]\gt ; The Heating Eflect of the in Precise surements of Electrical Resistance .
By B. T. GLAZEBROOK , F.R.S. , W. R. BOUSFIELD , K.C. , and F. .
SMITH .
( Received July 29 , 1911 .
) ( the National Physical Laboratory .
) In a paper on the specific heat of water , * one of us describes an experiment in which the passage of a current of amperes through an oil-cooled manganin resistance of .
wire produced an increase in the resistance corresponding to an increase of temperature of C. As the cooling surface of the resistance was 16 sq .
cm .
per watt , such a large increase of re was thought to be improbable , and the change of resistance was attributed to some other cause .
However , the effect of the passage of the current was similar to that resulting on raising the temperature , and , because of this , it was proposed to call the change a thermoid effect .
With platinum wires similar results were obtained , the increase of resistance nearly proportional to the square of the current and inversely proportional to the radiu of the wire .
The change of resistance with varying current was inyestigated by Dr. E. H. Griffiths in 1893 a fine platinum wire , he found the increase of resistance to be proportional to the square of the applied voltage .
With to this increase , Dr. Griffiths remarks:\mdash ; " " It seelned absolutely immaterial whether the current was on for only a few seconds or indefinitely Previous experiments on an alloy of copper , manganese , and nickel had shown that the resistance of the alloy changed appreciably in course of one and a-half hours when a current of ampere was passed through it , and Dr. Griffiths states that the rise appeared to be a function of the current rather than of the time .
If , in Dr. Griffiths ' determination of the mechanical equivalent of heat , he had neglected the rise of resistance of the wire with increasing current , an error of about 1 part in 60 would have been introduced .
The question as to whether the rise of resistance is due entirely to increase of temperature or due in part to some other effect of the current is of extreme importance for precise electrical measurements .
Moreover , since in many measurements it is necessary to use large currents , it is desirable to " " The Specific Heat of Water and the Mechanical Equivalent of the Calorie at Temperatures from C. to ' Phil. Trans , vol. 211 , pp. 199\mdash ; 251 .
' Phil. Trans , 1893 , pp. 361\mdash ; 504 .
VOL. LXXXV.\mdash ; A. Messrs. Glazebrook , Bousfield , and Smith .
[ July 29 have some easy means of calculating the rssistance for any current which may be used .
The change of resistance due to the heating effect of the current may be calculated if the temperature coefficient of resistance and the thermal emissivity are known .
Our experiments show that the thermal emissivity is a funcbion of the temperature of the wire , but for a small range of temperature the change is small .
If we take the mean emissivity over a range of temperature not greater than , the rise of temperature due to a current may be calculated with a fair of accuracy by means of equation ( 2 ) which follows , it being assumed , of course , that the rise does not exceed If resistance of wire , length in centimetres , radius in centimetres , current , difference of temperature between the wire and its when a steady temperature results , mean thermal emissivity of the surface in calories per second , 's equivalent , we have , ( 1 ) or .
( 2 ) The maximum variation of temperature over a cross-section of a wire has been shown by Dr. A. Russell* to be where is the specific resistance and the thermal conductivity .
In the case of the anin coil used at endon , and we may take , so that for a current of 5 amperes the difference of temperature between the axis and outer skin of the wire was of the order of C. In all that follows , we assume , therefore , that over a cross-section of a wire the temperature is uniform and that no strain is likely to occur because of the temperature gradient .
The usual form of expression for the change of resistance with temperature of a resistance wire is , ( 3 ) where and are constants .
When the surroundings of the wire are maintained at a constant temperature and a current is passed through it , the temperature of wire is raised , and it follows from ( 2 ) that its resistance may be wriften in the form , ( 4 ) where } , and and have the same values as before .
If there be some other effect of the current which increases the resistance , Theory of Electric Cables and Networks , ' p. 216 .
11 .
] Heating in Measurements of then an additional term must be added to ( 4 ) , which would be of the form , where is a constant .
To decide whether or not this is so , we must be able to calculate , which involves a knowledge of the thermal emissivity .
The most direct way of determining the latter is to measure the temperature of the wire while the current is passing through it .
Such a measurement could not be made by means of thermo-junctions , as any arrangement of such would not only alter the conditions of cooling near the junction , but it might also be equivalent to increasing the diameter of the wire .
