]\gt ; XI .
BAKERIAN LECTURE.\mdash ; The Effects of Temperature and .
Pressure on the hennal Conductivities of Solids.\mdash ; Part The Effects of Low Temperatures on the hermal and Electrical Conductivities of Certain Approximately Pure Ietals and CHARLES H. LEES , D. Sc. , F.R.S. , of Physics in the East London College , University of London .
Receiyed November 8 , \mdash ; Read in Abstract December 12 , 1907 , \mdash ; Received in Revised Form , with Additions , March 24 , and Read as Bakerian Lecture , March 26 , 1908 .
[ PLATES So-31 .
] DURING the last fifty years a considerable amount of attention has been bestowed on the question of the val.iations of the thermal and electrical conductivities metals with the temperature , but the results obtained by different observers , especially of thermal conductivities , differed so widely from each other that the answer to the thermal part of the question long remained doubtful .
In recent years , however , there has been an accumulation of evidence in favour of a slight decrease of thermal conductivity with increase of temperature from C. to C. in the case of most of the metals .
The experiments of LOBENZ , more especially the careful work of JAGER and DIESSELHORST , Shave contributed greatly to this result .
In both these cases the experiments were limited to the ange of temperature between C. and C. , and it seemed advisable , in view of the importance of both questions in the electronic theories of conduction of heat and electricity in metals , to extend the range over which the theories could be tested , particularly in the ection of low temperatures , where the experiments of DEWAR and had already furnished information as to the electrical conductivities .
The present paper contains all account of the measurements of the thermal and electrical conductivities ried out for this purpose , *Part I. , " " The Effect of Temperature on the Thermal Conductivities of some Electriced Insulators appeared in the ' Philosophical Transactions , ' , vol. 204 , pp. 433-466 ( 1905 ) .
A good account of the subject will be found in article in the new edition ] ( 1906 ) of WINKEI , ' Handbuch der Physik , ' Band III .
L. LORENZ , ' Ann. der Physik , ' 13 , p. 422 ( 1881 ) .
S W. JAGER and H. DIESSELHORST , ' Abh . .
Phys.-Techn .
Reichsanstalt , ' 3 , p. 269 ( 1900 ) .
J. DEWAR J. A. FLEMING , ' Phil. Mag 36 , p. 271 ( 1893 ) .
( 437 .
) 29.7.08 and the results obtained .
The first section deals with measurements of the thermal , the second with measurements of the electrical , conductivities of certain metals and alloys , and the third section compares the results with the electronic theories .
SECTION I.\mdash ; THE THERMAL MEASUREMENTS .
Outline of Method Used .
When the thermal conductivity of a substance which conducts heat readily has to be determined , it is advisable to give the material the form of a wire or thin rod and to allow the heat to flow along the axis .
By this arrangement it is easy to insure that the difference of temperature to be measured is not too small , and it has been a favourite one with observers .
Sometimes the rod has been exposed to the air of the room , as in the original experiments of BIOT , *sometimes enclosed in a yessel as in those of WIEDEMANN and FRANZ .
As the present experiments were to be carried out at temperatures down to that ofliquid air , the former method could not be used without entailing an excessive loss of heat from the sides of the rods .
The method of enclosure was therefore adopted .
The scale of the apparatus was mainly determined by the necessity oflowering its temperature to that of liquid air and maintaining it at that temperature .
This leads to a reduction in the size of the rods , and must be compensated by an increase in the accuracy of the heat and temperature measurements .
This was secured by communicatin the heat to one end of the rod and measuring it electrically , and by determining the difference of temperature at two points of the rod by means of two platinum thermometers .
The Apparatus .
The metal rods used , , fig. 1 , were circular cylinders 7 to 8 centims. long , 585 centim. in diameter , and were accurately turned out of larger rods .
In use they were placed vertically , the lower end fitting into a copper disc , , of centims. diameter , centims. thick in the centre , 1 centim. thick at the circumference , which in its turn fitted accurately into the lower end of a copper cylinder , , of centims. internal , centims. external , diameter , centims. long , closed at the top .
On the rod three thin brass , and centim. long , were placed , the fit being of such a nature that the sleeves could be slid along the rod with the fingers without the application of more than a small force .
S * J. B. BIOT , ' Traite de physique ' ( 1816 ) .
G. WIEDEMANN and R. FRANZ , ' Ann. der Phys 89 , p. 497 ( 1853 ) .
I had not succeeded at this date in making thin mica rings which would stand the wear of use with the whole of the rods , see p. 428 .
S For the accurate turning and fitting of the rods and sleeves I am indebted to Mr. E. T. COOK , the Mechanician of the University of Manchester .
CONDUCTIVITIES OF METALS AND ALLOYS AT LOW TEMPERATURES .
383 The outer surface of each ring was coyered with shellac varnish , thinly dusted with fine marble powder , and the rin then allowed to dry .
On the top of this insulatin layer , in the case of the sleeve , a No. 40 single silk-covered platinoid wire , centims. long , was wound , so that the winding occupied the middle portion of the sleeve , leaving a margin of centim. at each end .
On the insulatin layer of each of the two sleeves A and an equal length of single silk-coyered No. 40 pure platinum wire* was similarly wound .
The ends of the fine platinoid wire were soldered to the ends of two 55-centim .
lengths , , of No. double silk-coyered and shellac-varnished copper wire , and the ends of the fine platinum wire to two 48-centim .
lengths , or , of the same copper wire .
To obviate on the fine wires , the copper wires of each sleeve were placed in parallel grooves in the edge of a thin disc of wood , , attached to the sleeve by a few turns of silk thread , , wound round the sleeve , wood and wires .
In each case a total length of centims. of the thin wire used was required to make connections to the ends of the copper leads , the remaining length , centims. , being actually wound on the sleeve .
The three sets of copper leads passed out of the copper enclosure through three holes in the thin part of the copper disc , which formed the bottom of the enclosure .
At the holes extra insulation was provided for the wires , and they could be fixed in position with respect to the disc by three small wedges of wood placed in the holes of the disc .
The copper enclosure was supported on a wire frame , by means of which it could be placed in a straight Dewal tube , V , of centims. interllal diameter , 25 centims. internal depth , so that the bottom of the enclosure was 2 centims. above the bottom of the tube .
The copper leads were brought up to the mercury switch arrangements described on p. 384 , by means of which the Fig. 1 .
difference of the resistances of the two platinum coils and the actual resistance of one of them could be found .
The current supplied to the heating coil , and the EMF at its terminals , were measured as described on p. 384 .
Around the outside of the copper tube an insulated platinoid wire , , having the same resistance as that of the fine platinoid wire of the heating coil , , was wound , and whenever the current was switched off the heating coil on the rod , it was switched on to that of the tube , so that the amount of heat supplied to the apparatus should be the same throughout an experiment .
In addition , a further coil , , was wound on * Obtained specially from Messrs. JOHNSON , \amp ; Co , the outside of the tube to allow the temperature of the tube , if necessary , to be rapidly raised .
The The mercury switch to which the leads from the platinum resistance coils were brought was so arranged that when the difference of resistance of the coils was to be foumd , they formed two of the arms of a resistance bridge , the other two arms of which consisted of two coils , of about 2 ohms each , of manganin wire together and adjusted to equality to within 1 part in 10,000 .
A dial resistance with mercury contacts , allowing a variation of resistance from to ohm in steps of ohm , could be placed in series with either of the platinum resistance coils in the bridge , and provided the adjustment necessary for balancing the bridge approximately .
The resistance required to give an accurate balance was calculated from the deflections of the galvanometer for the two values of the dial resistance nearest to the balance value .
As the leads to the two platinum coils were made as nearly as possible equal to each other , the resistance so found is the diflerence between the resistances of the two coils .
Similarly , when the resistance of one of the coils was to be determined , that coil was , by means of the mercury switch , made one arm of a resistance bridge of which the two equal manganin coils formed two other arms , and the fourth consisted of a resistance box giving 10 ohms in steps of ohm , and the -ohm dial in series .
Th leads again bein as nearly as possible equal , the resistance of the fourth arm , when the bridge was balanced , was equal to the resistance of the platinum coil .
The current used in balancing the bridge was derived from a Leclanche cell connected to the bridge through a resistance of 53 ohms .
The connections of the cell could be reversed if necessary .
The galvanometer was of the moving-coil type , and had a resistance of 22 ohms .
It could be used either with or without a shunt of 10 to 50 ohms , as occasion required .
