]\gt ; A Recording Calorirneter for Explosions .
By BEIlTItAM HoPKlNSON , Professor of Mechanism and Applied Mechanics in the University of Cambridge .
( Communicated by Professor H. L. Callendar , F.R.S. Received December 13 , 1906 , \mdash ; Read January 31 , 1907 .
) The determination of the rate of loss of heat to the walls of a vessel after an explosion within it is a matter of considerable scientific interest and of practical importance .
Hitherto such determinations , if we except the recent work of Dugald Clerk on the loss of heat in the engine cylinder , have been based upon a study of the fall of pressure during the cooling of the gases after the explosion .
From the pressure the mean temperature can be deduced , and thence , if the specific heat is known , can be found the rate of heat loss at any moment .
Such a calculation is , , obviously unsatisfactory , because the only available values of the specific heat of gases at temperatures above 1500o are based upon explosion experiments , and involve doubtful assumptions as to the amount of loss before combustion is complete .
Some means of determining the loss of heat at any instant without any knowledge of specific heat is therefore both for finding the law of cooling of hot gases confined in a closed vessel and for placing on a satisfactory basis the specific heat values obtained from explosion experiments .
I have devised a simple means of doing this which appears to be capable of considerable accuracy .
It consists essentially in lining the explosion vessel as completely as possible with a continuous piece of copper strip and recording the rise of resistance of the copper strip during the progress of the explosion and the subsequent cooling .
Knowing the temperature of the copper and its capacity for heat , the heat that has flowed into it from the gas may be calculated from the resistance , certain corrections being applied for the heat which the copper has lost to the Up to the present I have only used the apparatus for the investigation of the loss of heat after an explosion of coal gas and air , but it might , I think , be applicable , with certain cations , to finding the heat loss , during and after the combustion of solid explosives .
The explosion vessel is shown in section in fig. 1 .
It consists of a cast-iron cylinder 1 foot in diameter and 1 foot in length on to which are bolted two end plates .
The cylinder was first completely lined with wood inch thick , and the end plates covered with pieces of cork .
Thirty-nine turns of copper strip of the quality used for electric lighting purposes , and of a high degree A Recording Calorimeter for Explosions .
of purity , were then wound on the inside of the curved portion , a clearance of about 1/ 20 of an inch being left between successive turns .
The strip was : inch wide by 1/ 25 inch thick .
The ends of this piece of strip were brought to terminals outside the vessel .
The end plates were similarly covered with parallel pieces of copper strip of the same dimensions , as shown in fig. 2 , the ends being brazed to connecting pieces .
The on the ends were electrically connected outside the vessel to the strip in the cylin drical part .
The whole when put together formed an explosion vessel having a capacity of about cubic feet which ( except for the uncovered portions on the ends where the cocks , etc. , came through ) was completely lined with an electrically continuous length of copper having an approximately uniform section of Prof. B. Hopkinson .
[ Dec. 13 , 1/ 100 of a square inch .
For recording the pressure I used an optical indicator , consisting of an iron piston which was forced by the pressure against a piece of straight spring held at the ends .
The displacement of the spring tilted a mirror about a fulcrum , and the mirror cast an image of a fine hole illuminated by an arc lamp on to a photographic film carried on a revolving drum .
This indicator was repeatedly calibrated by dead weights , and I think its readings are to be trusted to within 1 per cent. of the maximum reached .
The mixture was fired by an electric spark at the of the vessel , and it was at atmospheric pressure and temperature before firing .
For recording the temperature of the strip , it was connected as shown in fig. 3 .
is a D'Arsonval galvanometer having a stiff suspension and a periodic time of about 1/ 15 of a second .
Its resistance is ohms .
It is placed in series with the copper strip and with a resistance of about of ohm .
A constant current of about 8 amperes is maintained in strip A Recording lorimeter for Explosions .
by means of the battery ( storage cells ) and the bank of lamps L. A constant current of about amperes is maintained in the resistance by means of the battery ( six storage cells ) and an external resistance the current being in such a direction that the E.M.F. at the terminals of opposes that at the terminals of the stlip .
The resistance of the strip ohm at the temperature of the room , the result of this arrangement is that there is an approximate balance of electromotive in the FIG 3 .
galvanometer circuit before the explosion , and the galvanometer then shows little or no deflection .
