20 THE COLLIERY GUARDIAN. July 3, 1914. molecules may be called the explosive unit, and the number of such in unit volume is proportional to the percentage strength of the mixture. The time of explo- sion T depends on the number of such units per cubic centimetre, and on the mass per unit volume of inert gas which transmits energy without raising it. The minimum time of explosion is to be found in that mix- ture in which the retarding influence of excess molecules which cannot enter into full combination is balanced by the accelerating influence of the energy set free by their partial combustion. If from this point the rate of explosion falls in proportion to the increase in the mass of inert gas, and the number of explosive units is diminished in the same proportion, it follows that T will vary as the square of the change. That is true as long as the units are so close that the space between them is not great compared with their volume after explosion, but forms, as it were, a shell around each. As the limits of inflammation are approached, they are so far apart that the inert gas forms a sphere around each unit and the explosive unit is small compared with the volume of inert gas. The change of inert mass then varies as the square of the distance between the units and the time of explosion as the cube of the change of percen- tage. (The rate of change of the volume of a sphere varies as the square of its radius.) The period known as the time of explosion would thus appear to be capable of a simple mechanical explanation. The above argu- eo so 4-o 20 AMPERES <}or Fig. 3. Influence of cir cuit voltage on least igniting current in different percentage mixtures of methane and air. Iron poles. Continuous current. Fig. 5. 9 5 per cent, mixture of methane and air, most difficult to ignite at 100 periods per second. Iron poles. Alternating current. ment may explain why simple gases have higher limits of inflammability because of the smaller volume and mass of the explosive unit. The shape of the curves connecting alternating igniting current and percentage strength of mixture has probably the same origin. They are, in the more regular cases—the paraffins for example —parabolic at the minimum and cubic towards the limits. Ignition by Alternating Currents. The curves expressing the experimental relation between voltage and igniting currents, and between cur- rent at different frequencies and percentage of gas, are essentially different in type from those given above for direct currents. Consider first the ignition of the mix- ture of methane and air just giving complete combus- tion. At all voltages much larger currents are required to produce ignition than with direct current. Below 100 volts the currents are very large, from 200 to 600 volts are nearly constant, and at some point above this rapidly fall, to about 0*2 ampere at 2000 volts. Thus, in places where electricity is used in the possible presence of inflammable mixtures, as, for example, in coal mines, alternating currents are much safer than continuous, where open sparking may occur, and the higher the frequency (up to 100) the greater the safety. The igniting current at 200 volts with continuous voltage is 0’5 ampere; with alternating current at 200 volts and 100 periods per second the least igniting current is 20 ampdres. This is the root-mean-square value and is given in the curves, but the ignitions are always obtained from sparks giving the maximum flash, that is when the circuit is broken at the crest value of 28 amperes. The arrangement of the resistance and inductance of the circuit was the same as that used with continuous currents, so that the energy of any one spark is proportional to the current, not to its square. The alternating igniting current at this voltage is 56 times greater than the least continuous current at the same voltage, and the difference shows beyond doubt that the energy of the continuous current spark is not trans- mitted thermally to the gas, for in that case the energies of ignition in the two cases should be the same. The shape of the curve of fig. 4, similar to the isothermals of vapours below the critical temperature, suggests that one of the factors in the electric circuit may become critical. It is clear that it can only be a time effect, and that time must play an important part in the phenomena of ignition. The horizontal part of the curve disappears as shown, both at high and low frequencies’, for at the highest frequencies it is practically a continuous arc, and at the lower it approximates to the direct current hyperbolic type. At moderate frequencies the arc in reversing strikes across the ionised vapour in the break gap and may persist for several periods. If the break occurs in less than 0’05 second—the minimum time of ignition—the critical change of energy in the gas may never pass into the explosion stage. There should then be some period at which the remarkable influence of frequency reaches a maximum. The highest frequency in the present case at which it was possible to make measurements was 100 a second, and it is seen from fig. 5 that the maximum is being here approached. At frequencies of several hundred periods per second igni- tion will no doubt be obtained by smaller currents. The curves are of the usual trial and error type and show that the approach to the maximum is a statistical one, but the effect of raising the frequency is to avoid, as far as possible, the ignition of the gas and the curves would be symmetrical about the point of maximum Fig. 4. Influence of frequency in igniting current in a 9’5 per cent, mixture of methane and air. Iron poles. Alternating current. difficulty of ignition. If the ignition of a gaseous mixture is a thermal process alone the frequency of the current should have no effect as distinct from the energy of the spark, but it has, in fact, an exceedingly great influence Table III. Gas. Formula. Formula weight. Least