The bridge key was a thermo-electric one which , in its normal position , disconnects the cell from the bridge , but leaves the galvanometer connected , and , when depressed , disconnects the galvanometer for an instant , connects the cell , then re-connects the galvanometer .
By this means thermo-electric effects in the bridge have no influence on the observations .
Method of Experirnenting .
In setting up the apparatus for an experiment , the rod , sleeyes , and edges of the copper disc forming the base of the tube were smeared with olive oil , and the joints between rod and base of tube , between rod and sleeves , and between the base of the tube and the tube itself , made in such a way that the oil excluded all air .
Good thermal contacts were thus secured .
None of these contacts come in the direct line CONDUCTIVITIES OF METALS AND ALLOyS AT LOW TEMPERATURES .
385 of flow of heat within the region of measurement , but they influence certain small corrections , to be considered presently .
Before the rod was placed finally in the tube the leading wires from the coils were wedged in the holes at the base , and the distances apart of the three sleeves measured to centim. The apparatus when fitted together was lowered into the Dewar tube , the connections made , and after a few minutes , during which the temperature throughout became uniform , the differences , if any , between the resistances of the platinum wires were measured .
An electric current was then sent round the heatin coil on the rod , and was adjusted so that a suitable difference of resistance between the upper and lower platinum coils was obtained .
The heating current was then switched from rod to tube , and after 5 to 10 minutes the difference of resistance of the platinum coils , and the actual resistance of the lower coil , were found .
If a second observation , taken 3 to 5 minutes later , gay a different value for the difference of resistance , a further interval of 3 to 5 minutes was allowed and the difference again tested , and this was repeated till the difference of resistance became constant .
The current supplied to the heatin coil and the E.M.F. at its terminals were then read , and the current switched from tube to rod .
After 5 to 10 minutes the difference of resistance of the platinum coils was again found , and obseryations taken till it remained constant .
The resistance of the lower coil was then measured , and the current in the heatin coil and the E.M.F. at its terminals read ; then the current was again switched from rod to tube and the measurements repeated .
The mean of the two differences of resistance found when the heating current flowed round the tube , subtracted from the difference found when the heating current flowed round the rod , gives , so long as the rate of rise of temperature of the apparatus is regular , the difference of resistance which would be produced if the heating current were supplied to the rod and the temperature of the tube were kept constant ( see p. 397 ) .
The Dewar tube was then filled with liquid air and , when the apparatus had cooled down to the temperature of liquid air , the remaining liquid was poured out by tilting the tube .
Observations of difference of resistance between the platinum coils , with the heating current round the tube , were then made as before till the ence became steady , the resistance of one coil was determined , and the heating current and voltage measured .
The current was then switched from tube to rod , and the observations repeated .
In this way observations with the heatin current alternately round rod and tube were made throughout the gradual rise of temperature of the apparatus from that of liquid air to some temperature above that of the air of the room , the rate of rise of temperature being , if necessary , increased by a constant electric current sent through the supplementary heating coil wound on the copper tube .
VOL. CCVIII.\mdash ; A. 3 Theory of the Apparatus .
If a thin rod of uniform cross section is entirely surrounded by a vessel kept at constant temperature , and has heat supplied to it near one end , while the other is in good thermal contact with the wall of the vessel , the distribution of temperature within the rod , from its of contact with the vessel to that at which the heat is supplied , will be represented by the equation , ( 1 ) where is the excess of temperature of the rod at the cross section situated centims. from the end of the rod in contact with the vessel , over that of the vessel , is the perimeter , and the area of cross section of the rod , the heat lost per second from 1 sq .
centim. of the surface of tha rod when its temperature exceeds that of the enclosing vessel by C. , is the thermal conductivity of the rod , and is the amount of heat which crosses the section of the rod at any point per second .
In calculating from the energy supplied and the heat lost from the surface of the bar beyond , we may , if the point is near the free end of the rod , and the heat is supplied to the rod at a uniform rate between and the free end , take the mean temperature of the surface of the bar between and the end to be identical with the temperature which would be observed at , a point which divides the distance between and the end , in the ratio 1 : 2 if the above equation for held throughout the heated portion of the rod .
* If is the area of the surface of the bar beyond and is the total heat supplied to the bar , . . . . . .
( 2 ) Hence sin. ( 3 ) If the temperature excesses and at two sections and are observed , then Hence and , . . . .
( 4 ) * This ratio is exact if the loss of heat from one square centimetre of the surface of the bar may be neglected in comparison with the heat generated per square centimetre .
CONDUCTIVITIES OF METALS AND ALLOYS AT LOW TEMPERATURES .
387 where , \amp ; c. , are written for the functions , \amp ; c. , whose values for small values of the denominators do not differ much from 1 .
From this equation may be calculated if an approximate value of it known and substituted in the functions which involve it .
In the experiments to be described these functions have values not differing greatly from unity .
A sufficiently close approximation to the value of for this substitution may , in fact , be found by putting The effect of temperature on the dimensions of the rods is neglected , as it does not appear likely to mfluence the value of the conductivity found by so much as 1 part in 300 .
In the actual apparatus the heat was supplied to the rod by the passage of the electric current through a platinoid wire wound on an independent sleeve placed on the bar , and the temperatures were measured by the resistances of platinum wires wound on similar sleeve It is therefore necessary to determine the effect of the sleeves on the result previously given .
Effect of the Sleeves .
Let be the excess of temperature of the rod and that of the sleeve at points in a plane perpendicular to the axis of the rod , distant from the central transverse section of the sleeve , the isothermal surfaces bemg assumed , as in the previous calculation , to be planes perpendicular to the axis .
* See note , p. 382 .
The distribution of temperature throughout a rod of length , radius , and conductivity , which receives heat at a uniform rate per square centimetre per second through a strip of breadth of its curved surface at one end , and loses it at a correspondin strip at the other end and at no other point , is given by the equation , where is the distance of a point from the central transverse section of the rod , its distance from the axis , and and Bessel { unctions for unreal arguments .
The mean temperature over a strip of the surface of the rod of breadth , whose centre is from the central transverse section , is given by where is the total flow of heat .
If the heat were generated at a uniform rate throughout the material of the rod within the distance of the end , the flow of heat would be linear and the temperature at would be given by A comparison of the numerical values of the first few terms of the two series , in the case of the shortest rod used , shows that even in that case the error committed in assuming the surfaces in the rod to be plane is not so large as 1 in 500 .
Let be the areas of cross section of rod and sleeve respectively , their outer perimeters , and the conductivities of their materials .
The thickness and conductivity of the thin layer of olive oil between rod and sleeve are small enough compared to those of the rod and sleeve to allow the heat conducted through the layer parallel to the axis of the rod to be neglected .
The heat conducted across a length of the layer from rod to sleeve in 1 second is equal to ) We have then the following equations for the steady distribution of temperature in rod and sleeve : In rod . . .
( 5 ) In sleeye .
( 6 ) The last term of the second of these equations is small and may be neglected .
* Taking the temperature at the central transverse section to be zero , we have as the solutions of equations ( 5 ) and ( 6 ) , . . . . .
( 7 ) , ( 8 ) where is the flow of heat through the central transverse section of rod and sleeve , and .
If length of free rod equivalent in thermal resistance to the length of rod enclosed by sleeve , we have . . . . . .
( 9 ) To determine whether any further simplification is possible , we note that in the apparatus used the constants have the following values:\mdash ; centim. , sq .
centim. , to according to the rod used , sq .
centim. , to according to the temperature , to centim. , and centims. Hence lies between and The error troduced by neglecting in estimating , the length of free rod equivalent in thermal resistance to the length of rod enclosed by sleeve , is about 1 per cent. in the worst case , i.e. , that of shortest rod and lowest conductor .
The error produced in the ation of is about per cent. CONDUCTIyITIES OF METALS AND ALLOYS AT LOW TEMPF , RATURES .
Thus will vary between the limits and , having the former value in the case of the worst conductin rods at low temperatures , and the latter in the case of good conductin rods at high temperatures .
The errol in the estimate of , if we write , will even in the latter case amount to 3 per cent. , and the error in due to it will not exceed per cent. We have , then , the simplified equation or ( 10 ) The first term on the right is obviously the correction due to the increase of cross section at the slay , and the second that due to the ends of the sleeve , at which the flow does not immediately take advantage of the increase of section .
We have seen that in the aboye equation varies from to according to the rod under test , while yaries from at 10 to at 30 absolute .