When the explosion takes place the copper strip is heated and its resistance rises , and since the current in it remains constant during the short time occupied by the cooling of the gas to ordinary temperatures , the potential at the terminals of the strip rises by an amount proportional to the increase of resistance or to the increase of temperature .
Since the potential at the terminals of the resistance remains constant , except for the small disturbance due to the passage of the galvanometer Prof. B. Hopkinson .
[ Dec. 13 , current , the galvanometer deflection from the reading just before the explosion will be proportional to the rise of potential between the terminals of the strip or to the rise of temperature .
The mirror of the galvanometer reflected on to the moving film au image of the same small hole as was used for the change of pressure .
and a simultaneous record was thus obtained of the chance of temperature of the strip and of the pressure in the vessel .
One such record is shown in fig. 4 .
Curve A is the pressure reckoned downwards from the atmosphere line .
Curve is the galvanometer deflection reckoned upwards from the zero line The galvanometer is thrown into slight oscillation , owing to the very rapid change of temperature which occurs when the hot gas first comes into contact with the strip .
The white dots on the lines are due to the fact that an alternating current arc was used for illuminating the hole ; they are very useful in the diagrams , because they enable corresponding points on the two curves to be identified correct to about 1/ 500 of a second .
On the pressure curve , 1 mm. deflection corresponds to a pressure of .
per square inch The barometer in this case stood at 753 mm. , equivalent to lbs. per square inch .
The temperature just before explosion was C. , or absolute .
Allowing for the contraction of volume of 3 per cent. , which occurs in the combustion of a mixture of one part of Cambridge coal gas and seven parts of air , it follows that 1 mm. on the pressure diagram corresponds to a rise of temperature of C. On the curve the galvanometer deflection , 1 mm. is equivalent to a rise of resistance in the strip of ohm , or to a mean temperature rise of C. , assuming a temperature coefficient of .
The total weight of strip is 2870 grammes , and the specific heat is taken as the quantity of heat corresponding to 1 mm. on the galvanometer curve is , therefore , 222 calories .
The values assumed for the heat capacity and temperature coefficient are probably very nearly correct , since the strip is nearly pure copper ; but it is difficult to determine them directly with sufficient accuracy .
In order to confirm their correctness as far as is necessary for the present experiment , a portion of the strip was wound on a wooden frame and was electrically heated by passing a current of about 350 amperes through it The doubling of the zero line is due to the fact that the mixture failed to ignite at the first attempt .
The current was left on for a minute or so while this was being investigated , and slightly heated the strip .
At the second ( successful ) attempt the position of the zero line had therefore shifted upwards to an extent corresponding to the rise of temperature .
All measurements refer to the original film .
The reproduction is approximately three-quarters of the original size .
Hopkin son .
Soc. Proc. , Al .
, Plcpte 1 .
1906 .
] A Recording lorimeter for Explosions .
for seconds .
The amount of energy put into the strip in this process was measured by a ballistic wattmeter , the suspended coil of which was placed ( with a high resistance in series ) as a shunt across the terminals of the strip , while the fixed coils carried the current .
The resistance was measured just before and just after the passage of the currenlt .
In this manner a direct relation was established between the energy put into the copper and the rise of resistance produced thereby .
It was found to be the same , within 1 per cent. , as that deduced from the assumed temperature coefficient and specific heat .
Of the heat which passes into the copper , some part is lost to the wooden backing behind it , and it is the balance only which is directly measured in the diagram .
The percentage of heat so leaking out is a correction which increases from less than 1 per cent. second after up to about 20 per cent. 1 second after .
In order to determine the amount of this correction , recourse was had to the method of electrical heating described above .
A current of about 350 amperes was passed through the strip in the cylindrical part of the vessel for seconds , and the resistance was determined immediately after the current ceased .
The amount of energy put in under these circumstances was found to be per cent. greater than the heat accounted for by the rise of temperature in the copper strip .
The loss of heat to the backing was therefore of that which had gone into the copper .
The loss of heat depends on the manner in which the temperature of he copper rises , which is not quite the same in the explosion vessel as in the electrical heating .
In the latter case the rate at which work is done on the copper is nearly constant , and the temperature rise is nearly proportionate to the time , whereas after the explosion the temperature rises at first very rapidly and then remains more nearly constant .
