]\gt ; X. The Absolute of and R. WALLACE , B. Sc. , Assistcmt Lectnrer in ysics , College , or .
, by KELVIN , Pres. Received March 2 , \mdash ; Read Malch 23 , 1893 .
THE experiments described in this paper were un dertaken at gestion of Professor A. GRAY , M.A. , tTniversity College , Bangor , the object of contributing to the resnlts still necessary to establish the experimental work bearing on the absolute thermal conductivity of metals on a more satisfactory basis .
They were also intended to furnish a determination of the absolute conductivity of vure copper at different temperatures .
The data already accum u on the thermal of iron and are due chiefly to NGSTRoM , FORBES , NEubfANN , TA 1T and MITCHELL , and more recently to KIRCHHOFF and HANSEMANN and LORENZ .
FORBES fi publisbed , in 186 and 1865 , the details of his determination of the conductivity of iron , and showed that its value diminished with rise of Some years later TAIT confi r this result , and in addition gave determinations of the conductivities of copper , lead , \amp ; c. His results for copper showed that , unlike iron , its conductivity increased with increase of temperature .
STRoM , in 1861 , published results for iron and copper , evidently obtained with great care and by a very reliable method , which showed that the thermal conductivities of bobl ] metals decreased with increase of .
As pointed out by TAIT , both NGSTRoM and FORBES omitted to correct their results for change of specific heat with temperature , but the application of this correction does llot materially affect the nature of their results .
NEUMANN also determined the absolute conductivities of * See table giv en below , p. 588 .
' Trans. Roy .
Soc. dinb .
} , pp. 1 .
, and vo ] .
24 , pp. 73-110 .
A prcliminary communication had previously been made to the British Associntion Belfast in 1852 .
rans .
Roy .
Soc. Edinb 1878 .
S Communicated to the Royal Swedish Academy in , 1861 .
German and lations his paper are to be fonnd in POGGENDOBFF'S 'Annalen ' for 186.3 in the ' Phil. ) ( firsf half-year ) .
'Ann .
de Chim vol. 66 , pp. several lnetals , but gay no results bearing on the change of conductivity with rature .
More recently KIRCHHOFF and HANSEMANN* and LORENZt have results for iron and copper , but their methods were not so reliable as their and the results for copper are exceptionally low .
Although there can be no doubt as to the substantial accuracy of most of results obtained by the experimenters mentioned aboye , yet it is evident that .
on the change of conductivity with temperature are very unsatisfactory , that a good deal of work has yet to be done on the subject .
The method of determination adopted in the here described that due FORBES , with the exception that a single thermo-electric couple was employed for tion of temperatures instead of a number of ometerS .
In addition the uniformity secured by measuring all temperatures on the same scale , there is advantage that by this method the holes in the bar for the insertion of one junction of the couple may be very much smaller than those required for thermometers , may therefore be drilled at shorter distances apart , thus admitting of a accurate determination of the distribution of temperature along the bar .
as no mercury is put in the holes , no lining of iron is necessary in the copper Another advantage in the use of a thermo-electric couple is , that in observing cooling of the short bar used to determine the rate of cooling at different tempera .
tures , thejunction serted in the cools at practically the same ate as the , in the case of a thernlometer , the of the thermometer always behind that of the bar ; an experiment showed that at temperatures between C. and C. the temperature of the thermometer may be nearly a higher than that of the bar in which it is in .
QJerted even when this bar is sufficiently massive to cool rather he bars used in the experiments were of iron and copper .
The iron bar was inch square bar , about feet long , of ordinary commercial wrought-iron , its surface filed up and very lightly polished with black-lead to secure uniformity radiating power .
The copper bar was a round hard-drawn bar half-an-inch in diameter and 7 feet long , of practically pure electrolytically deposited copper , prepared by BOLTON and Co. , Oakamoor Mill , Staffordshire .
The sectional area of the ctly uniform and its surface was smooth polished .
