1462 WHITE AND PRICE THE IGNITION OF CXXXVIH1.-The Ignition of Ether-Alcohol-Air and Acetone-A ir Mixtugm in Contact with IJeated Surfaces. By ALBERT GREVILLE WHITE and TUDOR WILLIAMS PRICE. OWING to the large number of fires which had occurred during 1917 and 1918 in solvent-recovery stoves in which cordite was being dried it was decided t o investigate the conditions under which mixtures of the vapours of ether alcohol and acetone with air would ignite. This was rendered all the more necessary by the fact that the information available on this subject was scanty and often contradictory. As a general rule the actual rise in temperature necessary to cause the explosion of such an explosive as glyceryl trinitrate is much lower than that needed to ignite an explosive gaseous mix ETHER-ALCOHOL-AIR AND ACETONE-AIR MIXTURES ETC.1463 ture. On the other hand the lower thermal capacity of the gas and its comparatively far greater mobility render it more susceptible to ignition in many cases. There was thus some justifi-cation for the idea that the solvent-air mixture was probably responsible for many of the recovery-stove fires that had occurred, particularly when it is remembered that fires had been less fre-quent in “final” stoves than in recovery-stoves that more fires had occurred with cordite from which a mixture of ether and alcohol was being removed than when the solvent was acetone and that ether when mixed with air is indubitably more dangerous than acetone under similar conditions. The factors affecting the ignition of a combustible gaseous mix-ture are many and the influence of some of them is not particu-larly well understood.The problem of safety when dealing with such a mixture under manufacturing conditions divides itself naturally. into two parts the one dealing with the ignition of the mixture and the other with the propagation of the flame from one portion of the mixture to another for example from one build-ing to another. This division is seen to be inherent when it is remembered that by using a sufficiently powerful source of ignition it is possible to ignite almost any combustible gas-air mixture whereas the propagation of the flame is a totally different matter particularly if the gas-mixture considered is at rest. I n such a case there appear to be definite limits for the proportion of combustible gas to air in a mixture which propagates flame, these limits depending ultimately only on the direction of pro-pagatibn and the nature of the combustible gas used a t ordinary temperature and pressure.The experimental work is accordingly divided into three sections : (1) The ignition-temperatures of various gas-mixtures including ether-air alcohol-air and acetone-air mixtures. (2) The limits for the propagation of flame in these mixtures. (3) The investigation of a few miscellaneous facts concerned more particularly with various means of ignition. The fact has not been lost sight of that ease of ignition and propagation of flame might be enhanced by the presence of some impurity in the solvent vapour-air mixture.Accordingly experi-ments have been carried out to ascertain the effect of adding slight amounts of glyceryl trinitrate and of the peroxides of ether to the gas-mixtures dealt with. This was the. more necessary as glyceryl trinitrate even though present in the stove vapours in minute quantities is known to be a source of possible danger and also because there appears to be a tendency to assign any otherwise inexplicable explosion or fire wit,h ether t o the influence of thes 1464 WHI!CE AND PRICE THE IGNITION OF peroxides (compare Neander Chem. Zeit. 1902 26 336 and others). This idea seems to have arisen chiefly from the fact that ether which had given trouble had generally been kept for some time and also because it was well known that the exposure of purified ether to light caused the formation of compounds which appeared t o contain active oxygen.The methods used to prepare these compounds were those given by Baeyer and Villiger (Ber., 1900 33 3387; 1901 34 738). Naturally great care was exercised in freeing the solvents used from such impurities. The ether used was twice distilled from acid permanganate and washed several times first with a concentrated solution of potassium hydroxide in water and then with a dilute one. It was then washed several times with distilled water dried distilled and again dried for several days over sodium. On fractionating twice with a Young and Thomas still-head a fraction boiling within 0.05O of the boiling point of the pure substance was colleFted each time. The alcohol used was ordinary absolute alcohol which was twice heated under reflux for four hours over fresh lime then twice over calcium turnings for two hours and refractionated as for ether the fraction collected boiling within 0.05O of 78*4O.The acetone was purified by converting it into the sodium iodide compound collecting and distilling the double compound. The product was then carefully dried and fractionated twice as in the case of the other compounds. The purity of the solvents used can be gauged from the fact that the acetone obtained had D,"0.7808, as low a figure as any published. These solvents were carefully preserved in a dark cupboard. Except when otherwise specified percentages can be taken to mean percentage by volume. Tubes are also often specified by their diameters.Thus a tube 5 cm. in diameter would be referred to as a 5 cm. tube. SECTION I. The Zgnition- t emperatures of Et her-A Icohol-A ir and Acetone-Air Mixtures. With the exception of two figures for ether in air 1033O given by McDavid (T. 1917 111 1003) and 190° by Alilaire (Compt. rend. 1919 168 729) the ignition-temperatures found in the literature for the solvents in question are spontaneous ignition-temperatures which can be taken to be the temperatures a t which the substances dealt with (surrounded by oxygen or air a t the same temperature) will burst into flame without the application of any spark or other local high temperature. Two sets of these figures, which are intended for engine work are given. Thus Hol ETHER-ALCOHOL-AIR AND ACETONE-AIR MIXTURES ETC.1465 (Zeitsch. angew. Chem. 1913 26 i 273) gives the spontaneous ignition-temperature of alcohol in air as 510° of acetone as 570°, and of ether as 400O. Moore ( J . SOC. Chem. Imd.,. 1917 36, 109) gives the spontaneous ignition-temperatures of ether and alcohol in air as 347O and 51B0 and in oxygen as 190° and 395O respectively. It will be seen that only the figure given for the ignition-temperature of ether in air by Alilaire which was pub-lished after the completion of our work appears to be sufficiently low t o make an ignition of the solvent-air mixture under presenb clay conditions of recovery seem feasible. A paper by Perkin (T. 1882 41 363) on the luminous incom-plete combustion of ether is interesting in this connexion. Accord-ing to him this phenomenon was first discovered by Davy who noticed a pale phosphorescent light round a hot platinum wire.Doebereiner noticed the same thing and also remarked that when ether was dropped into a retort heated on the sand-bath to looo and upwards or into a platinum capsule exposed to the vapour of boiling water the spheroidal state is produced accompanied by a blue flame visible only in the dark and not capable of setting fire to other substances lachrymatory vapours of lampic acid being formed . Boutigny and Miller also noted this flame ‘and the products of the combustion and proved that metal or porcelain dishes were equally effective in producing these phenomena. Boutigny gives the temperature a t which ether begins to burn with this flame as a little below that of fusing lead and so agrees with Perkin and others that the temperature necessary must be about 260°.As will be seen later however a temperature much below this is sufficient to produce this flame in ether-air mixtures. According to Perkin this blue flame has a comparatively low temperature (it has since been designated a ‘(cool” flame). The fingers may be placed in it with impunity. It will neither char paper nor ignite carbon disulphide and a lucifer match may be held in it for some time before being ignited. He also states however that ether vapour burning with this blue flame when in large quantities or more especially in a confined space rapidly increases in tempera-ture and quickly enters into ordinary combustion. Perkin also examined other substances for (( the luminous appear-ance accompanying incomplete combustion .” Only traces of blue flame were obtained with the alcohols up to amyl methyl alcohol giving none.The generally accepted definition of ignition-temperature is that temperature to which a gas-mixture must be heated a t least locally, for the speed of the reaction to be such as to become self-suppor 1466 WHITE AND PRICE THE IGNITION OF ing. This temperature is not that at which a flame appears but that a t which self-heating becomes sufficient to cause ultimate inflammation. That the determination of an ignition-temperature is a matter of great difficulty can a t once be seen when it is considered that the temperature a t which such a reaction would become self-sup-porting must depend on the rate of dissipation of heat in the system as well as on many other factors.For instance if the time taken to bring the gas-mixture up to the ignition-temperature is appreciable the composition of the gas-mixture alters and i f a solid is in contact with the heated gases even below their ignition-temperature as proved by Bone and Wheeler (Phil. Trans. 1906, [ A ] 206 I) and by Meyer and Preyer (Ber. 1892 25 622) the most divergent results are obtained for the amount of combination that takes place. From their work the German investigators concluded that it was impossible to determine an ignition-temperature. We decided to try the soap-bubble method described by McDavid (Zoc. c i t . ) and if that proved unsatisfactory as was anticipated from Meunier’s work (Conzpt.rend. 1907-1912) to attempt to make use of a method in which the amount of heated surface brought into contact with the gas would be known thereby elimin-ating the most obvious defect in the soap-bubble method as pub-lished. To this end it was decided to pass the various gas-mixtures of which the ignition-temperatures were required into certain uniformly heated vessels. I n this way the temperature a t which ignition could be obtained in each vessel would be known. By taking vessels of the same material having different ratios of surf ace to volume and plotting the ignition-temperature against surface per unit volume it was anticipated that by ext.rapolation it would be possible to eliminate to a great extent disturbing variations due to surface action.The most obvious vessels to use were tubes of various diameters and a series of these was accord-ingly chosen. It was quite realised that the longer time taken to heat a large bulk of gas would affect the results t o some extent but direct experiment in which some of the products of combustion were introduced into the gas-mixture to be used soon proved this to be almost negligible in the case of ether-air mixtures of ordinary concentration. It is to be noticed however that in this method the temperature determined has been called the sub-ignition-temperature-the minimum temperature a t which combination in a gas-mixture becomes self-supporting. This appears t o be the practical temperature required as the phenomenon obtained may or may not give rise to ordinary combustion depending on circum ETHER-ALCOHOGAIR AND ACETONE-AIR MIXTURES ETC.1467 stances. Its identity with the ignition-temperature of a mixture depends entirely on what is understood by “ignition,” and a “flame ” in those cases in which cool flames are possible phenomena. Heating a gas-mixture quickly to its ignition-temperature is always assumed t o cause ultimate inflammation. In many cases, an ordinary flame causing more or less complete combustion cannot be obtained by heating a gas to its sub-ignition-temperature and in those cases in which such a flame appears it is only produced through the cool^' flame of incomplete combustion. For a fairly concentrated ether-air mixture the cool flame obtained is very similar in appearance to an ordinary flame but for very dilute mixtures it becomes practically a travelling phosphorescent flow.The Soap-bubble Method.