]\gt ; [ 229 ] VI .
On the Heat Dissipated by at Ifigh Part IV .
Thervnal nissivity in High-pressure By J. E. .
Inst. C.E. , A.M. Inst. , Past 1851 Exhibition John Fellow of the Owens College , lchester .
by ARTHUR SCHUSTER , F.R. Received ) ruary 7 , 1901 .
CONTENTS .
l'age Introduction . . . . . . . . . . . . . . .
229 Apparatus . . . . . . . . . . . . .
230 Experimental Work . . . . . . . . . .
) Causes of Error . . . . . . . . . .
233 ResuIts obtained . . . . . . . . . .
234 On the Variation of missivity with Pressure . . . . .
237 An Analytical Study of the Total Heat . . . . .
) Some Experiments on the Loss of Heat by Convection . . . . . . . . . .
243 On the Influence of Experimental Conditions . . . . .
The Dimensions of the diator . . . . .
244 The Dimensions of the Enclosure . . . . . .
244 The Temperature of the Gas . . . . .
Average Temperature of the Gas . . . . .
245 A Numerical Comparison of the Heat dissipated by Conductivity in Solids and Emissivity in Gases . . . . .
246 Explosive Properties of Nitrous Oxide . . . . .
246 Emissivity in Liquid Gases . . . . .
) Conclusions . . . . .
) PART Thermal issivity in pressure The question of the heat dissipated by a hot body in gases at ordina1y pressures has received considerable attention during recent years .
The subject has been * For Parts I. , II .
, III .
see ' Philosophical Transactions of the Royal Society , ' , vol. 191 , t ) .
501 , 1898 .
Part I. was entitled ' ' Emissivity of a Bright Platinum Surface in Air and other ases Part II .
, , " " A Bolometric Study of the Law of Thernlal tion.\ldquo ; Part III .
, , " " On the tion of the Intrinsic Brilliancy of Pl tinum with \ldquo ; MR. J. .
PETAVEL ON THE HEAT DISSIPATED BY experimentally treated in many rent ways .
The rate of cooling of a body of known specific heat has been directly measured by DULONG and PETIT , NARR , MACFARLANE , NICHOL , STEFAN , BRUSH , BOTTOMLEY , WINKELMANN , KUNDT , and ' ECKERLEIN , GRAETZ , \amp ; c. By CHRISTIANSEN'S method , the value of the conductivity has derived from the fall of temperature per unit length along the axis of a cylinder carrying a constant flow of heat .
SCHLEIERMACHER , SALA , AYBTON , and others have preferred to measure the quantity of electrical energy dissipated per unit time .
These experiments have , however , been carried out at or below the atmospheric pressure , and the question of the heat dissipated in gases at high ssures has rarely been touched upon .
From the ordinarily accepted principles of the Kinetic theory of gases , it may be shown that the conductivity of any perfect gas is independent of pressure .
The experimental work of STEFAN and of KUNDT and WARBURG has gone far to confirm this law as far as ordinary pressures are concerned .
It will be seen , however , that at higher pressures , only a small proportion*of the loss of heat is due to conductivity , and the question as to whether the theoretical law is strictly correct , though well worth investigation , is not of primary importance .
For the above reasons the present work has been restricted to a study of the total heat dissipated at exceptionally high pressures and temperatures .
The Apparatus .
The method employed is the same as that described in 1898 in the tirst part of the present series .
It will therefore be sufficient to recall here that the measurements are made by means of a calibrated as an electric thermometer according to system , and heated by an electric current , the readings of the current passin through the wire , and the electromotive force at the potential berminals , being made by a potentiometer .
The standard resistance had a temperature coefficient of , and as it was iently cooled , the small outstanding variations in temperature involved no correction .
The results of the comparison of this coil WOLFF , with two standards of reference are given below :\mdash ; Standard ohm Standard ohm in terms of the in terms of the working standard .
working standard .
On the 27th September , 1895 On the March , 1900 The original resistance of the working standard was , according to the Reich stalt ificate , at C. and at C. On the 3rd March , 1900 , its esistance having increased about per cent. , was therefore * About at 160 atmospheres .
PLATINUIsI SURFACE AT HIGH TEMPERATURES .
ohm .
From similar observations it was deduced that the electromotive force of the CLARK'S cell ( No. 5217 A ) had fallen by per cent. during the last years .
