]\gt ; [ 161 ] VI .
On the of Temperatnre : Experiments the Cavendish Laboratory , Cambridge .
H. L. CALLENDAR , B.A. , Scholar of Trinity College , lmbridge .
icated by J. J. THOMSON , F.R.S. , Professor of Physics in the Universi .
of Cambridgc .
Received June 9 , \mdash ; Read June 10 , 1886.* [ PLATES 11-13 .
] General Account of the Experiments , and Discussion of the Results .
IN view of the enormous discrepancies at present existing in estimates of high temperatures , it is exceedingly desirable that strictly comparable thermometric standards should be issued by some recognised authority .
Professor J. J. THOMSON , in the course of a conversation which I ] with him towards the close of 1885 , suggested that such standards could be issued in the form of platinum wire , the change of ical resistance with temperature being determined by comparison for each specimen before issuing .
The object of the present investigation was to test whether , in spite of the B. A. report on the SIEMENS pyrometer ( 1874 ) , pure platinum wire might not be possessed of the necessary qualifications for such a standard .
Let us first consider what qualifications are necessary .
( I. ) Such a standard should always give the same indication at the same temperature , should be free from secular change of zero , and should possess the widest possible range .
It should be portable and readily copied , so that standards might be multiplied , and , if the original were lost , it might be replaced by means of its copies .
The copies should be capable of accurate comparison and verification .
( III .
) It would be convenient if it agreed very approximately with any standards already in use , and if , failing this , the relation between the new and pre-existing standards could be accurately ascertained .
It would be a great additional advantage if , besides being useful as a standard , it could also be applied directly to all kinds of pracfical investigations .
With regard to the first point ( I. ) ; the of the platinum thermometer * The portions enclosed in square brackets have been added or altel.ed subsequently , June , 1887 .
18.8 .
87 MEASUREMENT OF therefore possesses in an eminent degree the qualifications necessary for a scientific standard .
There seems no reason to expect that the temperatul'e coefficient should be liable to secular change , seeing that violent treatment of these specimens for some months has not produced any measurable alteration .
III .
As it is a very important question to relation between ture by platinum wire and temperature by air thel.mometer , considerable space is devoted to the discussion of experiments bearing on the question .
The present standard of thermometry is REGNAULT'S normal air thermometer ; this , while posseveral advantages as an ultimate st. , is , nevertheless , so difficult to use that the discrepancies at present are bewildering , and it cannot be used at all in ordinary work for most purposes ; all measurements have , therefore , to be referred to it indirectly , which is a great .
But until the thermodyllamic thermometer shall assumed a more practical form , the air gives the best approximation to absolute temperature .
It happens that the platinum thermometer may be more accurately compared with this standard than any other .
The practical difficulty of all thermometric comparisons is that of maintaining an enclosure at a constant and uniform temperature .
This was to a great extent avoided in the present case by enclosing the spiral of fine platinum wire inside the bulb of che air thermometer itself .
There are , of course , some mechanical difficulties in the way , and it was found necessary to devise a modified form of .
thelmometer for the purpose .
Full details of the experiments and methods are given in subsequent sections .
The modified fonn of air mometer is also minutely described , as it may be found useful in other and similar investigations .
Various sources of error are also discussed , which may have considerable effect in experiments at high temperatures .
The investigation of these is still incomplete , but does not present any difficulties which may not be overcome with the aid of the platinum thermometer .
Among these we may mention surface actions between the gas and its enclosure ; the question of the possibility of eliminating the capricious changes of volume of the envelope ; and of determining the amount of dissociation to which gases are subject at high temperatures .
The experiments are , therefore , necessarily incomplete ; they are published chiefly on account of the great practical importance of settling on a universal standard for the more accurate comparison of different measurements of temperatnre .
[ The results of the comparison between the platinum and air thermometers are best represented graphically by drawing the curve of difference of temperature .
If stand for the temperature Centigrade by air thermometer , and pt for the temperature by platinum wire , that is to say , for the function ; the observations ( Plate 13 ) are seen to agree fairly well with the parabola .
( d ) OF TEMPERAT cannot be well represented in an empirical formul:a .
The simple parabolic formula , , corresponds exactly with ( d ) ( p. 163 ) , namely where S , and gives an accuracy of the order through a range of 60 C. Othe .
formulae may readily be found which give nearer approximations in special cases , and through small ranges .
For instance , ) exponential formula corresponds to the dotted line , Plate , fig. 11 , and exactly out the characteristic inequalities in the platinum-iron comparison-curve , greelng with it far more closely than the rabola ; but in most other cases the exponential formula has the disadvantage .
BENOIT and MATTHIESSEN use the parabolic formula , but MATTHIESSEN expressed his results as variations of conductivity , so that his formulae are quite inapplicable except between and C. ] I have endeavoured to give the fullest data in each case for the verification of my results , and to make the calculations in each case accurate within the limits of probable error , but among so large a mass of figures I can hardly hope to have escaped an occasional slip .
My most sincere thanks are due to the authorities at the Cavendish Laboratory for the facilities which they gave me for the prosecution of these experlments , and especially to Professor THOMSON for the kind interest he has taken in the subject , and for valuable suggestions and assistance .
In overcoming the mechanical difficulties which are always the most important part of an investigation like the present , I received great help from Professor THOMSON'S assistant , Mr. D. S. SINCLAIR , though , owing to the many calls upon his time , he was unable actually to make much of the apparatus .
I had intended to make many other experiments , but my progress has been much retarded by my own want of mechanlcal skill , and the necessity of making all my own apparatus .
I am at present intending to pursue these experiments to higher temperatures ; and have already worked out the experimental details of the application of the platinum thermometer to questions in the theory of heat where I consider greater accuracy may be obtained by its use .
Authorities consulted .
On the variation of the electrical resistance of metals with temperature .
All experiments almost up to date are collected in the first volume of the last edition of ' Die Lehre von der Elektricitat , ' by G. WIEDEMANN .
Original papers by BENOIT 111 the ' Comptes Rendus , ' and MATTHIESSEN in the ' Phil. Trans , are of the chief importance , but nearly everything has been collated .
In Thermometry the best general information up to date is to be found in the Enc .
Brit 9th edit .
, Articles " " Heat\ldquo ; and " " Pyrometers already referred to .
I have also studied with great care the whole of the classical researches of in the ' moires de l'Institnt , ' vols .
21 , 26 , \amp ; c. MEASUREMENT OF If the bulb itself cannot be used , a mel .
CUl.y thelmometel is made the same tube as the bulb .
Since , however , this method ceases to be plicable ) the boiling point of mercury , and is moreover exceedingly liable to inaccul.acy , except between and 10 C. , it was decided to measure the variations of the } coefficient and assume that the coefficient varied in the same way .
The method adopted was exceedingly simple , and is , shol'tly , as follow Lincar of Hard A glass tube is heated in a thick iron tube in a long gas furnace , the variations of its length observed by means of reading microscopes , and the mean temperature of the portion heated is given by observing the resistance of a platinum wire extending down the of the tube , assuming that when the temperature is steady the mean temperature of the wire is the as } of the of the tube it occupies .
by this method were made on ) piece of glass tube , the same as that used in making the air thermometers No. and No. 5 .
The ends of the tube were drawn down to 5 millims. diameter , to diminish the conduction of heat , nd to ellsure that the exposed should be at the temperature of the atmosphere ; copper leads were fused on to the ends of a piece of the stan dard platinum wire , centims. long , which was then inserted down the axis of the glass tube .
The glass tube was heated by being placed an iron -pipe , 67 centinlB .
in a gas ful.nace of the same length , supported on a with levelling screws for focussing .
The copper platinum junctions were exposed to the air temperature equal distances , about 1 cm .
beyond the ends of the iron tube , and , being screened from the heat by bright tinned plate , no appreciable -electric effect was produced , as the junctions were very slightly , and nearly equally , A length of about 63 cm .
of the glass tube was left in its originalstate ; the down ends projected about 16 cm .
beyond the iron tube , and were practically at the air temperature for their whole length beyond the copper platinum jullctions ; the variations in length of these portions of the tube may be neglected in compal.ison with other possible errors ; it is , therefore , assumed that the mean temperature of the platinum wire is the mean temperature of the of the tube it occupies , whose length is variable .
If the coefficient of expansion of glass accurately constant , or if the temperature of the whole portion heated were uniform , this assumption would give mathematically corl.ect results .
Since both these conditions al.eneal .
lie satisfied , the corrections to be applied for the variations at either end are negligible .
We take , therefore , the length of the part of the glass tube , whose variations we measuring , to be the distance between the ends of the platinum wire , and its mean temperature to be that of the platinum wire .
MEASUREMENT OF TEMPERATURE .
MLNT OF TEMP about twenty times as important in a mercury thermometer , where they product notable chcmges of zero .
One great advantage of the air thermometer is that they can usually be neglected .
The bulb is connected with the gauge by a tube of capillary bore , part of which is heated with ] the bulb , and part of which is at the known atmospheric temperature , a length of about 5 centims. being at intermediate and unknown temperatures .
If the bore be between three and five-tenths of a ] and the capacity of the bulb 50 cllb .
centims. , the uncertain error will be much less than , and may be neglected .
Some kinds of glass are liable to be cted on chemically by the gas at temperatures .
