FIFTY YEARS OF EXPERIMENTAL RESEARCH ON THE EFFECT OF WATER VAPOR ON THE COMBUSTION OF CARBON MONOXIDE*

V. A. Bone

The placement in the pages of Uspekhi of an article devoted—at first glance—to a special question calls for explanation. The study of problems of combustion, in particular of the mechanism of gas explosion, is becoming a most urgent research task. Every step forward in our knowledge of the mechanism of combustion may find its equivalent in many millions of savings for the socialist economy or in an increase in the defensive capacity of our country.

Despite its already substantial history, this area of physical chemistry is still far from a period of “maturity”; it is still very poor in general guiding ideas. A decisive turning point here should be expected from the introduction of the methods and ideas of chemical physics, and from the assimilation by its means of the accumulated empirical material. With regard to the experimental study of gas explosions, we are most of all indebted to the brilliant works of Dixon and his school. The lecture by Prof. Bone—a pupil of Dixon and a leading investigator of combustion processes—is devoted to a review of these works.

INTRODUCTION

When the Council of the Society invited me to deliver this lecture in place of my esteemed teacher Prof. H. B. Dixon, whose sudden death on September 18, 1931, took from us a brilliant pioneer in the field of the study of combustion, I decided to devote my lecture to a review of the experimental work of the last 50 years—work proceeding from his epoch-making discovery of the effect of moisture on the combustion of carbon monoxide.

* A lecture delivered at the Chemical Society in London on December 11, 1930, and published in Journ. Chem. Soc. London (1931), 338–361. Translated by A. Sokolik.

For, as he often said, “then they let out a hare which, despite the united pursuit of the dogs, has still not been caught.”

In my exposition I shall deal only with the principal experimental discoveries, insofar as they are connected with the various theories proposed for their explanation, since this is required for historical completeness. In the present case I do not undertake to defend any particular point of view. I shall try to describe certain facts quite objectively and to present the experiments that illustrate them, leaving my listeners to draw conclusions from them.

My review may be divided into two parts: the first, covering the period from 1877 to 1900, when only purely chemical theories existed, and the second, devoted to the physico-chemical approach to the study of the process of combustion, characteristic of recent work.

PART I

PURELY CHEMICAL THEORIES

Dixon’s Discovery (1877–1880)

In 1873 Dixon, who had received a classical education at Westminster, came to Oxford, where, under the influence of the late Harcourt, his scientific activity began. Three years later, again at Harcourt’s suggestion, he undertook to verify Bunsen’s experiment (1853) on explosions of detonating gas with an ever-increasing content of carbon monoxide. The results of these experiments, as was then believed, contradicted the law of acting masses, proclaimed by Berthollet in 1805, and led to the erroneous conclusion that a continuous change in the composition of a gaseous mixture causes a discontinuous (“per saltum”) change in the rate of the reaction passing into explosion.*

In 1877, while checking Bunsen’s work, Dixon accidentally passed a series of sparks through a mixture \(2CO + O_2\), enclosed at 300 mm in an eudiometer over mercury, the gas being—

* Annalen, 85, 137 (1853), Bunsen, Gasometrische Methoden.

came into contact with caustic potassium for some time. To his surprise, the mixture did not explode even when the experiment was repeated.

Supposing that something was wrong here with the carbon monoxide itself, obtained from magnesium formate, he repeated the experiments with carbon monoxide prepared in different ways and obtained the same result. He then suggested that the only new (unusual) condition of the given experiment was the dryness of the gas mixture.

And indeed, after adding a little water vapor to it, he obtained an explosion at the very first spark.

In further investigating this phenomenon he carried out several experiments, drying the gas over phosphorus anhydride. A mixture of \(2\mathrm{CO} + \mathrm{O}_2\) dried in this way for several days did not explode from a series of successive sparks of a Ruhmkorff coil even at atmospheric pressure. But the addition of negligible traces of water vapor was sufficient for the mixture to become explosive. This important discovery was published by Dixon in a report to the British Association in 1880,* and he demonstrated three experiments, two of which will be reproduced here.

Fig. 1.

In the first experiment, sparks passed between two platinum wires across a gap of \(1\) mm in a mixture of \(2\mathrm{CO} + \mathrm{O}_2\), dried over phosphorus anhydride for 8 days and enclosed in a glass tube (Fig. 1a), and no explosion occurred. In the second experiment, the spark did not cause an explosion in a mixture dried in the same way and enclosed at atmospheric pressure in a vertical glass tube (Fig. 1b), sealed at the bottom, and at the top se—

* B. A. Reports, 503, 1880.

closed stopcock. But as soon as a drop of water is admitted through the stopcock, after a few minutes the spark produces an explosion. In the third experiment, in gas dried in the same way and enclosed at atmospheric pressure in a sealed vertical glass tube (Fig. 1c), to the wall of which a piece of caustic potash is fused, a spark likewise does not produce an explosion.

But it is enough merely to heat the caustic potash from the outside (which thereby gives off an insignificant amount of moisture) for the mixture to explode instantaneously when the spark is passed.

Further Experiments of Dixon (1880–1893)*

Dixon first of all attempted to establish whether water vapor acts purely chemically or only as a “third body.”

For this purpose he passed a spark through a series of mixtures \(2\mathrm{CO} + \mathrm{O}_2\), dried over phosphorus anhydride, to which about 1% of various, likewise dried, gases was admixed. In doing so he found that, whereas all hydrogen-containing substances render the gas explosive, other impurities have no effect:

while HCl has an effect,	\(\mathrm{CCl}_4\) has no effect
\(\mathrm{CH}_4\)	\(\mathrm{C}_2\mathrm{N}_2\)
\(\mathrm{NH}_3\)	\(\mathrm{NO}\)
\(\mathrm{H}_2\mathrm{O}\)	\(\mathrm{SO}_2\)
\(\mathrm{H}_2\mathrm{S}\)	\(\mathrm{CS}_2\)
\(\mathrm{C}_2\mathrm{H}_4\)

From this he concluded that the explosibility of a mixture of carbon monoxide with oxygen is due to the presence of a hydrogen-containing gas or vapor capable of yielding (upon combustion) water vapor. Its action is reduced to the role of a “carrier” of oxygen to carbon monoxide,** so that:

\[ \text{1. } \mathrm{CO} + \mathrm{OH}_2 = \mathrm{CO}_2 + \mathrm{H}_2 \qquad \text{2. } 2\mathrm{H}_2 + \mathrm{O}_2 = 2\mathrm{H}_2\mathrm{O}. \]

In another series of experiments, carried out during 1880–1884, he compared the mean rates of propagation of flame between two points separated from one another

* Phil. Trans., 175, 617 (1884); 184, 97 (1893).
** B. A. Reports, loc. cit.

over 1 m, in a brass tube with an internal diameter of 13 mm at atmospheric pressure for mixtures of varying degrees of humidity—from “thoroughly dried” to “saturated with water vapor at 35°” (in the latter case the flame speed was measured at this same temperature). The experiments showed that the average flame speed increases sharply with increasing moisture content:

