]\gt ; Prof. B. Hopkinson .
[ Jan. 19 , heated gas bears to that of the gas .
Burstall , who has measured the temperature in a gas-engine cylinder by means of platinum wires , did not fully investigate this point , and his results are , in consequence , open to doubt .
* Befo.re proceeding to a detailed account of the apparatus and records obtained it will be convenient to state shortly the principal conclusions reached .
The experiments were all made on mixtures of air and Cambridge coal-gae having an average " " higher\ldquo ; calorific value of 680 British Thermal Units per cubic foot at C. and 760 mm. The composition of the gas is given in an appendix .
The mixture was fired at atmospheric pressure in a vessel of dumpy cylindrical form and of a capacity of cubic feet , which is shown in section in fig. 1 .
The combustion was started by an electric spark at the centre of the vessel .
Explosion of Rich ture .
With nine volumes of air to one of gas the maximum pressure varied from 76 to 82 lbs. per square inch above atmosphere ; and was reached about 1/ 4 of a second after firing ; on fig. 2 is shown a facsimile of a pressure diagram ( curve A ) .
It was found with a mixture of this composition ( air/ gas ) that : ( 1 ) The flame spreads from the spark with a velocity which varies somewhat in an apparently accidental manner , but which is roughly 150 cm .
per second .
The spread of the flame is of a rather irregular character , and differs htly in different directions ; Mallard and Le Chatelier found for a mixture containing 17 per cent. coal-gas , a velocity of flame propagation of 125 cm .
per second along a tube of 2 or 3 cm .
diameter .
( 2 ) The flame reaches the walls when the pressure is of the order of 15 to .
per square inch above atmosphere .
At this point , however , only a small portion of the walls is in contact with the flame , namely , that nearest the spark , and most of the gas is still unignited .
As the flame spreads a greater and greater area of the walls comes into contact .
' Proc. Inst. Mech. Engrs 1901 .
The only attempt to apply the methods of platinum thermometry to gaseous explosions of which I am aware , is this one by Burstall .
He could not use very fine wire because it melted , and it seems probable that on this account temperatures are a good deal wrong , even in the latter part of the explosion , and his results give no information as to the initial stages of the burning and throw no light upon the question of the velocity of combination .
He used a rotatiug contact maker , which made contact at definite epochs in the cycle , but did not give a continuous record of temperature in any one explosion .
" " Recherches sur la combustion des Melanges Gazeux Explosifs Bxplosions of Coal-Gas and Air .
with it ; until the flame completely fills the vessel and is losing heat to every part of it .
At this point the pressure is still a little short of the maximum , being about .
per square inch when the maximum pressure ched .
is 82 lbs. Maximum pressure is atlained in less than 1/ 30 of a seoond after the flame completely fills the vesseL ( 3 ) At the centre of the vessel the temperature of the gas rises very rapidly , after ignition , to about 1225o C. It reaches that figure , within in a time which is certainly less than 1/ 20 of a second , and probably less than 1/ 50 of a second ; in other the combustion is complete within about 4 per cent. in that time .
The temperature remains nearly steady .
during the earlier part of the spread of the flame , the pressure during this time remaining very nearly tant .
The combustion at the centre takes place very nearly at constant pressure , and is complete within 5 per cent. before the pressure has risen more than a couple of pounds above atmosphere .
From this result , if it be assumed that no heat is lost until the flame reaches the walls of the vessel , it follows that the capacity for heat at constant pressure , reckoned between and 1200o C. of the products of the explosion , is about times that of the same lume of air .
.There is no doubt that the flame radiates some of the heat of combustion , but it is improbable that the loss from this cause exceeds 15 per cent. If that percentage be assumed then the capacity for heat is times that of air .
( 4 ) In the adiabatic compression of the gas in the centre , which takes place in the later stages of the explosion , the temperature rises to a point which is considerably above the melting point of platinum \mdash ; probably about 1900o C. If the ultimate temperature , corresponding to a compression of atmospheres absolute , is C. , then the average value of ( ratio of specific heats ) for these gases is between 1200o and 1900o C. ( 5 ) A platinum thermometer , placed about 1 cm .
within the walls at the furthest point from the spark , reaches its maximum temperature , which varies accidentally , but is between 110 and 130 C. , within 1/ 30 of a second after the attainment of maximum pressure .
There is here but little adiabatic compression or rise of temperature after ignition .
The gas at this point has been compressed to about five atmospheres before ignition , and the temperature due to this compression is about 20 C. Experiments by Petavel , which coal-gas and air were exploded after compression to about 77 atmospheres , show that the rise of tempeTature on explosion is nearly independent of the pressure .
If herefore there were no loss of heat , the temperature at this point would rise to about 1400o C. At the same instant the temperature at the centre is 190 C. Consequently , even if the 'Phil .
Mag May , 1902 .
Prof. B. Hopkinson .
[ Jan. 19 , explosion were to take place in a vessel impervious to heat , there would be a difference of 50 C. between the maximum and minimum temperatures in the vessel .
