Measurement of the Temperatures of Stationary Flames1 - Industrial

Ind. Eng. Chem. , 1928, 20 (10), pp 1004–1008. DOI: 10.1021/ie50226a006. Publication Date: October 1928. ACS Legacy Archive. Cite this:Ind. Eng. Che...
0 downloads 0 Views 731KB Size
I,VDUSTRIAL A S D ENGINEERI*VG CHEMISTRY

1004

ical analysis in the manner just described, with the photographic measurements of all flames from the particular gas, affords strong support to our theory of diffusion flames. An illustration of this procedure for the case of carbon monoxide flames is given by the following series of experiments. Flow of carbon monoxide = 4 cubic feet per hour Flow of air = 5.55 cubic feet per hour

R = 0.631 inch L = 0.408 inch

L / R = 0.645 The carbon monoxide contained 3.2 per cent nitrogen. A small water-cooled sampling tube was introduced vertically into the flame and small samples of gas were withdrawn from different points on the axis of the tubes. The samples were taken a t such a rate that the shape of the flame and the conditions of flow were not affected. Duplicate determinations were made in every case. The results of the analysis are given in Table VII. Table V I 1 Y

Inches

h-2

c o t

Per cent

Per cent

co

0 2

Per cent

Per cent

Vol. 20, No. 10

lated analysis a t the point in question. From chemical analysis the value of the coefficient of diffusion of carbon monoxide was calculated to be 0.104 square inch per second according to the procedure described above. This value was used to determine the theoretical heights of a number of carbon monoxide flames. I n Table VI11 the data concerning these flames together with their measured and calculated heights are given. R

L

Inch

Inch

0.63 0.63 0.63 0.63

0.42 0.42 0.42 0.42

Table VI11 CARBON FLAME HEIGHT MONOXIDEExperiTheomental retical Cu. f f . / h r . Cu. ft./hr. Inches Inches 4.2 3 0.92 0.94 8 0 6 1.75 1.75 3.25 3.75 1.02 1.02 5.55 6.25 1.75 1.76 AIR

I n the last two experiments the flame was “inverted”i. e., air was passed up the inner tube and carbon monoxide up the outer tube. The agreement between the experimental and calculated heights of these carbon monoxide flames needs no further comment. Conclusion

I n Figure 8 these results are shown graphically, together with the curves which were calculated theoretically on the assumption that the coefficient of diffusion for CO = 0.104 square inch per second. The agreement between the theoreticaI curves and the experimental points is evident. The method of calculating the concentrations of the various constituents yielding the theoretical curves shown on Figure 8, while complicated and laborious, is sufficiently straightforward to require no detailed description. The procedure adopted was to consider the interdiffusion of the two gases to take place as though no flame were present. The mixture of gas so calculated for any given point was then assumed to react and the products thus obtained gave the final calcu-

The theory of diffusion flames proposed herein shows such good agreement with the experimental facts that we feel justified in the hope that its adoption, in essentials a t least, may lead to a better understanding of, and further contributions concerning, this very common and interesting class of flames. The authors feel that a more fundamental investigation of the phenomena described here on the basis of the kinetics of the chemical reactions involved might yield both interesting and profitable results. Acknowledgment

The authors wish to express their appreciation to the Combustion Utilities Corporation for permission to publish this work and also to A. Hall and P. S. Roller for valuable assistance rendered in the course of this investigation.

Measurement of the Temperatures of Stationary Flames‘ A. G. Loomis and G. St. J. Perrott FLAME LABORATORY, PITTSBURGH

EXPERIMENT STATION,

The concept of temperature as applied to flames is discussed. A number of proposed methods for measuring the temperatures of flames are critically reviewed and the optical method of Kurlbaum-Fery is described and examined in detail. This method depends upon comparing the brightness temperature of a continuous radiator with the brightness of the radiation from the flame colored with an alkali-metal vapor at a given spectral line. From a consideration of the laws of radiation it is shown that the true flame temperature is equal to the brightness temperature of the comparison radiator, as read with an optical pyrometer, when the spectral line is just reversed as seen in a spectrometer. Curves representing flame temperature as a function of air-gas ratio as measured by the line-reversal method are given for Pittsburgh natural &s, methane, propane, and carbon monoxide. These results are compared with measurements depending on the flame gases heating a 1 Published by permission of the Director, U. S. Bureau of Mines. (Not subject to copyright.)

u. s. B U R E A U OF M I N E S ,

PITTSBURGH, P A .

solid radiator contained in the flame and with the calculated results for the maximum temperature attainable at complete combustion.