\(2\mathrm{CO}+\mathrm{O}_2\)	Humidity, in %	Average flame speed in m/sec
Dried by passage over fresh \(\mathrm{P}_2\mathrm{O}_5\)	—	44
Dried by passage over \(\mathrm{P}_2\mathrm{O}_5\) that had been in use	—	69
Dried by passage over conc. \(\mathrm{H}_2\mathrm{SO}_4\)	—	103
Saturated with water vapor at \(6^\circ\mathrm{C}\)	0.9	125
Saturated with water vapor at \(8^\circ\)	1.06	155
Saturated with water vapor at \(12^\circ\)	1.34	200
Saturated with water vapor at \(35^\circ\)	5.5	226

Subsequently, in his Bakerian lecture to the Royal Society in 1893, devoted to the “rate of propagation of gaseous explosions,” he published the following detonation velocities for mixtures of \(2\mathrm{CO}+\mathrm{O}_2\) of varying degrees of humidity:

\(2\mathrm{CO}+\mathrm{O}_2\)	Humidity, in %	Detonation velocity in m/sec
Thoroughly dried with conc. \(\mathrm{H}_2\mathrm{SO}_4+\mathrm{P}_2\mathrm{O}_5\)	—	1264
Dried only with conc. \(\mathrm{H}_2\mathrm{SO}_4\)	—	1305
Saturated with water vapor at \(10^\circ\mathrm{C}\)	1.2	1676
Saturated with water vapor at \(20^\circ\)	2.3	1703
Saturated with water vapor at \(28^\circ\)	3.7	1713
Saturated with water vapor at \(35^\circ\)	5.6	1738
Saturated with water vapor at \(45^\circ\)	9.5	1693
Saturated with water vapor at \(65^\circ\)	24.9	1526
Saturated with water vapor at \(75^\circ\)	38.4	1266

As these experiments showed (and they were confirmed by other investigators as well), the flame speed in the mixture \(2\mathrm{CO}+\mathrm{O}_2\), both before and after detonation, increases with increasing degree of humidity only up to a 5.6% concentration of water vapor. With a further increase in moisture content, water acts only as a diluent. In discussing

of any theory devoted to the role of water vapor, one should remember this important fact.

The lowering of the flame velocity when moisture is removed from the gas mixture can be demonstrated by the following experiment. When a “moist” mixture of 20% CO + 80% air is ignited in a long horizontal tube, the flame propagates very rapidly. But the velocity at once decreases sharply if the gas mixture has first been passed through 80% sulfuric acid, and it is not ignited at all by the flame of a candle after more intensive drying with 98% sulfuric acid.

Discovery of H. B. Baker (1883–1902) *

With the exception of the experiments with “detonation velocities,” all the experiments described above were carried out by Dixon during the Oxford period of his activity (1877–1886), when (in 1883–1884) Baker worked together with him. And, for completeness of this part of the review, mention should be made of Baker’s discoveries, especially those relating to the oxidation of carbon. Working first in Dixon’s Oxford laboratory, and then independently in London, he found that carefully purified charcoal, heated red-hot in excellently dried oxygen, burns very slowly, without a visible flame, giving chiefly carbon monoxide; the amount of carbon dioxide formed (always very small) is inversely proportional to the degree of dryness of the system.

If, for example, after a week’s drying over phosphoric anhydride, the combustion products after passing oxygen over strongly heated charcoal contain: CO₂—5.0; CO—40.0 and O₂—55%, then after two weeks’ drying they contain: CO₂ = 2.0–1.7; CO—39.5–27.8 and O₂—58.5–70.5%.

Moreover, in 1884 Baker showed that in a glass tube filled under atmospheric pressure with—

* I. 47, 349, 1885; 81, 400, 1902; Phil. Trans., 178, 571, 1888. Journ. Chem. Soc. London.

dried over phosphorus anhydride with oxygen, thoroughly purified and dry sulfur and phosphorus can be subjected to repeated ignition without any traces of their combustion. Traces of moisture, however, immediately cause vigorous ignition. This gave chemists grounds to believe that moisture in general promotes combustion. Finally, in 1902, Baker crowned his remarkable discoveries by showing that carefully purified and perfectly dry detonating gas at a temperature corresponding to red heat does not react at all, whereas undried gas explodes. The same result was readily obtained by Andrews and myself at a temperature of 525° with the difference that, under our conditions, the moist mixture reacted without explosion. Baker also observed that more than 20 “extremely weak sparks” can be passed through a mixture dried over phosphorus anhydride without producing an explosion.

On the method of drying. Needless to say, these experiments aroused universal interest, and many attempts were made to repeat them; but some of these yielded negative results only because certain essential precautions were not observed, the importance of which is difficult to overestimate. A description of them would require much space. And since the most important of them may be found in the paper communicated by Baker and myself to the Society,* it will suffice here to say that, in the main, they amount to the following: 1) exceptional cleanliness, especially of the inner surface of the vessels, 2) exceptional purity of all gases and substances used, 3) the absolute absence in the apparatus of organic grease for stopcocks and rubber connections, 4) the use, as the final drying agent, of redistilled phosphorus anhydride purified and entirely free of lower oxides—a requirement first substantiated by Stenhouse in 1890–1893—and 5) the use, for the manufacture of explosive and reaction vessels, of special glass with minimal adsorption capacity with respect to moisture.

* Journ. Chem. Soc. London, 1601, 1604, 1929.

Personally, in experiments where it was necessary to obtain the most intensive drying, I refrain from using calcium chloride and sulfuric acid for preliminary drying, replacing them with solid caustic potash (but not “purified in alcohol”) and with a coil immersed in liquid air. For the final drying, however, I take redistilled and purified phosphoric anhydride, which must be snow-white and friable.

The drying capacity of phosphoric anhydride. As regards the drying capacity of redistilled and purified phosphoric anhydride used in these experiments, as early as 1887 Morley carried out careful measurements over several months of the quantity of moisture in air passed at a rate of 3 liters/hour: 1) over “moist calcium chloride” and 2) through a glass tube 8 cm long and with an internal diameter of 2 cm, filled with pure redistilled phosphoric anhydride.

He found that in the latter case, in 40,000 liters of air there remained 1 mg of moisture, or approximately \(1/32\,000\,000\) part by volume. Repeating this experiment in 1904, he came to the conclusion that (as he had supposed earlier) the residual “elasticity of water vapor” is most probably the vapor pressure of phosphoric anhydride and that, consequently, drying a gas with such phosphoric anhydride, even with such a short contact, may be considered absolute.*

It is clear from this that prolonging the drying process where it is required to attain “intensive dryness” is not due to any difficulty in drying the gas itself; rather, it is necessary in order to extract the last traces of moisture stubbornly retained by the walls of the vessel.

To be certain of this, it is necessary during the drying process periodically to heat the walls to \(150\text{–}200^\circ\), while at the same time cooling that part of the vessel where the phosphoric anhydride is located.

* Amer. Journ. of Science, 34, 199, 1887; Journ. Amer. Chem. Soc. 26, 1171, 1904.

If all these measures are taken, then for drying to the greatest possible extent it is sufficient that the gas remain in a sealed vessel of hardened glass, of 100 cm³ capacity, over pure redistilled phosphoric anhydride for 250 days.