-The results obtained during our investigation of the soap-bubble method of determining ignition-temperatures have already been published (this vol. p. 1248). By using different; igniting surfaces it was found that the ignition-temperature of a 5 per cent. ether-air mixture as determined by this method could vary from 907O to 1064” whilst that of a 12 per cent. mixture could vary from 870° to 1035O. The results obtained seemed to be erroneous and by using several gas-mixtures it was shown that the method could scarcely be trusted even for com-parative results. The Exhausted Tube Method.-The apparatus used in this method is shown in Fig.1. The heated vessel consisted of a long glass tube sealed a t one end and closed at the other by means of a rubber stopper. The tube could be kept a t any desired tempera-ture by means of an electric furnace the exact temperature inside the tube being registered by means of a copper-constantan couple except when that temperature was more than 500* when a nitrogen-filled mercury thermometer was used. The ignition tube was connected to the glass reservoir containing the mixture under examination by means of a glass lead passing through the rubber stopper. A three-way tap was inserted between the tube and reservoir in such a manner that the tube and lead could be con-nected a t will to the reservoir or to a Gaede box pump. The reservoirs were of 15 to 17 litres capacity and the tubes used a t first were a few cm.longer than the furnace which was 50 cm. long. The reservoir was filled with any required mixture by exhausting it and allowing air to sweep a known weight of the solvent from the filler shown in the figure. The filler was con-nected to the three-way cock by means of rubber tubing but the end of the filler always projected into the tap tube. In this way, none of the solvent escaped introduction into the reservoir and a knowledge of the molecular weight of the solvent together with VOL. uxv. 3 1468 WHITE AND PRICE THE IGNITION OF the temperature and pressure gave by means of a simple calcula-tion the percentage volume occupied by the solvent in the reservoir. A portion of the lead between the three-way tap and the furnace was connected to the remainder by means of rubber joints.These joints enabled tubes of various internal diameters to be introduced in order to vary the rate a t which equalisation of pressure in the tube and reservoir took place. An experiment was conducted as follows. The tube was exhausted to a pressure below 2 cm. and connected t o the reservoir by turning the three-way cock as rapidly as possible; an observer looking through the FIG. 1. c: Mt4L f Yo& T n CTCR sealed end of the tube (shielded by a plate of glass) reported whether ignition had or had not taken place. Shock Zgnition.-Preliminary experiments using 5 to 15 per cent. mixtures of ether in air and a tube 2 cm. in diameter gave results varying with the diameter of lead used.It was also found that changing the length of the tube from 50 to 100 cm. affected the temperature a t which ignition was obt.ained. These irregulari-ties were presumably due t o the differences in time taken to fill the tube with the gas-mixture. Accordingly a tube 7.5 cm. in diameter was substituted for the one previously used so that the effect of changing the diameter of the lead could be investigated more easily. With a lead of 1 mm. diameter ignition took plac ETHER-ALOOHOGAIR AND AUETONE-AIR MIXTURES ETC. 1469 a t temperatures near 200° but on using a lead of 5 mm. in diameter it was found possible to obtain ignition at from 50° to 60°. These resullx were obviously too low and observation showed that they were influenced by the position of the end of the tube relative to the furnace.Experiments were carried out to elucidate this phenomenon and it was found possible to ignite ether-air mixtures a t the ordinary temperature and to ignite other gaseous mixtures a t temperatures well below those commonly considered as their ignition-temperatures. The apparatus used consisted of a glass tube 7.5 crn. in diameter which was connected to a reservoir of 16 litres capacity by means of a glass lead 80 cm. long and of 1.9 cm. internal diameter. By using on the reservoir a cock of 1.5 cm. bore it was possible to equalise the pressure in the two portions of the apparatus very suddenly. Under these conditions, it was found that when the tube and lead were exhausted and kept a t the same temperature as the reservoir (16*5O) ether-air mixtures containing 5 t.0 15 per cent.of ether were ignited on opening the reservoir cock. The ignition invariably took place within 15 cm. of that end of the tube remote from the reservoir. In most cases it resulted in a pale blue flame which travelled quietly along the tube but sometimes gave inflammation sufficiently violent to shatter the glass. When using 7.5 cm. tubing ignition could be obtained easily with 60 and 90 cm. lengths even when the pressure in the reservoir was less than half the atmospheric. Ignition occurred in a 150 cm. tube only when the pressure in the reservoir was greater than three-quarters of an atmosphere. When a 300 cm. tube was used no ignition could be obtained at the ordinary temperature ; a similar negative result was obtained with bottles 30 cm.long and 11.2 to 12.5 cm. in diameter that is of approxim-ately the same capacity as the 90 cm. h b e . A pad of soft leather in the closed end of a tube of optimum length seemed t o prevent ignition a t the ordinary temperature and a plug of cotton wool prevented ignition in precisely the same way. Replacement of the 7.5 em. tube by one 2 cm. in diameter brought about the same result. It appeared to be immaterial whether highly purified ether or the ordinary commercial variety was used for these experi-ments. Amongst the other gases tested were mixtures of hydrogen and the vapours of acetone and carbon disulphide with air. Dilute carbon disulphide-air mixtures ignited a t the ordinary temperature acetone-air mixtures below 250° and hydrogen-air mixtures below 450° but these experiments were not continued.The information a t present available makes it appear highly prob able that this ignition is due to the shock caused by the sudden 3 1 1470 WHITE AND PRICE THE IGNITION OW stoppage of the gas rushing into the exhausted tube. In this con-nexion a statement made by Sir Charles Parsons in his inaugural address a t the Bournemouth Meeting of the British Association (1919) is of interest. This was to the effect that during the work of a committee appointed by the Admiralty in 1916 to investigate the cauge of abnormal propeller erosion it was discovered that by allowing water to rush into an exhausted conical vessel a pressure of more than 220 kiloa.per sq. mm. was recorded a t the apex of the cone. That ignition of a gas-mixture can be produced by a compression wave was demonstrated by Bradshaw (Proc. Roy. SOC., 1907 [ A ] 79 236) for mixtures of carbon disulphide in oxygen and for electrolytic gas. Another factor the actual rarefaction must not be forgotten. Investigations by Mitscherlich (Ber. 1893 26 399) on the temperature necessary to explode mixtures of hydrogen and oxygen seem to show that the explosion point is reduced very appreciably by lowering the pressure. Again Labillardiere Friedel and Ladenburg Stock and Guttmann and others have shown that the temperature necessary for the ignition of mixtures of the hydrogen compounds of phosphorus silicon and antimony with oxygen is presumably lowered by reducing the pressure and that explosion has been known to follow sudden rarefaction.In the cour8e of the work described in this paper the maximum reduction in sub-ignition-temperature of ether-air mixtures apparently obtained by reduction in pressure alone was only 7O that is from,187° to 180° as shown in Fig. 4. This appears to leave a fair margin for other factors. The importance of this “shock ’’ ignition is obvious whether it be considered from the theoretical or practical point of view and it may quite well account for hitherto obscure ignitions met with in the course of solvent-recovery and mine work. The phenomenon underlying shock-ignition must also invalidate a good deal of research work. A t t empt to E Zimina t e Sh ock-ignition .-One conceivable method of avoiding shock-ignition in the determination of sub-ignition-temperatures would appear t o be that in which the gas is given no appreciable flow before being stopped.An attempt t o realise these conditions was made by joining a small bulb on to a gas reservoir in such a way that the distance between the bulb and reservoir was as small as possible. A three-way tap was inserted between the reservoir and bulb to enable the bulb to be exhausted before an experiment. The bulb which was m‘ade of glass was kept a t any desired temperature by being almost entirely immersed in a bath of mercury. When once a connexion had been mad ETHER-ALCOHOL-AIR AND ACETONE-AIR MIXTURES ETU. 147 1 between the bulb and gas reservoir the tap was turned so as to close both bulb and reservoir.I f ignition took place a flash was easily observed where the bulb-stem emerged from the bath. The bore of the bulb-stern was varied in different determinations. The results obtained when using a 4.5 per cent. mixture of ether in air are given in table I. From them it will be noticed that the effect of the size of lead is not completely eliminated by this method. TABLE I. Showing the Subignition-temperatures of a 4.5 per cent. Ether-Air Mixture obtained by the Bulb Method using Various Leads. Sub-ignition-temperature when ,lead was L Diameter r- . of bulb in cm. Ordinary tube. Capillary. Fine capillary. 