N INVESTIGATION of the temperature of stationary flames is being carried on a t the Pittsburgh Experiment Station of the U. S. Bureau of Mines in connection with studies of flame propagation. Experiments

A

are being made with stationary flames preliminary to measurements on flames produced by coal-mine explosives. Although Kurlbaum2 in 1902 described a method for the measurement of the temperature of carbon-containing flames such as the candle, illuminating gas, and acetylene flames, nevertheless very few measurements of the temperatures of stationary flames are recorded in the literature. Recourse is taken to the calculated flame temperature, utilizing the A H of the reaction, the specific heats of the products, and the degrees of dissociation of carbon dioxide and water. Owing to our lack of knowledge of the true specific heats, especially

* Physik. Z.,8, 187, 332 (1902).

October, 1928

INDUSTRIAL AND ENGINEERING CHEMISTRY

above 2000” C., and particularly the amount of radiation from a given flame, these calculated maximum temperatures must bear an uncertain relation to the actual temperature. It is highly desirable, therefore, to measure the temperatures of flames under known conditions of air-fuel ratio, design of burner, etc., and to supplement these measurements of temperature with radiation measurements under exactly the same conditions. I n this paper some of the established methods of flame temperature measurement are compared and the results obtained with several gases using an optical method due to Kurlbaum, as modified by Fery. It is generally stated that a flame cannot have a temperature in the strict sense of the word, because it is not a system in equilibrium. Chemical reactions are taking place; part of the energy may be radiated by chemiluminescence and a certain time taken for the transformation of the remaining energy into kinetic energy of the products. Ames3 says that “for a body not in a state of statistical equilibrium, it is not allowable to use the word ‘temperature.’ This means that the word should not be used with reference to a single molecule or to such phenomena as occiir in most flames, or in an electric spark or discharge. It is true that bodies p!aced in flames, sparks, etc., may assume definite temperatures, but this does not affect the statement just made.” Haber’ believed, however, that the region of a flame in which reactions are taking place is confined to surfacesfor example, the boundary of the inner and outer cone of a Bunsen flame-and that in the main body of the flame little reaction is taking place and thermal equilibrium may be nearly attained. He has shown that the water-gas equilibrium adjusts itself very rapidly in the flame of a Bunsen burner, and has calculated the temperature in the inner cone from gas analyses made in the flame separator of Smithelk and Ingle.6 Maxwell conceived the temperature of a body as “its thermal state considered with reference to its power of communicating heat to other bodies.” If we made this our criterion of flame temperature, we might employ as the other body a thermocouple inserted in the flame. Values obtained by this means differ from one another depending on t,he size of the couple, but to a certain extent corrections may be made for this variation. Other possibilities arise in connection with measurements of radiation from flames, and it appears that for certain wave lengths, at any rate, a flame sufficiently thick may radiate as a black body. The method of spectralline reversal finally adopted by the writers seems to be most free from theoretical objections, and the experimental manipulations involved are made easily and rapidly. The various methods employed by other workers are discussed in the following text. Methods Used by Other Investigators

Several well-recognized methods may be used for such measurements. Nichols6 introduced thermocouples into the flame to various accurately measured distances from the median plane of the flame and by extrapolating to distance zero found t = 1760” C. for the middle of the Bunsen flame, and also the temperature gradient in the flame. Waggener7 introduced platinum, platinum-rhodium thermocouples of decreasing thickness into the flame and by extrapolation to a thermocouple of zero thickness corrected for loss due to conduction and radiation, obtaining 1760” C. for the Bunsen flame.

