DIXON : THE UNION OF HYDROQEK AND OXYGEN IN FLAME. 661PRESIDENTIAL ADDRESS.Delivered at the ANNUAL GENERAL MEETING, March 18th, 1910.l3y HAROLD R. DrxoN, M.A., Ph.T)., F.R.S.IN his presidential address last year, my distinguished predecessorin this Chair quoted with approval that clever definition whichdistinguishes a theory from a hypothesis -f‘ a theory is a suppositionwe hope to be true; a hypothesis is a supposition we expect to beuseful.” I do not know whether the majority of scientific peoplehope that the theories they employ are true; I feel very sure theybecome convinced of their truth by a long habit of using them.Supposition’s that have stood the criticism of Time, and have beentuught to u s as accepted theories, take possession of our mindswhether we will or no.I have heard St. Claire Deville, whoprofessed his disbelief in all theories, lapse unconsciously into theatomic theory in moments of controversy. Such theories becomealmost an essential part of our mental apparatus, and perhaps nogreat harm results if the majority of us believe in them as an articleof scientific faith. we expect tobe useful.” I n its inception it is a more or less piausible guess-often arrived a t by analogy. We try it (with proper scepticism) tosee if it will “ work.” If it serves we try it again with less mentalreserve. The hypothesis suggests certain consequences undercertain conditions; if we find these consequences follow in a fewtrials we are apt to regard our hypothesis as verified,” and webegin to think in terms of it-especially if we have published it..By and by our hypothesis becomes crystallised in our system, antii f further consequences are not in accord with ib, we-to reform thelanguage of the unreformed Chamber-disregard the consequences.But the hypothesis, however useful it may be as a means of winningnew facts, or new views of facts, is by no means proved to be trueby successful prediction.We must be continually on our guardlest we become the bondmen of our own hypotheses, although I donot think that we Britons, as a race, are the worst offenders.“Chemists,” wrote Stas, “ t h e instant they see certain facts arereproduced with an uppenranwe of regularity, believe this is a simplenatural law; moreover, they have contracted the habit of con-sidering that the law has been demonstrated the moment they havemade any measurements not greatly differing from it.”The danger lies in the hypothesi662 DIXON : THE UNION OF FIY~ROGEN AND OXYGEN IN FLAME.I am not claiming for myself exemption from the commoninfirmity.I have made hypotheses, and very possibly have allowedthem to bias my judgment. Moreover, I have suffered what Huxleycalled that great tragedy of science--“ the slaying of a beautifulhypothesis by an ugly fact.” But I regard this as a proper anddesirable ending, and am glad t o have helped in the “happydispatch,” not only of my own, but of those of some of my fellow-workers.I would urge, then, t h a t i t is the present wsefdness of thehypothesis, not its truth, t h a t is important: we ought to value i tonly so long as it is a working hypothesis, that is, is guiding andsuggesting work in the laboratory.We ought, indeed, to do unt,oour hypotheses as we would do unto professors-other professors-inan ideal university. The moment the hypothesis ceased to workin the laboratory it should be quietly superseded.My text might have been illustrated from almost any branch ofchemistry; I have chosen the union of hydrogen and oxygen in flame,partly because I am more familiar with this subject, and partlybccause the reactions involved are comparatively simple. It wasmainly with the hope of finding, among these or other gases,reactions simple enough to be interpreted and to throw light (‘onthe course of a chemical change,” that I started these researchesunder the guidance of Mr.Vernon Harcourt more than thirty yearsago. Though I have, no doubt, wasted much time, I have neverregretted the choice of subject, and have never been tempted toturn aside.The Direct Union of Hydrogen and Oxygen.I first studied the union of hydrogen and oxygen in flame a.spart of the investigation on the mode of burning of carbonmonoxide. When steam, either present to start with or formed inthe flame, was found to be necessary for the propagation of a flamein carbon monoxide-oxygen mixtures, I ventured on my f i r h thypothesis-I think a moderate one. I f there is some constitutionaldisability preventing oxygen from direct combination with carbonmonoxide, may not the latter take oxygen from steam, and thehydrogen, so liberated, re-form steam by direct union wit.h oxygen.I need not a t this point discuss what special disability preventsthe oxygen from burning the carbon monoxide directly.Manyhypotheses Eave been made to account for it, for example, that theoxygen, per se, is too stable (Lothar Meyer, Beketoff), or that gasesonly react in equal volumes (Mendeleeff). With the aid of severalof my old students I have shown, in communications to this Society,that neither the stability of the oxygen, nor the “law of equaDIXON : THE UNION OF HYDROGEN AND OXYGEN IN FLAME. 