The variations being in opposite directions tend to cancel , and the outstanding error as far as the energy measurements are concerned is not sufficient to affect the results .
On the other hand the temperature measurements are , of course , independent of the absolute value of the standard .
The CLARK'S was kept in the er chamber of a Berthelot calorimeter ; only very slow variations of its temperature were therefore possible , and the agreement of the electromotive force with the temperature indicated by the attached thermometer was thus ensured .
The enclosure was formed by a steel cylinder centims. in outside diameter centims. bore and 45 centims. long ( see fig. 1 ) .
It was surrounded by a water jacket .
The temperature of the enclosure was estimated by a thermometer , the spherical bulb of which just fitted a hole drilled in the walls of the steel cylinder , half-way along its length ; the depth of the hole being such that the bulb was equidistant from the inside and outside face of the wall .
After the thermometer was in position the hole was filled with mercury to ensure satisfactory contact .
The ends of the steel enclosure were closed by the fibre plugs , held in position by the gunmetal nuts , .
The current was led in and out by the cop per bars , ( about 1 centim. in diameter ) , passing through the centre of the plugs .
One end of the radiating wire was silver-soldered to the copper terminal , the other end was connecterl to by a flexible connection ( 200 , No. 40 , S.W.G. ) .
The radiating wire was kept taut by means of the light spring S. One of the potential terminals was connected to the steel cylinder , the other kept insulated and passed through the fibre plug .
The enclosure was kept at a constant temperature by means of a water circulation .
The same pure platinum wire was used as a radiator in all the experiments , but owing to two slight accidents the calibration is not qnite the same throughout .
The wire was calibrated four times during the course of the It is , of course , necessary to eliminate the cooling effect of the terminals , and to ensure this , the heat dissipated by the middle third of the wire was alone measured .
In the experiments made in 1898 the radiating wire was placed vertically .
At the pressure of one or two atmospheres that was then used , consistent results could be obtained by this disposition , and it had the advantage that the wire was an axis of symmetry with regard to the convection currents .
At high pressures , however , owin to increased convection , it is impossible to keep a sufficient length of radiating surface at a uniform temperature unless the axis of the radiator is horizontal .
For the reasons given in Part I. ( see page 506 and Plate 20 ) , the absolute value of the emissivity is rather greater when the radiator is horizontal than when it is vertical .
The pressures recorded were obtained from the readings of an air manometer , which was calibrated in the ordinary way by weighing mercury .
The glass mano meter tube ended in an air bulb , contained in a strong steel vessel resembling a small Cailletet bottle .
The relative volumes of the air bulbs were ulated so that one instrument began reading at 7 atmospheres and the other at 100 .
The tubes were so arranged that the level of the mercury in the reservoir could be estimated , and the correction for the weight of mercury column calculated .
An ordinary Bourdon gauge was attached to each manometer for convenience of reference during the experiment .
All joints and connections were made metal to metal without intermediate packing .
Where high pressures are to be used , the importance of designin the connections in this manner cannot be exaggerated .
A metallic joint once made will remain tight for any length of time .
And on the other hand , if at any time it is necessary to change the connections , the joint can be broken and re-made in a few minutes .
Experimental Work .
The same routine was observed throughout in taking the observations .
The enclosure having been filled with gas at about the required pressure , the water circulations were started .
As soon as the temperature had become constant , the pressure of the gas , and the temperatures of the standard resistance , of the pressure auge , and of the enclosure , were read , and any other observations that might be of importance , such as the barometric pressure , the temperature of the room , of the circulating water , \amp ; c. , recorded .
In all cases two readings of the current and force were taken , so as to cancel any thermo-electric effects by reversing the current .
The differences in the readings shown in Table I. were due in great part to inequalities in the resistance PLATINUM SURFACE AT HIGH TEMPERATURES .
of the two sides of the reversing switch .
They were in a large degree eliminated in subsequent work .
As however the mean of the two readings is always taken , of this kind hardly affect the results .
The method of working out the observations has in all cases been identical to that shown in Table I. , but for the sake of brevity and clearness the final results have alone been given .
The emissivity expressed in C.G.S. units is equal to the number of therms ( watergramme-degrees ) dissipated , per square centimetre of surface of per second , per degree Centigrade above temperature of the enclosure .
During the experiments the temperature of the enclosure was usually about C. The results of the observations are recorded in Tables II.-VI .