One of us has shown that in the case of mercury resistances the temperature and resistance can be measured simultaneously , forming a very convenient mercury-resistance thermometer .
The temperature is measured by the expansion of the mercury , and the temperature of a wire can be measured in a very similar way .
Our arrangement for measuring the expansion is shown in fig. 1 .
The wire experimented with was of iron ; it was cm .
long , 0.02 cm .
in diameter , and had been carefully annealed by passing a comparatively large current through it for one hour , the current being then gradually diminished in value so as to ensure slow cooling .
The wire I is connected to two copper leads and is stretched taut by means of a phosphor bronze wire which passes over an axle , 4 mm. in diameter .
The wire passes round the axle three or four times and is then soldered to it .
The wheel is cm .
in diameter and the spring ensures Messrs. Glazebrook , Bousfield , and Smith .
[ July 29 , that the wires are always stretched taut .
A small mirror is secured to the axle and reflects a beam of light so as to produce an image at a distance of 1 metre .
is an insulating support of ebonite .
When I increases in length , is pulled upwards , and the amount by which the point is raised is tically magnified about 1000 times .
In general , the amount by which is raised is greater than the increase of length of I. If the angle is nearly the mechanical nific tion is considerable .
However , we found it more convenient to make about l50o , in which case the mechanical is about 2 .
Our estimate of the total magnification was 2000 , which means that for the cm .
length of wire we used , an increase of temperature of 1o should produce a deflection of mm. The apparatus was calibrated by immersing it down to the point , first in cold and afterwards in hot water .
The wire quickly took up the temperature of the water and the remainder of the apparatus was regarded as constant in temperature .
The results are given in Table I. The means thus devised of measuring the temperature of a wire during the passage of a current through it rendered it easy to measure its resistance and temperature for any value of the current .
The change of resistance with temperature could be determined by a separate series of observations and the validity or invalidity of equation ( 4 ) established .
It was of course important to ensure that in any adopted for measuring the resistance of the wire the other resistances involved did not , or , if so , changed by known amounts .
Dr. H. Griffiths , in the paper already referred to , describes several arrangements for measuring the 1911 .
] in Measurements of Electrical Resistance .
545 change of resistance with increasing current .
Another ement , in which a mercury-resistance thermometer is employed , has also been described by one of us and has been used in making the observations at Hendon .
Fortunately .
at the National Physical Laboratory such standards of resistance are in use as enable a simple Wheatstone bridge to be used with manganin resistances in three of the arms , of such dimensions as to exclude error from the cause mentioned .
In general , the bridge consisted of two ratio coils of manganin , having nominal values of 10 and 1000 ohms respectively , a heavy current manganin standard of ohm , and the wire under observation .
All of the resistances were immersed in a bath containing 7 gallons of well-stirred paraffin oil maintained at a constant temperature of or of C. When a balance was obtained ( by the 10 or 1000 ohms coils ) the same current passed through the ohm and the wire ( usually of 1 ohm resistance ) , and since , in all of our measurements , the current density in the ohm standard was small , we assumed there to be no appreciable change in the istance of this standard .
The results of our observations show that we were justified in assuming this , but as a check we also used ratio coils of 10 and 10,000 ohms , and a heavy current manganin standard of ohm in series with the wire .
When the wire had a nominal resistance different from 1 ohm the 10,000 ohms was replaced by a variable resistance .
At the same time as the resistance was measured the temperature of the wire was indicated by the deflections of the spot of light .
For values of the current less than 1 ampere no readings of the deflection were taken ; for higher currents several readings were taken , as it was found that the of observation might amount to three or four degrees .
The , values of , and the increase of temperature of the wire as thus recorded are given in Table II .
Messrs. Glazebrook , Bousfield , and Smith .
[ July 29 , Afterwards the change of resistance with temperature of the wire was determined , and these results are plotted on the chart ( fig. 2 ) .
If , when a current is passed through a wire , the rise of resistance is due to the heating effect only of the current , then for a given resistance the temperature as deduced from the resistance-temperature chart should agree with that obtained from the expansion of the wire .
Table III gives the rise of temperature obtained in the two ways The values of obtained in the two ways being in such close agreement , and the small differences being sometimes positive and sometimes negative , we conclude that the heating effect of the current was the sole cause of the observed increase of resistance .