663iiiolecules,” can be the cause, for carbon monoxide will not explodewith ozone, with chlorine monoxide, or with nitrous oxide.Ontlic other hand’, the dried gases unite readily ujithout in&uming intlic presence of red-hot platinum, and in the burning of driedcyanogen the carbon monoxide first formed will burn in excess ofoxygen-either prolonging the flame, as in explosions, or burningwith a separate flame, as in the Smithells’ flame-separator. Steam,then, is cessary for the burning of carbon monoxide only under.certain corrditions. How does it act ? Carbon monoxide was foundto take o ygen from steam and liberate the hydrogen when thetwo were ieated together in a variety of ways. I n explosion ofmixtures I f carbon monoxide, oxygen, and steam, in which theoxygen was insufficient for complete combustion, the steam wasfound to give up its oxygen to burn the excess of monoxide.Moreover, the propagation of the flame through the mixture wasfound to increase in velocity the more steam was added up to6 per cent.of the mixture. It was therefore possible for thecarbon monoxide to take oxygen from steam in a flamc, and ifthe liberated hydrogen united directly with oxygen, water wouldbe re-formed, and the whole operation be completed by therepetition of some simple cycle.So the next point to investigate was-Does hydrogen unitedirectly with oxygen in a flame, or does it only do so through thedecomposition of or by the intervention of water?My first experiments on the union of hydrogen and oxygen werehigh temperature experiments. I found that an clectric spark,which would fire a damp mixture of hydrogen and oxygen, wouldalso fire a similar mixture however carefully dried.Many differentelcotrodes were tried, and they were submitted to every process Icould devise to prevent the possibility of steam being formed inor on the surface of the electrode. The spark always determinedexplosion. Moreover, I have analysed (on a rapidly-moving film)the flame of an explosion in electrolytic gas from its starting pointa t the spark as it spreads into the unburnt gas. The experimentswere made in exactly similar tubes, all other conditions being thesame-except that, in each pair, one tube was well dried and theother was moist. I n no case could any difference be detectedbetween the dried and moist gases either in the initiation or tlicspread of the flame.Dr. Brereton Baker, than whom I could citeno more careful or patient experimenter, has come to the same con-clusion. But it may be objected: “ Some steam molecule may existin the path of the spark, and that starts the reaction; once started,the water formed serves to spread the flame.” We cannot, of course,prove that no stcam molecules exist in the dried gas-indeed, 664 DIXON : TEE UNION OF HYDROGEN AND OXYGEN IN FLAME.should say it was very probable that there are some. But we havesome evidence as to the proyress of the reaction in the explosion ofhydrogen and oxygen, and that evidence w0 can use to test thepoint whether the flame, once started, is propagated by direct unionof hydrogen and oxygen molecules, or whether that union onlytakes place through’the intervention of a steam molecule.Both Moritz Traube and Professor Armstrong have put forwardthe view (on different grounds) that for the formation of steamthere must be a simultaneous reaction between three molecules :H, + OiH, + 0,.Now, the “ explosion-wave ” is a true physico-chemical constant.The genius of Berthelot was not at fault when he described theadvancing locus of high pressure and of rapid chemical change asZ’onde explosive.The “ wave ” was propagated, according t oBerthelot, from layer to layer by successive shocks between themolecules, resulting in chemical action ; so that the explosion-wavediffered from a sound-wave in that the former varied, not only withthe physical, but with the chemical nature of the gaseous mixture.The maximum velocity of the explosion-wave could be predicted,said Berthelot, by calculating ths mean velocity of the gaseousproducts before they had lost by conduction or radiation any of theheat of the chemical change.My own work on the explosion-wavebegan in an attempt to confirm or disprove my hypot.hesis thatsteam was the “carrier” of oxygen in the explosion of carbonmonoxide and oxygen. The results obtained led me to use theexplosion-wave as a, means of tracing the course of other changes ingases. For instance, it was found most useful in showing that thecarbon in gaseous compounds of carbon was not burnt directly tocarbon dioxide in explosions, but, in every instance that could betraced, the carbon dioxide, appearing as an end-product, had beenproduced by a secondary and not by a primary reaction in thewave-front.After many experiments on the diluting effect ofvarious gases on the propagation of the explosion-wave, I was ledto advance a “ workipg hypothesis ” somewhat different from thatof Berthelot. I n my hypothesis the gases are not burnt ‘( cold,” butthe unburnt molecules are heated in the wave-front by ‘‘ shock,”and so are burnt (‘ hot,” and consequently the product is raisedto a higher temperature. A formula based on this hypothesis wassuccessful in predicting the velocities of the explosion-wave in anumber of different gasesDIXON : THE UNION OF HYDROGEN AND OXYGEN IN FLAME. 