]\gt ; IX .
The Emissivity of Thin in Air .
W. E. AYRTON , F.R.S. , and H. KILGOUR .
Received July 2 , \mdash ; Read November 19 , 1891 .
[ PLATES 11-15 .
] I. General of the Experiments .
IN 1884 it was observed experimentally that whereas electric current required to maintain a thick wire of given material , under given conditions , at a given temperature , was roughly proportional to the diameter of the wire raised to the power three-halves , the current was more nearly to the first power of the diameter if the wire were thin .
When this difference in the viour of a thick nd a thin wire was first noticed it was regarded as quite unexpected .
But , as pointed out by one of us in the course of a discussion at a meeting of the Royal Society , the unexpected cter of the result was due to people having assumed that the loss of heat from radiation and convection per square centimetre of surface per 1c excess temperature was a constant for a kind of surface and independent of the size and shape of the cooling body , although as early as 1868 Box had drawn attention to the great difference that existed between the rate of loss of heat from unit area of a horizontal cylinder and per unit area of a sphere .
The interchange of heat between unit area of a body and the enclosure might be independent of the shape of the body as far as radiation alone was concerned , but it seemed nearly obvious that the cooling by convection must be ected by the shape of the cooling body .
The very valuable investigations that have been made on emissivity by Mr. MACARLANE , Professor TAIT , Mr. CROOKES , Mr. J. T. BOTTOMLEY , by Mr. ScHLEIER .
MACHER , had for their object the determination of the variation of the emissivity with changes of the surface and with change in the density of the gas snrrounding the cooling body , it was not part of these investigations to the change in the emissivity that is produced by change in the shape and size of the cooling body .
Indeed , so little has been the attention devoted to the large change that can be brought about in the value of the emissivity by simply changing the dimensions of the cooling body , that in Professor EVEBETT'S very valuable book on Units ancJ Physical Constants , the absolute results obtained by Mr. MACFARLANE are given as the " " results of experiments on the loss of heat from blackened and polished copper in air at atmosphere pressure and no reference is made either to the shape or to the size of the cooling body .
[ November 19 , 1891.\mdash ; Since this paper was sent in to Royal Society , a new 29.7.92 .
edition of this book has appeared , and , in consequence of a suggestion made to Professor EVERETT , the word " " balls\ldquo ; has been added after the word " " copper\ldquo ; in this new edition , as well as the following paragraph:\mdash ; " " Influence of Size .
" " According to Professor AYRTON , who quotes a table in 'Box on Heat , ' the coefficient of emission increases as the size of the emitting body diminishes , and for a blackened sphere of radius cm .
may be stated as ' ' .
" " The value in MACFARLANE'S experiments was 2 The laws which govern the loss of heat from thin cylindrical conductors have not only considerable scientific interest in showing how the shape of a body affects the convection currents , but they are of especial importance to the electrical engineer in connection with glow-lamps , hot-wire-voltmeters , fuses , \amp ; c. We , therefore , thought it desirable to ascertain the way in which the law of cooling for thick wires , which the diameter raised to the power three-halves , passed into the law for the cooling of thin wires , involving only the power of the diameter .
For this object the investigation described in the present communication was commenced at the beginning of 1888 .
A considerable number of preliminary experiments having been conducted for the purpose of arriving at the best conditions to be adopted in the investigation , we finally , in , 1888 , decided to measure the emissivity at different temperatures from nine platinum wires , having respectively the diameters of 1 , 2 , 3 , 4 , 6 , 8 , 10 , 12 , and 15 mils , or thousandths of an inch .
Wires of these sizes having been ordered from Messrs. JOHNSON , MATTHEY , and Co. , they were found , when received , to have the following diameters at C. : \mdash ; ; ; ; ; ; ; ; Throughout this paper we have given diameters of the wires both in mils and in millimetres .
It may seem unscientific to mix up nensions in thousandths of an inch with dimensions in centimetres , but , while it is convenient , for the purposes of comparison , to use one square centimetre as the unit of area in experiments on THERMAL EMISSIVITY OF THIN WIRES IN emissivity , it is also convenient to have some concrete conception of the wires spoken of .
Now , fine wires are practically known in this country as wires of 1 , , 2 mils , \amp ; c. , diameter , a wire of 1 mil diameter being , for example , the thinnest that has been practically used in the construction of electrical apparatus .
A better idea is , fore , obtained from stating that the diameters of wires are 1 , 2 , or 3 mils , than from saying that they are , or millim. Attached to several of the curves are the diameters of the wires expressed in millimetres .
These numbers are stated to four decimal places , but it would have been better to have given , as in the above Table , only three significant figures , this being the probable limit to the accuracy of the measurement of the diameters .
Suspecting that some of the results of published experiments on the currents required to fuse wires had been much influenced by the cooling action of the blocks to which the ends of the wires were attached , we started by making a calculation on the length necessary to give to our wires so that the loss of heat by conduction should not introduce any important error into the determination of the emissivity .
To do this it was necessary to calculate the distribution of temperature along a wire through which a steady current was flowing , and from which heat was lost by radiation , convection , and conduction , and it was further necessary to improve on the calculation one of us had published on this subject in the 'Electrician , ' for 1879 , by now taking into account the fact that the emissivity , as well as the thermal and electric conducting powers of the wire , were different at different points in consequence of the difference of temperature .
Such a calculation has not , as far as we are aware , been hitherto made , it having been assumed in all previous inyestigations that the effect due to the variation of the thermal and electric conducting power of the material with temperature , as well as the variation of the emissivity per square centimetre with temperature and with the diameter of the wire , could be ected .
Until we had completed the experiments described in this ) we ] of course only employ , in this calculation , values that we had guessed at as something near the truth for the emissivity of platinum wire for different diameters and at diffelent temperatures .
