]\gt ; Thermal Conductivity of other Gases .
By GEORGE W. TODD , M.Sc .
, Research Scholar of the University of Birmingham .
( Communicated by Prof. J. H. Pointing , F.R.S. Received June 2 , \mdash ; Read June 24 , \mdash ; Received in revised form July 27 , 1909 .
) Although there has been a number of experimental determinations of the ther1n , al conductivity of air , its value even to two figures is still uncertain .
It is true that the values hitherto obtained are of the same order as that deduced from the kinetic theory by Clausius , Maxwell , and others , but they differ themselves to the extent of 20 per cent. There have been only two methods of experiment .
Stefan , Kundt and Warburg , and Winkelmann observed the rate of cooling of a thelmometer in an envelope of rarefied air , the outer surface of which was kept at a constant temperature .
Schleiermacher and Schwarze measured the rate at which heat passed from an electrically heated wire stretched along the axis to the surface of a cylinder maintained at a lower constant temperature .
Repetitions of these two methods have not cleared away the uncertainty as to what is the correct value of the thermal conductivity of air .
Under these circumstances it might be interesting to an account of method for determining this quantity .
It will be noticed that in this method , which was ested by Prof. Pointing , the conductivity is measured at ordinary pressures , there being no need to take precautions against convection currents by the reduction of the pressure as is necessary in the other methods .
If a layer of air is enclosed bebween two parallel good conducting plates maintained at different temperatures , the colder plate will receive heat by radiation and conduction through the air .
There is no reason to suppose that convection can take place if the plates are horizontal , with the distance between them small compared with their linear dimensions , and if the upper one is at the higher temperature .
Communication with the surrounding air could be cut off by an insulating ring .
Let one of the plates be movable up and down , its surface being always parallel to the surface of the other plate .
Then assuming that no complications are introduced by the , the quantity of heat passing in each second from the upper plate to the lower at a distance from it is where is the radiation between the plates and is constant provided the temperature of each plate is kept constant .
Mr. G. W. Todd .
[ July 27 , If is varied and the corresponding values of are plotted on a curve a rectangular hyperbola is obtained , the horizontal asymptote giving the value of the radiation .
The conductivity of the air then be obtained , knowing the dimensions and temperatures of the plates , from .
In practice , however , the insulating ring must radiate to or absorb radiation from the lower plate .
In the experiments to be described the ring was attached to the upper hot plate , and therefore , being warmer than the lower plate , would radiate to it .
The conduction of heat down the ring is a very small amount if the ring is thin .
In the experiments the ring consisted of a cylinder of paper attached to the upper but not to the lower plate , so that only a small fraction of the heat conducted down would get to the lower plate .
The problem of finding an expression for the radiation effect of the ring is apparently a difficult one .
However , it is possible to deduce an expression for the radiation from a vertical plane of finite height to a horizontal strip extending perpendicularly from it to infinity .
In this case it is proportional to the product of the width of the strip and the height of the vertical plane .
We may take it , then , that with plates of a areasurrounded by a ring of small height the radiation effect is proportional to the distance between the plates .
So that Plotting and , we obtain a curve whose asymptotes are and Putting , the equation becomes which is represented graphically thus : 1909 .
] Thermal Conductivity of Air other Gases .
The slope of the asymptote gives the constant whence the thermal conductivity is determined .
Experiments were made with two forms of the apparatus of quite different dimensions , but the methods only differed in detail .
Both sets of experiments are described , as it appears desirable to justify the method by showing that it gives concordant results when dimensions are varied .
In the earlier experiments , too , the surfaces were varied between which the conduction and radiation took place .
The earlier experiments were made using two circular plates about 20 cm .
in diameter cut from sheet copper 3 mm. thick .
One surface of each was tulned as flat as possible in a lathe .
These plates were soldered one to the base of a copper steam chamber , and the other into the top of what may be called the calorimeter , the turned surfaces being outside .