665Velocity of explosion-wave.Gas mixture. Calculated. Found.H, + 30, 1740 17122H, + C1, 1832 1849C,N, + 0, 2725 2728C,N, + 0, + 2N, 2166 2163C,H,+20,+8N, 1727 17348H, 3- 0, 3554 3535I find it very hard to believe that these extremely closeapproximations can be mere coincidences.There are probably many cases in which the reactions are notso simple as the above.I find that the explosion-wave in a mixtureof equal volumes of ethane and oxygen does not give the sameproducts as are obtained in the explosion of ethylene, or ofacetylene, with its own volume of oxygen. The ethane is not burntwholly to carbon monoxide and hydrogen, but appears to form (asProfesor Bone has shown at lower temperatures) acetaldehyde andsteam, the acetaldehyde yielding methane and carbon monoxide.For other gaseous mixtures, especially those in which steam isformed a t a very high temperature, the formula gives a higher valuethan the rate found.For instance, the found velocity of theexplosion-wave in electrolytic gas is a long way below that given bythe formula. I “ explained ” that by t.he further supposition thatthe dissociation of steam was appreciable in the explosion-wave ofelectrolytic gas. I n support of this view, I showed that about1 per cent. of unburnt gas* remained behind after the explosion-wave had gone by, and proved that this could not be attributedto the cooling effect of the walls.I do not believe to-day in the truth of my working hypothesisof the explosion-wave. It embodied a number of assumptions, someof which I have myself shown to be erroneous.But I still think itwas sufficiently parallel to the truth to be useful: I made no higherclaim for it.The theory of the explosion-wave is not to-day dependent on thehypotheses of Berthelot and myself. Sufficient experimental datawere secured to allow a mathematical treatment of the subject.Professor Schuster was the first to suggest that Riemann’s equationfor the propagation of an abrupt variation in the density andpressure of a gas might be applied to the propagation of theexplosion-wave, since the necessary uniformity of type is maintainedowing to the continued reinforcement of the wave by the successivechemical changes. In 1899 Mr. D. L. Chapman, acting on thissuggestion, worked out from Riemann’s equation an expression for* Mem.Manehester PhiE. Soc., 1888. The fact was rediscovered in 1897 byF. Eniich (Monatsh., 18, 6), who attributes it t o the cooling action of the walls666 DIXOK : THE UNION OF HYDROGEN AND OXYGEN IN FLAME.the propagation of the explosion-wave, making, of course, ailassumption as to the specific heats of the gases formed at the tem-perature of the explosion. He pointed out how it might bcpossible to use the easily determined velocities of the explosion-waveto give the specific heats of the gases concerned. I n 1906 ProfessorJouguet, without knowing Mr. Chapman's work, developed onsimilar lines a very complete theory of the explosion-wavc.sa I-Ichas most justly chastised the wal:t of mathematical rigour in myhypolhesis and in that of Berthelot.We have reached the rightresult, he says, by a kind of chemical intuition. But as our resultshave led to M. Jouguet's generalisation, I, for one, will hope theverdict of my fellow-chemists will be: " Guilty, but please do itagain." Fundamentally, I think, M. Jouguet and I have the sameconception. We both think there is a preliminary heating up of thegas just before it burns; we both think that the wave-front ismoving as a sound-wave with a velocity, relative to the unburntgas in front, t w i c e that calculated from the ordinary heat of com-bustion of the gases. But whereas I had supposed this increasedvelocity should be set down to increased temperature, M. Jouguetpoints out that the burning gas is driven forward e n masse with thcvelocity of sound, and the wave-front is travelling like a sound-wave in this moving gas, and therefore has a motion relative to theunburnt gas in front double that of a sound-wave in the burninggases.The theoretical conclusions of M.Jouguet concerning the forwardmotion of the gas behind the wave-front are entirely borne out bymy photographic analysis of the explosion-wave. One can showthe retardation of a sound-wave meetilag the on-rushing burninggas; one can show that