152 LEWIS STUDIES IN CATALYSIS. PART X. XX1.-Studies in Catalysis. Part X . The Applicah-ility of the Radiation Hypothesis to Heterogeneous React ions. By WILLIAM CUDMORE MCCULLAGH LEWIS. IN the previous papers of this series the radiation hypothesis has been applied exclusively to reactions in homogeneous systems. A mode of applying the hypothesis t o reactlions in heterogeneous systems including heterogeneous catalysis having suggested itself to the author more than two years ago it may notl be out of place to indicate1 it briefly here. In attempting to elucidate the mechanism of any chemical or physical process two complementary methods of treatment may be employed. I n the first' the process is considered from the point of view of the material or molecular changes involved; in the second from the point of view of the concomitant or precedent energy exchanges.The radiation hypothesis belongs t o the second method of treatment. The two met'hods are not distinct in the sense t'hat the results obtained in one often furnish a clue to the solution of a difficulty met with in the other. It is necessary, however to possess in the first place some information regarding the most probable material mechanism of the process considered before introducing considerations based on the energy exchanges involved. I n the case of heterogeneous reactlions and catalysis, Langmuir's theory of the spatial distribution of molecules and atoms a t the interface between two phases will be adopted as a basis for the material changes occurring the energy changes being then dealt with from the point of view of the radiation hypothesis.Langmuir's theory (compare J . A4 mer. Ghem. Soc. 1916 38, 2221) is essentially an extension of the work of the Braggs on crystal structure. The surface of a solid is regarded as a checker-board on which atoms or molecules of gases may be condensed by being united to certain atoms in the surface itself. This adsorption effect is ascribed directly to valency in some cases the surface being almost entirely covered or saturated in others only a small fraction of the surface being thus occupied. According to Langmuir this surface layer does not consist of several layers of molecules or atoms in which the density varies continuously. Instead the change from solid to homogeneous gas is abrupt'.This is based on the idea that it is only a layer one molecule or atom in thicknes LEWIS STUDIES IN CATALYSIS. PART X. 183 which would be held sufficiently firmly to the surface especially at the moderately high temperatures a t which heterogeneous reactions proceed in general with measurable velocity. Langmuir’s experi-mental When a gas molecule strikes a surface it is in general condensed. The rate a t which it evaporates depends on the chemical or specific nature Df the molecule and of the layer of atQms in the surface of the solid. Thus nitrogen in which the atoms are already very com-pletely saturated possesses only a feeble external field of force, and in the moZeci(Zar forin therefore will be only slightly adsorbed. Langmuir has found that hydrogen in the atomic form pro-duced by heating a wire in dry hydrogen a t very low pressures has a remarkable tendency to be adsorbed this being regarded as due to the unsaturated affinity of the hydrogen atom.Langmuir has calculated that in this case the adsorbed layer of gas is just one atom in thickness. Oxygen is likewise easily adsorbed by metallic (tungsten) filaments. This adsorbed layer is exceedingly stable, and is evidently distinct from the formation of the compound WO, which volatilises easily in comparison. On Langmuir’s view, the oxygen is retained on the surface in the atomic form. A mole-cule or atom which is strongly adsorbed is capable of displacing one which is feebly adsorbed. Hence addition of a strongly adsorbed gas-which in certain cases may be the resultant of the reaction-may cover the surface of a solid more or less completely, the surface being thereby “p~isoned” with respect to a reaction in which the reactants are only feebly adsorbed.Langmuir has given several instances of such effectls. The essential point for our present purpoee is the dissociation partial o r complete which many substances undergo into tthe atomic statel on being condensed on surfaces the cause of such dissociation being the localised valencies or lines of force which hold the atoms of the condensed substance t o certain atoms of the surface of the solid.