Hence , after the completion of the experiments , we took up the mathematical investigation again , substituting for the emissivity such a function of the diameter of the wire and the temperature of the point as we had experimentally found it to be .
The investigation by which we finally arrived at tlJe calculated distribution of temperature along the wire is given in S V. of the paper .
The rate at which heat was lost by any one of the wires was measured by the product of the current passing through it into the P. D. ( potential difference ) maintained between its ends , while the ratio of the P. D. to the current gave the resistance of the wire , and therefore its temperature .
As the variation of resistance with temperature of different specimens of platinum is known to differ , it was not considered sufficiently accurate to deduce the temperature of the wire experimented on by using some supposed temperature coefficient for platinum ; consequently the variation of resistance , with temperature of each piece of platinum wire employed , experimentally determined up to 30 C. The investigation , therefore , divided itself into two distinct parts , .
:\mdash ; , the measurement of the power required to be given to platinum wires of various diameters , so as to maintain them at val.ious tempelatures above that of the enclosure .
, the determination of the law connecting the resistance with the temperature for each piece of wire employed in Fig. 1 .
And the second investigation also consisted of two parts , since a very considerable time had to be finally spent in determining the errors of the thermometers that had been employed in measuring the variation of the resistance of the wires with temperature .
Fig. 2 .
For carrying out the first part .
of the investigation , each wire was placed horizontally along the axis of a water-jacketed cylinder shown in section ( fig. 1 ) , centims. long , centims. internal , and centims. external diameter , the interior surface of the cylinder being coated with dull lamp black .
A stream of water of constant temperature entering by the pipe at leaving by the pipe THERMAL EMISSIVITY OF THIN WIRES IN AIR .
at , flowed through the water jacket and prevented the surface of the cylinder from becoming warmed by radiation and convection from the hot wire under experiment .
Fig. 2 shows the water-jacketed cylinder in position .
For inserting the wire to be tested into the enclosure-cylinder , it was attached at its two ends to a lamp-blacked brass carrier ( fig. 3 ) , which could be slid , between guiding pins ( fig. 1 ) , along the bottom of the inner cylinder into a definite position .
The two ends of the wire were soldered to the tips , of two wires , 1 millim. in diameter , which were made of platinum-silver so to have but a small heat conductivity .
Two short pieces of pl tinum wire , nlillin ] .
in diameter , to be used as terminals for the voltmeter , were soldered with the slightest touch of solder to two points about 6 millims. distant from the ends , of the wire under test , it having been determined that at this distance from the termina]s the temperature would not be much lower than the averag temperature of the wire .
In the case of the two finest wires tested , having diameters of and millim. respectively , it was not found necessary to adopt this arrangement , and the ends of the wires were themselves used as the voltmeter terminals , as seen in fig. 4 ; the platinum-silver wires , , were , however , filed down quite thin to preyent the fine wire being cooled by heat being conducted away from its ends .
The shaded portion of figs. 3 and 4 represents an ebonite cap used to insulate one end of the wire , from the brass carrier To prevent draughts entering the enclosure-cylinder , each of its ends was closed with an ebonite plug , ( fig. 1 ) through two holes in which passed the platinum- silver wires , , used as the terminals for the main current , and the fine pIatinum wire used for measuring the P. D. Connection was made between these latter and the wires which led to the voltmeter itself by their dipping into two mercu1y cups ( fig. 2 ) , carried from the stand of the apparatus , and which were turned into position close to the holes in the ebonite plugs , after the latter had been inserted into position .
For the experiments on the variation of resistance of the wires with temperature which were conducted during May , June , and July , 1888 , the bwo ends of each wire were fastened to two thick rectangular copper bars , about 7 millims. thick by 20 millims. wide .
The fastening was effected by making a fine saw cut about 6 millims. deep in the end of the bar ( fig. 5 ) inserting the end of the wire to be tested into the and gently hammering t , he copper so as to grip the wire .
The copper was then heated in blowpipe , and solder having a higher melting point than 35 C. , run in so as to fill up all interstices .
These bars , , dipped into oil contained in an iron pot wrapped round with asbestos cloth to loss of heat , and were supported from a wooden cross-piece , which ] carried the thermometer , the guide for the stirrer , and itself guided the handle of the stirrer , the up and down motion of which was limited by two pins .
Four thermometers of different ranges were used in these tests , and each of them THERMAL MISSIVITY OF THIN IN AIR .
was subsequently compared with a Kew standard thermometer .
difficulty was introduced into the carrying out of these comparisons from the fact that it is , or , at any rate , was not three years ago , possible to obtain a standard thermometer from the Kew Observatory to read , say , from to C. , with a short wide chamber at the base in which the mercury expanded below 20 C. All that could be obtained was a long thermometer which had been carefully tested between C. and C. , and the remainder of whose tube had been simply calibrated for uniformity of bore .
The consequence was that when we desired to compare one of our thermometers reading say , from C. to C. , with the Kew standard thermometer , their bulbs were very far apart when they were both immersed in oil so that the top of the mercury column of each was just at the top of the oil ; secondly , whereas we had kept each of our thermometers with its bulb close to the wire whose resistance was being tested , and , therefore , at a fixed distance below the surface of oil while it was being used , the Kew standard thermometer had to be continually lowered further and further into the oil as the temperature rose in order that correct readings could be obtained .
By adopting , however , the following device , a satisfactory , although very laborious , comparison between the four thermometers and the Kew standard thermometer was finally carried out in the autumn of 1889 , by three of the students of the Institution , Messrs. MULLER , STEPHENS , and WIGHTMAN .
The thermometer to be tested was placed in a deep oi ] bath in exactly the same position , relatively to surface of the oil , as that in which had previously been used .
