]\gt ; INDEX SLIP .
, J. A.\mdash ; On the High-Temperature Standards of the National ical Laboratory ; an Account of a Comparison of Platinum Thermometers and Thermojunctions with the Gas Thermometer .
Phil. Trans. , A , 1904 , pp. $43-384 .
Platinum Ihermometers , Coml)arison of , with Gas Thermometer at High Temperatures .
HABKIIB , J. A. Phil. Trans. , , vol. 203 , 1904 , pp. 343-384 .
Thermojunctions , Comparison of , with Gas TheImometer at High Tem- peratures .
HABKmB , J. .
Phil. Trans. , , vol. 203 , 1904 , pp. 34S-384 .
[ 343 ] XI .
On the -Temperatnre th National : an .
cconnt of a Comparison Th metions ?
oith the ) : J. A. HARKER , D. Sc. , Fellow of College , , Assistcmt the Communicated by R. T. GLAZEBROOK , F.R.S. , from the tional Received January 20 , \mdash ; Read February 11 , 1904 .
CONTENTS .
Page I. Introduction II .
The gas thermometer\mdash ; description . . . .
345 III .
Gas thermometer details\mdash ; scale , reservoirs , \amp ; c. 347 IV .
Barometer and auxiliary measurements 348 .
Preparation of the gases . . . .
348 VI .
Filling of the reservoir . . . . .
350 VII .
Dilatation of the porcelain reservoir . . .
350 VIII .
Pressure coefficient of the reservoirs . . . .
351 IX .
The platinum thermometers . . . . . .
352 X. The resistance box for platinum thermometry XI .
The potentiometer for thermocouple snrenleltts 353 XII .
Standard of electromotive force . . . . . .
356 XIII .
The thermojunctions . . . . .
357 XIV .
Formulae for metions . . . .
358 XV .
Determination of fixed points -point .
359 XVI .
Electric furnaces . . . .
361 XVII .
Furnace correction . . . .
362 XyIII .
Exploration of furnace 362 XIX .
Method of calculation of gas thermometer experiments 364 .
Specimen determinations of ice-and steam-points .
365 XXI .
Accuracy of constants of gas thermometer . . . . .
366 XXII .
Summary of calculation of gas thermometer tnres .
Series II .
368 XXIII .
Method of calculating comparison experiment .
369 XXIV .
Summary of experimcnta and constants used in calculation . . . .
372 Conclus]ons 377 XXVI .
Table of values of corresponding to given values of for . . . .
378 XXVII .
Table to calculate change in for a given small change 384 369 .
) 13.8.04 .344 DR. J. A. ON THE HIGH-TEMPERATURE STANDARDS I. Introduction .
IN a paper " " On the Comparison of Gas and Platinum Thermometers read before the Royal Society in Dr. P. CHAPPUIS and the author described a series of experiments in which several istance thermometers , constructed of wire of specially purity , were compared with the gas thermometer at a number of steady temperatures from below zero to above the boiling-point of sulphur , and in one set of measurements to just short of 60 C. The results were such as to substantially confirm the conclusion of CALLENDAR and GRIFFITHS that the indications of platinum thermometers may be reduced to the norlnal scale by the employment of CALLENDAR'S well-known difference formula where difference between , the temperature on the normal scale , and " " platinum\ldquo ; temperature .
The constant S for pure platinum wires is approximately , the three temperatures chosen for its determination being and the boiling-point of sulphur .
The paper concludes with the sentence , " " until further investigations have been to the relations of the various gas scales at high temperatures and as to the influence of the initial pressure and the effect of impurities and traces of water vapour in the employed , and until exact determinations been made up to high temperatures of the coeffi-cient of expansion of the material used as thermometric reservoir , we thin that for the purposes of high-range thermometry a scale by the parabolic formula from that of the platinum thermometer will suffice .
In the present state of our knowledge any attempt to improve on such a thermometric scale would be attended with such uncertainties as would probably render it futile Since that time , however , a substantial advance has been made in our knowledge , direct determinations of the expansion of porcelain up to high temperatures having been made by different observers , , Mr. BEDFORD , at Cambridge , and Messrs. HOLBORN and DAY at the Reichsanstalt .
A discussion by Dr. CHAPPUIS of the results obtained by these ) servers and their influence on high-range thermometry is found in the 'Philosophical Magazine , ' ( 5 ) , October , 1900 , and February , 1902 .
An examination of the difference formula for the platinum thermometer shows that it can only represent a ality over a limited rallge , the value of pt for a wire having a S of reachin maximum about 1700o , a value numerically not far exceeding such as may safely be attained .
It would not therefore be surprising if the formula which actually remarkably closely at low ranges should be found to ' Phil. Tr , vol. 194 , pp. 37-134 .
BEDFORD , ' Proc. Phys. Soc XVII .
, Part III .
, p. 148 , and ' Phil. ' HOLBORN and DAy , 'Ann .
Phys vol. 6 , 1901 , p. 136 .
OF NATIONAL LABORATORy .
.345 give erroneous results ) ttempel.atul.es well below the maximum to which the materials used in the construction of platinulm mometer can be subjected without injury .
The investigations dealt with in the present papeleeJl c out the National ) tory d two , and consist mainly tion of the of and the author the thermometel up to validity of the difference two thermometers made of wires of purity , by comparison of these instruments with the .
With these instruments were also compared sinrultaneously lnoj u , whose electl.onotive force } of had been ) special re the Reichsaustalt Il .