Hence varies from to according to the rod , and from at 10 to at 30 absolute .
The variation of with temperature has , however , only a very small effect on the value of , as will be seen from the following table of values for and VALUES of in centimetres . .
17 .
50 For .
297 If for simplicity of calculation we take the mean value of as holding throughout the whole range of temperature for any single bar , we make an error in the calculation of which will not exceed per cent. even in the case of the worst conductor tested , LIPOWITZ'S alloy , and will be insignificant for most of the other materials .
Thus we have the following values for and of the formula ( 4 ) for , where distance from enclosure to near edge of first sleeve .
, where distance between the near edges of first and second sleeves .
, where distance between the near edges of second and third sleeves .
and were in every case made centim. Estimation of the Srncdl Difference of Temperature between the Rod and the Resistance Coils .
In addition to a slight difference between the temperature of the rod and that of the sleeve , , at the central transverse section of the sleeve , there is a further slight difference between the latter and that of the wire of the platinum resistance , which may both be calculated with sufficient accuracy for the present purpose by taking where centims. is the perimeter , centim. the thickness , the conductivity , of the layer of oil between the rod and sleeve , centims. the perimeter , centim. the mean thickness , the conductivity , of the layer of silk and shellac between sleeve and wire , and centims. the perimeter of the outer surface of the wire .
From the aboye we obtain Within the range of temperature covered by the experiments yaries from at absolute to at absolute .
Taking the mean value as sufficient for the present purpose , we find .
( 11 ) In the work which follows , the mean temperature of the platinum coil wound on the sleeve so as to cover centim. of its length is taken as identical with the temperature at the centre of its length , the difference between the two being too small to influence the results appreciably .
Obtained from the minimum thickness , centim. , by the addition of a correction suggested by ' Electricity and Magnetism , ' I. , p. 280 .
The thermal conductivities are only known roughly , so that the final value of the correcting term is not likely to be very accurate .
Its influence on the values of the conductivities is , however , very small .
CONDUCTIVITIES OF METALS AND ALLOYS AT LOW TEMPERATURES .
391 Corrections due to the to the Platinurn Resistance Coils .
It has been stated ( p. 383 ) that only centims. of the total length , centims. , of platinum wire were actually wound on the sleeye ; the remainder , centims. , served to make connection to the No. double silk-covered copper wires leading to the resistance bridge .
Owing to its comparatively large cross section the whole of the copper wire may be assumed to have the temperature of the enclosure , but the temperature in each centim. of platinum wire will vary from that of the enclosure to that of the coil of which it is the continuation .
This will make the mean temperature of the wire as determined by its resistance slightly less than the temperature of that part of it on the sleeve , and will also lead to the conduction of a small amount of heat from the rod to the leads .
We have therefore to determine the magnitude of these effects .
If we measure along the platinum wire from its junction with the copper wire , we have for the temperature excess at the equation where is , as before , the temperature of the wire on the sleeve , and where is the perimeter , the cross section , of the wire , the conductivity of its material , and the external conductivity .
The mean value of is given by For the platinum wire used , measured over the silk centim. , ( mean ) , sq .
centim. , ; therefore , and centim. Hence and The resistance of the wire will therefore be the same as if , at each end of it , the length on the sleeve were increased by centim. and the projectin ends remoyed .
The effective length of the wire as used is , therefore , not the total length , centims. , but a shorter length , centims. Hence temperature excess of wire on sleeve times mean temperature excess of whole wire .
Hence , in the original equation for ( ( 4 ) , p. 386 ) , we must take and , where and are the mean temperatures over the whole of each platinum wire .
Hence we have as the connection between the temperature of the rod and the observed temperature of the platinum wires , mean value . . . . . .
( 12 ) The amount of heat conducted from the sleeve* along the wire at its point of contact with the sleeve is equal to the value of at , which is For the two platinum wires of the upper sleeye this amounts to , which must be deducted from the heat supplied to give the heat which flows down the rod to the thermometers .
Corrections for Leads in Power Circuit .
As in the case of the platinum resistance coils , so in that of the heating coil , the No. 40 platinoid wire constituting it was not wholly wound on the sleeve , but at each end centim. projected and served to make contact with the No. copper wires supplying current to the coil .
Part , therefore , of the difference of potential measured was expended on the leads and on that part of the platinoid wire not on the sleeye .
On the other hand , the passage of the current through the latter portion of the wire raised its temperature above that of the wire on the sleeye , and some of the heat generated in it would in consequence reach the sleeye by conduction along the wire .
We require to determine this amount .
If watts are spent in the platinoid wire , and is the total length of the wire , gram-degrees of heat are generated per second in each centimetre of the wire .
The equation for the distribution temperature excess along the wire in the steady state will therefore be , . . . . .
( 13 ) where have the usual meanings .
The solution of this equation may be put in the form , ( 14 ) * The amount conducted from the eve through the wooden support of the leads was too small to be taken into account .
CONDUCTIVITIES OF METALS AND ALLOYS AT LOW TEMPERATURES .
393 where is the temperature at the junction of the platinoid wire with the copper lead , that of the wire on the sleeye , is measured from the copper lead , and is the length of the platinoid wire between the copper lead and the sleeve .
If we neglect the heat generated by the current in the copper lead , and suppose the length of the lead to be sufficient for its temperature excess to fall to zero before the wire emerges from the containing tube , its temperature excess at a point from the junction with the platinoid is given by the equation where refer to the copper wire and have the usual meanings .
Since the flow of heat at the copper-platinoid junction is continuous , we must have , at i. e. , Hence The flow of heat out of the exposed platinoid wire into the sleeve is given by the value of at i.e. , by The total flow into the sleeve from the two wires will therefore be .
( 15 ) VOL. CCVIII.\mdash ; A. 3 For the platinoid wire the values of the constants are sq .
centim. , ( mean ) , centim. over silk , ( mean ) .
Hence , and For the copper wire , sq .
centim. , ( mean ) , centim. over silk .
Hence The { low of heat into the sleeve from the two wires per second will therefore be Thus of watts spent in the wire , only i. e. , ' reach the sleeve , where , is the excess of the mean temperature of the sleeve over that of the surrounding tube .
The watts were measured at the ends of the copper leads , for the resistance of which a further correction must be made , a correction which will depend on the temperature of the leads and therefore of the tube .
Each lead of copper wire , centim. diameter , was 55 centims. long , and of this length 33 centims. was in the Dewar tube , 22 centims. in air .
The resistance per metre at the temperature of the air , to C. , may be taken as ohm , and the total resistance of the leads as ohm at the temperature of the air .
When the platinum temperature of the Dewar tube is , we may , with sufficient accuracy , take the resistance of the leads to be ohm If , therefore , A amperes flow through the coil , and give a difference of potential of volts at the terminals , the watts spent in the platinoid wire Hence the watts expended on the sleeve , ( 16 ) where is the excess of the m-CONDUCTIVITIES OF METALS AND ALLOYS AT LOW TEMPERATURES .
395 It has been stated that the power supplied was measured by an ammeter in circuit and by applying a voltmeter to the ends of the heating coil for the few seconds necessary to take a reading .
Owing to the resistance of the voltameter , ohms at C. , not bein very great compared to that of the heating coil , ohms at C. , the measured value of the volts at the nds of the heating coil is less than the value when the voltameter is not in circuit , and the readings obtained must , if is the resistance external to the coil , and the electromotive force in circuit is constant , be multiplied by to give the true volts on the coil , i.e. , the volts on the coil .
Since the current was derived throughout from three storage cells kept well charged , it will be sufficient in the factor involving to write , and we have volts on coil .
Hence the watts spent on the sleeve Hence , the total quantity of heat imparted to the bar per second , is given by the equation* Or , since very nearly , Hence .
( 17 ) In the small term it will be sufficient if we write Thus , for the original equation , p. 386 , we must substitute * For simplicity , the loss of heat along the wires of the upper temperature measuring sleeve , p. 392 , is treated as if it took place from the heating sleeve .
Hence in the equation ( 4 ) for ( p. we must write for and for We thus obtain Finally , substituting for the temperatures of the rod and in terms of and the observed mean temperatures of the platinum resistances by means of equation ( 12 ) , p. 392 , we have that is , kq , ( 18 ) where .
centims. Effect of the use of the Variable State of Temperature Distribution .
So far the theory has been worked out on the assumption that the temperature distribution throughout the apparatus is steady .