* “Pyrometers and Pyrometry,” Mining

M e t , 7, 37 (1920).

by Lamb, p. 300, Longmans, Green & Co., 1908. 6 J. Chem. SOC.( L o n d o n ) ,61, 204 (1892). 0 Phys. R e v , 10, 234 (1900). Yerhondl. deul physrk. Ges., 14, 78 (1895).



Burkenbusch* likewise introduced a thermocouple directly into the flame and compensated for heat loss by radiation by heating the wire electrically. To accomplish this he represented the watt-consumption of the couple as a function of its temperature when suspended in a vacuum. The couple was then suspended in the flame and again electrical energy was added and represented as a function of the temperature of the couple, giving a curve that was of less slope than the first curve. The intersection of these two curves was taken as the temperature of the flame, because the radiation loss was assumed to be the same in each case, and when the couple and the flame are at the same temperature no heat is added from nor lost to the surrounding gases, a condition entirely duplicated in vacuo where there is no gas for such addition or loss of heat. This method was improved by S ~ h m i d t . ~ Instead of tl thermocouple he used a platinum wire, measuring its temperature optically and determining its heat loss by radiation in absolute units when heated electrically outside the flame by measurements with a thermopile. Inside the flame the consumption of electrical energy was measured a t the same time as the temperature of the wire (optically); this was likewise put in absolute units and the sum of the loss through radiation and convection given. His measurement by this method gave 1800” C. a t the hottest point of the Bunsen flame. It is not clear how Schmidt corrected for heat lost by convection when the wire was heated in air. A method recently proposed from the National Physical Laboratorylo is quite similar to the method of Schmidt. A refractory metal in form of wire is heated electrically in vacw and the relation between temperature and heating current is determined by an optical pyrometer. The same wire is then inserted into the flame and the relation between temperature and heating current is again determined. When the results are plotted graphically the point of intersection of the two lines is assumed to give the temperature of the flame,for a t the temperature r e p resented by this point the electrical supply is sufficient to balance the radiation loss, whether the wire is an uacuo or in the flame, so that the surrounding gas in the flame neither imparts nor abstracts heat from the wire. The temperature of the Bunsen flame measured by this method was 1765” C. A sample of the gases undergoing combustion may be withdrawn from the flame a t given points through a water-cooled tube for analysis and, together with the heat of combustion, the specific heats of the combustion products, and equilibrium data, the temperature may be found, as was done by Haber and Hadsman” in the case of the acetylene-oxygen flame. The conductivity of the flame containing a known concentration of an alkali salt may be measured, and from this e is the equilibrium constant K of the reaction 144 +M+ calculated. The equilibrium constant for the ionization of a metal vapor is connected with the temperature by means of Saha’s equation:lZ

+

log10 K =

- 5048 V + 2.5 loglo T - 6.56 T

where V is the ionization potential of the metal vapor present in the flame. Measurements of this kind with coal-gas flames have been made by Barnes13 and by BennetP who, however, were not making measurements of flame temperatures, but rather proving the validity of Saha’s equation. The general method of measuring the emissive and absorp Wud Ann. 67, 649 (1899). Deul phys Ges , 11, 87 (1909). 10 National Physical Laboratory, Report for 1926, p. 63. 11 Z Physrk Chem , 67, 343 (1909). 12 Phd Mag , 40, 478 (1920). 13 Phys. R e v , 23, 178 (1924). l4 Phtl. Mag , 171 63, 127 (1927). see also Noyes and Wilson, Astrophys. J , 57, 20 (1923). 8

9

‘ ”Thermodynamics of Technical Gas Reactions,” English translation

1005

ISDUSTRIAL AiYD ENGINEERIAVGCHEMISTRY

1006

tire power of the non-luminous flame may be used, since the radiation spectrum of such flames contains the bands of carbon dioxide and water in the infra-red. SchmidtI5 found that the law of black-body radiation held quantitatively ~ 4.4~.From the measured values €or both bands a t 2 . 7 and of the emissive and absorptive power of the Bunsen flame in the field between X = 2 . 4 1 and ~ X =4.6p, Schmidt was able to obtain the true temperature of the flame, making use of the radiation laws of Kirchhoff and of Planck. Henning and TingwaldtlGhave very recently used this method to measure the temperature of the acetylene-oxygen flame.