Traube, Mendeleev, and Armstrong (1885–1892)

One of the first investigators drawn to this field by Dixon’s discovery was Traube. In 1885 he proposed a popular lecture experiment demonstrating the extinction of a flame in a mixture of carbon monoxide and air dried over sulfuric acid at atmospheric pressure. It should be noted that, when the air is replaced by similarly dried oxygen, the flame also does not go out.

Traube also showed that if a flame of hydrogen or carbon monoxide is directed onto the surface of cold water or ice, hydrogen peroxide is formed.

By directing a carbon monoxide flame with a blowpipe onto a piece of ice and collecting the water thereby formed, we easily detect in it the presence of hydrogen peroxide, using titanosulfuric acid as the reagent (obviously the basic salt \( \mathrm{TiOSO_4} \)) and starch with a solution of potassium iodide.

Traube*, however, went further and, attempting to cast doubt on the possibility of the reaction postulated by Dixon,

\[ \mathrm{CO} + \mathrm{OH_2} = \mathrm{CO_2} + \mathrm{H_2} \]

at high temperature, concluded that the participation of water vapor in the combustion of carbon monoxide proceeds according to the following reaction cycle, in which hydrogen peroxide plays an essential role:

\[ \begin{aligned} 1)\quad & \mathrm{CO} + \mathrm{O:O} + \mathrm{OH_2} = \mathrm{CO_2} + \mathrm{H_2O_2},\\ 2)\quad & \mathrm{H_2O_2} + \mathrm{CO} = \mathrm{H_2O} + \mathrm{CO_2}. \end{aligned} \]

In 1886 Dixon subjected this theory to severe criticism, considering Traube’s view erroneous that the reduction of water vapor in a carbon monoxide flame was impossible—

* Ber., 18, 1890 (1885).

of the kind. However, in 1891 Mendeleev, although rejecting Traube’s first equation, nevertheless supports his scheme, which assigns hydrogen peroxide the intermediary role in the reaction that Dixon assigns to water vapor. Mendeleev on the basis that “interaction in equal volumes precedes all others,” considers the following cycle of bimolecular reactions to be the most probable one for the combustion of carbon monoxide (Principles of Chemistry*, 1891):

\[ \begin{gathered} 1)\ \mathrm{CO}+\mathrm{OH}_2=\mathrm{CO}_2+\mathrm{H}_2,\quad 2)\ \mathrm{H}_2+\mathrm{O}_2=\mathrm{H}_2\mathrm{O}_2,\\ 3)\ \mathrm{H}_2\mathrm{O}_2+\mathrm{CO}=\mathrm{CO}_2+\mathrm{H}_2\mathrm{O}. \end{gathered} \]

In 1885–1886 Armstrong*** entered into the controversy, advancing the proposition that two absolutely pure substances cannot enter into a chemical reaction and that, for this, the presence of an electrolyte closing the electric circuit is necessary. In the present case the role of the electrolyte must be performed by “current-conducting water” (i.e., water in which

* Journ. Chem. Soc., 49, 106 (1886).

* We quote several excerpts from Principles of Chemistry* that more closely explain Mendeleev’s point of view on this question:

“It seems to me that the matter (the phenomena described, discovered by Dixon and Baker) can be explained by the fact that \(\mathrm{H}_2\mathrm{O}\) with \(\mathrm{CO}\) partly gives \(\mathrm{CO}_2+\mathrm{H}_2\), while hydrogen with oxygen gives \(\mathrm{H}_2\mathrm{O}_2\), which with \(\mathrm{CO}\) forms \(\mathrm{CO}_2\) and \(\mathrm{H}_2\mathrm{O}\). Consequently, water is regenerated and again serves the same purpose (283).

The opinion, always put forward by me, concerning the primary origin of hydrogen peroxide and the formation of water already through its decomposition, has recently begun to spread, especially thanks to Traube.

It may perhaps most simply explain the necessity of traces of water in many reactions, for example, in the explosion of a mixture of carbon monoxide with oxygen, and perhaps the very theory of the explosion of detonating gas and the combustion of hydrogen will gain in clarity and truthfulness if we take into consideration the preliminary formation of hydrogen peroxide and its decomposition...

Since the formation of \(\mathrm{H}_2\mathrm{O}_2\) from \(\mathrm{H}_2\) and \(\mathrm{O}_2\) corresponds to a smaller quantity of heat than the formation of water from \(\mathrm{H}_2\) and \(\mathrm{O}\), it may be that the temperature of the flame of detonating gas depends on the pre-formation of hydrogen peroxide (221) (up to the 6th ed. of Principles of Chemistry, 1895). (A. S.).”

* Pres. Address, Chemical Section, Brit. Assoc., 1885; Proc. Roy. Soc. 40, 287 (1886); Journ. Chem. Soc., London 49**, 112 (1886).

dissolved traces of various substances contaminating the walls of the vessel).

At the same time, oxygen will play the role of a depolarizer:

\[ \begin{array}{c|c|c@{\quad}c|c|c} \mathrm{O} & \mathrm{H_2O} & \mathrm{CO} & & \mathrm{OH_2} & \mathrm{CO_2}\\ \Vert & & & \longrightarrow & & \\ \mathrm{O} & \mathrm{H_2O} & \mathrm{CO} & & \mathrm{OH_2} & \mathrm{CO_2,} \end{array} \]
\[ \text{before}\qquad\qquad\qquad\text{after} \]

All these mutually opposed points of view gave rise to a lively polemic throughout the 1890s, especially between Dixon and Armstrong, who could in no way reconcile their sharply contradictory positions. Armstrong stubbornly insisted on the necessity of accepting his point of view, while Dixon considered utterly improbable, on kinetic grounds, a theory that presupposed the simultaneous collision of five molecules. Their polemic, however, aroused lively interest in this question and served as the occasion for numerous and important experiments which otherwise, perhaps, would never have been undertaken. Quite recently, in 1925, Armstrong again came forward with the assertion that carbon monoxide is “not in itself a combustible gas,” proposing an even more elaborated explanation of the essential role of water in this reaction.*

Dixon opposed to Mendeleev’s theory the fact that, whereas an undried equimolecular mixture of carbon monoxide and nitrous oxide explodes without difficulty, the same mixture after intensive drying no longer explodes from a spark. According to Mendeleev, however, intensive drying cannot in any way hinder the reaction, since here, as before, the equality of the volumes of both components is present. And no reply followed to this observation.