In the case of each series the pressure indicated is the absolute pressure of the gas when no current is passing through the radiating wire .
The pressure readings recorded have been corrected for the ture of the manometer , the height of the mercury column and the departul.e of air from BOYLE'S , this latter correction being made according to AMAGAT'S results .
The importance of having some data with regard to the total heat dissipated cannot be denied , but our knowledge of the phenomena is not at present sufficient for a general theory to be evolved .
It was therefore thought preferable to reduce the results in such a manner that they should not inyolye any assumptions which later work might prove to have been unjustifiable .
It must be admitted that the emissivity as above defined is dependent to a certain degree on experimental conditions , such as the size and shape of the radiator and enclosure .
But the .
is gained that , with these conditions specified , it is a well-defined quantity , and that when the laws which govern the transfer of heat through high-pressure gases become accurately known , the generalisation of the present results will present no difficulties .
The choice of the size of the radiator and enclosure was made with a view to mimmise the objection that has been pointed out .
Before passing to other considerations , two possible causes of error require investigation .
If what SMOLUCHOWSKI DE SXOLAN has called the discontinuity of temperature\mdash ; the equivalent of what is known to electricians as contact resistance\mdash ; exists between the enclosure and the contained gas , the observations will give results too low .
This effect , however , is inverselv proportional to the pressure of the gas , and above 1 atmosphere would correspond to an effective increase in the radius of the enclosure of less than centim. It need not , therefore , be brought into consideration .
A more serious cause of error lies in the fact that the walls of the enclosure being 2 centims. thick , the inside surface might be considerably above the tempera* All the results are referred to the area of the radiator as measured at temperalure of about C. 'Phil .
Mag vol. 46 , p. 192 , 1898 ; E. GEHRKE , ' Annalen der Physik , ' 4th , vol. 2 , p. 102 , 1900 .
VOL. CXCVII.\mdash ; A. 2 ture indicated by the thermometer ; this again would cause the emissivity to be under-estimated .
Taking the maximum values of the temperature and pressure , we find that in .
at 160 atmospheres and 1200o C. the emissivity is , and the heat dissipated C.G.S. units per second , per unit length .
The inside diameter of the enclosure is centims. , its superficial area sq .
centims. per unit length .
Taking the conductivity of steel as , this would correspond to a fall of temperature of C. Thus temperature interval we estimated at 1200 , really was 1198 .
This cause of error would be very serious were it not for the fact that it decreases not only with the square of the temperature but also with the pressure .
At 160 atmospheres and 50 C. the flow of heat is C.G.S. units , making the error C. ; at 100 atmo spheres and 50 C. the error is C. , and for 30 atmospheres and 50 C. C. Over the greater part of the range of observation the error due to this cause is therefore below one-third per cent. but it to about two-thirds cent. at the highest and pressures Results obtained .
In all cases the heat dissipated by a hot body surrounded by gas is the sum of three distinct quantities , all three being functions of the temperature , and one at least being also a function of bhe pressure .
The formula for the total radiation expressed in words will therefore read\mdash ; Total heat dissipated Convection Conduction Or Emissivity . . . . .
Of these four quantities , two , Conductivity and Radiation , have been determined by previous experimenters , and the determination of a third will enable us to solve the equation for any value of or .
The heat carried off by convection not being directly measurable , our only resource is to determine it by difference .
The experimental results will be found plotted in figs. 2 to 6 .
It will be seen that in all cases above about 10 atnnospheres the emissivity is practically a linear function of the temperature , from which fact we conclude that the loss by radiation must be relatively very small .
We may thel.efore write for any given pressure Both the values and as the pressure rises : closer observation shows that for any one gas they may be expressed as exponential functions of the pressure .
We thus obtain for the total emissivity an expression of the form . . . . .
( ii .
) , .
235 7emper\amp ; tuae in degree6Centycdde .
ernpsrsture u MR. J. PETAyEL ON THE HEAT DISSIPATED BY Fig. 5 Fig. 6 .
where emissivity in C.G.S. units , pressure in atmospheres , ( Temperature of the Radiator \mdash ; Temperature of the Enclosure ) in degrees C. The values of the constants and , are given in the following table:\mdash ; PLATINUM AT HIGH TEMPERATURES .
In figs. 2 to 6 the full lines represent the formulae , the constants of which have just given ; the points are the actual observations .