* We have now to see how far the radiation hypothesis may assist in extending this view of the mechanism of the process. I n general the velocity constants of heterogeneous reactions are characterised by possessing smaller temperature coefficients than those which are possessed by reactions in homogeneous systems.This means on the basis of the considerations developed in earlier papers that the critical increment in the het’erogeneous process is * The catalytic effect of traces of moisture in the activation of molecules and atoms and therefore possibly of surfaces is not considered in the present paper.. The facts hitherto recorded point to the conclusion that water is effective where ions are required t o enable the reaction to proceed. results support this view in many cases. So much for the nature of the material changes involved 184 LEWIS STUDlES IN UATBLYSIS. PART X. less than it would be for the same process occurring in the homo-geneous system.This in fact appears to be the basis of the accelerating or catalyt'ic effect of a given surface as viewed from the energy required to effect the chemical change. It has already been shown that the reactivity of a substance depends on the magnitude of its critical increment that is the amount of energy which must be added per molecule or per gram-molecule in excess of the average energy content in order to bring the molecule into the active state. The higher the critical increment the smaller is the reactivity or rate of reaction of the substance. This increment is taken account of by the exponential term which appears in the velocity expression developed in previous papers. The term referred to is e - E / R T where E is the critical increment per gram-molecule i? the gas constant per g-ram-molecule and T the absolute temperature.It is this quantity that governs tthe magnitude of the temperature coefficient of a reaction and as is evident the greater tjhe value of E the greater is the temperature coefficient. Let us suppose that a given reaction occurs in a homogeneous system the sum of the critical increments of the reactants being El whilst the sum of the critical increments for the same reaction when a heterogeneous catalyst is present is 3,. Then E ) E . The ratio of the velocity constant in the presence of the catalyst to that when the catalyst is absent is given essentially by the ratio e-Ez:RT/e-Ei/RT or e(Ei-EzYR2'. This is in general a large posi-tive quantity; it may be referred to as the catalytic factor.Let us suppose that the process considered involves th6 dissociation of a gaseous molecule. If this occurs in the homogeneous phase the critical increment is large of the order of 50,000 to 100,000 calories per gram-molecule. This energy has to be supplied by absorption of the radiation present in the system and the greater the amount of energy required the higher must be the temperature in order that a sufficient number of quanta of high frequency may be avail-able. If on the other hand a catalyst is present which is capable of condensing or adsorbing the gas in the atomic fom then the energy required is essentially that of sublimation or de-sorption of the atomic resultants from the surface diminished by the energy of adsorption or condensation of the molecular reactant.Such effects are in general small of the order 5000 t o 10,000 calories per gram-molecule. Hence in this case the catalytic factor would be e(W000-j030)lRT which for the temperature T = 1000° would corre-spond with fe22.5 or 1010 approximately. Itl is evident that the effect which we have been considering is of very great magnitude, and to this extent is in agreement with the known high efficienc LEWIS STUDIES IN CATALYSIS. PART X . 185 of heterogeneous catalysts. From the point of view of the energy changes involved therefore t’he action of a catalyst is to be ascribed t o the substitution of relatively small energy terms of the nature of de-sorption or sublimation effects in place of true critical energies of activation or dissociation.In general the problem is not so simple as the case just considered. Frequently more than one reactant is involved and in some cases partial activation or polar-isation of one or more of the reactants may be effected without such reactant coming into direct contact with the surface of the solid itself. This will naturally occur when the surface is already covered by a reactantl which possesses high capacity of adsorption. I n general however the function of the catalyst is to bring a t least ane of the reactants into the active form which would other-wise only be attained in the homogeneous phase by exceedingly high temperature conditions. The possibilities which present them-selves will be rendered somewhat clearer by a preliminary examina-tion of one or tlwo actual cases.The Reaction between Oxyyeiz und Sulphur. We shall first of all consider the reaction S + O,= SO, as occur-ring in the homogeneous gaseous state. Since the resultant contains two atoms of oxygen the process does not require complete dissociation of the oxygen molecule a a preliminary step. Instead a partial activation or polarisation of the oxygen molecule is sufficient. A value for this quantity may be obtained from a consideration of the thermal decomposition of ozone which has been measured by Chapman and Jones (T. 1910, 97 2463) the reaction being shown to be bimolecular. The details of the calculatioIf will be given in a subsequent paper but it may be stated here that the critical increment of ozme per gram-mole-cub obtained from Chapman and Jones’s results is 