For instance , hydrogen or coal exerts a powerful reducing action on lead glass at a dull red heat , and the action proceeds more slowly at much lower temperatures .
Perhaps nitrogen is , .
this reason , { he most suitable thelmometric material .
Water vapour and carbon dioxide also are well known to exert a disintegrating action , and are for other reasons also especially to be avoided .
The Gauge .
It has usually been the custom to confine the air to constant volume in the air thermometer by a column of mercury whose level is adjusted up to a fixed mark before taking an observation .
The use of mercury necessitates a very wide tube , otherwise the effects of capillarity produce large errors : the volume of the air unheated must , in consequence , be large and also variable with the shape of the meniscus .
Moreover , for accurate work , it is necessary not only to thoroughly dry and clean the bulb of the air thermometer , but also the tubes and mercury of the manometer which is in direct connexion with it .
This it is practically impossible to do perfectly .
Again , it is in most observations exceedingly inconyenient to have a mercury manometer rigidly connected with the bulb of the air thermometer , and in such close proximity with the source of heat that the mercury becomes unequally heated .
The chief adyantage of mercury is that its vapour tension is so small ( about 001 centim. at ordinary temperatures ) that it is negligible .
Pure , howevel possesses this same qualification , and has nore of the above disadvantages .
Since it wets the tube , the col.rection for capillarity is quite constant .
If the bore of the tube be about centim. , the capillary elevation is about centim. With a gauge tube of this bore the correction volume may be made quite small , and may be accurately known , the shape of the meniscus being constant .
This enables us to use bulbs of smaller capacity , which are , for many purposes , more convenient , without unduly increasing the magnitude of this correction .
The use of sulphuric acid in the gauge tube is also attended with several other advantages .
The air thus rendered very sensitive , the slightest change of temperature is at once indicated , and it is very easy to see when the temperature is steady .
The mass of air under .
H. I , .
CALLENDAR ON THE PRACTICAL observation is pel.fectly confined , and the same mass of air may be kept under obser .
vation for a lengthened period , and is maintained perfectly dry .
Since the bore of the gauge tube is small compared with that of the mercury manometer , a tnge of of .
in the bulb first takes effect in the small-bore only lter the reading of the wide tube melcury manometer very slightly : this enables the instrument to be used when the temperature not absolutely constant , as the reading of the gauge can be instantly recorded , and the mercury manometel read at leisure .
It might be thought that errors would be introduced by and viscosity of the acid in so small a tube ; but it can be easily shown by direct expel.iment that this is not the case to an appreciable extent .
If an open gauge be con nected to a merculy manometer , the acid exactly and perfectly follows evely eadjustment of the manometer , and its capillarity and density may thus be in terms of mercury .
It has been suggested that the acid might have some action on the confined mass of , eithel absorbing it or giving off : it would be difficult to prove that this is not the case .
On the other hand there does not seem much reason to believe that it is .
of the gauge is that , when the instrument is at constant , snch as that of melting ice , by altering the adjustment of the mercury manolneter , the reading of the gauge is altered consistently , and a great llumbel of quite dependent observations may be taken , which ought , of course , on eduction , to lead to almost identical results , the mean of which will be to a great extent freed scale and personal errors .
The gauge is with a scale , and is carefully calibrated ; it is difficult to make an error of 02 centim. in reading the scale , and all errors divided by bout 7 in the reduction to mercury , so that they may practically neglected as far as the pressure is concerned ; they become more important in applying the volume correction , especially if the volume of the bulb be small .
The extl.emely simple of gauge shown in the diagram may be used with great advantage , but it is liable to some slight in conveniences .
Supposing the air in the bulb to be at atmospheric pressure and temperature , of 1o C. in its temperature alters the reading of the gauge by nearly 2 centims. If the limb of the gauge be only 10 centims. long , it has thus a range of only about without readjustment of the mercury manometer , so that , when tempel.ature is changing , the manometer requires to be continually readjusted .
More elaborate forms were , , actually used in most of the experiments one of which is shown in fig. 2 , Plate 11 .
The gauge tube carries a bulb which is ated , and whose volume is about * A very complete answer to this and other objections to the use of sulphuric acid will be found in S18 of ir WTLLIAM article on ' Heat in the ninth edition , 1880 , of the 'Encyclopaedia Britannica , ' vol. 11 , p. 561 .
MEASUREMENT OF MPERATIJ R equal to that of the bulb of the air thermometer : a little below the bulb the gauge tube widens out into a cylindrical , from which a side tube communicates with the three-way tap , which also serves to close the entrance to the large cylindrical bulb D. A side tube , near the lower extremity of this limb of the gauge , serves for the introduction of the requisite sulphuric acid , and can be sealed off when required .
The bulb serves to determine the volume of the bulb of the air thermometel at any time , on the principle of the volumenometer , and so to check any alteration in the volume of the bulb without the necessity of dismounting and filling it afresh ; it may also be used for several other purposes .
When it is desired to fill the bulb of the air thermometer with pure dry air by repeated exhaustion and re-admission at a high temperature , the whole cf the acid ( if the gauge is already full ) first drawn back into the bulb , which it completely fills , and the tap is turned thus ( ) : the contained air is then exhausted , and the dry air re-admitted repeatedly by way of the tube .
When this has been satisfactorily effected the tap is turned thus ( T ) , and the acid allowed to flow down into , completely cutting off the mass of dry air in the bulb .
The mass of air enclosed may also be easily adj usted by means of the manometer , to give any desired sure a desired temperature .
The volume of the bulb gives the necessary range to the gauge in heating or cooling ; if the tap be turned thus ( ) , so as to put and in munication and cut off the manometer , the instrument may be to itself to cool heat through a large range .
When the air thermometer has been thus left to itself to cool or heat , it often happens that the pressure is only roughly known , so that , if communication were opened direct between the gauge and the mercury manometer , a sudden and disastrous movement of the acid in either direction might take place if the manometer did not happen to be exactly adjusted .
It is convenient , therefore , to introduce the small auxiliary piece of apparatus on the way between the gauge and the mercury manometer .
The tube is of small bore , about 1 mm. , and is used as a kind of safety indicator to show when the pressure is the same on both sides of the three-way tap ; for this purpose it is half filled with mercury .
The pressure of the air in is always known to within 20 cm .
or so of mercury ) so that the manometer can be approximately adjusted ; the tap is then turned thus ( T ) ; the tube indicates the residual difference of pressure on either side of ; the manometer is adjusted till this difference vanishes , and is then put into direct communication with the gauge .
If these precautions were not taken , the acid might sometimes get sucked back into the bulb or the pressure , either of which events would ruin the series of observations .
the of xpansion o The sulphuric .
is for effecting this comparison by a different i Similar thermonleterH , filled with ] different at approximately the same pressure , are exposed to the source of heat and connected to the same mercury manometer , so the extel'nal circumstances are precisely the same for both , and the difference of their pressures at any is indicated ) the stllphuric acid in the gauge .
In a rison of air and between and the diiference amounts to about 401nm .
of acid , a very surable quantity .
In the first experiments the manometer belonging to a " " JOLLY\ldquo ; air thermometer was used .
It had a ] lnirror scale of millimetres divided with considerable accuracy , but it was found th the tubes were of too small diameter , and the mercury was apt to stick Another manometer was therefore constructed , with wider tubes , about cm .
internal diameter , which was found to be exceedingly satisfactory ; the same rnirror scale was used , and same adjusting apparatus .
The scale is only divided to millimetres , but it is possible , with practice , to estimate with considerable accuracy to the tenth of a millimetre .
It appears from the many servations made with the ratus that the probable error of a single observation is bont this .
It would have been theoretically preferable to take the with a kathetometer , but this would have involved great loss of time , and there was one to spare .
To avoid errors of , \amp ; c. , cadings were taken as follows:\mdash ; The scale was fixed vertically in a vertical plane , with the aid of a plumb-line .
A telescope capable of sliding up and down , being cl mped at any height on a vertical axis , was levelled by focussing it on its own nage in the mirror at the distance of about a metre ; this gave magnifying power of about three neters as compared with the aked eye a distance of 10 To eliminate perso1lal errors in estimating the of a lnillimetre , a series of independent observations at ) oints of the scale were taken at each constant .
The close agreement of these shows the apparatus performed satisfactorily .
The mirror scale was tested for scale errors by comparing it with the standard metre , with the aid of reading microscopes ; the eryors were found in all cases to be less than01 cm .
, and are therefore ected .
In taking observations , the reading of the top of the meniscus in each ] is recorded and the rence set down in an cent column ; this is then corl.ected for and reduced to zero , allowing for the linear expansion of the glass scale .
OF The tempel.ature of the mercury is repeatedly taken in the course of the obseryations by the aid of a ] icate thermometer immersed in the mercury of the open .
It almost ] differs slightly from that of the air .
The correction for temperature is only applied to the nearest tenth-millimctre , partly because the readings only accurate to order , and partly because , since the temperature is varying and is ) the same oughont , the correction itself is iable to a slight uncertainty .
To ayoid in accuracy , the calculations were usually carried to the next figure , to which , however , but little weight can be attached .
1r he Barometer .
Readings of the atmospheric pressure were taken the standard Fortin meter .
Each reading is subject to the following ctions : 1 .
The most important is the correction for temperature .
The scale is of brass , graduated in English inches , and is correct at a ature of Fahr. The scale errors are negligible .