To obtain uniformity of temperature oughout the oil bath , the oil was kept constantly agitated by means of a stirrer driven by an electromotor , and further , the heat was applied not merely at the bottom of the bath but along the whole of its long vertical sides , which were covered with several layers of asbestos cloth to prevent the flames warming one part of the surface more than another .
The bulb of the thermometer was surrounded by a bobbin of wire , the resistance of which was very carefully measured for many dings of the thermometer .
Then the Kew standard thermometer was inserted in place of the thermometer to be tested , but now , as the temperature rose the coil and the standard thermometer were depressed together so as to keep the coil alwa .
surrounding the bulb of the thermometer , and so that the level of mercury in the thermometer tube was always only just above the level of the oil .
The tcmperatures were then read off on the standard thermometer which caused the coil of wire to have exactly the same resistances as before , and which were , therefore , the true corresponding with the readings of the thermometer to be tested .
In consequence of the unwieldy length of a Kew standard thermometer reading to C. , it wa necessary to use an oil bath 80 centims. deep for this experiment .
Much thought and labour was given to the piece of apparatus , and many devices were introduced into it which it is nob necessary to describe here , first because after we had completed this of the investigation we learnt that much more suitable standard thermometers might have been obtained from abroad than could be purchased from the Kew Observatory ; secondly .
because if one were making such an investigation again , one would use as standards of comparison the coils of accurately known resistances at high temperatures which can now be purchased , but which we were advised three had not then the necessary degree of accuracy .
of liesisfrmcc of We will now return to the main part of the investigation , and it , to be understood tcmperatures mentioned the temperatures as corrected by using the results obtained from the comparison of the thermometers with the standard thel.mometer .
The first point was to determine the variation of resistance with temperature of the copper bars , and the thick copper wires that went to the Wheatstone 's bridge .
The total resistance of the and wires at C. was ohm , and , as the tvlref ; were protected by an asbestos screen from the action of the heat , the total variation in the resistance of the copper bars and wires combined was very small , being only ohm between C. and C. This yariation was , therefore , almost negligible compared with the variation of the resistance of even the thickest platinum wire experimented on , for which the following are the results .
TABT I.\mdash ; Piece of the tintlll Wire that had been used in the Experiments on Emissivity .
10th , 1888 .
ameter 1 mils , or millim. Length unknown , but about centims. Although the exact length of the above wire was unknown , the results are not the less useful for giving the law of variation of resistance with temperature , for , as pointed out in SIII .
, " " Results of Emissivity Experiments the observations contained in Tables I. to V. were employed to give the ratios of the resistances of one of the wires at any two temperatures , and not the specific resistance at any one temperature .
Hence it was unnecessary to know the lengths of the wires used the experiments , the results of which we recorded in Tables I. to , and the lengths where mentioned are only approximate in some cases .
THERMAL EMISSIVITY ] THIN WIRES IN AIR .
The next piece of wire used was of the same diametel , and also bit of the actual wire that had been used in the emissivity experiments .
The length of this piece , however , accurately known .
TABLE II.\mdash ; Piece of tinum Wire that had been used in the Experiments on Emissivity .
May 14th and 15th , 1888 .
Diameter 14 mils , or 0356 millim. Length centims. Curve No. 1 , fig. 7 ( Plate 11 .
) 0.477 0.478 The equation connecting the resistance and the temperature of the specimens was determined in two distinct ways .
First , a ( No. 1 , fig. 7 ) was drawn connecting all the values of temperature and resistance obtained on May 14th and 15th , then three points on this curve were selected , , those corresponding ature . .
From these , on the assumption that , the resistance at temperature could be expressed in terms of , the resistance at the temperature .
by the equation the values of , and were calculated .
The values 0 .
ohnl t ( 1 ) .
bl LSSf } S. W. E. AYRTON AND H. KILGOUR ON Next , with the object of ling the magnitude of the error that would be introduced into the values of and by a small erroy having been made in reading the temperature , the preceding calculation was repeated on the assumption the second temperature instead of being C. was C. The values then obtained were 0.00354 . . . . .
( 2 ) .
( 000896 Thirdly , the method of least squares was applied to all the observations made on May 10th , and given in Table I. In this way there were obtained the values ( 3 ) .
Lastly , the method of least squares was applied to all the observations made on 14th and 15th , and given in Table II .
, the values thus obtained being 0 .
( 4 ) .
compare the values of , and obtained from the curve recording the observanions of May 14th and 15th with the values obtained by applying the method of least squares to the same observations , we may examine the value of the rence between , the resistance of the wire at C. , and , the resistance at C. ; using the values of , and given in ( 1 ) , we find using the values of , and given in ( 4 ) we have \mdash ; Again the ) coefficient of increase of resistance per C. , between C. and C. when obtained from the curve alone , is , whereas the mean coefficient per 1o between C. and C. , using the means of the values of , and given in ( 1 ) , ( 3 ) , and ( 4 ) , is We may , therefore , conclude that ib is not necessary to use the lengthy method of least squares to obtain the values of , and ) that the lues obtained by HERMAL SIVITY THIN AIR .
using three points on a curve which graphically records the results of the on temperature and resistance are accurate enough for practical purposes .
threepoint method was , therefore , alone adopted for the remaining .
Indeed , for the purpose of determining the temperatures of the wires in the emissivity experiments which corresponded with the yarious observed resistances ; it was found most easy and most accurate to simply read off the temperature at once from the curves which recorded the results of the experiments made on the variation of the with temperature for the particular wires .
Specimens of such curves are seen in cllrves 1 , 2 , 3 , 4 , 5 , 6 , figs. 7 and 8 .
( Plates 11 and 12 .
) We were , however , led to study the various formulae that had been published by different invesf , igators on the connection betw eenl the tance and temperatul.e of different metals , and this led us to carry out a considerable amount of calculation which brought to light some interesting results .