For the reception of the thermo-electric junction , holes about a millimetre diameter were in these bars at , distances apart which preliminary exper had shown to be suitable .
The depth of the holes was about to threeof the diameter of the bar , and care was taken in drilling that their axes were right angles to the bar , and passed approximately through its central axis .
*WIEDEMANN'S ' Annalen , ' 9 , p. 1 ; 13 , p. 406 .
WIEDEMANN'S ' , p. 422 .
See For } ' There ) of Hent , ' SS 298-300 .
CONDUCTIVITIES OF IRON AND COPPER .
The arran gement of apparatus finally ) ployed was that shown in fig. 1 .
One end of , experimental bar was immel.sed in a vessel of nlelted lead ined at a constant temperature , and the of its length was to the air , but protected from draugbts and external radiation by a wide trough of sheet zinc , arranged as shown in the figure .
The molten lead was maintained at a constant temperature by a thermostat .
Th. , consisted of an air thermometer ] laving its bulb immersed in ] ) ad , and controlling , by means of a column of mercury in the -shaped bend of its stem , the passage of the gas at } thejunction of the entrance and exit tubes and .
By means of this the could be maintained at a temperature constallt within a ange of about ] a degree for any length of time .
The the radiation from the heating apparatus by enclosing this atus in a box , lined with sheet iron , off box by a wooden screen lightly packing the space between the box and screen with cotton wool .
The , shown in the figure was intended to shield the part of the bar within the box from ular heating by the flame of the burner , and erved to prevent contact between the cotton wool and hot bar .
The space between this tube the .
was packed at one end with asbestos to preyent current of hot air passing out of the box along the bar .
couple employed was one of a good soft iron silver The were about 5 millim. in diameter , and the vere made by fusing the wires together under borax and filing to the required size .
' howed that with this couple deilection on the ometer scale MR. R. W. WART ON ABSOLUTE THERMAL between C. C. , strictly to the difference between the tempeatures of its junctions .
The nometer in use with the couple was a low resistance reflecting galvanometer , adjusted so as to give a deflection of about four isions for a difference of 1o C. bebween the temperatures of the The scale deflections could be easily read to half a division , so that the difference between the ) of the junctions could be estimated to about one-tenth of a degree .
The ibution of temperature along the bars in the stationary state was deterlnined by .
one junction successively into the small holes driled in the bar , while the othel junction was maintained at a constallt temperature36 by inserting it in hole in a short iron esting o the bottom of the zinc ough T. The deflections obtained were , in this way , proportional to the difference of temperature between the points 011 the bar and the temperature of the surroundin air .
It was found necessary to carefully insulate the ends of the of the couple from one another just above the junctions , and from the bars in which they were ; this was done by filing down the wires , inserting a thin slip of silk between them , and then whipping them round with a fine silk thread .
The outer surface of the thread slightly at the higher temperatures , but not to destroy its efficiency .
In taking a series of observations of the distribution of along the , the heating was allowed to go on for five six hours , and then the deflections corresponding to each hole along the .
were observed , up and down the bar , time after time , until a set of perfectly constant and concordant readings were obtained .
It was not always possible to this end ; a change in the temperature of the room , or any cause tending to produce aught in the , destroyed the constancy of the results , but by repeated observations under fayourable circumstances , and by shielding the bar from draughts by mealls of the trough , it was found possible to obtain perfectly constant data , except the case of the copper bar , for the holes nearest the heated end .
It was found the temperatures of these holes when above ls C. to 20 C. yaried continually , through a range a maxinum value of about 1o .
the hole nearest the end , and quickly diminishing to zero at the third or fourth hole from the end .
This fluctuation in temperature was probably due , partly wallt of massiyeness of the .
and partly to the formation of a film of oxide 011 the surface of this part of the bar ; but by keeping the temperature from going much above 20 C. and by the mean of ten or twelve successive eadings the mean temperature of these holes was determined with certainty to * This temper ture was indicated by a thel. .
placed with its bulb in a hole drilled in the close to small hole which the junction was placed .