But it has been stated that the measurements were made during the gradual rise of temperature from that ofliquid air .
We have therefore to consider the effect of this on our fundamental equation ( 4 ) .
Let be the temperature of the bar at a cross section distant from that where the bar joins the enclosing tube , and let V be the temperature of the tube .
Then satisfies the conditions , at end of the bar , the heat being supposed imparted to the bar at its free end .
CONDUCTIyITIES OF METALS AND ALLOYS AT LOW TEMPERATURES .
Let , where is independent of , and satisfy the following conditions , at at - , , at at then obyiously satisfies the conditions laid down for .
The solution of the equations for is Hence when Since is independent of , we have at any instant for all values of with inite - with zero Since , however , cannot be observed at the same instant with zero and finite , temperatures are obseryed alternately with zero and finite , and the distribution with finite compared with the distribution with zero calculated for the same instant by interpolation between the observations with zero made immediately before and after .
In all but three or four cases it was only necessary to take the mean of the observations with zero taken before and after .
Since is a continuous function the error committed by this method of observation will be very small .
of Platinum Thermometers .
As the observations of conductivity were made over the range of temperature between about C. and C. , the temperatures selected for standardising the thermometers used were the boilin point ofliquid oxygen , the freezing point and the boiling point of water .
The two thermometers were slipped on to a short rod of brass 2 centims. in length and of the same diameter as the rods used throughout the work .
The rod and thermometers were then placed in a short brass tube about 3 centims. long , one end of which was closed by a brass plate .
Through a cork in the other end of it the thermometer leads and a pair of compensating leads similar to those of the thermometers passed to a series of mercury cups , which allowed either of the thermometers or the compensating leads to be connected to a resistance box .
The coils of this box were of platinum silver wire , and were multiples of the B.A. ohm at C. The fixed arms of the resistance bridge were the 10- and 1000-ohm coils of the box respectively , and any error in the ratio of these two was eliminated by substituting a l-or 2-ohm standard resistance coil for the thermometers and finding its apparent resistance .
This enables the box readings to be converted to true ohms .
The following is a record of one of the tests .
Hartmann and Braun 2-ohm standard correct at C. Temperature coefficient .
Temperature of room , C. Hence standard ohm .
Standardleads apparent ohms of box .
Leads , , ohm Standard ohms , , ohms Other tests agreed to within one part in 5000 .
The resistances recorded in what follows have all been reduced to standard ohms .
The current used in the bridge was derived from a pint Leclanche cell connected to the bridge through a resistance of 53 ohms .
The test of the resistances of the measuring coils at the low temperature was made with the arrangement shown in fig. 2 .
A is the short brass cylinder containing the measuring coils , the leads of which , and the compensating leads , pass through a cork closing the top of the cylinder .
is the bulb of the oxygen vapour-pressure thermometer* EFG used to determine the temperature on the constant- volume hydrogen scale of the liquid air in the Dewar vessel V. represents six short copper rods placed round and to keep the temperature of the liquid air uniform .
is a bulb containing crystals of potassium permanganate which on heating furnish the oxygen .
is the sealed tube through which the thermometer was exhausted before use .
is the mercury gauge to determine the vapour pressure of the oxygen condensed in the bulb B. The difference of level of the mercury in was initially centims. , and the height of the barometer centims. Liquid air having been poured into the Dewar tube V till it was almost full , and the opening of the tube plugged with cotton wool , the bulb was heated so as to generate the oxygen , which condensed in B. The vapour pressure indicated initiaHy * I adopted this method of standardisation on the stlggestion of Dr. M. W. TRAVERS , F.R.S. , and found it most convenient .
CONDUCTIVITIES OF METALS AND ALLOYS AT LOW TEMPERATURES .
399 was centims. , corresponding , according to TRAVEBS ' determinations , to C. absolute , and rose during the first 40 minutes to centims. , corresponding to a temperature of C. During the next 16 minutes the following measurements were taken:\mdash ; Elliott B.A. box , ratio arms 10 : 1000 .
11 52 manometer . . . .
centims. 11 54 compensating leads ohm ( apparent ) .
11 56 lower coil . . .
, , 1158 upper , , . . .
, , 12 manometer . . . .
centims. 12 2 upper coil ohm ( apparent ) .
12 lower , . . .
0.8295 , 12 6 compensating leads , , 12 8 manometer 4251centims .
Barometer centims. , temperature C. From these observations we deduce that , at a mean pressure of centims. of mercury at C. , or centims. of mercury at C. , at sea-level in latitude under which oxygen boils at C. absolute on the constant volume hydrogen thermometer , the resistances of the coils were : Upper coil apparent ohm true ohm .
Lower , , , , , , , The determinations of resistance at the freezing point of water were made by placing the brass tube containing the coils at a depth of about 5 centims. in clean ice shayings contained in a brass can with holes in the bottom to allow the ice to drain .
After the apparatus had stood for half an hour the following observations were made :\mdash ; 1 compensating leads ohm ( apparent ) .
1 2 lower coil . . .
ohms 1 4 upper , , , , , 1 8lower , , . . . . .
, , 1 10 compensating leads ohm From these observations we deduce that the resistances of the coils at the melting point of ice were :\mdash ; Upper coil apparent ohms true ohms .
Lower , , , , , , , , In the resistance determinations at the boiling point of water the coils with the brass rod through them were placed in a narrow brass tube 12 centims. long , the top closed by a cork through which the leads passed , and the whole placed in a hypsometer with the lower 11 centims. of the tube exposed to the steam .
Owing to the uncertain behaviour of the shellac on which the insulation of the coils depended in the presence of moisture at this temperature , the coils were well dried by being kept at C. for about an hour and a half before the following measurements were taken : \mdash ; 3 40 compensating leads ohm ( apparent ) .
3 42 lower coil . . . . .
ohms 3 44 upper , , , , , , 3 48lower , , . . . . .
, 3 50 compensating leads ohm Barometer centims. at C. centims. at C. at sea-level in latitude From these measurements we conclude that the resistances of the coils at a temperature of C. Upper coil apparent ohms true ohms .
Lower , , , , , , , , Although the slight excess of the resistance of the upper coil over that of the lower coil is not proportional to that of either coil , it is so small in amount that it will be sufficient in defining the platinum temperature scale for the coils used and in reducing it to the standard hydrogen scale to use the means of the observed resistances at each temperature .
We thus have for the platinum temperature corresponding to a resistance of either coil , . . . . .
( 19 ) and as the basis of the connection between the platinum and hydrogen scales the following : Temperature on hydrogen scale .
Resistance .
C. absolute C. . . . .
ohm .
, , C. ohms .
In order to compare the platinum used with that used by other experimenters , the ' ' fundamental constants\ldquo ; of CALLENDAR are calculated as follows:\mdash ; ' ' Fundamental coefficient\ldquo ; ' ' Fundamental zero\ldquo ; *CHAPPUIS , ' Travaux et moires d Bureau International , ' , p. 108 .
CONDUCTIVITIES OF METALS AND ALLOYS AT LOW TEMPERATURES .
401 The temperature on the platinum thermometer corresponding to C. or C. absolute on the hydrogen thermometer is The S of the difference equation ( 20 ) is then given by , The platinum wire is , therefore , nearly identical in properties with that used as a standard by DEWAR and FLEMING , the fundamental constants of which CALLENDAR has calculated , and appears to be rather less pure than that used by me in the determination of the heat conductivities of some substances which are electrical insulators at low temperatures .
The connection between the two temperature scales ( 20 ) , and between the conductivities according to the two scales , ' . .
, , ( 21 ) is shown in the following tableS:\mdash ; TABLE of Connection between Temperatures and Thermal Conductivities on the Platinum Scale used and the Hydrogen Scale . .
Phil. , p. 100 ( 1895 ) .
'Phil .
hIag 47 , p. 213 ( 1899 ) .
'Phil .
Trans , 204 , p. 451 ( 1905 ) .
S Compare CALLENDAR , 'Phil .
Mag 47 , p. 214 ( 1899 ) .
VOL. CCVJI . .
3 of Resistance Bridge .
The standardisation of the resistance bridge used in determining the resistances of the temperature coils during a conductivity experiment was effected by substituting standard coils of , and ohms for one of the temperature coils , and adjusting the bridge to a balance as in the experiments .