Since these results , however , are not specially connected with this present investigation on nissivity , we purpose presenting them in a separate communication later on .
TABLE III .
\mdash ; Piece of the Platinum Wire that had been used for the Experiments on Emissivity .
June 11th and , 1888 .
Diameter , mils , or millim. Length , centims. Curve No. 2 , fig. 7 ( Plate 11 ) stan .
' ' 0 .
0.455 ' The following results able IV .
) were not obtained with a piece of that been actually employed in the emissivity experiments , but with a piece of wire cut off the same reel from which was taken the wire used in the emissivity expel.iments .
Before , however , making the measurements given in IV .
, the piece of wire used was heated by a strong current , so as to be ightly luminous for inutes , in order to bring it to the same as that of the piece that had been used in the emissivity experiments .
bJFiSSRS .
W. E. AYRTON AND H. KILGOUR ON THE TABLE IV .
July , 1888 .
Diameter , mils , or millim. Length unknown .
Curve No. 8 , fig. 7 ( Plate 11 ) .
The curves recording the results given in Tables III .
and IV .
are shown in Curves 2 and 3 , fig. 7 .
TABLE V.\mdash ; Piece of Platinunl that been used in Experiments Emissivity .
July 27th , 1888 .
Diameter , mils , or millim. Length , centims. JULY 20th , 1888 .
Diameter , , or millim. Length unknown .
Resisbance , in ohms .
0.707 ' THEBMAL ] MISSIVITY OF THIN WIRES AIR .
JULY 26th , 1888 .
Piece of the preceding wire .
Length , centims. Temperature .
' JULY 12th , 1888 .
Diameter , 6 mils , or millim. Length , centims. Fig. 14 .
JULY , 19th , 1888 .
Diameter , 4 mils , or millim. Length , centims. Curve 4 , fig. 8 ( Plate 11 ) .
Temperature .
istanc .
' The smallness of the variation of resistar ) of the last wire with shows that it is not a pure platinum wire , and that it possibly contains a of silver or iridium . .
W. AYRTON AND H. KILGOUR ON THE 5th , 1888 .
Diameter , mils , or millim. Length , centims. Curve 5 , fig. 8 ( Plate 12 ) .
JULY 3rd , 1888 .
Diameter , 2 mils , or ] .
Length about centims. Curve No. 6 .
fig. 8(Plate 12 ) .
S84 JULY 11th , 1888 .
neter , millim. Length about centims. Tcmperatnrc ohms .
Resistance , in ohms .
THERMAL EMISSIVITY OF THIN WIRES IN AIR .
III .
Results of Emissivity Experiments .
To determine the temperature of any one of the wires used in the emissivity experiments from its observed resistance , we might proceed as follows : calculate the specific resistance and then determine from the preceding , 1 , 2 , \amp ; c. and 8 ) , the temperature at which a piece of the same wire had the same specific resistance .
To do this it would be , of course , necessary to know the lengths of the two pieces of the same wire used in the two sets of experiments .
Now , whereas in each of the emissivity experiments a straight piece of wire of about 28 centims. employed , the length of which could be measured with considerable accuracy , it was only possible to use shorter pieces for the experiments on the variation of resistance with temperature , for the latter pieces were of the wires that had actually been used in the emissivity experiments , and detaching a piece of wire from the apparatus seen in figs. 1 to 4 , and attaching it to the clamp shown in figs. 5 and 6 , necessarily shortened its length .
Since , therefore , there was this greater difficulty in measuring the exact lengths of the wires used in the experiments recorded in Tables I. to V. , curves 1 , 2 , \amp ; c. , figs. 7 and 8 , it was decided to regard these experiments as the relative resistances of each particular wire at different temperatures , but not the specific resistance at any particular temperature .
In order then to use these curves ofrelative resistance with , 2 , \amp ; c. , for the purpose of determining the temperature of the wires in the emissivity experiments , we must know the resistance of each of the latter wires at some one temperature .
This might have been ascertained by means of the Wheatstone 's bridge , but , as a relative resistance rather than absolute resistance was required , following method was employed instead , since it avoided the possible introduction of an error that might have arisen from some want of agreement in the unit of resistance of the Wheatstone 's bridge and the units of current and P. D. The ratio of P. D. to current , or the resistance of the wire , having been measured for a number of currents , a curve was plotted connecting the resistance of the wire for various currents flowing through it , and by continuing this curve until it cut the axis along which resistance was measured we obtained the limiting value of the ratio P. D. to current for current nought , that is , the resistance of the wire when at temperature of the enclosure .
Fig. 9 ( Plate 13 ) shows the curve thus obtained for the thickest wire , viz. , that of 14 llils , and similar curves were drawn for all the othel wires for the purpose of ascertaining their respective resistances when at the temperature of the enclosure .
Let and be the and diameter of the wire in centimetres , be the temperatures of the water jacket surrounding the enclosure cylinder ABCD and 2 ; be the temperature of the wire when A amperes are passin through it , and when V volts is the P. D. between the ends of the wire , be the emissivity , that is the NIESSR W. E. AYRTON AND H. KILGOUR ON THHi number of calories ( gramme C. ) lost per second per square centimetre of surface per 1o C. excess temperature , then 0 .
' and are not , however , absolute constants for a given wire , since as the wire gets hotter hotter its length and diameter become larger and larger .
If be the coefficient of linear increase of platinum per C. , which , according to KOHLRAUSCH , is , then allowing for the lengthening and thickening of the wire as it increases in temperature , we , if and be the values at C.\mdash ; Now may be regarded as equal to where is a correction that must be added to the obseryed excess temperature , the amount of this correction depending on the ] of .
The values of and the amounts of the correction for different values of are given in the following Table VI .
, and are shown plotted in fig. 10 ( Plate 13 ) .