4.8 178.0" 181.0" 184.0" 4.1 179.0 184.0 186.0 3.5 180.5 185.5 188.0 An interesting point observed in these experiments was the fact that when using a 4.5 per cent.mixture near its apparent ignition-temperature explosion invariably occurred. On the other hand, when using the bulbs specified above with an ordinary capillary lead and the same gas-mixture on no occasion did an explosion take place when the temperature of the bath was greater than 1 9 7 O . Above this temperature a luminous flash was observed and nothing more. A 6 per cent. mixture gave simiIar results but a 10 per cent mixture gave no explosion a t any temperature tried. An 8 per cent. mixture behaved in precisely the same way as a 10 per cent. mixture except that on one solitary occasion a violent explosion shattered the bulb. It is possible that a good approxim-ation to the correct sub-ignition-temperature could be obtained by using a very fine capillary tube and a fairly large bulb but the experiment would not be without danger.Final Apparatus.-It was found that by making use of a fairly long tube and allowing the sealed end to project well out of the furnace effects of shock-ignition were apparently eliminated. I n this case the point a t which shock-ignition would have occurred in normal circumstances was well outside the heated zone and on no one occasion was an ignition in the final apparatus observed to start outside the furnace. The lengths of tube used were as follows: 2 cm. tube. 100 cm. 4 cm. tube. 130 cm. 6.6 cm. tube. 130 cm. The chief difficulty encountered was that of deciding when an This was generally easy in the case of ignition had occurred 1472 WHITE AND PRICE THE IGNITION OF shock-ignition as not only was the flame fairly easily seen but the products of combustion had a characteristic and powerful odour.In the case of dilute mixtures of ether in air and ether-alcohol-air mixtures containing small quantities of ether however the matter was quite otherwise. It was found almost impossible to distinguish between an ignition and the glow given below the sub-ignition-temperature by combustion on the surface of the glass until travel-ling was taken as the criterion. This was of course due to the fact that the ignition a t the lowest possible temperature of a mix-ture containing ether and air invariably commenced with what has been termed the cool flame. This flame often requires a completely darkened room in order to be visible but whenever any appreciable quantity of combustible mixture is present it is liable t o develop more or less rapidly into ordinary combustion and possibly, detonation.That the volume of mixture present is an important factor can be seen from the fact that on no occasion did ordinary combustion develop in the 2 cm. tube within 150° of the sub-ignition-temperature. On the other hand when using mixtures containing from 5 to 10 per cent. of ether in either of the other tubes ordinary combustion was liable to develop and certainly did develop if the temperature was a few degrees above that necessary for inflammation. Test experiments carried out with and without the thermo-couple and thermometer in the tube showed that the presence of these instruments did not appear to affect the result obtained.Fig. 2 gives the results obtained for the sub-ignition-temperatures of various ether-air mixtures. It will be noticed that in the case of the 2 cm. tube the results for dilute mixtures differ slightly according to the lead used. This is presumably due to the fact that combustion takes place to a relatively greater extent in the case of the smaller lead making i t more difficult to see the flame a t the same temperature. On the other hand above a certain limit the internal diameter of the lead does not appear to affect the results obtained with the other tubes. Surface action appears to be negligible in the case of ether-air mixtures of any appreciable concentration if the tube has a diameter of a t least 4 cm.That the longer time taken t o heat the larger bulk in a wide tube made little difference in the case of ether-air mixtures of concentration greater than about 4 per cent. was proved by adding 1.5 per cent. of the products of combustion of an ether-air mixture to a 5 per cent. mixture of ether in air. The sub-ignition-temperature was only raised 2O. It was found that no difference in sub-ignition-temperature could be detected when purified ether was replaced by the commercial article. It is interesting to note from the shap ETRER-ALCOHOL-AIR AND ACETONE-AIR MIXTURES ETU. I 473 of the curve in Fig. 2 that the ignition of ether-air mixtures by this method depending as it does on the preliminary production of the cool flame appears to be a molecular process requiring only a certain intensity of molecular movement for its production.The form of curve connecting minimum igniting current with composi-tion of ether-air mixture appears to be quite different. I n Fig. 3 can be seen the sub-ignition-temperatures of certain ether-alcohol-air mixtures as determined in the 2 cm. and 4 cm. FIG. 2. x Percentage of ether in mixture. A. 2 cm. tube 1 and 3 mm. leads. B. 4 cm. tube 3 mm. lead. C . 5& cm. tube 3 mm. lead. tubes. 'The number of points shown on each curve is not great, owing to the fact that the general form of the curves had been previously found when using smaller leads. Each point on the diagram is the mean of three determinations agreeing t o within 20 in the case of temperatures below 220° and to within 5 O in the case of temperatures above this point.A consideration of the results given shows that the sub-ignition-temperature of an ether-alcohol-air mixture falls very rapidly when the amount of ether in the mixture is increased from 1 to 1474 WHITE AND PBICE THE IGNITION OF per cent. This drop is seen to take place with a smaller per-centage of ether in the case of the 4 cm. tube and in both tubes with a smaner percentage of ether in the case of the mixture con-taining the smaller amount of alcohol. The curves also indicate the manner in which the addition of alcohol to an ether-air mix-ture affects the sub-ignition-temperature. The elevation in sub-ignition-temperatures is roughly proportional to the amount of alcohol present provided the ether-content lies between 2 and 5 per FrG.3. 480" 180 Percentage of ether in mixture. cent. tubes are given in t.able.11. For comparison some results obtained with 4 and 5.5 em. TABLE 11. Showing the Sub -ignition- t enzperatures of Various E t her-A lcohol-Air Miztures containing 2 per cent. of Alcohol as determined in 4 and 5.5 em. Tubes. Sub-ignition-temperature for > Percentage of 5-5 cm. tube. ether in - 4 cm. tube, mixture. 5 mm. lead. 3 mm. lead. 3 mm. lead. 1 495" 500" 470' 2 220 222 217 3 207 210 203 t 201 205 19 ETHER-ALCOHOGAIR AND ACE)TONEI-AfR MIXTURES ETC. 1476 is 4 The sub-ignition-temperature determined with the 5.5 cm. tube seen to be higher in every case than that obtained with the cm. tube. This is probably due t o the fact that owing to the larger diameter of this tube the time of heating is necessarily longer and hence slow combustion occurs more readily.With this tube too a slight but perceptible difference in sub-ignition-temperature is obtained when using 3 mm. and 5 111111. leads as shown in table 11. It was seen that in the case of dilute mixtures of ether in air the size of lead used if small affected the temperature obtained. This effect was exceedingly marked in the case of ether-alcohol-air mixtures containing little ether. With small leads it was found impossible to obtain consistedt results. For instance for a mix-ture of 2 per cent. of alcohol and 1-25 per cent. of ether in air, when using a 1 mm. lead with the 4 cm. tube a flash would occasionally be obtained below 300° but one could only be certain of an ignition well above 400O.It was found however that if the cock connecting the tube and reservoir was turned on slowly, no ignition was ever obtained below 400". These differences vanished when leads of 3 mm. or more were used. The sub-ignition-temperatures of various alcohol-air mixtures are given in table 111. TABLE 111. Showilzg the Szct,-igrtition-temperatzlres of Vapiozcs A tcohol-Air Mixtures as determined in 2 em. 4 em. and 5.5 ern. Tubea. Sub-ignition-temperature for Percentage of F A I aloohol in 2 crn. tube 4 cm. tube 54 cm. tube mixture. (3 mm. lead). (3 mrn. lead). (6 nun. lead). 2 515" 500" 620" 3 505 490 605 4 486 470 600 6 480 465 496 I n the 4 cm. tube an explosion was often obtained with 4 and ij per cent.alcohol-air mixtures a t 485O. The evidence as to the slower heating of the gaseous mixture in the largest tube is con-firmed by the figures given here as the temperatures found in the case of the largest tube are higher than those for the smallest,: It thus appears from the above figures that the nearest possible approach to the correct sub-ignition-temperature of an ether-alcohol-air mixture is obtained by means of a 4 cm. tube. Itq was found that almost identical results were obtained in the 2 cm. and 4 cm. tubes whether 2 or 3 mrn. leads were used and that it was not important whether a 3 or 5 mm. lead was used for the 3 I 1476 WHITE AND PRIUE THE IUNITION OF 5.5 cm. tube. The sub-ignition-temperatures considered most likely to be correct are summarised in table IV.TABLE IV. Showa7tg the Sub-ignitiomtemperatures of Various Ether-Alcohol-Air Mixtures. Percentage composition of mixture by volume. c -Ether. 3-16 8 4 3 2 1 8 4 3 2 1 0 0 0 0 Alcohol. 0 2 2 2 2 2 4 4 4 4 4 2 3 4 5 3 Air. 97-86 90 94 95 96 97 88 92 93 94 95 98 97 96 95 Sub-ignition-temperatures. 187' 189 198 203 217 470 194 200 220 255 465 600 490 470 465 The sub-ignition-temperatures of some acetone-air mixtures as determined in tubes of various diameters were found to vary in a manner similar to those of alcohol-air mixtures. The results given below were obtained when using a tube 4 cm.in diameter with a 3 mm. lead. TABLE V. Showing the Sub-ignition-Temperatures obtained for some A c e t one-A ir Mix t ur es . Percentage of acetone Sub-igni tion-in acetone-air mixture. temperature. 4 500" 8 500 Saturated at 15" 505 The ignition found for acetone was very faint a t the sub-ignition-temperature but grew in intensity very rapidly as the temperature of the tube increased. ZnfZuence of Pressure .-During the course of preliminary work on the determination of sub-ignition-temperatures of ether-air mixtures it was found that consistent results were not obtained for successive experiments when the mixture in the reservoir was not renewed after each determination. This was apparently due to change of pressure inside the reservoir.Several experiment ETHER-ALCOHOL-AIR AND AOETONE-AIR MIXTURES ETC. I477 were therefore made in which the pressure in the reservoir was reduced by pumping out the gas-mixture before firing and it was found that when the pressure in the reservoir after an experiment was plotted against the sub-ignition-temperature determined the curves shown in Fig. 4 were obtained. Other mixtures tested in the same way gave precisely similar resulk and it was finally found FIG. 4. 190 180 Pressure in cm. that the easiest method of determining the accurate sub-ignition-temperature of any given mixture under 760 mm. pressure was by starting with a gas-mixture under about 900 mm. pressure and plotting a small portion of the pressure-sub-ignition-temperature curve.It will be observed that the minimum sub-ignition-temperature The results given in Fig. 2 were obtained in this way. 3 J" 1478 WHITE AND PRIOB! TBE IGNITION OB’ appears to be given a t a lower pressure in the cam of a mixture rich in ether than in the case of a dilute mixture. This is prob-ably due to the fact that there is a minimum quantity of ether per unit volume necessary to give visible luminosity under any given conditions. In@aence of the Material of tlte Tube.