]\gt ; ' [ 119 ] IV .
On the Determination of the Boiling-Point of Sulphur , and on Method of Standardising Thermometers by reference to it.\mdash ; Experi- ments the vendish Lmbridge .
H. L. CALLENDAR , M.A. , Fellow of Trinity College , Cambridge , E. H. GRIFFITHS , M.A. , Sidney Sussex College , bridge .
Communicated by Professor J. J. THOMSON , F. R. Rscoived November 29 , \mdash ; Read December 18 , 1890 .
[ PLATES 7 , 8 .
] GENERAL SUMMABY or THE VoRK .
THE results of many careful experiments by different observers in the course of the past few years have shown that electrical resistance tbermometers afford the jcurate and convenient method of temperature through a very wide range .
The object of the following paper is to illustrate a method of the indica tions of snch thermometers to the absolute scale of temperature by means of a single observation at some known fixed in addition to and 10 C. The boiling-point of sulphur happens to be the most convenient to choose for this .
We have therefore made a very careful determination of this point by reference to the normai air thermometer , and have a tion of the method and apparatus which we have found most suitable for standardising resistance thermolneters by means of it .
a which one of us had the honour to make to the Royal Society in June , 1887 , " " On the Practical Measurement of Temperature it was shown that , if be the temperature by normal air thermometer , and be the temperacure by platinum-lesistance thermometer , the difference between them is very closely represented from to 70 C. by the formula , -pt@ . . . . . . .
( d ) .
The value of the constant 8 for a particular of wire was found to be *CALLENDAIf , ' Phil : Trans , 1887 , p. 161 .
22.6 .
91 alene ( benzophenone ( and sulphur as determined by CRAFTS and AULT .
Curves were drawn with these data for each mometer g the difference , , in terms of .
It was found the upper portions of these curves differed materially ( see Plate 8 ) from the parabolic formuIa ( d ) above given .
In order to compare the obtained with those deduced from the @-formula , a thermometer ( L ) was constructed out of a piece of the spiral used in the air thermometer experiments of 1887 , for which the value of the constant 8 was known to be Observations were taken with this thermometer in sulphur vapour in the same apparatus which had been ) for standardising the other thermometers .
The results gave as the platinum-temperature attained by the thermometer .
This gives , assuming @1570 .
The value thus found was nearly six degrees lower than that given by AULT .
This showed either that the value of @ was wrong , or the thermometer did not attain the temperatnre of the sulphur vapour , or that the number given by REGNAULT was too high .
But when ) number thus found was used to calculate the appropriate values of @ for each of the other thermometers it appeared that the values of for all the other boiling- and freezing-points , deduced by means of formula ( d ) from the observations with all the different thermometers , were in remarkably close agreement .
This suggests a very simple method of standardising such thermometers .
Assuming that the ( d ) formula holds good , it is only necessary to observe the platinum-temperatllre of sulphur .
with each thermometer , in order to find the appropriate value of @ , and to reduce all the platinum-temperatures to the air standard .
It was felt , however , that the value of @ assumed for the standard wire required verification .
The wire bad been very roughly treated , and used for all sorts of purposes since the experiments of 1887 , and might well have altered in the interval .
Two new thermometers , and , were therefore constructed out of the remainder of the old spiral , and were very carefully compared with the air thermometer at a temperature very near the boiling-point of sulphur .
Owing to the importance of this point , the experiments are very fully described in the following pages , Part I. The result agrees perfectly with that found in 1887 , and shows that the value of the 8-coefficient has not altered appreciabIy in the interval .
The apparatus used for this ( 'omparison , although useful as a constant high* GB IFI ITII , supra , p. 43 .
a Meyer tube , which proved to be more convenient for the purpose .
The platinum thermometers , which had been dised as above mentioned , were used for this determination with various screens and other precautions to guard against errors due either to super-heating of the vapour to of heat by radiation and condensation .
The experiments with the Meyer tube are described in Part IT .
They show conclusively that errors due to radiation , \amp ; c. , are of serious magnitude at this temperature , and that , in order to find the true boiling-point , it is not sufficient merely to insert a thermometer in the vapour , as we find has generally been the practice among previous observers .
According to our experiments with carefully screened platinum mometers , the temperature by normal air thermometer at constant pressure , of the saturated vapour of sulphur , boiling freely under a pressure of 760 mm. of mercury at C. and C.G.S. ( sea-level in latitude ) , is .
We have taken great pains to verify all the work of reducing the observations , and have carefully tested all the apparatus employed .
We believe that the above result may be relied on to the tenth of a degree .
The number given by REGNAULT is nearly four degrees higher than this .
His results have been so widely quoted and adopted that we have thought it necessary to add a s'uort criticism of his method , and to an explanation of the discrepancy .
A table is also given containing the boiling- and freezing-points of substances , reduced from the previous observations with the other thermometers .
* There can be no doubt that , even at low temperatures , the platinum thermometer is a far more accurate and sensitive strument than the mercury thermometer .
With suitable apparatus it is easy to read to C. , and here is no trouble about shifting of zero .
We have made a careful comparison of the platinum and air thermometers between and C. , The experiments are described in III .
, we find that the @-formula holds ately between these limits .
In setting up the apparatus and taking the observations , we have received great assistance from Mr. G. M. CLARK , B.A. , of Sidney Sussox College .
Our best thanks are due to him for many suggestions .
* GBIFFITIIS , ' Phil. Trans , 1891 .
MESSRS .
H. L. CALLENDAR AND E. H. ON TlfE PART I. of Platinum and Air at th -Point of Sulphur .
When we found that the preliminary experiments with thermometer made the boiling-point of sulphur so much lower than the value given by REGNAULT , we at first endeavoured to account for it by supposing that the wire had altered in the four years ' interval , or the previous thermometer experiments were faulty .
In any case they were felt to require verification .
An iron-tube apparatus was therefore constructed , into which an air thermometer could be introduced , along with two platinum thermometers .
The apparatus was so arranged that it could be kept at a steady temperature by a continuous flow of sulphur vapour for any length of time .
Two new platinum thermumeters , and were made out of the remainder of the wire of the old spiral , and were compared with the air thermometer in the iron-tube apparatus on two different days .
Description of the Air Thermometer .
The form of instrument employed is shown in Plate 7 , fig. 1 .
It was similar in every respect to that described and figured in the paper*above referred to , except that the bulb was formed of ordinary glass , and not of hard glass .
A piece of tubing , about cm .
in diameter , was taken , and cleaned in the usual way .
One end was closed in the blowpipe , and the other was drawn out into a fine capillary tube , a metre long .
This was passed through a small brass tube , the end of which was split into four for a length of 15 cm .
, and bound round the bulb with wire , to serve as a support and to protect the capillary tube .
The brass tube was firmly fastened to a wooden stand carrying the volume- and pressure-gauge .
The volume of the bulb was determined by water calibration up to a- marked zero- point on the capillary tube .
In the thermorneter care was taken to subject it up to the point to the temperature to be measured .
A short distance beyond this point the brass tubo was cooled by winding closely round it several turns of thin lead tubing , through which water was made to circulate .
The volume of the capillary tube , for some distance on either side of the point , was only .
per cm .
The uncertainty of a centimetre or two in the position of the neutral point would not therefore introduce appreciable errors .
As a matter of precaution a platinum wire was originally inserted in the brass tube , to take its mean temperature , but it was found that the corrections were always so small that they could be neglected .
A length of about 30 cm .
of the capillary tube was bent into the form of a narrow , shown at rved to give the end free play , and to protect the tube from strain .
* ' Phil. Trans A. , 1887 , p. 167 .
DETERMINATION OF THE BOILING-POINT OF SULPHUR .
The end of the capillary tube had been previously cut off , and ground flat to fit the end of the gauge-tube .