The corrections are but small , but they have nevertheless been taken into account in determining the values of the emissivity given in the following Tables VII .
, VIII .
, and IX .
TABLE VI .
THERMAL EMISSIVITY OF THIN WIRES IN AIR .
Diameter , 14 mils , or millim. Length , centims. Water-jacket , C. May 3rd , 1888 .
Fig. 11 .
( Plate 14 .
) The first eight observations in the above ] were made to enable the early part of the curve in fig. 9 to be plotted , so as to find the resistance of the wire that corresponded with current nought ; and this the curve shows to be ohm .
wire , then , at bhe temperature of the water-jacket , viz. , C. , has this resistance , ohm .
MESSRS .
W. Hi .
AYRTON AND H. KILGOUR ON THE TABLE VIII .
Diameter , mils , or millim. Length , centims. Water-jacket , May 2nd , 1888 .
Fig. 11 .
As in the case of Table VII .
, the first nine observations in Table VIII .
were made only for the purpose of enabling the early part of the curve in fig. 9 to be drawn .
In the following Table IX .
, which gives the results for the remaining wires , the first eight or nine observations that were made in each case are omitted , since , although these observations were of value in enabling the early part of the curve , such as is shown in , to be drawn for each wire , the difference between the temperature of the wire and of the water-jacket was too small to enable the corresponding emissivities to be accurately calculated .
TABLE IX .
Diameter , mils , or millim. Length , centims. Water-jacket , C. April 28th , 1888 .
Fig. 11 .
THERMAL EMISSIVITY OF THIN WIRES IN AIR .
TABLE IX.\mdash ; continued .
Diameter , mils , or millim. Length , centims. Water-jacket , C. April 2 , 1888 .
Fig. 11 .
Diameter , 6 mils , or millim. Length , centims. Water-jacket , C. April 2 , 1888 .
Fig. 11 .
TABLE IX.\mdash ; continued .
Diameter , 4 mils , or millim. Length , centims. Water-jacket C. April l , 1888 .
Fig. 12 .
( Plate 14 .
) We have already drawn attention to the fact that the above wire , of 4 mils or millim. in diameter , has a much smaller variation of resistance with temperature than the other wires , and therefore that it is probable that this wire has some iridium or silver in its composition .
From the preceding table we that the variation of emissiyity with temperature is also much smaller with this wire than with the others .
THERMAL EMISSIVITY OF THIN WIRES AIR .
TABLE IX.\mdash ; continued .
Diameter , mils , or millim. Length , centims. Water-jacket , C. May 4th , 1888 .
Fig. 12 .
Diameter , 2 mils , or millim. Length , centims. Water-ja ket , C. April 4 , 1888 .
Fig. 12 .
bIESSRS .
W. E. YRTON AND H. KILGOUR ON THE TABLE IX.\mdash ; continued .
Diameter , mil , or millim. Length , centims. Water-jacket , April l , 1888 .
Fig. 12 .
Using the values of temperature and emissivity given in the preceding Tables VII .
, VIII .
, and IX .
, the curves shown in fig. 11 have been drawn for the five thicker wires , viz. , those having diameters of 14 mils , or millim. ; mils , or millim. ; mils , or millim. ; mils , or millim. ; and 6 mils , or millim. , respectively ; and the curves shown in fig. 12 have been similarly drawn for the four finer wires , viz. , those having diameters of 4 mils , or millim. ; mils , or millim. ; 2 mils , or millim. ; and 12 mil , or millim. , respectively .
The curves for the four finest wires on fig. 12 and for two of the thicker wires on fig. 11 , appear to be fairly regular throughout their whole length , while the curves for the 14 , the , and the mils wires appear doubtful for temperatures below C. On examining the curves ws see that : 1 .
For any given temperature the emissivity is the higher the finer the wire .
2 .
For each wire the emissivity increases with the temperature , and the rate of increase is the greater the finer the wire .
For the finest wire the rate of increase of emissivity with temperature is very striking .
3 .
Hence the effect of surface on the total loss of heat ( by radiation and convection ) per second per square centim. per 1o C. excess temperature increases as the tempera ture rises .
The wire of 4 mils diameter appears to be an exception to these rules ; but we have already seen reasons for that this wire was not drawn from pure platinum like the rest , but possibly from platinum-iridium or platinum-silver .
THERMAL EMISSIVITY OF THIN WIRES IN AIR .
On comparing the loss of heat from the wire of mi ] diameter when at 30 C. with that from the wire of 6 mils diameter when at C. , both being in an enclosure at C. , see that formel loses per square centimetre per second , not -or 58 times 15\mdash ; 10 as much heat as the latf , er ( which it would do if the emissivity were the same ) but instead , or 3480 times as much heat ; arising from the fact that the emissivity , that is , the number of calories ( gramme ) lost per second per square .
of surface per 1o C. excess temperature of the mil wire at 30 C. , is 60 times as great as that of the 6 mils wire at , the latter varying very rapidly with the temperature near C. From the curves on figs. 11 and 12 ) following table been drawn up giving the emissivities of the various wires at eight useful even temperatures .
TABLE X.\mdash ; Emissivities at\mdash ; The next step taken was to express mathematically the law connecting the emissivity of a wire with its diameter for a fixed temperature , and this was done for three fixed temperatures , viz. , 10 C. , 20 C. , and 30 C. Using the values of the ' emissivity contained in the three columns headed with these temperatures in Table X. we obtained:\mdash ; For 10 C. , . . . .
( 5 ) , , , 20 .
, . . .
( 6 ) , , , 30 .
, . . .
( 7 ) , where is the diameter of the wire in mils or thousandths of an inch .
The following Tables , XI .
, XII .
, and XIII .
, give the actual emissivities , the emissitives calculated by means of the three preceding formulae respectively , the differences , and the percentage differences .