-A consideration of the results obtained with various tubes shows that it was impossible to eliminate surface action entirely probably owing to the fact that in the case of the larger tube the heating of a body of gas is necessarily slower. It thus became a matter of importance to dis-cover whether the material of the tube had any influence on the result obtained.The simplest method of effecting this appeared to be by fitting a glass tube with a thin sleeve of the material under test. Accordingly the 4 cm. tube was fitted with sleeves long ,enough to project beyond the furnace on either side these sleeves being made of various metals that might conceivably be used in a manufacturing plant in the presence of gas-mixtures such as those considered. The results obtained are giSen in table VI. TABLE VI. Showing the Results obtained f o r the Sub-ignition-temperature of Ether-Air and Alcohol-Air Mixtures in a 4 cm. Tube provided with an Internal Metallic Sleeve. Sub -ignition-temperature. Material of 4.3 per cent. sleeve. of ether in air.Glase ........................... 187’ Copper ........................ 175 Iron ........................... 178 Lead ........................... 180 Zinc ........................... 184 Galvanised iron ............ 184 10.5 per cent. 5 per cent. of ether in air. of alcohol in air. 187O 466O 176 420 178 400 180 -184 e 184 -The figures given for copper and iron in the case of the alcohol-air mixture can only be regarded as rough approximations owing to the rapidity with which these metals oxidised at the tempera-ture necessary for ignition. When the metal was oxidised to any appreciable extent different results were obtained. For instance, the sub-ignition-temperature in the case of a copper sleeve oxidised in the course of ten experiments was 4 7 0 O .In the case of both ether-air and alcohol-air mixtures the ignition commenced as a cool flame but was invariably more violent in the presence of metals more particularly copper and iron than with glass. Influeme of the Velocity of t h e Gas-mixtu~e,-As the gases dealt with in solvent-recovery are generally in motion it wa ETHER-ALCOEOL-AIR AND AOETONE-AIR MIXTURES ETC. 1479 decided to investigate to some slight extent the effect of the velocity factor on the sub-ignition-temperatures obtained. Arrangemente were therefore made by which the gas-mixture in a reservoir could be displaced by means of water and the gas from this reservoir made to displace the contents of a second reservoir. The gas-mixture from this second reservoir was passed into a 4 cm.tube 120 cm. long which was kept a t any desired temperature by means of an electric furnace. The far end was partly closed by means of a thick glass plate. It WELS found that once a steady state had been attained the velocity of the gas in the tube could be measured sufficiently accurately by estimating the rate a t which water was introduced into the first reservoir. An observer looking through the plate glass could easily see if ignition occurred. The aub-ignition-temperature for zero velocity was taken to be the lowest temperature at which ignition occurred after the gas aupply had been cut off. The results obtained for two ether-air mixtures are given below in table VII. TABLE VII. Showing the Effect of the Velocity of the Gasmixture &owing through a 4 cm.Tube on the Szlbignition-te?nperature observed. 4 4 per cent. of ether in air. 14 per cent. of ether in air. - - Velocity in Sub-ignition- Velocity in Sub-ignition-cm. per second. temperature. cm. per second. temperature. 0 1 8 7 O 0 186" 6.0 195 1.0 189 12.6 202 5-5 195 - - 8.0 197 - 13.0 202 -The experiments carried out were sufficient to indicate that for very small velocities increase of velocity causes an elevation of the sub-ignition-temperature observed. The velocities dealt with on the manufacturing scale are however of a totally different order, ranging from 100 to 400 cm. per second in various pipes. Influence of the Presence of Glyceryl Trinitrate Ethyl Hydrogen Peroxide and Diethyl Peroxide in the Ether Used.-Several attempts were made t o find if the presence of glyceryl trinitrate in an ether-air mixture affected the sub-ignition-temperature.I n no case was any such effect discernible. I n the experiments for which the results are given in table VIII the glyceryl trinitrate was introduced into the reservoir by passing the air used for making up the mixture through a calcium chloride tube in which glyheryl trinitrate was spread over glass wool the tube and contents bein ‘1 480 WHITE AND PRICE THE IGNITION OF kept a t 40° to 45O. Resulta obtained in a 2 cm. tube were similar to those given below which were determined ’by using a 4 cm. tube. TABLE VIII. Showing the Effect of the Presence of Glyceryl Trinitrate Ethyl Hydrogen Peroxide and Diethyl Peyoxide on the Sub-ignition-temperature of Ether-Air Mixtures.Composition of Mixture. Sub-ignition-temperature. ........................... 5.3 per cent. of ether in air 5.3 per cent. of ether and 1.5 per cent. of diethyl 4-7 per cent. of ether and 0.5 per cent. of ethyl 6.3 per oent. of ether in air saturated with glyceryl 187” peroxide in air .................................. .;... 189 3-7 per cent. of diethyl peroxide in air ............ 189 hydrogen peroxide in air ........................... 182 trinitrate at 20” ....................................... 187 The presence of glyceryl trinitrate did not appear to affect the flame given by the mixture but the presence of the peroxides caused a very fierce flame and generally an explosion. It was considered inadvisable to try to determine the sub-ignition-temperature of ethyl hydrogen peroxide in air.The per-oxides were found to be exceedingly dangerous to handle; even diethyl peroxide exploded violently on one occasion during distillation. SECTION Ir. The Limits of Propagation of Flame in Ether-Alcohol-Air and A cet one-A. ir Miz t ures. Many references are to be found in the literature to the limits of inflammability of mixtures of ether and alcohol with air and some figures are also given for acetone-air mixtures. The limits determined by various workers are given below. Ether-Air Mizture.-Limits of inflammability. first edition 1912 p. 73). m n ! Ghem. Eng. 1916 14 190). 2.7 t o 7.7 per cent. by volume (Brunswig “Explosives,” 50 to 60 grams per cubic metre for lower limit (Marchis Met.2.9 to 7.5 per cent. by volume (Lewes J . Soc. Arts 1915 761). 2.9 to 7.5 per cent. by volume (Schwartz “Fire and Explosion 0.058 to 0-195 gram per litre (Meunier Compt. rend. 1907, Risks,” first edition p. 35). 144 1107) ETHER-ALUOHOL-AIR AND ACETONE-AIR MIXTURIOS ETU. 1481 A 1cohoLAir iKixture.-Limib of inflammability. 4.0 to 13.7 per cent. by volume (Brunswig Zoc. cit.). 4.0 to 13.6 per cent. by volume (Lewes Eoc. cit.). 3.95 to 13.65 per cent. by volume (Bunte and Eitner J . Gaa-3 to 8.4 per cent. by volume (Thornton PTOC. Roy. Soc. 1914, [ A ] 90 280). A cetone-Air Mizture.-Limits of inflammability. beleucht 1901 44 835). 5 to 12 per cent. by volume (Brunswig loc. cit.). 2.15 to 9.7 per cent. by volume (Wheeler and Whitaker T., 1917 111 267).It will be seen that the results obtained vary considerably owing t o the different conditions under which the experiments were carried out and the various igniting sources used. This is to be expected, as the conditions governing the propagation of flame were not properly appreciated until recent years. The definition now adopted is that suggested by Coward and Brinsley (T. 1914 105, 1859) in which inflammability is regarded as a specific property of a mixture independent of the size and shape of tho vessel in which it may happen to be contained and also of any particular type of igniting arrangement. They propose to define a gaseous mixture as inflammable per se a t a stated temperature and pressure if and only if it will propagate flame indefinitely the unburnt portion of the mixture being maintained a t the original tempera-ture and pressure.On this definition inflammability is a property of the mixture itself although a function of the temperature and pressure. Dilution-limits however rarely vary much throughout the usual range of variation of laboratory temperature and pressure. It will thus be seen that in order to obtain satisfactory results in the estimation of dilution-limits it is necessary to use a vessel (1) of such size that any cooling of the gas-flame by the walls can be neglected and (2) of sufficient length to enable a sound judgment t o be made as to whether a flame would propagate indefinitely or no. For every gas-mixture examined propagation-limits were deter-mined for three directions-upward horizontal and downward.Preliminary work seemed to indicate that in a glass tube 5 cm. in diameter by using a sufficiently powerful initiator it; was possible to force a flame through a mixture below the limit of pro-pagation to the end of the tube unless it was a t least 120 CM. long. All the tubes for limit work were therefore made a t least 150 cm. in length. The tubes consisted of: (1) Glass tubes 2.5 cm. in diameter 1482 WHITE AXD PRICE THE IGNITION OF (2) Glass tubes 5 cm. in diameter. (3) An iron* tube 5 cm. in diameter. (4) An iron tube 15 cm. in diameter and 300 cm. long. The glass tubes and the 5 cm. iron tube were 150 cm. long. The glass tubes were closed a t both ends by gas-tight stopcocks of 4 mm. bore and a similar cock was fitted t o one end of the iron tube, where a 5 cm.glass observation piece was cemented for observa-tion purposes. The cocks in the case of the 15 cm. iron tube were of brass and of 6 mm. bore. I n this tube inflammation was observed through three equidistant windows of thick plate glass, which were cemented into holders on the tube. These windows gave much trouble and it was found impossible to render them gas-tight by using cement alone. Accordingly caps were soldered over each window in such a way that by inserting a 3.7 cm. rubber stopper into the observation hole left in the cap the whole apparatus could be made gas-tight. The stoppers were removed immediately before firing. This tube was filled by means of a filler similar to that shown in Fig. 1 the end of the filler being made to project well through a tightly fitting piece of rubber tubing drawn over one of the cocks.The filling was carried out precisely as described previously a correction being always applied for the vapour of the solvent present in the air above the liquid in the filler. In the case of all the smaller tubes the fillers were provided with ground-glass joints fitting pieces sealed on to the tubes concerned. All the air passed in t o make up any mixture containing alcohol or acetone was carefully dried by passage through a calcium chloride tube. The calibration of the tubes was carried out by weighing the quantity of water necessary to fill them. Ignition was effected by passing a spark from an induction coil between two electrodes of stout platinum wire separated by an air gap of 1 cm.the current being obtained from six 2-volt accumulators. I n the glass tubes originally used the platinum was sealed through the glass but as good sealing glass became scarce, this was found to be impracticable and the electrodes used for all the tubes consisted of platinum in glass mounted in rubber stoppers. The original method of mixing the gases was by allow-ing the tube to remain for several hours but this became in-advisable when rubber was brought into contact with the solvent-laden air. A little mercury enabled efficient mixing to be carried out by shaking the tube but i t was found that this affected the results obtained and finally small glass beads were used. Corn-* This and the other iron tubes used in the investigation consisted of terne-Plate that is sheet-iron coated with tm alloy of lead and tin ETHER-ALCOHOL-AIR AND AOETONE-AIR MIXTURB~B ETU.1483 parative tests showed that when using an adequate number of beads Hhaking.a 5 om. tube for twenty minutes gave satisfactory mixing It was alao shown that under these conditions the same results were obtained whether the electrodes were held by small rubber stoppers or were sealed through the glass. Throughout the course of the work the only tube that caused trouble was the 5 cm. iron tube. The various cements used for fastening the glass observation cap to the main body of the tube seemed to hold solvent and the results obtained for this tube cannot be conaidered FIG. 5. 