The density of the mercury is normal at Thus for a height of cm .
the correction is cm .
at Cent. , since the brass scale is incorrect ; at Fahr. , the .
ss scale is corl.ect , the mercury correction is cm .
, and the change in the correction is 0123 cm .
per degree Centigrade .
The correction is uncertain , because it does not { ollow that the hermometer indicates the mean temperature of the mercury and scale .
2 .
The correction for capillary depression and vapour tension of mercury amounts altogether to about cm .
Cent. 3 .
The corrections for gravitation and height above sea-level need not be applied , since absolute measures of pressure are not required ( except for determining the temperature of saturated steam at atmospheric pressure ) .
4 .
The level of the cistern of the barometer is at a height of 4 metres above the manometer .
A constant correction cm .
is applied to allow for this .
5 .
Sources of Erro .\mdash ; The atmosphel.ic pressure varies incessantly and capriciously , so that the reading of the barometer cannot be beyond the nearest tenthmillimetre NAULT ) , even if all the mechanical adjustments could be perfectly effected and the temperature were certain .
The barometer is , therefore , read seyeral times in the course of each series of observations , and the applied to each reading to the nearest tenth-millimetre .
6 .
The constant errors of the attached thermometer sufficiently small to be negligible .
In reading the barometer , precautions were taken to avoid inaccuracy in the mechanical adjustments , the probable error this llly about Paris .
de nstitut , ' tom .
21 , p. C9 .
UBEMENT OF RATU R Let , and Then , finally , . . . . . .
( 3 ) It is , in general , useless to proceed to a third approximation , because the tainty of the corrections is at least of the same order , as a rule .
The value of , which may be called the " " corrected\ldquo ; pressure , is obtained directly from the observations by applying the small corrections in the manner detailed below .
is proportional to the absolute temperature .
The ratio is theoretically constant for each filling , though , as a matter of fact , it is yariable with the " " zeroerrors\ldquo ; of the air thermometer .
The value of is calculated from by the aid of logarithms , the logarithm of being tabulated once for all for each series of observations .
This is the only part of the reductions where logarithms are required .
The correction terms being , from their very nature , uncertain , often to the extent of nearly 1 per cent. , it is quite sufficient to perform all the multiplications they involve by the aid of a small slide-rule results correct to 1 per 1000 .
This method saves trouble , paper , time , and mistakes .
Red , of the .
The correction of the mercury columns for temperature is most easily applied by a graphic method : by ruling two series of straight ]ines , giving the correction to be applied to any length of mercury column less than 1 metre for each degree of temperature Centigrade , both for the glass millimetre scale and for the English standard barometer with the brass scale .
The barometer lines need not extend through the whole range , since its reading varies far from 76 centims. Let be the height of the barometer , corrected for temperature , \amp ; c. , as aboye described .
Let , be the readings of the right and left limbs of the mercury manometer ; and let be corrected for temperature and added to H. Let be the reading of the gauge when the pressure is the same on the in both limbs .
Let be the recorded reading in centims. Let density of 1 .
density of mercury Then is the pressure due to the acid , expressed in terms of mercury : is usually so small that no temperature correction need be applied .
Both and the reduction factor are determined by previous experiment with the acid and gauge used .
MEASUREMENT OF IPERATURE .
rections to be applied to the Thermometer .
Air does not follow exactly the equation of elasticity of a perfect gas Since , however , it follows this law very approximately , its true equation of elasticity may be written ' where is an unknown function of the variables which must be very small .
The experiments bearing on this deviation are those of REGNAULT on the compressibility of gases at different temperatures and the comparison of air thermometers ; and , in particular , the experiment of THOMSON and JOULE ( ' Phil. Trans lS62 ) " " On some Thermal Effects of Fluids in Motion From all these experiments RANKINE deduced as the approxunate equation of elasticity It may , of course .e , be written in our in the form , ( E ) in which for air TABLE .
The following Table gives approximately the values of this correction at different points for the standard air-thermometer:\mdash ; tempel aturec tonstant deonstant p *This is the form notation given by CLATJSIUS , Iechanical Theory of Heat , ' p. 227 .
MEASUREMENT OF TEMPERATURHi .
81 Changes of Zero In spite of all precautions , slight changes of zero pressure were observed with all the air thermometers , amounting in some cases to upwards of a millimetre of mercury in a week ; these were probably due to surface condensation .
In support of this it may be remarked that the vortex atom theory of gases requires the existence of an air film of swollen vortices sticking to the sides of the containing vessel ; the phenomenon is also well attested by mally experiments ; in particular , we may mention that of Professor BUNSEN " " on the condensation of on the surface of -spun glass It is still necessary , however , to determine in what way the film varies with the temperature .
I am investigating this , but the experi1nents are not yet completed .
the present we may assume that it exists and see what kind of error it will produce .
Let be the mass of air enclosed in grammes ; , its pressure , volume , and absolute temperature ; the difference of the heats of unit mass of air ; the surface density of the air film in grammes per square centim. ; and the surface of the air thermometer in square centims. is a function of the pressure and temperature , and also apparently of the nature of the gas , and of the surface , of its past history , and of the time .
We shall have the ( ) First approximation , .
Solving this to a second approximation , being small , we have neglecting other corrections .
Let , be the pressures corresponding to and C. respectively , then , according to REGNAULT , ; according to BALFOUR STEWART , S .
In experiments roughly .
In BALFOUR STEWART'S case , so that , if we assume gramme per square centim. , we shall derive , both from and BALFOUR STEWART , the value , corrected for surface condensation ; also , in the case where , as in my air thermometer ( No. III we should expect to find the value *Simi ] are changes were observed by REGNAULT .
'Phil .
Mag Series 5 , vol. 17 , p. 161 .
Vide ' Mem. Inst tom .
21 .
S Vide ' Phil. Trnns .
, ' 1863 .
MEASUREMENT OF TEMPERATUR furnace ; this renders the wire very soft .
( If the wire is not annealed , its resistance will be slightly diminished on the first heating , owing to a change of structure .
) It is then wound on a glass tube and anne led again , to make it preserve a perfect spiral form .
Measurement of tloe sistance .
If the resistance be too small , chance errors , such as change contacts , are unduly magnified .
To make the resistance large involves either incollveniently fine wire or an awkwardly long .
In the actual experiments , therefore , ) were used with resistances varying from 5 to 20 ohms .
The most rapid nd accurate way of measuring such a resistance is by the WHEATSTONE bridge method with ordinary post-office box , and a sensitive THOMSON mirror galvanometer with and scale .
Resistances of 10 and 1000 are taken out in the arms of the balance , so that the adjustable resistance is 100 times that of the sp when the balance is obtained .
This gives the value direct to four figures , and a fifth may be obtained ) accurately observing the galvanometer throws and interpolating .
The box used was a very accurate box of B.A. units by Messrs. ELLIOTT .
On the extraordinary accuracy of the boxes the reader may be referred to the paper by GLAZEBROOK , " " On the Determination of the Ohm in Absolute Measure\ldquo ; ( see ' Phil. rans .
' 1883 , p. 262 , whence it appears that an uncertainty of 1o C. in the tenlperature of the box will give an error of about 0004 in the absolute value of a resistance , but that ratios of resistances , if the temperature of the box be uniform may be determined with an error of only about00002 .
sensitive galvanometer employed had a resistance of about ohms , being approximately that most suited to the given arrangement of resistances .
It was arranged so as to be extremely sensitiye .
A deflection of one scale division corresponded to a difference of 0001 ohm when the resistance to be measured was 10 ohms .
It was also nearly dead-beat to facilitate the reading of deflections for interpolation .
The battery consisted of from four to six ] elements in series ; as circuit had a resistance of about 500 ohms , they worked steadily .
Sources of Thermo-electric effects were made as small as possible by careful arrangement of the junctions ; they were then completely eliminated by using a rocking mercury commutator in the battery circuit , for rapidly reyersing the current and observing when the reversal had no effect on the galvanometer ; it is important , however , to make the thermo-electric effect small , because , if it is large and variable , it greatly interferes with accurate observation of the resistance by making the galvanometer unsteady .
There were only two screw contacts ; these were between the thick copper leads EASURFMENT O 181- ) whole work of the reduction of a single coJnpleteol ) selvation .
The method of observing is as follows :When the temperature is nearly steady , the nate resistan ce having been ascertained , the galvanonletel circuit is closed and the moment observed when reversing the battery CUlTent with the rocking commutator does not affect the galvanometer ; the simultaneous reading of the IJ gauge and the time are recorded , and the readings of the manometer and bal.ometel.al.e taken .
The column gives the absolute temperature corresponding to the mean pressure the temperature Centigrade observed the ected value of the resistance reduced to 1 at C. by dividing by the value of the value of calculated from the fornlulae ( ) ( see page 194 ) , and I constructed an experimental ument in November , 1885 , of lead glass at the Cavendish Laboratory .
It was necessarily imperfect from want of skill in glass working , }thd lack of the requisite assortment of lass tubing , However , it well suited to test the ) acities of the nlethod , and the obsel.vations taken with it as nearly as may be expected with subsequent oved apparatus .
Details:-cubic centims. , The observations are not all given , as they less accurate for several reasons .
Most of them were taken with the temperature varying rapidly , and the mercu1y manometer belonging to a " " Jolly\ldquo ; air thermometer was used , the tubes of which were too narrow for purposes of accuracy .
