906 THE COLLIERY GUARDIAN. May 2, 1913. factors the first must undoubtedly exercise the pre- ponderating influence. The effect of the other factors named is small in comparison, and they operate in contrary directions. The Experiments.—The method of experiment adopted has been that of passing an electric spark from an induction coil between terminals placed at the centre of the mixture of inflammable gas and air contained in a spherical vessel of about 2 litres capacity. The mixtures were prepared by adding measured volumes of the pure combustible gases to air, free from carbon dioxide, con-^ tained in graduated glass gas-holders over concentrated sulphuric acid. The manner of determining the lower-limit mixtures was that of “ trial and error.” For example, a mixture of methane and air containing 6*1 per cent, of methane having been tried and found to propagate inflammation on the passage of an electric spark, a second mixture was prepared containing 5’9 per cent, of methane. This also propagated flame. The percentage of methane was therefore further reduced by 010 in a new mixture, and so on until two mixtures were obtained, differing in their content of methane by 0’20 per cent., one of which enabled flame to be propagated, whilst the other did not. The lower-limit mixture was taken to be that containing the mean percentage of methane contained in these two mixtures. The lower-limit mixture could be distinguished with certainty from that containing just insufficient combustible gas. For the momentary passage of the electric spark sufficed to promote the inflamma- tion of all the gas contained in the globe in the former case, and on further sparking no signs of combustion could be observed. Whilst in the latter case, although the flame of the burning gas might appear to travel nearly through the whole mixture on the first passage of the spark, and some doubt might exist as to whether it had not, in fact, travelled throughout; on causing the spark to pass a second time a “cap” appeared above it, showing that the mixture still contained combustible gas, and this cap remained whilst passage of the spark was continued, growing gradually smaller in size, until all the gas had been burnt. As a further check, samples of the residual gases were withdrawn from the globe after experiments with mixtures near the lower limit and complete analyses made. Results of Experiments. — Before considering the influence of the size and nature of the source of heat used to cause ignition, or the effect of variations in the initial temperature and pressure of the mixtures, the details of the experiments made with the different gases when these conditions were constant are given. They may be summarised as follow:— Lower-limit mixture Gas. contains between Per cent. Per cent. Methane 550 and 5 70 Ethane 300 „ 320 Propane 2’15 , 2’20 n-Butane 1’60 , 170 n-Pentane 135 , 1’40 &o-Pentane 1’30 , 1’35 The results obtained for methane are in good agree- ment with those obtained by Coquillon, and by Mallard and Le Chatelier, but are quite at variance with those obtained by Teclu. The effect of the spark on mixtures containing as much as 4’50 per cent, of methane makes it impossible to believe, as Teclu states, that mixtures containing 3’50 or 3 67 per cent, of methane can pro- pagate flame under normal conditions as regards the initial temperature and pressure of the mixtures, and it must be concluded that either his method of analysis or his method of preparing the mixtures was faulty. Altogether it would appear that the figure given, 5’60, represents the smallest possible percentage of methane in a methane and air mixture that can enable self-pro- pagation of flame to take place when the initial tem- perature and pressure of the mixture are atmospheric. The Influence of the Size and Intensity of the Initial Source of Heat.—In discussing the influence of the volume of the initial source of heat on the propaga- tion of flame in explosive mixtures of gases, Mallard and Le Chatelier put forward considerations to the effect that a small electric spark of high temperature is insufficient to cause the inflammation of weak mixtures, which can, however, readily be ignited by the flame of a lamp of less intensity than the electric spark, but of larger volume. It would appear that above a certain limit no increase in the size and intensity of the source of heat can affect the percentage of combustible gas that must be present to form a lower-limit mixture. The only effect that can follow an increase in the magnitude of the source of heat introduced into a mixture containing less combustible gas than is required to form a lower-limit mixture is an increase in the size of the aureole surrounding it. A number of experiments were made with mixtures of methane and air and ethane and air, in which the length of the spark from the induction coil was either 10,15, 30, or 50 mm. The longest spark was found to be no more effective than the shortest, and no great difference could be observed in the height of the cap formed. Finally, by employing a Wehnelt electrolytic interrupter in circuit with the primary of the induction coil, using a current of 110 volts, a “ flaming ” spark was obtained, 30 mm. long, of large volume and great intensity. Comparative experiments were made with this source of ignition with most of the mixtures enumerated. The results were, in general, as have been anticipated. The percentage of combustible gas required to form a lower-limit mixture suffered no reduction in consequence of the more violent means of ignition, but the size of the aureole surrounding the spark was greatly increased. Moreover, with mixtures that were capable of propagating flame, the flame appeared to travel more rapidly, and a greater pressure was momentarily pro- duced, than when the ordinary spark was used to cause ignition. It has been suggested that the reason for this increased rapidity of combustion may lie in the great ionising power of the Wehnelt spark, and that, possibly, ionisation from an external source and ignition by means of an ordinary spark might produce similar effects to the combined ignition and ionisation of the Wehnelt spark. This point will be tested. The Influence of Small Variations in the Initial Temperature and Pressure of the Mixtures.