A short double funnel of ] was drawn to fit them both closely , and sealing-wax was run from either side .
This form of joint was generally used by REGNAULT .
It is perfectly air-tight if properly made , and is certainly quite unobjectionable in the case of a constant-pressure air thermometer .
of Gerrnan The expansion of the glass forming the bulb was determined by the method described in the previous pape using the remainder of the piece of glass tubing from which the bulb had been drawn .
The glass tube was enclosed in two coaxial tubes of brass , separated from each other by strips of asbestos cloth , and was heated in a thick iron tube in a long gas furnace .
The whole apparatus was supported on levelling-screws for focussing .
The gas furnace was screened with bright tin plate , and all the precautions were taken which have been previously described .
The mean temperature of the portion heated was determined by means of a known platinum wire ( the same piece as was used in thermometer L ) , extending down the axis of the tube , and supported at intervals by mica plates .
The ends of the tube were drawn down to.5 mm. diameter , to diminish conduction of heat , and to imitate as closely as possible the actual condition of the bulb .
They projected on either side for a distance of about 15 cm .
beyond the gas furnace .
A large and fairly rigid brass tube with end pieces projecting at right angles was supported , so that reference marks at either end of it could be brought into focus under the reading microsoopes close to the ends of the glass tube .
Its temperature , which was always very nearly that of the surrounding air , was taken by means of a delicate mercury thermometer placed inside it .
Corrections were applied for the small variations in its temperature , and the changes in the length of the glass tube were determined by direct comparison with it .
Five series of observations were taken with this apparatus on different days .
The highest temperature reached in each series was a little over 40 C. It was always taken first to avoid the permanent contraction .
After the first series , the tube was found to have contracted by per cent. of its length ; but towards the end of the experiments it appeared to reach a steady state .
The whole of the observations on the linear expansion of this glass were very closely represented by the formula This gives.for the cubical expansion between and 10 C. the factor * Phil. Trans A. , p. 206 .
thermometer , after it had been dismou nted , the conclusion of the whole series of experiments .
It was very carefully and completely filled with pure ercury , and placed in melting ice for an hour and a half .
It was then heated steam at C. , till some time after the mercury ceased to drop out .
The mercury which had run out was very carefully weighed and found to be grms. The weight of the mercury filling the bulb at up to the zero ( deducting the weight filling the known volume of the capillary tube exposed to the air temperature ) , reduced to vacuo , was found to be grms. , whence , the volume of the bulb at C. , is that the density of mercury at C. is grms. per If be the of the bulb at and C. , and the corresponding values of the densities of mercury , we have the following equation : The linear expansion experiments give ; whence REGNAULT'S experiments on the absolute expansion of mercury give for the expansion between and the value .
The value found by assuming the linear expansion is too large , therefore , by one part in a thousand only .
This degree of accuracy is amply sufficient for our purpose , as the whole correction for the expansion of the bulb only amounts to at the temperature of boiling sulphur .
Changes in the Volume of the Bulb .
The volume of the bulb was determined from time to time during the course of the experiments .
It is interesting to trace the changes which it underwent .
Before mounting , the volume of the bulb was found by direct calibration to be c.c. At the close of the experiments , after repeatedly boiling in sulphur , it was found to be .
The volume at intermediate times was determined by observations in ice and steam in the following manner:\mdash ; Shortly after mounting , the bulb was beated in a gas furnace , and dried by exhaustion at a temperature of about 25 C. Observations taken with it next day in ice and steam , assuming the volume of the bulb to be , gave the value as the absolute temperature of .
ice by air thermometer at DETERMINATION OF THE BOILING-POINT OF SULPHUR .
constant pressure .
This gives the value for the dilatation between and C. , which is approximately that found by BALFOUR STEWART , CHAPPUIS , and the most careful of recent observers .
At the close of the experiments the bulb was repeatedly exhausted while in boiling sulphur , and filled afresh with dry air .
Observations taken with it next day in ice and steam , before it was dismoumted ( assuming the volume of the bulb to be , gave the value , which is practically identical with that previously found .
We may , therefore , reasonably assume the same value of this constant in all cases .
We can then use the observations in ice and steam to calculate the volu1ne of the bulb on intermediate days .
This is the most accurate way in which the volume of the bulb can be determined without dismounting it .
the conditions are the same as those under which value of the constant is determined .
Any error in the value of calculated by this method will not affect the temperatures deduced to any greater extent than an equal error of observation in determining In a trial experiment in the iron tube apparatus on Tuesday , September 9 , in which the sulphur distilled over too fast to allow good observations to be taken , the bulb was exposed to a temperature of for nearly a quarter of an hour .
Next its volume was found to have diminished to .
Showing that , although it had been subjected to no stress whatever , the contraction was very rapid at first .
On Friday , September 12 , the thermometer was kept sulphur vapour for nearly two hours .
The temperature appeared to rise slightly owing to the contraction of the bulb , for a short time after the other thermometers had become steady .
After the experiment the was found to have diminished from .
to On Saturday , September 13 , the thermometer was again heated in sulphur vapour .
Several observations were taken , which gave no indications of change .
The bulb was then exhausted fifteeD times , and dry air readmitted .
In spite of the great stress to which it was thus subjected , at a temperature not far below its softening point , its volume was afterwards found to have diminished very little , namely , from to This seems to indicate that the bulbs of mercury thermometers may be practically freed from rise of zero by a prelirninary boiling in sulphur before filling them with mercury .
The sulphur can readily boiIed for hours a common test-tube or * REGNAULT gives for air at constant ] , for air at constant volume .
BALFOUR STEWABT ( ' Phil. Trans 1863 ) finds for air $ , t constant volume .
CHAPPUIS Archives des Sciences , ' Geneve , vol. 20 , 1888 , pp. 5-36 ) finds fop nifrogen at } pressure the value The two latter observers appear to have used acourate methods to have taken precautions than REGNAULT .
of asbestos to prevent its catching fire .
The method of boiling in mercury , which often adopted , takes a very long time , and would probably give less perfect results , at least in the case of thermometers intended for use at temperatures above The Pressure-Gauge .
The gauge used with this thermometer for measuring the variations in the volume .
and pressure of the enclosed air , was the same as that ibed and lred in the paper before referred to .
* The figure is reproduced for convenience in Plate 7 , fig. 2 .
The -tube FHJK represents the sulphuric acid pressure-gauge for indicating the small differences between the external pressure and the pressure of the air enclosed .
The bulb , together with the rest of the tubes and taps , represents the gauge for measuring the dilatation at constant pressure .
The use of the sulphuric acid pressure-gauge is particularly advantageous in connection with a constant-pressure air thermometer .
Its employment makes it possible to take a series of readings very rapidly when the temperature is varying slightly .
The reading of the gauge can be instantly recorded at any desired moment , and the other observations , such as weighing the mercury and reading the barometer , may be made comparatively at leisure .
These and other advantnges have been already explained in the previous paper .
They have since been very fully confirmed and illustrated by ( ' Phil. Mag Aug. , 1888 , p. 149 ) .
The gauge for measuring the variations of volume of the enclosed air consists essentially ( see Plate 7 , fig. 2 ) of a large bulb A containing mercury , and communi cating with the bulb of the air tbermometer by the way ED .
It is surrounded by a water bath ( shown in fig. 1 ) , whose temperature can be accurately observed by means of a delicate thermometer .
Mercury can be either introduced , or withdrawn from this bulb in weighed quantities , by ) eans of the taps and , from weighed beakers of mercury , into which the tubes open at .
Thus , the volume , temperature , and pressure of the in the bulb A at moment can be accurately determined .