In no case was the of than a fifth of a .
Thu bar was a short masve iron bar , the one used cooling , .
OF IRON AND COPP h R. within a of .
A much this would no on the results given below , which for temperatures below 20 C. rates of cooling of the at diflerent temperatures were determined by heating a piece of bar , about a foot long , in a sand-bath to temperature than required , and observing its rate of cooling under the same conditions as those to which the experimental was exposed .
For this purpose it was suspended in the position originally occupied by the long , the thernl 0-electric junction insel.ted in hole drilled at its middle point , and the deflection of the galvanometer observed at regular intervals during the cooling .
It found that the of of cooling were not trustworthy for the first few minutes after it had su that it was to the bar to a temperature higher than the highest at which the rate of cooling was required .
With this precaution it was found that the data obtained for rates of under similal conditions ) well together\mdash ; for example , the data given in columnls and of Table I. were obtained on diffel.ent days but under similal conditions .
Experiments that the presence of a thin of oxide on the SUl.face of the copper bar had perceptible effect on its rate of cooling , it was found tlJat the film did not unless the heating was continued for a long time , so that by the time of in the sand bath it was possible to observe the cooling with the surface in the different initial stages of oxidation , and the rate of cooling found to be practically the same in all cases .
In reducing the results , the graphical method was largely used\mdash ; the distribution of temperature the and the cooling observations were plotted in the usual way , and from the curves so obtained differential curves were plotted , in the one case , the dient of at any the , in the other , the rate of at any temperature .
In the case of curves slJowing the stationary state of the bars , the scale adopted was such represented actual distances ( the the lengths of the ordinates in the ) in scale divisions .
1 the differential curves , however , showing the dient of any point the ) ordinates were on ten times this scale ' .
so that one netre represented about of a degree Centigrade .
The cooling cul .
Ves were on a lIlewhat larger scale in overlapping sections .
The scale varied , according to the data to be plotted , from one in which 4 centims. represented one minute and 5 millims. one scale division ( 1 millim. to of a degree Centigrade ) to one which 1 centim. represellted one minute and 4 millims. one scale division .
to size of the scale , it was necessary to draw the curves in sections , experience showed that ) sections must overlap considerably , in ordel to secure continuity .
The curves showing the rate of cooling at temperature were drawn with bscissae on the same scale as the observational curves , but the tes were in case a scale limes as great .
As it is ] to dnce these curves on the scale which t were drawn , two ) ecimen C are given on reduced in figs. .
) .
) of the teml ) ture curve for the copper bar , fig. II gives the upper part of for the iron The imental results and the details of their reduction are set out in Tables given below , which , taken in coltnection with the followin o general ] nation of the notation dopted , te ] } successive steps of the reductions .
Let denote the absolute thermal conductivity at normal section of the bar at distance from the heated end , where : is bhe gradient of temperature .
, if denote the of cross-section of the bar , denotes the heat which crosses he section considered in unit time .
This heat is by radiation from surface of the .
beyond this section , hence , if denote the length of a short element in part of the bar , its mean tempel.ature , and its rate of cooling this temperature , then the heat lost per of time from the surface of this element is , where denotes the specific heat , and the density of the of the bar at the temperature this it follows that , if all the units involved be consistent , we have . . . . .
( 1 ) .
corrections this relation becomes . . . . .
( 2 ) , where are the values of at C. , the temperature of the cross ection of the bar , and the coefficient of linear expansion of the metal .
The value of for any cross-section can evidently be found directly from ( 2 ) by calculating the corresponding value of 2 , where has its proper value for each element of the bar the reduction )yever be more conveniently made by taking as equal to , where the specific heat at a mean emperat .
ur which is given by ives ns , , where denotes the factor involving In the reductions given below it wiIl be seen that the absolute ductivity , ollnccted with is point , in tbc ] communicated to the Royal uiuted o to mo .