Similar differences must exist whenever a mixture is fired at discrete points ; the temperature at the firing points will always be raised much above the mean by adiabatic compression after ignition .
( 6 ) It seems certain from these experiments that in a mixture of this strength combustion is for practical purposes everywhere complete at the time of maximum pressure , and that subsequent to that time the gas is a mixture of , and inert gases in chemical equilibrium .
( 7 ) At the moment of maximum pressure the distribution of temperature is roughly as follows:\mdash ; Mean ( inferred from pressure ) . . . .
1600 Centre near spark 1900 ( b ) 10 cm .
within the wall . .
1700 ( c ) 1 cm .
from wall at end 1100to1300 ( d ) 1 cm .
within the wall at side 850 At points , and , the gases can have lost but little heat at this time , and the ences of temperature are almost wholly due to the different treatment of the gas at different places .
At ( a ) it has been burnt nearly at atmospheric pressure , and compressed after burning to about six and -half atmospheres absolute , while at ( c ) it has been first compressed to about six atmospheres as in a gas-engine , and then ignited without any subsequent compression .
At the point ( d ) much heat has been lost , since this is the first point on the wall reached by the flame ; the gas here is ignited when the pressure is about two atmospheres , its temperature rises instantly to 1300o , and at once begins to fall .
( 8 ) Up to the time of maximum pressure convection currents in the gas have had but little effect upon the distribution of temperature , which is mainly determined by the treatment accorded to the gas at rent points .
After combustion is complete , however , the motion of the gas set up by the explosion or by convection currents rapidly obliterates these initial differences .
Half a second after maximum pressure the distribution of temperature is as follows : mean temperature of the whole of the gas ( calculated from the pressure ) is about 110 C. The mean temperature , exclusive of a layer 1 cm .
thick in contact with the walls ( shown by the resistance of a long wire stretched from the centre to , fig. 1 ) , is about 1160o C. The temperature of the hot core is fairly uniform , though it varies in an accidental manner about the mean value ; thus , at centre , in different explosions , I have found temperatures ranging from 1100o to Bxplosions of Coal-Gas and Air .
1200o C. at this time .
The mass of gas during cooling may therefore be described as a hot core in which the temperature is approximately uniform ( though it varies accidentally as the result of currents ) surrounded by a thin layer wherein the temperature falls to the temperature of the walls .
I find , by calculation , that if such layer were 3 cm .
thick , and if the fall of temperature were uniform , the mean temperature inferred from the pressure would fall short of that of the hot core by about the observed amount , viz. , C. Explosion of Weak Mixtures .
The explosion of a weak mixture containing 12 volumes of air to 1 of gas differs markedly from that of a 9 to 1 mixture .
In the latter case there is no time for the buoyancy of the light burnt gas to materially affect the ationo of the flame , though it doubtless causes the apparent velocity of propagation to be slightly greater in an upward direction than downward .
Convection currents have no material influence on the phenomena until after the attainment of maximum pressure .
But in the weaker mixture their effect is important from the outset .
The small portion of the gas first ignited instantly rises with increasing velocity .
At the same time it grows , by the ignition of that surrounding it , but at a rate which soon becomes considerably less than the velocity with which it is rising .
In spite of the very slow propagation of flame from point to point , however , the combination of the gases , once initiated , is rapid .
Thus the temperature of a wire placed close to the spark rises within second to about 1000o C. , and then remains nearly stationary for some time .
A few centimetres below the spark the temperature will rise rapidly and then fall ; the flame reaches the wire , and is then carried upward and away from it , the wire being cooled by the current of cold , which follows in the wake of the ascending flame .
About 1 second after ignition , and while the pressure is still less than .
above atmosphere , the upper ]lalf of the vessel is filled with burnt gas which is in contact with , and losing heat to , the upper half of the walls .
In the lower parts of the vessel the gas is still unburnt .
The last portions of gas to be ignited are those immediately under the spark , and from 10 to 20 cm .
away from it .
A wire placed at this point shows a gradual rise of temperature , due to the adiabatic compression , followed by a sudden rise , due to ignition , slightly before the time of maximum pressure .
In general , a platinum wire placed anywhere within the vessel shows , at some time before maximum pressure , a sharp rise in temperature lasting about 1/ 10 of a second , after which the temperature is steady for a time , and then falls slowly .
There are fluctuations of temperature both up and down , but these are plainly Prof. B. Hopkinson .
[ Jan. 19 , accidental effects of convection currents , and are due neither to the ignition of portions of unburnt gas nor to the slow combination of gas already ignited .
My experiments seem to show that in the weakest inflammable mixtures and in strong mixtures alike , the combustion once initiated at any point , is completed almost instantaneously .
Moreover , the complete inflammation of the gas is , even in the weakest mixture , nearly simultaneous with the attainment of maximum pressure .
One-tenth of a second before maximum pressure there are undoubtedly places to which the flame has not spread , and it is extremely improbable that there are any such places 1/ 10 of a second after .