However , we think it desirable to add that when electrical measurements are made with wires not carefully annealed , the passage of a current through them does often produce permanent changes in the resistance .
In addition , therefore , to raising the temperature of a wire , a current may so alter the molecular structure as to change , temporarily or permanently , the specific resistance .
Very similar effects may be produced by a cycle of temperature , and it is at present difficult to say whether or not the char ) in the specific resistance is due entirely to the change of temperature of the wire .
With well-annealed wires the changes are small .
It is now possible to calculate , the thermal emissivity of the surface of the wire .
For the iron wire experimented with this with the temperature , the values being as follows:\mdash ; Heating in Measurements of Electrical Resistance .
On the diagram fig. 2 ) the values of and have been plotted in addition to the values of and temperature , the scales of temperature and being so chosen that for all practical purposes a single curve connects and and and .
If and are plotted the resulting curve may be made identical ( by suitably choosing the scale of ) that connecting and The above-described experiments with the iron wire having established the fact that the increase of resistance is substantially due to the increase of temperature , we give the results of some measurements on other wires .
In two of our experiments fuse wires were used .
The resistances , lengths , and radii of these were measured , and the melting points were determined in hot glyceriue .
Portions of the wires were then placed in an electric circuit , and the current gradually increased until they fused .
This Messrs. Glazebrook , Bousfield , and Smith .
[ July 29 , was done in air as well as in well-stirred paraffin oil .
The results are as follows:\mdash ; In the case of wire No. 2 the meltin point is above that of the boiling point 19 C. ) of the oil used .
This may explain the large mean value of in oi .
These experiments show clearly that very large temperature differences between the wire and the oil may exist , even with a wellstirred batb .
Other experiments were made with wires of manganin and of platinum silver , and curves were plotted connecting resistance and temperature and resistance and .
When the scale of temperature was suitably chosen the resistance-temperature curve was in each case practically coincident with the curve .
The mean value of was of the order The experiments hitherto ecorded were made at the National Physical Laboratory .
We now come to others , most of which were made at Hendon .
Observations were made on pairs of wires of similar materials , No. 1 of each pair being about cm .
diameter and No. 2 being about cm .
in diameter .
The wires were bare and immersed in well-stirred paraffin oil , the current gradually increased from to 3 amperes .
In the case 1911 .
] Heating in Measurements of Etectrical of the platinum wires and of one copper wire , currents of and 4 amperes were also employed .
Table gives the comparative resistances , the temperature of the bath being practically constant .
For all of the above , curves connecting and the resistance have been plotted and in all cases practically straight lines result .
The temperature coefficients of resistance are given in Table and these , together with the curves enable the relation between , the rise of temperature , and to be found .
The mean value of for the range of temperature has been calculated from the equation where is the length of wire having a resistance In Table we give the diameters of the wires , the temperature coefficient of resistance , the resistance per unit length at C. , and the mean value of the thermal emissivity .
It will be observed that is greater for the small wires than for the larger ones ; we have already seen that it increases slightly with the temperature .
The mean value is about twice that observed at Teddington with wires of iron , manganin , and platinum silver .
The cause of this may be partly due to a difference in the paraffin oils , but we believe if ; to be principally due to the fact that much more vigorous stirring is employed at Hendon .
The importance of uniform stirring must not be overlooked .
Dr. E. H. Griffiths tried the effect of variations , and found , as would be expected , that the more perfect the stirring the less the wire became heated , but within the limits of stirring employed in his experiments on the mechanical equivalent of heat , no correction was thought to be necessary .
Messrs. Glazebrook , Bousfield , and Smith .
[ July 29 , In the case of copper wire No. 1 , the following results were obtained with a current of 3 Resistance .
Normal stirring Very vigorous stirring No stirring These figures seem to show that the normal stirring employed at Hendon was vigorous , and that the value of resulting from such stirring is as great as would be likely to be realised in ordinary practice with bare wires immersed in paraffin oils .
In the case of wires covered with silk and shellac , the heating effect is greater than with bare wires , and is best found experimentally by measuring the variation of resistance with temperature and of resistance with current .
We now briefly indicate the importance for precise electrical measurements of the heating effect of the current .
For such electrical measurements as those involved in the determination of the mechanical equivalent of heat , it is clearly of importance , especially when coils with large temperature coefficients of resistance are used .