a sound-wave, following a little behind thcwave-front, travels with almost the same velocity as the explosion-wave itself.One other point of some importance: I found experimentallythat an increase in the initial temperature of the unburnt gas wasaccompanied by a diminution in the rate of explosion.Myformula did not account for this, but. I find the equations ofM. Jouguet lead to this result, when a moderate rise in the specificheats of the products of combustion is assumed.Messrs. Chapman and Jouguet have raised the very importantquestion whether the specific heats of gases at high temperaturescan be calculated backwards from the observed velocities ofexplosion. I would point out that a determination of the velocityof the explosion-wave for one mixture would not definitely fix thespecific heats of the products of combustion. For instance, if we* Joiiriut.1 de Mathe'matipes, 1905-6DIXON : THE UNION OF HYDROGEN AND OXYGEN IN FLAME. 667ttake the simple case where cyanogen burns t o carbon monoxide andnitrogen :the velocity of the explosion-wave was found to be 2728 metresper second when the initial temperature was 10’.Working back-wards from the observed velocity, I find that the mean specific heatof the products may lie betweenCZN, -k 0, == 2CO + N,0.5 1 0.61000 10004.5 + --T and 4-5 + -T,according as the compression of the gases is assumed to be more orless. The observed rate corresponds with the following pressuresand temperatures :Specific heat of CO and N,.4*5+ --TPressure in atins. Temperature, C.57.21 5966” 0-51100051.21 5736” 0.5510004*5+ -- T4.5 + O?oT100047‘16 5541”But by measuring the velocity of the explosion-wave when thismixture is “damped down” with inert nitrogen, or with argon,it is possible to limit the range of specific heats correspondingwith the observed velocities, and also to compare the specific heatsof nitrogen and argon at high temperatures.I am makingexperiments on these “ retarded ” explosions.The exceptions to the application of Chapman’s and of Jouguet’sequations-for example, in the explosion of carbon monoxide andoxygen, where we believe on other grounds that steam intervenes;or in the complete burning of carbon compounds, where we haveother reasons for believing that the formation of carbon dioxide isa secondary action-appear to me greatly to strengthen the viewthat the flame is propagated as a wave.If, then, the flame in explosions of gases is really propagated id9a pressure-wave, that is, by collisions of molecule with molecule,the chemical change involved in the wave-front must also proceed bythese collisions; and since the wave dqes not wait for the rareoccurrence of a triple collision, the primary chemical change mustbe uni- or bi-molecular, n o t termolecular.As Sir Joseph Larmor *has put the case, “Imagine the scale of magnitude of a gas a t apressure of one atmosphere to be magnified so that the diameterof each moving molecule becomes about one inch, there will bein the model roughly about one molecule in each cubic foot, andMem. illanchester Phil.soc., 1908.* “On the Physical Aspects of the Atomic Theory.”VOL. XCVII Y 668 DIXON: THE UNION OF HYDROGEN AND OXYGEN IN FLAME.a molecule will have to travel about a hundred feet before it,encounters another one.Such binary encounters will thus happenwith some frequency, and from some of them combination mayensue. But the chance of three molecules coming together simul-taneously is negligible; the only way in which a termolecularcombination can arise is by one of the molecules attaching to itselfanother, and this pair going off together to meet a third. . . . Itappears to be commonly recognised that direct termolecular com-binations occur seldom; the inference from the present line ofargument is that in gaseous reactions they do not occur at all.”It has been suggested that Sir J. J. Thomson’s hypothesis-thatchemical action in gases begins on the surface of condensed particlesof liquid water-might be applied to the explosion-wave.I findit difficult to believe either that sufficient water particles can existin the dried unburnt gas to allow the flame to proceed a t theobserved velocity, or, if such condensation really took place, thatthe velocity of explosion would not be increased by the addition ofwater vapour.To sum up, I believe the reaction between hydrogen and oxygenin an explosion is a direct one, and does not take place throughthe intervention of steam, for three reasons, namely:(i) That well-dried mixtures of electrolytic gas always explodewith a spark;(ii) That the velocity of explosion in a well-dried mixture isgreater than when steam is added;(iii) That the explosion-wave is propagated as a pressure-wavethrough the gas.The Union of Hydrogen and Oxygen at Low Temperatures.Whether steam or water facilitates the slow combustion ofhydrogen and oxygen below the inflammation temperature isanother question.The most