The etor employed for this is duplicate ) the one used ) DAY the .
It obtained the same , of Berlin , presented to the ) Sir NOBLL For this lnunihcent and for the kindly , and advice by the President of the Reichsanstalt , USCH , .
HOLBORN in procuring for us the the construction of electric naces , the laboratory is ) eatly indebted .
The instrument is specially designed for pid work at high tempel.atures , and was arranged so that measurements could be made with any desired initial pressure and with bulbs of different materials .
The principle employed by CHAPPUIS , in the two thermometers at Se vres , all the snrel n , depend upon the determination of ) single length .
though undoubtedly capable of giving by .
the most results , becolues somewhat .
of pressure needed .
For this reason , thel.efol.e , in the the nlanometer is al.ranged so as to measure directly the difference of height between the level of a very short metal point , to which the luercnr in the closed , fig. 1 , is adjusted , and the lllercury surface in the longr tube during the ]lleasulelnent\fnof ; g4 con Inunicates with the atmosphere by the tap H. * The tubes A and mean of cone j the lower part closed iron reservoir in the base plate of the apparatus , into which mercul.y can enter the reservoir by means of the tube itnd steel tap D. The fine adjustment of the height of the mercury to the in A is by steel with c -shaped head the bottom .
the ) ratus and working on a steel let into of the * In the form of the ) A ioined to the reservoir below ) a glass tap , through the sidc tube of which the of the , into .
It however , that a source of ngor in the csults coml)arisons VOL. CCIII .
Fig. 2 .
Joint between platinum nsck of reservoir .
metal cap ; ) , split metal collar ; plati- nun , this case drawn to small ] size for lower 4 or 5 ceIltims .
; asher soldered to capillnry .
Fig. 1 .
OF PHySICAL tTORY .
III .
The manometer scale , nearly metres long , is graduated into millimetres and carries a slider with vernier and fine adjustment screw by which the position of the mercury colunnll in could be read directly to 02 millim. Over the part of the scale used in these experiment , no corrections were required .
Accurate setting of the mercury to the point in ] ) dosed limb of the manometer facilitated by the use of it short-focus telescope to stand .
Measurements of the ] .
of this point on the scale , as given ernier readings when the hting c of the slider in line with the point , were taken { requently th ] hout t. The nsfer of leyel eiiected by Quincke lmicroscope fixed on a table which could be attached to the thermometer supports .
different the ) to tube A having for different reasons several tinleH rebnilt .
For some early up to of ] vere fixed capillary stem ched b fusion .
the later series at lres reservoirs of Berlin l silnilar to those employed by , and HARKER , were kindly ordered for ns ) ] the Berlin .
They were made in manner described ) HOLBORN DAY in their paper in 'Wied .
Ann. , ' , pp. 830 , 8.31 .
For attachment of the ' to the ) tintlm c ] .
to the nnonet e joints made by or cement too to be trusted for long periods , so the joints were in ] cases in in fig. 2 .
brass washer soldered on to the platinum pillary is clan ) ) tight against the end of the porcelain or } neck of the ) hollow ] ) which on to two of ) ) collar , a thin leather , on which smeared a little lnarine glne , being to make the joint .
collar is prevented from sliding ) the neck by fitting closel.y into rounded channel cut into the glass or by a small wheel .
) of.joint perfect satisfaction .
lcing lost the ) cing f ) the mcrcnry ) of the recurrence of this nvoided ) the snbseqnent ddit oement embracing the of the , it extrcmoly difficnlt to the ication to be frcc enough to permit of its constant use shutting .
commnnic tion b ) ccn the the bulI ) and at the snmc to prevent slow of into A the tho bulb was under low pressnre .
The was therefore nlodified 1 into ) plain stcel and introduc.ing a smallcr perfcctly into the side ) it not needed after the ) onoe filled .
onditions the ngement C be kept perfoctl gas tigh DR. J. A. HARKER ON RATURE STANDARDS IV .
fs .
The barometer used was a Fortin with inch tube , made by HICKS and standardised at Kew .
It was compared from time to time with an old Royal Society 's standard and found to have the same correction , millim. The temperatures of the mercury columns in the gas thermometer were obtained by means of two auxiliary thermometers disposed at convenient heights , having their bulbs immersed in short test tubes filled with mercury .
These are called ( 2 ) and ( 3 ) the exa mples of gas thermometer readings given ] ater .
The temperature of that part of dead space over the mercury tube A is given by a thermometer inserted into the metal stopper carrying the reference point .
This thermometer is labelled ( 1 ) .
The temperature of the platinum is kept constant by surrounding it a of rubbel tube , through which a circulation of water from a supply vessel kept in the room is maintained during the experiments , the temperature of the flow being given by the thermometer ( 4 ) .
In making the correction of the pressures of the various mercury columns to latitude and sea level , the latitude of the thermometric laboratory was taken as and the height above sea level of the cistern of the barometer as 10 metres .
The relation V. of the The nitrogen used in the earliest experiments was obtained from atmospheric air by the ordinary method of absorbing the carbonic acid by potash and the oxygen by hot copper , but for the sake of continuity with the work of CHAPPUIS and the author , in which chemical nitrogen was used in all the high-temperature experiments , the method of preparation employed by them was used in all the later work .
Into a flask containing 100 grammes of potassium bichromate , dissolved in air-free distilled water , was introduced gradually by means of a tap-funnel a strong solution containing 100 grammes nitrate of soda and 100 grammes nitrate of ammonia .