MESSRS .
W. E. AYRTON AND H. KILGOUR ON THE TABLE XI.\mdash ; Wire at 10 C. TABLE XII.\mdash ; Wire at 20 C. TABLE XIII.\mdash ; Wire at 30 C. The values of the emissivity given in the last table are plotted in the curve shown in .
1@ ( Plate 15 ) .
The statement not unfrequently that the current required to maintain a wire of a given material at a given temperature above that of the surrounding envelope is proportional to the diameter of the wire raised to the power three halves , is equivalent to stating that the emissivity is independent of the diameter .
Now if we may THERMAL EMISSIVITY OF THIN WIRES IN AIR .
assume that the three formulae ( 5 ) , ( 6 ) , 7 ) given above for may be used not merely for the platinum wires when fine , but also as giving at any rate a rough approximation of the value of when the wire is as much as 4 or 5 millims. thick , we conclude that for a temperature of 10 C. the value of in the must be something of the order 220 mils , or , so that the neglecb of the second term may not make an error in of more than 5 per cent. , and something uf the order inch , or millims. , if the error is not to exceed 1 per cent. ; that a.temperature of 20 C. , the value of in the formula must be something of the order 244 mils , or millims. , so that the neglect of the second term may not make an error in of more than 5 per cent , , and } , .
of the order inch , or millims. , if the error is not to exceed 1 per cent. ; and that for a temperature of 30 C. , the value of in the formula must be something of the order 267 mils , or millims. , so that the neglect of the second term may not make an error in of more than 5 per cent , and something of the order inch , or millims. , if the error is not to exceed 1 per cent. Generally , then , although it may be possible to obtain only very rough approximations of the values of the emissivities of thick wires by using the three formulae that we have deduced from the experiments on wires , still it follows that to assume that the emissivity is a constant for wires whose diameters vary from a small value up to 1 inch is to make a large error in the case of the greater number of the wires , and an error of hundreds per cent. in the case of some of them .
The formulae ( 5 ) , ( 6 ) , ( 7 ) given above have been calculated from the results of experiments made on wires varying from to 14 mils , and the method of calculation employed makes the percentage difference between the obseryed and calculated emissivity very small at each end of the range as well as in the middle .
We may , therefore , use the formula to obtain some idea of what the emissiyity is likely to be for a wire somewhat ] than that used in the experiments , say , of 075 mil in diameter .
Using formula ( 7 ) to obtain the emissiyity at 30 C. , we find it to be .
We can now make an approximate estimate of the current density , or amperes per square centimetre , it would be necessary to employ with a platinum wire of 075 mil in order to keep it at a temperature of 30 C. , when the enclosure was , say , at C. From the tables that we have given of the resistance of ] atinum wire at different temperatures , we see that the resistance of .
of wire millim. in diameter is about 2 at C. Hellae the ance of 17 centims. of wire mil .
, or millim. in diameter , will be about 2 , or ohms .
If , therefore , be the required current density , the rate of production of heat in calories ( gramme-centigl.ade ) per second will be which must equal the number of calories lost per second , viz. , , or therefore , amperes per square centimetre .
If the wire , instead of being platinum , had been made of copper , and if the emissivity for a copper wire , mil in diameter , were the same as that of a platinum wire of the same diameter , the current density that would be required to keep the copper wire at a temperature of 30 C. , when the enclosure was at C. , would be about that is , 122,600 amperes per square centimetre , or 790,500 amperes per square ) .
[ It may be useful to give the general formula which we have arrived at for .
the current A amperes , required to be sent through any long thin platinum wire mils in diamete , to maintain it at a temperature of 30 C. when the enclosure ( having about the dimensions shown in fig. 1 ) is maintained at a temperature of C. From the various ements given in the paper it follows that the resistance per cubic mil of platinum at 30 C. is about ohm .
Hence , equating the rate of production and loss of heat per mil length of the wire , we have , therefore , about .
Also , if , he wire having the conditions described above be inches long , the watts that must be expended on it equal oximately ( THERMAL EMISSIVITY THIN WIRES IN AIR .
IV .
Comparison of the Results toith those by MM .
DULONG and PETIT .
All the emissivity in figs. 11 and 12 ( Plate 14 ) have the same general shape , being concave to the axis of temperature , and it is interesting to see how far this general shape agrees with the results obtained in the classical researches of MM .
DULONG and PETIT for the loss of heat from thermometer bulbs in air .
The formulae they developed for the loss of heat by radiation and con vection in air lead to the following expression for the emissivity:\mdash ; where is the temperature of the enclosure , the excess temperature of the cooling body , a constant having the value if the temperature be measured in degrees Centigrade , the pressure of the gas , a constant depending on the nature of the surface of the cooling body and of the enclosure , and a constant depending mainly on the nature of the gas surrounding the cooling body and but very slightly on the nature of the surface of the cooling body .
Substituting the value of and expanding in powers of we have .
If be used to stand for the expression in the brackets and for , calculation shows that and have the following values for the different values of Values of And using for and the values found by Mr. HOPKINS for polished limestone cooling in air at 760 millims. pressure , contalned in an enclosure at C. , the expression given above for becomes hence , substituting the values given in the last table for , we have of Consequently , if the law connecting the loss of heat from thermometer bulbs in air with the difference of temperature between the bulb and the enclosure is of the same nature as that connecting the loss of heat from very fine horizontal wires in air with the difference of temperature , we should expect to find that the curve connecting the values of and in the last table would , when plotted , be everywhere concave to the axis of .
But this is not the case , for we find on plotting this that while it resembles our curves in being concave to the axis of , for values of less than about 20 C. it changes its curvature at about this point and becomes distinctly convex .
It is , of course , to be remembered the maximum value of in the experiments of MM .