An idea of the arrangement of the apparatus may ) gathered from the accompanying rammatic representation ( fig. 2 ) .
Steam was sent the steam chamber from the boiler situated above , and the temperature of the upper plate was indicated by a thermometer whose bulb was immediately above it .
The steam chamber was fixed firmly by means of screws to a thick annular piece of wood which held uP the whole of the apparatus above it as shown in the tYram .
The rate at which heat passed through the lower plate into the calorimeter was determined from observations of the rise in temperature of a uniform flow of water through the calorimeter .
The latter was 2 cm .
in Mr. G. W. Todd .
[ July 27 , depth , and constructed to ensure the perfect circulation of the stream of water .
This was done by up the interior , by means of partitions parallel to the plate , in a way similar to that shown in fig. 6 .
As the outflowing water comes immediately from the " " \ldquo ; copper plate the temperature of the latter was taken to be the same as that of the outflow .
An inyestigation described below showed that this was not quite true , but the correction is less than 1 per cent. This temperature was read on the thermometer ( fig. 2 ) , and the difference in temperature between the outflowing and inflowing water was measured by the thermojunctions JJ of copper and constantan wires .
By means of a screw , the table which supported the er could be moved up and down in a direction always perpendicular to the horizontal plane .
Both the base of the table and the ring of the tripod were provided with levelling screws .
The position of the calorimeter was determined by observing through the microscope-cathetometer the coinciding of a silk thread stretched across a strip of glass at X with the cross wires in the microscope .
The contact position of the plates was ascertained by connecting them in seeris with a Leclanche cell and a galvanometer .
The temperature of the lower plate was kept as constant as possible by regulating the temperature of the inflowing water with the heating coil of manganin wire in series with a variable resistance .
An admirable insulating ring was made by ting ordinary foolscap paper round the edge of the steam chamber , the contact between the paper and the copper being made as air-tight as possible , to prevent the possibility of convection streams .
While fitting sufficiently well , the calorimeter was able to slide freely up and down in this double-walled cylinder of paper .
This ring was , perhaps , the best possible .
Its thinness and low conductivity would prevent much conduction from the steam chamber to the calorimeter , and the film of air enclosed between its double walls would help to reduce the conduction , if any , from the air between the plates to the outside .
If there were convection currents , they must have been too feeble to have had any appreciable effect on the curves .
It may be possible that small eddies existed close to the insulating edge , but there could have been no rapid translation of the air in the rest of the space .
During an experiment , the flow of water was maintained constant by the ordinary simple device of allowing the water from the mains to enter a reservoir from which an outflow tube carries away water rising above a certain level .
Thus a constant " " head\ldquo ; is obtained independent of the alterations in pressure which may occur in the mains .
In these experiments the reservoir was suspended by a cord passing over a pulley fixed to the 1909 .
] Thermal Conductivity of Air and other ceiling , so that the flow through the calorimeter could be altered when desired .
The volume of water flowing through in each second was measured by noting with a stop-watch the time taken for it to fill a graduated litre flask .
The galvanometer was of the D'Arsonval type and had a resistance at C. of ohms .
Its terminals were soldered directly to the copper leads from the thermojunctions .
If the volume of water flowing through the calorimeter in each second is and the galvanometer reading is , the quantity of heat received per second by the lower plate is where is the deflection on the galvanometer scale corresponding to a difference of 1o C. between the inflow and outflow junctions .
But where is the thermal conductivity , is the radius of the plates , and is the temperature difference between the plates .
Therefore we may put , where so that The deflection per 1o C. difference between the junctions was divisions .
The radius of the copper plates was cm .
, so that The thermal conductivity of air has been calcuIated by finding the constant A graphically , as indicated at the beginning of the paper , and substituting it in this equation .
Each of the following gives the experimental data for one determination of the conductivity .
The figures opposite give the ance apart , in centimetres , of the plates , and the figures immediately below give the corresponding galvanometer deflection:\mdash ; Surfaces Outflow temp. .