The agreement between the calculated and observed results will be seen to be satisfactory .
The increase of emissivity with pressure in the case of air is shown in fig. 7 .
Fig 7 .
On the Variation of )with Fressure .
To return to the general formula for the heat dissipated .
From the results so far obtained , we may write for pressures above 10 atmospheres Emissivity , and therefore Heat lost by convection in therms per sq .
centim. , per .
degree temperature interval , per second . . .
( iii .
) , where are constants , the pressure , and and the temperatures of the radiator and the enclosure .
The numerical values of each of these three terms of the above for air at temperatures between 100 and 1000 and pressures between 10 and 160 atmospheres will be found in Table VII .
, and the values of the emissivity for the other gases studied are recorded in Tables VIII .
to XI .
If we consider the temperature constant , we obtain for the variation of emissivity with pressure and we have seen that the exponents of are different for every gas studied .
It is not easy to reconcile this fact with the theory that conductivity , viscosity , and specific heat are all three independent of pressure .
It must be remembered , howeyer , that the theory is deduced on the assumptions that the molecular paths are straight lines , that the radius of the molecular sphere of action is very small compared with the mean free path , and that the cohesion of the gas is a negligible quantity .
Neither of these hypotheses seems altogether justifiable at pressures of one or two hundred atmospheres .
*The numerical value of the heat lost by conductivity is merely given to show that , as far as our present knowledge goes , it forms a very small proportion of the total loss observed at high pressures .
Though the conductivity of gases at pressures is fairly accurately known its temperature coefficient is as yet uncertain .
No observations are available above C. , and even at lower temperatures there is much discrepancy between different observers .
For air , for instance , WINKELMANN gives EICHHORN , SCHLEIERMACHER , and ECKERLEIN .
Again , the effective temperature of the gas is uncertain , but to make sure of 1tot under-estimating the by conductivity we have used the maximum value DULONG and PETIT Annals Chimie Physique , found for the loss due to convection and conduction per degree temperature interval At constant temperature the variation with pressures would be where a constant , pressure , for air , for hydrogen , and for In the formula ( ii .
) above is of the order , and for small values of , the emissivity is mately equal to where constant , pressure , and ( see p. 236 ) is for air , for hydroge ] ] and for PLATINUM SURFACE AT HIGH TEMPERATURES .
The results of WARBURG and VON have shown that for gases .
their critical temperature , there is a great increase of viscosity as the critical is approached .
MEYER explains , and at the same time confirms this result , by a calculation based on the assumption that the gas is a mixture of double and simple molecules .
The relatively large value of the exponents of 1 ) ( see page 236 ) in the case of carbon dioxide and nitrous oxide , may be accounted for by the above fact .
It has been shown that the conductivity of a gas is proportional to its viscosity .
Taking into account that even the so-called permanent gases diverge considerably from BOYLE'S law at the higher pressures , it seems probable that the constancy of this conductivity is far from absolute .
A glance at figs. 2 to 6 will show that in every case a sharp fall occurs in the curve of emissivity when eases below a certain limit ; and from such measurements as have been made at small temperature intervals there can be no doubt that at pressures above 10 or 20 atmospheres the conductivity accounts for only a very small proportion of the total heat dissipated , but none the less the conductivity may have considerably increased in absolute value .
The present experiments do not , therefore , definitively settle the question .
To eliminate convection at the higher pressures , the temperature interval between the radiator and the enclosure would have to be a small fraction of a degree .
An apparatus designed to measure up to 1000 degrees is not suitable for this kind of work .
It was therefore thought preferable to leave the subject for a future investigation , rather than record values which vould only be a rough approximation .
On the other hand , at ordinary pressures the constancy of the conductivity has received ample experimental verification , and there is no difficulty in the absolute amount of heat lost by each of the three factors : radiation , convection , and conduction .
An Study of the Total Heat Dissipated .
Our first object must be to detel.mine what proportion of the loss is due to conduction alone .
In the present case we have two coaxial cylinders : the one the the other the enclosure , and we must determine what is the thermal conductance of the air filling the vening space .
carbon dioxide .
When it is taken nccount that DULONG and PETIT'S ived from observations made at pressures below 1 atmospherc and under tobally different conditions , even this approximats agreement is not devoid of interest .
'Wied .
Ann vol. 17 , p. 390 , lS82 .