10,690 cals.Further the heat evolved a t constant volume when two gram-molecules of ozone decompose into three gram-molecules of oxygen has been determined with accuracy by Kailan and Jahn (Zeitsch. nnorg. Chem. 1910 68 243) the value being 69,000 cals. Apply-ing the quantum expression (compare T.; 1917 111 1086) to the process 20 4 30, we obtain 69,000 = 3 3 ’ 0 - 21,380 whence E’o =30,127 cals. or 30,000 cals. in round numbers. The symbol E’ denotes the critical increment per gram-molecule required for the partial activation o r polarisation of oxygen which will permit three molecules thus activated to react to form two molecules of ozone. A molecule possesses in general different degrees or stages of activation and this may not be the one required in the case o 186 LEWIS STUDIES IN CATALYSIS.PART X. the union of oxygen with sulphur. All partial activations are, however small quantities compared with the activation required to cause complete dissociation of a molecule. So far as order of magnitude is concerned the above value may be employed in this preliminary investigation. I n the temperature range 200° to 500° t9he vapour of sulphur consists mainly of the molecular form S8. Preuner and Schupp (Zeitsch. physikal. Chem. 1909 68 148) have measured the equilibrium of the reaction 4S = 35,. The mean value of the heat effect is 26,500 cals. This heat is absorbed in breaking down 3S8 molecules to 4S6 molecules. The same authors have obtained a fairly accurate value for the heat absorbed namely 58,000 cals.in the gaseous reaction S6 = 3S,. Hence the process $3 + S requires an absorp-tion of heat equal t o 21,542 cals. We have now to consider the heat absorbed in the dissociation of S2 into the atomic state’. Buckle (Zeitsch. anorg. Chem. 1900 78 169) has measured by an explosion method the equilibrium of the reaction S,=28 in the gaseous state over the temperature range 2000° to 2500O. The results do not lead to an accurate value for the heat effect. Budde takes the value 120,000 cals. per gram-molecule. Ton Warten-berg (Zeitsch. m r g . Chem. 1908 56 320) estimates the heat effect to be 90,000 cals. approximately. It has been shown (com-pare T. 1918 113 471) that $he critical increment in the case of the dissociation of a molecule into atoms is connected with the heat absorbed by the relation -Q,=E-$RT.At T=2000° the value of E obtained from Budde’s results is therefore 122,000 cals., but this is liable to considerable error. On the radiation hypo-thesis this energy should be given by Nhv where v is the frequency of the light absorbed N the number of molecules in one gram-molecule and h PJanck’s constant. Martens (Ann. Physik 1902, [iv] 8 603) has calculated that sulphur should possess a band in the. ultra-violet region a t h = 226 pp. The corresponding frequency is 1 3 . 3 ~ 1014 and therefore Nhv or the critical increment per gram-molecule should be 125,550 cals. This is remarkably close to the value calculated from Budde’s data.I n fact the agreement is partly accidental. It is probable that the value obtained from Martens’s data is the more correct. It follows that the heat of dissociation of diatomic sulphur into the atomic state in a gaseous system is 123,000 cals. per gram-molecule. Hence the energy absorbed in the process is8 -f 2 s is (123,000 + 21,540) or 144,500 cals. in round numbers. As might be expected the chief factor in the total energy change from S to atomic sulphur is t’he single process of dissociating the S molecules. The critical increment We have now to consider the activation of sulphur vapour LEWIS STUDIES IN CATALYSIS. PaRT X. 187 required to produce two gram-atoms of sulphur in the gaseous state from the corresponding quantity of S8 molecules is 147,000 cals., and therefore the critical increment per gram-atom is 73,500 cals.We have now to consider the formation of sulphur dioxide from oxygen and sulphur the latter consisting of S8 molecules the system being entirely gaseous. The partial critical increment of the oxygen is taken to be 30,000 cals. per gram-molecule. Hence the total critical increment of the system (S+O,) under the con-ditions stated is (73,500+30,000) or 103,500 cals. The heat of formation of sulphur dioxide from solid sulphur and gaseous oxygen is 69,400 cals. per gram-molecule (Berthelot) (compare Perguson Proc. Nat. Acad. Sci. 1917 3 371). The heat of vaporisation of sulphur is 12,000 cals. per gram-atom in round numbers. Hence the heat of formation of sulphur dioxide from its gaseous components is 81,400 cals.Employing the relat'ion : Heat evolved = Eresultants - Ereactants, we getl 81,400 =ESo - 103,500 whence 3s 0 = 184,900 cals. per gram-molecule. It follows from this value that the frequency of the effective radiation is 19.6 x 1014 and the wave-length A = 153 ,up. Sulphur dioxide is known t o have an absorption band in the extreme ultra-violet region beyond 200 pp (compare Garrett ?Id. Mag. 1916 [vi] 31 SOS) but the position of the band has not as yet been located. Tho above exceedingly