504 THE COLLIERY GUARDIAN. September 15, 1916. gases with one kind of spark, in others with another, clearly depends on an intimate, but at present unknown, relation between the physical or chemical properties of the molecule, possibly on its structure and the nature or duration of the spark. Secondly, direct evidence that the steps are due to selective absorption is given by the ignition of hydrogen with alternating current break sparks. In a mixture of 25 per cent, of gas in air, and at a pressure approaching 401b. to the sq in., a step is obtained having the well- known form of fig. 1. This type of change occurs in nature wherever there is selective absorption. In spectra it is known experi- mentally as Kundt’s law, and is there represented mathematically by a dispersion formula of the type :— a2 — il 2 4- 4- ____________-L M /*° + _X12 + X2 +’ each of the variable terms making the refractive index p, pass through an infinite value as the wave length X coincides with A4 X2 . . . . In the present experiments with methane, at pressures lower than atmospheric, the least igniting impulsive Hand Break. Amps. 2-0 0-2 0-4 0'6 0-8 Atmospheres. Fig. 5.—Continuous Current, 100 volts, Crossed Platinum Rods. spark is measured by the primary current of the induc- tion coil, which, when broken, causes a single secondary spark just giving ignition. This varies with the pressure (as in fig. 2). The two curves given are with gas prepared by different methods, and are seen to agree singularly well when the difficulties of obtaining the same conditions for every spark and precisely the same gas mixture are known. The fact that the steps are so clear is a proof that the selective action is definite and capable of quan- titative determination. If is is to be admitted that the primary difficulty in ignition is the finding by each combustible molecule of oxygen atoms with which to combine, the steps of fig. 2 may be explained as follow : The collision frequency between combining molecules, or atoms when there is dissociation, can be changed by varying either the total or the partial pressures, that is, either the pressure in the explosion vessel or the percen- tage of combustible gas in the mixture. The latter has previously been shown to give rise to steps, and in fig. 3 they are also found by change of total pressure. The Fig. 6.—Apparatus for Testing the Influence of Rate of Break. Vacuum Pump. Inflammable , Mixture Air reduction of the total pressure by one-half has the same effect on collision frequency as halving the percentage of gas in a mixture at atmospheric pressure. The mixture used in the present work being CH4 4 04, or 9-35 per cent., a change to CH4 4 O3, or 12 per cent., has the same effect on the number of collisions as a change of pressure in the ratio 9*35/12, that is, 0*78. The 9*35 mixture being at atmospheric pressure (and the elec- trical conditions suitable for selective absorption to occur), a step might be expected at 0*78 atmosphere. There is a step at 0*8 atmosphere. The next step, corresponding to CH4 4 O2, or 17*1 per cent, in air, would be at 9*35/17*1, or 0*54 atmosphere. There is a small step at 0*6, and a large one at 0*5. In fig. 3 it is shown that steps arise from change of pressure at successive multiples of an atmosphere. In the same w^ay, they ane to be expected at sub-multiples of an atmosphere, |, J, |..., and there is evidence for each of these. By a well-known theorem an oscillation of any frequency can be replaced by two, of higher and lower frequency approximately equidistant from it. Thus, if by nearness to the strong effect at half pressure the 0-54 step is suppressed, one should appear at about 0-61 atmosphere, and this is found to be the case. At pressures higher than an atmosphere the collisions with oxygen are increased. Thus, CH4 4 O5, or 7*5 per cent., should on the present view give rise to a step at 9-35/7-5, or 1-24 atmosphere, and, as shown in fig. 2, such a step is found. The interpretation of these steps must, from the nature of the case, be tentative. After consideration of every possible cause this appears to be the only one capable of explaining the facts. When the mixtures are compressed before ignition there are also steps in the least igniting current curve which are very suggestive. The pressures at which they occur are 1, 2, 3, 4, and 5 atmospheres, absolute, as shown in fig. 3, the abscissae of which are, however, pressures above atmospheric. From this there can be no doubt that the steps are caused by the collision frequency reaching successive multiples of that at atmospheric pressure. It is, there- fore, a physical effect, and supports the conclusion that, in gaseous explosion, collision mechanics of the simplest kind are of importance. The pressure most favourable to ignition by impulsive sparks is between two and three atmospheres absolute. There -.is reason to believe that the steps continue to much higher pressures, and this is now being investigated. The ignition of methane by impulsive discharge, therefore, proceeds per saltum as the pressure is changed, this being an example of selective absorption of a kind to be considered later. The step is not to be regarded as discontinuous, but to follow the usual laws of dispersion by a resonating system. Ignition by Condenser Discharge. This differs from the above in two important features : (1) The best igniting spark is independent of the pressure from the lower limit at half an atmosphere to just above one atmosphere, and so forms one step. (The value of 0-1 0-2 0-5 0-3 07 06 0-4 0 Magnetic Break! CH 0-2 0-4 06 0-8 10 1-2 14 Amps. 0'8 r Atmospheres. Fig. 7.