When gently heated on a water-bath this mixture gives off a steady stream of which is passed over hot copper and copper oxide to destroy any oxides of nitrogen it may contain , and collected over weak boiled solution in a gas holder .
From this it passes through solutions of silver nitrate and strong potash , boiled sulphuric acid , and over solid caustic potash to a larger -tube phosphoric anhydride , where it is left several days before use , or is collected in larger quantity in the dry gas holder over mercury .
The tightness of all the joints and taps was ultimately made so perfect that no measurable inward leak of air could be detected by the attached manometer , even when the whole was left standing under reduced pressure for some months .
A sketch of the gas preparation apparatus is shown in fig. 3 .
OF V. :350 .
J. A. ON ATURE STANDARDS VI .
of the The experiments of HAPPUIS and the author , 'Phil .
rans .
' vol. 194 , p. 93 , showed how important it is that ) revious to and during the ] of the reseryoir with the whole old ) ted to a temperature ag that with .
of err dur it ] to obtain an accurate value for a the nbourhood of the -point of sulphur unless it had been previously heated for considerable period to a ature considerably higher than this .
Although very little is known as to the behaviour with change of telnperature and pressnl.e of the fihus of condensed by their surface tension adhere closely to the of lass or pol.celain vessel , clnd how they influence the coefficient of expansion of gas enclosed in it , it seems at any rate isable to remove the existing fil ] in to be used high-temperature work as far as possible by repeated heating and evacuation .
Accordingly for each new filling during this work the procedure was to heat the reservoir very strongly , exhausting it ( and the connecting tubes leading to the pump so thoroughly that an electrodeless -tube attached by side tube owed stroI green fluorescence both when the reservoir was hot and the next after cooling .
This only attained after much laboul by taking special care in the anipulation of all the taps and joints in the circuit , nd by blowing together into one piece the whole of the long canal leading from the gas thermolneter in the lnain laboratory to the pump and gas ion apparatus in a small room on an upper floor .
After the had stood this test the gas was filled into the reservoir to 100 to 200 millims. pressure at least three times in each case before the final filling and pressure adjustment were made , the bulb being heated while as strongly as expedient in the particular case .
In 1898 and 1899 , and HARKER , in the absence of precise data as to the expansion high lres of of Berlin ) orcelai they -ployed in their gas tbermometer , year o to content themselves with extrapolation of the expression for the expansion obtained by them from determinations made in the Benoit-Fizeau dilatometel between the limits and , confirmed by further ts made a weight thermometer method .
The values for the expression they employed seemed the time to ) amply confirmed by the measurements of HOLBORN and WIEN , lllade in 189 which gave for the mean coefficient of linear expansion between the limits and 60 the value , the value calculated by HAPPUIS and HARKER for the same temperature limits being .
On the other hand , the later experiments of HOLBORN and DAY made on Berlin porcelain rods heatod horizontally in an electric furnace show fairly conclusively that the expansion of porcelain cannot be represented by any simple function for more than a OF NATIONAL few hundred that ) be applied outside the of actual experiment ithout risk of ions e To obtain the most the )nsion of Berlin porcelain , of ) luethods ovel the to ) ined tures 1 out ) ClfAl ] in paper in the in ' nded to in of HOLBORN .
too be ) just eell confirlled of ) in 4of the der llstalt , ' vllosc 1 to in FizealI dilatometer in close of I have taken values of dilatation veel l calculated of a reservoir .
of nd whence v of olume . . .
In the work of and account taken of of the bulb by changes of internal value of of olulne w deduced tions external ) ssure on the reservoir ) .
With the uyed the whole is only very small , a change of 1 metre in producing a change of about of the whole v of resel .
at ordinary temperatures .
The of bigh changing elastic constants of the porcelain here involved , quite , it deemed to onnit this correction for the ) escnt .
J. A. HARKER ON TIII ermometer .
tinum t chosen for the comparison at high temperatures with the one of six constructed in 1902 at the laboratory the in the report of the ndards ( nittee for 1903 , the oldel one , lettered which had been elnl)loyed by and in their work at Sevres .
Since the date of the Sevref$ ) in 1898 therlnonleter h ) ) eserved as a standard and used coml ) tively seldom .
Its istic dimensions Length end of pol.celai tube to below the wooden collar , centims. ' of tube , The head of the old pattern with , the leads being entirely of platinum , inch , anl the " " ) \ldquo ; wire , incl ) diameter .
Thel.lnometer BA is similar in to ' and HARKER'S thermometers , illustrated in ) , the head being anged to be capable of cuum f considerable time .
follows :\mdash ; Length of tube to under head centims , diameter of porcelain tube Length of illica cross ]llillins .
Thickness of " " bulb\ldquo ; inch .
4 ' lead\ldquo ; , , , Both })\ldquo ; leads constructed from the stock of wire of extra high by JOHNSON and for the British tion .
Details as to the of these itish AtioI 1 are given in the endix to the of the British Association , 1903 . .
The resistance box used was the one ally described by in ' Natnre , ' November 14 , 1895 , pp. 39 to , which subsequently been modified by the Caml)cientific Instrument by substitution of Doulton-ware plug holders the usual blocks to which the individual resistances were connected , In this form the box was used by CHREE in his investigations described in 'Proc .
Roy .
, 11 .
3-58 .
the period covered present work the coils were standardised in the manner described by RKER and ( loc. cit. , p. 5 and great was taken oughout as to the cleanliness of the plugs and to prevent the accumulation of the black deposit which ] a tendency to form in the plug holes .