4 A. as trustworthy as those obtained with the others.The mixing of the content of the 15 cm. iron tube was done very efficiently by rolling a 12.5 cm. perforated hollow copper ball from end to end. The apparatue used for determining the upper limit of propa-gation for alcohol-air and certain ether-alcohol-air mixtures con-sisted of a 5 cm. glass tube jacketed by enclosure in a wider glass tube so that hot water could be continuously circulated round it. The arrangement used i s shown in Fig. 5. Two sets of this apparatus were fitted up one as shown in th 1484 WHITE AND PRICE THE IGNITION OF sketch arranged for experiments on downward propagation with the electrodes at the same end of the tube as the ground-glass joint for filling the tube and the other arranged for upward propaga-tion in which the electrodes were a t the end away from the ground-glass joint.For horizontal propagation either of the above tubes was used and placed horizontally before firing. Some difficulty was encountered in sparking the mixtures contained in this apparatus but by enclosing the leads in glass tubes this was finally overcome. Naturally mixing in these tubes could only be accom-plished by allowing the tube to remain for some time. The procedure in the case of any mixture can be seen by a consideration of the results given below. Experiment 21.-Glass tube 5 cm. in diameter. Solvent mixture used 75 per cent. of ether and 25 per cent. of alcohol (by weight). Lower limit downward propagation. Temperature 19O. Percentage of solvent-vapour in gas-air mixture (by volume).2-70 ................................. Complete ignition. 2.40 ................................. Flame just started. 2-00 ................................. Complete ignition. 2-50 ................................. Partial ignition. 2.55 ................................. Flame went nearly to 2.57 ................................. Flame went very LimiC2.57 per cent. the end. slowly to the end. I n the case of a lower limit an accuracy of 0.02 per cent. was aimed a t ; in the case of an upper limit 0.05 per cent. was taken. I n every instance just before firing the cock furthest removed from the electrodes was opened ta allow a free passage for the gases. When the limit of propagation was being determined for a mixture of ether and alcohol a liquid containing the requisite proportions of these two solvents was made up and used.A test experiment showed that this gave the same result as was obtained when the two solvents were weighed into the tube sepmately. Ether-AZcohol-Air Mixtures.-The results obtained for ether-alcohol-air mixtures in glass tubes are shown in table IX. The experimental results obtained with the 2.5 cm. tubes are not so trustworthy as those determined in larger tubes as can be seen from the results themselves. For example the result obtained for the lower limit of an ether-air mixture is least for downward propagation and other anomalies could be pointed out in the same way. These irregular results were probably due to the fact that it was only for downward propagation that the flame travelled more or less steadily.For upward propagation it was sometime TABLE I X . Showing the Limits obtained for the Propagation of Flame in Ether-in Glass Tubes. The limit figures given show the percentage volume occupied by Direction Percentage composition of upw/ards. Horizontally. mixture (by weight). Diameter - b -100 0 2-5 18.50 2.35 6.16 75 25 YS 6.95 3.15 7-35 Y Y - 3-52 -Y9 - 4.35 -1 Y - 5-02 -100 0 5 15-75 1.93 8.00 75 25 Y 11.70* 2.40 10.95* 60 50 1 ) 10.70* 2.89 10*36* 25 75 9 ) 12*00* 3-53 11-50* 0 100 I t 18*95* 4-24 13*80* Lower upper limit. limit. Of tube upper Ether. Alcohol. in cm. limit. Per cent. Per cent. Per cent. 50 50 25 75 0 100 * The figures marked thus were determined in the jacketed tube at 60° and are, the others whioh were determined at air-temperatur 1486 WETTE AND PRIOE THE IGNITION OIT obviously jerked out.Wheeler’s work has shown that only a very slight change in the limits takes place when the diameter of the glass tube is increased beyond 5 cm. so that this was the largest size of glass tube used. His work however was chiefly carried out with permanent gases. The most noticeable fact brought out in table IX appears to be the great difference between the results obtained for the upper limit for upward propagation and those obtained for propagation in other directions. These differences which are normally due to convection currents set up by the flame are not so great for per-manent gases. The difference bet.ween the upper limits for hori-zontal and for upward Propagation is most marked and it almost appears as if 8 different type of propagation were brought into being.It is quite conceivable that the heated gases rising in the tubs are responsible for the initiation of a cool flame. The difference is not so great in the case of alcohol as it is for ether. The results for ether-alcohol-air mixtures in a 5 cm. glass tube are shown in Fig. 6. The results given for the upper limits were all determined a t 60°. The graphed results for the lower limits and the upper limit for downward propagation form good approximations to straight lines but this is not the case for the other two curves A t first sight it seems very strange that for the upper limit for upward propagation the results for a mixture containing equal weights of ether and alcohol should be more than 6 per cent.less than the corresponding figure for either ether or alcohol. A possible explanation appears to be afforded by an examination of the sub-ignition-temperature curves given in Fig. 3. From these it will be seen that the addition of alcohol to an ether mixture raises the sub-ignition-temperature so that the slope of the two upper curves near the 100 per cent. ether point is in the direction to be expected. Similarly the slope near the 100 per cent. alcohol point can be explained when it is remembered that it takes an appreciable quantity of ether to cause any deoided lowering of the sub-ignition-temperature of an ether-alcohol-air mixture. That there is no apparent irregularity corresponding with that seen when the percentage of ether is between 1 and 2 per cent.on the sub-ignition-temperature curves may be due to the fact that ao few points have been determined on the limit curve; but it is far more likely to be due to the fact that the sub-ignition-temperature obtained for an ether-alcohol-air mixture containing say 2 to 2.5 per cent. of ether is due solely to the ignition of the ether present and that the alcohol takes little part in the reaction. These remarks do not apply to curve C Fig. 6 ETHER-ALUOHOGAIR dlqg AOETONB-AIR MEETORBIS ETC. 1487 aa in that case the flame obtained is undoubtedly that of ordinary combustion. The limits of inflammability of a mixture of two or more gases with air are connected with the limits of the components of the mixture by Le Chatelier's rule which states that if n d n" .. FIG. 6. 0 Percentage of ether. Pwcetttags of akohol Composition of aolvelzt mixture uaed (by weight). 28 60 76 100 are the percentages of various combustible gases in a limit mixture which will just propagate flame and N I?/ iV// . . . the limiting percentages of the separate gases that can propagate flame then n n' n" - + + __ + . . . =1. N N' N" In table X are given the values for Le Chatelier's constant 1488 WEITE AND PRICE THE IGNITION O$ calculated from the results found for various ether-alcohol-air mixtures in 5 cm. glass tubes. TABLE X. Showing the Value obtained for Le Chatelier's Constant for Various Ether-Alcohol-Air Mixtures in 5 em. Glass Tubes at 20k20, Value of constant in mixture of following percentage composition by Maximum weight.percentage < \ variation ). Direction of 26 ether. 50 ether. 76 ether. from Limit. propagation. 75 alcohol. 50 alcohol. 26 alcohol. unity. Downwards 0.984 0.971 0.983 3 Horizontal ... 0.833 0.774 0-830 33 Upwards ... 0-646 0.589 0.663 41 Downwards 0.989 0.967 0.980 3 Horizontal ... 0.995 0.985 0.992 1 Upwards ... 1.004 0-993 1.00'7 1 The figures used for the upper limits in this table were all found at 60° those for the lower being determined as usual within 2* of 20°. It will be seen that in all cases for the lower limit and for downward propagation in the case of the upper limit Le Chatelier's rule holds for mixtures of ether and alcohol with air, and consequently the limits for a mixture containing any propor-tion of ether and alcohol can be calculated with an error of not more than 3 per cent.For horizontal and upward propagation, however the rule breaks down entirely. As would be expected, the greatest deviation from the rule always occurs for a 50 per cent. mixture. I n table XI are given the.results obtained when the same mix-tures were ignited in 5 cm. and 15 cm. iron tubes. The upper limits for ether-air mixtures are given in table XII. TABLE XI. Showing the Results obtained in Iron Tubes of 5 and 15 cm. Diameter for the Lower Limits for the Propagation of Flame in Certain Ether-Alcohol-Air Mixtures at 20 k 3". Percentage composi-tion of solvent mix-ture by weight. Ether. Alcohol. -100 0 100 0 75 25 50 50 25 75 0 100 Lower limits for propagation of flame.Diameter of f l I tube in cm. Upwards. Horizontal. Downwards. 5 2.24 2.29 2.34 15 1.73 1-80 1-93 15 2.24 2.30 2-46 15 2.81 2.89 3.02 15 3-48 3-53 3.69 16 4.16 4.23 4.37 ETHER-ALCOHOL-AIR AND ACETONE-AIR MIXTUBES ETC. 1489 TABLE XII. Showing the Upper Limits for Propagation obtained for Ether-Air Mixtures in 5 and 15 ern. Iron Tubes. Value of limit and direction of pro-pagation. A Diameter / I of tube. Upwards. Horizontal. Downwards. 