Steam chamber temp. Rate of flow .
in sec. Mr. G. W. Todd .
[ July 27 , Surfaces blackened .
Outflow temp. .
Steam chamber temp. Rate of flow .
in sec. .
Thus Outflow temp. .
Steam chamber temp. Rate of flow .
in sec. .
Thus Outflow temp. .
Steam chamber temp. Rate of flow .
in sec. .
Thus The mean of these values obtained with blackened surfaces gives The mean of all the results gives , for the value of the thermal conductivity of air at ordinary pressures and in the neighbourhood of C. , results would each be increased by probably about S per cent. if a correction were made for the difference in temperature between the lower plate and the issuing water .
The apparatus was dismantled before the author realised the necessity for the correction .
Another form of the apparatus was then constructed , and the opportunity was taken to introduce some improvements .
It appeared better to the plates thicker and of some alloy harder than copper which could be worked into a plane surface without fear of bending , although with thicker plates it would be necessary to correct for the rence in temperature between the upper and lower surfaces .
With a larger area , the edge effect would probably 1909 .
] Conductivity of Air other be reduced to an imperceptible amount at the small distauces used in the experiment .
As the determination of the conductivity depends on the accurate measurement of the distances between the plates , care was taken in designing the new apparatus to have the plates held in more rigid supports .
The position of the bottom plate was adjusted by means of three levelling screws with fine threads , and sensitive spirit levels on the supporting base indicated when the plate was horizontal .
A wooden table being of no use with sensitive levels , a rigid one was specially built up , consisting of slate slabs cel1lented to the tops of four earthenware drain pipes , which again were cemented to the brick flool .
The micro-cathetometer , by means of which the position of the lower plate was determined , rested on this slate table .
The temperature of the upper surface of the top plate was measured by a thermocouple , one junction pressed against the plate and the other immersed in boiling water .
Instead of measuring the difference in temperature of the water and the orimeter by reading galvanometer deflections , a null\ldquo ; method was used .
The electromotive force produced at the thermojunctions was balanced against the fall of potential down a copper wire of known resistance and carrying a known current .
A sketch of the arrangement for supporting the plates is given below .
The steam chamber was fixed to a cast-iron , supported by three steel rods screwed into the base B. These three rods , for simplicity , are shown in the as two .
The steam chamber could be raised or lowered by means of three screw heads resting in conical bearings drilled in projecting portions of the cast-iron ring ( see section ) .
The calorimeter rested on a smaller and thinner cast-iron supported Mr. G. W. Todd .
[ July 27 , by three steel rods fixed into a strong -shaped cast-iron base , details of which are given in accompanying diagrams ( fig. 4 ) .
Two adjustable spirit- were fixed at right angles to each other on this base .
By the aid of these and three } screws , the calorimeter could be moved up and down and the surface of the plate kept parallel in all its positions .
The spirit-levels were about 3 inches long and contained ether , so that the bubbles would respond readily and with as little viscous restraint as possible .
A displacement of the bubble through one division corresponded on one to a fraction over 3 seconds , while one division on the other indicated a tilt of almost exactly 3 seconds .
As the levelling screws enabled adjustment to less than one-tenth of one of these divisions , and as the plates were more than 30 cm .
in diameter , it will be seen that differe1lces in level between opposite ends of a diameter must be less than mm. The plates were turned from rolled-sheet brass .
inch thick .
These would be sufficiently strong to avoid risk of bending , and , being made of brass , better surfaces could be obtained on them .
For the careful turning of the surfaces of these plates and for the fitting of them to the steam chamber and calorimeter , I am indebted to Mr. G. O. Harrison , the mechanician in the Physics Department of the University of .
I must also express my gratefulness to him for the construction of the calorimeter , which is described later .
Two shallow dishes having the same diameter as the , plates were " " spun\ldquo ; out of sheet copper inch thick .
One served as the steam chamber and the other was converted into the calorimeter .