Experiments with cyanogen and the oxygen theory

(1886–1896)

Parallel to this discussion there appeared a series of studies devoted to the combustion of cyanogen. The most important of them we now—

* Proc. Roy. Soc., B, 98, 202 (1925).

let us consider. Let us note that cyanogen is one of the most endothermic gases—according to Berthelot, its molecular heat of formation is 77, whereas for acetylene it is 47 Cal. And since the carbon of cyanogen burns in two sharply separated stages, in the first—to carbon monoxide, and in the second—to carbon dioxide, the study of the combustion reaction of cyanogen is most directly connected with the study of the combustion of carbon monoxide. As early as 1886 Dixon* observed that, when cyanogen is exploded in an excess of oxygen, a complete (theoretical) yield of carbon dioxide is obtained, independent of the moisture content; but at that time the full significance of this observation was hardly realized, since only in 1892 was it shown, again by Dixon, that in the explosive wave a detonating cyanogen–oxygen mixture burns in two stages:**

1) at the wave front . . . . $C_2N_2 + O_2 + O_2 = 2CO + O_2 + N_2,$
2) behind the front . . . . . . $2CO + O_2 + N_2 = 2CO_2 + N_2.$

In 1894 Smithies and Dent demonstrated the presence of the same two stages of combustion by direct analysis of the gases in the inner cone of an ordinary cyanogen–air flame, burning at atmospheric pressure in a Smithies separator burner. Thus final confirmation was obtained that the first product for this reaction is carbon monoxide. And since carbon monoxide at the moment of formation is probably at an exceptionally high temperature, its subsequent combustion to carbon dioxide even without the aid of water vapor thereby becomes natural.

In 1890 Beketov*** showed that a mixture $2CO + O_2$, dried over phosphorus anhydride, becomes explosive after the addition to it of about 10% of likewise dried cyanogen. From this he concluded that the combustion of carbon monoxide is due to the presence of atomic oxygen, whose activity is determined by the high temperature of the flame,

* Jour. Chem. Soc., 49, 384.
** Phil. Trans., 183, 116 (1893).
*** Bull. Acad. Sci. No. 5, 2, 175.

caused by adding cyanogen to the combustible gas. Moreover, in his opinion, water vapor also promotes the combustion of carbon monoxide, insofar as the atomic oxygen needed for the reactions is obtained from it much more easily (i.e., at a lower temperature) than from molecular oxygen, which (as Beketov asserts) is stable at \(2000^\circ\).

Although at that time this point of view impressed Dixon more than all the others, as being the closest to his own, he nevertheless considered that, even if it were correct, only part of the difficulties was removed by it. And he asked: “If carbon monoxide cannot split the oxygen molecule, then why is this possible for free hydrogen?” In addition, the heat of formation of carbonic acid from carbon monoxide is greater than that of water from hydrogen, so that “Beketov*, having removed some difficulties by his theory, has confronted us with new, still greater ones.” Nevertheless, in order to confirm the oxygen theory he carried out several experiments that did not give a positive result.

Mixtures \(2\mathrm{CO} + \mathrm{O}_2\) and \(\mathrm{C}_2\mathrm{N}_2 + 2\mathrm{O}\), dried for seven days over phosphoric anhydride, he mixed with one another in the ratios \(4:1\) and \(2:1\), thus obtaining complex mixtures of the following composition:

	(1), %	(2), %
\(\mathrm{CO}\)	53.3	43.3
\(\mathrm{C}_2\mathrm{N}_2\)	6.7	11.7
\(\mathrm{O}_2\)	40.0	45.0
	100.0	100.0

When a spark was passed through, in both cases an explosion occurred, with the cyanogen burning completely, while the carbon monoxide burned in the first case by 66%, and in the second by 87%. This confirmed the correctness of Beketov’s observation (though not the conclusions he drew from it), since complete—

* Obtained in the dissociation of water.

** The molecular heat of formation of water is \(68.4\ \mathrm{Cal}/g\text{-mol}\), while the heat of formation of \(\mathrm{CO}_2\) from \(\mathrm{CO}\) is \(68.2\ \mathrm{Cal}/g\text{-mol}\) (A. S.).

combustion of carbon monoxide increased with an increase in the intensity of the flame exciting the reaction (with an increase in the content of cyanogen in the complex mixture). As a by-product, a guarantee was obtained of the purity of the Dixon cyanogen used from hydrogen cyanide, since otherwise the combustion of carbon monoxide would have had to be complete in both cases.

After this Dixon prepared a mixture of carbon monoxide and ozonized oxygen of composition \(36\mathrm{CO} + 8\mathrm{O}_3 + 56\mathrm{O}_2\).

After drying it for a week over phosphoric anhydride in the tube shown in Fig. 2, he passed between platinum wires a spark 7 mm long, without obtaining an explosion, although a deep luminous halo was distinctly formed around the spark. Dixon regards this result as unfavorable to Beketov’s oxygen theory, although perhaps not sufficiently convincingly refuting it.

Fig. 2.

Fig. 3.

On the other hand, incomparably more convincing evidence against this theory was provided by the remarkably bold experiments, in conception, carried out by Dixon jointly with Russell in 1897.*

In the glass tube shown in Fig. 3, a previously well-dried mixture of carbon monoxide, chlorine dioxide, and oxygen of composition \(60 : 29 : 11\) (by volume) was sealed over phosphoric anhydride.

After 15 days of drying, a spark was passed between the “platinum wires,” “whereupon a blue flame passed along the tube, lingering somewhat at the end of the tube,” so that there could be no doubt that the flame

* Journ. Chem. Soc., 71, 601.

To p. Bóna

TABLE I

No. 1

Shows explosive combustion before ignition.

No. 2

Shows the negative effect from the discharge of a capacitor, capacitance \(0.5\,\mu\mathrm{F}\), at \(1000\ \mathrm{V}\) (the spark is the white spot in the center of the picture).

No. 3

Shows explosive flame during the discharge of a capacitor of capacitance \(1.0\,\mu\mathrm{F}\) at \(1000\ \mathrm{V}\).

Carbon monoxide

Hydrogen

CO 53%
H₂ 47%

Spectra of CO flames.

covered the whole space. Meanwhile, despite the fact that there was more than enough chlorine dioxide for the complete combustion of carbon monoxide, as gas analysis showed, no less than 60% of it remained unburned. Dixon always presented this result as an almost conclusive argument against Békétov’s theory, since the experiment did not show that “oxygen, just liberated from a compound, exhibits greater activity toward carbon monoxide at high temperature than ordinary oxygen.”

Returning to the combustion of cyanogen, we shall now show an experiment first carried out by Smithells and Dent in 1894.* Into the lower part of a quartz “flame separator” there are introduced separately, through a long “drying tube” with phosphorus anhydride: a) cyanogen from a mercury gas-holder and b) air. The two gas streams, mixing as they pass through the apparatus, leave the upper opening of the tube in the form of a homogeneous mixture. On igniting it, we obtain a cyanogen–air flame with two cones. You see the inner cone, of a beautiful lilac color—in it the cyanogen burns, giving carbon monoxide and nitrogen—and the outer cone, of the usual bluish-green color characteristic of the combustion of carbon monoxide. The separation of the flame into two cones is obtained for the first time when the composition of the cyanogen–air mixture is close to 1 : 3.3. We can now separate the inner cone from the outer one by moving it downward so that a gap of 4–5 cm (vertically) is formed between them. Carefully lowering a jar with air dried over sulfuric acid over the outer cone, we see that its blue flame is rapidly extinguished. If, however, we reduce the vertical distance between the two cones by about half and repeat the experiment again, we shall see that the outer cone now is not extinguished, but continues to burn quietly in air dried over sulfuric acid. This experiment, reproducing the observation of Smithells and Dent, argues that carbon monoxide at the moment of its forma—

* Journ. Chem. Soc., 65, 603.

formation in a cyanogen flame acquires, for a short time, the ability to combine directly with oxygen.

