electrons with molecules, which are known to become appreciable in some gases under similar conditions; for example, in H2O the current would increase by about 5 per cent. between successive slits with this value of Z/p and a pressure of 1 mm. So no deductions are made from the observations with Z/p=32. 5. Conclusions. The probability h of attachment at a collision may be estimated by means of the formula if the mean free path L, of an electron at 1 mm. pressure of gas, be deduced from measurements of the viscosity of NH3. Accordingly, from Roth and Scheel's Konstanten der Atomphysik,' L=2 x 10-2 cm. ; and since from figs. 8 and 9 with Z/p=12, k=43, and a/p=045, This method, however, is not a reliable one by which to examine the dependence of h on the energy k of the electron, *Deduced from the relations a=hu/1W and W=815 Zel/mu. as it is known that L itself may vary considerably with ; so the following argument is to be preferred : If E denote the energy (in volts) lost by an electron at a collision with a molecule, then for a given gas E is a function of k alone. The number v of collisions made by an electron which travels 1 centimetre along the direction of the electric field, in the steady state of motion, is such that the work Z (volts) done by the field is equal to the energy lost in these collisions. The probability of attachment in this distance is represented by a and also by vh. vh=a; h/E=a/Z=(a/p)/(Z/p). The values of h/E for different values of k may then be determined from the curves in figs. 8 and 9, and represented by a curve as in fig. 10. This shows that as k increases from 5.5 to 64 the ratio h/E also steadily increases from 6 × 10−3 to 8·1 × 10−3, i. e. by a factor of about 14. It is easy to show that E=7x 10-16 W2 volts where W is the velocity of the stream of electrons in the direction of Z; and as none of the many experiments made to date in different gases † has revealed an instance where W decreases as increases, it may be concluded generally that E never diminishes as k increases ‡. It then follows that in NH3 increases by a factor not smaller than 14 when k varies from 5.5 to 64. Some information concerning the variation of the mean free path L may be obtained in a similar way from a curve whose ordinates represent L2/E, since it is easy to deduce from the above expressions for h and h/E that L2/E=k/11(Z/p), and so to determine this curve (fig. 10) by means of fig. 8. This curve indicates that, as k increases from 5.5 to 45, the ratio L/E increases from 2.34 to 8.2, and as E is not diminishing, it follows that L increases in this range by a factor of at least 3.5. The conclusion that h may vary considerably with k in NH, and in air § makes it certain that the method used by L. B. Loeb and H. B. Wahlin is liable in general to give erroneous results, for it depends on the assumption that his a constant characteristic of the gas alone. There are other weaknesses in this method, some of which have already been pointed out T, so in referring to the table on p. 513 of his book, in which are given the values of n(1/h) obtained with this method, he states: "It is questionable whether the values are accurate in more than order of magnitude. They do differ, however, so widely in order of magnitude that even these crude early results give a good idea of n." Further on he adds: While these results are not of more than qualitative value. . . ." We are inclined to doubt whether they even "give a good idea of n" on comparing Wahlin's value 10-8 for h in NH with our lowest value, which is about 2 × 10−5; these are by * J. S. Townsend and V. A. Bailey, Phil. Mag., Nov. 1922, p. 1045. † H2, O2, N2, He, A, Ne, CO, CO2, NO, N2O, CH, C5H12 The experiments of Townsend and Glasson, which might indicate the contrary, correspond to very different conditions, where the energies cause intense ionization by collisions. S V. A. Bailey, loc. cit. 6 L. B. Loeb, Kinetic Theory of Gases,' p. 510. no means of the same order of magnitude. To explain this discrepancy by assuming that our gas contained more of potent impurities than his, requires that ours should have contained a proportion of these at least 1000 times greater than did Wahlin's; there is, however, no reason to suppose that our samples of NH, were less pure than his. On the other hand, if it be argued that the disagreement is attributable to the difference in the energies of the electrons, then it has to be admitted that when the mean velocity of the electrons changes from 1.1 x 107 to about 2.6 x 107 cm./sec., the probability of attachment h increases by a factor of at least 1000; a change of such magnitude cannot be ignored, even if only "qualitative" results are sought. Since Loeb admits (p. 513) the greater reliability of the methods we have used, it would appear that the results obtained with his method, and set out in his table, may be quite misleading. E. M. Wellish has expressed the view that "an electron cannot effect a permanent union with an uncharged molecule to form a negative ion unless the relative velocity at collision exceed a critical value characteristic of the molecule concerned. . . . It is probable that the circumstances of an encounter as well as the relative velocity will determine the effectiveness of a collision so that only a fraction of these impacts will result in the formation of ions." We are unable to see the necessity for his assumption of a minimum critical velocity, for the experimental facts he adduced in support of it can be understood from our point of view without requiring any additional hypothesis. The NH4Cl and CaO used was very kindly prepared for us by Mr. G. J. Burrows, of the Department of Chemistry. We are also much indebted to the Colonial Sugar Refining Co., Ltd., for the supply of carbon-dioxide snow, and to the Commonwealth Oxygen & Accessories Co., Ltd., for the supply of liquid air, both without cost to the University. It is to be hoped that their example will find imitators in Australia, where the policy adopted with striking success by the industries of other countries, of assisting research in pure and applied science, is still somewhat disregarded. In conclusion we desire to record our appreciation of the skill and resource shown by Messrs. G. C. Barnes and H. Taylor in the construction of the apparatus used in this work. *E. M. Wellish, Am. Journ. Sci. xliv. p. 26 (July 1917). Phil. Mag. S. 7. Vol. 6. No. 40. Dec. 1928. 4 B CVII. Spark Ignition. By E. TAYLOR JONES, D.Sc., Professor of Natural Philosophy in the University of Glasgow*. [Plates XXI. & XXII.] THE substance of the following communication was contained in a lecture to Section A of the British Association at Glasgow on September 10th, 1928. chief topics discussed are the nature of the action of an electric spark in producing ignition of an inflammable gaseous mixture, and the conditions which determine whether a spark will or will not ignite a given mixture. One of the earliest experimental results published on this subject was the observation by Thornton † that the heat dissipated in a spark just sufficient to cause ignition is less if the spark is produced by the discharge of a condenser than if it is produced (as in "low tension" or "inductance" sparks) by separating the electrodes from contact so as to interrupt a current in an inductive circuit. An explanation of this result, based on the view that spark ignition depends upon the volume of the gas which the spark can by its own heat raise to the ignition temperature ‡, was suggested by Taylor Jones, Morgan, and Wheeler §. A condenser spark of very short length between metal points being regarded as an instantaneous point source of heat in a uniform medium, the temperature in its neighbourhood is represented by Fourier's expression * Communicated by the Author. + Phil. Mag., November 1914. See Wheeler, Trans. Chem. Soc. cxvii. p. 903 (1920); also the Third Report of the Explosions in Mines Committee of the Home Office,' 1913. The argument for this view may be stated as follows:-If we suppose that a small spherical volume of the gas is heated by the spark to the ignition temperature, the gas within this volume is burnt, and its temperature is raised further by the heat resulting from the chemical action. At the surface of the sphere there will, therefore, be a large temperature gradient and rapid loss of heat by conduction. The rate of cooling of the sphere due to this cause is proportional to the ratio of its surface to its volume, and is very great if the sphere is very small. Consequently the small flame started in the sphere will soon become extinguished by the conduction from its surface, and will therefore fail to spread throughout the gas, unless the volume of the sphere exceeds a certain minimum value. § Phil. Mag., February 1922. |