It is safe to assume in dealing with a 12/ 1 mixture that 1/ 5 of a second after maximum pressure ( when the loss of pressure by cooling is still less than 5 per cent. there is present in the cylinder a mass of , and inert gas in complete chemical equilibrium .
In the 9/ 1 mixture this state is , of course , attained much sooner .
The difference in the behaviour of the weak and strong mixtures is wholly due to the very slow propagation of flame in the former ; in a 9/ 1 mixture the flame seems to travel about 10 times as fast as in the 12/ 1 mixture .
In the 12/ 1 mixture maximum pressure is about .
above atmosphere , and is reached about seconds after the passage of the spark .
nearly half that time , at least half the area of che vessel has been in contact with the flame .
It seems probable , therefore , that the proportionate loss of heat before the attainment of maximum pressure in a weak mixture is considerably greater than in a strong one .
In other words , if the explosion were adiabatic the ultimate pressure produced would exceed the maximum actually observed by a gl.eater proportion in the case of the weak than in that of the strong mixture , and this mainly by reason of the greater loss of heat .
Description of Fig. 1 shows a section of the vessel along its axis , with its principal dimensions .
A is the sparking point\mdash ; nearly at the centre of the vessel .
, and are three platinum thermometers ; , in the case shown , is close to the spark , at a distance of about 30 cm .
, and about 1 cm .
from the walls of the vessel .
Each thermometer consists of a coil of about 5 cm .
of pure platinum wire , inch diameter .
The coil is hard-soldered to two thicker platinum wires sealed into the ends of glass tubes , and the thicker wires are soldered to stout copper leads .
The copper-platinum junctions , being just within the tubes , are protected from the flame .
Owing to this protection , and to the rapidity of the changes of temperature to be measured , 1906 .
] Explosions of Air : spread of the flame through the vessel .
The minimum ordinate and maximum temperature at are reached at the moment of maximum pressure or very shortly afterwards .
There is no pause in the rise of temperature followed by a more rapid rise , such as was observed in the case of B. The reason of this is , of course , that there is here no appreciable rise of pressure after combustion .
The way of correcting for the inertia of the galvanometer is to smooth out the oscillations by taking the means of successive maxima and minima , This method of correction is satisfactory if the current does not vary very much in the course of an oscillation .
Applying it to the curve we the table:\mdash ; The time is reckoned from the point marking the com1neucement combustion to a point half-way between the maximum and following minimum ( or vice versed ) whose mean is shown in the second column .
The third column is the total i'esistance of the circuit , obtained by dividinero the second into number determined by substituting known resistances for the thermometer coil .
The fourth gives the rise of resistance obtained by racting from the third column the of the circuit other than the thermometer coil ( ohms ) and the resistance of the thermometel coil at C. .
The last column ives the temperature of the wire calculated in the usual way from the resistance .
The temperature coefficient was determiued to be between and C. : the factor was iven as by the Scientific Instrnment Company who supplied the platinum wire .
naximum temperature shown on curve is about 1250o C. , and occurs very nearly at the moment of maximum pressure .
This estimate of .
B. Hopkinson .
[ Jan. 19 temperature , however , may be a good deal wrong , because the diagram , being on a smaller scale , cannot be measured so accurately as B. The velocity of propagation of the flame was found by taking records in which thermometers and were used .
In three different explosions the times taken to travel this distance ( about 30 cm .
) were respectively and second .
In another explosion a pyrometer was placed just inside the wall at the point .
It was found that though this point is if anything rather neal'er to the spark than , the flame reached it of a second later , a further illustration of the irregular manner in which the flame is propagated .
A large number of diagrams were taken with a weak mixture in which the proportion of air to gas was 12/ 1 .
The thermometers were placed in all sorts of positions .
The general character of the results is indicated above .
One diagram is reproduced in fig. 4 .
In this case one wire only was used , and it was placed about 15 cm .
from the spark and vertically below it .
It will be observed that at first the temperature rises very slowly .
Mors than two seconds afoer ignition ( at A ) it is only about 21 C. , and such heating as has then occurred is almost entirely due to adiabatic compression.* The flame now reaches it ( A ) and the temperature rises in 1/ 10 of a second to 1300o C. The pressure has now attained its maximum value above atmosphere ) and about seconds have elapsed since the spark passed .
The temperature remains steady for a while and then falls ; but there is ) perceptible rise after the pressure begins to fall .
In a large number of trials I as unable to discover any point at which inflammation occurred later than in the diagram here shown , so that it is probable that all the gas is nited at the time of maximum pressure .
Moreover , this must be nearly the last portion of the gas to be ignited , and here , if anywhere , the effects of slow combustion of gas already nited would be perceptible in a rise of temperature after maximum pressure .
In other words the gas is in chemical equilibrium a very short time after maximum pressul'e .
The Specific Eeat at Constant Pressure and the of Reaction .
Returning now to the record shown on fig. 2 with a 9 to 1 mixture , the temperature of the wire at the centre of the yessel is plotted in terms of the * Owing to the heating effect of the current the wire starts about hotter than the gas .