This conclusion was confirmed two years later by comparative measurements of the duration of combustion in the detonation phase of “wet” and “dry” mixtures of \(C_2N_2 + 2O_2\), carried out by Dixon together with Strange and Graham*. Two identical “detonation tubes,” fitted at equal distances from the spark gap with glass windows, were set vertically side by side. One of the tubes, previously thoroughly dried, was filled at atmospheric pressure with a mixture of \(C_2N_2 + 2O_2\), dried for several days over phosphoric anhydride. The other tube was filled with the same mixture, but saturated with water vapor at \(13^\circ\) (\(H_2O = 1.5\%\)). Both mixtures were ignited simultaneously, and the flame, passing by the glass window, was photographed on a film moving in a vertical plane.

To Dixon’s surprise, at both low and high film speeds the two images were stretched in exactly the same way, which, as he says, can mean only that “the carbon monoxide and oxygen formed as a result of the combustion of cyanogen can react without the mediation of water vapor up to the point when the gas mixture cools, and that the duration of this reaction is not noticeably altered in the presence of water vapor.”

The conclusiveness of this experiment depends (and he knew this well) on how free the cyanogen used in the experiment was from hydrogen cyanide; but the attention Dixon paid to the purity of the gases he used, and especially the results described above of his experiments with the mixture \(CO—C_2N_2—O_2\), give almost no grounds for doubting this, just as he himself did not doubt it.

All these data, obtained from experiments with cyanogen, taken together with Dixon’s other experiments, very little

* Journ. Chem. Soc., 69, 773 (1896).

consistent with Beketov’s theory of “atomic oxygen,” at the same time correspond well to the point of view according to which the direct oxidation of carbon monoxide is conditioned by the prior “excitation” of its molecules, which in all

Fig. 3a. I—Dry mixture \(C_2N_2 + 5O_2\); II—the same mixture with an admixture of \(1.3\ \mathrm{cm}\ H_2\); III—the same mixture with an admixture of \(0.9\ \mathrm{cm}\ H_2O\).

probability also occurs in the combustion of cyanogen; and this conclusion, it would seem, is not refuted by observations recently made in Garner’s laboratory, where an “appreciable difference” was found in the total radiation emitted in the explosion of a “moist” and a “dry” mixture \(C_2N_2 + 2O_2\)*.

* Nature, 125, 705, (1930).

The experiments mentioned by Bone are described in Nature in Tavad’s article. If they do not refute Dixon’s overly general scheme of cyanogen combustion in two stages, then, in any case, after them this reaction cannot be regarded as an exception with respect to the participation of water (as was considered on the basis of Dixon’s observations).

Indeed, whereas explosions of the dry mixture \(C_2N_2 + 2O_2\) proceed barely audibly, even a slight admixture of hydrogen and water imparts a strong sound effect to the explosion (even at \(300\ \mathrm{mm}\ \mathrm{Hg}\)). Likewise, as in the combustion reaction of CO with an admixture of water or hydrogen, the radiation is markedly reduced, as is seen from the diagram in Fig. 3a.

It is extremely important to establish how this change in radiation is connected with the change in electrical conductivity and the rate of propagation of the explosion. (A. S.).

PART II

PHYSICOCHEMICAL EXPERIMENTS

Although at the end of the last century general interest was concentrated chiefly on the chemical aspects of the phenomenon, nevertheless a large number of investigators tried to explain them from the physical point of view, especially when the active role of moisture was established not only in the process of combustion, but also in other reactions in which oxidation has no place.

Lothar Meyer on the Influence of Temperature (1886)

As early as 1886, Lothar Meyer, repeating some of Dixon’s experiments, found that a mixture \(2\mathrm{CO}+\mathrm{O}_2\), after being dried for 6 days over phosphorus anhydride, does not explode from a spark in an eudiometer at 156 mm, but can be made to react, though without explosion, at the same pressure simply by passing through it for two minutes a continuous series of powerful sparks; in this case combustion proceeds to completion. This, in his opinion, proves that the combustion of “dry” carbon monoxide seems to depend only on temperature, and that for direct oxidation a higher temperature is necessary than for indirect* oxidation (i.e., through water vapor). To this Dixon replied with an experiment showing that after preliminary heating of carbon monoxide and air dried over sulfuric acid, it is nevertheless impossible to obtain a steady flame. But this experiment can hardly be considered convincing.

Experiments on the Influence of Ionization (1893–1904)

In 1893 Thomson made an extremely important discovery, consisting in the fact that after intensive drying a gas becomes electrically nonconducting even at a considerable potential difference. Proceeding, however, from the electrical nature of the forces binding the atoms in a molecule, he

* Ber., 19, 1099 (1886).

On the article by Bond

TABLE II

No. 1

No. 2

No. 3

No. 4

Effect of protrusion on flame velocity

Advances in Physical Sciences, Vol. XII, Issue 4.

came to the conclusion that the presence in a gaseous medium of droplets of a liquid with such a high dielectric constant as water possesses should, in all probability, weaken the bonds between atoms and thereby increase the reactivity of the molecule. Moreover, in the discussion of this question at the meeting of the British Association in 1910, he drew the attention of chemists to the fact that “in the process of combustion not only atoms and molecules take part, but also electrons—particles of considerably smaller dimensions and very high velocities. They may precede the explosive wave, preparing the way for it by ionizing the gas.” As early as 1894 Braun pointed out that in an explosive wave “ionization” takes place, the result of the combined action of high temperature and chemical interaction. In addition, Turpin, working in Dixon’s laboratory, showed that the flame in the explosive wave of detonating gas \((2H_2 + O_2)\) retains conductivity for \(0.001\) sec., i.e. for almost the same length of time as the luminosity lasts.**

Although Dixon pointed out the difficulty of assuming the presence of liquid water droplets in an explosive flame—for example, in the detonation flame of the mixture \(2CO + O_2\), the moisture content of which is below the saturation point at ordinary temperature (while the flame temperature is not below \(3000^\circ\))—nevertheless he never weakened his interest in this theory. In 1896 he, and independently of him Baker,* attempted to subject a well-dried mixture \(2CO + O_2\) to the action of X-rays while simultaneously passing an electric spark through it, but no ignition was obtained; later, in 1914, together with Campbell and Slater, Dixon** obtained no signs of detonation either in “moist,”

* B. A. Reports, p. 501.
** ZS. f. phys. Chem., 13, 155 (1894).
*** Studies from the Physical and Chemical Laboratories, Owens College, 1, 294 (1893).
**** Journ. Chem. Soc., 69, 789, 1308 (1896).
***** Proc. Roy. Soc., A, 90, 506.

... in a “dry” mixture \(2\mathrm{CO} + \mathrm{O}_2\), placed in a strong magnetic field (up to 1000 gauss).*

But the negative results of these experiments can hardly be regarded as definitive. And it would be highly desirable to repeat them, using the much stronger “ionizing” agents and magnetic fields available to modern experimental technique.