.

1-01. 20, s o . 10

radiator if the latter is cooler than the flame; but if it is hotter the bright lines are reversed and appear dark upon the brighter background of the continuous spectrum. Final adjustment is made by the observer at the eyepiece of the spectrometer with the rheostats controlling the heating current through the comparison radiator until the lines are just a t the point of reversal. The brightness temperature of the comparison radiator now is the same as the temperature of the flame, and this temperature is determined by means of the optical pyrometer, keeping the current constant through the band lamp. Theory of M e t h o d of Line Reversal

At exact reversal the following relation must hold EF PF E R = E R (1) v here EFand ER are the respective spectral brightnesses of the flame and the comparison radiator for the wave length A, and PF is the permeability of the flame a t the same wave length. If the absorptive power of the flame is given by A F and the reflective power as RF.then we have EFIAF = (1 -k RF/AF) E R (2)

+

k Figure 1-Line-Reversal

Apparatus

since RF

It mill be noted that all of the above methods except those of Kurlbaum-Fery and of Schmidt are strictly limited to flames whose temperatures lie below 1800' C., because no material that does not melt or oxidize may be used for thermocouples or electrodes or wires, while a water-cooled tube introduced into the flame to remove samples of gas for analysis may cause serious errors due to conduction and radiation. The method of Schmidt in the infra-red is difficult to apply and demands very special equipment. Thermocouples of platinum, platinum-rhodium are easily contaminated by the flame and very erroneous results are obtained from this cause: and there ill always be the question of catalytic action due to any material introduced into the flame with attendant abnormal temperatures. h l e t h o d of Line Reversal

The optical method of Kurlbaum? subsequently modified by Kurlbaum and Schulze,'7 and by Feryla may be used a t

any temperature above 900" C., and has recently been used by Henning and T i n g ~ a l d t to ' ~ measure the maximum temperature of the acetylene-oxygen flame, 3100" C. The method of Kurlbaum-Fery is one in which a continuously radiating body, such as a tungsten band lamp or an electric arc, is viewed through a spectrometer, with the flame colored by means of an alkali-metal salt, between the radiator and the spectrometer. The method may be understood by reference to Figure 1. The tungsten band lamp a, serving as the continuous radiating body, is heated from a set of storage batteries by means of appropriate resistances and is focused by means of lens b into the center of the flame c, just above the tips of the cones. The brightness temperature of the band lamp is read by means of the optical pyrometer the screens y and A allowing radiation to emerge from the same point of the band. The image of the band together with the flame is focused on the slit of the spectrometer e by means of the lens d. A screen, i, allows radiation from an area of the flame equal to the area of the band-image t o reach the spectrometer. The pair of spectral lines from the flame colored with sodium chloride vapor at X = 0 . 5 8 9 ~are seen as bright lines upon the continuous spectrum from the comparison f 3

16

A n n P h w i k , 29, 1027 (1908).

Z.Physik, 48, 805 (1928) 17 Deut p h y * Ges , 6, 428 (1903) 16

18

10

Comp. r e n d . , 157, 909 (1903) 805 (1928).

Z. Physik, 48,

+ PF + A F = 1.

If the reflective power of the flame is so small that it may be neglected in comparison to the absorptive power, then EFIAF = E R

(3)