careful experiments made to decidethis question are those of Dr. Brereton Baker. In 1902 he describedto this Society experiments in which he showed that when thegreatest care had been taken to purify the gases, the moist gasesexploded when the tube containing them was heated in a flamewhile the dried gases did not,. I n some of the tubes which did notexplode, a small quantity of steam had been formed, and in onetube, which had been dried for two days only, water was visiblyproduced by slow combustion.I n another experiment Dr. Baker heated a fine silver coil inthe dried gases, and succeeded in melting it without exploding themixture.The importance of these experiments induced me to repeat themDIXON : THE UNION OF HYDROGEN AND OXYGEN IN FLAME, 669and Dr.Baker was good enough to assist me by placing at mydisposal an ingot of his pure silver, and some of his fifteen-timesrecrystallised barium hydroxide. To provide against the infirmitiesof age, I enlisted the services of Dr. E. C. Edgar (whose work onthe atomic weight of chlorine proves his manipulative skill) to makean entirely independent series of experiments. As our experimentshave not been published, perhaps I may be permitted to give abrief summary of them here,The tubes, in which the gases were collected, were made of Jenaglass; into the walls of these were fused platinum hooks, fromwhich hung coils of thin silver wire carefully drawn from Dr.Baker’s ingot. The tubes were connected by a glass joint eitherdirectly to the electrolytic vessel, or to a condenser and dryingtube sealed to that vessel.The tubes, as well as the other parts ofthe apparatus, were cleaned by hot chromic and nitric acids andby steaming; they were finally dried in a current of air.The tubes were filled with electrolytic gas by placing them in atrough holding pure mercury (Fig. l), and partly evacuating themuntil the mercury rose nearly to the silver; the gas was then slowlyadmitted. After several repetitions of this operation, the piecea1’ Y 670 DIXON: THE UNION OF HYDROGEN AND OXYGEN IN FLAME.of phosphoric oxide (redistilled in oxygen) were introduced, andthen the last traces of air were eliminated by again emptying andfilling.I n other tubes no mercury was admitted, but after the admissionof phosphoric oxide the glass was sealed off, and the air pumpedout, while the tube and silver coil was maintained at a red heat.On cooling, the electrolytic gas was admitted, the pumping repeated,and the tubes, after being finally filled, sealed off.For the “ dry ”experiments, forty days were allowed before the silver coil washeated.Dr. Edgar agrees with me in the following observations-(i) thata higher temperature was required to start the reaction in thedried than in the moist gas; (ii) that combination occurred bothin the moist and in the dried gases, and that liquid water wasdeposited in the lower (cool) portion of the tube.In four experiments performed by Dr.Edgar, he finally fusedthe silver wire without exploding the dried gases. His account,written at the time, says : “ I n each experiment, after the mercuryhad risen almost to the silver, the wire was fused withoutexploding the gases.” I can entirely confirm this description. Iwatched the mercury rising in the tube and drops of water con-densing; on raising the temperature of the wire, the mercury rosemore quickly, and finally the silver fused without an explosionwhen the mercury was near the coil. I confess I did not think atthe moment that;, when the silver fused, the gas in the heatedupper portion of the tube must have been mainly steam.Onreflection it was clear that the experiment should be tried ofattempting to melt the silver by rapid heating. Two of the tubeswhich had been drying for forty days exploded when the coil wassuddenly heated to a bright red-heat. Our experiments show that,with the purity and dryness we were able to obtain (no doubt lessperfect than in Baker’s experiments), hydrogen and oxygen donot unite so readily as moist gases in contact with a heated silverwire, but such gases can be exploded by a sufficiently hightemperature.Professor Le Chatelier has defined the ignition point of gasesas that temperature at which the initial flameless combination heatsup the gas (more or less rapidly) until it inflames. ProfessorNernst has given a similar definition.Below the ignition point,combination occurs, but the heat evolved is not sufficient to over-come the cooling effect of the surroundings. I n a tube heated to540°, for instance, electrolytic gas slowly combines, the amountof steam formed in unit time gradually diminishing as the changeproceeds. A t 560° the heat evolved by the more rapid combinatioDIXON : THE UNION OF HYDROGEN AND OXYGEN IN FLAME. 