In his Manchester lecture of 1909, “On the Influence of Moisture on Chemical Reactions in Gases,” Baker described experiments which led him to the conclusion that “ionization in the absence of water is chemically ineffective” and that the combined action of “ions” and water vapor is necessary.**

Flammability of the dry mixture \(2\mathrm{CO} + \mathrm{O}_2\)

Apparently, before 1903 no attempts had been made to compare the flammability of \(2\mathrm{CO} + \mathrm{O}_2\) mixtures with different degrees of humidity. It was only in that year that Girvan presented to the society an article*** in which he reported that, whereas a series of powerful sparks from an induction coil could not ignite gas dried at \(-80^\circ\), ignition always occurred when the “drying temperature” lay between \(-50^\circ\) and \(-35^\circ\).

And after drying at \(-15^\circ\), the mixture exploded already from a single spark. In addition, he also observed that an explosion in gas dried at a temperature from \(-35^\circ\) to \(-50^\circ\) always proceeded more slowly and more mildly than in moist gas.

Thornton’s remarkable experiments of 1914–1916 on the electrical ignition of explosive gas mixtures****

* Here, in Bon’s account, there is a certain inaccuracy: Dixon, placing different sections of the explosion tube between the poles of an electromagnet, tried to obtain a change in the rate either in the detonation phase or in the initial phase (at the point of ignition, where the rate of propagation of the explosion is relatively low). But in both cases the experiment gave a negative result. (A. S.).

** Mem. Manchester. Phil. Soc. 53, No. 16, 7 (1909).

*** Girvan P. 19, 230. 1903.

**** Pros. Roy. Soc., A, 90, 272; 91, 17; 92, 9, 381.

To the article by Bock

TABLE III

5

6

Effect of the electric field on the flame velocity

Uspekhi fizicheskikh nauk, vol. XII, issue 4.

provided decisive proof that, for given conditions of spark formation (electrodes, type of discharge, voltage, etc.), the ignition of a definite mixture at a given temperature and pressure requires a definite “minimum ignition current.” For example, to ignite a “moist” stoichiometric mixture of carbon monoxide with air at room temperature and atmospheric pressure, a condenser discharge at 100 V requires a minimum capacitance of about 2 μF. In this connection the natural question arises whether the problem of igniting strongly dried mixtures of carbon monoxide is reduced first of all to obtaining a sufficiently powerful spark. All the more so since, in Dixon’s early experiments, the spark used was from an induction coil of very moderate power.

Although I often discussed this supposition with him, only after the war did I obtain in my laboratory the opportunity to test it experimentally. I shall describe some of the most important results of our experiments, with which Dixon was directly connected up to his death.

Experiments at Imperial College (1920–1930)

1. The effect of progressive drying on the electrical ignition of mixtures \(2\mathrm{CO} + \mathrm{O}_2\).

a) Together with Weston* we investigated the effect of progressive drying of the mixture \(2\mathrm{CO} + \mathrm{O}_2\), from “saturated” at room temperature to “dried with calcium chloride,” on the minimum capacitance required for ignition by a discharge between platinum electrodes from a condenser charged to 110 V. The results, which lie on the hyperbola shown in Fig. 4, indicate a rapid increase in the minimum capacitance with increasing degree of drying, so that at the limiting drying with calcium chloride (\(\mathrm{H}_2\mathrm{O} =\) about \(0.03\%\)) this capacitance is almost 30 times higher than that required for the mixture “saturated at \(17.4^\circ\)” (\(\mathrm{H}_2\mathrm{O} = 2.0\%\)).

* Proc. Roy. Soc. A, 110, 614 (1926).

In discussing these results it should be recalled that the discharge of the capacitor is oscillatory, extremely brief, with a frequency of about a million per second. It begins with impact ionization of the gas between the electrodes, its maximum being reached in the very first oscillation. In addition, very high temperature and local pressure are produced in the discharge, which is manifested in the crackling sound produced by the spark. The igniting ability of the discharge depends on the character of its first oscillation, which is determined by the properties of the circuit.

Fig. 4.

b) The next stage of our experiments was to determine the possibility of exploding, by means of a sufficiently powerful spark of a capacitor discharge, the mixture \(2\mathrm{CO} + \mathrm{O}_2\), dried over phosphorus anhydride to the greatest possible extent. Accordingly, a carefully purified and preliminarily dried mixture \(2\mathrm{CO} + \mathrm{O}_2\) was sealed at atmospheric pressure in a cylindrical glass vessel of about \(100\ \mathrm{cm}^3\) capacity, with platinum spherical electrodes, over redistilled phosphorus anhydride (Fig. 5).

The inner surface of the vessel had first been completely cleaned, and the electrodes were kept red-hot for 20 hours in vacuum and in oxygen at \(2\text{–}3\ \mathrm{mm}\), in order to ensure the removal from the system of organic impurities and hydrogen.

Ten such vessels were prepared by the method described* and were kept for a “drying period” of from 150 to 1000 days, during which time the gas remained in contact with phosphorus anhydride.

* Ibid., 615; 123, 285 (1929).

In addition, every two weeks the vessels were heated to \(2000^\circ\) with simultaneous cooling of the lower part, containing phosphorus anhydride, in order to remove the water film adsorbed by the inner surface of the vessel. Thus here, undoubtedly, an extreme degree of drying was attained (all the more so since it is reached already after 250 days and even, probably, earlier).

In the experiments subsequently carried out, 8 vessels were tested; two of them were left in reserve. In each of the vessels tested a discharge was produced from a condenser charged to 1000 V, the gas mixture sustaining a discharge up to a capacitance of 0.5 \(\mu\mathrm{F}\) (the total energy of the discharge being equal to 0.25 joule) and invariably exploding at a capacitance of 1 \(\mu\mathrm{F}\). The minimum capacitance necessary for ignition was approximately 0.75 \(\mu\mathrm{F}\). In all cases the flame spread through the whole vessel faster than the eye could follow, with from 74 to 89% of the gas burning. On 11 November 1927 Dixon was present at the explosion of one of these vessels (No. 5), whose drying period had lasted 555 days and which on the preceding day had sustained the discharge of a condenser of 0.5 \(\mu\mathrm{F}\) at 1000 V without any sign of ignition.

Fig. 5.

The explosion was obtained with a discharge of twice the capacitance at the same voltage, and 87.8% of the gas burned. Dixon openly acknowledged this experiment as maximally convincing. The appearance of these very important experiments is shown in the photographs obtained for vessel No. 6, whose drying period lasted 758 days (Table I, Nos. 1, 2, 3).

No. 1 shows the vessel before ignition; in the lower part of the vessel the phosphorus anhydride is clearly visible. No. 2 was taken in a darkened room and shows the negative result obtained with the discharge of a condenser of capacitance 0.5 \(\mu\mathrm{F}\) at 1000 V (the spark is the white dot in the center; it should especially be noted that around it there is not the slightest sign of a halo that would indicate the presence of a combustion process-

No. 3 reproduces an explosion as a result of ignition by a discharge with a capacitance of \(1\,\mu\mathrm{F}\) and \(1000\,\mathrm{V}\) (we note that the flame completely filled the whole vessel). In this case the combustion proceeded to \(89.25\%\).