About for hydrogen for all other gases .
Let us divide bbis space into a number of concentric cylinders , having a length equal to the distance between the potential terminals of the wire a thickness The thermal resistance of each of these cylinders will be ' where is the adius of the cylinder and the conductivity of the gas under considel.ation , in this case air .
The total thermal resistance will thei.efore be Integl.ating between the limits radius of the enclosure and ?
radius of the radiatol , we have Total resistance , the conductance Total flow of heat per degree Centigrade of temperature interval per second .
Now the emissivity is defined as the flow of heat .
unit surface of radiator per degree Centigrade per second , and the aboye expression divided by the superficial area of the wire will thus be equal to the part of the emissivity due to conduction alone .
\mdash ; . . . . .
( iv .
) , where is the part of the emissivity due to conduction alone .
In the present case , thus : Invel.sely , if the convectio1l is zero and the radiation conductivity .
By measurin the emissivity at low temperatures and pressures , we can reduce , though not entirely suppress convection , and , by subtracting the heat lost by radiation , obtain a comparison with the standard determinations of the conductivity of gases .
The value thus calculated will always be in excess of the true conductivity of the gas by an anlount proportional to the heat lost by conyection .
In the following table is the emissivity C. , and atmosphere ; the radiation calculated from J. T. BOTTOMLEy'S experiment and is the conductivity at C. according to WINKELMANN:\mdash ; 'Roy . .
Proc vol. 66 , p. 276 , 1900 .
PLATINUM SURFACE AT HIGH fflS .
The fact thus brought out , that the convection is a maximum in carbon dioxide and a minimum hydroge may cause some surprise , for the coefficient of viscosity of the former is only half that of the latter .
It must , however , be remembered that the force causing convection is the buoyancy of the heated gas which is proportional to its density .
The heat conveyed by the stream of gas is proportional to the volume of the gas in motion , its mean rise of temperature , its mean velocity , its specific heat , and its density ; thus we have for a given temperature Convection Where a constant , the density of the gas , the specific heat , the viscosity .
According to the above formula , the relative convection in carbon dioxide , oxygen , air , and hydrogen is represented by the numbers 20 , , and .
If we take into consideration that the experimental numbers , and are obtained from the difference of two nearly equal quantities , the agreement may be considered satisfactory .
We are now in a p5sition to express in absolute units the heat lost respectively by convection , conduction , and radiation .
* This fact is confirmed by the experiments of KUNDT and WARBURG ( see ' Pogg .
Ann vol. 156 , p. 179 , 1875 ) .
They found that , in a certain apparatus , the heat sipated was independent of pressure up to 30 millims. in the case of air , and up to 154 millims in the case of hydrogen .
This relation between the heat lost by convection and the density , viscosity , and specific heat of a gas only holds good when the speed of the convection currents is very low ; at higher temperature and pressures the phenomenon is more complex .
VOL. CXCVII.\mdash ; A. 2 I At one-tenth of an atmosphere and 10 C. , we have for the various gases\mdash ; Pressure , Atmosphere .
Temperature interval , C. * And if , for the present , we assume the conductivity constant the values at 100 atmospheres and 10 C. would be\mdash ; Pressure , 100 Atmospheres ; Temperature interval , The result is remarkable ; for instance , in the case of air , at of an atmosphere 12 per ) .
of the total loss is due to convection , and at 100 atmospheres 92 per cent. is due to this same cause .
The heat lost by convection is in air 100 , and in hydrogen 500 times greater at a pressure of 100 atmospheres than it is at atmosphere .
For temperature-intervals of several hundred degrees it is impossible experimentally to eliminate either of the three effects of convection , conduction , or radiation , but the first of the three can be greatly diminished .
To this end the apparatus was filled with lightly packed glass-wool .
The proportion of the total volume actually occupied by the glass was per cent. * The loss of heat is expressed in therms per square centimetre of radiating surface , per second per degree Centigrade of temperature interval .
PLATINUM SURFACE AT HIGH TEMPERATURES .
Emissivity Air .
As may be seen in the above table the result was to reduce convection to one-half or one-third of its former value .
The effect on the total heat lost is of course least at low pressures , at which conduction plays the most important part ; the decrease is about 14 per cent. at 1 atmosphere , and twice or three times as much at higher pressures .
On the Influence of dittons .