The box unit is very exactly ) the nine coil values anged on the binary scale OF THE NATIONAL PHYSICAL LABORATORY .
from 5 up to 1280 units , with an additional coil of 100 units for the determination the fundamental interval .
The coils are of fine platinum-silver wire wound into loose hanks not paraffined , so as to follow with as little lag as possible the tempera- ture of the massive copper tank in which they are placed .
The working of the resistance box has under the conditions which obtained for this work been sufficiently good to render unlikely the introduction of any systematic errors .
The sensitivity the platinum thermometer bridge was such that even at the highest ranges a change of about C. could be detected , though an accuracy of at 1000o C. is probably about the limit in the absolute measurement of temperature even under favourable conditions with resistance coils of platinum-silver , such as those employed heye , with a temperature coefficient of 00026 per degree C. The flexible leads which were never detached from the thermometers during any series were carefully checked before and after the experiments , and the possible checks , got by using different combinations of coils in measurement of same temperature , always agreed reasonably well .
XI .
The Potentiometer for Thermocouple surements .
The potentiometer used in these was specially and made in the laboratory for accurate thermocouple work , and was described in detail in a paper in the ' Phil. Mag. ' for July , 1903 .
It is direct reading , and is arranged so that for Fig. 4 .
Connections to potentiometer .
ordinary thermoelectric work where the maximum electromotive force to be measured is of the order of 02 volt the individual settings could be made to about icrovolt if required .
A diagrammatic representation of the connexio1ls is shown fig 4 VOL. CCIII .
\mdash ; A. 2 a plall of the top of the instrument in fig. 5 .
It is designed with a view of rendering possible the use of a short slide-wire of large cross section .
The balancing coils , on which the fall of potential is adjusted to a definite value , are in two rows , the centre GALV THERMO JUNCTION5 .
ST .
CELL ACCRES Fig. 5 .
Plan of potentiometer .
row of the box consisting of 20 coils of of an ohm each .
In series with these is a second row , immediately behind the bridge-wire , consisting of 11 coils of of an ohm each .
By means of an arrangement of thick copper bars connected with the ends of the slide-wire , which has a total resistance of 02 ohm , any two adjacent coils of this latter series may be put in parallel with the slide-wire .
The 11 coils of 01 , two of which are thus shunted , are always exactly equivalent to 1 ohm .
ordinary thermoelectric work the fall of potential along these two sets of coils is adjusted so that each of the back row represents 1000 microvolts , each coil of 01 ohm being therefore 100 microvolts .
The slide-wire , 200 millims. long , is provided with a divided scale on which part of its length can be easily estimated .
It will be seen that the slide-wire thus connected acts like a vernier to the small coils .
The adjustment of the electromotive force is made by a standard Clark or Weston cell and the auxiliary set of coils in the back row , a feature of the instrument being that without any external alteration either form of standard may be used at will .
The five coils to the left are permanently connected in series , but are arrallged so that any coil may be cut out of circuit when required .
Their values are 100 , 40 , 1 , 5 , and 5 ohms respectively .
Those to the right are a set of 10 simple series coils ohm each , arranged so that a connexion can be taken from any one of them to OF THE NATIONAL PHYSICAL LABORATORY .
the long copper bar just in front .
Suppose we wish to use as a standard a Clark cell whose electromotive force at the prevailing temperature is volts .
It is obvious that we may make the fall of potential over the 1 ohm balancing coils have the desired value of 1000 microvolts by putting into circuit coils 100 , 40 , 1 , 5 , and three of the series , leading from the third hundredth by means of the copper bar to No. 17 of the set , when altogether we shall have ( in the back row , in balancing coils and bridge-wire , making in all ohms .
Should a Weston cell having an electromotive force of volt be substituted for the Clark , the only alteration necessary would be to short-circuit coils 40 , 1 , and 5 , and to move the connector from the third to the sixth of the set of hundredths .
The compensating current is furnished by a small secondary cell , in series with which is a dial resistance capable of variation up to 200 ohms by steps of 005 ohm .
The four thermojunction cuits provided are connected to a selector switch , by means of which each Qsively or any two of them connected in opposition may be brought into circuit , and the change-over from the standard cell connexion required in the preliminary adjustment is made at the two-way switch at the front left-hand corner , by means of which the galvanometer may be put into the circuit desired .
Constructional details are given in the paper referred to .
All the coils employed are of selected manganin , carefully annealed , and all connexions are made by mercury caps and copper pieces , the only metals employed anywhere in parts carrying current being copper and manganin .
The values of those coils in the box which were used in this work as determined at the conclusion of the comparison experiments are given in the following table .
Any alteration in their relative values which had taken place since the first standardization is undoubtedly so small as to be quite negligible compared with other errors in thermocouple work .
* It is obvious th in } ) down to obtain a standard of thermal electromotive force\mdash ; 100 or 1000 microvoIts\mdash ; so long as the relative values of the coils employed remain the same , their absolute value is of no moment .
A matter of great importance , however , is to measure the value of each coil in exactly the same way as it is used .
Accordingly , for this standardization a potential method of measurement was employed in all cases , the current and potential leads being connected exactly as in actual work .
COIL Values in ohms at C. Coils in main row nominal value each .
No. Mean value of the 20 tenth-ohm coils Total , , , , slide-wire set of Il coils shunted by slide-wire Maximum variation in the resistance of slide-wire set for different positions of the slide-wire connector Value of nominal 100-ohm coil The auxiliary set of 01 ohm coils for temperature compensation were all found to be within part of their face value .