671self-heats the mixture to the inflammation point.. I f the gas,however, can be maintained at 560° by contact with a sufficientlylarge surface, the combination proceeds as before, but more quickly.I n those experiments, such as those of M. HBlier, in which electro-lytic gas is drawn over small pieces of heated porcelain, exposinga very large surface to the gas, temperatures of 800° and highercan be employed without producing explosion.I n the experiments which Dr.Coward and I published last yearon the ignition point of gases, we brought together hydrogen, heatedin an inner tube, and oxygen, heated in an outer tube: we foundit necessary to use a wide outer tube for the oxygen, and a certainrapidity of flow of the hydrogen in the inner tube in order toobtain a constant ignition point. When the region where theheated gases mingled was removed from contact with a heatedsurface, a minimum ignition temperature was obtained. When thegases met close to a heated surface, the ignition point was raised.That electrolytic gas can combine isothermally at high temperaturesif the surface contact is large is a sufficient explanation of theseobservations.Now the question raised by Dr.Baker is this: Is the initiationof slow combustion in electrolytic gas, and. the velocity of thereaction when it is started, influenced by the presence of steamor water particles? I n his latest experiments he has found thatwhen moist hydrogen and nitrous oxide are kept at 530°, thepresence of an ionising agent, such as thoria or radium bromide,greatly increases the rapidity of the reaction; in the carefullydried gases the ionising agent had no effect.Although I agree with Dr. Baker that steam (or water) inter-vmes in the initial action of hydrogen on oxygen at moderatetemperatures, I do not think it necessary for the propagation ofa flame once started in the mixture.There are many similardifferences in gaseous reactions. Light will not explode driedhydrogen and chlorine, it does so in presence of water; but oncestart a flame in mixed hydrogen and chlorine, and it is propagatedas an explosion-wave through the dried gases. A platinum wireheated to whiteness will not fire a mixture of cyanogen and oxygen,flameless combination taking place round the wire; but a sparkimmediately sets up the explosion-wave in the same mixture.What is the First Product of the Union of Hydrogen and Oxygenin Flame?Do hydrogen and oxygen directly form steam, or do they formhydrogen peroxide in a flame? I f we accept the view ofMendcEeff that all primary reactions between different substance672 DIXON: THE UNION OF HYDROGEN AND OXYGEN IN FLAME,are bimolecular-a view enforced by Sir J.Larmor for gaseousreactions-we must suppose that a collision between a hydrogen andan oxygen molecule can only result in the primary formation ofhydrogen peroxide, or in the formation of steam and a free oxygenatom.The following reasons, among others, have been brought forwardfor the view that hydrogen peroxide is the first product in flame :(i) That hydrogen peroxide is found in the water rapidly con-densed from a hydrogen flame;(ii) That the mixture of equal volumes is more sensitive to aspark;(iii) That the mixture in equal volumes has a lower ignitionpoint than any other mixture.(I.) As, I think, Moritz Traube first showed, a jet of burninghydrogen playing on to the surface of water produces peroxidein the water.Some years ago, when I repeated Traube’s experi-ment, I thought that the mere heating of the water would producesome peroxide, but I found afterwards that this was incorrect. Itrequires some burning to take place near the surface of the water.But the presence of the peroxide in the condensed water might beexplained either by its direct formation from hydrogen and oxygen,o r by the union of the liberated oxygen atom with a molecule ofwater. Those who favour the first view consider tho peroxide tobc dihydroxyl, H-01-0-H, those who take the second view considerthe peroxide t o bc oxygenatcd water, r <I>=O.HThere are rcactions of hydrogen peroxidc (and o f a llralineperoxides) which may be appealed to in support of both views:possibly both forms may exist in solution.The second formulaseems to me to be supported by the remarkable researches OPCarl Harries 011 the decomposition of ozonides by water, in whichhe shows that the loosely joined oxygen atoms attach themselvesto water molecules to produce hydrogen peroxide quantitatively.The recent experiments of Manchot * confirm the olderobservations that ozone as well as hydrogen peroxide exists ina hydrogen flame. I f the ozone is formed owing to the liberationof an oxygen atom, hydrogen peroxide may be formed by theunion of an oxygen atom with steam. Engler’s experiment ofburning a hydrogen jet so as to bore a hole into ice, showsplenty of peroxide in the condensed water in the hole.I haverepeated this experiment both with ice and with a lump ofsolid carbon dioxide. When a large jet of hydrogen is used,* Ber., 1909, 42, 3948DIXON : THE UNION OF HYDROGEN AND OXYGEN IN FLAME. 