in the Final Report of the Royal Commis- sion on Accidents in Mines, Appendix xxiv., Plate 1. Dr. Kellner’s description of the apparatus at once suggested a possible explanation of the differences between our results. In our experiments the gas and air were thoroughly mixed by the Sirocco fan before entering the gallery; in Abel’s experiments the gas entered by a vertical tube through the floor, and being lighter than air would flow upwards as it was drawn along the gallery. In spite of the care which was believed to have been taken to secure uniformity of mixture, it does not seem possible that this uniformity can have existed in Abel’s experiments— since all later work, including our own experiments, show that a flame is not propagated through a uniform mixture of methane and air containing 3*5 per cent, of methane. If complete admixture were not effected? before the gas reached the lamp flame an apparently poorer mixture might ignite than when complete mixing took place ; and, moreover, the dust either by carrying gas down to the flame, or by passing through the flame and then rising to the gas stream—might, ignite by means of convection a gas stream that would not otherwise be inflamed. An inflammation of the unmixed gas stream, once started, might strike back along the upper part of the gallery, and apparently produce a “ general inflammation.” To test this hypothesis we adapted our gallery by bringing in a pipe for delivering coal gas at a point 4ft.. beyond the fan and 16 ft. from the lamp. The pipe was arranged so that its open end could be made flush with the floor of the gallery, or the pipe could be pushed through the opening in the floor so that it could deliver the gas some inches above the floor level. Two-feet beyond the gas pipe an opening in the roof allowed the drum sieve to be attached to the gallery. When coal-gas entered at an opening flush with the- floor and the current was maintained at speeds varying, from 320 to 350 feet per minute, but little difference could be observed whether the lamp-flame was 3 in. or 8 in. from the floor. In each case small streaks of flame detached themselves and passed away with the current in discontinuous lines. The gas-stream, so introduced into the current of air, appears to separate into long threads, which are broken before they reach the lamp- flame, and so form short disjointed filaments of flame darting along in nearly parallel lines. On raising the inlet of the gas pipe a few inches a marked difference was observed in the effect on the lamp flame. An oil lamp being placed on the floor,, with its flame (lj in. high) protected by a screen, the gas was turned on until the flame began to “ tail.” The gas was then kept constant for 2 minutes and no cloud or flare of flame was formed. The lamp was then raised. 5 in., when immediately streaks of flame floated away, and in a few seconds flares attached themselves to the upper part of the gallery, and finally the flame struck. back along the roof. On adjusting the oil lamp so that the flame was 6 in. from the floor, the same stream of air and gas was passed for several minutes without producing any cloud, or flare. Then it was noticed that a burning particle of soot was carried from the flame and produced am attenuated cloud of flame along the gallery. A minute later a similar particle was carried upwards and pro- duced a flare which attached itself to the roof; after a. few seconds the flame struck back through the gallery. Three times in the course of the day the gas was fired backwards by the rising of a red-hot spark from the oiL flame. When a Bunsen burner was substituted for the oil lamp, and adjusted to give a non-luminous flame 2 in. high, in the centre of the gallery a well-marked cap stretched from it. No visible sparks came from the- flame, but occasionally portions of the cap were- detached and floated away. When calcined magnesia was now introduced by turning the drum-sieve, the lamp-flame became luminous with a red glow, and flame- clouds were formed and floated away at intervals of barely two seconds so long as the fine dust was passing. On rotating the drum-sieve more vigorously a denser dust-cloud was formed, and red-hot particles could be followed by the eye as they emerged from the flame carrying an aureole with them. On two occasions a red-hot particle of magnesia rose from the flame and ignited the gas above, whence a flame struck back along the roof. We did not observe that the calcined magnesia, falling into the current from the sieve above, had the • power of depressing the gas stream. We would call attention to the great difference between the flares produced in the uniform and in the non-uniform mixtures. In the well-mixed gases the flares were roaring flames charring wood immediately ; in the streaky mixtures the flares were noiseless, were - visibly discontinuous, and bad hardly any charring , effect on wood.