From Kirchhoff's law it follows from equation (3) that the true flame temperature is equal to the true temperature of the comparison radiator, if this is a black body, or to the apparent temperature as measured by an optical pyrometer if the radiator is not a black body. Three possible questions arise T\ ith the method outlined: (1) Does the temperature of the flame remain unchanged when it is colored n-ith a metal salt? ( 2 ) Does Kirchhoff's law apply accurately to emissive and absorptive processes for a , spectral line? (3) Can the reflective power of 1~ the flame be neglected when it is colored with a metallic salt? (1) It was shown by Kohn20that for the Bunsen flame the same tem- ? Ism perature was obtained within the experimental 3 error (10') with the line- e reversal method, the 1 flame being colored with C- li60 a sodium salt, and for thF non-colored flame using the method of Schmidt, as outlined above This was also proved by Henning and Tingwaldt'e in the case of acetylene-oxygen flames by measuring the total radiation from the 8 9 10 11 I? TOTAL HYDRbCARBOhS PERCENT colored and non-colored flames by means of a Figure 2-Pittsburgh Natural Gas vacuum thermoelement. Kirchhoff's law as embodied in equation (3) can apply (2) only when the emissive power EF and the absorptive power AF of the flame apply to exactly the same process-that is to say, it is necessary for a given spectral line that the absorbed energy be completely converted into heat or radiation of the same wave length, and conversely the energy emitted in the particular spectral line must represent only the heat energy of the molecule

,

20

Ann. P h y s i k , 44, 749 (1914)

'

October, 1928

ISDC7STRIdL ;ZAYDE S G I S E E R I S G CHE-IIISTRY

or the absorbed energy a t the same wave length. It is necessary, therefore, in measurements by the line-reversal method t o deal with spectral lines for which in emission as well as in absorption reversed energy transitions are possible-that is, resonance lines, such as the sodium D-lines, the red line of lithium, or the green line of thallium. The conception of a resonance line implies (1) that its final orbit is the ground orbit of the atom, and ( 2 ) that its initial orbit is the (energetically) next highest orbit from which the return t o the ground orbit, and only to this, is possible, being accompanied by the emission of monochromatic light. The heat energy from the combustion processes in the flame containing an alkali-metal vapor will lift the electron from its ground orbit, 1 S , into the “energetically” next highest orbit, 2 P. According to the principle of selection the atom thus excited can undergo the transition 2 P +1 S as the only way in which the excited atom can revert to its unexcited state, thus giving out monochromatic light of the same wave length as it absorbed. It is clear from this that only resonance lines may be used, for Kirchhoff’s law would not apply in cases where intermediate quantum states are possible; and in that case the true flame temperature would not be equal to the brightness temperature of the comparison radiator. In measuring the temperature of luminous gases it is also necessary to take into account whether the radiation intensity observed is due entirely to the kinetic energy of the molecules or in part t o chemical or electrical processes of some kind, for in the case of chemiluminescence the laws of radiation cannot apply. For example, the radiation equation of Planck is derived esplicitly from the assumption that the vibrating resonators obtain their energy from the heat motion of the molecules, and 9 io 11 12 only in that cast: can kT be MUIHANE. PER CENT substitutsd for the average Figure 3-Methane energy U of a resonator. In cases of chemiluminescence we should have to put in place of kT a functionf(i) of the intensity i of the process excited by luminescence, and T would therefore disappear from the radiation equation.*l KohnZ0has definitely shown bv measurements of emissive and absorptive powers within the range 900’ to 1800 ’ C. that Kirchhoff’s law quantitatively applies to flames colored with alkali-metal vapors, and therefore they are pure temperature radiators and chemiluminescerice effects are entirely absent. In general, th‘e condition for pure temperature radiation will be more nearly approached the higher the temperature of the flame under the same conditions otherwise,” and this fact makes possible the measurement of flames with very high temperatures such as the acetylene-osygen flame, by the line-reversal method. (31 It was proved by Henning and Tingwaldt’g by experimental measurement that the reflective power of such colored flames is zero. ITe should expect this result from the work of R. IT.iVood, who has shown that a very high concentration of mercury vapor is necessary for metallic reflection, whereas in these flames we are dealing with very low concentrations of metal \-apors.

I t is e\-ideiit that the method of line reversal has much to commend it. from the theoretical standpoint. It gives values indicating the ability of the flame to impart heat to the sodium atom and, owing to t’heintinlate mixture of the metal atoms with the rectct’ing gases of the flame, temperature equilibrium should be established rapidly and the values determined by this method should approximvte the true temperahre of the flame. From the experimental standpoint the method is ideal because of the simplicity of the experimental technic and the rapidity with which the measurements may be made. Experimental M e t h o d

The burner j (Figure 1) consisted of a small brass box containing t’hree rows of five quartz tubes in a row, each tube t2

Pringsheim, Physik. Z.,14, 129 (1913). Franck and Jordan, “.4nregung yon Quantenspriingvn,” p. 192.