By the methods of the Fourier analysis of the conduction of heat , however , if the loss of heat for any one surface variation of temperature is known it can be calculated for any other .
The surface temperature of the wood is the temperature of the copper .
We determine experimentally the flow of heat into the wood when its surface temperature is made ( by electrical heating ) to vary as a linear function of the time .
From the result is calculated the heat-flow when the temperature varies in the manner given by Curve in the diagram .
Details of these and other calculations are given at the end of the paper .
In addition to the heat which passes into the copper and , ) the copper , into the backing behind it , heat also goes into those parts of the walls are not covered .
The total exposed area covered by the copper strip is 3200 sq .
cm .
The total surface of the vessel is 4000 sq .
cm .
There are therefore 800 sq .
cm .
, or about 1/ 5 part of the total , uncovered .
Of the VOL. LXXIX.\mdash ; A. Prof. B. Hopkinson .
[ Dec. 13 , uncovered area , however , approximately sq .
cm .
are the narrow strip left for insulation between the adjacent turns or pieces of copper strip .
These turns are separated by mm. on the average , and since the thickness of the strip is 1 mm. , any gas that reaches the backing which is exposed between the turns must part with the greater part of its heat to the of the strip .
This part of the uncovered area may be regarded as protected by the copper , though not actually covered by it .
The balance of 220 sq .
cm .
consists mainly of two large patches on the endplates , and of an annular space round the copper on each end-plate .
It will receive heat at approximately the same rate , on the ayerage , as does the copper .
The whole heat which the gas has lost , therefore , exceeds that which has gone into the copper in the ratio 4000/ 3780 , or by 6 per cent. This correction is the most uncertain point in the whole experiment .
It is underestimated , because some heat , no doubt , finds its way between the turns .
But the error on this account cannot be very As a preliminary test of the accuracy of the copper strip calorimeter , I have calculated the heat accounted for by it at points so far down the cooling curve that the specific heat of the gas may be considered as known .
We may take , for example , a point about 1 second after ignition , when the gas temperature is 54 C. At this point the ordinate of Curve ( after the oscillations ) is mm. , equivalent to 7850 calories .
The correction to be added for loss of heat to the backing ( see Appendix 2 ) is here 20 per cent. ; the total heat which has passed into and the copper is therefore 9420 calories .
Multiplying this by the factor , we get 10,000 calories as the heat which has gone into the walls .
Before firing , the mixture was of the following composition , approximately : Cambridge coal gas cub. per cent. Air ( including some water vapour ) , , Total , , , , the volumes being reckoned at C. and 760 mm. The calorific value of the gas , determined in a Boys calorimeter , was found to be 670 British thernffi units , or 170,000 calories per cubic foot .
The heat produced in the explosion is therefore 14,000 calories , if the products are all cooled to about C. If we suppose the cooling to be stopped at 54 C. , the heat evolved will be less by the latent heat of the steam produced ( about 30 grammes per cubic foot of coal gas ) by the amount evolved by the gaseous constituents in cooling from to C. The 1906 .
] A Recording lorimeter for Explosions .
latent heat item is 600 calories .
For the other item we must have recourse to the results of Holborn and Austin , who have determined the specific heat of steam , CO2 , and air by direct heating at constant pressure up to so C. These results are shown in the second column of the following table , which also gives the composition of the present in the explosion vessel and the heat evolved by each in cooling through the range named .
The specific heats are expressed in calories per cubic foot at constant volume , and are the mean values between and C. The amount of steam is not accurately known , since no measurement was made of the degree of saturation of the gas and air before combustion ; moreover , the steam does not all condense at the lowest temperature of , and some of it is , therefore , not cooled the full range as gas .
But the error on either account is ver .
small .
Steam and Total Steam and Total Steam and Total Steam and Total Steam and Total Steam and Total Steam and Total Steam and Total Steam and Total Steam and Total Amount .
Amount .
Steam and Total Amount .
Steam and Total Amount .
Steam and Total Specific heat .
10.7 8.4 6.3 Heat evolved in cooling Adding the heat of condensation ( 1480 calories ) we get 3807 calories evolved in cooling the products of the combustion from 54 to C. The heat evolved by the products of the explosion in cooling down to 54 C. is then 14,000\mdash ; 3807 , or say 10,200 calories .
This agrees very closely with the heat found by the copper strip oalorimeter , 10,000 calories .