The 9th of these vessels (not yet tested) is here before you on the table, and we can give you the opportunity to witness its test. It was filled on July 15, 1929, with a well-dried mixture \(2\mathrm{CO}+\mathrm{O}_2\), and, consequently, its drying period “over phosphorus anhydride” lasted 514 days. Its electrodes are connected to a capacitor of capacitance \(0.25\,\mu\mathrm{F}\), charged to \(650\,\mathrm{V}\).

You see that when the capacitor is discharged through the gas mixture, despite a perfectly distinct spark, no explosion occurs, since the energy of the spark is considerably below the minimum required for ignition.

Having increased the capacitance of the capacitor to \(4\,\mu\mathrm{F}\) and again charged it to \(650\,\mathrm{V}\), we thereby stored twice as much energy as is necessary, and, as you see, now upon discharge an explosion occurs and the flame rapidly fills the entire vessel. (When the vessel was opened several days later it was determined that the combustion products contained \(\mathrm{CO}_2\)—66.4; \(\mathrm{CO}\)—22.0; \(\mathrm{O}_2\)—11.1; \(\mathrm{N}_2\)—0.5%, so that the combustion had proceeded only to 76%.) The conclusion forced upon us by these experiments may be formulated as follows: an extremely pure mixture \(2\mathrm{CO}+\mathrm{O}_2\), dried to the greatest possible degree over phosphorus anhydride, can nevertheless be ignited if sufficient power of the spark discharge is provided.

2. Spectrographic data

The fact that the flames of carbon monoxide and of hydrogen, as can be seen here, differ sharply from one another both in color and, in general, in outward appearance, has always seemed to me difficult to reconcile with Dixon’s original point of view, namely, that the oxidation of carbon monoxide as a whole “amounts to the alternating reduction and oxidation of water molecules,” although it cannot be denied that in the presence of water vapor the partial combustion of carbon monoxide

and occurs in this way. To clarify this question, Veston* undertook a systematic study of the spectra of the flames of hydrogen, carbon monoxide, and their mixtures in various proportions. With the kind cooperation of the eminent contemporary spectroscopist Prof. Fowler, we obtained three typical spectrograms of CO flames with an excess of oxygen, shown in Table 1 below (from 500 Å in the visible to 3200 Å in the ultraviolet region).

No. 1, taken from the flame of pure (undried) carbon monoxide, shows a sharp and undoubtedly continuous spectrum with groups of the so-called “water-vapor lines” (“steam lines”), standing out distinctly in the region \(3200+ +3060\) Å. With moderate drying of the gas or with an increase of pressure to approximately 10 atm these lines disappear completely, while the continuous spectrum remains unchanged. When the pressure is lowered the spectrum is overlapped by the characteristic band spectrum of carbon monoxide, which (in the words of Prof. Fowler) is “entirely different from the better-known bands of the oxides of carbon observed in vacuum tubes.”**

No. 2—the spectrum of a flame of pure hydrogen—shows nothing except several groups of “water-vapor lines” near the ultraviolet region, with maximum brightness at \(3200—3060\) Å. The sharp differences between these two spectrograms are of great significance. With the gradual addition of hydrogen to burning carbon monoxide, the continuous spectrum characteristic of it rapidly disappears, and the “water-vapor lines” begin to predominate in the spectrum.

Finally, at 50% CO/50% \(\mathrm{H_2}\) (“water gas”) the entire spectrum consists only of the latter, as is seen from spectrogram No. 3. Similarly, if two gas jets of hydrogen and carbon monoxide are mixed so that at the burner orifice an equal-volume mixture of them is formed,

* Proc. Roy. Soc. A, 109, 177, 523 (1925).

** The band spectrum of the flame, attributed by Fowler to carbon monoxide, according to the latest studies is emitted by \(\mathrm{CO_2}\) molecules (see V. Kondrat’ev, ZS. f. Phys., 63, 322 (1930). (A. S.).

then in this case a flame is obtained which can hardly be distinguished in appearance from the flame of pure hydrogen.

From the point of view of these data it is difficult to resist the conclusion that, in the combustion of undried carbon monoxide in equally undried air at normal pressure, two reactions proceed simultaneously:

a) direct interaction between “excited” carbon monoxide and oxygen, giving radiation corresponding to the continuous and banded parts of the spectrum and determining the characteristic bluish-green color of the flame, and b) interaction between carbon monoxide and water molecules, causing the appearance in the spectrum of the “water-vapor lines.” With the gradual addition of hydrogen to carbon monoxide, the ratio between (a) and (b), initially very large, rapidly decreases, until, in a 50/50 mixture, reaction (b) becomes predominant, practically displacing reaction (a). These spectrograms made a strong impression on Dixon. And, in his words, they “for the first time furnished convincing proof that in an ordinary carbon monoxide flame both the direct and the indirect reactions proceed simultaneously.”

The necessity of the presence, for reaction (a), of “excited,” but not necessarily “ionized,” carbon monoxide is indicated by the totality of the data obtained from the study of explosions of carbon monoxide with air at high initial pressures, and also from recent determinations of flame velocities for the entire explosion region of “wet” mixtures of carbon monoxide with oxygen.*

These experiments showed that both in the initial “uniform period” and in the detonation phase there is a sharply expressed maximum of velocity for the composition \(3.8\,\mathrm{CO} + \mathrm{O}_2\), and this point is shifted little when the mixtures are diluted with nitrogen, argon, or helium.**

* Proc. Roy. Soc. A, 130, 542 (1931).

** In the work by Bone mentioned in the text it was shown that the maximum velocity lies not for the stoichiometric mixture, but for a mixture of CO with oxygen rich in fuel. Dilution with oxygen lowers this velocity, and Bone interprets this fact as follows: the flame velocity

In this same connection it should be mentioned that there is a substantial body of data showing that hydrogen accelerates the combustion process of carbon monoxide considerably more than an equivalent amount of water. In this case their action may be at least twofold—at very low concentrations, physical, and at higher concentrations, chemical. In the latter case the possibility of several reaction mechanisms is probable. Moreover, the direct oxidation of carbon monoxide is strongly favored by pressure, so that at very high pressures combustion proceeds exclusively by this path. But even in this case a small admixture of water can accelerate the reaction by a purely physical action. Judging both from personal statements and from published articles,* Dixon essentially accepted this point of view.

3. Combustion in an Electric Discharge

It is also necessary to mention the independent work of my colleague Finch and his co-workers, devoted to the “cathode combustion” of carbon monoxide, work that shed a bright light on the mechanism of combustion in an ordinary flame.**

Accepting our point of view, according to which “unexcited” molecules of carbon monoxide are inert with respect to oxygen, the authors of the work suppose that direct oxidation requires “excited” (but not “ionized”) carbon monoxide and atomic oxygen. In addition, they consider possible the “auto-oxidation” of carbon monoxide \((2\mathrm{CO}=\mathrm{C}+\mathrm{CO}_2)\), with the carbon, combining directly with oxygen, again giving carbon monoxide.

As for the role of water and hydrogen, they believe that the latter, during combustion, forms hydrogen peroxide and “excited” water molecules. Both the one and the other can oxidize carbon monoxide much better than oxygen itself—

is determined chiefly by the concentration of CO, which indicates the leading role in the combustion process of previously “excited” CO molecules, and not of oxygen. (A. S.)