1007

5 mm. in diameter, except for the measurements with carbon monoxide when the diameter of each was 4 mm. The tubes were completely encased m-ith asbestos paper to prevent the introduction of secondary air from below. The sodium chloride solution (just under saturation) was sprayed into the flame by means of a small glass nozzle, k , of the type used by Barnes13 and by Bennett.I4 The gas stream entered the mixing chamber 1 directly a t the source of the air stream, insuring complete mixing. The gases were introduced either from ”“ cylinders containing the gas under pressure or from a large gasometer ISW and pumped into the mixing chamber by means of a small 18M gas-tight pump a t rates that could be determined on a sensitive ori- g fice flowmeter. The air was sup- $ 1 plied from a larger pump a t rates read on a second orifice flowmeter. ,lso Both gas and air were passed through several large bottles to smooth small pressure fluctua- 1-40 tions. Samples of the gas-air mixtures were drawn off for analysis l ~ w 4 6 just before entry into the burner. PROPANE PER CE\T These analyses were made on a FigureI-propane Bone-Wheeler apparatus.23 The brightness temperature of the tungsten band lamp, after the adjustment to line reversal as described above, was read on an optical pyrometer of the disappearing filament type. It n-as calibrated against a pyrometer of the rotatingdisk type which had been calibrated by the Bureau of Standards. Readings were made through a red (X = 0 6 6 5 ~ ) filter glass as usual, and since the brightness comparison in tlie spectrometer was made in the yellow (X = 0.589pj, a correction is necessary. By employing the relation between true temperature T and brightness temperature SA.as derived from JVien’q equation, me have

6

1

1

X 2 303 log ex

T

SA

Ca

1

1

_ - - =

atid

T--S,=

A’. 2.303 log ex’ C2

wliereJX refers to red light (0.665~)and A’ to yellow light (X=O.589pj, ex the spectral emissivity, and c2 = 1.433 an. deg. Subtracting we have

vliere SA is the brightMw -, ness t e m p e r a t u r e obt++i, i 1 -t__ I 1 served with the pyrometer and SA. is the temperature of the flame. F r o m t h e d a t a of 1 Forsythe and Worthing24 E for spectral einiqsivities 1880 of tungsten as a function 32 34 36 38 40 CARBOY MONOXIDE, PER CEhT of t h e t r u e a n d t h e , brightness temperature, Figure 5-Carbon Monoxide transposition curves were plotted for the temperature range covering these experiments, and this additive temperature (from 25” to 32“ C.. depending on the temperature) applied as a correction to the readings. The absorption of lens b was determined by focusing the pyrometer on the image of the band lamp and then on the ~

23 24

Grice a n d P a y m a n , Fuel, 3, 236 (1924). /1sllophrr. J., 61, 146 (1923).

INDUSTRIAL AND ENGINEERING CHEMISTRY

1008

Vol. 20, No. 10

band lamp directly, with the lens removed. This absorption amounted to 29" C., and is a negative correction to the flame temperature, just about canceling the color correction described above. The cooling effect of the water in the salt-solution spray was found experimentally by inserting a platinum strip in the non-luminous flame and reading its temperature on the pyrometer, with and without pure water being sprayed at the same rate as used in the measurements with the flames. The amount of cooling was 15" C. for all the flames except the carbon monoxide flame, where it was 5" C. because less air was used. The measurements of Henning and Tingwaldtlg showed that the salt in the spray had no cooling effect.

a function of heating current when using a bright platinum strip gave 1770"C. when corrected for emissivity. The agreement to within 20" C. between the two methods is regarded as satisfactory when it is remembered that they are so different in character, the one depending upon a comparison of brightness temperatures of a tungsten radiator and of sodium vapor radiating due to the temperature that it assumes from the gases of the flame, the other depending upon the heat gained by a solid radiator, corrected for radiation loss, from the flame gases. The method of line reversal is much more convenient and rapid, however, and from an experimental standpoint would be chosen in preference to all others.