673burnt with an outside blast of air, and a very small jet ofnitrous oxide is introduced into the centre of the hydrogen flame,a piece of ice may be introduced so that it is surrounded byhydrogen, and a hole may be bored in it by the small central jetof nitrous oxide burning in hydrogen. The central jet may thenbe fed by oxygen, and the samples of condensed water compared.While the oxygen produces plenty of peroxide, only a trace is foundin the water condensed from the nitrous oxide flame. This, ofcourse, only shows that in the burning of hydrogen and nitrousoxide, steam is formed directly without the splitting off of anoxygen atom.In the explosion-wave the greatest velocity of propagation is notgiven by electrolytic gas.The addition of hydrogen increases thevelocity of the wave so long as it is set up in tho mixture; this,of course, is due to the mobility of the hydrogen. But the factthat the addition of equal volumes of oxygen and nitrogen t oelectrolytic gas retards the wave exactly in proportion t o theirdensity is an argument that hydrogen peroxide is not the primaryproduct of the change, for the two gases appear to be equallyinert to the mixture 2H,+ 0,.Ifhydrogen peroxide were directly formed, the immensely high tem-perature of the flame and the very rapid cooling behind the flameshould be precisely the conditions most favourable for a high yieldof the endothermic peroxide.I have carefully examined theproducts of combustion of the explosion-ma.ve in such mixture, andcould detect no peroxide. On the other hand, when the explosion-wave traverses electrolytic gas, about 1 per cent. of unburnt gasis left behind, showing how rapid the cooling is.II. The experiments of Emich on the sensibility to expIosion bysmall sparks of different mixtures of hydrogen and oxygen led himin 1897 to suggest that hydrogen peroxide was the first product ofthe reaction on the ground that the mixture in equal volumes w amost sensitive to the spark. In a later paper (1900) hbe withdrewthis conclusion, although his work is still quoted as an argument infavour of the direct formation of peroxide.Whether or not alocal heating of a small mass of gas by a, spark will set up generalinflammation must depend inter alia on the rate a t which heat isconveyed away from the heated region; a gas richer in the rapidlymoving hydrogen molecules will convey heat away more quicklythan one poorer in hydrogen. Dr. Coward has recently foundtlhat the mixture in equal volumes is not the most sensitivetQ a spark. The addition of oxygen still further increases thesensibility.Consider the explosion-wave traversing the mixture H, + 0,674 DlXON : THE UNION OF HYDROGEN AND OXYGEN I N FLAME.111. But a stronger argument has been drawn by K. G. Falk*from his experiments on the ignition point of gaseous mixturesheated by adiabatic compression.Acting on the very ingenioussuggestion of Professor Nernst, he has compressed different mixturesof hydrogen and oxygen in a steel cylinder by means of a weightfalling on a piston, and has determined the volume of the compressedgas by measuring the lowest point reached by the piston. Heconsiders the whole mass of gas is heated equally throughout untilit reaches the ignition point, and then that the whole detonatespractically instantaneously, when the explosion stops the descent ofthe piston. On the assumption that the piston had no time tomove downwards appreciably after the ignition point was reached,he calculates the temperatures of ignition of different mixtures :Ignition point.605”540514530571The last four figures are plotted out, and the dotted curve drawnthrough them on page 676.Falk says: “The fact that themaximum affinity is shown by the mixture H,+O,, proves thatH,O, must be the first product of the reaction between hydrogenand oxygen.”Falk gives as his reasons for the assumption that the gas detonatesinstantaneously throughout its whole mass, once the ignition tem-perature is reached, the fact that the explosions are of greatviolence, and that the compression was very nearly the same in thecase of electrolytic gas whether the weight fell from a lower or ithigher altitude. What he found to be nearly true of the rapidly-firing mixtures, 2H,+ 0, and H, + O,, he has assumed to be alsotrue of the less rapidly-firing mixtures with excess of oxygen, andof mixtures of hydrogen and air.When I read Falk’s first paper, it occurred to nie that thedescending piston must send in front of it sound-waves, which wouldbe reflected from the bottom of the cylinder back to the pistonagain, and so bn, producing by their collisions zones of higher tem-perature, a t one of which the flame would eventually start.Falkstates in his second paper that Professor Jouguet has made thesame criticism. By photographing the explosion produced by theadiabatic compression on a rapidly-moving film, I have shown thatthe flame does start from a point, and in the more slowly burningmixtures the spread of the flame throughout the gas takes anappreciable time.Moreover, the point of ignition can be made to* K. G. Falk, J. Amer. Chm. Xoc., 1906, 28, 1517 ; 1907, 29, 1536DIXON : THE UPU’ION OF HYDROGEN AND OXYGEN IN FLAME. 