Dr. E. H. Griffiths , who employed a platinum wire for the heating , allowed for this change of resistance , and pointed out the importance of determining the resistance for various currents .
In Dr. determination of the mechanical equivalent the current was measured by measuring the potential difference on a coil of ] atinum silver through which the current flowed .
coil was immersed in wellstirred oil , and Dr. Barnes believed the increase of temperature of the wire to be less than C. While this may be possible , it is of interest to point out that in Dr. Barnes ' experiments the maximum current which passed through one platinum-silver wire was 1 ampere .
Each wire was about 100 cm .
long , cm .
radius , and 4 ohms in resistance .
If these values are substituted for , and in equation ( 2 ) , we find that the rise of temperature , is equal to .
If the cooling effect of the stirring was the same as at the National Physical Laboratory the value of is , equivalent to an increase of the resistance of 6 parts in 10,000 .
If , on the other hand , the cooling effect of the stirring corresponded to the most vigorous stirring up at Hendon , the rise of temperature would be C. , equivalent to parts in 10,000 of the resistance .
For the co1Ilparison of standards of resistance the heating effect- of ths current is sometimes overlooked , especially when a comparison of methods is made .
For instance , it is generally thought that in bridge methods for the 1911 .
] Heating in Measurements of Electrical Resistance .
comparison of resistance much larger currents can be used than in the potentiometer method , because in the bridge methods the currents pass through the resistances for a few seconds only .
In general , however , this is a mistake , for the temperature of a resistance coil in oil rises to within a tenth of a degree of the maximum temperature within a few seconds of the circuit being closed .
If the maximum increase of temperature of the wire is and if is the increase of temperature seconds after making the circuit , we have , ( 5 ) where is the density of the material of the wire , and is the specific heat .
Experiments were made on a manganin wire mm. in diameter , and the resulting values of were in air and in oil .
The value was about .
For a current of ampere the maximum elevation of temperature of the wire in air was , the actual temperature being .
By means of equation ( 5 ) we have calculated the time in seconds , after closing the circuit , corresponding to various values of .
The times and temperatures of the wire are given in Columns 1 and 2 of Table As a check on these values we measured the change of resistance of the coil ( by observing the galvanometer deflection ) during the first minute of the passage of the current of ampere , and from the resistance values we estimated the temperature with the aid of a resistance-temperature chart .
These times and temperatures are recorded as " " observed\ldquo ; values , and are given in Columns and 4 .
In oil a current of about amperes was required to raise the temperature to , but since is about twenty times the value in air , the rate of increase of temperature is much greater .
The final temperature of is reached within in less than 5 seconds after closing the circuit .
Messrs. Glazebrook , Bousfield , and Smith .
[ July 29 , APPENDIX.\mdash ; By W. BousFlELD , ( Received September 20 , 1911 .
) It is a matter of interest to determine the exact relation of the aetual effiect to the radius of the , but the experiments previously described were not sufficiently rable qse to enable this to be done .
Differences in stirring produced notable differences in the results , and even differences in the winding of the wire on the mica plate were material .
To bring these variable elements under control , the wire was soldered to two heavy copper electrodes and lightly stretched itudinally in a glass tube , through which flowed a current of cooling water .
Owing to the velocity of the water and to the fact that the wire was led into and out of the main tube by side tubes at right angles , the flow was necessarily turbulent .
Some transverse vibration of the wire produced by the water current also tended to mere gliding flow .
The diameter of the tube was cm .
and the rate of flow ( except in one experiment ) was 100 .
per second , giving a mean velocity in the direction of the length of the tube of about 135 .
In the subjoined experiments the wires used were the soft platinum wire of cm .
diameter which had been used for a previous experiment in the stirred oil bath , and a series of smaller wires drawn from a portion of the same wire so as to be completely comparable as ards lltateria The for the experiment on the first wire of the series are iven below .
In the bridge arrangement , , the resistance to be measured , was in series with a mercury thermometer resistance , and the other arms of the bridge consisted of manganin resistances , one of 100,000 ohms and the other , , adjustable .
current of 5 amperes was passed through the wire for an hour before taking ths measurements .
The second column in the table records the value of as determined by an ampere balance in the circuit , the accuracy of which could be relied upon within 1 in 5,000 .
The temperature of the inflow water during the experiment was C. The length of the wire was cm .