XII .
Standard of Electromotive Force .
The standards of E.M.F. used with this potentiometer were two milar H-form cadmium-sulphate cells with saturated solution of the type employed at the Reichsanstalt , and made up as part of a large batch of similar ones by Mr. F. E. SMITH in 1902 .
From Mr. SMITH'S measurements as to the relation of the E.M.F. of these cells to the standard Clark cells of the laboratory , and from other data , it is practically certain that the error committed in assuming their E.M.F. to be identical with those at the Reichsanstalt is not greater than 1 part in 10,000 , which corresponds to a tenth of a degree at 1000o C. with the thermojunctions employed .
For the E.M.F. of each of these cells , which throughout the work were never found to differ by more than volt , the Reichsanstalt official value , namely volt at C. , was assumed .
The following table gives the value of the total resistance used in the potentiometer at different temperatures to adjust the E.M.F. over each coil of the main row to exactly 1000 microvolts : The table of the values of the coils in the potentiometer shows that the actual resistance corresponding to a nominal value of ohm is .
The relation of this to the mean value of the main set of tenths , namely 100024 ohm , is well OF THE NATIONAL PHYSICAL LABORATORY .
within 1 part in 10,000 of its nolninal value , and an inspection of the table shows that the cumulative effect of individual coil errors is very small .
No corrections were therefore applied to any of the potentiometer readings .
Special precautions were taken to avoid the effect of temperature variation both on the accumulator furnishing the compensating current of 01 ampere and on the standard cells , both being placed in double-walled boxes surrounded by a thick layer of cork clippings .
Under these conditions the daily temperature range in the boxes was found to be reduced to about one-fifth of the value outside them , and the compensating current could generally be kept to within of its value for an hour at a time without adjustment .
XIII .
The Therraojunctions .
The thermojunctions used in this research were composed of pure platinum with platinum containing 10 per cent. of rhodium , and were all 6 millim. diameter .
They were obtained from HERAUS of Hanau , through Dr. HOLBORN of the Reichsanstalt , and were compared by him with the standard junctions of the Reichsanstalt at a number of fixed points .
In a letter to the Director of the Laboratory , Dr. HOLBORN says : " " Two elements were compared at four points , the melting-points of , Sb , Ag , and Cu , and gave the following results:\mdash ; Microv .
lts .
Before commencing comparisons with the gas thermometer , three independent determinations of freezing-point of silver were taken in an electrically heated crucible furnace with one of these junctions N.P.L. 2 ; in these experiments two different observers took part , and the three results were : microvolts .
The mean of these : 9087 , agrees very closely with the datum given above .
Junction N.P.L. 2 was selected for the comparison , while N.P.L. 1 was reserved as a * In all experiments with thermojunctions here referred to , it is fo be understood the cold junction is at C. master standard , was ared with No. 2 in a } ) ecially arranged electric furnace at tures up to 1200o C. before and after the investigation .
The results of these risons show conclusively that Junction No. 2 bas not tered any material Iteratio during its protracted heatings at high temperatures .
* .
of the two junctions , as given by a comparison made at the close of the work , is shown in the table:\mdash ; Microvolts .
\mdash ; No. No. 2 .
in XIV .
Formnloe for Thermojunctions .
From the values of the E.M.F. of N.P.L. 1 and 2 , as determined above by HOLBORN , a formula involving two powers of the temperature was calculated by least to give the relation between L.M.F. and temperature to represent the 's scale .
The formula gives at the four melting-points given above much smaller than their probable The formula for HOLBORN and D standard junction ( llly the vahles of the temperature tained from the gas thermometer with of idium and employing the revised data for the expansion of this erntures , is table gives side by side the E.M.F. of HOLBORN ctnd junction ) , and of our own junctions at temperatures above the of second set of risons the lstre of some inches uf the tinnm wirc decidedly impired , did not to nied by the sm lest c the .
of the jutlction .
lIoLBORN , WIEI ) .
Annalen , ' 1900 , vol. 2 , p. ) .
The change from the tho }nsion of the bnlb involves correction of the scale to 1000o .
OF THE NATIONAL PHYSICAL LABORATORy .
It will be observed that both the formulae just quoted do not apply at lower temperatures , being nearly C. in error at , and that therefore extrapolation downwards even over a narrow range is llot permissible .
The error of the mula for NPL 2 was determined to be at XV .
of the Fixed Points Sulphur Boiling-point .
The determinations of the fixed points and for the gas and platinum thermometers were made in baths specially built for each kin of instrument .
The ice-points were taken glass vessels of a capacity of about 6 and 3 litres respectively , consisting each of an invel.ted glass } ) -jar with draining nent below , and surrounded by a thick packing of cork clippings .
Very little melting of ice took place even in 12 hours , the upper surface of the ice being protected by a thick felt covering wrapped round the stem of the instrument .
The block-ice previous to use was always well washed and finely divided an ice plane , and was repeatedly tested for dissolved impurity , method adopted } ) ' to ascertain the amount of chlorine present by addition of silver nitrate to the ainiI t It was found satisfactory , except on one occasion .
The steam-point apparatus for the gas thermometer was of the usual type with concentric tubes , but was arranged to be easily nanged from the vertical to the horizontal position by suitable couplings of wide compo tubing , connecting it to boiler .
The steam- and zero-points for any series of comparisons were always taken with the reservoir in the position in which it was used in that series , and in all cases the amount of stem emergent was made as nearly as possible the same in the comparisons .