As a further check , we may make a similar comparison at the end of half a second from the time of firing .
At this point the gas temperature obtained from the pressure diagram is 84 C. , which is just beyond the limit of Holborn and Austin 's specific heat determinations .
At this point we have : Heat in copper strip , mm. . .
6720 calories Heat lost to backing , 12 per cent. of the above 810 Heat to uncoyered part , 450 Total heat lost to walls The heat evolved in the products from to C. and in condensing the steam is 5180 calories .
The heat evolved in coolin the burnt products to should , therefore , be ) ries .
Prof. B. Hopkinson .
[ Dec. 13 , According to this estimate , therefore , the copper strip calorimeter is about 800 calories , or 10 per cent. wrong .
There is , however , a considerable possible error in the estimate derived from the calorific value of the gas at this point .
In the first place that calorific value may itself be in error by as much as 2 per cent. , or 280 calories .
Again , Holborn and Austin consider that their measurements of specific heat may be in error by as much as 3 per cent. , equivalent to a possible error of 120 calories in this experiment .
Finally , it is possible that some of the gas is still unburnt .
There are several small connected to the vessel , and there is a small annular clearance space between the endplate and the end of the wooden lining of the cylindrical part .
These spaces amount in the egate to only 1 per cent. of the whole volume of the vessel , but they are so situated that in the progress of the flame unburnt gas is compressed into them .
Thus at maximum pressure these places will be filled with unburnt mixture at a pressure of about 7 atmospheres .
On account of the large surface in these places it is probable that the flame does not penetrate into them at once ; the gas trapped in them burns slowly , coming out of its retirement into the main body of the vessel as the pressure falls , and then .
At time second the pressure is 3 atmospheres , and there may still be as much as 2 or 3 per cent. of the mixture remaining unburnt in the spaces refe1Ted to .
Since 3 per cent. of the mixture represents 420 calories , this goes far to account for the discrepancy between the observed and calculated values of the heat loss .
These experiments establish , I think , the substantial accuracy of this form of calorimeter .
The most uncertain feature is the allowance to be made for the uncovered area .
This has been taken as equal to 6 per cent. of the covered area , which is an inferior limit .
The close agreement of the calculated and observed values of heat loss , 1 second after firing , shows that this allowance is not far wrong .
The other correction , that of the loss of heat to the backing , though far larger in amount , is capable of pretty accurate determination .
It rests really upon a comparison of the rise of resistance produced by the application of the same amount of work to the copper strip when exposed to the air and when in contact with the backing .
It is independent of the absolute meaning of the ballistic wattmeter and of any knowledge of the specific heat or temperature coefficient of the copper .
The of Heat Loss .
The full line curve , fig. 5 , shows the heat loss per square centimetre of surface in terms of the time .
This is curve of fig. 4 , with the oscillations 906 .
] A Recording Calorimeter for Explosions .
slnoothed out and corrected for the loss of heat to the backing .
It may be noted that only the absolute values oIJttlis culve are affected by the uncertainty as to bhe area of bhe copper .
The comparative heat losses at different 1J0 doubt tely shown .
This detel.mination of heat loss is , of , quite unaffccted by the uncertainty to when combustion is complete .
The loss of heat about 1/ 20 second after ignition , when the flame first comes into contact with the copper .
At first the loss goes on at a very great rate , and by the time maximum pressure is reached ( when the flame is in contact with the whole surface of the strip and heat to every part of it ) , about 1700 calories , or 12 per cent. of the gross heating value of the gas , has passed in to the .
The rate of loss of heat at this point is about 10 calories per second per square centimetre , and the mean gas temperature is C. At second from ignition the rate of heat loss Prof B. Hopkinson .
[ Dec. 13 , is about calories per second , and the mean gas temperature is 1300o C. The mean temperature is reduced in the ratio between these two points , the product of mean temperature and pressure is reduced in the ratio but the rate of loss of heat at second is only 1/ 3 of what it is at maximum pressure .
A study of these figures into prominence one aspect of cooling after an explosion , which has , I think , hardly received the attention it deserves .
The immediate effect of the explosion is to raise the temperature of the mass of gas to a mean value of about 1800o C. , and the boundary of this hot body is suddenly cooled by contact with the walls to about C. , which surface temperature remains practically constant during the time of cooling .