The standardisation of each of the resistances of the dial used in the measurement of the difference of the resistances of the two platinum coils , from which the difference of temperature of the two points and of the test bar is calculated , was effected by substituting for it a standard coil of ohm by Hartmann and Braun , and examining the effect of the substitution on the balance of the bridge .
The resistance , ohm , found by adding together the resistances of the ten coils found in this way , agrees very well with the value found by direct comparison of the dial with a Wolff -ohm standard after the experiments were C0lnpleted .
In the tables of observations which follow , the values of the resistances given are the corrected ones .
ndardisation of Ammeter Voltmeter .
The Weston milliamperemeter No. 5491 , which was used throughout the work to measure the current through the coil in which the heat was generated , was standardised by comparison with a Kelvin balance which had been checked by the copper voltmeter .
It was found to read 1 per cent. too high .
The values of the current given in the following tables are the corrected values .
The Keiser and Schmidt moving-coil voltmeter was compared by a potentiometer method with a cadmium and a Clark cell , either separately , in series , or in opposition .
The electromotive force of each cell had been determined by comparison with a standard Chrk .
It was found to read per cent. too high .
The values given in lowing tables are corrected values .
of the oefficient " " \ldquo ; of Heat the Surface of the Rod .
The final for the thermal conductivity of the material of the rods , equation ( 18 ) , p. 396 , involves a knowledge of , the heat lost in 1 second from 1 sq .
centim. of surface of the bar when that area is 1o C. hotter than the tube .
As the terms involving only affect the result to a small extent , an approximate value of the quantity is sufficient .
To determine this value , the heatin coil and upper temperature measuring coil were placed on a brass rod centims. long , so as to di vide the rod into three equal CONDUCTIyITIES OF MTALS AND ALLOyS AT LOW TFMPERATURES .
403 parts .
This rod with its coils was suspended in the centre of the tube .
A short brass rod 2 centims. long was fitted into the base of the tube , and on the end projecting into the tube the lower temperature measurin coil was placed .
The coils were connected up as in an experiment to determine conductivity .
On sending a small current through the heating coil the temperatm.e of the rod in the centre is raised , and when equilibrium between the electTical ) of heat and the loss of heat from the surface of the rod had been attained , the excess of tempel.ature of the rod ovel the temperature of the shortel one continuous with the ] tube was obseryed .
The temperature of the latter and the supply of power to central rod were also measured .
To get rid of the effect of the slow change of temperature of the enclosure the heatin current was then switched from the rod to the enclosure , and the difference of temlJerature between and enclosure measured after the steady state had been attained .
The same series of observations was made at about , 20 and 30 absolute temperature .
The total quantity of heat , to the bal.pel .
second by the beating coil is given in gram-degrees by equation ( 17 ) , , as where amperes sent through coil , volts at ends ofleads , absolute temperature of enclosure .
excess of temperature of rod over that of enclosure .
If is the total surface of rod and sleeves , and the coefficient of loss surfaces of rod and sleeve be assumed to be the same , since both were slightly oiled surfaces , we have Hence Now for sides and ends of sleeves , , , , , , , rod , sq .
centims. ( 12 ) , p. 392 , and ( 19 ) , 1 ) .
where is the mean temperature of the platinum temperature coil on the rod , its resistance , that of the coil in thermal contact with the enclosure .
Thus C. , .
( 22 ) The table on p. 405 gives the observations and their reduction .
Although the determination of the general law of variation of the coefficient with temperature is not necessary for the purpose of the present work , it may be of to out that the observed values of for the three values of the mean temperature of rod and enclosure measured on the hydrogen tenlperature scale do not agree with any of the formulae which have been proposed to express that variation .
LORENZ* ) oposed the fornlula , but WAGNER has recently found it unsuitable and gests h , while I found , where , suitable over aratively small range of temperature .
The present obseryations suggest the expression for the case of a small rod at low temperatures in an enclosure only a few , say 10 , degrees coolel than itself is the mean temperature of and enclosure on the hydrogen thermometer scale .
Results oConductivity In the bless of Observations and Results which follow , the values of the quantities observed are given after correction for errors , if any , of the observing instruments .
addition , a few of the most important quantities involved in the calculation of the conductivities are given , order to facilitate comparison of the different experiments with each other , and to enable errors in the calculations to be more readily detected .
* L. , ' Ann. der Physik , ' 13 , p. 582 ( 1881 ) .
, 'Beiblatter , ' 27 , p. 534 ( 1903 ) .
C. H. LEES , ' Phil. Mag 28 , p. 429 ( 1889 ) .
CONDUCTIyITIES METALS AND ALLOY AT LOW TEMPERATURES . .
CONDUCTIVITIES OF MFTALS AT LOW CONDUCTIVITIES OF METALS AND ALLOyS AT LOW TEMPERATURES .
409 VOL. CCVIII.\mdash ; A. 3 .
co .
co : * .
CONDUCTIyITIES OF METALS AND ALLOyS AT LOW TEMPERATURES .
CONDUCTIVITIES OF METALS AND ALLOYS AT LOW TEMPERATURES .
CONDUCTIVITIES OF METALS AND ALLOYS AT LOW TEMPLRATURES .
415 CONDUCTIyITIES OF METALS AND ALLOyS AT LOW ] VOL. CCVIII.\mdash ; A. 3 CAL OFESSOR C. H. LEES : THERMAL AND ELECTELECT CONDUCTIVITIES OF METALS AND ALLOYS AT LOW TEMPERATURES .
, OF IsfETALS AND ALLOyS AT LOW .
42 ] The results of the observations are embodied in the curves on the opposite page .
From the smooth curves drawn as nearly as possible through the points given by the observations , the table on p. 426 of values of the conductivities at various temperatul e been constructed .
Discussion of the Ihermal Results .
An inspection of the CUl.ves of results over the temperature range C. to C. , to which for comparison the results obtained by JAGER and DIESSELHORST* at C. and 10 C. have been added , shows that over the whole range from C. to C. the conductivity of a pure metal changes comparatively little .
As to the character and magnitude of the small rates of change of conductivity with temperature at C. , the two sets of experiments are in close agreement throughout .
As the temperature rises there is a slight decrease in the conductivity of zinc , copper , iron , cadmium , tin , and lead over the range covered by the expel.iments .
The absolute values of the conductivity at C. , given by the present iments , agree closely with those given for these metals at this temperature by JAGER and DrESSELHORST , except in the case of cadmium , for which the present value of at C. is about 2 per cent. less than that obtained by JAGER and DIESSELHORST .
Silver appears to have a maximum conductivity at about C. , and to decrease very slightly in conductivity up to C. The value of at C. , according to the present experiments , is about 3 per cent. less than that of JAGER and fiSELHORST .
Steel shows a slight tendency to a maximum about C. , the change with temperature at C. agreein with that found by JAGER and DIESSELHORST , although the absolute value at C. is about 5 per cent. higher than for their steel .
Aluminium has a minimum conductivity about C. , and a rate of change with temperature about C. , which agrees with that found by JAGER and , although the absolute value is about 4 per cent. higher than theirs .
The alloys tested all incl.ease in conductivity with increase of temperature .
In the case of manganine the absolute value and the rate of increase with temperature at C. agree with those given by JAGER and DIESSELHORST .
When the slight unavoidable differences in purity of the specimens used by JAGER and DIESSELHOBST and by myself are taken into account , the close agreement between the results obtained by methods which differ so materially from each other must be considered most satisfactory , for not only have we possible differences in chemical composition to contend with , but , according to the researches of BEILBy , for each * W. JAGER and H. DIESSELHORST , 'Wissenschaftl .
Abhandl . .
Phys.-Techn .
Reichsanstalt , ' 3 , p. 270 , ( 1900 ) .
In cases where they give two or more values for the same metal , that value has been used to which they attach the greatest weight .
'Phil .
Mag 8 , p. 258 ( 1904 ) ; 'Roy .
Soc. Proc , 76 , p. 462 ( 1906 ) , and 79 , p. 463 ( 1907 ) .
CONDUCTIVITIES OF METALS AND ALLOYS AT LOW TEMPERATURES .
VOL. CCVIII.\mdash ; A. 3 I PROFESSOR C. H. LEES : THERMAL AND ELECTRICAL CONDUCTIVITIES OF METALS AND ALLOyS AT LOW TEMPERATURES .
metal there is a hard and a soft state , with different physical properties .
If in the future a still closer agreement between the results obtained by different observers is demanded , it will be necessary to specify both chemical composition and microscopic structure of the materials tested .