The construction of the steam chamber was simple , one of the plates being soldered on to one of the shallow copper dishes , holes having been previously drilled in latter for the entrance and exit of the steam and for fixing in the supporting rods .
The continually condensing steam would be carried away by two narrow tubes soldered as low down as possible at opposite sides of the steam chamber .
1909 .
] Conductivity oj Air other A supply of steam from a large boiler entered by way of a flexible metal tube $ inch in diameter , and a similar tube conducted the steam back again through a condenser to the boiler .
The top of the steam chamber was oovered with cotton wool to diminish the condensation inside .
An important part of the apparatus was the calorimeter , by means of which the rate of transmission of heat across the space between the plates measured .
As in the previous experiments , partitions were introduced to provide proper circulation of the water .
The construction can be best unders.tood from drawings .
A ledge ( fig. 5 ) was turned on the edge of plate so that it could be slipped into the copper dish for soldering .
In addition , another ledge was turned inside the first for the purpose of holding the first partition .
This partition was made of -inch brass and flattened so that the distance between it and the under surface of the plate should be fairly constant all over .
Round the edge of the partition -inch holes were drilled as near together as possible .
The circulating water comes up through these holes , flows along the under surface of the plate taking up its temperature , and makes its exit down a -inch tube in the centre .
This first partition was held in position by fine pins through the previously mentioned ledge .
The other two partitions were of thinner material and riveted together with six spacing pieces inch in diameter .
Three longer pieces riveted to the edges of the middle larger partition supported the two partitions on the bottom of the copper case of the calorimeter ( fig. 6 ) .
After soldering the middle partition to the sides of the case , the plate with the upper partition fixed to it was slipped on and soldered in .
The central tube was screwed into the top partition and soldered through a hole in the bottom of the calorimeter to the lower partition , after which the hole was soldered up .
Mr. G. W. Todd .
[ July 27 , This calorimeter rested on a cast-iron ring fixed by steel rods to a firm base .
Two glass rods , holding a small glass plate over which were stretched crosswires of silk , were cemented across two of these steel rods , and the position of the intersection of these fibres could be read on the cathetometer to mm. At both the inflow and outflow tubes of the calorimeter there were five junctions of copper and constantan ( fig. .
The wires were passed through rubber bungs fitting into glass tubes and were insulated inside by ebonite rings , through each of which 10 small holes were drilled .
junctions were bent over so as to be in the hne of flow of the water .
As in the previous experiments , the inflowing water could be kept at a constant temperature by means of a heating coil .
The electromotive force set up by the difference in temperature between the two sets of junctions was measured by balancing it against the fall of potential down a copper wire ( fig. 8 ) .
By means of a switch either the junctions at the inflow and outflow , or the junction in contact with the upper surface of the top plate , could be thrown into the potentiometer circuit .
The current was supplied by accumulators and kept constant at ampere by adjusting the resistance in the circuit .
The heat received by the calorimeter per second is .
The difference in temperature gives rise to an E.M.F. , , where is the resistance of the wire per centimetre , the current it calries , and is the length required to balance .
But for small temperature differences , where is E.M.F. per 1o C. difference .
Thus 1909 .
] Thermal Conductivity of Air and other Gases .
We have also .
( 2 ) Putting and , where ; ( 3 ) and thus A is obtained , as before , by plotting and X. The difference in temperature between the upper and lower surfaces of the layer of gas will be less than the differenc.e in temperature between the steam in the steam chamber and the water the calorimeter , partly because of the temperature slope through the plate and partly because of the difference in temperature between the metal and the issuing water .
We will deal with these separately .
The rate at which heat passes through the top plate is and through the bottom plate where conductivity , temperature of steam chamber , thickness of plate , temperature of oufflowing water , and temperature of surface of plate in contact with air .
Thus , ( 5 ) and .
( 6 ) is practically equal to ; therefore by subrracting the last two equations .
The correction is , as we might expect , very small .