* ' Phil. Trans , 1887 , Plate 12 : The reader who wishes to use an apparatus of this kind should refer to his paper , as the explanations given are very full and detailed ; but in our opinion the form of pressure-gauge which he describes is neither so simple nor oonyenient aS that shown in Plate 7 , fig. 2 .
The ] three-way tap , , can be made to servi , all the purposes of the two taps shown in his figure .
It can also be turned in the middle of the acid , so that it oannot move either way , whatever the variations pressure .
This is a most essential provision of great practical convenience , the absence of which is a serious defect in his form of gauge .
The add , ition of the volume-gauge ( Bulb , \amp ; c. , Plate 7 , fig. which makes it possible to use tho air thermometer at constant pressure as well as at constant volume , is also a great advantage , and rs apparatus muoh more complete .
DETERMINATION OF THE BOILING-POINT OF SULPHUR . .
Method of taking In taking a series of observations at constant pressure with this instrument , the bulb A is first filed from the reservoir ] the level of the mercury stands at some point in the tube DE .
The level is recorded by means of the scale S. The temperature of the water-bath is also taken at the same time .
These two numbers , as will presently appear , give the zero point from which the volumes are calculated .
They remain the same throughout any one series of observations .
Mercury is then allowed to run out by the tap into a weighed beaker at until the sulphuric acid gauge FJIJK shows that the pressure of the air in the bulb is nearly equal to that of the atmosphere .
The water-bath Qurrounding the bulb is well stirred ; its temperature is recorded ; and also the temperature of the air .
The reading of the acid in the pressure-gauge , is taken at the same instant as the resistance of the platinum thermometer .
The height of the barometer is immediately observed , and the weight of the , which has been allowed to run out , is taken at leisure .
The pressure-gauge FG has a range of about , so that , if the temperature is varying slowly , a considerable number of observations may be taken for one hing of the mercury .
Formuloe .
The temperature is calculated by means of the following formulae:\mdash ; If be the pressure of the air enclosed , the volume of the bulb , and temperature , and if be the volume of any portion of the air con tained in the capillary tube , or other parts of the apparatus , at a temperature , the formula of the air thelmometer may be stated as follows:\mdash ; . . . . .
( 1 ) , where mk is a constant depending on the mass of air enclosed , and 2 implies the summation is extended to all parts of the apparatus which it occupies .
To find we deduce from formula ( 1 ) , . . . .
( 2 ) .
The values of the quantities included in this formula , are deduced from the quantities observed as follows:\mdash ; Reduction of the Pressure , The observed height of the barometer is expressed in centimetres of mercury at C. at sea-level in latitude .
A correction of cm .
is added to for for scale errors , for the error of the attached thermometer , and for any difference of level between the cistern and the air thermometer .
' .
The bal.ometer used is one of the Fortin pattern by CASELLA , with a scale very clearly divided in inches , and a vernier reading to 001 inch .
It was carefully compared with the Cavendish Laboratory standard metre scale , which has been verified at the Board of Trade Standards ' Department .
The scale errors were found to be exceedingly small , but the ivory point was set085 cm .
too high .
To find the pressure of the air in the bulb , a small correction for the difference of level of the acid in either limb of the pressure-gauge has to be added .
A preliminary experiment gave as the reading of the gauge when the pressure was the same on either limb .
The factor for reduction to mercury was found by comparison with a mercury manometer to be .
If be the corrected height of the barometer we have , therefore , . . .
( 3 ) .
Calculation of the Volume .
The volume of the bulb at the temperature is found from the equation already given , ( 4 ) .
The volume of that part of the capillary tube which was expcsed to the temperature of the air , was found by mercury calibration to be c.c. The volume of the gauge-tubes CDE , , from the end of the capillary tube to the zero point of the scales at and , was found be 0 .
The volume per cm .
of the gauge-tube FG was .
If be the reading of the gauge , the total volume at the air-temperature is , therefore , ( 5 ) .
The volume of air in the bulb , at the temperature of the ath , is found as follows .
Let be the total weight of mercury filling the bulb A up to the zero point of the scale , and let be its density at the temperature the bath , when the reading of the scale is taken , before running out the ry .
Let be the weight of mercury run out , and ] be its density at the temperature of the water-bath at the time of observation .
Then , if be the coefHcient of expansion of mercury , we have .
The volume of the meroury remaining in the bulb will be .
The voJume per cm .
of the tube DE is .
If the mercury stood at before ru1ming out ; DETlJtMINATION OF THE BOILI -POINT OF SULPHUR .
a small correction has to be added to the above expression , the point of the scale being the zero from which the volume of the bulb A is reckoned .
The total volume of the bulb up to the zero point is at the temperature .
If be the coefficient ofexpansion ofglass , its at will be The volume of the air in the bulb A at the temperature , is the difference between the volume of the glass envelope and the volume of the mercury in it at the time of observation .
It is given by the pression , which reduces to ( 6 ) .
The value of the coefficient is 000151 , so that the second term in the above expression is always very small .
The volume of the bulb A is approximately 48 .
The whole correction may be neglected , provided that the rise in temperature of the water-bath is less than a tenth of a degree .
It is not , as a rule , worth while to reduce the weight of the mercury to vacuo , since the tion amounts to only 1 in 20,000 .
Sources of Error .
The great advantage of using the thermometer at constant pressure is that we can dispense with the mercury manometer .
The apparatus it involves is generally cumbrous , the observations are slow and laborious , and it is difficult to rely on the result beyond the nearest tenth-millimetre .
In reading the barometer , on the other hand , we are measuring the height of a mercury column under much more favourable conditions .
A greater degree of accuracy can be attained ; a thetometer is not required ; and it is mparatively easy to verify the scale .
The operations of weighing the mercury and of the volume-gauge are , capable of being performed with very great accuracy .
An error of a decigram in the weight would produce an error of only in the resulting value of at C. , and at C. The mercury thermometer used for taking the temperature of the water-bath could be relied on to at least a fiftieth of a degree , and the mean of several observations would probably be cort'ect to .
It had been carefully compared with the platinum thermometer and its dings were by this means corrected to the air standard .
An error would produce an error in of only at .
C. , and of at 44 C. An error cm .
in reading the sulphuric acid pressure-gauge would produce an error at C. , and at 44 C. * GRIFFITHS , ' Brit. Assoc. Report , ' 1890 .
hundredth of a degree , shows that this method is capable of giving very good results .
Variations in the Constant , The observations in ice and steam give the equations ( 7 ) These may be used , as above explained , to calculate , if is known ; or to calculate , if is known .
They cannot be used to calculate both and because the variations in mk may amount , under certain circumstances , to nearly one part in a thousand in the course of a few days .
The variation is generally most marked for a short period after the bulb and volume-gauge have been exhausted .
If the value of is determined immediately , and again after an interval of .
two or three days , the value found on the second occasion is always slightly the lower , as if some of the air had been either absorbed by the sulphuric acid or by the surface of the glass , or included between the glass surface and the mercury in bulb A. It is herefore desirable to allow the whole apparatus to settle for some time , before observations are taken with it .
In the present case a whole week was allowed to elapse between the first filling and the use of the thermometer to determine the boiling-point of sulphur .
The value of mk observed on Thursday , September 11 , ' the day before the first boiling in sulphur , was .
The value found on Friday , September 12 , immediately after boiling in sulphur for two hours , was differ by less than one part in ten thousand , and we may therefore assume that the pparatus had reached a fairly steady state .
In the series of observations between and 10 given on ) , which were taken within a day of exhausting and refilling the air thermometer , a diminution in ' the value of from to , amounting to one part in three thousand , was observer .
In order to allow for this , it was necessary to assume that the diminution took .
place uniformly , and the value of at intermediate times was deduced by means of the time observations which were taken throughout the series .
DETERMINATION OF THE BOILING-POINT OF SULPHUR .
Method of Calculation .
The whole set of formulae are collected here for reference . . . . . .