The heat passing in to the walls immediately after the flame touches them is drawn almost wholly from the surface layer of gas in contact with them , and as it has to be transferred so short a distance its flow is at first extremely rapid .
This surface layer , however , soon parts with its heat , and further supplies have to be drawn from the inner portions of gas , the cool surface layer now acting as heat insulation .
Thus the rate of flow rapidly diminishes .
The of the mass of flame may , in fact , be expected to follow substantially the same law as that of a solid body similarly treated .
The Fourier analysis shows that if the surface of a body of large extent uniformly heated to temperature be suddenly cooled to temperature zero , and tke surface temperature then maintained constant , the heat-loss per unit of surface up to any time will be at first proportional to the square root of the time , being equal to , where is the thermal conductivity and the thermal capacity per unit volume of the solid .
Strictly true of a body of infinite extent having a plane boundary , this law holds approximately for finite bodies so long as the central portions have not been much cooled .
It is interesting to enquire how far the of the flame follows the same law .
On fig. 5 I have shown in dottPd line a curve whose ordinate is proportional to the square root of the time , reckoned from a point second after ignition .
This point was chosen because loss of heat does not begin at all points of the strip at the same time ; the flame first touches the strip about second after ignition , and it is not in contact with every part until second after ignition .
It will be seen that except for the first fiftieth of a second the two curves agree well up to about second , when the heat loss becomes considerably less than if it followed the square root law .
This would equally happen with a solid body of finite dimensions when 1906 .
] A for Explosions .
the cooling had penetrated to the interior ; the square root law only holds accurately for a body of infinite dimensions .
The Fourier analysis is strictly applicable only to a solid body in which heat transfer takes place only by conduction , and in which the thermal capacity and conductivity are independent of temperature .
In the , the transmission of leat takes place mainly by convection .
The effects of convectionl may probably be mathematically represented in very much the same way as those of conduction ; convection consists in the bodily transfer of gas from a place of high temperature to a place of low temperature , and the corresponding rate of transmission of heat will , for a iven amount of itation , be roughly proportional to the temperature gradient .
There will be an effective conductivity , depending mainly on the state of motion of the gas .
But though in a general way the loss of heat may be expected to go on in the gas according to much the same laws as in the solid body , the differences between the two cases are so great that there is not much a priori ground for expecting exact quantitative agreement .
The closeness with which the loss of heat after the explosion follows the square root law may be to some extent accidental , and hardly in itself justifies making quantitative applications of the same kind in othel cases .
It seems to me , however , to justify drawing a few general inferences .
The most important is that the rate of loss of heat from a hot gas is not only a function of the mean pressure and temperature , it really has very little to do with these quantities .
It is dependent essentially on the temperature gradient in the surface layerjust within the walls , and this , in any practical case , is mainly determined by the history of the surface layer .
No valid inference as to the rate of loss of heat after an explosion can be drawn from experiment on heat conduction or convection in a gas under steady conditions of temperature .
Nor can it be assumed that two masses of gas , in similar vessels and at the same pressure and mean temperature , will lose heat at the same rate , though that assumption is often made .
Another point is that the rate of heat loss may be expected to increase considerably with the density of the hot gas , the temperature and state of agitation being the same .
According to the solid body theory , the rate of loss of heat is proportional to .
Now in a gas , the thermal capacity per unit volume is proportional to the density .
Moreover , the effective conductivity , may also be expected to increase more or less in portio to the density .
For represents the rate of heat transfer by the motion of the hot gas , and the quantity of heat carried by a given volume of gas at a given temperature is proportional to the density .
It would not be surprising if the loss of heat after an explosion were found to be proportional to the density of the gas Prof B. Hopkinson .
[ Dec. 13 , in spite of the fact that experiments under steady conditions of temperature show that the rate of removal of heat from a hot body by gas surrounding it does not increase in proportion to the pressure of the gas .
I think it possible that a good many hitherto obscure phenomena connected with explosions may be explained by considerations such as these .
* Specific Heat terminations .
By deducting from the total calorific value of the gas the heat ] up to any moment as determined by calorimeter , the internal energy ( thermal and chemical ) in the gas at that moment can be obtained .
Since mean , 1907.\mdash ; Since writing the above I have had the advantage of reading two papers by Lord Rayleigh ( ' Phil. Mag March , 1899 ) .