W. OF IRON AND first calculated from the values of , and 2 , and bhe diffusivity afterwards calculated by by the proper values of , the thermal capacity of unit ] of the material of the bar .
To obtain the values of and the ] sities of the iron copper were carefully determined in the usual way , and , as the iron was not pure its specific heat at different temperatures could not be obtained from vious results ; it therefore determined by calorimeter , as described below .
For the specifc heat of copper the results given by BED were adopted .
It will be evident that if ; and are expressed in the same units , the absolute value of these units need not be known , hence , what follows , are given in scale divisions ( galvanometer deflections ) , the absolute value scale division in degrees has but little on the results obtained .
IRON .
I. 1.\mdash ; Table showing distribution of temperature along iron ] dnring ste , state .
Temperature of Between C. and 10 C. , * ' NIemoires couronnes de Acad. dc Bruxelles , ' vol. 27 , CONDUCTIVITIES OF IRON AND COPPER .
Table showing values of , the absolute thermal conductivity of at temperatures between C. C. From these values ( between 10 C. and if we assume the conductivity at .
to be given by we have , on reduci n method of least squares , or .
of Specific of Iron .
Advantage was taken of the snowfall in January to use calorimeter this determination .
The calorimeter was prepared for by the method described by BUNSEN in his paper given ] in the 'Phil .
, p. 161 , and all the precautions there detailed were obseryed .
The tube used for determining the change volume attendant on the melting of ice was of very fine bore , but was perfectly clean , so that there wafi no trouble with the sticking of mercury thread .
This tube was carefully calibrated , and a calibration curve drawn from which a length measured in any part of the tube could be converted into a length corresponding to the uniform bore taken as the basis of the calibration .
The heater used was made specially for the work .
It consisted of a hollow cylindrical oil bath shown in section at .
This bath was heated up to temperature of aboub 25 C. , and then ickly placed in the wooden case shown in the figure and packed round on all sides with cotton wool .
Thus packed , the bath took nearly a day to cool , and could , therefore , be used for a complete series of determinations at temperatures below 20 C. The piece of iron to be heated was suspended by a silk fibre in the cylindrical Mn .
R. W. STEWART ON ABSOLUTE THERMAL space in the centre of the heater in close proximity to the bulb of the thermometer , , which passes through a cork in the lid of the case .
The silk thread passed through the glass tube , , and was attached at the other end to a small hook in the end of the slide , .
The tube , , was just long enough to be in contact with the piece of iron , that when the slide , , was withdrawn quickly the thread broke and the iron free to fall through the holes in the double bottom of the case , between which the slide , previous to withdrawal , acted as a screen .
When it was desired to make a determination the heater was lifted on to the lid of the calorimeter case and placed in a marked position , so that the holes in the bottom were directly over the mouth of a paper funnel , , acting as a guide to the calorimeter tube , .
When placed in position , the thermolneter was watched and the slide , , withdrawn , when the required temperature was indicated .
The iron at once fell straight into the calorimeter tube and the heater was then removed , the mouth of the funnel , , dosed with cotton wool , the vation of the mercury-thread in the indicating-tube proceeded with .
In the meanwhile , a second piece of iron was suspended in the heater so as to be ready for another determination .
The cooling of the heater was so slow that there was no donbt about the temperature of the iron when dropped into the calorimeter , and the time of fall into the tube , , was too small to admit of any appreciable cooling .
Experiment also showed that placing the heater on the lid of the calorimeter case , even for a .
time , had no effect on the calorimeter .
In the reduction of the results obtained the latent heat of water and the specific heat of ice were assumed to be correct as given by BUNSEN .
The rapid melting of the snow and some accidents with the calorimeter limited time of working and preyented any extension of the results given .
The principal data of two sets of experiments are given in the following Table .