Nearly five centims. of the fine wire was outside the bulb , and the resistance measurements were rendered less accurate by some small therrno-electric effects and a slight uncertainty in ] the esistance of the comlecting wires , and by neg.lecting to take the of the esistance box .
capillary tube was also of the same bore as the tube of the gauge , as there was none finel to be had the spot ; this of course enders the volume correction more uncertain .
Table A-l .
contains the obsel.vations taken when the temperatul.e was approximately steady .
The column shows how very nearly they agree with the formulae .
Some of the tions , howevel , taken when the was rying rapidly about 25 C. , lleally two degrees out .
MEASUREMENT OF TEMPERATURE .
Details of Air T'hermometer No. II .
Volume of bulb cub. centims. Total volume of tube cub. centim. ; diameter , 03 centim. The gauge tube was of very uniform bore : cub. centim. per centimetl.e .
Cent. Mean coefficient of expansion of glass between and Cent. , , by mercury weight thermometer .
From the observations in ice and steam , on February 13 , we have To find the other temperatures , P. For the platinum spiral the resistance of the connexions was , at Cent. , 0139 ; at Cent. , 0221 .
The value of this small correction at intermediate tenlperatures was found by ruling a straight line on logarithm paper sufficient accuracy .
Applying corrections , we find .
ohms , colog As the gauge tube had been made unnecessarily long , the 'q ' correct on was large and negative .
By direct comparison with the mercury of the manometer , the ratio of the density of acid to mercury at was determined to be 1382 and the capillarity centin ] .
Thus , when .
; and , when centim. From these details the " " correction diagram\ldquo ; is constructed which is shown in Plate XI .
, fig. 5 , for the purpose of illustrating the method of reducing the observations .
The ordinate scale , on one side , gives values of , on the other , values of for the same abscissa scale of .
A ifference of in makes instead of 0133 when , so variations of may almost be neglected .
Only line is therefore drawn , corresponding to ] C. ( The lines of the section paper are only indicated , every tenth line being reproduced in the diagram .
) [ When the apparatus was cooling the lower part of the bath as some 50 hotter than the upper .
The platinum spiral was hung vertically in this Instrument , and was deformed by own weight , so as to be massed chiefly in the lower and hotter port on of the bulb .
The observations taken during cooling are , therefore , affected by error amounting to two or three degrees , the mean tempel.ature of the being higher than that of the air .
] I 8 R. H. ON PRAt NI MEANT OF ObservIII .
Since lead-glass begins to soften at about 45 C. , it was considered advisable to make an instrument of harder Jass .
Air thermometer No. .3 was con- structed of combustion tubing .
Since luard glass cannot well be joined on to other tubing , the capillary tube was formed by drawing out a piece of colnbustio tubing , which formed the bulb ; the platinum spiral was then the end out , and the whole cleaned and calibrated .
To avoid leakage , one of the electrodes passed out through the whole length of the capillary tube , and the fine wire itself was sealed through the glass at the other end , a length of 1 centim. perfectly coated with glass .
A capillary tube formed in this way is of course exceedincrly , and some delicacy of manipulation is required , but it appears that the instrument cannot be otherwise made of hard without glass fn.nace in which a bulb can be blown .
This thermometer was suitably mounted and filled with dry air at a temperature of 47 C. , and the series of observations recorded in Table -111 .
were taken with it .
It was heated in a thick iron tube constructed to fit it , and the temperature maintained steady by regulating the gas supply .
It showed no tendency to leak , till at the conclusion of observation ( 10 ) it was unfortunately cracked by too sudden exposure to the air after heating in the steam bath .
The chief objection to this thermometer was the small volume of the bulb , and the large volume correction , which necessitated a third approximation ; the value of equation 2 , p. 181 ) large , all the uncertain effects of surface condensation are unduly magnified .
For instance , for observation ( 7 ) we obtain for the value of the temperature or , according as we use the first or second value of The mean deviation of observations ( 1 ) , ( 6 ) , ( 7 ) , from the curve ( e ) is For this reason , I made a larger and better instrument on the same plan , } ) it was unfortunately broken just as it was completed .
In spite of this accident , I hoped to make another series of experiments with a better thermometer , but a most un fortunate accident has completely disabled my right hand for the last month and put a stop to my experiments .
Details of ermometer No. 3 .
Volume of bulb .
cub. centims. Capillary cub. centim. per centim. centims entim .
MEASUREMENT OF TEMPERATURE .
OF RATU R Note : Observation Ir .
1 is rked doubtful in book of notes , as there was small glycerine bubble in the air tube connecting the mercury manometer with the sulphuric acid gauge ; moreover , it disagrees with observations taken a few minutes after with the same instrument , when the bubble had been dislodged ( see II .
4 , 5 , \amp ; c. , next Table ) .
[ Nevertheless , as the temperature was steady , it is not liable to ) constant error ( see p. 187 , end Table VI.\mdash ; Thermometer No. 2 .
The above are all single observations , yet they are remarkably consistent with other , and give a very smooth curve .
They are given to show the capabilities of the method , and the great sensitiveness of a fine platinum spiral , It is usual to apply the method of least squares to calculate the empirical formula which best suits the observations .
This has the advantage of giving a grea } .
appearance of accuracy and npleteness .
In the present case , however , the , if any , ld not compensate the trouble , and the simpler method of selected numbers gives equally good results . .
CALLENDAR ON PRACTlCA 1 If we formula , takino data the , calculate the values of the constants , we shall a or less unsatisfactory formula of interpo ] ation .
To show how much values of for data , the following Table may be : \mdash ; The formula represents the observations fairly well within the ental , but is quite inadmissible for extrapolation , for it has a maximum about 1090o C. , whereas I have observed values of exceeding he second formula , , . . . . .
( ) has only two constants , and represents the curve nearly as well , but is subject to the same ection .
The formula ( c ) is approximately the rest parabola , but does not represent the observations so satisfactorily at low temperatures .
* has incidentally investigated the question , unfortunately gives no numbers , but represented his results graphically by a series of straight lines for each interval of , using the curve thus obtained to give the temperature in terms of the resistance .
His extended to 1000o , but he does not publish curve .
In 1871 SIEMENS suggested that the law of change of resLstance with temperature was ; where , are constants and the absolute temperature .
The nearest empirical formul of the SIEMENS type calculated from the above data is . . . . .
( ) This differs radically from the other empirical formulae .
It represents the observations about as well as the parabola , but at high temperatures approaches * For correction of is formula , see Appendix , p. 22 , ' ' vol. 26 , p. 287 , 188 OF the limit00252 , and does not vanish or give a maximum value for R. SIEMENH ' formula makes ative when found for platinum which happens to nearly with , except that ( s ) is hter and steeper .
It has been usual to use a formula of the type for extrapolation , which corresponds to Formula ( d ) , pp. , 1 For instance , BENOIT gives , and recently HUMPIDGE* ; with these we may compare my formula .
From these we deduce , as on p. 165 : BENOIT S24S ( roughly ) .
HUMPIDGE S071 CALLENDAR @1.57 BENOIT took the point of cadmium to be s6 C. , instead of about But the discrepancies in the values of S seem too large to be entirely attributable to errors of observation .
Formulae of extrapolation , unless very accurately verified through a wide range , are unsatisfactory , since they disagree so widely at high temperatures , and a very small change in the data from which they are calculated makes a great difference in the value of the coefficients ; whereas , if instead a formula of the type were used , it seems probable that different platinum wires would give almost identical results .
In support of this , the reader is referred to the " " of \ldquo ; ( p. 200 ) .
For the value of in this formula we substitute the mean coefficient between and 10 C. For instance , in the present paper we use the formula , and call the temperature so calculated " " by } wire just as we speak of " " by mercurial nometer although mercurial thermomete ] differ much more widely and irregularly from each other .
'emperatures so calculated will be strictly comparable , and the actual relation between platinum and absolute temperature may be subsequently by.a more htened and skilful * Roy .
Soc. Proc vol. 39 , No. 239 , Feb. 1885 .
See noires de l'Institut , ' tome 21 . .
H. ON CTIC A generation : meanwhile , the wire may be used as a standard , and preserve the record for compaison with future determinations .
And , since the range of the platinum thernlometer is very great , and different wires may be accurately compared hout their range , we may expect that by adopting such a standard we ] at rate secure consistency in place of the discrepancies at present prevailing .
[ An onential formula of the type was found to represent the comparison of platinum and iron very approximately ( see p. 202 ) .
If we assume from Series .
the data nd calculate the values of the constants for platinum , we obtain .
By an odd coincidence it happens that this formula satisfies almost exactly the atiollS at varying temperature in Series A. .
Values of calculated by this formula ( e ) are therefore given in Tables A. .
and A. .
for comarison .
of fferent TVires .
Different specimens of wire have very rent coefficients ; when , however , the resistance-variation of one specimell of wire has been determined by direct comrison with an air thermometer as above described the resistance-variations of other wires may be compared with the standard by a method as accurate as the comparison of resistances , and practically independent of all knowledge of temperature .
It is only necessary the temperature should be approximately uniform and steady enough to allow observations to be taken .
The nlethod consists in ding the wires to be compared side by side on a cylinder of non-conducting material , such clay , porcelain , or asbestos .