In one of them he investigates atically the law of cooling of a mass of air by conduction , when the surface temperature is suddenly changed , the air being enclosed in a spherical enclosure of constant volume .
The Fourier analysis is applied with the modifications necessitated by the fact that when any portion of the air is cooled below the temperature of the remainder it contracts and thereby causes all the other parts to expand , so lowering their temperature in a manner independent of conduction .
It is shown that if the radius of the sphere be 16# cm .
the excess of the mean temperature of the air over that of the surroundings will fall to half its initial value in 26 seconds .
In the second paper , experimencs are described in which this result is verified , by observing the rate of fall of pressure in a mass of air , contained in a spherical enclosure , after it has been suddenly heated by adiabatic compression .
The experiment is closely analogous to observing the fall of pressure after an explosion , with the difference , however , that the range of temperature is only a fraction of a degree Centigrade as compared with ) or .
The mathematical analysis only applies when the range of temperature is small .
It is , however , interesting to compare the rate of fall of temperature in the two cases .
My explosion vessel happens to be of about the same dimensions as the sphere in Lord Bayleigh 's experiments , and it will be seen from the pressure-curve ( fig. 4 ) that the pressure falls to about half of its maximum value in second , that is , in about 1/ 70 of the time taken for a similar amount of cooling in Lord Rayleigh 's experiments .
This result must be ascribed mainly to the effects of convection after the explosion ; for neither the thermal conductivity nor the thermal capacity ( which is the other constant determining the rate of fall ) is very greatly different in the two cases .
It is known that the gases are set into violent vibratory motion by the explosion , and it seems possible that the greatly enhanced " " effective conductivity\ldquo ; is due to this fact .
Possibly , radiation may also play some part in accelerating the cooling .
The curve of heat loss , calculated by Lord Rayleigh for the spherical mass of air , agrees fairly well with a curve whose ordinates are proportional to the square root of the time , until the temperature has fallen to about of its initial value .
The fall of temperature in half that time is very nearly of the fall occurring in the whole time .
If it went as the square root of the time the ratio would be 2- } or .
After the explosion the temperature falls to 3/ 5 of its maximum value in about second from the time of ignition , and , until then , the curve of heat loss is closely similar to Lord Rayleigh 's curve , the deviations from the square root law being in the same direction and of about the same amount .
1906 .
] A Recording Calo.imeter for Explosions .
temperature of the gas is known a curve of specific heats at all temperatures up to 1800o C. could thus be obtained .
To use Dugald Clerk 's phrase , however , these would be " " apparent\ldquo ; specific heats , involving the assumption that all the gas is burnt .
Now , though that assumption may very likely be true for a vessel with a smooth surface and without pockets , it is certainly not true in the vessel with which I experimented , and in which , as indicated above , the gas trapped in the } ) ockets is a serious proportion of the whole .
I will , therefore , defer all discussion of the information which this calorimeter may be expected to give as to the variation of specific heats , until I have fully investigated the question of delayed combustion .
I desire to acknowledge the very able assistance that I have receiyed in this investigation from Mr. L. du B. Hugo , who was until lately a student and assistant demonstrator in the Engineerin Laboratory , Cambridge .
He carried out most of the experimental work under my direction , and made several suggestions which materially facilitated it .
I have also to thank Professor Callendar for the kind interest which he has shown in the work .
APPENDIX .
1 .
Temperature Distribution in the Copper Strip .
Different parts of the length of the strip receive heat at different rates , but the total heat given to any element of length is equal to the rise of temperature of that element multiplied by its thermal capacity .
Since the strip is of uniform section from end to end , the thermal capacity of an element is proportional to its length .
For the same reason the increase of resistance of an element is proportional to the rise of temperature multiplied by the length .
The rise of resistance of the whole length , therefore , measures the heat given to the whole length , provided that the temperature may be assumed uniform over the section .
This assumption can be readily justified .
The thermal conductivity of copper may be taken as .
At maximum pressure heat is flowing into the copper , to the curve , fig. 5 , at the rate of about 10 calories per square centimetre per second .
The corresponding temperature gradient just within the surface of the copper is per centimetre .
If this temperature gradient were the same throughout the thickness of the strip , the temperature on the side away from the gas would be less than on the exposed side .