high value for the critical increment of sulphur dioxide requires that the molecule should be correspond-ingly stable. Thus it should not be possible to decompose it into its components by a quartz mercury lamp since quart.z does not transmit wave-lengths longer than about 185 pp.As an illustra-tion of its stability it may be mentioned that von Wartenberg (Zoc. c i t . ) was unable to detect any sensible dissociation of sulphur dioxide even a t 2200O abs. For our present purpose it is more important t o observe that the critical increment of the reactants (S+O,) is also very high namely 103,500 cals. The numerical values given above refer t o the reaction non-catalysed. If however the reaction is carried out in the presence of solid or fused sulphur heterogeneous catalytic effects enter. This has been shown experimentally by Bodenstein and Car0 (Zeitsclt. physikal. Chem. 1910 75 30) the sulphur acting as a positive catalyst. The result of the positive catalysis is that the critical increment of the system (S + 0,) is much less than the value given above.From the temperature coefficient obtained hy Boden-sstein and Caro in the region of 250° in the presence of solid sulphur it is found that the critical increment of the reactant 188 LEWIS STUDIES IN CATALYSIS. PART X. ( S + O ) lies between the limits 31,308 and 34,184 cals. the mean value being 33,000 cals. in round numbers. It is possible to account approximately for the order of magni-t,ude of the critical increment obtained when heterogeneous catalysis occurs by supposing that the oxygen is already activated a t the temperature chosen before coming into contact with the sulphur surf ace the increment. of partial activation of oxygen being of the order 30,000 cals. as we have seen already. The sulphur itself is already in the atomic state in the surface layer of the solid and consequently does not require further activation.The heat of volatilisation of the sulphur dioxide per gram-molecule is a quantity of the order 5000 cals. so that in all the apparent increment is of the order 35,000 cals. which agrees moderately well with that observed. I n the above case the catalytic factor a t 250° is e(103j500-3585°0)/Rp or e681000/RT or 1028 approximately. These numbers are simply employed for purposes of illustration; sufficient data have not yet been accumulated to permit of more exact calculation. If such changes in the critical increment are brought about as a result of catalytic effects it is necessary t o conclude that in general the heat effect of a process will be modified by the catalyst, and if this is the case the variation of the equilibrium constant of the reaction in the surface layer with temperature will be affected, so that finally the equilibrium constant of the catalysed reaction will differ from that of tlhe non-catalysed reaction.This con-clusion is in general agreement with that arrived a t by Bancroft ( J . Physical Chem. 1917 21 573) on false equilibria and the effect of heterogeneous catalysis on the position of the equilibrium. The Union of Oxygen and Hydrogen. Bodenstein (Zeitsch. physikal. Chern. 1899 29 665) has found that the temperature of the termolecular velocity constant corre-sponding with the reaction 2H2+0,=2H20 is 1.75 for loo ov0r the temperature range 482" to 509O. The reaction proceeds under the conditions employed almost entirely a t the surface of the porcelain containing-vessel.From the above value of t'he tempera-ture coefficient it would follow that the critical increment for two gram-molecules of hydrogen and one gram-molecule of oxygen is 66,000 cals. and therefore for one gram-molecule of hydrogen and one ha!frgram-molecule of oxygen the increment of the reactants is 33,000 cals. Bodenstein's results have however been crit'icised by Bone and Wheeler (Phil. Trans. 1906 [ A ] 206 l) who find that the reaction is not termolecular but approximately unimole LEWlS STUDIES CATALYSIS. PART X. 189 eular especially unimolecular with respect to the hydrogen. The reaction which appears to occur is therefore H,+O=R,O. Bone and Wheeler have given data for the reaction f r m which the temperature c d c i e n t and critical increment of the reactants may be calculated when nickel is the catalyst.For the temperature range 473O to 493O abs. the critlical increment of the reactants is calculated to be 35,000 cals. which agrees fairly well with the value obtained from Bodenstein’s results for the porcelain surface. Over the temperature range 493O to 513O abs. the results obtained by Bone and Wheeler give an increment of 52,000 cals. in round numbers. This is considerably greater than that obtained a t the lower range of temperature and indicates that the catalytic effect is relatively less efficient a t the higher temperature due pre-sumably to diminished adsorption of the reactants. In both cases, however the increment is a relatively small quantity very much smaller than would be expected from the process occurring in the homogeneous phase f o r the molecule of oxygen