By this arrangement we may be sure that their mean temperatures are the same , and we have only to measure their resistances several tinles in rapid succession when the temperature is steady , or oscillating slightly about a maximum , take the mean of the tions .
The resistance of the standard wire gives the mean temperature , which is the same for both coils .
If , however , the resistance-variation of the standard wire be not considered known with sufficient accuracy , this does not affect the accuracy of the comparison as such ; the resistancevariation of the others is simply to temperature on the scale of the standard The simultaneous resistances are measured by precisely the same apparatus and the same method as before .
They are therefore liable to the same errors and corrections ; that is to say , they are corrected for the resistance of the connecting wires and the temperature of the resistance-box : the latter is less material , since it affects both wires proportionately .
The resistances reduced to unit at C. ; thermo-electric effects are especially to be avoided , and are , of course , eliminated as before .
That changes of structure may not occux , producing permanent changes in the resistance , the wire must be first annealed .
To prevent chemical changes such as * ' llomoires de l'fnstitut , ' tom .
21 , ASUREMENI OF TEMPERATUR oxidation , which would produce the same effect , the wires are sealed up in a hard glass or porcelain tube nearly exhausted , from which the air has been removed by substituting or The insulation resistance from one wire to the other can also be measured with the same bridge if less than a megohm .
It may be neglected if it is greater than 100,000 ohms .
This method affords also a fairly accurate means of measuring the temperature variation of the resistance of non-conductors , which begin to conduct electrolytically when heated .
The chief difficulty is polarisation , but this may be avoided in the usual way .
It would , of course , be possible by this method of comparison to draw up very accurate Tables of the resistance variation of metals and other conductors referred to platinum as the standard .
Such Tables would , vever , be of little practical use , apart from the particular specimens of wire for which the comparisons were made , except in so far as they tended to elucidate the phenomenon itself A Table of empirical formulae of the type has been published by BENOIT for most common metals .
The method of comparison was not adopted , as far as I can discover , in spite of its obvious advantages , and as he used different specimens of wire his results do not agree very closely with mine .
It is , however , from a thermometric point of view that this method of comparison is of the greatest possible importance .
For this reason we proceed to give full details of a comparison between two platinum wires .
Since the same apparatus was used for all the comparisons it will be described once for all .
The Diagram of Connections ( Plate 11 , fig. 3 ) shows the double Wheatstone bridge arrangement adopted .
The arm of the balance , AB , is divided in the ratio ; the adjustable resisAC 1000 tances , AD , , are taken from post-office boxes , and may have any value up to 10,000 .
The sensitive galvanometer may be put in circuit between and , or and , or and , so that the resistances of either DB or BE , or their ratio , may be observed without the necessity of constantly altering the resistances in the arms , AD , except in so far as is necessary to follow the slow chan , of temperature .
A complete rvation of both resistances can be thus made in a few seconds .
The actual arrangement the apparatus is shown in Plan ( fig. 4 ) , which is lettered to correspond with the diagram .
The ends of the wires to be compared are fused on to copper tails , and the junctions arranged close side by side at to avoid the production of thermo-electric currents .
represents the rocking commutator the battery circuit to eliminate any residual effects .
The copper tails are in turn soldered to thick copper rods , passing , air-tight , through an indiarubber cork which fits the porcelain tube in which the wires are heated .
The copper rods are amalgamated , and dip into mercury cups , numbered 1 , 2 , 3 , 4 , 5 , on the plan .
Cup No. 2 is connected by a thick copper rod to the binding-screw , , of the resistance-box .
Cups 2 and 3 OF for the purpose of interpolation .
The close .
of the values of the atio r obtained from different resistances and at considerable intervals , shows the accuracy of the resistance-boxes and the reliability of the method .
Observations taken at lower temperatures were even more concordant .
The porcelain tube was enclosed in an iron tube to render the temperature more uniform and steady .
anceE 1eftM 9 Time , .
Quantity observed connexionsResistanceof Tnsulation re-l sistance / The following Table X. shows the reduction of some of the observations .
The first pair of observed resistances are reduced from the observations rocorded in the lOUS Table .
The erence of reduced resistances is found by the simplest possible reduction from the corrected ratio by means of the formula .
The value of is only approximate , and the difference of the mean coefficients is found by dividing by the approximate value of the temperature .
The mean of the standard wire is assumed to be constant , the relative variations of the other tabulated ; the column of temperatures is calculated by assuming for each wire a constant coefficient equal to the mean value between and 10 ; the difference column shows how very closely the two platinum wlres aglee : although their temperature coefficients differ about 3 per cent. , yet the temperatures deduced by usin either wire agree to nearly 1 in 1000 .
Several other observations were taken at different times with the same comparison coils ; they all agree within the hmits of error of the resistance-boxes .
No zero variation of either wire could be certainly detected ; and observatious taken with the temperature varying at the rate of 1 per minute agree to 1 in 5000 with the best . .
H. L. NDAR ON THE PRACTICAT , MEASUREMENT T ] !
]MPERATURE .
Comparison of and The comparison of and iron is interesting because ironl behaves so .
A considel.able number of comparisons were made by different methods ; the most accurate is the last series , VI .
, for which the same apparatus was used as in the comparison of two platinum wires .
All the series , owever , agree with each other as nearly as can be expected , the different modes of treatment , that ordinary commercial soft iron wire was used .
The first experiments were made at the beginning of November , 1885 , with unannealed iron wire wound side by side with platinum on a strip of asbestos board , and heated in an air bath up to 20 C. , a mercury thermometer being used to show the temperature was steady .
The following simultaneous values of the resist nces were observed : Platinum . . . . . .
Iron . . . . . .
Series I. Temperature , pt . .
01547 Afterwards the wire was partially annealed , and the sanJ e coil hcated in paraffin in the same air bath , with the following results:\mdash ; Platinum . . . .
Iron . . . . . .
icel .
Sories II .
Temperature , . .
order to command higher temperatures , a new comparison coil was made and enclosed in an exhausted tube , to prevent oxidation of the iron , which was first carbonised by annealing in coal gas ; the whole was heated in an iron tube by a regulated gas furnace .
The following observations taken:\mdash ; Platinum . . . . . .
Iron observed . . . .
Series Temperature , pt . . . .
investigate more particularly the effect of annealing , another comparison coil was wound the iron wire unannealed , and observations were taken with it up to 10 C. , with the following results:\mdash ; Order of observation . .
2 4 5 1 Resistance of platinum . .
Resistance of iron . .
Series Temperatnre , pt . . . .
146 Before annealing , the resistances at C. were :Resistance of iron , ohms ; resistance of platinum , .
After )ealing , resistance of iron , ohms ; platinum , .
So that the resistance of the iron was considerably increased , while that of the platinum remained practically unaltered .
The following observations were then taken:\mdash ; Besist llce.iron ovedPlatinum .
Series V. Tempel'aturc , They show how slightly the temperature coefficient was affected by annealing .
Curve I in the diagram , Plate 13 , fig. 11 , is constructed from the following observations , which were taken by heating platinum iron comparison coil in a vacuum in a porcelain tube which was continuously exhausted to remove the steam which is given off by the clay when first heated , and which vitiated the previous series .
Iron . .
PlatinumTemperaturept .
other observations were taken with the same apparatus at different times .
The resistance of the iron , when once annealed , was not found to be much permanently altered if due precautions were taken .
The curve representing these and other observations belonging to the same series at once suggests an exponential formula .
If we assume that the resistance variations of and iron may both be represented by formulae of the type , ( ) and if , be any two observed values of the resistance of the platinum coil , and , those of the iron coil observed simultaneously , we shall obtain at once , by equating values of the temperature for each observation , the conditions Whence the ratio constant .
By taking for , the extreme values at either end of the series we obtain the mean value of the ratio for the whole series to be * By some oversight I used the value.76034 in the calculations .
MEASUREMENT OF TEMPERATU R Testing the observations in pairs by this condition , the value of the ratio is found nearly constant ; a small deviation from the curve produces a large variation in the ratio , which is a ratio of small differences , and is of course immediately deducible from the differential equation corresponding to Formula .
The comparison gives the values of the ratios : , but not the absolute values of the coefficients .
Its accuracy is not therefore affected by that of the particular values assumed for the standard wire constants and Tf we take the observations on the comparison of two platinum wires and treat them in the same way , we shall find , from observations ( 1 ) and ( 5 ) , the mean value of the ratio So that , if we assume From this we obtain so that , if we take the value of previously found , namely , , we shall find So that is nearly equal to If , we have evidently and being the values of the resistances at the same temperature .
For iron , assuming the same values of and , we shall find The chief difficulty of the comparison is that of protecting the iron wire from alteration when it is maintained at a red heat for some time .
In Table XI .
the obseryations given in Series VI .
ars compared with the above exponential formula .
The observations were mostly taken on different days , and the resistances observed at the air temperature in the intervals show the direction in which the correction for zero variation should be applied , but not its amount .
It is noteworthy that in all cases the correction for this would tend to reduce the small differences between the formula ( e ) and the observations .
As it is , the mean difference is only TABLE XI.\mdash ; Comparison of Platinum and Iron .
It does not appear that any equally simple and convenient empirical formula could be found to repl.esent the observations so closely .
[ For the explanation of the small deyiations , see Appendix , p. 228 .
] Compa ) of Pure tin wire , diameter025 centim. Coil wound on cylinder of asbestos in paraffin in a glass tube .