The resistance of the column of tap water in the tube between the thick copper electrodes ( measured by means of the Kohlrausch * Our best thanks are due to Messrs , Johnson and Matthey , who were kind enough to give special attention to the drawing of the various wires required for the purposes of the investigation .
1911 .
] Heating in surements of Electrical wheel-bridge and telephone before soldering in the wire ) was 82,000 ohms , i.e. , about 600,000 times the resistance of the wire .
The correction of for the conductivity of the water is therefore insensible .
Platinum Wire No. 1 .
Diameter , cm .
Water flow , 100 .
per second .
The calculated values of which are given in the table are from the expression This gives for in the expression the value which may be compared with the value in the oil bath with " " nonnal\ldquo ; stirring , which was Hence by this arrangement , with a water flow of 100 .
per second , the effect is reduced to one-fifteenth of its former magnitude .
If a platinum wire of cm .
diameter and 1 metre in length , having a resistance of about half an ohm , be used as a standard resistance for carrying currents up to 5 amperes , mounted as above described in a tube of about 1 cm .
diameter , with a regulated flow of water of about 100 .
per second , it is therefore easy to apply the necessary correction , so as to obtain an accuracy of 1 in 50,000 on the resistance measured , and the error introduced by the conductivity of the tap water will be insensible .
In order to test the influence of the nature of the surface of the wire on the resuIt , the same wire was coated with platinum black electrolytically , and afterwards submitted to a white heat in the flame of a Bunsen burner , whereby the platinum wire acquired a dull grey matt surface .
The result was a small further reduction of from to In order to test the influence of water velocity on the resuIt , the wire was submitted to a continuous constant current of a little over 5 amperes 640/ 25 , and the bridge measurements were taken with the water flow varying from 100 .
per second to .
per second .
The results of this series of experiments are given in the subjoined table .
The inflow temperature of the water was C. The outflow temperature was Messrs. Glazebrook , Bousfield , and Smith .
[ July 29 , sensibly the same as that of inflow for the first four or five measurements .
For the last the outflow was about higher than the inflow .
The calculated values of are obtained from the expression This expression ives the value of with infinite water flow as The observed\ldquo ; values of are obtained by from the observed .
values of R. The series of values of illustrates well the great importance of effective and regular stirring in securing a reliable correction for the effect .
Even with the very small flow of .
per second , the mean temperature of the cooling water is only higher than in the first experiment , but the value of is 20 times as great , and the increase in the resistance is nearly 2 per cent. With slow velocity of flow the turbulence ceases , and is replaced by a slow , gliding motion of the water along the wire .
It is to this that the increased resistance is to be attributed , and not to the slight increase of temperature of the water .
To determine the actual law of the effect as depending on the radius of the wire , foul other wires were taken , drawn from the same material , and the resistances measured with varying currents and a uniform water flow of 100 .
per second .
The results are given below ; the resistance of each wire measured with an indefinitely small current being taken as unity .
All these series of values satisfy the equation 1911 .
] Heating in Measurements of Electrical Resistance .
Experiment on Five Platinum Wires with uniform Water Flow of 100 .
per second .
The values of ether with the diameters of the wires , are given in the following table .
The values for the diameters are each the mean of 10 measurements .
If in a diagram there are set out the values of ) as ordinates , and the values of as abscissae , the result is a straight line , which gives the relation The values of calculated from this expression are given in the last column of the table , and they correspond very accurately with the observed values .
Hence , for a series of wires of similar material , similar surface , and with the same water flow , we have constant .
If be the resistance of wire per unit of length and the specific resistance , we have , and therefore constant .
556 in surements of Electrical Resistance .
Furthermore , that the whole effect is due to rise of temperature of the wire , and within the range for which the temperature-resistance curve may be taken as a line , we have where is .
the ordinary temperature coefficient .
Also by equation ( 1 ) and since and , we have , or Thus it follows for a selies of wires of the same material that constant .
It thus appears that the emissivity\ldquo ; of a round wire in contact with a liquid cooling agent is proportional to the inverse square root of the radius , when the other conditions of convective action of the cooling agent at the surface of the wire are kept constant .
The Properties of Systems .
III.\mdash ; The Osmotic Pressure ofColloids .
By W. M. BAYLISS , F.RS .
( Received June 30 , 1911 .
) [ This paper is published in Series , vol. 84 ( No.