The compression of the gas to 50 lbs. above atmosphere will cause its temperature to rise C. The wire will then be hotter than the gas , but the difference.will be somewhat less than , since the heat to be dissipated is less because of the increased resistance .
At temperatures of and over , resistance is so great that the current in the wire can have but little effect upon its temperature .
Prof. B. Hopkinson .
before explosion , both gas and air are saturated with moisture at C. and .
per square inch , the products of the combustion will be as follows:\mdash ; assuming that the and occupy their proper molecular volumes and the voIumes being reckoned at C. The heat produced by the combustion when the gas is burned in a calorimeter is approximately 620 B. Th. U. , the products being cooled to C. Of this , however , about 60 B. Th. U. are to be ascribed to the condensation of the steam produced in the explosion .
If the cooling were stopped at C. ( when condensation begins ) only 550 B. Th. U. would be obtained , about 10 heat units due to the cooling of the gases other than steam from to .
Moreover in the as it actually takes place a certain portion of the heat is radiated from the surface of the flame.* The amount In the paper as originally written there was a note dealing with this point , in which I gave my reasons for supposing that the heat radiated could be neglected , and the specific heat was calculated on that assumption .
Some remarks made by Professor Callendar at the discussion of the paper , however , have led me to look into the matter further , with the result that I must now admit the probability of a loss by radiation during the early stages of the explosion of 15 per cent. of the total heat then being generated .
Professor Callendar stated that he had observed that an ordinary Bunsen flame radiates from 15 per cent. to 20 per cent. of the total heat of the gas used , which is a much greater proportion than I had thought possible ; it is reasonable to suppose that the proportion of heat radiated from the flame at the centre of my vessel will at least be of the same order , though the flame is somewhat colder than the Bunsen flame .
Among the reasons that I gave for neglecting the radiation was the fact that no rise of temperature , other than that due to the adiabatic compression of the gas surrounding them , was observed in any of the wires in the outer part of the vessel ( e.g. , that at , fig. 1 ) , until they were actually in the flame .
On working out the actual figures , however , I find that radiation from the flame could not certainly be detected by such rise of temperature unless it amounted to 15 per cent. of the heat generated , in which case it would just be apparent .
An estimate of the specific heat , unaffected by radiation errors , can be obtained from the explosion of a weak mixture , such as that the diagram of fig. 4 .
In this case the burning of the last portion of the gas to be ignited , which is that round about the thermometer coil , takes place in an inward direction , so that there can be but little radiation from the flame surface .
The gas just before ignition has been compressed to .
per square inch above atmosphere , and its temperature is very nearly that due to such compression , viz. , C. On combustion the temperature rises to C. The products of combustion of 1 cubic foot of gas in this case consist of about cubic feet of steam and and 10.8 cubic feet of and O. The internal energy is the heat of of -Gas and Air .
this loss is quite uncertain ; but it is improbable that it exceeds per nnt If we assume that figure , there remain 470 B. Th. U. as the heat evolved in cooling the products at constant pressure from 1230o to C. The same volume of air cooled through the same range would evolve 370 B. Th. U. Thus the average volume ecffic heat of the products is times that of air .
At constant volume the ratio would be , taking for the products as .
If , on the other hand , we assume that the radiation can be neglected , the ratio of the specific heats of air and products is 550/ 370 or about .
I think it is fairly certain that the true value is even and , and probable that it is nearthe lower figure .
This , of course , is the average value ; at the upper temperature the ratio will be a good deal greater .
At C. the specific heat of the products is about times that of air .
Of the heat evolvedin cooling , 290 B. Th. U. are accounted for by the nitrogen and oxygen present .
The balance of 180 B. Th. U. ( assuming the 15 per cent. loss ) is the heat capacity of the cubic feet of and .
The average volume specific heat of these gases is accordingly about 24 times that of air over the range to 1230o C. It is obvious that the uncertainties in the amount of radiation , the composition of the gas , etc. , are such that this can only be regarded as a very rough approximation to the truth .
It is not difficult to measure the original correct to 2/ 10 mm. and the ordinates in the second column are probably within that amount of the truth , and certainly within 4 .
The correspondin .
errors in the temperature , at the highest temperature measured , are about and C. respectively .
The order of accuracy of the results is further shown in the following table , which gives the temperature found at corresponding times in six different explosions .
composition of the mixture is nominally the same in all except the second , in which the preliminary steaming was omitted , so that the gases were very much drier than in the other cases .
combustion ( 550 B. Th. U. ) plus the heat required to raise the mixture to C. before firing ( 63 B. Th. ) ; total , about 610 B. Th. U. evolved in cooling from to C. Of this the and account for 427 B. Th. U. , leaving 183 for the cubic feet of and .
The capacity of heat for the latter is accordingly times that of the same volume of air .