current, I. P min. N. 1 x pm. I N N x pm. Nc x pm. Methane ch4 16 2'7 10'2 9 27'5 0'30 91'8 51'0 Ethane c2h6 30 4’0 6'9 15 28'3 0'27 103 55'2 Propane c3h8 44 6’45 4'8 21 30'6 0'30 100 52'2 Butane c4h10 58 8'8 3'75 27 32'6 0'32 100 51'8 Pentane c5h12 72 9'2 3'05 33 26'7 0'28 96 51'8 Mean values 29’1 0’294 98’3 52'4 on the magnitude of the current required to be broken, an influence which can only be accounted for by the rapid alternation of the voltage between the poles pre- venting the free escape of ions from the spark. There is again one frequency at which the horizontal portion of the curve is longest. This is when the fre- quency is about 50, and is much more marked when the poles are nickel. The time of duration depends on the inductance of the circuit. In all these alternating current experiments the circuit was non-inductive, that is the power-factor was never less than 0’95, and the currents had a pure sine wave form. Relation of Igniting Current to Percentage of Gai in Mixture. Paraffin Series.—Working with alternating current more trials are required than with continuous currents to obtain the least igniting current, for the poles may be separated at a value approaching zero, or at a crest value of the current. In the one case there is no arc, in the other it is at a maximum. The final critical igniting current is, however, little affected by the uncer- tainty of break. The curves of fig. 6 obtained for the paraffin gases are remarkably uniform in type and fit closely into a regular scheme. Their most striking feature is that the higher hydrocarbons are the most difficult to ignite by break- sparks, a fact which must profoundly affect our measure of inflammability. They can be ignited in weaker mix- tures, but not so readily. Pentane takes three times more energy to ignite by alternating currents than me- thane. Another interesting feature is that the curves are as nearly as possible symmetrical between their upper and lower limits of inflammability, the maximum inflammability being at the mean of the extremes. In the family of curves for continuous current in fig. 2 the minimum currents are the same throughout, in that for alternating currents in fig. 6 the product of the minimum current and the percentage at which it is found is the same for every gas. The heat of combustion of pentane is four times greater than that of methane and it can- not be suggested that the gas is more difficult to ignite because of its greater heat of combustion, for the alter- nating igniting currents of ethyl and methyl alcohol are the same. Atomic Nature of Break-Spark Ignition. But if it is true that combustion is the formation into more stable compounds of molecules electrically unsaturated and that no combination takes place until they are ionised either wholly or in part, it would follow that the igniting current should be proportional, if the process is molecular, to the number of molecules of com- bustible gas in the mixture, or if it is atomic, to the number of atoms of one or more of the constituent elements. From Table III. it will be clear that, as in the continuous current case the igniting current was most nearly proportional to the number of hydrogen atoms in a molecule of the combustible gas, it is here propor- tional to the total number of atoms in each molecule of combustible gas or in each explosive unit. It follows AMPERES 18 r— is 12 15 IO PENTANt BUTANE- PROPANE ETHANE- 1 ethane 17 16 Fig. 6. Least igniting currents for paraffin gases in air. Alternating current. 200 volts, 36 Iron poles. IO II 12 13 14 15 l< PER CENT OF GAS IN AIR that the product of the number of molecules of com- bustible gas in the most inflammable mixture and the number of atoms in each of its molecules should also be constant, and this is seen to be the case in the last column, Table III., N is the number of atoms in explo- sive unit, Nc, the number of atoms in a molecule of combustible gas, pm, the percentage of gas in the most inflammable mixture, and I the lowest point of the curves :— For the paraffins, continuous current ignition above the point of perfect combustion varies as the percentage of combustible gas, alternating current ignition as the square of excess of either of the combining gases on each side of the point of maximum inflammability. There can be no doubt that the shape of a molecule must in some way affect its resistance to being broken up, so that great uniformity can scarcely be expected in numerical results of this kind. The methane molecule CH4 must differ from the higher homologues of the form CH3—CHa—CHa—CH,—and the results with ethane, propane, and butane agree well together in all cases. The explanation of the difference in type of the benzene and carbon-disulphide ignitions may be that in the latter case there is explosion (without intermediate stages), and that in the former, however ignited, there is first very full dissociation of the molecules around the spark. Cause of Difference between Continuous and Alternating Current Ignition when the Gas Percentage is Varied. The continuous current curves derive their peculiar form from the fact that the igniting current increases in proportion to the number of molecules of combustible gas in unit volume of the mixture. The alternating current curves in general are entirely different in shape, and are nearly symmetrical within the limits. The mini- mum current, either continuous or alternating, must clearly occur at that mixture in which self-ignition is most easily started, and since with continuous current ignition the forcing action of the spark quite masks any changes in the gas in contact with it, it is interesting to examine the conditions most favourable to alternating current ignition. All the paraffin gases are most easily ignited by altern- ating arcs in mixtures giving combustion neither to CO nor CO2, but to the mean of these, and these again agree with the means of the upper and lower limits. It