, at . . . . .
( 5 ) .
at where . . .
( 2 ) , . . .
( 3 ) .
in ice ( 7 ) .
in steam The values of the small correction terms were worked on a small slide-rule , with the exception of the barometer correction .
The latter was taken from a large scale diagram giving the whole correction for temperature , scale errors , gravity , \amp ; c. , for any height between 70 and 80 cm .
The multiplications and divisions involved in calculating 2 were generally performed on a FULLER'S Spiral Slide-Rule , results correct to one part in twenty or thousand .
Logarithms , as a rule , were used .
only in the final division .
The whole work of calculating a single observation is shown in the following table .
The limit of accuracy in a single reading of barometric pressure was about 001 inch , corresponding to an error of at C. The results cannot , therefore , be relied on beyond this point , but the calculations were carried to the next figure to avoid .
inaccuracy in the hundredths .
DETERMINATION OF THE BOILING-POINT OF SULPHUR .
drawn off in the bIow-pipe , and the upper end was closed with sealing-wax to keep out the damp .
The ends of the three electrodes were amalgamated and dipped into ebonite mercury cups Nos. 1 , 2 , and 3 , the single electrode being in No. 2 .
In taking observations the thick copper leads from the resistance box were first placed in cups and 2 , to measure the resistance of the platinum spiral plus that of the connecting wires ; then in cups 1 and 3 to measure that of the connecting wires alone ; and , stly , in Nos. 2 and 3 to measure the spiral and connections again .
Deducting the value found in the second observation from the mean of the first and third , we obtain most probable value of the of the spiral itself at the time corresponding to observation 2 .
This form thermometer is extremely easy to make , and gives very accurate results , provided that a good resistance-box is used .
Some preliminary experiments made with a double coil of this kind showed that , if it was kept dry , the insulation was amply sufficient up to a temperature of at least 50 C. , but that it fell beyond this point .
Thermometers and were also made from the same wire as that used for L. They were intended for ltse at higher temperatures with a special form of wire bridge , and were arranged and insulated in a rather different manner , as shown in Plate 7 , fig. 3 .
The two wires were interwound in a double screw thread , passing oIes in a thin mica plate AB , which sufficed to keep the wires from touching each other and to preserve the of the spirals .
Each spiral was furnished with a double and single electrode symmetrically arranged .
The six electrodes were insulated and kept in place passing through holes in thin mica wads , one of which is shown at , which were cut to fit the glass tube containing the thermometer .
In the earlier experiments in the iron tube apparatus electrodes of copper were used .
These were subsequently replaced by electrodes of platinum to which the ends of the spirals were fused in the oxyhydrogen blow-pipe .
This form of thermometer is rather difficult to make , but it has the advantage of very perfect insulation even at high temperatures .
Moreover , by using it with a suitable wire bridge , the resistance of the connecting wires can be automatically iminated , and the platinum temperature can be read off on a scale of equal parts without the necessity of any calculation .
The wire bridge intended for this thermometer could not be made in time for this series of experiments .
The resistances of the two spirals were therefore observed separately with a resistance-box in the same way that of thermometer L. Measurement of the Iiesistance .
The resistance-box used in the present case was a very good Post Office box of the dial pattern made by Messrs. ELLIOTT BROS .
The coils were of German silver , and had a mean temperature coefficient per C. It was kept in a large copper water-bath to ensure steadiness and uniformity of temperature .
The temperature of the coils was taken by a delicate mercury thermometer inside a dummy coil in the centre of the box .
The resistances observed were corrected to C. The temperature of the box never varied far from this point .
Resistances of 10 and 1000 units were taken out in the arms of the bridge ( except when deter1nining small resistan ces , such as those of connections , \amp ; c. , when 10 and were used ) .
The qtance t be measured was generally about 5-20 ohms , and it was found that this arrangement gave bhe best results .
A battery of two Leclanche cells* was generally used .
The battery circuit connected the junction of the two lower to that of the two higher resistanoes .
Thus the cells had to work through a resistance of about 500 ohms .
The current through thermometer was generally less than one three-hundredth of an ampe .
The current-heating of the wire must therefore have been negligible .
A sensitive mirror galvanometer was used having a resistance of about 14 ohms .
The sensitiveness was generally adjusted , so that one scale division corresponded to about one part in a hundred thousand of the resistance to be measured .
The last two figures in each resistance measurement were determined by observing ga]vanometer throws and interpolating .
To facilitate readings , the galvanometer was adjusted so as to be very nearly dead-beat .
A mercury commutator was used in the battery circuit , and the swings were observed by reversing the battery .
This method has the of doubling the siti veness , and of eliminating all possible errors arising from thermo-electric effects , and from the shifting of the zero of the spot of light on the scale .
The resistance of each thermometer was repeatedly observed in ice and steam .
If be the corrected values of the resistances at and , and be the resistance at any other temperature , the corresponding value of the platinum temperature pt is deduced by the formula . . .
( 8 ) .
Sources of Error .
The sources of error be avoided in using platinum resisbance thermometers have been very fully discussed in a previous mmunication , but it will be convenient to recapitulate them briefly here .
* In some cases ( in bhe iron tube observations , and with thermometer H ) , six cells were used with 100 ohms in the battery circuit .
DETERMINATION OF THE BOILING-POINT OF SULPHUR .
1:35 A r.esistance-box is necessa , ry for accurate work .
The dial pattern is much to be preferred .
Any change in the plug or screw contacts in the low resistance , bridge must be refully avoided .
In the high resistance arms they are less material ; but even hers they are sometimes serious .
For instance , in using a P.O. box of the ordinary pattern , it was found that the change from 499 to .
) , involving a sudden change of ten in the number of , produced an error of a tenth of a in the resulting platinum-temperature .
Thermo-electric effects are the most serious source of error .
It is important to arrange the platinum-copper junctions close together .
Any small residual thermoeffects in other parrs of the circuit are not likely to give any trouble , and can be easily e]iminated .
Variations in the resistance of the connecting wires become a serious source of error , if not properly allowed for .
In our experience , the safest method of avoiding such errors is to use double electrodes , and to measure their resistance at each observation , as above described ( p. 133 ) .
The chief objection to this nethod is that ( unless a wire bridge of special pattern is used ) a complete observation of the resistance of the spiral and its connections takes two or three minutes ; but this is imma , terial , provided that the temperature is fairly steady .
If a resistance-box is used , and rapidity of observation is important , the best plan is to use counections of low as in thermometer the variations of which may be disregarded .
The of this method are , that a water cooler must be uQed to keep the upper part of the stem at constant temperature , and that the em of the thermometer must always be immersed to nearly the same extent .
hick electrodes have also the drawback , that they increase the loss of heat by conduction along the stem , and that , unless they are similar and are symmetrically arranged , they are likely to give rise to large thermo-electric effects by unequal cooling of the junctions .
If double electrodes are used , the connections to the resistance-box must be made by mercury cups .
The wires should be freshly amalgamated at frequent intervals to ensure good contacts .
Changes of Zero .
The changes of to which platinum thermometers are liable , when used at temperatures below 50 C. , very rarely amount to so much as 1 in 10,000 .
The following Table contains all the observations of the fixed points of and taken after they had been fitted awith platinum electrodes .
The observations were taken on different days , different heights of the barometer , and , different temperatures of the resistance-box .
We are here working up to the limit of accuracy of the resistance measurements , and it is certain that the small discrepancies shown in the Table are due to errors of observation .
Fortuitous errors of this kind produce the greatest effect on the platinum-temperatures deduced when they alter the value of the divisor .
We have found it best , therefore , in calculating the platinum-temperatures , take the mean value of DETERMINATION OF THE BOILING-POINT OF SULPHUR .