If we take this figure and calculate from it the heat required to warm the products of the explosion of 1 cubic foot of gas and 9 of air from to C. , the result is 464 B. Th. U. , whereas the total heat evolved in that explosion is 650 B. Th. U. The discrepancy may be accounted for by supposing a radiation loss of about one-sixth part ; and if such loss be umed , the specific heat of the products at constant pressure works out to about times that of air .
The as within the flame surface is of course unaffected by radiation , and its treatment must be sensibly adiabatic until convection currents come into In criticising this table it is to be remembered that the calorific value and composition of the gas were not determined from day to day , and the calorific value may have varied between limits differing by 5 per cent. Moreover , the amount of water vapour present must also have varied somewhat .
The combustion of 100 volumes of gas in 900 of air produces volumes of steam , in addition to this there might be present ( at C. 760 ) anything up to 23 volumes of moisture before explosion , according to the dryness of the gases .
The high temperatures shown on September 8 may be explained by the fact that there the gases were nearly dry , whereas in the other cases they were nearly saturated .
This result also serves to show the considerable influence of the amount of moisture present\mdash ; due probably to the high specific heat of steam at such temperatures .
In deulating the mean temperatures in the last column , the results of September 8 are left out of account .
The temperatures are those of the wire , uncorrected for radiation or time-lag .
The Batio of Specific Heats .
In the adiabatic compression of any substance we have .
( 1 ) We have here to deal with a mixture of gases , 79 per cent. of which is perfect gas .
The remainder , and , is not perfect , but its deviation from the gas law const .
is dependent on dissociation , the effect of which is limited in amount to a reduction of the absolute density to twothirds of its value at ordinary temperatures .
If dissociation were complete the density of the mixture would become 10 per cent. less than at ordinary temperatures .
In other words the characteristic equation is , ( 2 ) where is a usly increasing quantity , which at very high temperatures Explosions of Coal-Gas and Air : attain a value 10 per cent. in excess of its value at C. At 1200o C. it is improbable that dissociation is complete , or that has nearly reached the limit of its increase .
Now differentiate ( 2 ) and we may now write ( 1 ) in the form , where , ( 3 ) and is a quantity which changes slowly with and R. The integral of ( 3 ) is approximately .
Now has been shown by the experiments to be more than 25 per cent. greater at 1200o C. than at C. It is unlikely that the term increases in anything like .
so large a proportion , since cannot increase by more than 10 per cent. We may , therefore , confidently expect a substantial diminution in as the temperature and pressure rise .
Now for a perfect gas , where is the ratio of the specific heats .
and this relation is also nearly for the mixture in question .
At C. the value of for this mixture is not accurately known , but it is probably not greater than , making something less than .
It may be expected , therefore , that at 1200o C. is less than , and less than An inferior limit to at 1200o C. can be deduced from the experiments as follows : In the explosion described above , the wire melted at a pressure of atmospheres .
The temperature of the gas must then have been considerably above the melting point of platinum ; in all probability it was 1900o C. * At a pressure of one atmosphere the temperature of the gas was 12:30o C. Calculating from the equation it follows that the average value of between 1200o C. and 1900o C. , is .
Since is a diminishing quantity , its value at 1200o must be greater than the average value between these limits , viz. , greater than We may , therefore , assert with a high degree of probability that the value of for this mixture is between and at 1200o and that the ratio of the specific heats at that temperature is between and .
For the of correcting the temperatures in the neighbourhood of 1200o C. , the value was assumed for , and for a small correction this is no doubt near enough to the truth .
The Measurement of Temperature .
The temperature of the gas differs from that calculated from the resistance of the wire from four causes , the eflects of which must be considered:\mdash ; * See the section " " Measurement of Temperature below .
VOL. LXXVII.\mdash ; A. 2 Prof B. Hopkinson .
Jan. 19 ( 1 ) The law connecting the resistauce of platinum and its temperature has not been experimentally verified beyond 1000o C. Up to that however , it has been found that the temperatures given by the platinum and gas ermometers agree within about C. I found that a sample ofplatinum wire whose purity is indicated by its temperature melted when its temperature as shown by its resistance was 1670o .
The ] temperature must have been slightly under 1710o .
It is probable that extrapolation up to 1500o C. will be within the general order of accuracy of these experiments .
( 2 ) The ends of the wire are colder than the middle because of conduction As the diameter of the wire is only 1/ 2000 of its length , it may be expected that this correction will not be important .
Suppose that the temperature fallt at the end of the wire by 1000o in 1 mm. , and that the heat conductivity ol the platinum is the same as when cold , , then the amount ol heat conducted away per second by the wire diameter will be calorie per second .
Now the heat supplied to the by the gas per millimetre length is shown below to be about calorie second for every 10 by which the gas is hotter than the wire .
A comparison of the two nitudes shows at once that at a distance of 1 mm. from the end the temperature of the wire must have become sensibly uniform .
This inference has been verified by comparison of the temperatures shown by wires of very different length ( ranging from 3 to 10 cm .
) under the same circumstances .
( 3 ) The temperature of the wire lags behind that of the gas , when the latter is changing rapidly .