* Nature, 122, 805 (1928).

* Proc. Roy. Soc., 124, 303 (1929); 125, 352 (1929); 129*, 314, 656 (1930).

...of hydrogen. These, and perhaps other possible reaction schemes, are subject to future discussion, since the question has proved to be far more complex than it seemed at first.

It is possible that the correct solution lies in the fact that, for the oxidation of carbon monoxide by hydrogen peroxide, water, and oxygen, different degrees of “excitation” of its molecules are required, and possibly also some “excitation” of the oxidizing agents themselves.

4. Photographic Data

The analysis of explosive flames by the photographic method, first applied 50 years ago by Mallard and Le Chatelier, and then developed and improved in the 1890s by Dixon and his collaborators and, finally, recently brought to an still greater degree of accuracy in the new high-speed camera constructed by Fraser in our laboratory, now represents such a powerful research tool that it is appropriate to conclude this review with the data obtained with its aid on the influence of drying on the propagation of flame in the mixture \(2\mathrm{CO} + \mathrm{O}_2\).

It should be recalled that the principle of this method is reduced to photographing a flame, moving with velocity \(x\) along a horizontal glass tube, on a sensitive film moving in a vertical plane with a known velocity \(y\), so that the trace of the flame represents the resultant of the two velocities.

A) The influence of progressive drying on the flame velocity. Of the many excellent photographs taken by Fraser,* four, of particular interest, are reproduced in Table II. Here an explosive flame, moving through the mixture \(2\mathrm{CO} + \mathrm{O}_2\) at atmospheric pressure along a horizontal tube (Fig. 6) (35 cm long \(\times\) 2 cm internal diameter), was photographed on film moving with a constant velocity of 148 cm/sec in the first three photographs and 94 cm/sec in the fourth.

* Phil. Trans., A, 228, 202 (1929).

In all cases ignition was produced by the discharge of a capacitor between platinum electrodes in the middle of the tube. From this point the explosion propagated along the tube in both directions toward its ends.

No. 1—with a “moist” mixture of $2CO + O_2$ ($H_2O = 1.7\%$). Both flames arise simultaneously at the spark and at first move with acceleration up to the “shoulders” at a distance of 5 cm.

After this the propagation proceeds considerably more slowly, already with a constant velocity of about 1200 cm/sec, to the ends of the tube. In this case combustion proceeds to completion, to 100%.

No. 2—with a mixture of $2CO + O_2$, dried for 7 days over $CaCl_2$ ($H_2O = 0.03\%$). Here, in comparison with No. 1, there is a noticeable slowing of combustion.

Fig. 6.

The almost constant velocity of the flame after the first “shoulder” is now reduced to 420 cm/sec; combustion proceeds only to 97.2%, and the duration of the afterglow is twice as great as in No. 1.

No. 3—with a mixture dried for 80 days over $P_2O_5$—shows how sharply combustion is slowed when the last 0.03% of moisture is removed from the mixture; in this case the total duration of luminescence is now almost 14 times greater than in No. 1 and almost 8 times greater than in No. 2; combustion is only 89%.

In No. 4—with a mixture of $2CO + O_2$, dried for 223 days over $P_2O_5$—the extreme limit of drying is reached, with a further slight slowing of combustion. Nevertheless the flame invariably reaches the ends of the tube, combustion proceeding to 92%, while the duration of luminescence is 15 times greater than in No. 1.

The data from the photograph are reduced to the following table:

	No. 1	No. 2	No. 3	No. 4
Time (milliseconds) at which the flame reaches the ends of the tube	16.9	23.6	287	308
Total duration of glowing (in milliseconds)	27.5	50.0	388	416
% combustion	100	97.2	89	92

B) The influence of a strong electric field on the propagation of the flame through an intensely dried mixture \(2\mathrm{CO}+O_2\). 1) On repeating experiment (No. 4) with a mixture \(2\mathrm{CO}+O_2\), dried for 245 days over phosphoric anhydride, but in a longer (60 cm) tube, the slow propagation of the flame could at first be followed with the eyes, until the flame stops and remains in place for about 53 milliseconds (Table III, No. 5). Then a further, very slow propagation toward the end of the tube begins, after which the flame retreats slightly back into the partially burned mixture, where it is extinguished.

The average velocity of the flame is only 72 cm/sec, and the combustion process is prolonged here to 550 milliseconds; only 26.5% of the entire mixture burns.

2) Assuming that the accelerating action of moisture on the direct oxidation of carbon monoxide is initially of a more electrical than chemical character, we repeated experiment No. 5, placing the intensely dried mixture \(2\mathrm{CO}+O_2\) in a strong electric field, expecting to obtain some effect on the propagation of the flame.

Into exactly the same explosion tube, filled with the dry mixture \(2\mathrm{CO}+O_2\), two electrodes were additionally fused in at a distance of 30 cm from the central ignition point (Fig. 7). After 240 days of drying over phosphoric anhydride placed at both ends of the tube, the mixture was ignited at the center of the tube in exactly the same way as before; in this case the two additional electrodes were connected to the poles of a Wimshurst machine, which operated throughout the entire combustion process.

From photograph No. 6 (Table III) it can be seen that two

flames spread from the spark gap during the first 100 milliseconds extremely slowly and at almost constant velocities of 43.5 and 50 cm/sec.

This initial phase is then followed by another, marked by so considerable an acceleration of both flame fronts as they approach and cross the region of the field electrodes, that the flame rapidly reaches the ends of the tube. At the same time the glow increases and spreads backward, so that the entire tube is completely filled with flame. The total duration of the glow, from beginning to end, is 375 milliseconds, and the percentage of combustion is 89.6. Another

Fig. 7.

important fact was that the flame advanced much more rapidly in the direction toward the negative than toward the positive pole of the field.*

Dixon, who showed exceptional interest in these experiments, expressed himself as follows about these two photographs: “It is evident here that the resistance created for the reaction by the dryness of the gases can be overcome by an electrostatic field; and, as it were, the action of the field is manifested considerably more strongly at one pole than at the other…

Thus the problem of the combustion of this gas becomes one of the most interesting problems of physical chemistry.”**

With these last printed lines of his my survey ends.

In conclusion, a few more words in his memory. Dixon was not only an outstanding master of experiment, but also a brilliant mentor to those who had the good fortune to be his pupils.

* On the latest experiments with an electric field see Bone, Proc. Roy. Soc. 132, 1 (1931).

** Nature, 124, 582 (1929).

And his principal merit lies in the fact that he founded in our country a school of researchers of combustion processes, one that absorbed the best traditions of Robert Boyle and Humphry Davy. In a letter addressed to me, dated August 26, 1927, on the occasion of the publication of the book dedicated to him, Flame and Combustion in Gases, he wrote:

“You, I think, know that I regard as the best reward for all my work the fact that, whereas 50 years ago, when I set about repeating Bunsen’s experiments, no one in England was interested in questions of gas combustion, now we have an active English school, with a substantial number of experienced scientists, with an ardent thirst for the profound study of gas reactions.”