Gases

Comparison with Calculated Temperatures

The methane was obtained from a natural-gas well and analyzed 97 per cent CHI; the propane was of commercial grade. Both gases were stored in cylinders under pressure. The Pittsburgh natural gas (85 per cent methane, 14 per cent ethane, 1 per cent nitrogen) was taken from the city mains. The carbon monoxide was prepared from formic acid, dehydrated by phosphoric acid, and purified with potassium hydroxide solution together with an absorbing tube of Cardoxide and charcoal; its purity was better than 99 per cent by analysis.

It is interesting to compare the measured values of temperature with the calculated temperatures. The calculations were based upon the A H of the reaction and the specific heats of the gases, both from the data of Lewis and Randall, and taking into account the degrees of dissociation of water and carbon dioxide. Calculated maximum temperatures are as follows: Pittsburgh natural gas, 2000" C.; methane, 2000" C.; propane, 2050"C.; and carbon monoxide, 2230" C.; the measured values, 1875" C., 1875" C., 1930" C., and 1960" C., respectively. The difference is due to the radiation from the particular flame under investigation, and to possible inaccuracy of the specific heat data. It is to be noted that the maximum temperatures as measured correspond to gas-air mixtures somewhat higher than the theoretical mixtures for complete combustion. This is perhaps due to the fact that some air is drawn into the flame from the surrounding atmosphere. Further measurements are under way for many other combustible gases, mixed with air and with oxygen. It is planned to measure the temperatures of these flames with various designs of burners, and supplement these measurements of temperatures with measurements of total radiation.

Results

The results of the measurements by the line-reversal method are shown in Figures 2, 3, 4, and 5. The corrected temperature is plotted as a function of per cent gas in the gas-air mixture, except in the case of Pittsburgh natural gas where per cent total hydrocarbons is plotted. Comparison with Solid Radiator in Flame

It was thought advisable to compare the results by the line-reversal method with measurements by another method that should differ from it as much as possible. The method as proposed by the National Physical Laboratory as outlined in the introduction, was chosen, using a natural gas-air mixture of 10.83 per cent total hydrocarbons, which gives a temperature of 1750" C. according to the method of line reversal. The intersection of the curves of temperature as

Acknowledgment

The writers' thanks are due to J. E. Crawshaw, explosives engineer, for his assistance in making temperature calculations, and to J. S. Brown, junior explosives chemist, who assisted in the experimental measurements.

Radiant Energy from Flames W. E. Garner DEPARTMENT OB PHYSICAL CHEMISTRY, BRISTOL UIIVERSITY. ENGLAND

The thermal and the chemiluminescence theories of the radiant energy from flame are discussed and the conclusion is reached that the emission is very largely chemiluminescence. New experimental evidence on the radiation from the carbon monoxide flame is in agreement with this conclusion. The study of the radiant energy from flames offers a line of approach to the problems of catalysis of the processes of combustion, and this is illustrated by reference to experimental work on the catalysis of the carbon monoxide flame by hydrogen. I t is concluded that the action of hydrogen is twofold in character. It acts as a catalyst in the chemical sense when the hydrogen percentage exceeds 0.02, and as a conserver of chemical energy within the flame throughout the whole range of concentrations up to 2 per cent. The latter type of catalysis is termed "energo-thermic," and in the above example it is con-

cluded that either the proton or the electron is the effective agent. The chemical energy is conserved within the flame by collisions between protons or electrons and the newly formed products of the combustion process.

HE study of the radiation from flames is in the embryonic stage of development. Ittshows promise, however, of becoming a highly specialized branch of knowledge, which will play an important part in the elucidation of the mechanism of the processes of combustion occurring in flame. The advance of modern physics has made us aware of the almost infinite variety of "unit mechanisms" which can occur during the interaction between molecules and between molecules and radiation. These mechanisms have been classified and certain laws concerning them have been made known. There exists, therefore, a much broader, and a t the same

T