It was suggested to the author that the difference in the temperatures of the outflowing water and of the receiving plate might be increased by the formation of a stagnant layer of water immediately beneath the plate , and so the second correction might be necessary .
Although the distance between the plate and the partition below it was only inch , it was a point which needed an experimental test .
the plate was made one junction of a thermocouple , the junction being in the plate 1 inch from the edge ; the other junction was placed in the outflowing water .
Any difference in temperature between the junctions was indicated by the deflection of a galvanometer connected in series with them .
The deflection was found to be more than that required by the temperature Mr. G. W. Todd .
[ July 27 , slope in the plate , indicating that there was either a nant layer , or some contact difference of temperature , between the plate and the flowing water .
This deflection , however , which corresponded to a fall from the plate to the water of C. , was independent of the velocities used in the expeliments and practically indepe.ndent ( the change being about 10 per cent. between the highest and lowest heat flux ) of the heat fluxes .
So that in the last equation has to be diminished by C. Constants of the Apparatus .
Mean thickness of top plate cm .
bottom plate , , Radius of plates Thermal conductivity of at C. Current through'potentiometer wire Resistance of wire at C. ohm per cm .
E.M.F. per lo C. difference in inflow and outflow junctions ) volt betweenjunction in steam cham , ber and that in boiling water , , ( Corresponding to cm .
of potentiometer wire .
) Therefore and Experimental observations are shown in the following table .
is the outflow temperature , which was kept as near as possible to C. by the flow of water through the calorimeter and by adjusting the current in the heating coil .
The third column gives the number of seconds required for the outflowing water to fill a litre flask .
At each distance this was measured several times and the mean taken .
The variations amounted to 1909 .
] Conductivity of Air other two- or three-fifths of a second , and were evidently experimental and not real variations in the rate of flow .
The fourth column gives the flow in cubic centimetres per second .
is obtained from the fifth column .
The lengths of potentiometer wire required for balance are given under .
From 20 to 30 minutes were required for steady conditions .
After this time the balance point moved to and fro about 1 per cent. from a mean position .
Mean boiling water and wire cm .
Thus and .
From th .
curve .
But a value not very different from that obtained in the first experiments .
Mr. G. W. Todd .
[ July 27 , Thermal Conductivities of other The thermal conductivities of other gases may be obtained by comparison with air .
To do this , one has only to fix the distance between the plates and provide a gas-tight insulating ring .
Using the usual notation , the heat crossing from the hot to the cold plate is Also where , denoting the sum of the radiation between the plates , the edge effect , and the conduction of the ring , is constant , whatever and , if the temperatures of the plates are kept constant and provided the plates do not tarnish so as to affect the radiation .
Whether they did so or not could , of course , be determined by repeating the measurement of heat flow with air between the plates .
We have , then , for air , If the space between the plates is occupied by a gas of conductivity whence .
The plates were separated by a narrow ring of ebonite of such dimensions that the area of contact with the plates was small compared with the area of the gas layer .
The enclosed space was sealed up by painting round the edges of the plates and ring with shellac varnish .
Two holes in the ring at opposite ends of a diameter served as inlet and outlet for the gases .
At each hole the ebonite was cut " " flush\ldquo ; with the plates and the entrance or exit tube , having on the end a short piece of rubber tubing , was pressed against the plates over the hole .
The contact between the rubber and plates was sealed with shellac varnish .
The gas to be experimented upon was passed from a gasholder sulphuric acid into the space between the plates , displacing the air or gas already there .
Mean thickness of ebonite ring , cm .
Internal diameter , .
Therefore , if the other constants are the same , 36 Mr. G. W. Todd .
[ July 27 , The following table gives the results obtained for one or two gases .
Air was taken first , then carbon dioxide , oxygen , and air again .
The second determination of for air was made to see whether any change had taken place in the radiation .
In the case of nitric oxide , the reading of the potentiometer wire slowly increased , after the supply of gas had been stopped for some time , from 275 to 387 cm .