Resistance of copper connexions Resistance in melting ice of platinum , of tin hypsolnetel at , platinum , tin Insulation resistance at C. , 170,000 B.A. resistance at C. , 300,000 B.A. of Results ; Means reduced .
Tin observed Tin calculated Platinum Temperature 89 .
68 .
54 .
205.3 The temperature coefficient of this platinum coil , taken from the observations in ice and , comes out 003462 : almost , within the limits of error the same as for other from the same reel .
The chief difficulty encountered arose from the softness of the tin neal its point .
The last obsel.vation is rather uncertain the metal gavewav before the temperature sufficiently steady .
The coil was OF mended , but the wire gave way below 10 C. It was , therefore , decided to try by a slightly different method , and in particular to protect the wire strain .
Some difficulty was found at first in fusing the fine tin wire on to the copper connecting wires .
Solder evidently inadmissible if the } ) iments w to be pushed up to the melting point of the tin .
The esistances , calculated from the mula , whence for purposes of comparison , show that , as be expected , the observations between and 10 are very consistent and continuous .
It is probable , however , that , as the last obseryation seems to indicate , the simple formula will only esent the observations well between narrow limits .
Other formulae for tin BENOIT : , whence MATTHIESSEN : , whence A small spiral of the standard platinum wire was made , having a resistance of about 1 ohm , and its ends fused on to short of stouter platinum , and these on to thick copper leads .
Resistance of the whole , cold , at C. , of copper connexions , about0140 B.A. The fine coil was raised to a bright orange ( in full daylight ) by being solid LETCHER gas ilame ; its resistar ) varied slightly on either side of varying draughts of air ; half the thick platinum was heated red-hot .
This was continued for about all hour .
When the whole had cooled again to the temperature of the air , which had risen to , the resistance was measured , and found to be 10502 B.A. The increase is almost exactly accounted for by the higher temperature .
Readings were then taken for some time with the coil in vaxious parts of the flame , and with different states of flame , with the following approximate results:\mdash ; Mlt .
H. NDAR ON PRACTICAL In this experi1nent the wire is radiating freely , so that it , not the tempera .
ture of the flame , but a point at which the loss of heat by radiation balances the heat received from the impact of the burning gas .
The more rapid the current , the the nperature reached ; this accounts partly for the higher temperature to the spiral was raised in the edge of the flame , where the current is much stronger well as the supply of air better .
By suitably screening the wire it would probably be possible to raise it to a higher temperature .
The temperature can only be roughly estimated by the radiation , because the latter is selective .
If the wire be blackened ( platinised ) it appears much brighter , because it radiates better , but it is for that very reason not so hot , as the measurement of its resistance Received June 9 , 1887 .
Since the foregoinos .
paper was communicated to the Royal Society , so many more extended and accurate experiments have been made that it was thought desirable to delay the publication of the paper for a few months , in order that some account of them might be included .
It was felt that this would render the paper more useful without adding excessively to its length .
The observations with the air thermometer in particular have been rendered more accurate , so that the temperature may be determined by aid of the platinum wire with a probable error of less than a degree at C. A much improved method of representing the results graphically has also been adopted ( see Plate 13 ) , which shows at a glance the kind of accuracy attained .
More accurate and extended comparisons tend to confirm the general conclusions from earlier experiments : that the curves of resistance-variations of different wires of the same metal are similar , but not coincident ; that the usual parabolic formulae give an accuracy , of the order of 1 per cent. only , through a C. ; that the deviations therefrom are not readily expressible by any simple formulae .
Considerable improvements have been made in the primitive apparatus .
An improved form of gau ge was devised and ordered about the middle of June 1886 , which was .
subsequently found to work well and to possess many advantages over the kind ; the principle was exactly the same as that described on page 172 , but the mechanical provisions were complete and perfect .
They will be readily understood on reference to the diagram , fig. 6 , Plate 12 .
The sulphuric acid U-gauge proper represented by the tube FGHJK ; it has a three-way tap which may be turned in the middle of the acid , so that it cannot move either way , hatever v$xriations of pressure : this is a most important practical improvement .
The amount of acid is introduced through the side-tube , and the scale MEANT OF TEMPERATURE .
FG should be adjusted so that the acid stands zero when the pressure is the same on either limb of the gauge at and K. In addition to the -gauge there is another connection between the .
thermometer and manometer by way of the bulbs , which is filled with pure dry mercury , and is used for observations at constant pressure , for calibrating the bulb with air on the method of the menometer , and as an operl way through which the bulb may be exhausted and refilled with dry .
It communicates by a three-way tap with the manometer and ; and by with , and through a long rubber tube with a mercury reservoir , which may be adjusted at any height , and which communicates with the outer air through a drying tube containing and glass wool to keeI ) out damp and dust .
Mercury may be introduced or withdrawn to any extent by aid of the reservoir , whose height is adjusted to suit the pressure , or in weighed by the tubes , which dip into weighed beakers of mercury .
The bulbs , in the figure are convenient for taking up leakage , and other practical details .
To determine the pressure correction in terms of mercury corresponding to each reading of the scale , the way AMN is cleared by drawing off the mercury into the l.eservoir , the tap is turned off , and turned thus ( T ) , and is connected with the open air through the drying tubes , so that the atmospheric pressure is communicated by way of to the -limb of the -gauge .
Readings of the manometer and the gauge FG are taken at various stments , from which the probable error of a single pressure observation and the pressure correction are easily deduced ; capillarity is thus eliminated .
To dry and fill the air thermometer , the bulb is maintained at a temperature between 40 and 50 C. , is connected to the exhaust , and with the dry-air supply .
By means of the three-way tap the bulb is connected alternately with either .
Mercury is re-admitted to fill the bulb , when the operation is considered complete .
To take observations at constant pressure , mercury is allowed to run out from the bulb A into the weighed beaker at , till the sulphuric acid gauge indicates that the pressure is approximately that of the air .
The scale FG gives the small variations of pressure , and the weight of the mercury the auxiliary volume .
The temperature of the bulb A is carefully observed , and is kept steady by being surrounded with a water reservoir .
Total volume of Bulb A from to .
at C. The gauge above described arrived about the beginning of August , 1886 .
A new thermometer , No. 5 , was at once made of hard glass , similar to No. 3 , the expansion of which was known .
There are some difficulties in mounting an instrument of this description .
The mounting is diagrammatically illustrated in fig. 7 , Plate 12 .
The bulb is fitted with asbestos board in a brass tube AD , into which is screwed a smaller tube DC which carries and protects the electrodes of the spiral and the delicate capillary OF RATURE .
this arrangement of the junctions -electric effects were rendered almost ciable in.spite of the great sensitiveness of the pparatus and the low resistance of the galvanometer .
The correction for resistance of connections , owing to bhe use of so much fine wire , amounted to nearly 3 per cent. , but it was measured on each and could be vely accurately applied .
With the instrument above described upwards of 130 independent observations in several sets , at various steady temperatures , were taken during the days August 7 to 12 , 1886 .
They we ] ewol.ked out on much ) the same plan as that explained below on page 218 .
To save space , only the observations at steady temperatures in Table A. The experiments were interrupted or- August 12 by the cracking of one of the glass joints of the gauge ; an attempt to inend it only made matters worse .
It was sent to the maker street ) for repairs ; and meanwhile other experiments were undertaken .
The breaking of the gauge was particularly unfortunate , as I had so far taken no observations at constant pressure , 410r which the gauge had been specially designed .
A great many observations were taken with the temperature varying to test the working of the ) paratus ; these on reduction ] were never found to differ by more than a degree from the curve , and often happen to fall exactly on it .
To insert them would only confuse the diagram .
The earlier observations of Series , which were taken .
just after filling , before the apparatus had reached a steady state , give results which are too by two or three rees .
This is an error of the air thermometer , and not due to .
of the resistance of the platinum ; for the observations in ice and steam show that the platinum spiral suffered no change during this series of observations ; but the value of the constant fell slightly , probably owing to surface absorption .
The air thermometer was filled with dry air on the 9th August , at a temperature of C. , after repeated exhaustion ; and it seemed probable that there would be a tendency for the air pressure to fall rather fast at first , owing to surface condensation ; between observations ( 3 ) and ( 4 ) the gas was left unaltered for hours .
The change in the difference column from to is in the direction of absorption , but condensation probably took place most rapidly as the instrument cooled at lower temperatures .
The historical element seems to enter here : the dates and times of the observations are therefore given ; heating gradually dissipates the air-film , which may take some time to condense again , so that an observation ( e.g. , No. 12 ) taken at a lower after a higher temperature may give results which are too high ; but this requires further investigation .
The later experiments seem to throw doubt on the suggestion offered a year ago ( p. 181 ) as to the value of explaining discrepancies in the values of the coefficient of dilatation according to different observers : the apparatus is not particularly suited for determining the absolute value of the constant , which depends on the difference of pressures in ice and steam , on which the probable error with my apparatus may amount to a quantity of the order of one in a thousand .
OF \mdash ; Comparison of Platinum and Air Thermometers\mdash ; ( continued ) .
Mercury manometer .
MEASUREMENT OP TABLE A.-v.\mdash ; Comparison of Platinum and Air Thermometers .
W. 244902 W. W. 98.535 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 22.46 W. 83.345 A. .\mdash ; Comparison of Platinunl and Air .