which has to be dissociated is very stable.We have now to attempt to account for a quantity of the above order of magnitude on the basis of the energy-mechanism out-lined. Let us assume in the first place that the oxygen is adsorbed and exists in the atomic state attached to certain posi-tions on $he suTface of the catalyst. It is necessary that an activated or polarised molecule of hydrogen shall come into con-tact with an oxygen atom. It is only necessary for the hydrogen to be partly activated. Bohr (Phil. Mag. 1913 [vi] 26 1 476, 857) has investigated the energy changes which occur in the mole-cule and the atom of hydrogen in.various processes involving the removal and addition of an electron. Bohr has calculated that the process of transferring an electron so as to give rise to a system consisting of a positively charged hydrogen atom and a negatively charged one requires an absorption of energy of 21,000 cals. per gram-molecule of hydrogen. We shall employ this value in the present case although there is evideiice that a somewhat higher value is probably more correct. The latent heat of vaporisation of wat8er is in round numbers 9000 cals. per gram-molecule’in the neighbourhod of looo. As before we shall assume that the heat of de-sorption of the water produced in the reaction is of the same order of maghitude. Hence we would expect the critical incre-ment of the process to be of the order 30,000 cals.per gram-molecule of hydrogen and per gram-atom of oxygen. This agrees moderately with the observed value. SufficientIy accurate data are not as yet available for calculating the critical increment of the reactants of the same reaction in th 190 LEWIS STUDIES IN CATALYSIS. PART X. homogeneous gaseous state. It is necessary to dissociate the mole-cule of oxygen and this appears to require a quantum of energy corresponding with approximately the region h = 200 pp whence the critical increment per grain-molecule is of the order 140,000 to 150,000 cals. That is the total increment! of the reactants, reckoned per gram-molecule of hydrogen is 21,000 + 140,000/ 2 or 91,000 cals. The catalytic efficiency is therefore given by the ratio order 1017.These figures are merely illnstxatixre but they serve to indicate the great influence on the velocity which is to be expected on the basis of the treatment suggested. I n dealing with the union of oxygen and hydrogen i t has been assumed above t.hat the oxygen is condensed in the atomic form on the catalyst the subsequent chemical change being H + 0 = E120. From a number of observations made by Bone and Wheeler (Zoc. c i f . ) i t appears that hydrogen is preferentially adsorbed. I n such cases the most probable reactpion because it involves the minimal critical increment. would be represented by H,+O,==H,O, in which the hydrogen and oxygen are partly activated but neither of them is completely dissociated.The formation of water would result from the subsequentl decomposi-tioii of the hydrogen peroxide'. The idea that. hydrogen peroxide is an intermediate stage is of course not new. It appears from such considerations that the specific nature of the catalyst may determine the actual mechanism of a given reaction to a large extent. e-3%00o/RT c %000/RT or e % O O f l / ~ ~ . At 5000 this factor is of the I -The Union of Oxygen and 1S"ilicon. I n the reactions just considered the critical increment of partial activation of oxygen has been taken to be 30,000 cals. approxim-ately this being the value required for t.he formation of ozone. As already pointed out more than one stage of activation may be anticipated up to the limiting activation which corresponds wit?h complete dissociation of the molecule into atoms.Each activation corresponds with a certain size of quantum of radiant energy that! is with a certain frequency. The general conclusion reached in connexion with absorption spect'ra is that frequencies are related to one another in terms of even multiples of some fundamental fre-quency that is various degrees of activation are similarly related. A low degree of activation of the oxygen molecule requires 30,000 cals. of energy to be absorbed per gram-molecule and therefore higher degrees of activation would require 60,000 90,000 cals ., etc. up to the limiting value of complete dissociation whic Ll3WY.S STUDIES IN CATALYSIS. PART X. 191 appears to correspond with a quantity of the order 140,000 to 150,000 cals.Sufficient information is not as yet available t o enable us to say how many of these possible degrees of activation may actually manifest themselves. As an example of partial activation of oxygen which is apparently considerably greater than 30,000 cals., we may take the case of the formation and decomposition of an exceedingly stable coImpound silica or quartz. To decompose a molecule of quartz it is’evident that a quantum in the very extreme ultra-violet portion of the spectrum is required, in order t o supply the necessary energy. It is well known that quartz commences to absorb radiation sensibly beyond the wave-length 185 pp. s. Richardson (Phil. Hny. 1916 [vi] 31 463) finds that the dispersional wave-length of quart’z is 105pp. It does not necessarily follow that the dispersional wave-length or frequency is thab required for complete dissociation of the molecule.That in the case of quartz however the necessary wave-length cannot differ much from 105 pp is rendered probable by the follow-ing consideration. In a quartz mercury vapour lamp it is gener-ally believed t-hat the quartz remains undecomposed ; otherwise it would be difficult t o account for the life and permanence of the lamp. That is quartz can only be decomposed by a wave-length which is shorter than any emitted by the mercury vapour. 