The steam bath as arranged for the horizontal position is shown in fig. 6 .
The bulb of the gas thermometer , resting on a small piece of cork , occupies the centre of the inner tube , through which steam brought direct from the boiler by a well protected wide tube is circulated .
The steam issuing from the outer jacket is Fig. 6 .
Steam bath .
condensed and returned to the boiler as shown .
The excess of the steam pressure over that of the atmosphere , which in these experiments was seldom over 1 millim. of water , is indicated by a small graduated water gauge .
The sulphur-points of the platinum thermometers were taken in the usual manner in the well-known Callendar form of boiling-point apparatus .
The only departure from previous custom being that for the glass boiling tube was substituted one of thin weldless steel , which is more durable and can be heated up quickly without being removed from its asbestos cover .
Careful comparisons of this form with the older glass apparatus showed no systematic discrepancy .
For the sulphur boilin-point CALLENDAR'S old value C. at normal pressure *An ordinary iron tube such as gas or steam pipe cannot be used for this purpose , as owing to conduction from the flame of the burner upwards there is a tendency for the sulphur vapour to become superheated .
OF NATTONAL ICAL LABORATORy .
was taken .
The sulphur-points were always taken on days when the pressure was not removed from 760 millims. to eliminate the uncertainty as to the coefficient to use when reducing to llormal pressure .
XVI .
Electric Furnaces .
Two different electric furnaces were } ) loyed in this work .
Their dimensions were ] , but they differed in that in the first the heating-wire was wound uniformly , and in the second an approximation to a logarithmic spiral was made at each end , the turns being gradually crowded , so that the cooling effect of the ends was in a great measure compensated by the additional heat supply , Both furnaces were wound with wire of pure nickel about diameter .
The heating current was supplied from a special battery of 56 accumulators reserved for this purpose , which was divided into four groups of 14 cells , capable of being coupled in series or parallel , as desired .
A set of large well-ventilated manganin resistances , formed of two No. 9 wires in parallel , and capable of carrying 100 without undue heat , was arranged so that the external resistance of the circuit could be altered by steps of 025 ohm up to ohms , thus enabling any desired amount of energy to be put into the furnace at will .
The construction of the furnace and the disposition of the different instruments within it is shown in fig. 7 .
The nickel heating-wire is wound upon the inner tube of unglazed biscuit porcelain , and in order to prevent the turns becoming short-circuited when hot , the whole of the wound portion is covered with a thin layer of " " purimachos\ldquo ; which is baked on at a moderate heat .
The leading-in wires are doubled or trebled in all cases .
The bulb of the gas thermo- meter is supported on a small bridge of fire-clay resting on the furnace bottom , and the standard platinum thermometer and thermojunction are arranged as shown , great care being taken that neither the wires of the junction nor the thin porcelain VOL. CCIII.\mdash ; A. 3 .362 .
J. A. HARKLR STANDALDS capillary tubes used to cover them shall anywhere touch the furnace wall .
As previous experience with cras and platinum thermometers , whose walls were of porcelain , had shown how very much more slowly the transfer of heat took place through this material than through metal or glass , even when surrounded by a stirred liquid , it was judged preferable to make a comparatively small number of hightemperature experiments , in which great constancy of temperature was attained for some time previous to and during the observations , rather than to attempt to obtain mean values from more extended series under less perfect conditions .
With this object , recordel was connected to a second platinum thermometer placed in the furnace , and , dul.ing the adjustment of the temperature and the comparisons , records from this instrument were taken on an open scale , The use of the recorder greatly shortened the time necessary for the establishment of a steady temperature by guiding the observer as to the manipulation of the resistances in the heatin circuit .
addition to the large set of resistances in the heating circuit , a set of coils of 01 ohm each was placed close to the recorder , and it was found that a change of one step on this set made all the difference between a steady state and a gradual rise or fall in the temperature of the furnace when equilibrium had been nearly established .
desired , it was quite easy to keep the furnace temperature constant to about a fifth of a degree for half-an-hour at a time , at temperatures as high as 1000o C. , but in most of the experiments the temperature was intentionally allowed to rise very slowly .
Without these precautions , comparisons between instruments of such widely differing " " lag\ldquo ; as a bare thermojunction and a gas thermometer with porcelain reservoir , whose walls were 2 millims. in thickness , would have undoubtedly been liable to serious error .
XVII .
Correction .
In order to investigate the distribution of temperature throughout the space filled by the gas thermometer bulb , which was about 130 millims. in length , a pair of thermojunctions , quite independent of the standard , were arranged so as to measure the temperature difference between the centre of the furnace and points further out , and so obtain a correction to be applied to the readings of the gas thermometer , to educe its indications to what it would have registered had the whole of the bulb been at the same temperature as the middle poin XVIII .
of For this purpose a thin wire junction of platinum with platinum iridium was chosen on account of its great sensitiveness at high tempel.atures .
This was made up to work differentially , and was composed of a piece of platinum iridium between two pieces of platinum , thus forming two junctions , which were both placed in the hot space , entering the ful.nace from opposite ends .
The wires were stiffened by threading OF NATIONAL PHYSICAL LABORATORy .
them through thin porcelain capillary tubes , and the two junctions of the platinum with the copper forming the rest of the circuit were placed together in ice .
The .
given by this element for a difference of 1o between its two hot junctions is given in the table below .
* This is obtained by direct comparison over the whole range of a simplejunction up of wires from the same reels , which had been similarly treated .
ture in oC .