The fact that the thermal conductivities of most of the pure metals decrease with increase of temperature , while those of all the alloys tested increase with increase of temperature , raises the question whether the increase with temperature observed in the case of aluminium both in JAGER and DIESSELHORST'S experiments and in the present ones , and in that of nickel in the present experiment , cannot be attributed to small amounts ( less than 1 per cent. ) of impurity remaining in the specimens tested .
While there is much to be said in favour of this in the two cases mentioned , it seems to me that the apparent existence of a maximum of conductivity of silver , and possibly of steel , and a minimum conductiyity of aluminium , shows that we must guard ourselves against stating that the thermal conductivities of all pure metals decrease as the temperature rises , and limit ourselves to the less general statement that the thermal conductivities of most pure metals decrease with rise of temperature over the range C. to C. , while those of all the alloys tested increase over the same range of temperature .
On comparin these results with those obtained previously for non-metallic substances , *mostly electrical insulators , it will be seen that on the whole metals and non-metals are affected in the same way by change of temperature , i.e. , both tend to conduct heat better at low than at high temperatures , while alloys conduct better at high than at low temperatures , and in this respect resemble glass , the only mixture of non-conductors of electricity which has up to the present been inyestigated .
SECTION 2.\mdash ; THE ELFCTRICAL CONDUCTIVITY ENTS .
The effect of temperature on the electrical resistivity of the material of each rod was determined by sending an electric current along the rod and through a standard small resistance in series with it .
The rod was supported on a pair of knife edges , 4 centims. apart , and the difference of potential between the two was compared with that between the potential terminals of the standard resistance by the method shown in fig. 4 .
is the rod , two short copper cylinders provided with slightly conical holes into which the ends of the rod were gently forced , are the knife edges which support the rod and serve as potential terminals , is the standard resistance of or ohm .
The current was supplied from a storage cell , and could be adjusted by means of the resistance LEES , ' Phil. Trans. Roy .
Soc , 191 , p. 399 ( 1898 ) ; , 204 , p. 433 ( 1905 ) .
42@ PROFESSOR C. H. LEES : THERMAL AND ELECTRICAL A second circuit , insulated from the first , consisted of a small resistance provided with potential terminals , a storage cell , and a resistance box R. One of the potential terminals of could be connected directly to one of the potential terminals of , or to one of the knife edges on the rod , while the other could be connected through a low-resistance galvanometer to the other terminal of , or to the other knife edge .
In either case the resistance was adjusted till no current passed through the galvanometer .
If is the resistance which gives a balance on the rod and that on the standard resistance , the resistance of the rod between the knife edges is equal to times that of the standard resistance , the electromotive force in each circuit being supposed constant during the observations , and the resistances of the cell and being negligible compared to R. In order to overcome the effects of thermo-electric currents the circuits were made through a " " thermo-electric key which , when up , cuts the cells out of but leaves the galyanometer in circuit .
On depressing it , the first operation is to break the galvanometer circuit , then make both cell circuits , and lastly re-make the galvanometer circuit .
The frame supporting the rods during the test is shown in fig. 5 .
It consists of four strips , , of sheet copper , 1 centim. wide , and centim. thick , insulated from each other by strips of mica slightly wider than the copper , the whole shellacked and bound together with shellacked silk thread .
The upper end of each strip is soldered to 30 centims. of copper wire of 054 centim. diameter , which connects it with the rest of the circuit shown in fig. 4 .
The outer copper strips , and , end in short lengths of flexible conductor , to the ends of which the copper caps , which fit on to the rod , are soldered .
The middle copper strips are bent at right angles at their lower ends and form the two knife edges and in which the rod is supported .
To insure good contact , the rod is bound to Fig. 5 .
the frame by a few turns of silk thread passing over it at each knife edge .
The temperature of the rod was measured at the centre of its length by a platinum thermometer , consisting of about 31 centims. of No. 40 silk-covered platinum wire identical with that used in the thermal conductivity observations , and having a resistance at C. of about ohms .
This was wound at the centre of a thin ring of mica which just fitted the rods .
It was made by winding round one of the rods , whose surface was oiled slightly , a very thin strip of mica which had been shellac varnished , binding wire round it and baking it .
On cooling , a ring which stood throughout the experiments was obtained .
The ends of the fine platinum wire CONDUCTIyITIES OF METALS AND ALLOyS AT LOW TLMPERATURES .
were soldered to the ends of two lengths each of 33 .
of double silk-covered copper wire , centim. in diameter , Q. In order to prevent of the thin wire , the copper wires rested in grooves in the edge of a small block of wood , , the other edge of which rested on the mica ring , and seyeral turns of thread were taken round ring , block , and leads .
The whole was shellac varnished and stoved .
The platinum thermometer formed one arm of a resistance bridge , the equal arms of nich each consisted of an ohm coil .
The adjustable arm consisted of two resistance boxes with units and tenths , and the dial mercury cup resistance ( p. 384 ) giving hundredths .
The current was supplied by a Leclanche cell through a resistance of 5 to 10 ohms , and the galvanometer was a small pointer galvanometer allowing a determination of resistance to ohm , and of temperature , therefore , to C. The frame , rod , and thermometer were enclosed in a brass tube centims. long , centims. outside diameter , with walls about centim. thick , lined inside with mica .
Around this tube a few turns of insulated manganine wire were wound .
Through this wire an electric current could be sent if at any time it was necessary to raise the temperature of the apparatus .
The rod was not concentric with the tube , owing to the space occupied by the aIne .
The space between rod and tube was filled with clean sand so as to facilitate communication of heat between the two .
The upper end of the tube , out of which the insulated leads passed , was closed by a split cork .
Method of Experimenting .
In making an experiment the following was the routine adopted:\mdash ; The rod was first lightly oiled at the centre of its length , and the lnica rin of the platinum thermometer slipped over it , the oil assisting the motion and excludin air from the space between rod and inner surface of ring .
The excess of oil was then wiped away .
The knife edges were then cleaned if necessary .
The ends of the rod were next inserted into the end caps and the rod then placed on the knife edges .
The threads holding rod in its place were then would round it , the rod moved slightly to ensure good electrical contact between it and the knife edges , and the rod and frame placed in the brass tube .
The sand was then poured in and the top of the tube closed .
The brass tube and contents were then placed in a straight Dewar vacuum vessel , 20 centims. deep , 4 centims. wide , and the top closed by a of cotton wool .
The various leads were brought to copper mercury cups at which connections were made to the circuits .
The temperature of the rod was determined by balancing the bridge , and the resistance in series with the rod was so adjusted that the values of ( fig. 4 ) necessary to balance when the supplementary resistance was connected to the rod and standard resistance respectively were convenient .
After a few minutes ' interval , during which any temperature differences in parts of the apparatus had time to disappear , the ature and the values of for balance were determined accurately .
Liquid air was then poured into the Dewar tube , and when the temperature had become steady , the observations of rature and of values of were repeated .
The liquid air was then poured out of the Dewar tube , and the temperature rose slowly .
During the rise several observations of temperature and of were made at intervals .
After three or four observations had been taken the emperature of the apparatus was raised or by sending a current for a few minutes through the manganine wire wound on the outside of the brass tube .
When the rate of rise of temperature was again normal , observations were again taken .
These steps were repeated till the original temperature of the rod was reached .
In no case did the value of the resistivity at the end differ appreciably from its value at the beginning of an experiment .
Reduction of the Electrical Conductivity Observations .
In order to obtain the temperature of the rod from the measurements of the resistance of the platinum thermometer and its leads , we require to know the resistance of the leads and the constants of the thermometer .
The resistance of the leads was determined by placing an equal length of the same wire alongside the leads and determining its resistance for various temperatures of the rod during one of the experiments .
The values thus found were subtracted from the resistances determined by the bridge in each experiment .
The constants of the platinum coil were found by comparison with the two coils previously used in the heat conductivity measurements .
The three coils were made in the same way of the same wire , and the new coil agreed throughout with the old " " \ldquo ; coil .
The constants used in the reduction of the resistance measurements to temperature in the former experiments were therefore used in the present case .
As it was not the practice to wait till the temperature was quite steady before observations of temperature and resistance were made , there may be a slight difference of temperature between the rod and thermometer .
Attempts were made to determine the magnitude of this difference by or lowerin the temperature of the zinc rod at about the same rate as was usual in an experiment , i.e. , 1o C. per minute , and comparing the results .