The remaining factors which influence the total amount of heat dissipated .
the dimensions of the enclosure and radiator and the temperature of the gas.ae Let us first consider the effect of a change in the diameter of the radiator .
We have found above ( p. 240 ) that the heat lost by conduction is , where is the radius of the enclosure , the radius of the radiator , and the conductivity of the gas .
If the ratio is very we may , within a limited ra1Ige , considel as constant and put *Some experiments on this question were recorded in Part I. , ] ) the additional co1tsiderations may be found of interest .
MR. J. E. PETAyEL ON THE HEAT DISSIPATED BY Simultaneously with conduction ( i.e. , the dissipation of heat by molecular motion ) we have convection , by which the heat is carried off by the general upward flow of the gas .
With an infinitely small radiator each molecule comes in contact with the radiator at the temperature of the enclosure .
With a radiator oi large area a certain proportion of the molecules which strike its upper surface have become heated by previous contact .
The heat lost by convection decreases therefore as the diameter of the radiator increases ; it is obvious , however , that it can never become zero , but must tend towards a constant quantity .
Such experimental observations as are available point to the same conclusion , the loss by convection being best represented by an expression of the form The loss by radiation is independent of the size of the radiator and may be put equal to a constant , .
The variation of emissivity with the size of the radiator is therefore expressed by the following formula:\mdash ; Emissivity . . . . .
This is the same expression as was established empirically by AYRTON and KILGOUR .
Within the range of their experiments to millim. both and were constants .
In an investigation on the effect of the size of the enclosure , formula ( iii .
) is not suitable .
It was established that the length of the cylinder was sufficiently great in comparison with its diameter to render the end effects negligible .
We therefore cannot put .
If , however , we take the simple case of a spherical enclosure and radiator , and reason by analogy , we can easily obtain the information we desire .
Proceeding as on page 240 , the thermal resistance of the elementary spherical shell is Integrating we get for the total resistance The part of the emissivity due to conduction is therefore * AYRTON and GOtJR Phil. Trans have shown that an expl.ession of this form holds good for the smallest diamet , ers that can practically be used .
It would , however , be unwise to extend the formula much beyond the limit within which it been experimen tally proved .
It is evident that for the expression is at fault , for the value of the emissivity cannot exceed NC where is the number of impacts and the molecular specific heat .
PLATINUM SURFACE AT HIGH TEMPERATURES .
If is made very great is nearly constant with regard to and is equal to , showing that if the enclosure is above a given size , the heat lost by conduction is practically independent of its actual dimensions .
The theoretical treatment of the variations of the emissivity with the temperatures of the gas offers some difficulties .
On the one hand the conductivity increases in proportion to the viscosity , on the other the convection is a function of the density divided by the viscosity .
Taking the gas at the same density , both at the high and the low temperature , we may safely predict that at low pressures , where the effect of conductivity predominates , a rise in temperature will correspond to an increased value of the emissivity .
At high pressures , however , with the data at present available , a theoretical solution is not possible .
The results of two sets of experiments made with carbon dioxide are given below\mdash ; Emissivity in Carbon Dioxide in C.G.S. Units .
We see that even at 20 atmospheres an increase of 34 per cent. in the absolute temperature causes an increase of about 10 per cent. in the ] of the emissivity .
In the case of carbon dioxide , and probably of all easily condensible gases , when the pressure is near that which would bring about a change of state , the emissivity is diminished by a rise of temperature .
Finally the question of the average temperature of the gas in the enclosure in its relation to that of the radiator offers some interest .
It can be determined from the variation of pressure , for the apparatus constitutes a rough form of constant volume thermometer .
On the other hand , when the heat lost by convection is small , the average temperature can be calculated from the equations on If we diyide the gas into a series of concentric cylinders , the boundaries of these cylinders will be isothermal surfaces , and the fall of temperature from the inner to temperature of the radiator-temperature of the enclosure in degreos Centigrade .
the outer surface of each cylinder will be to the thermal resistance .
By plotting the tempel.atures in terms of the volume of the successive cylinders , rating the curve and dividing by the total volume , we obtain the average of the gas .
This works out as 15 per cent. of the temperature of the radiator if we take the temperature of the enclosure as the zero of our scale .
The same ratio experimentally determined is about 6 per cent. This relatively low value is accounted for by the fact that each gaseous particle does not only part with its heat to the succeeding one , but streaming upwards , comes sooner or later in contact with the upper wall of the enclosure .