—The experi- ments that have been made on the influence of the initial temperature and pressure of the mixtures in determining their lower limit values have been confined to such small variations as may occur in coalmining practice. That is to say, the highest temperatures and pressures employed have been but little more than can exist in the deepest mine at present worked. Such small variations in temperature and pressure do not appear to have any measurable influence on the lower- limit value of methane and air mixtures. With the initial temperature as high as 35 degs. Cent., and the pressure 800 mm., no self-propagation of flame could be observed with mixtures containing less than 5’60 per cent, of methane. II. The Sampling and Analysis of Mine Gases. Sampling.—No one method can be given as the best, in all circumstances, for sampling mine gases; much depends upon the locality from which the sample is to be taken, and upon the nature of the gases to be sampled. For example, if it is desired to estimate the amount of carbon dioxide in a return airway, sampling over water is inadmissible; whereas if a sample from a borehole is being taken, displacement of a liquid by the gas (which may be issuing only in small quantity) is probably the most convenient, or the only, method employable, and, since the gas from boreholes is usually heavily charged with water, there is not much objection to the use of water as the liquid to be displaced. Fig. 1. A convenient method of taking samples of the air in a roadway or other open space is that advocated by Dr. Haldane of sucking the air to be sampled through a small narrow-mouth bottle by means of a rubber tube reaching to the bottom. Undoubtedly the most accurate, and in some respects the most convenient, method of sampling in all cases is to allow the gas to enter a vessel from which the air has been previously thoroughly exhausted by a vacuum pump. The form of vessel most suitable for collecting the sample according to this method is shown diagram- matically in fig. 1. It is of glass, and holds about 75 cubic centimetres. In the absence of suitable means of exhausting the tubes, they can be filled with liquid and emptied where the sample is to be taken, air in any connecting pipe being swept out through the three- way tap, as before. Mercury is the best liquid to employ, but, failing this, a mixture of pure glycerine and distilled water, in equal proportions by volume, is suitable. Even this liquid is, however, not suitable for use when it is required to determine accurately small percentages of carbon dioxide. The sample having been obtained and the sampling tubes brought to the laboratory, their contents should at once be transferred to storage tubes for analysis. The tubes should not be more than four-fifths filled with gas. Analysis.—The form of apparatus that has been employed in the laboratory at Eskmeals by Dr. Wheeler, for the analysis of mine air, is shown diagrammatically in fig. 2. It comprises essentially three parts, namely:— (1) A water-jacketed combination of measuring (A) and pressure (B) tubes, communicating, through a glass tap 0, with a mercury reservoir D ;'(2) an absorption vessel F standing over mercury in a trough ; (3) an explosion tube E, fitted with firing wires and connected with a mercury reservoir. The principle of measurement employed is that of the measurement of the pressure of the gas at constant volume. For this purpose the gas is brought to a certain “ constant-volume ” mark on the measuring tube A, and its pressure (in millimetres of mercury) read off on the pressure tube B. Instead of using a number of large absorption vessels each containing a particular reagent (which is used unchanged many times Explosion Sami. Fig. 2. over in successive analyses), all the absorptions are carried out over mercury in the one absorption vessel F, in each case with a comparatively small volume of the particular reagent, which is always used fresh and is at once discarded after use. Details are given of the mode of operation, the preparation of reagents, &c. Ill* Effect of Inert Dust upon Explosions of Gas and Air in a Small Tube. A glass tube, If in. in internal diameter, and 19 ft. long, was stopped up at one end by an indiarubber bung, through which a mixture of gas and air was admitted so as to fill the tube. A f-inch spark from an induction coil served as means of ignition, the spark gap being placed 3 in. from the closed end of the tube. After a few trials explosions were obtained with great regularity, and, so far as could be judged by the appear- ance and sound, of uniform strength. Dry stonedust was used of specific gravity 2 6, which had been passed through a 200-mesh-to-the-inch sieve. Fluedust of specific gravity 2’4 was also employed without previous screening. When 3oz. of dust (about four teaspoonfuls) was distributed in a thin layer on the bottom of the tube extending from the closed end to about 3 ft., the explosion went on unchecked to the end of the tube. When, however, the dust extended to 4f ft. from the closed end, the explosion stopped at 8| ft. When the dust extended from the closed end to a distance of 10 ft., the explosion stopped at 10J ft. Next small dust zones, consisting of about 10 grammes of dust spread along a length of 8 in., were placed at various points along the tube. The explosions were then arrested, as follow:— Distance of dust Distance from zone from closed end closed end of tube to of the tube. which the flame went. Ft. in. Ft. in. 6 6 7 4 10 0 13 0 11 6 13 10 13 0 14 3 14 6 15 6 16 0 16 6 After each experiment five or six explosions were made without putting in fresh dust. Usually for about three of these the dust remaining in the tube (some of which was carried back by the back suction which followed the explosion) was enough to stop the flame as