675vary by altering the velocity of the piston. With electrolytic gasthe photographs show that the explosion-wave is very quickly setup, but the flame always starts f r o m one point. The mass of thegas i s n o t fired instnntaneousZy, although the time required in thiscase for the flame to reach the piston is negligible. With such amixture as H, + ZO,, the time required t o set up the explosion-waveis not negligible.I f we accept Nernst’s definition of the ignition point of amixture of gases as that temperature at which a mass of the gasrapidly self-heats itself by combination until it bursts into flame,and if this flame starts at some point in the mixture and spreadsfrom that point, there will be two periods during which the pistonmay continue to move after the “ ignition point ” has been reached :(i) the period from the beginning of rapid self-heating until theflame appears; and (ii) the period required for the flame to spreadthrough the gas and stop the descent of the moving piston.Just as the time required f o r the flame to set up the explosion-wave differs in different mixtures, so does the time required for theself-heating from Nernst’s ‘‘ ignition point ” to the actualappearance of the flame.When a mixture of air and hydrogen (5 vols.to 2) was com-pressed in the cylinder, and the descent of the piston was arrestedby means of a steel collar, I found the ignition temperature a fewdegrees only above that of electrolytic gas.This was in agreementwith our previous determinations at atmospheric pressure. Butwhen the piston was allowed to descend until it was stopped bythe explosion of gases, as in Falk’s experiments, I could obtainignition points, not only as high as Falk’s (649O), but, by increasingthe velocity of the piston, I got temperatures of 700°, 800°, or 900°at pleasure. A similar variation was found for the mixtureIt was evident that the error due to the movement of the pistonduring the ‘‘ pre-flame ” period of self-heating might be far greaterthan the error due to the movement of the piston a f t e r the flamehad started.I n determining the temperatures of ignition of mixtures ofhydrogen and oxygen, it was necessary, therefore, to stop the descentof the piston artificially the moment the gases were brought to theself-heating temperature.This was effected by having on the endof the piston a steel head, which was caught by a steel collar at acertain point in its descent. By varying the thickness of the collar,the piston could be stopped a t any point. By a series of trials acompression was found which fired the mixture, and a slightly lesscompression, which did not. The mean between the two was takenH, + 40,676 DIXON : THE UNION O F HYDROGEN AND OXYGEN IN FLAME.as the ignition point,. On repeating each experiment several times,nearly concordant ignition points were obtained.For the sake of comparison with Falk’s numbers, I have calculatedthe mean ignition points from my compression-volumes, using thesame ratio of the specific heats (y=1’40) as Falk has done. Thetemperatures so calculated are plotted in Fig. 2 with the continuouscurve drawn through them:Ignition pointsMixture. (y=1*40).H5 + 0 2 542H4 + 0, 536H, + 0, 53013, -t 0, 525H6+ 0, 557”Ignition pointsMixture. (y = 1 *40).H, + 0 4 520”H, + 0 5 516HZ + O6 512H,+O, 509 ir, + 0, 507570”56556055555054554053553052552051551050525 50 100 150 200 250 300 350 400Vols. of 0, t o 100 vols. of HO_.My ignition point for electrolytic gas agrees closely with thatfound by Falk, an agreement which shows that this mixture musthave a very short period of flameless combustion, and the spread ofthe flame must be very rapid. I find the mixture in equal volumeshas a lower ignition point than electrolytic gas, but it is not themixture of “maximum affinity.” The further addition of oxygenregularly lowers the ignibion point. I would submit, therefore, thaOBITUARY NOTICES. 677the ignition points of hydrogen and oxygen mixtures do not provethat hydrogen peroxide is the first product of the reaction.I began this address with a note of warning, and would wishtbo end on the same note. I have spoken of the use of hypothesesand hinted a t their abuse. I n criticism it has been my desire-atleast since the fierce invectives of youth lost their sweetness-to begentle with my adversary’s and stern with my friend’s hypothesis.If ever any of my old research students look back on my criticismsof their work with feelings from which time has removed thebitterness, I think they will admit that a t least I saved them fromtlie publication of hypotheses, useful perhaps at the time, butcertainly untenable in the light of fuller knowledge