No difference of temperature of the thermometer so large as the experimental error , C. , could be detected between observations which gave the same yalue of the resistance of the rod with rising or falling temperature .
A rough calculation from the dimensions of the mica ring on which the .
thermometer was wound showed that the difference of temperature of rod and thermometer could not exceed one-tenth of the change of temperature of the rod in one minute .
Whenever in an experiment the rise of temperature has exceeded C. CONDUCTIyITIES OF METALS AND ALLOYS AT LOW TEMPERATURES .
431 per minute , a correction equal to one-tenth the rate of rise has been subtracted fi.om the observed temperature .
The determination of the resistance of the rods depends virtually on the measurement of the current through them and the difference of potential at two points 4centims .
apart .
As the rods are about centims. long , and the current was introduced through a strip about millim. broad , situated at the ends of the curved surfaces of the rods , the current may , to the degree of accuracy aimed at in the present paper , be assumed to be uniformly distributed throughout the cross section of the rod in the interval between the two knife edges at which the potential difference is measure If is the resistance found , the resistivity of the material , A the area of cross section of the rod , and the distance between the knife edges , we may therefore write * The distribution of potential throughout a rod of length and radius and resistivity , through which an electric current flows which enters and leaves by strips of breadth of the curved surface at each end , and is of uniform density at each strip , is given by the equation , where is the distance of a point from the central transverse section , is its distance from the axis of the rod , and is the Bessel function unreal lment .
Hence the potentials at the points on the surface at which the knife edges touch the rod differ from that at the central section by , or , if is the total current , .
In the apparatus used centim. Hence and the equation reduces to to within 1 part in 500 .
The above statement is therefore correct to 1 part in 500 .
If is the radius of the rod , and the distance apart of the edges at C. , at any temperature C. we have where is the mean coefficient of expansion of the material of the rod , and that of the coppcr frame holding it , between C. and C. The values of the mean coefficients of expansion of metals at low temperatures are only known in a few cases , but from these cases it does not seem likely that the value of the correcting factor exceeds 1 by more than 1 part in 200 , in any of the experiments .
Results of Electrical Conductivity Experiments .
The tables of resistivities at diflerent temperatures em the observations:\mdash ; Ag Density at C. Ag Density at C. Ag Density at C. Ag Density at C. Ag Density at C. Ag Density at C. Ag Density at C. COPPER ROD .
Soft-drawn High-conductiyity Copper .
Density at C. ivityohm p ' ALUMINIUM ROD .
Al Density at C. ivity , ntim cistivityohm .
ZINC ROD .
Pure Redistilled .
Density at C. CONDUCTIyITIES OF METALS AND ALLOYS AT LOW TEMPERATURLS .
433 CADMIUM ROD .
Pure Redistilled .
Density at C. TIN ROD .
Pure Tin ( KAHLBAUM ) .
Density at C. BRASS ROD .
Approx .
70 Cu , Density at C. VOL. CCVIII.\mdash ; A. 3 IRON ROD .
Fe Density at C. NICKEL ROD .
Ni Density at C. STEEL ROD .
approx. 01 .
Density at C. LEAD ROD .
Pure .
Density at C. PLATINOID Approx .
Cu , 15 Ni , Density at C. OF METALS ANY ) GERMAN SILVER ROD .
Approx .
Cu .
, 15 Ni , Density at , Resistivity , H. ohm scale .
centim. cube .
C. ALLOyS AT TEMPERATURES .
435 MANGANINE ROD : Cu , 04 Ni , 12 Mn .
LIPOWITZ ALLOY prox .
Bi , 25 Pb , 14 Sn , 11 Cd .
Density at C. .3 PROFESSOR C. H. rfHERMAL AND ELECTRICAL The results given in the foregoing tables are shown graphically in Plate 30 , in which the crosses indicate values obtained by observation .
To get a more compact diagram the scale has been varied to suit the substances tested , nd the words which follow the names of the substances on the diagram , e.g. , " " 1/ 10 scale indicate tha the values plotted in the diagram are that particular fraction of e.g. , 1/ 10 of the values observed .
The values found by JAGEf and DIESSELHORST at C. and 10 C. are shown by the broken lines to the right .
The lines for the same material- in the two cases are joined by a dotted line or by a bracket .
As the electrical resistivity of a metal is nearly proportional to the absolute temperature , i.e. , the electrical conductivity multiplied by the absolute teIhperature nearly a constant , and as this product is required in the discussion which follows , table of values of the quantity at a number of temperatures within the range covered by the experiments is given on p. 437 .
It has been calculated from the values of the resistances taken from the curves of Plate 30 .
Discussion of the Electrical Results .
A comparison of the values of the electrical resistivities obtained in the present experiments with those obtained at C. and 10 C. by JAGER and DIESSELHORST , shows that the agreement is not quite so good as in the case of the thermal measurements .
This is not likely to be due to errors of obseryation , as the electrical measurements admit of greater accuracy than the thermal .
It seems rather to point to the greater influence on the electrical properties than on the thermal , of small differences in chemical composition , or in physical structure .
This is in keeping with the results obtained by GRUNEISEN* in the cases of copper and iron .
The influence of impurity on the electrical resistivities of metals is , in fact , so marked , particularly at low temperatures , that a comparison of the resistivities found in the present experiments with those found by DEWAR and FLEMING ( D. and by RLINGH ONNES and CLAY ( O.C. ) , will serve as a test of the degree of purity of the materials used .
GnuNEIsEN , ' Ann. der Phys 3 , p. 43 ( 1900 ) .
MATTHIESSEN , 'Ann .
der Phys 122 , pp. 19 , 68 ( 1864 ) .
DEWAR and FLEMING , 'Phil .
Mag 36 , p. 271 , ( 1893 ) ; FIBMING , ' Proc. Roy .
Inst June , 1896 , p. 9 .
IERLINGH ONNES and CLAY , ' Comm .
' No. ( 1907 ) .
CONDUCTIVITIES OF METALS AND AT LOW TABLE of Electrical esistivities , and of the Ratio of Resistivities at C. and at C. in ohms per centimetre cube .
Copper . . .
Silvor . . . .
Aluminium Zinc . . . .
Cadmium Copper . . .
Silvor . . . .
Aluminium Zinc . . . .
Cadmium The values of the ratio in the case of DEWAR and FLEMING'S metals have been calculated from their tables of results with the help of CALLENDAR'S table , connecting their platinum temperatures with hydrogen temperatures .
From this table it follows on the strength of MATTHIESSEN'S rule that impurity raises the resistivity , that the materials used in the present experiments were , in some cases less and in some cases more pure than those used DEWAR and FLEMING .
In most cases the difference is not large , but in the cases of cadmium and tin the materials used have a considerably less resistivity , and in the case of iron a considerably resistivity than those used by DEWAR and FLEMING .
In the case of nickel and iron the ratios of the resistivity at C. to that at C. , are much larger than the corresponding ratios calculated from their observations .
The total amount of impurity present in each of these specimens is less than 1 per cent. , so that the influence of even small quantities of impurity seems to be very marked .
table of values of the of electrical conductivity into absolute temperature , given on p. 437 , shows a general tendency for the product to increase with of temperature for all the metals tested except iron and nickel , and these contained a little immpurity .
The increase is relatively small in the case of lead , * NICCOLAI'S resuIts , 'Accad .
Lincei Atti , ' 16 , p. 906 ( 1907 ) .
His resuIts at low temperatures have not been used , as there is indication that the temperature dings of the thermo-junctions used by him have becr !
reduced to the hydrogen scale .
OALLENDAR , hil .
Mag 47 , p. 214 ( 1899 ) .
CONDUCTIyITIES OF METALS AND ALLOYS AT LOW TEMPEPATURES .
439 which , however , was very free .
All the alloys show a decrease of the product as the temperature decreases .
A comparison of this table with the table of thermal conductivities give1l on p. 426 will show that there is some similarity between the two .
SECTION COMPARISON 0F THERMAL AND ELECTRICAL CONDUCTIyITIES .
In 1882 L. LORENZ showed by measuring both the therlnal and electrical conductivities of bars of a ) number of different metals , that quotient of the thel.mal by the electrical conductivity was not constant , as WIEDEMANN and had supposed , but increased with the temperature , so as to be nearly proportional to the absolute temperature , a result which he had anticipated on theoretical grounds .
His results did not attract much attention till the publication of RIECKE'S electron of conduction in 1898S and the confirmation of LORENZ'S theory for pure metals the temperatures C. and C. by JAGER and DIESSELHORST , in 1899 by a method which giyes the of the two conductivities directly .