To test the effect of this , wires of two different and 2/ 1000 of an inch respectively\mdash ; have been placed as close together as practicable , and their temperatures compared in the same explosion .
* The result is shown on fig. 5 , and was confirmed in several experiments .
Both wires were within 5 cm .
of the LQparks , so that the rise of temperature in the thin wire is similar to that shown on fig. 1 .
It melted shortly before the attainment of maximum pressure .
At the point A the temperature of the thick wire is rising at the rate of 5000o per second .
At the same moment the temperature of the thin wire is rising about 1300o per second .
Since the thick wire has four times the mass of the thin , it is receiving heat about 15 times as fast .
Now , rate at which a very fine wire receives heat from a gas , by conduction , is almost independent of its diameter and , if this be true of our two wires , the difference of A somewhat similar method has been posed by Professor Calendarfor determining the radiation correction , ' Proc. Inst. Mech. Eng 1901 .
Prof. B. Hopkinson .
here described shows that if the temperature is changing at the rate of per second , a wire diameter will be at least 20 hotter or colder than the gas .
In a gas-engine at 120 revolutions per minute the mean temperature falls from 1600o to 1000o C. in about second , so that of this diameter untrustworthy measurement of temperature , except , perhaps , at the extreme end of the stroke .
* ( 4 ) One other source of error remains to be considered , namely , that due to radiation .
In consequence of this the wire must always be somewhat colder than the , since it has to receive heat at the same rate as it loses heat by radiation .
The correction necessary for this may be deduced from the comparison of thin and thick wires described in the Iast paragraph .
At the point A the thick wire is getting hotter at the rate of 5000o C. per second .
Its diameter is , hence assuming that the specific heat of platinum at 1000o C. is , the capacity heat of the wire is water unit per centimetre .
It is therefore heat at the rate of calorie per second , plus any loss due to radiation .
It will presently appear the latter item is small compared with the first .
The gas at this point has a temperature of 1250o in round numbers .
It follows that with a temperature difference of 25 C. between gas and wire , the latter will receive heat at the rate of calorie per second per cenGimetre .
Now the heat radiated by a platinum wire at C. has been found by Bottomley to be about calorie per second per square centimetre of surface .
Petavel has extended Bottomley 's investigation by comparing the radiation at various temperatures up to the results of the two researches it appears that a wire at C. will radiate calories per second per square centi1netre .
The wire diameter therefore radiates calorie per second per centimetre of length .
To supply this amount of heat the gas must be hotter than the wire .
This is the colTection applied in in addition to that for The conductivity of the gas increases with the temperature .
to the kinetic theory in its simplest form , the conductivity should be proportional to the square root of the absolute temperature ; but it is found , in fact , to vary more rapidly than this , though always less than in proportion to the temperature .
So far as I am aware , there are no experimental data for temperatures above C. , and I assume that the conductivity then varies as the square root *The smallest wire used by Professor Burstall was diameter .
The error due to time-lag would be that of a wire diameter , but is still of the order of C. Bottomley , 'Phil .
Mag vol. 49 .
Petavel , ' , vol. Prof. B. Hopkinson .
[ Jan. 19 of coal-gas or hydrogen , and air , is less than two-thirds of that which would attained by an equal volume of air on the fiddition of the heat of combustion .
A similar phenomenon appears in gas-engine indicator diagrams .
In most such diagrams the expansion line is somewhat above the adiabatic expansion curve for air , though it is known that much heat is being lost during expansion .
But though the fact is well established , there is still much controversy about the cause .
Confine the attention for the moment to closed-vessel explosions , such as those described in this paper , in which the is strictly homogeneous before gnition .
At the time of maximum pressure the rate of loss of heat to the walls is just equal to the rate at which the chemical energy of the gas is being converted into thermal energy .
This statement requires some qualification in the case of strong mixtures ( a point which is dealt with below ) , but is probably very nearly true of weak mixtures , which possess most interest from this point of view .
It follows that at the moment of maximum pressure the gas has not quite attained chemical equilibrium .
At some time after maximum pressure equilibrium is reached for practical purposes , and the gas then consists of a mixture of , and inert gas , whose internal energy is all thermal and is equal to heat of combustion less the heat lost up to that time .
Messrs. Mallard and Le Chatelier and others assumed that this state was reached within a very short time of maximum pressure ; and they utilised their records of explosion pressures to deduce the specific heat of the products of combustion .
From a study of the curves of pressure during cooling they calculated the loss of heat during burning , and so determined the internal energy of the burnt mixture at or about the time of maximum pressure .
This was assumed to be all thermal energy , and so the capacity for heat was determined .
The calculation of the loss of heat is open to serious criticism , since the state of the gas cooling is very different from its state while burning is in progress .
But at the time of maximum pressure the correction is still a small one , and the possibly considerable error in it does not affect their broad result , which was that at 1200o C. or over , the specific heat of the products of combustion in the gas-engine must be much greater than that at ordinary temperatures .
lt has , however , been urged ) Clerk* that the assumption that chemical equilibrium is almost simultaneous with maximum pressure is unjustifiable .