[ Note by H. L. CALLENDAR , added March , 1891.\mdash ; Subsequent experiments with this thermometer make it almost certain that the small apparent changes of zero shown in the above table were due to changes in the mercury cup and screw connections or to errors of the resistance-box .
I have since completed and compensated the resistance coils for the wire bridge mentioned on p. 133 .
With this apparatus I find it comparatively easy to read to a thousandth of a degree , and the thermometer no longer shows any changes of zero .
I have made many other thermometers of this pattern with mica insulation , and I find them all remarkably constant even when used at temperatures as high as 1300o C. ] Description of the Iron Tube Apparatus .
designing this piece of apparatus*we were chiefly guided by a study of the experiments of REGNAULT on the vapour pressures of sulphur .
The difficulties which he encountered are very fully described by him in his paper in the 'Memoires de l'Institut , ' vol. 26 , 1862 , p. 526 .
A short description of the apparatus he finally adopted may not be out of place .
It consisted essentially of two coaxial cylinders of thick wrought iron , closed below , and bolted toget , her at the top .
The outer cylinder was heated in a furnace .
The space between the cylinders was half filled with sulphur .
The inner cylinder was open to the air .
The air thermometer was made to fit the inner cylinder , and was not itself in actual contact with the sulphur vapour .
A small iron t , ube connected the apparatus with a wider tube , in which the sulphur condensed .
REGNAULT states that he experienced great difficulties with this apparatus , owing to the super-heating of the vapour .
This arose partly from the clogging of the small exit tube with viscous Phur , but chiefly from the conduction of heat along the thick iron walls of the cylinder .
We have endeavoured to avoid these sources of error in our experiments by giving the apparatus a somewhat different form , shown in Plate 8 , fig. 1 .
Lengths A of two-inch gas-pipe were screwed together in the form of an inverted by means of a four-way piece .
The ends of the tubes at A and were fitted with caps of the ordinary pattern .
The whole of the vertical limb was filled with sulphur up to the level S. This required about a kilogramme of sulphur , occupying , when melted , a volume of 500 The lower cap was heated in a Fletcher 's circular gas-furnace , shown at by means of a powerful blowpipe .
The top of the furnace was covered with a piece of asbestos card , and the whole of the remainder of the apparatus was clothed with a double layer of asbestos board and .
The level of the sulphur stood at 15 cm .
above the part of the tube exposed to the flame .
The vapour had , therefore , to rise * We also received great assistance from Mr. THOMAS , of Jesus Lane , Cambridge , who personally snperintended the construction of the apparatus .
blDCCCXCI.\mdash ; A. MESSRS .
H. L. CALLENDAR AND E. H. GRIFFITHS ON THE through a long column of liquid before reaching the thermometers .
It appeared probable that the chance of super-heating would thus be minimised .
The horizontal limb A wa , s about 30 cm .
in length , and was provided with openings at both ends for the insertion of the platinum and air thermometers .
The end , through which the air thermometer was inserted , was closed by a split cork made by nailing together pieces of asbestos board .
Some trouble and annoyance was caused at first by the escape of burning sulphur at the variousjoints and openings , till it was found that asbestos board , when soaked in water , formed an adhesive putty , which , if applied wet , at once stopped the escape .
With this apparatus a preliminary experiment was tried , in which the at was closed with a screw-plug , and the sulphur was allowed to distil over through a tube near the end of the horizontal limb .
This arrangement , however , did not prove satisfactory .
The whole of the sulphur distilled over in about a quarter of an hour , which did not suffice for the attainment ol a steady and uniform ternperature .
The apparatus was , therefore , reconstruoted , with the addition of a condenser , as shown in Plate 8 , fig. 1 .
The sulphur vapour was made to pass round by A and to the condenser , which was fitted with a wide glass chimney a metre long , in which the condensation of the vapour could be easily observed .
The condensed sulphur flowed back to the boiler by the small tube , which was bent in such a way as not to interfere with the insertion of the air thermometer .
The tubes A leading to the condenser , were of -inch bore , and so ran little risk of clogging .
By way of precaution , however , a small blow-pipe flame was made to play on thejoint until the circulation was fully established .
Obser vations Results .
Observations were taken with this apparatus on two separate days with very concordant results .
Within half an hour or less of lighting the furnace on each occasion , the heavy brown sulphur vapoul .
made its appearance in the glass tube condenser , and gradually rose with a well-defined surface until the condensation balanced the supply .
The air-blast was regulated each day so as to maintain the vapour at a constant level of 40 cm .
above , thus ensuring a very uniform and steady circulation through the whole apparatus .
The quantities observed , and the results of each observation , are given in Table III .
The readings of the platinum and air thermometers were taken 'fllmultaneously by different observers .
The fractions given in the columns headed " " Resistances\ldquo ; represent the observed swings of the galvanometer , and show the sensitiveness of the apparatus .
The roll-sulphur used in these experiments was said to be prepared from .
It smelt strongly of that gas when freshly crushed .
It appears that sulphur prepared in this way is quite pure enough for the purpose , and that the presence of a percentage of impurities does not materially affect the temperature of the vapour .
The sulphur used on the Friday was poured out of the apparatus before it solidified .
It was to be much discoloured , owing to the presence of various impurities due to residues of oil , lead , \amp ; c. , used in fitting the iron tubes together .
Fresh sulphur was used on the Saturday , this , on examination , appeared to have suffered hardly any change .
The mean values of the platinum-temperatures , observed with thermometers and were : On Friday , On Saturday , The small difference is exactly accounted for by the difference of the barometric pressures on the two days .
This would seem to show that the temperature of the vapour was practically unaffected by the impurities present in the sulphur used on the Friday .
The observations of the baronoeter taken during the course of the experiments on Friday , failed to show with certainty any variation amounting to more than001 inch .
The mean value of all the readings was therefore taken and the same value was assumed in the reduction of all the observations .
It is probable that the slight apparent oscillations of the temperature , shown by the air thermometer , are partly due to real oscillations of the barometer , of which no account could be taken .
The observations of the barometer on Saturday showed a very steady and continuous rise .
A time-chart was therefore constructed from which bhe ost probable value of the barometric pressure , corresponding to the time of each observation , was deduced .
The very close agreement in the values of is certainly , in part , fortuitous ; but the observations must , at least , be regarded as proving that the air thermometer had reached a ver steady state , and that the volume of the bulb was not undergoing any further change .
It is almost certain that the further diminution of volume from .
to , took place when the bulb was sub.sequently exhausted while still exposed to the high temperature .
The value is therefore used in the reduction of the observations taken on Saturday , as well as of those taken on Friday .
The mean values of deduced from the observations with the platinum thermometers and , by assuming the value @ in formula ( d ) , On Friday , C. On urday , C. DETERMINATION OF BOILING-POINT OF SULPHUR .
The mean values of deduced from the simultaneous observations with the air thermometer On Friday , C. On Saturday , C. The value found on Friday is seen to agree perfectly with that deduced from the observations with the platinum thermometers .
The value of deduced from Saturday 's observations , is too low ; but it is probable that the value of , used in reducing the observations taken on that day , is a little too great .
The same of the air thermometer constants were used in the reduction of both sets of observations .
They were determined by readings in ice and steam , taken on Friday evening shortly after the boiling in sulphur .
Tt was , unfortunately , impossible to re-determine after the experiments on Saturday , because the meter had been filled with a fresh lot of air .
A diminution of only 1 in 6000 in the of would suffice to account for the small discrepancy observed , and it is certain , from the measurements taken on preceding days , that was diminishing at about this rate per diem .
We may , therefore , conclude that these experiments , so far as they go , are a complete verification of the ] of @found in 1@87 , and show that the platinum wire has not altered appreciably in the interval .
PART II .
in the Meyer Tube .