CONDUCTIVITIES OF IRON AND ] TABLE showing the Data for Determination of Bore of Tube .
and reducing by method of least squares , we and , therefore , or where denotes the true specific heat at C. Determination of Density of Iron , and the of ( sd ) the ) of Unit Volumc .
The density of the iron found to be C. ) Ul.poses this may be taken as , and the density at will be .
Hence the value of ( sd ) at C. is given by .
of , the Diffusivity of Iron , in Absolute Units .
From the results obtained above the value of at C. will be given by .
That is , , or , .
is result represented with sufficient accuracy by .
CONDUCTIVITIES OF IRON AND COPPER . . .
The mean value of these results gives .
of Density Value of ( sd ) , the Thermal of Volume of Copper .
The density of the copper was found to be at C. Taking this as , we may take the density of copper at C. , as given by .
The specific heat of copper at C. is given by BE as .
This result may be expressed as 0892 , value of ( sd ) at C. is , or .
Value of , the of Copper , in Absolute C.G.S. Units .
From the obtain e above the value of at will be given by .
That is , , or , .
This result is probably represented with sufficient accuracy by the formula .
It will be seen that the results of the experiments here described go to show that for both iron and copper the conductivity with rise of temperature .
Recent determinations of the specific heat of pure copper ( probably identical with that used in these experiments ) give the specific heat at C. as 0928 .
formula gives .
, a value about 3 per cent. too high .
The determinations here referred to were made by J. , FRS , with ; steam calorimeter , and were kindly communicated to me by Mr. JAMES H. , B Sc , University.\mdash ; 12.4.93 .
] For convenience of comparison these results are given below , followed by a Table previous results .
Values obtained for Thermal Conductivities of Iron Copper from Experiments detailed in this Paper .
[ C.G.S. units .
] IRON .
CoppER .
ABLE showing the results of previous Determinations of the Absolute Thermal Conductivity of Iron and Copper .
[ C.G.S. units .
] The Emissive Powers of the of the Iron and Copper the data obtained in the cooling experiments involved in the conductivity research , it is evidently possible to calculate the emissivities of the surfaces of the bars at different temperatures .
As it be of interest to place these on record , they are given in the foUo table .
The iron bar was a inch square bar , and its face was blackened and lightly polished with blacklead .
The copper bar was a round bar of -inch diameter , and its surface was ) and polished .
CONDUCTIVITIES OF IRON AND COPPER .
TABLE showing the emissive powers of the surfaces of the Iron and Copper Bars .
The emissivities are given in gramme degrees of heat lost per square centim. of rface for each degree Centigrade difference of temperature between the surface and .
he surrounding air .
The results are corrected for variation of specific heat with emperature .
On comparison of these results with those published by Mr. D. by Professor for somewhat similar surfaces , it will be found that the greement is fairly close .
The experience gathered in the conduct of the experiments described above goes show that in order to obtain a thoroughly satisfactory determination of thermal \ldquo ; onductivity and its variation with temperature , the bars employed should be somewhat massive\mdash ; not less than an inch in diameter\mdash ; and their surface should be plated ( to prevent oxidation ) and well polished .
Further , to avoid the use of very lOng bars , and to diminish as far as possible the importance of the of heat from surface of the bar as an element in the determination , NGSTRoM'S method should o adopt , ed in preference to FORBES ' .
It is noteworthy that with FORBES ' method the influence of expezimental errors eems to cause an apparent decrease of conductivity with fall of temperature .
trace of this is seen in the tables giving the values of for iron and copper , but the reduction of an unsatisfactory set , of observations made with the copper bar before the use of the trough ( fig. 1 ) gave results which were not only much too low , but which decreased rapidly with fall of temperature .
In conclusion the author has to express his thaluks to Professor GRAY for constant belp and advice during the progress of the work .
* ' Proc. Roy .
Soc 1872 , p. 93 , or EVERETT'S ' Physical Constants , ' p. 107 .
'Proc .
Roy .
Soc. Edinb 1869-70 , p. 207 .