As a matter of fact , the gradient is less at points within the strip , and it is apparent that there can be no serious difference of Prof. B. Hopkinson .
[ Dec. 13 , temperature between one part of the section and another .
A small deviation from uniformity is without effect , the rise of resistance being , to a first approximation , proportional to the mean temperature across the section .
2 .
Determination of Heat Lost to Backing .
If the temperature of the plane surface of an infinite solid vary according to the equation , where is the time , then the Fourier analysis shows that the total amount of heat that has passed into the solid at time is where , , are the derived functions of and is the thermal conductivity and the thermal capacity per unit volume of the solid .
In our case the solid is the wood backing , and the temperature is that of the copper strip .
In consequence of the low thermal conductivity of wood , the temperature changes are confined to parts near the copper , and the backing may be treated as infinitely thick , at any rate for the purpose of finding a correction .
The heat accounted for in the copper is CF ( t ) , where is the thermal capacity of the copper strip per unit area .
The heat lost to the backing is to that in the copper in the ratio .
( 1 ) The values of , , etc. , are obtained at once from the curve in fig. 4 ) showing the rise of temperature of the strip in terms of the time .
In order to obtain the percentage of heat loss , therefore , it is only necessary to find the constant in some particular case in which and the loss of heat are both known .
For this purpose , that part of the strip which is in the cylindrical surface was placed in series with a fuse of No. 18 copper wire , a storage battery of 100 cells , and the series coils of a ballistic wattmeter .
The suspended coil of the wattmeter , in series with a non-inductive resistance of 8000 ohms , was placed as a shunt across the strip .
On closing the switch in the battery circuit , the fuse allowed a current of about 350 amperes to pass for about seconds before melting .
A raphic record of the current was taken with a quick period galvanometer , and it was found to rise within 1/ 50 second to365 amperes , it then fell in the course of seconds to 340 amperes .
Arcing then commenced at the fuse , and the current rapidly fell , ceasing altogether at seconds .
1906 .
] A Recording Calorimeter for Explosions .
wattmeter was calibrated in the ordinary way with steady currents .
Its period of oscillation was seconds .
The resistance of the slirip was measured in the ordinary way at definite intervals after the current had ceased .
From the observations of resistance a curve of cooling was obtained and the imnediately after the breaking of the fuse was thus calculated .
This correction for cooling was small .
A number of observations the following results:\mdash ; Work done on strip , as measure( by wattmeter ) ) calories Weight of coppel 1950 grammes Rise of te1nperature of copper C. Heat in copper , 3135 calories Heat lost to 542 The heat lost under these conditions is , therefore , per cent. of the heat in the copper .
the passage of the curlent , work is dove on the strip at a nearly uniform rate for second , and then at a rate for another second .
The heat loss will be very nearly the same as though the temperature had risen at a rate for the whole time .
Putting , etc. , in the expression ( 1 ) above , the expression for the fraction of heat lost in this case becomes whence In the heat loss at different points after the explosion , this value of the constant is assumed , and the values of , , etc. , are taken from the curve of copper temperature ( in fig. 4 ) .
For example , at mm. , mm. per second , , and so on , the rest of the terms being unimportant .
The percentage of heat loss obtained by substituting these figures in expression ( 1 ) is approximately 20 per cent. The specific heat of the strip was dete1nlined by the same ballistic wattmeter method , a weighed quantity of strip being wound on a wooden frame , so that heat loss only takes place to the .
air .
The rise of temperature produced by blowing a No. 18 copper fuse in series with this strip was C. , a tenlperature coefficient of .
The corresponding value of the specific heat was 0.0935 , deduced from the wattmeter reading .
A correction of 23 per cent. had to be applied to the A Recording Calorimeter for xplosions .
temperature rise for the cooling which took place before the resistance could be measured .
It should be added that in both ballistic wattmeter measuremeuts the reading was corrected for the fact that the period for which the current passed ( seconds ) was comparable with the period of the instrument ( seconds ) .
The correction was applied on the assumption that a constant torque acted on the suspended coil for seconds , and it amounted to an addition of per cent. It was found that the amount of energy put into the copper during the blowing of a copper wire fuse of given size was remarkably constant , differing by more than 2 or 3 per cent. from the mean .
This method of determining specific heats is , I think , capable of considerable accuracy , and is certainly convenient .