Difference for 1o in microvolts .
One of thejunctions was carefully laced and kept at the middle point of the thermometer bulb , and the second was arranged so that it could be pushed backward and forward into positions 2 , 4 , 6 , and 8 centims. to the right and left of this point , observations being made of the differential .
produced in each of these positions .
From a number of such observations made after the different comparisons and spaced over the interval .
to ] C. , curves were constructed showing the distribution of temperature over this space for each of the furnaces used , and by measurin these curves the difference between the average temperature of the bulb and that of the central point , where the standard thermojunction was placed , could easily be found .
For the compensated mechanical centre was found not to quite coincide with the position of highest temperature at higher ranges , though over the lower part of the scale the curves were practically symmetrical .
The ctions obtained by this method from the different tions were plotted as ordinates against furnace temperatures as abscissae , and from the mean curve thus obtained the values were deduced as the meall corrections at different temperatures : * In ctice , when the iunctions close )gether i the furnace possible contact , owing to the small secondary effects rising from Ylsymmetric hting of the junction , the E.M.F. round the circnit was not lways zero .
Whcn , howcver , the jll1lctions were so supported as to nowhere touch the furnace wcdl , the effect quite and allowed for in each cass .
DR. J. A. , ON THh ATURE STANDARDS From thiQ table the numbers in the column headed " " furnace correction\ldquo ; in Series II .
and III .
have been deduced by interpolation .
For the uncompensated furnace used in the earlier experiments the correction is uncertain to about C. , and the results are , therefore , only given to this accuracy in the final column representing the " " corrected gas thermometer temperature XIX .
Method of C'alculation .
The observations taken with the gas thermometer calculated according to the usual formula .
Let pressure at and let and be temperatures of " " dead space\ldquo ; with bulb at and at temperature Let relation of ) volume of the whole " " dead space\ldquo ; to the volume of the bulb at coefficient of nsion of the gas at constant volume between and the mean cubical coefficient of expansion of the porcelain bulb between and ; then therefore \mdash ; For a given filling is a constant if the zero-point remains constant .
The values of obtained from the steam-point determinations were calculated from the same formula , inserting the value of obtained from the Regnault-Broch table for the boiling-point of water under the prevailing pressure , as given in the example .
OF THE NATIONAL PHYSICAL LABORATORY , -POINT ; 3 .
tpri116 , 1903 .
It will be observed that during the 10 lninntes covered by the servations the barometric pressure fell rather than lnillim .
while the scale rose a corresponding amount .
On this occasion the variations in the temperature of the jacket water were so abnormally great during the preliminary period before the experiment , that it was decided to dispense with the circulation in the space round the platinum capillary and take the whole dead space as being at the temperature of the part below the stopper in che closed limb of the manometer , given by thermometer No. 1 .
STEAM-POINT ; 3 Apri116 , 1903 .
server , J. A. XXI .
of nometer Constant Determinations .
As an example of the kind of accuracy attained in the determination of the constants of the gas thermometer , the individual values of the zero- and steam-points taken before the comparisons in Series II .
is given along with the readings taken in one complete determination of each .
For calculation of experiments of this kind , where no gradual systematic drift is cted , instead of utilising for calculatin the individual values of the the single determinations of the ice-point , the mean of the ice-points is take1l for bhis purpose .
The difference between the value of found before and after the comparisons is within the limits of experimental error .
It will be seen from the table appended , which gives in an abridged form the calculation of the gas thermometer temperatures in the 13 experiments of Series II .
, that the difference has , however , been treated as real , and assumed to vary with the time .
A more serious change has , however , taken OF THE PHYSTCAL LABORATORY .
.367 place in the quantity of gas present in the bulb .
The change in this case has also been assumed as proportional to the time , since before the first experiment was made the whole furnace was maintained at a high erature for some time , and there is no sufficient evidence that the change occurred otherwise than regularly .
So far as could be found by testing on the pump after the conclusion of the series no leak could be discovered , even when the bulb was at a full red heat .
CONSTANTS for Gas Thermometer .
Series II .
I. Before the comparisons:\mdash ; Ice-points .
Steam-points .
II .
After comparisons:\mdash ; Ice-points .
Steam-points .
DR. J. A. HAliKER ON STANDARDS ; a@ .
OF THE NATIONAL LABORATORy .
XXIII .
Method of calculating Cornparison xperiment .
As an example of the method employed in making and calculating out an experiment , the results of the observations made in Experiment No. 13 of Series II .
are given in full .
During this experiment the steadiness of the temperature of the furnace was perhaps a little above the average , but the room temperature and that of the resistance-boxes and of the various mercury columns of the gas thermometer and barometer were rising more rapidly than usual .
According to observations taken on the thermojunctions and platinum thermometer , the steady state had been maintained for about 10 minutes before the first recorded reading on the gas thermometer was taken .
As only two observers were available , one of these took alternate observations on the gas thermometer and thermojunctions , while the second took the platinum thermometer readings .
By graphic interpolation the mean thermojunction readings , corresponding to the times at which the other instruments were read , were obtained and are given in the example .
The readings of the gas thermometer are made as independent of one another as possible by lowering the mercury each time before a setting , raising it again slowly so as to make a new meniscus .
The calculation to the accuracy here necessary of the air temperature corresponding to definite platinum temperatures is somewhat laborious , if the formula has to be applied for each obseryation .
The raphic methods used by HEYCOCK and cannot easily be made sufficiently accurate .