Spiral .
44.85 59.35 21.31 21.32 21.33 , while ed es .
MEASUREMENT OF TEMPERATURE .
TABLE A.-v.\mdash ; Comparison of Platinum and Air .
ASUREMENT OF legal ohm-box by ELLTOTT , nominally correct at C. , to which temperature the resistances are reduced .
) temperature coefficient of the box was mined to be 00023 per 1o C. , the of platinum-silver wire .
The relative values of the resistances were correct to more than 1 in 10,000 .
The resistances observed are given in legal ohms , but have to be corrected for the temperature of the box R. The correction for the resistal ) of connecting wires is By observing also the resistance of the capillary tube electrode , the mean temperature of the corresponding 15 cm .
of the capillary tube could be determined , and an allowance for its varying temperature approximately made by part with the bulb and part at air temperature .
In most cases the temperature was varying ; and an exact balance was observed with the rocking commutator as previcusly described , and the simultaneous reading of the -gauge was recorded .
When the temperature was too steady to admit of this a fifth figure was obtained by observing galvanometer throws and interpolating .
The column headed gives the reading of the level of the by the scale S. In taking observations at constant pressure , mercury was allowed to run out from the bulb Plate and weighed in a beaker .
The observed weight in grammes W. of the mercury and beaker is given in the manometer column ; the manometer was of course unnecessar .
in taking observations at the atmospheric pressure .
The temperature indicated by a delicate thermometer immersed in the water surrounding the bulb A is then given in the column .
The readings given of the heI mometers and barometer are mean or typical readings : all are not given , to save space .
The observations are numbered to correspond with Plate 13 .
No permanent change of resistance of the platinum wire was with certainty detected in this series of experiments , in spite of the prolonged heating to which spiral was subjected , and although it was in contact with the glass of the bulb for a considerable portion of its length .
The apparent change of ternperature coefficient from , Series IV .
, to , Series V. , was probably due to the fact that bhe position of the loose comparison-electrode was shifted so as accurately to correspond with the single electrode inside the capillary tube ; it had been accidentally misplaced through an error in measurement in Series IV .
The apparent change in the value of in Series .
from to ( the latter value exactly observed again one month latter , on April 18 ) was prob partly due to a the comparisonelectrode , the copper portion of which was accidentally fused in re-coating with hard glass on March 4 , so that it had to be re-made , and its resistance was slightly increased ; and partly , perhaps , to a change in the plug contacts in the 10-ohm arm of the bridge , which were thoughtlessly altered on that day to measure another .
The value should , of course , be used in reducing the observations from March 4 onwards .
On March 4 , while the bulb of the air thermometer was immersed in steam , it bl R. H. L. CALLENDAR ON THE filled afresh with dry air at 103 cm .
pressure , corresponding to a zero pressure of almost exactly one megadyne per sq .
cm .
By allowing the mercury.to run out till the bulb A emptied , and again measuring the pressure , and assumlng the formula , the volume of the bulb up to the Cu-Pt junctions was calculated to be .
A repetition of the same operation on March 8 with the same gave the same value .
The observations were all reduced , assuming as the volume of the bulb at C. , and its expansion to be that given by the formula on page 163 ( linear expansion of glass ) .
The correction for glass expansion amounts to nearly at 60 C. , and must be considered subject to greater uncertainty than the other elements of the calcutions ; the pparent discrepancies between observations at constant volume and pressure may be in part due to errors of this nature .
The method proximation used in previous portions of the paper in working out the results fails in the case of observations at constant pressure .
Latterly I have used a 's spiral slide-rule , which gives results correct to 1 in 10,000 at least .
is so expeditious and convenient .
that the formula 1nay be worked out directly with great rapidity .
If be the pressure of the air enclosed , V the volume of the bulb , its temperature , and any other portion of the volume which is at temperature , the formula of the air thermometer may be stated thus\mdash ; , . . . . . .
where 2 implies that the summation is extended to ] portions of the apparatus occupied by the mass of air under observatio ] ] , and depends on the mass of air enclosed , and is theoretically constant for each filling .
To make the statement more definite , will define to be such a function that the above equation is true , being and equal to one atmosphere : is , then , the temperature by normal air thermometer at constant pressure , and is nearly equal to the absolute temperature on the thermo-dynamic scale .
From Formula I. we obtain , as the value of . . . . . .
II .
To illustrate the corrections involved , we will proceed to calculate Observation ( 5):\mdash ; Mean value of resistance observed , legal ohms .
Correction for wires , corrected , .
Temperature of box , Temperature correction ( 54 ) corrected , .
Assuming from Observation ( 16 ) , .
Resistance of capillary ) electrode , .
Resistance of the platinum tion , 690 .
lIean temperature , cm .
of this portion of the capillary tub is therefore reckoned at air temperature , remaining 10 cm .
at the temperature of the MEASUREMENT OF TEMPERATURE .
bulb .
Thus : cm . .
The volume of this portion cm .
( by curve constructed from calibration data ) is 0 ; 30 cub. cm .
This has to be deducted from , the volume of the bulb , and added to , the volume at air temperature .
The volume of the bulb up to the platinum-copper junction of the electrode is calculated , assuming that air Boyle 's Law , from the observations ( 20 ) and ( 26 ) , both of which give the same result , In our case The glass expansion factor corresponding to is taken from the curve whose equation is given in the section on the expansion of hard lass , p. 168 , .
Small errors are less likely to be introduced if this correction is applied last of all to the value of To calculate the volume at air temperature 2 The volume from diyision of the scale , and of the scale to the platinum-copper junction of the capillary tube electrode was determined by previous calibration to be , and is assumed to be , at the temperature of the air , ; The mean value of , the reading of the gauge in Observation ( 5 ) , is The corresponding volume is Thus , and Mercury was allowed to run out by way of into a beaker till the difference between the pressures of the internal and external air came within the range of the gauge .
The weight of the mercury and beaker was , that of the beaker .
The same beaker was used in all the weighings here given .
Since the value of the temperature of the bulb A observed with the mercury thermometer cannot be correct to more than 1 in 5000 ( 1/ 20 degree ) , it is needless to reduce the weight to vacuo or apply any of the other small corrections .
The density of the mercury at C. being grms. per , the volume of grms. is .
Since the mercury stood at just ) efore it was allowed to run out , we must add .
to reduce it to Thus the auxiliary volurne was .
at a tenlperature C. by air thermometer .
Thus ; ; The acid stood at in the when the pressure was the same in either limb , and by experiment it was found that the reducing factor was required to give the pressure correction in terms of mercury .
Thus the required correct on Observation ( 5 ) is The barometer reading reduced to centimetres is cm .
, and the temperature C. ; correction for temperature , cm .
; correction for level and capillarity , cm .
MEASUREMENT OF TEMPERATURE .
The value is assumed in its reduction .
This is a little uncertain , because in Observation ( 2 ) the sun had been allowed to shine on half of the barometer in the room upstairs , causing the high reading F. of the attached thermometer , and rendering the temperature correction uncertain ; the same uncertainty , of course , applies to the reduction of Observations ( 3 ) , ( 4 ) , ( 5 ) , into which the same of mk enters .
A change of less than 1 in 1000 in the value of mk or V would suffice to account for the deviation of this ooroup from the curve .
Observations ( 4 ) and ( 5 ) indicate a difference of 1o C. between the values of the temperature by air thermometer at constant pressure and at constant density respectively at 54 C. This may be similarly due in part to errors in the assumed volume or expansion of the ] , or merely to changes in the temperature distribution between Observations ( 4 ) and ( 5 ) .
Between 20 and 30 C. the observations seetn to deviate in a regular way , as indicated by the dotted curve , from the parabola .
That this deviation is not merely due to errors ofobservation is , I think , indicated by the curves of comparison of different platinum wires , which seem to show that the increase of resistance does not strictly follow the parabolic law , but is affected by some minor and comparatively unimportant variations which , from their very smallness ( being of the order of 1o are difficult to take account of otherwise than in a graphic record .
Moreover , the observations in question were taken on different days and under different circumstances , so that it would be difficult to account for their agreement on any other hypothesis than that of some constant error aflecting either the platinum or the air therrnometer .
Observations ( 11 ) and ( 13 ) differ excessively from those in their immediate neighbourhood .
On reference to the Table A. , it will be noticed that they were both taken with the temperature varying much too fast for purposes of accuracy ; on the day in question the gas pressure was very variable , and it was difficult to get good observations at steady temperatures .
Observation ( 11 ) was further ated by the fact that some of the water had leaked from the glass case surrounding the bulb and its level had been allowed to get too low .
Observations ( 37 ) and were taken at constant density , and differ from ( 35 ) , ( 36 ) , in the same direction and for the same reasons that ( 5 ) differs from ( 4 ) .
This apparent difference is greater than that indicated by theory , and may be due to some error in the value of , the volume of the bulb ; further and more careful experiments are required to clear up this polnt .
Observations at steady temperatures from the previous series of experiments Tables are entered in the diagram by means of small .
circles ; they are seen to agree with the curve as well as can be expected , considering the varlous errors pointed out in each case to which the earlier experiments were liable .
Had the observations been more complete , the agreement would probably have more perfect .
In order to render them comparable with the results of Series A. , the same value of the coefficient of dilatation , namely , , has to be employed in their reduction in place of the coefficient , which is used in the bless A. .
and I11 .
in the earlier pol.tions of the paper .