5 15.45 7.96 6.70 15 23.30 22-30 6.50 A comparison of these results with those given in table IX shows that for all three directions of propagation the lower limit as found for 5 cm. tubes is greater when the tube is made of iron, as would be expected from the greater conductivity of this material.The reverse relation holds in the case of the upper limit obtained for ether-air for upward and horizontal propagation but in the case of the downward propagation the figure obtained for the iron is greater than that obtained for a glass tube of a similar size or even for the 15 cm. iron tube. As this result appeared peculiar, a fresh determination of this limit was made but the composition of the limiting mixture was found to be the same as that found in the first experiment. A comparison of the limits obtained in the 15 cm. iron tube with those previously determined (see tables IX XI and XII) is instructive. It will be seen that in every case the lower limit of a mixture is least in the case of the 15 cm.tube. The difference is very appreciable where a mixture contains a fair amount of ether but is not so great where alcohol is present in excess. The upper limit again is always found to be greatest in the case of the 15 cm. tube if we except the anomalous result. obtained for downward propagation determined in the 5 cm. iron tube. These results may be due to a decrease in the cooling effect of the walls or may possibly be due to turbulent motion in the gas caused by convection currents as found by Wheeler and Mason (T. 1917 111 1044) in the case of velocity of flame. An item in favour of the latter supposition is provided by the exceedingly high figure obtained for the upper limit for horizontal propagation in the 15 cm. tube. On the other hand the flame observed in the case of upward and horizontal propagation in the 15 cm.tube resembled very closely the cool flame of ether and the character-istic odour following such a flame was observed. The lower-limit results for iron tubes are shown graphically in Fig. 7. The figures in table XI11 show that Le Chatelier’s rule holds moderately well for the lower-limit results obtained in the 15 cm. iron tube 1490 WHITE AND PRXCE THE IGNITION OF TBLE XIII. Showing the Value Found for Le Chatelier’s Constant from the Figures obtained for the Lower Limit for Propagation wing Bther-Alcohot-Air Mixtures in the 15 cm. Zron Tube. Values of constant given by mixture of percentage composition by weight shown. Perc?ntage Direction of 26 ether. 60 ether. 76 ether.variation propagation 76 alcohol. 60 alcohol. 26 alcohol. from unity. / A I maxlmum DOWXIW~~~.H ............ 1.028 1.025 1.026 3 Horizontal ............... 1.028 1.036 1.022 4 UpwarilS ............... 1.039 1.038 1.031 4 FIG. 7. 60 25 0 0 26 60 75 100 Percentage of ether. Percedage of alcohd. Co?nposiCion of the ether-alcoho2 mixture wed (by weight). Acetone-Ether-Air Mixtures.-Owing t o the differences in resulte obtained for 6 cm. glass and 15 cm. iron tubes particularly for ether-air mixturea it .was decided t o determine the propaga ETHER-ALCOHOL-AIR AND ACETONE-AIR NIXTURES ETC. 1491 tion of flame limite of certain ether-acetone-air mixtures in glass and iron tubes as it was considered likely that results differing from those published by Wheeler and Whitaker (Zoc.cit.) for acetone-air mixtures would be obtained. The results of our experiments are shown in table XIV those of Wheeler and Whitaker being given in table XV. TABLE XIV. Showing the Propagation of Flame Limits obtained for Acetone-Ether-Air Mixtures using Various Iron and Glass Tubes a t 20 f 20. Percentage of solvent in limit mixture and position of r A \ mixture by Material Upwards. Horizontal. Downwards. diameter of Upper Lower Upper Lower Upper Lower Ether. Acetone. tube. limit. limit. limit. limit. limit. limit. 0 100 Iron 6 cm. - 3.80 - 3.90 - 4.00 Percentage com- direction of propagation. weight. and - -0 100 Iron 15 cm. 12-40 2.88 12.40 2-89 10.90 3-11 0 100 Glass 5 cm. 12.20 2.89 9.15 3.04 8.35 3-16 25 75 $ 9 11.20 - 8-56 - 7.75 -50 50 ?9 11.70 2.34 8.25 2.39 7.25 2.49 75 25 Y Y 13.20 - 8.15 - 6.65 -100 0 Y Y 15-75 1.93 8.00 2.05 6.15 2.15 TABLE XV.Showing the Propagation of Flame Limits as determined by Wheeler and Whitaker for Acetone-Air Mixtures in Glass Tubes of Various Diameters. Percentage of acetone in limit mixture and direction of propagation. Upwards. Horizontal. Downwards. - - - / \ Diameter of Upper Lower Upper Lower U&pp Lo27 tube in cm. limt. limit. limit. limit. 2-6 7.6 2-30 6.7 2.40 6.5 2-76 6.0 9.5 2.20 9.3 2-25 8-3 2.40 10.0 9.7 2-15 9.5 2.20 8.5 2-35 It will be seen that our results differ considerably from those In most cases they are considerably higher. previously published 1492 WHITE AND PRICE THE IGNITION OF The explanation is almost certainly to be found in the fact that in our method the solvent was weighed directly into the tube, whilst in the other the acetone present in any mixture was estimated by analysis.That our method is the more accurate as well as the easier is obvious but such a large discrepancy can only be explained by some abnormality in the behaviour of acetone during storage if the methods of analysis employed were not faulty. That such abnormal behaviour does take place when acetone vapour is stored over mercury is rendered extremely probable when the facts underlying the molecular association of acetone suggested in the paper by Wheeler and Whitaker are considered. I n this connexion too a quotation from the above paper might prove instructive. When discussing the analysis of the mixtures used the following statement is made “ A supply of air was saturated with acetone vapour at 15O and 760 mm.when it con-tained 13.5 per cent. (by volume) of acetone and used as a stock mixture from which the experimental mixtures could be prepared by the addition of air. . . . Consistent results were obtained by either of the absorption methods of analysis. Thus a mixture, known to c o n t a i n about 6.5 per cent. of acetone gave on analysis: “(1) Sodium hydrogen sulphite method . . . 6-50? 6.45 6.54, 6-55. “(2) Distilled water . . . 6.40 6.56 6.54 6-50 6.51 6.51. ‘‘ Absorption by distilled water seemed therefore to afford a sufficiently accurate method of analysis.” The mixture “known t o contain about 6.5 per cent. of acetone ” was apparently made up by diluting the requisite quantity of the stock solution with a calculated volume of air.The accuracy of the method of analysis appears to be assumed from the fact that results approximating closely t o the calculated figure were obtained. An examination of published figures for the vapour pressure of acetone a t 15O however shows that a stock solution of air saturated with acetone vapour a t this temperature a t 760 mm. pressure would contain more than 19 per centn. of acetone (by volume) the vapour pressure of acetone a t 15O being given by Sameshima ( J . Amer. Cheem. Soc. 1918 40 1482) as 147 mm. Extrapolation from Regnault’s results gives a very similar figure-about 150 mm. Raising Wheeler and Whitaker’s results in the ratio of 19:13*5 would often bring them much nearer t o those obtained by us.The results obtained for lower limits in the 15 cm. iron tube are strikingly similar to those obtained in a 5 cm. glass tube (see table XIV) and for upward propagation the upper limits are also very near. There are however very marked differences in the other figures given for the upper limits. It is worthy of note tha ETHER-ALCOHOL-AIR AND ACETONE-AIR MIXTURES ETC. 1493 the upper limits for upward and horizontal propagation are here identical in the case of the 15 cm. tube. The results obtained when using the 5 cm. iron tube show clearly what a large effect the conductivity of the material of the tube has in this case. The results obtained in 5 cm. glass tubes for the propagation of flame limits of ether-acetone-air mixtures are shown graphically in FIU.8. Percentage of ether. Percentage of acetone. Composition of solvent mixture wed (by weight). 0 25 60 75 100 Fig. 8. It will be noticed that the upper-limit curves present the peculiarities commented on in the case of ether-alcohol-air mixtures although to a less extent. The figures given in table XVI show that Le Chatelier’s rule holds moderately well for ether-acetone-air mixtures except in the case of the upper limit for upward propagation 1494 WHITE AND PRICE THE IaNITION OF TABLE XVI. Showing Values obtained for Le Chatelier's Coastant for the Limits for Propagation using Ether-Acetone-Air Mixtures 2% the 5 cm. Glass Tube at 2052O. Direction of Limit. propagation. Downwards Horizontal Upper Upwards Lower Downwards Lower Horizontal Lower Upwards Value of constant given by mix-ture of percentage composition by weight shown.25 ether. 50 ether. 75 ether. 75 acetone. 50 acetone. 25 acetone. 0.997 1.005 0.997 0.962 0.959 0,981 0.875 0.866 0.910 - 0.952 -- 0.953 - - 0.987 -T A \ Percentage [maximum variation from unity. 1 4 13 6 6 1 Znfluence of Temperature.-The influence of temperature on the limits of inflammability of gaseous mixtures has been studied by Bunk and Roszkosski ( J . Gasbeleucht. 1890 33 491 524 535, 553)) Taffanel (Compt. rend. 1913 157 595) Burrell and Robert-0011 (United States Bureau of Mines Technical Paper No. 121, 1916)) and Mason and Wheeler (T. 1918 '113 45). The experi-mental work of Bunk and Roszkosski appears to have been defective but the other workers found that the inferior limit of inflammability of methaneair mixtures was lowered by increasing the temperature of the gas-mixture before ignition Mason and Wheeler also showed that the upper limit of inflammability of methane-air mixtures became much greater under these conditions, so that increasing the original temperature of the gas widens the limits of inflammability of methane-air mixtures.This would be expected from the fact that the self-propagation of a flame t.hrough a combustible mixture is only possible when the heat due to the reaction between the combining gases is sufficient to make up for losses due t o radiation conduction and convection whether the heat lost is dissipated or utilised in raising adjacent layers of the gas t o the inflammation temperature.The heat; of reaction neces-sary and the heat dissipated must obviously be less in the case where the original temperature of the gaseous mixture is higher. A few figures are given below t o show t o what extent the limits of inflammability are affected by temperature ETHER-ALCOHOL-AIR AND ACETONE-AIR MIXTURES ETC. 1496 TABLE XVII, Showing some Results obtained b y Mason and Wlzeeler for the Downward Propagation of Flame in Mixtures of Methane and Air. Initial temperature. Lower limit. Upper limit. 20" 6.00 13.40 100 5.45 13.50 200 5-05 13.85 300 4.40 14-25 500 3.65 15-35 700 3.25 18.75 I n the circumstances it was decided that it would be unnecessary to do more than find the change in the upper limit of ether-air mixtures with temperature.These experimenb were carried out in the jacketed tube utilised for the upper limits of alcohol and ether-alcohol-air mixtures. The results obtained are shown in table XVIII. It will be seen that a decided rise in the upper limit for propagation takes place when the initial temperature of the mixture is raised through 40°. TABLE XVIII. Showing how the Upper Limit for the Propagation of Flame Varies with the Initial Temperature of the Ether-Air Nixture Used. Direction of propagation. Limit at 20'. Limit at 60'. Upwards ...... 