DULONG and PETIT was 24 C. , while in some of ours exceeded 30 C. We are , however , inclined to attribute the inability of the formulae of MM .
DULONG and PETIT to give even the general shape of the curves which we have obtained to the fact that the convection which occurs with thermometer bulbs hardly suggests the very great convective cooling that experiments show to occur with very fine wires at high temperatures.\mdash ; May 31 , 1892 .
] V. Calculation of the Distribution of Temperature along Platinum Wire Heated by Electric Cur.rent .
Let be the diameter of the wire in centimetres .
, , temperature in degrees centigrade at any point of the wire distant centimetres from the nearer of the two supports to which the ends of the wire are attached .
, temperature of the supports .
, , electrical resistance , in ohms , of a cubic centimetre of the wire at a temperature / thermal resistance of a cubic centimetre of the wire at a temperature THERMAL EMISSIVITY OF THIN WIRES IN AIR .
be the emissivity at a temperature for a wire of diameter , , number of calories corresponding with one watt-second .
, , current in amperes flowing through the wire ; then following the method of reasoning employed with such problems , we have ( 8 ) .
is in reality the reciprocal of the amount of heat in calories that would flow per second across one cubic centimetre of the material for 1o C. difference of temperature between the opposite faces and for a mean temperature of the material of is the number of calories lost per second on account of radiation and convection for 1o C. excess temperature from a square centimetre of the platinum wire of diameter and at a temperature In the preceding equation is strictly the diameter of each part of the wire at the particular temperature it is at .
As , howeyer , an increase of the temperature from C. up to 30 C. only increases the linear dimensions of platinum by about per cent. , may be taken as the diameter of the wire at C. and are known from the various curves connecting resistance with temperature and emissivity with temperature for each of the various wires experimented on .
The variation of the thermal resistance of platinum with temperature has not , , as far as we can learn , been examined , nor does it appear to be even known whether the thermal of platinum increases with temperature , as does the thermal lesistance of iron , or diminishes with the rise of temperature , as does the thermal resistance of copper and German silver .
Under bhese circumstances we decided to assume that the thermal resistance of platinum was a constant , and had the value it known to possess at ordinary temperatures , and to see what sort of result a mathematical assumption based on this result would lead to .
As a matter of fact , eyen the thermal conductivity of platinum at ordinary temperatures is not stated ] icitly in books , but it can be easily arrived at indirectly .
For we find that WIEDEMANN and FRANZ determined that and in the article " " Heat\ldquo ; in the ' Encyclopaedia Britannica , ' the value of the thermal conductivity of copper , as determined b NGSTRoM , is stated to be trustworthy .
From these numbers we deduce that the thermal resistance of platinum , at ordinary tempel.atures , is , which is the value we have taken for .
Equation ( 8 ) can then be written in the form ( 9 ) where .and and are therefore constants for a particular wire with a particular current flowing it .
We next tried to expand and as functions of from the curves which we had experimentally obtained connecting emissivity with temperature and resistance with temperature , bub we found that it was not possible to express these functions of in any such simple shape as would allow the integration of equation ( 9 ) to be effected analytically , and a result obtained suitable to be used for easily determining the value of for any value of .
We , therefore , ulted Professor HENRICI regarding the integration of equation ( 9 ) in a practical and we have to express our thanks to him for the warm interest that he has taken in this mathematical problem , and for the many suggestions that he has kindly made , and which have enabled us to arrive at the following solution .
We have also to thank one of our assistants , Mr. WALKER , for carrying out the graphical and numerical calculations contained in this section of the paper .
It is clear that the law of distribution of temperature along the wire will depend on the diameter of the wire among other , also on the current passing through it , the variation of temperature with llength of wire being the more ra , pid the thicker the wire and the greater the culTent passing through it .
We therefore selected for consideration a wire of mean diameter , namely , that of 6 mils or and we took the case when ampere was passing through it , which is the greatest current that was passed through this wire in the emissivity experiments .
Having selected this wire and current , the next step consisted in calculating and for different values of t. is C. , is 25874 , and the value of may be obtained from the emissivity curve for the 6 mils or millin ] .
wire given in fig. 11 .
Instead of calculating the resistance per cubic centimetre of the material for different values of ( t ) it is more convenient to write as , where is the length , centims. , of this 6-mil wire that was used in the experiments from the results of which the curve in fig. 14 ( Plate 15 ) has been drawn , and THF , RMAL EMISSIVITY OF THIN IN AIR .
is the resistance in ohms of this centims. of this 6-mil wire at any rticular temperature , .
Now , when A equals ampere , equals 0152 millim. , and is centims. , ; therefore , the equation ( 9 ) becomes .
To obtain a curve representing the first term on the right-hand side of this equation , the curve for , given for the 6-mil wire in fig. 11 , was altered by simple geo- metrical construction so as to give a curve for .
The ordinates of this curve were then altered , graphically , so as to give to a convenient scale a curve for , and , lastly , the ordinates of this curve were reduced , graphically , in the proportion of to To obtain the curve representing the second term on the right-hand side of equation ( 10 ) , the ordinates of the curve on fig. 14 were multiplied by 14917 , and then the curve was re-drawn to the same scale as the curve representing .
The ordinates of the two curves were then subtracted from one another , and a curve , having for its ordinates the difference of the ordinates of the last two curves , drawn , which gives the values of for any value of .
This curve is seen in fig. 15 ( Plate 15 ) , it is parabolic in shape , cuts axis along which temperature is reckoned at equals 31 C. , and has its vertex approximately in the line along which is reckoned .