Next year DRUDE'S electroll theory appeared , more recently H. A. LORENTZ has published one .
According to these theories the electricity and energy are supposed to be carried entirely by which move to and fro and come into collision with the molecules of the metal and with each other , and are assumed to have on average the same kinetic energy of translation as the molecules of a gas at the same temperature .
the first two both positive and negatiye movable , in the third the negative only .
On the former theories if we limit ourselves to two kinds of electrons caITying charges and where , and if are the nunlbers per cubic centimetre , .
velocities at a temperature their mean free paths , we have the thermal conductivity , and the electrical conductivity , wlJere is the mean kinetic energy of translation of a gas molecule at absolute temperature 1o * L. , ' Ann. der Phys 13 , p. 422 ( 1882 ) .
G. WIEDEMANN and R. FRANZ , ' Ann. der Phys 89 , p. 497 ( 1853 ) .
L. LORENZ , ' Ann. Phys 147 , .
429 ( 1872 ) .
S E. RIECKE , ' Ann. der Phys 66 , p. 353 and 545 ( 1898 ) .
W. JAGER and H. , 'Sitz.-Ber .
Akad .
Wiss .
Berlin , ' 38 , p. 719 ( 1899 ) ; ) der Phys.-Tech .
nstalt , ' 3 , p. 282 ( 1900 ) .
P. DRUDE , 'Ann .
der Phys 1 , p. 566 , 3 , p. 369 ( 1900 ) .
H. A. , ' Proc. Amsterdam , ' 7 , p. \amp ; c. Hence The value of , the charge on an electron , is known with a fair degree of accuracy to be electromagnetic units , * but the value of is still uncertain , as it depends on the number of molecules in a cubic centimetre of a gas .
temp The following table giyes a few of the values which have been found for by various methods and the values of of calculated from them : From this it appears that we are not yet in a position to calculate more than the order of magnitude of the constant which the electronic theories give as the value of In order to see how far the theoretical result agrees in other respects with the served facts , the values of at various temperatures for the metals and alloys tested have been calculated from the values of the thermal and electric conductivities given in the tables on pp. 426 and 437 .
The results are shown in the following table and in the curves on Plate 31 .
The COl.responding values calculated ffom JAGER and DIESSELHOBST'S experiments at C. and 10 C. are also given for comparison .
The mean of J. S. TOWNSEND'S ( ' Phil. Mag 45 , p. 133 ( 1898 ) ) , J. J. THOMSON'S ( ' Phil. Mag V. , p. 355 ( 1903 ) ) , and H. A. WILSON'S results Phil. ) .
KELVIN , 'Phil .
lIag 4 , p. 197 ( 1902 ) .
, 'Phil .
Mag 4 , p. 301 ( 1902 ) .
S J. H. JEANS , ' Phil. Mag 8 , p. 694 ( 1904 ) .
M. PLANCK , ' Ann. der Phys 4 , p. 566 ( 1901 ) .
H. A. LORENTZ , ' Versl .
Akad .
van Wet 14 , p. \amp ; c. ( 1906 ) .
J. J. THOMSON , ' Phil. Mag 5 , p. 355 ( 1903 ) .
H. A. WILSON , ' Phil. Mag 5 , p. .
This line is added since Professor TOWNSEND'S proof that when a gas is ionised by Rontgen rays there are only half so many positive as negative ions produced ' Proc. Roy .
Soc 80 , p. 210 , 1908 ) , seems to render it advisable to take into account the possibility of the atom in electrolysis having a charge UCTIVITIES OF METALS AND LLOYS AT LOW RATURES .
The curves show that there is a tendency for the values of the conductivity constant to collect in the neighbourhood of the value , within the range given by the electronic theories , and that this tendency is rather more marked in the values calculated from JAGER and DIESSELHORST'S results than in those calculated from the present experiments .
Lead , tin , and possibly cadmium give values independent of temperature over the wide range covered , and therefore support the statement made by L. LORENZ in 1882 .
Zinc , silver , copper and aluminium show a marked decrease of the constant as the temperature decreases .
Nickel , and possibly iron , show an increase as the temperature decreases .
This appears to be the characteristic behaviour of alloys .
The values for aluminium and copper are decidedly below , while those for nickel and iron are decidedly above , the value All the alloys give values above this , and , with the exception of brass , decidedly above .
This large value of the constant for alloys especially at low temperatures is in keeping with and extends the results obtained previously by GRUNEISEN , found that increase of impurity in copper and iron at ordinary temperatures increased the value of the constant .
It is necessary to bear this increase in mind in basing any argument on the values observed .
The high values for nickel and iron may , since both contained a little impurity ( about 1 per cent be ascribed to the impurity .
In the cases of the other metals , with the exception of aluminium , which contained a little impurity ( about 1 per cent it is not probable that any small amount of impurity present in them could have materially influenced the positions of the curves .
In the case of aluminium it seems possible that the curve at low temperatures may be a little high , owing to the small quantity of impurity present in the metal .
The only conclusion which can be drawn from the curves seems to be that the Lorenz law of constancy of the value of at different temperatures for each metal has only a restricted application , and that although the metals which support the law give nearly the same value , , for the constant , the other metals have values of the constant which differ from by amounts far outside the limits of experimental error .
These facts are not of such a character as to provide strong support for any of the electronic theories mentioned on p. 439 in their present forms .
A modified theory has been put forward recently by Professor J. J. to get over a difficulty he pointed out in the older theories , which made the energy , ' Ann. der Phys 3 , p. 43 ( 1900 ) .
It is worthy of notc th for lres of these InESSEN , ' Ann. der Phys 110 , p. 190 ( 1860 ) , found that the conductivities , and SCHULZE , ' Ann. der Phys 9 , p. 555 ( 1902 ) , that the thermal conductivitics , could be calculated from the relative of the constituents present .
added June 1.\mdash ; The variations of the values of for the same metals at higher temperatures will form the subject of a futurc communication .
] S J. J. TIrOMSON , ' Journ. Inst. Elect. , p. 455 ( 1907 ) , or ' Corpuscular Theory of hIatter , ' p. 86 .
CONDUCTIyITIES OF METALS AND ALLOyS AT LOW absorbed by the electrons in 1 gramme of a metal during an increase of temperature of 1o C. far exceed the observed specific heat of the metal .
In this theory the electrons pass from one doublet , consisting of a positively and negatively charged atom , to another similar doublet , which have arranged themselves along the lin of electric force , as in the old theory of electrolysis the molecules of the Grotthus chains were supposed to do .
If is the distance apart of the charges of the doublet , and is the distance between the doublets , THOMSON shows that is 9 times the valuo found on the former theory .
Since may vary in value from metal to metal , and in the same metal at different tempel.atures , this theory may by a proper choice of the laws of variation be brought into unison with the observed facts .
It is , however , open to doubt whether any theory which does not take into account the atomic differences of the various metals can give a satisfactory explanation of the whole facts .
The nearness to each other of the curves for metals belonging to the same chemical group , Plate 31 , suggests that atomic properties play their part in determining the behaviour of a metal to the flow of heat and electricity through it .
Electronic theories have regarded the free as the sole carriers of energy from molecule to molecule , and have therefore been unable to deal with the thermal conductivities of electrical insulators , in which there can be no free electrons .
Now the thermal conductivities of some substances which insulate electrically are as large as those of the worse conductin metals , , the thermal conductivity of quartz along the axis is , while that of bismuth is only to .
It seems , therefore , that a theory of heat collduction which ascribes it to free electrons alone must be a very partial account of it .
The fact pointed out on p. 427 , that the variation of the thermal conductivities of the metals with temperature is of the same character as , though less , the corresponding variations for electrical insulators , and that alloys vary in a similar to the only mixture ( glass ) of insulators yet investigated , both ) contention , that theories which depelld solely on free as carriers claim to be adequate expressions of the phenomena .
This inyestigation was commenced in Manchester , where a liquid .
plant ) available .
I have to thank Dr. E. C. C. BAILEY and the ities of Univel.sity College for kindly providing me with liquid air in London .
I have also to thank the Committee administering the Goyernment for Scientific Inyestigations for a large proportion of the funds out of which the ials used in the investigation have been purclJased .
temp er atur .
hy drog al .
B. .
Trans. , , vol. 208 , Plate 31 .
face page 440 .
Iemp er