The may have failed to attain equilibrium in either of two ways .
Firstly , the flarne may not have spread to every part .
There may be discrete portions of gas which are still in their pristine state and in which combustion is actually not started .
If I understand their views 'Inst .
C. E. Proc vol. 80 ( 1885 ) , p. 1 .
Explosions of -Gas Air .
orrectly , all , including Mr. Clerk , are agreed that this is not the fact , but hat at time of maximum pressure , if not some time before , combustion fairly started everywhere in the vessel .
My own experiments support this .
Mr. Clerk , however , suggests the second alternative , namely , that though the reaction is initiated everywhere , it is not complete .
On the analogy of other reactions we know that some interval must elapse between the beginning and the completion of the combination of these gases at any point .
Mr. Clerk contends that in the case of a weak mixture the interval is a long one\mdash ; that in places heat is still being produced by the transformation of chemical energy long after the time of maximum pressure , and this though every bit of gas is inflamed before that time ; and he seems to consider that even in strong mixtures a considerable proportion of the internal energy is in chemical form at the moment of maximum pressure .
He suggests that the " " suppressed heat\ldquo ; is to be largely , if not entirely , accounted for in this way .
My experiments do not support such a view as this ; they appear to me to prove that even in the weakest mixtures combustion , when once initiated at any point , is alinost instantaneously complete .
Moreover , they show that the specific heat of the products is very much greater at high temperatures than at low , and the extent of the difference seems to justify the view that it is main reason of the so-called " " suppression of heat It may be added that this rise in the specific heat is consistent with direct observations of that constant for , which have been made up to about 80 C. , and prove that increases considerably .
It was suggested above that in the case of strong mixtures maximum pressure does not necessarily mark equality between the rate of loss of heat and the rate of transformation of chemical energy into heat .
At the moment of maximum pressure the gas , in addition to possessing untransformed chemical energy and heat energy , is in violent motion , and , moreover , it is very far from being in thermal equilibrium .
The kinetic of the gas is being changed by viscosity into thermal , and , quite apart from want of chemical equilibrium , this might cause the pressure to be stationary , and in actual fact must certainly tend towards that result .
The fact that the gas is not in thermal equilibrium would not have any effect upon the pressure observed if it were a perfect gas having a stant specific heat ; in other words , the internal energy of a quantity of unequally heated perfect gas is dependent only on its pressure and not on the ibution of temperature .
The specific heat of the products of combustion in an explosion is , however , far from constant , being greater at high temperatures than at low , and it is easy to see that the result of this is to make the energy of an unequally Prof. B. Hopkinson .
heated mass of such products greater than that of the same quantity at the same pressure when the temperature is uniform .
Thus , if loss of heat to the walls were arrested at the moment of maximum pressure after an explosion of a strong mixture there would be a further rise of pressure , due solely to the attainment of thermal equilibrium .
I have shown that the differences of temperature amount to 50 C. , at least , in a 9 to 1 mixture ; in the present state of our knowledge it is impossible to say how much effect the equalisation of these differences might have upon the pressure , but that it would cause some rise there can be no question .
But though I consider that Messrs. Mallard and Le Chatelier and their followers were probably right in supposing that the was in chemical equilibrium at the lime of maximum pressure , or very shortly after , it seems to me that only a rough approximation to the specific heat can be obtained by a study of explosion pressure records only .
The uncertainty as to the loss of heat , the great differences of temperature between one part and another , and the violent motion of the gas all conspire to make the results inaccurate ; and if the pressure be observed at so long a time.after maximum pressure that the gas may be taken to be at rest and in equilibrium , the loss of heat will have become so great as to make the results wholly untrustworthy .
The circumstances of explosion in the cylinder are , of course , somewhat different from those obtaining in the closed vessel , where the gas before ignition is homogeneous and at rest .
In the gas and air are usually introduced simultaneously at a velocity , and the engine is designed so as to ensure , as far as possible , the thorough mixture of th ' incoming streams .
Nevertheless , there is a possibility that the mixture is not quite complete , since from the nature of the case it depends wholly upon mechanical agitation and not upon diffusion .
The mixture must certainly be eneous in the sense that a sample of , say , 100 from whatever part of the cylinder it be taken , will show the same composition .
On the other hand , it is probable that the .
composition , as shown by a sample very small compared with the dime1lsion of the vessel , and yet immensely great compared with molecular dimensions , will have very reatly varying composition to the precise point from which it is taken .
The structure of the gas may perhaps be of a streaky character , consisting of line of rich mixture embedded in a matrix of weaker mixture .
Moreover , the gases in an engine are probably in turbulent motion at the of ignition , and the movement of the flame will be somewhat different from that observed in a homogeneous mixture at rest .
The agitation of the gas will certainly accelerilte the spread of the flame ; the want of homogeneity Bxplosions of Coal-Gas and Air .
perhaps , retard the combustion at any given point .