This anomalous value of MEASUREMENT OF rEMPERATURE .
value of may be in error to the extent of about , as it depends on the air thermometer observations .
These , however , may be rendered still more accurate , and uncertainty fu1ther reduced ; they do not affect the claim of the platinum thermometer to rank as a practical thermometric standard .
The inum air thermometer gives us the means of comparing theory with experiment , and of testing the accuracy of the formula at high temperatures by taking observations at various essures in rapid succession , while the temperature is maintained nearly constant .
The platinum spiral may at least be relied on to give any small changes of temperature which may occur .
Nevertheless , for experiments of such delicacy it was felt to be necessary to intain the conditions as uniform as possible , because any slight variation might introduce errors comparable with the ces that might be expected to exist between the indications of the air thermometer at constant pressure and at constant density .
Consequently , when on return from Indis , in February , 1887 , I set to work again with the same glass air thermometer , a small gasometer was ordered in order to keep the gas pressure perfectly constant during any .
This want had not been so much felt in August , when there were few students at work and little was being used ; but the unsteadiness of the gas pressure was a great source of annoyance during the series of rvations given in Table A. , which were taken during February and March in full term time , pending the arrival of the gasometer .
On one occasion the temperature rose 10 in five or ten minutes , owing to change of gas pressure .
Such changes of temperature would have been less material , had the platinum spiral been uniform ; so that its mean temperature would have been more nearly that of the enclosed , The gasometer arl.ived about the end of April ; ib leaked badly , and had to be mended .
After that it performed satisfactorily , and a few observations were taken at steady temperatures one morning with its aid .
Unfortunately , the observations at constant density were pushed too far this and the capillary tube gave way at about 55 C. under two atmospheres ' pressure .
Air ermometer .
For temperatures it is necessary to use an air thermometer of porcelain .
The difficulties to be encountered are that the wire nlnst not touch the glaze and cannot safely be fused through the porcelain air-tight , at any point exposed to the high temperature ; and that expansion of the material must be determlned with some accuracy .
To meet these difliculties , a somewhat peculiar type of instrument has been made .
A cylindrical bulb of porcelain , AB , Plate 12 , fig. 8 , 3 cm .
diam. , 200 .
capac tie , opens at either end into thick side-tubes AC , , of small rectangular bore , 3 mm. Nll .
H. L. CALLENDAR ON THE PRACTICAL by 1 mm. ( one of which is shown in section , .
foll size , at E ) , which extend 6 inches on either side beyond the tube gas-furnace which the bulb is heated .
The sidetubes , are fitted to contain double electrodes of thickish platinum wire , to the of which , at A and , a single length of wire is fused , which extends through the bulb from A to B. By means of these double electrodes we can determine the resistance of the fine wire independently of its connections , and we can also determine the mean temperature of the side-tubes .
One end is fixed and connected with wax to the tube of the pressure-gauge .
The free end is closed with wax , and observed by means of a reading microscope .
The expansion of the bulb itself is thus simultaneously observed , and the electrodes air-tight at the points and , which are heated .
Moreover , the form of the bulb renders it suitable for .
in a tube gas-furnace , which is very convenient .
of Platinum Wires .
Owing to the importance , from a thermometric point of view , of the comparison of different platinum wires , another comparison coil was wound with greater precautions , to ensure perfect symmetry .
The wires compared were pieces from the reels 1 and 3 used in making the platinum spirals for the air thermometers of Series .
and V. respectively .
were fused on to platinum electrodes and wound on terra-cotta slips , as previously described ( pp. 199 , 200 ) .
The following obseryations were taken:\mdash ; TABLE An attempt was made to take an observation at a temperature of about , but the insulation was not quite good enough ; the temperature was allowed to fall again , and the following observations were taken:\mdash ; MEASUREMENT OF TEMPERATURE .
Number oThe value of is taken from the curve .
The observations are reduced as follows:\mdash ; Let be the resistance of the wire thab of the ratio , and the measured resistance of the connecting wires .
Then ( being small ) .
Let stand for , so that is a constant exactly analogous to in the case of the air thermometer .
Then ; .
And the difference may be pub into a form more convenient for calculation , namely , . . . .
( T ) The small difference may be accurately determined from the observations in ice and steam , for we have \mdash ; ; whence and the ] difference is accurately given by the formula , where is the ratio only quantity , therefore , requiring the fullest accuracy of observation is , the change of Since this is small , approximate values of the others will suffice .
From the observations ( 1 ) and ( 2 ) of Table .
the deduced value of is ; from Observations ( 7 ) , ( 8 ) , .
These are used in the reduction of Tables .
and .
respectively .
A rise of zero MEASUREMENT OF TEMPERATURE .
TABLE .\mdash ; Comparison of Platinum ( No. 1 ) and Commercial Iron Wire .
( The observations in this Table are the same a.s those in Table XI .
, p. 206 .
) Differen .
The value of is given by the formula . . . . . .
( ) The value of the difference is calculated from the formula . . . . .
( ) The temperature by iron wire is given by the formula , where TABLE F. .\mdash ; Comparison of Platinum ( No. 2 ) and Pure Iron Wire .
The formulae are as before , but the values of are different .
( See Obs. ( 2 ) .
) The deviations are taken from the last column and set down on Plate 13 , fig. 11 ; the abscissa is the approximate temperature by air thermometer ; the similarity of the curves shows that the resistance-variations of the wires are similar in character .
Herein we see the treachery of the common empirical formula .
Had we only the observations in Table II .
, we naturally suppose that the parabolic formula accurately represented the phenomenon , and that the numbers in the " " deviation\ldquo ; column errors of observation ; and regret that after all our trouble the results were so inaccurate .
On the other hand , had we only the observations in Table might be carried away ( as I was ) with the idea that the exponential formula was the true solution of the problem , and that the small deviations were accidental , as in this case they very well might have been .
But , having both series , .
and in every way independent , and yet exhibiting such a marked similarity , we are led to inquire what it means .
Each curve is marked by a small subsidiary wave at about 40 before the final upward rush .
This is not albogether fanciful , but is just such in position and magnitude as would be produced by an irregularity in the resistancevariation of the platinum of the order of taking place between 20 and 30 ; for in calculating the value of and we have assumed smooth parabolic formulae , which take no account of the small ularities which are shown to exist by the curves of comparison of platinum wirss , and one of which is indicated also by the air thermomeber observations .
In the curve F. .
the phase of this subsidiary wave is retarded .
Now the curve P. , containing the comparison between platinum wires ( 1 ) and ( 2 ) , which were used in the comparisons F.-l .
, F.-ll .
, respectively , shows that the curye of is affected by a deviation similar in magnitude and sign to that of , but taking place about later .
If this curious correspondence of the curves had been foreseen , the comparisons between the platinum wires would have been made more extensive and complete , and , in particular , the form of the curves between 10 and 30 would have been carefully explored .
Professor TAIT has observed that at about a dull-red heat the resistance of iron increases very rapidly ; the thermo-electric and mechanical qualities of the metal also change suddenly about this point .
It is well known that very slight impurities exert a very marked effect on the softening point of iron ; we need not be surprised , therefore , that the upward rush in the case of F. .
does not come within the limit of observation , like that of F. .
I propose to make a direct comparison between the wires F. .
and F. .
to confirrn this .
Unfortunately , my first impression on reducing the observations was that they disagreed completely ; the nature of the greelnent was not noticed till the apparatus had been dismounted and the wires used for other purposes , so that no more observations could be taken with the same coils .
The curves F. , F. , sufficiently illustrate the futility of applying empirical formulae calculated from observations between nalTOW limits for purposes of extrapolation .
I have known cases where an empirical formula , calculatsd by the method of least squares from observations at xarying temperatures with a raercuried thermometer between and C. , has been applied to estimate temperatures above 1000o .
This is an extreme instance of a method to be avoided .
Between the limits and , however , the parabolas\mdash ; MEANT OF TEMPERATURL .
, . . . . .
] .
, . . . . . .
both of which correspond exactly to the formula\mdash ; represent the resistance-variations of the wires I. and II .
with an accuracy of about 1 per cent. BENOIT gives for Iron\mdash ; ; ; for Steel\mdash ; ; These correspond to ( Iron ) and fe \mdash ; ( Steel ) respectively .
MATTHIESSEN , by obseryations between and 10 with a mercury thermometer , found for very pure iron the formula whence .
But this kind of formula is quite inadmissible for extrapolation , as it makes a maximum when .
Between the limits and 10 C. the formula of the parabolic type which agrees most nearly with it is \mdash ; He similarly found for two specimens of commercial platinum , between and ( 1 ) ; whence ( 2 ) ; whence Wire ( 1 ) at gives .
No. ( 2 ) gives .
These differ by half a degree , whereas the pure wires of Tables give approximately , and differ by only one or two hundredths of a degree .
MATTHIESSEN'S formula ( 1 ) makes a maximum when , and is of course inadmissible for extrapolation .
Comparison of Platinum-Silver Alloy .
This is of little interest , except that the platinum-silver wire is used for standard resistance-coils .
It is quite unsuited for thermometry , because of its small coefficient and variable composition .
frans. .
A. Il .
lA A4 of fl ll 6 6 6 / ' . .
/ ?
/ $ , ' '