15.75 17.05 Horizontal . . . 8.00 13.00 Downwards ... 6.15 7.48 Influence of Pressure.-The influence of pressure on the limits of inflammability of gases over any large range is by no means easy to predict although it is well known that the lower limit of in-flammability of many gas-air mixtures increases a t diminished pressures.Terres and Plentz ( J . Gasbeleucht. 1914 57 990, 1001 1016 l025) Burrell and Robertson (Zoc. cit.) and Mason and Wheeler (Zoc. cit.) have investigated the effect of pressure on the limits of inflammability of mixtures of methane with air. The general conclusions to be drawn from their work appear t o be that, below atmospheric pressures decreasing the pressure narrows the limits of inflammability but that above atmospheric preasure, increasing the pressure raises both tbe limits of inflammability. The work done by us was confined to pressures at or below atmo-spharic prwsure a i d the results are shown in table XIX and XX 1496 WHITE AND PRICE THE IGNITION OF TABLE XIX.Showing the irnnflztence of Pressure on the Limits for Horizontai Propagation of Flame in Et her-Air Mixtures. Pressure in m. 770 751 600 520 460 460 420 400 300 200 100 50 Percentage of ether in mixture. wk-At lower limit. At upper limit. 1.87 12.90 - 10.50 - 9.20 1-88 - 8.20 - 7.90 - 7.80 1.92 7.30 2.08 6.80 2.33 6.10 2.99 5-00 --TABLE XX. Showing the Influence of Pressure on the Limits for Downward Propagation of Flame in Ether-Air Mixtures. Pressure in mm. 600 600 400 300 200 100 50 Percentage of ether in mix-ture at lower limit. 6.20 6.20 6.20 6.20 5.90 5-50 No ignition.It will be noticed that the results obtained near atmospheric pressure in the case of these experiments differ appreciably from those found by the ordinary method as given in table IX. This is due to the fact that all the experiments given in tables XIX and XX were carried out with both cocks closed. The type of ignition obtained was also very different ; for instance the ignitions obtained when determining the upper limit for horizontal propaga-tion under 600 and 751 mm. pressure were characteristic slow cool flames which could not be seen except in a totally darkened room. It wits found that any attempt to determine the limits for pro-pagation of ether-air mixtures for pressures greater than atmo-spheric in the glass tubes available merely shattered the tube.The curves given by plotting pressure against the percentage of ether in the limiting mixture are given in Fig. 9. That for hori-zontal propagation is interesting. It was a t pressures above that a t the peculiar bend marked x that the cool flame became notice ETHER-ALCOHOL-AIR AND ACETONE-AIR MIXTURES ETC. 1497 able Below this pressure the flame was of a green colour and traversed the tube very rapidly in a manner similar to that noticed for the lower limit a t corresponding pressures. Influence of VeZelocity .-Under present-day conditions of solvent-recovery a good portion of the solvent-air mixture is often in rapid motion. It therefore became a question of determining so far as was possible in the laboratory the effect of velocity on the limits of propagation of solvent-air mixtures.Owing to the fact that no FIU. 9. Parcentage of etheriin mixture. real approximation t o manufacturing conditions could be attained, the work done was confined to ether-air mixtures. It has long been known that a moving mixture of combustible gas and air too weak to propagate flame can carry a cap of flame t o a great distance from an igniting source. Wheeler has also proved that a mixture below the lower limit of propagation when in a quiescent state can often inflame when agitated. The speed of propagation of flame is also notably dependent on the degree of mechanical agitation of a mixture and various experiments on the effect of agitation on gas-mixtures and on the rate of development of pressure when gas-mixtures are ignited are given by Clerk an 1498 WHITE AND PRICE THE IGNITION OF Hopkinson ( R e p .Brit. Assoc. 1912 ZOO) Clerk (“ Ganet Lecture,” Junior Institution of Engineers 1913) and Wheeler (this vol., p. 81). So far as we know however no figures have as yet been given for the effect of the velocity of a gas-mixture on the limits of its propagation of flame. The methods used for moving the mixture and determining its velocity were those described under sub-ignitisn-temperature. Great care had to be taken in determining the upper limit for pro-pagation against the gas-current as the flame passed with great velocity down the 10 mm. bent glass tube joining the limit tube t o the reservoir. This gave very little time for preventing the flame from getting into the reservoir.The results obtained are given in sable XXI. TABLE XXI. Showing the Effect of the Linear Velocity of an Ether-Air Mixture on its Downward Propagation of Flame Limits as determined in a 5 cm. Glass Tu6.e at 20*2O. Flame moving in the direction of gas current. , Percentage of ether in mixture at 0 2-13 6-15 1 1.97 6-40 3.5 1.95 6.50 9 1.95 6-66 Percentage of ether in mixture at Velocity in cm. per sec. lower limit. upper limit. Flame moving against the gas current. 0 2.13 6.15 1 - 6.25 3.5 - 6.25 9 - 6.25 It waa found to be impossible to find the lower limit of propaga-tion when the flame was moving against the stream as it merely became a question of the velocity of the stream as compared with that of the flame in the mixture used.It will be seen from the upper portion of the table that the velocity of the gas-current affects very appreciably the percentage mixture which will pro-pagate flame. That a portion of this change is due to turbulence caused in the gas however appears to be very likely for in the cases tried when the flame moves against the gas-current with any real velocity the upper limit is always the same. The figures given for zero velocity in table XXI are those found with one cock closed in the ordinary manner. As conditions are slightly different when both cocks are open it was decided t o try BU& ara experiment. The upper limit found under these condition ETHER-ALCOHOL-AIR AND ACETONE-AIR MIXTURES ETC. 1499 was 6.30 per cent. but the convection due to the flame probably caused this figure to be a little high on account of the air drawn into the tube during the passage of the flame.It almost appears as if when a flame is travelling against a gas-stream the turbulence due t o the velocity of the stream practically balances the effect of that velocity in hindering the propagation of flame. In&uence of the Presence of Glyceryl Trinitrate Diethyl Peroxide and Ethyl Hydrogen Peroxide. In the experiments to determine the effect of glyceryl trinitrate on the limits of propagation for ether-air mixtures the mixture used was charged with glyceryl trinitrate as described under the experiments on the determination of sub-ignition-temperatures. The results obtained when using ether-air mixtures containing diethyl peroxide and ethyl hydrogen peroxide are given in table XXII.TABLE XXII. Showing the Effects on the Limits of Propagation of Flame of adding Amounts of GEyceryl Trinitrate Diethyl Peroxide and Ethyl Hydrogen Peroxide to Certain Ether-Air Mixtures. Direction of propagation. Upwards . . . . . . Downwards ... Upwards . . . . . . Downwards ... 9 , 9 9 1 9 Limit. Lower Mixture used (with air). Ether. Up& Lower Ether saturated with glyceryl Upper Y f Lower Diethyl peroxide. 9 ) Y f trinitrate a t 20". , Ether Containing 25 per cent. of diethyl peroxide by weight. Ether containing 10 per cent. of 80 per cent. ethyl hydrogen peroxide by weight. Upper 9 Lower Upper $ 9 Percentage of ether in limit mixture .1.93 2-15 6.15 1.95 6.15 2-34 2-18 10.1 2-17 6.5 The peroxides of ether are calculated as ether in making up the percentage volume occupied by the solvent in the limiting mixture. It will be seen that glyceryl trinitrate appears to have no effect on either limit and that the peroxides have little effect on the lower limit but that they raise the upper limit very appreciably. The glyceryl trinitrate did not appear to affect the flames given, but the flames when peroxides were present were invariably fiercer than when ether alone was used except perhaps a t the extreme limit. VOL. cxv. 3 1500 WHITE PRICE THE IGNITION OF SECTION 111. Investigatioit of Various Means of Ignitio?E. A series of experiments was carried out in which sparks obtained by various means were used for attempting to ignite ether-alcohol-air mixtures.It was found that steel to steel emery to steel and pyrites to steel sparks appeared to be unable to cause the inflamma-tion of any of the many mixtures tested. Ferro-cerium to steel sparks however ignited most mixtures very readily. The igniting powers of a small gas flame and a moderately powerful electric epark appeared t o be of the same order and both almost invariably gave rise to ordinary combustion the limits of propagation of flame being naturally identical in the two cases. I n the case of ether-alcohol-air mixtures quick heating of a mixture up to but not far above its sub-ignition-temperature seemed t o give rise to a cool flame which had limits for propagation varying from those of ordinary inflammation.The Cool Flame.-The difference in propagation limits for the two methods of combustion was particularly noticeable in the case of concentrated ether-air mixtures as the flame travelled easily through a 20 per cent. mixture in a horizontal tube 4 cm. in diameter although the upper limit for the propagation of ordinary combustion in a 5 cm. tube would be 8 per cent. No deter-minations of the limits for propagation of a cool flame were made, but experiments carried out for other purposes indicate that it is unlikely that such a flame could propagate downward through a mixture containing much more than 6 per cent. of ether. This flame was occasionally observed when electrical ignition was utilised, more particularly with high concentrations of ether or low pressures.It appeared as stated by Perkin to require very little oxygen and the products of combustion were characteristic. It was found that the addition of less than 1 per cent. of oxygen to a mixture of 9 per cent. of ether in nitrogen was sufficient to give luminous combination below 220O. The increase of temperature caused by this flame in a mixture containing less than 3.5 per cent. of ether and heated to its sub-ignition-temperature wids insufficient to be indicated by the fine thermo-couple registering the tempera-!are of the gas in the ignition tube. The increase of pressure caused by it was also very small. This was measured roughly by ib effect on a column of mercury so arranged that after ignition t&e mercury in both limbs of a U-tube would be level.The mean of three experiments with a 3.9 per cent. mixture gave a momentar ETHER-ALCOHOL-AIR AND ACETONE-AIR MlXTURES ETC. 150 1 increase of pressure equal to 3 or 4 cm. Mixtures containing more ether gave far greater