The iron tube apparatus above described gave very good results as as steadiness of temperature was concerned , but it as liable to several defects .
It was cumbrous and difficult to set up , and , being very massive , it took some time to reach a steady state , and required a considerable blast to heat it .
The success of the glass-tube condenser suggested a much simpler form of apparatus .
A wide glass tube , AB Plate , was used , having a diameter of about 4 cm .
, and a spherical bulb at the lower end .
Tubes of this kind are commonly employed to heat VICTOR MEyER'S vapour-density apparatus .
For the sake of brevity , therefore , we shall refer to this piece of apparatus as a ' hIeyer\ldquo ; tube .
was cut off to suitable length ( about 40 cm .
) for the insertion of the thermometers .
The lower half was padded externally with asbestos .
The upper half was left to act as condenser , and the was covered with a piece of asbestos card , shown at , to prevent the catching fire .
The bulb was with sulphur , and made to stand in a circular hole cut in a piece of asbestos board , .
A preliminary experiment was tried , in which the tube was heated with a blow-pipe .
No difficulty was found in driving the sulphur our up length of time .
Thermometer was inserted , and some readings of its resistance were taken the sulphur vapour at , different levels in the condenser .
They showed that any increase in the blast merely sent the sulphur vapour higher up the tube till its effect was compensated by the increased area of condensation , and did not appreciably alter the temperature observed in the lower port , ion of the tube protected by the asbestos .
The result of the observations reduced to a pressure 760 , gave ; which is almost identical with the value found in the preliminary experiment with thermometer ( p. 120 ) .
To render the observations in this series comparable with each other , all the platinum-temperatures have been reduced to a standard pressure of 760 mm. of mercury at C. in lat. , and the corresponding air-temperature , , by formula(d ) is also given in each case .
The correction for pressure to be applied to the air-temperature , , of the boilingpoint of sulphur to reduce to 760 , is per m The pressure correction to be applied to the -temperature , pt , may be deduced from the formula ( ) .
By differentiating we obtain whence , if S , and C. , we obtain C. per mm. The blowpipe was soon found to be unnecessary .
By padding the tube with asbestos more carefully and to a greater height , the sulphur could be made to boil quite readily with a Bunsen burner ( Fletcher , No. 48 ) .
By simply adjusting the gas , the vapour could be maintained at a fixed level near the top of the tube .
The arrangement finally adopted for padding and supporting the tube is shown in Plate 7 , fig. 4 .
The bulb projects through a circular hole in a piece of asbestos board CD which fits it tightly .
The tube is filled with sulphur some 3 or 4 cm .
above the level of the board .
The padding of asbestos wool is kept in place by an outer tube of asbestos board bound with wire .
A cone of asbestos board surrounds the whole .
This gives the arrangement greater stability , and diminishes the rate of cooling .
The method and precautions to be observed in working the apparatus are a.s follows .
* Deduced from observations .
See GRIFFITHS , sulphur is all melted , the apparatus is placed on a tripod stand , the thermometers and leads are adjusted , and the gas is turned full on till the level of the vapour rises halfway up the condenser .
This generally takes from two to three quarters of an hour .
The time depends greatly on the thinness of the glass of the bulb\mdash ; the thinner the better\mdash ; and on the length of condenser-tube left bare above the asbestos packing .
When the level of the vapour reaches the desired point , the gas is turned down till it just suffices to maintain the level steady .
Some taken with thermometer in this apparatus gay the temperature .
; With this arrangement there are evidently two sources of error : ( 1 ) The thermometer is cooled by conduction along the stem , so that the sulphur condenses on the upper part of the stem and runs down it , cooling the bulb slightly below the temperature of the vapour surrounding it ; ( 2 ) The sulphur condensed on the sides of the upper part of the tube exposed to the air acts in a similar way , and the thermometer loses heat by radiation to the walls of the tube .
To test this some readings were taken with the spiral of thermometer inserted to various depths in the tube , wit .
the following results:\mdash ; With the spiral just immersed below the level of the asbestos padding , and about 3 cm .
below the top of the vapour , the temperature found was ; With the spiral about halfway down ; With the spiral partly i1nmersed in the boiling sulphur ; These results plainly showed that the sources of error suggested above were of serious magnitude , and that , in order to find the temperature of the vapour , it was not sufficient merely to insert a thermometer in the vapour of the liquid boiling freely in a glass tube , even if the tube were very carefully protected by thick wrappings .
The difficulties above described were overcome in the following manner .
To protect the thermometer from the condensed sulphur which ran down the stem , and to shield it from radiation to the upper parts of the tube , an asbestos umbrella Plate 7 , fig. 4 , was made to fit the stem , and was boumd on to it with wire a short distance above the spiral .
To protect it from cooling by radiation to the sides of the tube , a glass tube screen , , coaxial with the thermometer , was hung on this umbrella by three wires .
three holes , so arranged as to give free passage to the sulphur vapour , while not allowing an radiation to pass .
The glass-tube screen was protected by the umbrella from drippings of condensed sulphur , and must have attained very nearly the temperature of the vapour .
The thermometer and screens were kept in position in the centre of the tube by means of three-legged distance-pieces of wire twisted round them , some of which are indicated in the figure .
The observations indicated , as might have been expected , a slightly higher temperature than those in which the thermometer was exposed to radiation from the sides of the tube .
The gas was adjusted to keep the level of the vapour near the top of the tube .
The sulphur was boiling very steadily and quietly , and the thermometer did not show a.variation of more than a hundredth of a degree for about half an hour .
The result found was ; 443.77 ; showing a rise of C. in the value of In order to test the amount of cooIing produced by the condensed sulphur which ran down the stem of the thermometer , as compared with that due to radiation to the sides of the tube , a small cap of asbestos paper was fitted to the stem of thermometer just above the spiral , to divert the drippings .
The temperature observed was ; ; showing a rise of C. was then tried with the umbrella and glass-tube screen , arranged as above described .
The observations gave ; The effect of the glass-tube screen in stopping radiation is therefore to raise the observed temperature by C. , and the effect of diverting the dripping down the stem is to raise it by C. Both causes together produce a rise of C. , which is nearly the same as the rise found in the case of thermometer Seeing that the effect of radiation was so large , another experiment was tried with thermometer , in which two coaxial glass-tube screens were used .
It gave the result ; To test whether agreed with thermometer , an experiment was also tried with a single glass-tube screen , which gave ; a number practically identical with that found for when singly screened .
DETERMINATION OF THE BOILING-POINT OF SULPHUR .
The further rise caused by using the double screen seemed to indicate that glass tubing , though probably opaque to the radiation from the glass stem of the thermometer , might not be sufficiently opaque to the heat radiation from the platinum wire itself at this temperature .
An experiment was therefore tried with thermometer in which th inner glass screen was replaced by a tube of brass .
This was found to cause a still further rise , the result being Thermometer ; ; After the experiment the brass tube found to be rather thickly coated with sulphide .
It was therefore replaced by a tube of platinum foil on glass .
The naked portion of the Meyer tube at the top was cut off to a length of 3 cm .
, to the condensation .
It was felt that this arrangement would probably give the nearest approach to the actual temperature of the vapour , as a very small flame was sufficient to maintain the ebullition .
All four thermometers were used , and their fixed points were re , -deterrnined after the experiment .
The results found were Thermometer E. .
Thermometer . .
, , Assuming @ for thermometers , and , we find for the corresponding air-temperature the value .
We believe this to be within of the true temperature of the vapour of sulphur boiling ' freely under a pressure of 760 mm. Results of Previous Observer.s .
We have not been able to find that any previous observers , with the exception of REGNAULT , have seriously attempted the accurate determination of the boiling-point of sulphur by reference to the air thermometer .
The numbers given in CARNELLEy'S tables are either deduced from REGNApLT'S observations , or are rough experiments made to test the working of some vapourdensity apparatus by comparison with his results .