Since the value of S for pure platinum wire has been found to be , varying from one specimel to another within very narrow limits , the most suitable method of effecting this conversion was found to be to construct a table giving , for a sufficient number of points , the value of for given values of pt when S. A second table gives the correction to appl to the value of thus obtained , if the S the dard yalue by a small amount .
* The letters AB , BA refer to the position of the revcrsing switches ding to the potentiometer .
Owing to small Thomson Peltier effects , there small ] iffercnce b the two positions .
In this case it is rather ) the vorage .
There was no difficulty in setting to ltlicrovolt , so the readings are given to this figure , though it is not considered in measurement .
VOL. CIII. .
3 GAS Thermometer Readings .
Expt. 13 , May 1 , 1903 .
Observer , J. A. corr corr .
Mean temperature of dead space l4 . .
01230 , where taken from iunction l004 A second approximation , employing in the dilatation term , makes the final gas thermometer temperature C. THrRMOJUNCTION Readings .
Expt. 13 .
bseJve J. A. OF THE NATIONAL PHYSICAL LABORATORY .
7 ] From these observations plotted , the following were deduced as ] thermojunction readings simultaneous with the gas thel.mometer readings:\mdash ; .
From table of E.M.F. of junctions the value of the .
for 1000o C. and the difference for microvolts .
Whence the temperature corresponding t ) ( thermojunction ) PLATINUM Thermometer Readings .
Expt. 13 .
Observe ; W. Summary.\mdash ; gas ( found ) Furnace correction gas corrected thermojunction platinun ) thermometer XXIV .
First Set of Comparisons .
October , 1902 .
millims. Mean ice-point of gas thermometer before after , , , , Mean value of before after , Platinum thermometer ohms before commencing , , , after seyenth experiment , , , at conclusion , F.I. before commencing , at end , mean of all observations .
FIRST Set Comparisons .
Furnace wound uniformly .
and Gas Thermometel from from .
ected g RESULTS of Comparison arranged in Order of Ascending Temperature .
OF THI NATIONAL PHYSICAL LABORATORY .
\amp ; co 'd Set of risons .
April and May , 1903 .
millims. Mean of gas thel.mometer before comparisons after , , , Mean value of determinations before comp risons after , , , In this case there is a change in the zero-point greater than that observed in the first set and in the opposite direction .
In allowing for this , the change was supposed to have been proportional to the time .
Platinum thermometer ohms before commencing comparisons , , , at conclusion of F.I. before commencing at conclusion of mean of large number of observations extending ovel six months .
The platinum thermometer constants of this set are expressed in terms of a slightly different unit from that employed in first set six months earlier . .
OF THE NATIONAL PHYSICAL LABORATORy .
Third Set of Comparisons .
May and June , 1903 .
millims. Mean ice-point of gas thermometer before comparison after , , Mean value of from determinations before after 003661 , , Platinum thermometer ohms before commencing comparisons , , , at conclusion of F.I. before commencing at conclusion of mean value over a long period .
DR. J. A. HARKER ON THE HIGH-TEMPERATURE STANDARDS .
OF THE NATIONAL PHYSICAL LABORATORY .
XXV .
Concl usions .
These three sets of experiments show that : ( 1 .
) The readings of platinum thermometers and , which may be taken as representative samples of pure platinum , when reduced to the air-scale by CALLENDAR'S formula , employing his value for the boiling-point of sulphur , are in reasonably close agreement with the results given by the constant-volume nitrogen thermometer employing chemical nitrogen under low initial pressure and using the revised values for the dilatation of porcelain .
The divergence of the two scales only exceeds the probable error at the higher part of the range .
( 2 .
) The platinum thermometers , and the thermojunctions representing the temperature scale of the Reichsanstalt , based on measurements made with the gas thermometer with bulb of platinum iridium , agree very closely , the thermojunction giving apparently a slightly higher value throughout the range covered by the experiments .
As the result of these comparisons is to justify the application of parabolic formula up to 1000o C. , the tables previously alluded to for deducing the value of from pt over the whole range for which platinum thermometers are useful to , are given .
From to the yalues are giyen for each degree , from 20 to 1100o for each For the figures given in Table I. , in their final form , calculated by a method which renders unlikely the accumulation of errors greater than 1 unit in the last figure given , I am indebted to Mr F. J. SELBY .
I have also to acknowledge my indebtedness to Mr. W. HUGO , who assisted the comparisons by taking the observations with platinum thermometers , and especially to the Director of the Laboratory , Dr. GLAZEBROOK , for his continued interest and advice on many points .
* In the paper as printed this table has been shorte1ted by only giving the value of for intervals of 1o over the range to VOL. -A . .
J. A. HARKER ON THE H-T MPERATURE STANDARDS XXVI .
TABLE I. OF NATIONAL PHySICAL LABORATORY .
TABLE I.\mdash ; continued .
DR. J. A. HARKER ON THE HIGH-TEMPERATURE STANDARDS TABLE I.\mdash ; continued .
OF THE NATIONAL PHYSICAL BORATORY .
TABLE I.\mdash ; continued .
T. for parts .
erencel 0 .
1.023 .
J. A. ON THE MPERATURE STANDARi ) TABLE I.\mdash ; continued .
Proportional parts . .
5 .
515 .
7 .
721 .
9 .
927 parts .
OF NATIONAL PHYSICAL LABORATORy .
TABLE I.\mdash ; continued .
pt .
T. XXVII .
IT.\mdash ; To the Change in for a , S.