11 Comparison of the by in and by Emissivity in Let us now consider the numerical values of the total emissivity obtained at high pressures .
At 160 atmospheres , for instance , in air the emissivity is ( see formula ) Thus at 160 atmospheres air dissipates heat at the same rate as a body having a conductivity , therefore at about the same rate as glass or plaster of Paris .
The conductivity of water is about the same\mdash ; , but in this case the comparison is not a fair one , for placed in similar circumstances convection would also come into play , causing the total loss of heat to be much greater .
Hydrogen at 120 atmospheres behaves like a body having a conductivity .
Finally the emissivity in liquid carbon dioxide is about , which would correspond to a conductivity of , therefore higher than mercury or bismuth .
The conductivity in liquid nitrous oxide is nearly the same .
Near the critical temperature the emissivity in the liquids rises very rapidly ; it falls again instantaneously to less than one-tenth of its former value when the gaseous state is reached .
A point of some interest which was observed during the course of the work was the gradual change in the phenomenon of the decomposition of nitrous oxide .
At a pressure of 1 atmosphere the decomposition was very slow and more than half an hour had elapsed before the change was anything like complete .
PLATINUM SURFACE AT HIGH PERATURES .
At increasing pressures the decomposition became more rapid , and finally , in an experiment in which the apparatus had , at the start , been partially filled wit liquid gas , it took the form of a sharp explosion .
The ature of decomposition is for all pressures about the same , and lies between 800 and 900 degrees Centigrade .
Some measurements taken in liquid nitrous oxide and liquid carbon dioxide are Table .
Conclusions .
The general result of the investigation is to show that the law which connects the thermal emissivity with the temperature of the radiator and the pressure of the gas is\mdash ; where Emissivity Heat dissipated per second , per degree Centigrade temperature interval , per centimetre of area , expressed in gramme-degrees .
, are the constants given on page 236 , is the pressure in atmospheres , and the temperature interval .
The bearing on astro-physical problems is at once apparent , and the high effective conductivity of compressed gases must be taken into account when treatin questions relative to the mean temperature of the sun , its rate of cooling , or more generally , any problem dealing with the physical state of gaseous astral bodies .
Although extrapolations are always very uncertain , in the present case it seems justifiable to conclude that under the very great pressures with which astronomical physics has to deal , all gases or vapours would transmit heat at a higher rate than the best conducting solids known .
From a practical point of view , the results obtained explain the very rapid cooling of the products of an explosion , a fact which has not only been noticed during the course of experiments on ballistics , but has long been well known to all engineers engaged in the construction of gas engines .
The relatively ] cooling surface required , in the case of gases under high pressures , is a fact that had been previously noticed , and is already to some extent taken account of in the design of the modern compressors .
The laws of thermo-dynamics show that the efficiency of all heat engines is dependent on their temperature range and pressure : in compliance with these laws the pressures used are daily becoming greater .
It is therefore hoped that the present investigation may be of some use to those engaged in practical Before closing , I desire to express my indebtedness to Dr. MOND , the donor of the Davy-Faraday Laboratory , at which Institution the greater part of the work was carried out .
My sincere thanks are also due to Dr. ALEXANDER for the kind way in which he has facilitated the research by eyery means in his power .
PLATINUM HIGH TEMPERATURES .
IIT .
Emissivity in Oxygen .
TABLE .\mdash ; Emissivity in drogen .
VOL. CXCVII .
\mdash ; A. 2 .
J. I ON THE DISSIPATED BY TABLE V.\mdash ; Emissivity in Carbon Dioxide .
Temp. PLATINUM SURFAC AT HIGH ATURES .
TABLh \mdash ; Emissivity in Nitrous Oxide .
TABLE VII .
These msasurements are only given as preliminal.y and altogether approximate stated , the apparatus is not snitable for measurements depending on ture intervals . .
J. { ON HEAT DI SIPATED BY VIII.\mdash ; Emissivity in Air .
In this the four following the figures in , referring to pressures outside the range of the formlae , are from a direct interpolation of the observed values .
In every case the emissivit .
is expressed in water-gramme-deg1ees per square centimetre of radiating surface per second .
TINU AT HIGII XI .
Emissivity in Carbon Dioxide .
rees TABLE .\mdash ; Emissivity in Nitrous in degrees Uentigradc .
)