0. W. Richardson and Bazzoni (Phil. Mag. 1917 [vi] 34 285) have found that there is a limiting wave-length in the spectrum of a substance; t.hat is no wavelengtlh shorter than a certain value, characteristic of the substance can be emitted.I n t.he case of mercury vapour this limiting wave-length lies between 120 and 100 pp. The mean of these two limits is 110 ,up and we conclude on the above reasoning that quartz can only be decomposed by a wave-length shorter than this value. This points fairly definitely to S. Richardson’s value 105 pp for the dispersional wave-length of quartz as being the wave-length capable of decomposing the molecule. The critical increment corresponding with A = 105 pp is 270,000 cals. per gram-molecule of quartz an enormous quantity which is in qualitative agreement with the known stability of quartz. We have now t o consider the heterogeneous reaction On Langmuir’a view as applied in the present paper we regard the silicon as already in the atomic date.I f x is the necessary critical increment of oxygeii per gram-molecule then x is likewise the total crif8ical increment of the reactants. The heat of the reaction is known to be 184,000 cals. i n round numbers and hence, Si + O,= SiO, 192 LEWIS STUDIES IN (3AThLYSIS. PART X, on applying the quantum-heat exprsssion heat evolved = critical increment of resultants - critical increment of reactants, we obtain 184,000 = 270,000 - 5 whence x= 86,000 cals. Owing tol the error in the observed heat effect and in the value of the critical increment of quartz this value for the critical increment of oxygen may be regarded as agreeing approximately with the value 90,000 cals. expected from the lower degree of activation of the molecule.What is particularly importank is that even this value does not correspond with complete dissociation of the oxygen molecule. We may therefore conclude that the molecule of quartz 0 0 possesses the structure Si<l rather than 0:Si:O. This is an illustration of how a knowledge of the necessary critical increments -which in the present case unfortunately are not known with precision-may lead to information concerning molecular structure. One of the chief difficulties met wi’th in the kinetics of hetero-geneous reactions has its origin in the selective nature of the absorbability of the reactants and the resultants particularly the latter. The so-called catalytic (‘ poisons ” are now generally regarded its owing their effect to marked selective adsorption as a result of which the surface of the catalyst becomes covered with a layer of molecules and is thus no longer capable of catalysing the reaction.I n many cases the resultants of the reaction are adsorbed in this manner and consequently function as a catalytio poison. Since the extent of adsorption diminishes a3 the tmpera-ture rises it is obvious that when such poisoning effects are present the temperature coefficient of the reaction velocity over a certain range of temperature is not comparable with that over a different range for the total observed velocity depends not only on the true effect of temperature on the chemical process itself but likewise on the alteration in the ext’ent of active surface presented to the reactants. The simplest conditions are obviously those in which the adsorption effects are a minimum and such conditions will occur generally when the energy required for sublimation or de-sorption is small. I n the other cases where adsorption effeda are large it is necessary to correct the observed velocity aonstants for the change in the area of the effective surface produced as a result of the change in temperature. Thus in tlhe case in which the resultant is markedly adsorbed and therefore acts as a negative catalyst the temperature coefficient will possess too high a value, and instead of decreasing as temperature rises may even increase. A similar abnormal behaviour is to be anticipated when a reaction proceeds partly in the homogeneous gaseous phase partly in th THE ESTIMATION OE' THE METEOXYL GROUP. 193 surface far as the temperature rises the reaction bride to pre-dominate in the gaseous phase and therefore possesses a higher temperature coefficient. MUSPRATT LABORATORY OF PHYSICAL AND ELECTRO-CHEMISTRY, UNIVERSITY OF LIVERPOOL. [Received .January 227244 1919.