A similar investigation was made for the same wire for a current of ampere , and it was found that the curve for in this case cut the axis of temperature at C. The fact that is nought at a particular temperature tells us , of course , mathematically nothing about the actual value of , but from our general knowledge of temperature curves , we know that when a current is passing through a fine wire , as in our experiments , the temperature will rise rapidly along the wire in the neighbourhood of the supports , then rise more slowly , and at no great distance from the supports the temperature curye will become nearly flat , and will be absolutely flat over the middle portion of the wire .
We are , therefore , justified in assuming that and are at about the same point of the wire .
On examining the numbers in Table V. , which refer to the 6-mil wire , it will be not , iced that the mean temperatures of the wire for currents of and ampere respectively , C. and C. , which are almost exactly the NDcccxcn .
for which are for these two currents in question .
It is certainly a calculation based on the assumption that the thermal resistance of platinum is the same at all temperatures between C. and C. as it is at ordin ary tures , should have led to the result that , both for .current of mpere and a current of ampere passing this 6-mil should be nought , that is , the temperature curve should be flat at almost exactly the mean temperature that the wire had in each case .
It would , therefore , appear that the assumption regarding the thermal resistance of platinum having a constant value at different temperatures has not introduced any serious error .
If we assume that the vertex of the parabolic curve for is in the line along which is reckoned , which is very nearly the case , then ( 11 ) .
is minus the value of when equals nought , and is the value of when equals nought .
At the point of the wire where the temperature is the highest the temperature curve will be flat , that is , will be nought , and at the end of the wire where it is attached to the support the temperature will have some definitive value , , in other words when and when these two conditions only and without any reference to the length of the wire , it is possible to integrate the equation ( 11 ) , and the lesult we arrive at is ( 12 ) .
An examination of this equation shows that in order that may equal must equal infinity , and , therefore , in obtaining this integral we have tacitly assumed that the maximum temperature of the wire is only obtained at an infinite distance from its end .
But while great simplicity is obtained by this hypothesis , no more error is introduced by its employment than is met with by the use of the ordinary equation for the conduction of heat , and which leads to the result that two bodies initially at a different temperature take an infinite time to arrive at thermal equilibrium , even when connected together by a good conductor of heat .
On examining the curve for on , which as already stated has been THERMAL EMISSIVITY OF THIN WIRES IN AIR .
calculated for the 6-mi ] wire for the maximum current used , viz. , ampel .
, we find that .
The temperature of the enclosure in the ivity experiments with this wire was C. , and as the temperature of the point of the wire where it was joined to the support would be somewhat higher than this , we may take as about C. It is to be noticed that the higher this temperature actually was the more uniform must have been the temperature of the whole wire and the more accurate will be our experiments on emissivity .
Substituting these values for and in equation ( 12 ) , we have , and the following are the values of obtained for the corresponding values of The actual length of is 6-mil wire between the points of support was centims. , and the maximum temperature was C. Therefore , as the calculated temperature is .
at a distance of centim. from the support , it follows that in pe cent. of the wire , reckoning from the support , the temperature rises to about 95 per cent. of the value it has at the middle of the wire ; while at a distance of centim. from the support , that is , in about 5 per cent. of the whole length , the erature has reached about 99 per cent. of its value at the middle .
The calculated distribution of temperature along this 6-mil wire , when traversed by the largest current employed with this wire , is shown in fig. 16 .
The distance between the voltmeter wires was 27.46 centims. , a , s indicated in the figure ; and it is with the distribution of temperature over this portion of the wire that we have to deal in the emissivity experiments .
Of this length , about 2570 centims. is seen to be at a temperature a little over 31 C. ; while 176 centim. has a mean temperature .3 of C. The error , therefore , which we have made in assuming that the whole length of was at 31 C. ; not serious .
And , fortiori , the lengths which were employed with all the finer wires in the emissivity experiments and the position of the voltmeter wires , were such that the temperature was practically constant over all the length of the wire used in the emissivity calculations .
In the case of the wires thicker than 6 mils , the error arising from the conduction was necessarily greater ; and the calculations that we have made on the distribution of temperature along the 6-mil wire have led us to the conclusion that , if we were going to repeat the experiments , we should , in the case of the thickest wires employed , attach the voltmeter wires at about 2 centims. from the supports .
We do not , however , thinuk that any serious error has been introduced by attaching them only about 7 millims. , except , perhaps , in the case of the largest currents with the largest wires .
When plottin the curve for given in fig. 15 , it was noticed that this curve , where it cuts the axis along which is reckoned , was not strictly parabolic .
suggested that an idea might be obtained of the error ntroduced , by assuming the lower portion of this curye to be parabolic , if a tangent were drawn to the true curve for at about equals 31 C. , and if calculations were made on the assumption that this tangent line were itself the curve for Doing this , we obtain the following equation by integration for the temperature curve ( 13 ) where , as before , is C. , and where is minus the value of the ordinate of this tangent when equals nought .
When measured , is found to be 2388 .
THERMAL EMISSIVITY OF THIN IN AIR .
Solving equation ( 13 ) for various values of , we find If now a distribution of temperature curve be drawn , similar to that seen in fig. 15 , but using these values of and instead of those found from equation ( 9 ) , it is found that the two curves showing the distribution of temperature along the wire are not very different .
We may therefore conclude that the exact shape of the curve for for small values of , has but little effect on the resultant distribution of temperature curve , and therefore that no important error is made by assum i as we have done , that the true curve for is a parabola , with its vertex on the line along which .
is reckoned in fig. 15 .
@ 8 7.0 Nranh \amp ; .
Plata 13 .
Fig 9 \ldquo ; Amperes ff to for \mdash ; / c ) t ) / c ) t ) / c ) t ) / c ) t ) / c ) t ) / c ) t ) / c ) t ) / --\mdash ; - c ) t ) / --\mdash ; - c ) t ) / --\mdash ; - c ) t ) .
1892 .
-1A ' fflgour .
aeae- * / .
/ / .
1 .
\fnof ; m