This must have been due to the formation of nitrogen dioxide , the oxygen of the air diffusing in through the exit .
It was probable , too , that the polished surfaces of the plates had been attacked and the radiation thus increased .
A re-determination of for air showed this to be the case and , after the experiments , when the plates were separated a thin red film covered their surfaces .
In the second experiment with nitric oxide , one or two bubbles of air were sent in order to make the oxidation more rapid and perhaps more complete .
The wire reading increased more rapidly from 280 to 399 cm .
, almost the whole length of the potentiometer wire .
The first value for the conductivity of nitric oxide has been calculated , taking for air 4777 and the second value with for air 4946 .
Both values , however , are certainly too high , because of the formation of the higher oxide , which attacks the plates and increases the radiation .
The second value for nitrogen dioxide , taking for air as 4946 , is probably the right value for that gas .
The values in this table are of the same order as those obtained by previous experimenters .
According to the kinetic theory , the thermal conductivity of a gas is proportional to the product of its viscosity coefficient and its specific heat at constant volume , so that between wide limits it should be independent of the pressure .
The results given in this paper show that the conductivity at ordinary pressures is not very different from the conductivity determined at pressures of a few millimetres of mercury .
It may be interesting to give a table of some previous determinations of 1909 .
] Thermal Conductivity of Air other Gases .
thermal conductivities for comparison .
The columns and are the observed values of the conductivity at and 10 C. espectively .
In the fifth column , is deduced from these , assuming that the conductivity is a linear function of the temperature .
A Determination of the Badiation Constant .
The same apparatus may be used to determine the radiation constant .
In our equation ( see equation ( 3 ) ) , where and , the quantity is given by the intersection of the lines and will be proportional to , the constant of radiation , and if the surfaces are perfect radiators , or if the emissive power of the surfaces in terms of a " " black body\ldquo ; is know1l , can be calculated .
For the quantity of heat radiated per second will be ( equation ( 1 ) ) to Stefan.s law , , where is Joule 's equivalent and is the absolute temperature of the radiating surface .
Whence If the ratio of the emissive power of the surfaces to that of \ldquo ; black body\ldquo ; is , we must write The correction for the temperature gradient through the plates assumes bigger proportions because powers of the temperatures enter in .
From equations ( 5 ) and ( 6 ) , The ratio was determined for a metal surface painted black .
This surface at the temperature of steam was exposed to the junction of Boys radiomicrometer and the deflection compared with that produced by a Mr. .
W. Todd .
[ July 27 , " " black body\ldquo ; at steam temperature .
The black body was made by soldering a cylindrical brass vessel 8 cm .
long and 5 cm .
diameter , blackened inside and containing an opening cm .
diameter , into a larger cylinder 11 cm .
long and 8 cm .
diameter .
Steam could be passed through .
space between the cylinders .
In making a comparison of the painted surface with the black body , each in turn was set up before the radiomicrolneter .
When steam was passing freely the scale reading was noted and immediately the junction was exposed ; after 15 seconds the scale was again noted and the junction covered .
This procedure was repeated several times with one radiator and then with the other .
The exposed readings were plotted as ordinates and the unexposed as abscissae ( fig. 9 ) .
If we draw any ordinate AB , AB\mdash ; OA will give the deflection produced by the black body when the unexposed reading is OA .
AC \mdash ; OA will 6 the deflection produced by the black paint surface when the unexposed reading of the radiomicrometer is also OA .
The ratio ( AC-OA)/ ( AB-OA ) is equal to .
Six such curves gave for , giving a mean of The surfaces of the plates in the conductivity apparatus were coated with the same black paint , and experiments were made in the same way as for the determination of the conductivity .
Tables I and II observations of experiments .
The constants of the apparatus were the same as those in the conductivity experiments , so that and From Table I\mdash ; Thus ] ergs/ cm2 .
Table II\mdash ;