—Continuous Current, 240 volts, Platinum Points. Amps. 50 the least igniting capacity was 6 mfd. charged to 150 volts, at every pressure down to 0-5 atmosphere below which ignition by condenser sparks failed completely.) (2) As the pressure is raised ignition becomes easier, and there are steps, but now down, as 301b., 601b., and 85 lb. per square inch are approached. In fig. 4 there are steps at 1, 3, 5, and 7 atmospheres, but not in this case at 2, 4, or 6. Apart from their physical interest, these results are of some practical importance. The curve of fig. 3 is for ordinary magneto ignition; those of fig. 4 correspond to the lodge ignition by condenser discharge, the high effec- tiveness of which is well known, especially in poor mixtures highly compressed, which cannot be readily ignited by magneto sparks. Electromagnetic jump sparks and condenser discharge, therefore, proceed in opposite directions, one becoming easier as the pressure is raised, the other more difficult. The former are examples of the relatively slow process of ionisation by collision, which is more difficult at high pressure. Con- denser discharge, on the other hand, is one of the most sudden phenomena in nature, and the spark is, in addition, of high enough temperature to pit platinum freely. Most of the energy of charge is dissipated in the spark. This is equivalent to the combustion of a definite mass of the gas, and the heat of the spark is rapidly com- municated to the gas around. If the mass of the latter heated to ignition temperature is equal to or greater than that whoso combustion would set free as much heat as there is in the spark, self-ignition can proceed. A short calculation from the observed least igniting capacity will ■serve to show that there is more than sufficient energy for the purpose, the remainder being absorbed by the poles or 'radiated, but the ratio of division cannot at present be measured. Compression increases the mass of gas in contact with a spark of a given magnitude. It follows that at the higher pressures a smaller spark, provided that its energy is given to the gas as heat, should cause ignition. On this view the product of least igniting capacity and gas pressure would be constant as a first approximation. The occurrence of steps rising as the pressure is lowered is a modification of this by selective collision. The next two kinds of ignition illustrate a smoother type of charge. Ignition by Continuous Current Circuit Break Sparks. Below atmospheric pressure the least igniting current is found to increase, so that the product of the current and pressure is approximately constant, until at half an atmosphere there is a sudden and most remarkable increase of inflammability. The curve of fig. 5 dips sharply and rises to the lower limit at a third of an atmo- sphere. The simplest explanation of this is that it is a continuation of the selective action which gives rise to the steps at higher pressures. Every gas examined, that is, hydrogen, methane, ethane, propane, carbon monoxide and coal gas, exhibits this effect; in hydrogen it is so great that at the dip the igniting current falls almost to zero, the mixture being those for perfect combustion. If, li owe ver, the break of circuit is made slowly the dip is entirely wiped out and the curve is hyperbolic. o 100 it 35 u. 180u. 55 v. 04 0-8 1-2 0-5 03 0’8 06 0-4 02 09 07 0-1 —1---------------------------------------- 0 10 20 30 40 50 60 70 80 Lb. per square inch. Amps. 1-0 r Fig. 8.—Continuous Current, Electro-magnetic Break. 1-6 Atmospheres. Fig. 9.—Alternating Current, Break Sparks, 200 volts, 36 amps., Platinum Poles. 2’5 2-0 4-5 4-0 3-5 3-0 1-5 ro 0'5 0 0-2 0-4 06 0-8 1-0 1-2 1’4 16 2 0 0-5 20 1-5 1’0 Amps. 25 10 20 30 40 50 60 70 80 90 100 Lb. pei* square inch. Fig. 10.—Alternating Current, 100 volts, 80 amps. In order to examine the influence of rate of break, which could not be done with certainty with hand, the electromagnetic arrangement of fig. 6 was used. The current to be broken was led through a brass tube con- taining an iron plunger carrying at its lower end a platinum head. The explosion tube being set at 45 degs. the rod rested on the lower platinum point, making con- tact, and was drawn up by a separately excited solenoid, the current of which could be varied to give any desired speed of break. When the break was relatively slow the lowest of the three curves of fig. 7 was obtained; increas- ing the speed of break caused the least igniting current to take the form of the upper curves, each of these having the same rate of break throughout. There is, therefore, a critical relation between the collision frequency and the spark duration necessary for rhe observance of this effect. Above atmospheric pressure there is the singular result that the least ignit- ing current is almost independent of pressure, falling slightly at -the highest pressures. This is obtained at all voltages from 50 to 180, as shown in fig. 8. Continuous-current spark ignition is, therefore, midway between impulsive and condenser ignition; that is, it is not a simple energy effect, for this would cause a falling curve with or without steps, nor does it work by ionisation alone. It would appear to be a very fair mean between ionic and thermal ignition.