Of REGNAULT'S eight observations on the pressures of sulphur vapourt , various temperatures , only two were taken in the neighbourhood of the atmospheric pressure .
They are *E.g .
, the number C. attributed to .
MEYER corresponds to a pressure of 725 mm. at Zurich , and is deduced by ation from two of 's observations .
Berichte Deutscb .
Chem. Ges vol. 9 , p. 1225 .
) 'Memoires de l'Institut , ' vol. 26 , 1862 , p. 527 .
From The table calculated from empirical formula by WEINHOLD* gives , at 760 mm. The table re-calculated and smoothed by from observations gives .
These vahies are nearly a degree higher than that found from the observation at 760 .
The reason of this discrepancy is , possibly , that the empirical formula was calculated to satisfy the observations at higher temperatures , which were more likely to be vitiated by superheating .
It was part of REGNAULT'S amme to determine the vapour-pressures of sulphur through a wide range of tempera.ture .
He was therefore compelled to work with a very strong and thick iron cylinder , heated in a churcoal furnace .
The air thermometer itself was not in direct contact with the vapour , but was enclosed in a closely-fitting and massive iron tube , firmly bolted to the top of the cylinder , and with it .
REGNAULT describes , at some length , the difficulties which he with this apparatus from the of the vapour , and it is plain from the language he uses that he had no great confidence in his results .
In the iron tube apparatus which we ourselves used , the possibility of superheating was reduced to a minimum by making the vapour rise through a long column of liquid sulphur which was not directly heated. .
Want of space prevented the introduction of screens to guard against the cooling of the thermometers by radiation to the sides of the tube .
Thus , although the temperature attained by the thermometers was st , eady to C. for more than an hour , and although it was the same ( allowing for the difference of pressure ) on both days , we ca1mot assume that it was actual temperature of the sulphur vapour .
All that we can safely assume is that since all the thermometers had glass envelopes , the loss by radiation was very nearly the same for each .
The experiment may -therefore be taken as affording tccurate c between the platinum and air thermometers , and a satisfactory verification of those performed in 1887 .
The experiments in the Meyer-tube illustrate very forcibly the necessity of suitably the thermometers from loss of heat by radiation .
They sho# that errors from this source may readily amount to at .
In the last series experiments , however , which the thermometers were doubly screened and the condensation was reduced to a minimum , it is probable that the results obtained within C. of the true temperature of the vapour .
We therefore propose to assume the value for standardising platinumf * PocGENDoxrr , Annalen , ' vol. 149 , 1873 , p. 231 .
'Phil .
Mag vol. 20 , 1885 , 524 .
See also RAMSAY , ' Chem. Soc. Journ vol. 49 , p. 37 ( 1886 ) .
DETERMINATION OF THHi BOILING-POINT OF SULPHUR .
147 : thermometers .
We may observe that , even if this not the actual temperature of the vapour , it is probable that it will be the temperature by the thermometer in an apparatus similar to ours .
Any other known boilin g-point , such as that of mercury , well be used for this purpose ; but that of sulphur presents special advantages .
The material cheap , and readily procured in a state of sufficient purity .
It boils at a convenient temperature , and can be kept in glass vessels for any.length of time without alteration .
It is more suited for the purpose than mel .
Cury , because it boils more quietly and at a higher temperature .
It is also less likely to clack the glass vessels , and its fumes are not poisonous .
Method of Standardising Platinum The method which we recommend for standardising platinum thermometers is briefly as follows .
Observe the platinum-temperature pt of sulphur vapour an apparatus similar to that shown in Plate 7 , .
Calculate the ) ature t of the vapour at the observed rometric pressure , at 760 mm. to be 44 53 C. , and adding C. for each mm. of pressure above the standard .
he value of the constant @ ma , y then be by substituting the observed value of the difference ( \mdash ; pt ) in formula ( d ) .
instance , the observed in the case of thermometer is \mdash ; C. at C. This gives@ .
The value of the ratio for this thermometer is Another thermometer , which was made in a similar way to , with electrodes , but of different wire , having a coefficient , gave the value at , whence@ The value of .
for each of these thermometers is much lower than in the case of , and .
It appears , generally , that the lower the value of , the higher the value of S. Another wire used thermometers Phil. Trans. gives the values For many purposes an accuracy of the order of 1o C. at 45 is sufficient .
In this case , the trouble of making a special experiment with sulphur to find S , may be spared .
It appears likely that for values of in the neighbourhood of the value of S may be assumed to be without much risk of making an error of more than half a degree at 45 C. , and that for of the type the value of @ may be approximately found from the empirical formula .
MESSRS .
H. L. CALLENDAR AND E. H. GRlF ITHS ON ] of Mercury .
A determination of the boiling-point of meWury was also made in a similar apparatus with thermometel.sL , and .
The result found was C. The result deduced from previous experiments with thermometer , in a different apparatus , is C. The closeness of the agreement is indirect verification of the @-formula .
The agreement would probably have been closer , but the Meyer-tube cracked near the bottom in the course of the experiments , so that we have reason to believe that the result found with and is a small fraction of a degree too low .
REGNAULT gives seven observations on the boiling-point of mercury at atmospheric pressure .
They range from to .
The mean is 35 9 , which agrees as closely as could be expected with the number found above .
Boiling- Freezing-Points of some other Substances .
The results of an investigation into certain boiling- and freezing-points have been given in a previous which is referred to as " " Paper ( G ) " " throughout this .
It was therein pointed out that the values obtained depended on the following " " fixed points\ldquo ; :\mdash ; B.-P .
Naphthalene ( 760 mm. ) ( CRAFTS ) B.-P .
Benzophenone , , B.-P .
Sulphur , , and sufficient data for the correction of the deduced temperatures were given , in case these " " fixed points " " were found to be inaccurate .
The conditions under which the experiments described in Paper ( G ) were performed , closely resembled those prevailing in the Meyer-tube when the thermometer was unscreened .
This conclusion is borne out by the numbers given on p. 120 of this Paper , where the platinum-temperature of sulphur vapour in the same as that used in the previous experiments was found to be , a number almost identical with that given by a specimen of the same wire when unscreened in the Meyer-tube ( see p. 142 ) .
Our experiments have shown that the temperature of the unscreened thermometer , when placed in sulphur vapour in the Meyer-tube , is too low ; at the same time they enable us to apply the small thus rendered necessary .
The values of and in both thermometers and have shown signs of change in the interval which has elapsed since the spring of the year , but by assuming the old values of pt when in the vapour of sulphur under 760 mm. to be equivalent to , and calculating the values of @ for each thermometer , we shall avoid the GBIFFITHS , supra .
DETERMINATION OF THE BOILING-POINT OF SULPHUR .
subsequent changes of zero , and it is probable that any small errors due to radiation , \amp ; c. , will be eliminated , since all the ations of boiling-points were made with similar apparatus and under similar circumstances .
A small correction has first to be made in the numbers given in Paper ( G ) , owing to a scale error in the Sidney College barometer , which had previously escaped detection .
A careful comparison with the CAVENDISH Standard Barometer has shown a constant error of mm. The consequent correction is given in column of Table Iy .
Column I. gives the corrected values of pt for each thermometer , deduced from the numbers giyen in Paper ( G ) .
Applying a similar correction to the values of the platinum-temperatures of sulphur vapour in the old apparatus , we get pt .
under .
The values of , given in column II .
of the Table of Boiling-Points , are deduced from the above values of@ .
The circumstances under which the freezing-points of the metals were determined , preclude the possibility of loss of heat by radiation , \amp ; c. In reducing these observations , therefore , the corrected value of the platinum-temperature of sulphur vapour under 760 mm. should be assumed , in order to obtain the value of@ .