It is quite unnecesgry , however , to suppose that these influences have any substantial effect upon the gas-engine indicator diagram in order to explain its peculiarities .
follow at once cfrom the fact , which I think may be considered as ntablished , that the specific heat of the working substance is much greater than that of air , and from the very slow propagation of flame through a weak mixture .
When a fairly rich mixture is exploded in the gas-engine cylinder ( say , , the ignition being before the end of the compression stroke , the diagram shows a very sharp rise of pressure , followed at once by the falling curve of the expansion stroke .
The diagram is almost identical with that which would be given by the sudden addition of the heat of combustion ( less a small percentage of loss ) to the mixture of and inert gas , which results from the combustion , followed by expansion with some loss of heat to the walls .
The internal energy of such a mixture at 1500o C. is about one and a-half times that of the same volume of air at the same temperature and pressure ; hence the rise of temperature on explosion will be about two-thirds of that which would have taken place if the working substance had been air instead of the mixture referred to .
Moreover , the ratio of the specific heats of the mixture between and 1000o C. is something less .
The expansion line will therefore , if adiabatic , lie above the line constant .
There is no necessity at all to suppose that in this case we have in the cylinder anything but a mixture of , and inert gases in complete chemical equilibrium for at least nine-tenths of the stroke .
The natural rate of propagation of flame through such a mixture is high enough to fill the compression space before the piston has sensibly moved ; and the time required for this process will be still further reduced by the motion of the gas and ( probably ) by its high temperature before ignition .
When a very weak mixture is used , the form of the diagram is so far modified that the maximum pressure occurs later in the expansion stroke , the curye co1responding to which first rises then falls instead of falling for practically the whole of its course .
This is exactly what would happen in the explosion of a homogeneous weak mixture , the volume of which is made to increase rapidly while the flame is .
Plainly in such a case the rise of pressure due to the spread of the flame might be balanced by the fall due to increase of volume .
In the gas-engine the volume increases slowly at first , and the pressure then at an increasing rate as in the closed vessel .
As the middle of the stroke is approached the increase of volume gets more and more rapid , until ( at the moment of maximum pressure ) the spread of the flame is just able to keep pace with it .
Then the Explosions of and Air .
piston slows down , and the flame ] overtakes it , and fills the vessel before the end of the stroke .
Occasionally , vever , there may still be unburnt gas present at the moment of release .
Such extremely weak mixtures are , of course , not of much practical importance , because the gas is burnt in a very uneconomical way , much of it being ignited at a low compression .
lt would seem that in all cases of importance gas-engine the working substance may be treated as a mixture of CO2 , , and inert gas , in chemical equilibrium to which the heat of combustion , less a small percentage of loss , is added at the beginning of the stroke .
The slowness of propagation of flame will cause the attainment of equilibrium to be more or less delayed , especially in high-speed ines ; but the efforts of designers will naturally be directed to timing the commencement of the process and hastening its completion , so that every part shall be ignited under the best circumstances , that is , when the compression is a maximum .
The next step in the development of gas-engine theory must be to ascertain the properties of this working substance , viz. , its internal energy as a function of its temperature and pressure .
1 wish to express my indebtedness to Professor for some valuable criticisms ; and I have to thank Messrs. G. B. , of Christ 's College , Cambridge , and W. N. Duff , of Trinity College , for assistance the experimental part of the work .
The second column shows the volutue of oxygen required for complete coml)ustion ; the third and fourth , the resultant volumes of COz and The analysis was kindly given me by Mr. Auchterlonie , Engineer and Manager to the Cambl.idge Gas Coml ) and Reception of Electric Waves .
413 The oomposition of the heavy hydrocarbons\ldquo ; is somewhat uncertain .
Ore hundred volumes of gas require 576 volumes of air for complete mbustion .
One hundred volumes of gas burnt in 900 of air give about 1$3 volumes .
of steam , 57 of , and 780 of inert gases ; assuming that there is no dissociation .
On Methods whereby the diation of may be mainl .
V confined to Certain , and whereby the Receptivity of a Receiver be to Electric from Certain Directions .
By G. MARCONI , LL. D. , D.Sc .
icated by Dr. J. A. Fleming , F.R.S. Received March 15 , \mdash ; Read March 22 , 1906 .
) This Note elates to results observed when for usual vertical antenna employed as radiator or absorber in wireless telegraph stations there is substituted a horizontal conductor placed at a comparatively small distance above the surface of round or water .
When an insulated horizontal wire , AB , such as is shown in sketch 1 , is connected at one end to a sphere of a spark gap , the other sphere of which is eal.thed , and sparks are caused to pass between the spheres , it will be noticed on investigating the space around such an oscillator that the radiations emitted reach a maximum in the vertical plane of the horizontal wire , AB , and proceed principally from the end , , which is connected to the spark gap , whilst the radiation is nil , or reaches a minimum , in directions which are approximately from the direction in which the maximum eflect occurs .
I have also noticed that any horizontal conductor .
sufficient length