pressures. It was found that a 0.3 cm. mesh iron gauze or a 0.2 cm. mesh brass gauze prevented the passage of a cool flame down a glass tube 7 cm. in diameter. Discussion of R esul t s . The sub-ignition-temperature figures given above agree fairly well with ignition-temperatures previously published for alcohol and acetone. The sub-ignition-temperature given for ether-air mix-tures however whilst agreeing almost exactly with the ignition-temperature given by Alilaire differs notably from the other figures available.The difference is probably to be accounted for by the fact that in the methods employed to obtain these no account was taken of the cool flame of ether-the cool flame of alcohol can only be obtained near the temperature a t which an explosion or ordinary combustion occurs directly. This may be justifiable and necessary in the determination of ignition-temperatures but involves the neglect of a phenomenon which can and very often does give rise t o ordinary combustion under suitable conditions. Moreover these conditions are precisely those liable to obtain during solvent-recovery on the manufacturing scale namely the presence of a large volume of the solvent-air mixture and some degree of confinement. That ordinary combustion of a dangerous nature could be caused by heating an ether-air mixture in glass tubes to 1 8 7 O was proved again and again when using 4 and 5.5 cm.tubes particularly if the percentage of solvent lay between 5 and 9. That this result was a genuine one was proved by stopping a current. of ether in air flowing along a glass tube kept a t 1 8 7 O . Ignition of the ether occurred in every case. The method employed to determine sub-ignition-temperatures thus appears to be a practical one and also has the advantage of being easily adaptable to determine the effect of substituting for glass any material that might be used in manufacture. The ignition observed by Alilaire must indubitably have com-menced as a cool flame. 'The temperatures necessary to obtain such a flame in mixtures containing fair quantities of ether as shown by us are fairly near the temperatures attainable in a steam-heated building particularly when it is considered that the presence of metals lowers the sub-ignition-temperature appreciably.On the other hand the results previously obtained for the ignition-3 K 2502 WHITE AND PRICE THE IGNITION OF temperature of ether-air mixtures are far above those one can conceive of being attained in such a building except in the most extraordinary circumstances . The results show that a quiescent gas appears to be more easily ignited than one in motion but the experimental work covers only a very small range of velocities and in any case the propagation of flame is more easily and quickly carried out by gas in motion.The presence of glyceryl trinitrate in a gas-mixture as was anticipated from its amount does not seem to affect the tempera-ture of sub-ignition or the limits of propagation of flame. The peroxides sometimes present in ether in very small amounts can, however affect both its ignition and its propagating qualities if present in sufficient quantity. Our work seems to indicate that their influence in causing primary ignition could only be inappreci-able although it is quite conceivable that they could well affect the change from cool to ordinary flame. Reduction of pressure appeared to cause a lowering of the sub-ignition-temperature of the mixtures examined but the effect of pressure alone within the range of variation of atmospheric pressure can scarcely have a practical influence on the ignition of the solvent-air mixtures as for ether-air for example a reduction of pressure of 10 cm.near atmospheric pressure caused a variation in the sub-ignition-temperature of less than 2 O . The only phenomenon that could be expected to reduce the sub-ignition-temperature of ether-alcohol-air and ether-air mixtures below the danger limit is thus that described as shock ignition. With a difference of pressure of less than half an atmosphere it was possible by this method to ignite a gas-mixture a t least 170° below its sub-ignition-temperature so that it is quite conceivable that the development of sudden differences of pressure on the manu-facturing scale might easily be the determining factor in bringing about ignition of the solvent-laden air.Exactly how this is to be brought about can only be conjectured as our inside knowledge of gas-ignition particularly as regards this fresh phenomenon is very limited. The present work has shown how many accidents could happen but much remains to be done before any sound explanation can be given of such a conflagration as was described in The Times of March 28th 1919 when a bottle of ether exploded in a military hospital a t Southage. According to the same report, explosions of bottles of ether are of somewhat frequent occurrence. The results given in the older work for the limits for propaga-tion of flame in ether-air mixtures are 1.8 per cent. and about 9 per cent. the latter being apparently far .out whilst those for alcohol-air mixtures agree fairly well with our results particularl ETHER-ALCOHOLAIR AND ACETONE-AIR MIXTURES ETC.1503 as regards the lower limit. The change from 5 cm. glass to 15 cm. iron tube affects the results for the upper limit for horizontal and upward propagation in ether-air mixtures very materially the limits becoming well over 20 per cent. in each case instead of 8 per cent. for horizontal propagation and 16 per cent. for upward propagation. The extreme limits determined for ether-air mix-tures are thus 1.73 and 23.30 per cent. The upper limit for pro-pagation in alcohol-air mixtures in 5 cm. glass tubes was found to be 18.95 at 60°. As this figure was well above the highest con-centration of alcohol vapour obtainable during normal recovery, there was no point in repeating this in the 15 cm.iron tube. 'The lower limit of propagation for alcohol-air mixtures was only very slightly altered in the large iron tube falling from 4.24 per cent. in the 5 cm. glass tube to 4.16 per cent. The results for the propagation of flame in ether-alcohol-air mixtures obtained during this investigation are distinctly interest-ing. It is found that Le Chatelier's rule holds for all directions of propagation for the lower limit and for the upper limit for down-ward propagation. The rule does not hold for the other two direc-tions of propagation for the upper limit the discrepancies being very considerable in the case of upward propagation. Wheeler's work on acetone-air mixtures has already been discussed and it may suffice here to state that the limib given by him are 2-15 and 9.7 per cent.our results being 2.88 and 12-40 per cent. The lower limit of 5 per cent. given by Brunswig is obviously wrong but the upper-limit figure of 12 per cent. is very near that found by us. It will be seen that the effect of temperature and pressure on the limits for the propagation of flame in ether-air mixtures is quite material. The influence of the velocity of the gas-current was not examined throughout a sufficient range t o enable sound conclusions to be drawn as to its effect under manufacturing con-ditions but i t is fairly clear that a margin must be allowed for this factor when dealing with the limit results obtained. The presence of 1 per cent. of the peroxides of ether in ether-air mix-tures appears to have no appreciable effect on the lower limit for the propagation of flame and it is by no means likely that there would be sufficient peroxide present under practical conditions to affect the upper limit materially.Summary. The soap-bubble method described by McDavid (Zoc. cit.) gave for the ignition-temperature of ether-air mixtures results varying from 859" t o 1068O. The method seemed to be untrustworthy. 3 K* 1504 THE IGNITION OF ETHER-ALCOHOLAIR ETC. The other method used seemed to give the minimum temperature a t which the reaction in a combustible gas-mixture became self-supporting-called the sub-ignition-temperature. This temperature, which appears t o be the one required from a safety point of view, was 187O for ether-air mixtures in glass and varied from 187O t o about 500° for the different ether-alcohol-air mixtures used.It was about 500° for acetone-air mixtures. The presence of appreci-able quantities of metal in the vessels used lowered the sub-ignition-temperature. Decrease of pressure appeared to reduce the sub-ignition-temperature of a mixture but the presence of small quantities of glyceryl trinitrate or of diethyl peroxide had little effect on the sub-ignition-temperature of ether-air mixtures. The sub-ignition-temperature of such mixtures was lowered by the presence of ethyl hydrogen peroxide. The effect of slight velocities seemed to be to raise the sub-ignition-temperature of ether-air mixtures. When an exhausted vessel is quickly put into communication with a reservoir containing ether-air or carbon disulphide-air mixtures under specified conditions the gas can be ignited a t the ordinary temperature. This phenomenon has been termed shock ignition. The limits for the propagation of flame in mixtures of ether-alcohol-air and ether-acetone-air have been determined in 2.5 and 5 cm. tubes of glass and in 5 and 15 cm. tubes of iron. The extreme limits found were 1.73 and 23-30 per cent. for ether-air mixtures 4.16 and 18.95 per cent for alcohol-air mixtures and 2.88 and 12-40 per cent. for acetone-air mixtures. The upper limit for propagation in alcohol-air was determined a t 60°. Figures obtained with the 15 cm. iron tube often differed appreci-ably from those obtained with 5 cm. glass tubes. Le Chatelier’s rule was found to hold fairly well for ether-alcohol-air mixtures except for horizonkal and upward propaga-tion in the case of the upper limit. The only considerable devia-tion from the rule in the case of ether-acetone-air mixtures was observed for upward propagation and the upper limit. Increase of temperature was found to raise the upper limit for propagation in ether-air notably and reduction of pressure was found to narrow the limits. Increase in the velocity of the gas-mixture widened the limits materially. The presence of the per-oxides of ether scarcely affected the lower limit of propagation in ether-air but any considerable quantity raised the upper limit of such a mixture. It was found impossible t o ignite ether-alcohol-air mixtures by means of steel to steel emery to steel or pyrites to steel sparks bu THE CONDUCTIVITIES OF IODOANIL1"ESULPHONIC ACIDS. 1505 inflammation was readily obtained when using ferro-cerium t o steel sparks. Many of the properties of ether-air mixtures appear to be explained by the formation of a cool flame. Further work is contemplated on the phenomenon referred to as shock ignition. We desire to express our thanks to Messrs. Nobel's Explosives Co. Ltd. for whom the work was carried out and particularly to Mr. W. Rintoul Manager of the Research Section for kind per-mission to publish our results. We also wish to thank Mr. A. W. Sanderson for assistance in carrying out some of the experimental work. THE RESEARCH LABORATORIES, ARDEER FA~ORY STEVENSTON. [Received September 22nd 191 9.