Our experiments have shown that the difference in the platinum-temperatures given by thermometer when unscreened and when fully protected is ( see p. 144 ) .
We may assume , therefore , ttJat , had all the precautions been observed which we have since found to.be necessary , the values of pt would have been raised by a like amount .
The resulting values of S are ; ; . . .
; and the numbers given in column II .
of the Table of Freezing-Points are deduced from these values of@ .
Column III .
gives the result for each thermometer , re-calculated and corrected , as above described .
The close agreement of the resulting numbers is an indication of the validity of the method of reduction adopted .
We believe hat the results are within a tenth of a degree of the real values of The vacant spaces under thermometers and indicate that no observations were taken with them at those points .
and REGNAULT'S numbers , and is a reduced copy of the one that was used in Paper ( G ) for reducing observations with thermometer E. The continuous line shows the path of the curve obtained from the parabola , the abscissa being , the ordinate The scale on which these curves are plotted is about of the original .
It will be seen that the divergence up to a temperature of about is very small , the rapid divergence above that temperature being due solely to the difference between REGNAULT'S and our determinations of the -point of sulphur .
As pointed out in the previous paper , thermometers A to were to be regarded as trial instruments , and a summary only was given of the results obtained by means of them .
Since they were of different patterns , and the spirals formed of different wires , it may be interesting to give the mean of the results obtained by applying a similar correction .
Their values of @ range from to ( D ) .
The experiments were not performed with the same care as those with thermometers , and , and their results are more irregular ; in one or two cases they differ from the mean by as much as The mean values of deduced from the five thermometers are as follows :\mdash ; Aniline Naphthalene Methyl salicylate Benzophenone Mercury . . . . .
only ) These thermometers were not used for the determination of any freezing-points .
[ Note by E. IT .
GRIFFITHS , March 21st , 1891.\mdash ; Since the communication of the preceding paper , I have constructed and graduated another platinum thermometer .
The method of insutation and construction is similar to that described on p. 133 , and shown on , fig. 3 .
The thermometer contains two coaxial spirals , the smaller one being inserted within the larger .
The stem is 18 inches in length , and both ends of spiral are fused to double platinum electrodes .
The walls of the surrounding hard glass tube are of eggshell thinness .
The resistance of the spirals can be taken separately , in series , or in parallel arc .
Four resistances differing greatly in magnitude can thus be determined at any temperature .
The accuracy of the resistance-box is thus exposed to a severe test .
In order to reduce the stem resistance as much as possible , the stout wires forming the electrodes are , when taking observations , connected in pairs .
The method adopted is as follows:\mdash ; Let be the resistance of the longer spiral , and the resistances of the two arms of one of its electrodes , and , the resistances of the arms of the other electrode .
parallel arc and leave in the same manner by The resistances , and , being very nearly equal , both ends of the spiral will , in the second case , be at the same potential ; hence the second connector gives the total external resistance .
The value of can thus be determined by two observations .
* Similar connectors are used to determine , and and being the resistance of the two spirals when in series and parallel arc respectively ) .
The thermometer has been standardised in the manner described on p. , and all the utions observed which our experience has shown to be necessary .
The umbrella , coaxial tubes , \amp ; c. ( see Plate 7 , fig. 4 ) , were invariably attached to the thermometer when it was placed in a vapour .
I was anxious to test the accuracy of the method adopted for correcting the results of my former determinations ( se Table IV .
, p. 150 ) , and I have , therefore , by means of this thermometer , re-determined the boiling-points of mercury and naphthalene with great care .
The results may be su.mmarised as follows:\mdash ; PART III .
Comparison of Platinurn and Air Therrnometers Thermometer was constructed with the view of measuring tenlperatures between and .
It has already been described in a paper read before the British Association .
A short description , however , appears to be necessary .
* The validity of this method depends on the equality of , \amp ; c. I , therefore , carefully determined their resistances separately at different temperatures .
I found the limits of error to be within of the stem resistance , and since the stem resistance in no case exceeded of the total resistance , inaccuracy thus introduced might be neglected .
A fresh specimen of naphthalene gave C. , " " On the Comparison of Platinum and Mercury Thermometers at Low Temperatures ' Brit. Assoc. Rep 1890 .
repeated exhaustion and refilling with dry air at a temperature of three to four hundred degrees , was filled with melted anthracene .
This substance ( melting-point was found to be as good an insulator as paraffin at temperatures below The coil , consisting of about 56 inches of wire , having a diameter of rests on calcined asbestos , but has no contact with the glass envelope .
The spiral and asbestos are thus perlectly dry , and contained in an almost vacuous } ) .
The leads down the stem are of silver , with a ance of less than 001 ohm , and approach to within one inch of the spiral , and are connected with it by moderately thick platinum wires , thus any flow of heat from the spiral to the silver is diminished .
The instrument is connected with the resistance box by two leads about 6 feec long , each consisting of 50 copper wires ( B. W. G. 24 ) enclosed in an india-rubber tube .
Changes in the external resistance , caused by changes in temperature of the leads , could therefore be neglected .
The fixed points were determined at various times with the The first pair of observations were taken in the Sidney Laboratory , the second pair at the Cavendish , but on a different floor to the barometer , had therefore to be corrected for difference in elevation .
During the last pair of observations barometer was placed by the hypsometer ; these observations therefore are probably che most reliable .
In any case the diverger1ce of any one experiment from the mean would not affect by more than 1 in 15,000 .
Hence , mean and , and The comparison of and the air thermometer was lcted as follows .
Both thermometers were placed in a nearly cubical iron tank , holding about 16 gallons of water .
The air thermometer was inserted through a hole in the middle of one of the sides , the thermometer through a hole in the centre of the lid .
The two bulbs met in the centre of the tank , being in all but actual contact .
The sides of the tank were covered with several wrappings of flannel , as we desired to keep the temperature nearly constant during each group of tions .
The water was raised to the required temperature by a gas burner beneath the * The first four numbers have been slightly altered since their communication to the Association , owing to the subsequent iscovery of the constant scale in the X DETERMINATION OF THE BOILING-POINT OF SULPHUR .
be relied on much beyond , which corresponds to the limit of accuracy attainable in a single reading of the barometer ; but the last column shows that the small differences between consecutive observations , which are not affected by the uncertainty of the barometric reading , agree much more closely among themselves .
The following table gives the mean results of each group of five .
TABLE VI.\mdash ; Comparison of Platinum with the Air Thermometer .
The differences given in the Iast column are calculated from the formula . . .
( ) .
the group of observations at , where the discovery of a faulty connection led us to suspect the observations before their results were deduced , we see that the observed calculated values in no case differ by more than , except the last two observations .
These show an error of , both in the same direction .
In taking the two observations in question , it was unfortunately necessary to read the barometer by gaslight , when its temperature had just begun to fall .
It is very likely that the temperature of the mercurial column itself would lag slightly behind that of the attached thermometer .
This would account for the small discrepancy observed .
The results obtained by a comparison of with REGNAULT'S mercury thermometer No. 7 , and with the Kew standard , are given in the commun ication previously referred to .
* It was then pointed out that REGNAULT and Kew were in practical agreement .
REGNAULT says that the differences between No. 7 and the air meter were so small that they could not be determined with His mercury thermometers generally showed a difference of to as compared with the air thermometer at , but the observations are not consistent to .
The readings of thermometer No. 7 ( which was made of ordinary glass ) appear to approach the air readings more closely than is usually the case with mercury thermometers .
The comparison with the Kew andard was made by means of four thermometers .
having a range of temperature from to , divided to of a degree , and coJrected at Kew .
A comparison of the difference-curve with that whose ordinates are given in the above Table , shows that the Kew standard reads C. lower than our air thermometer at C. CONCLUSIONS .
The experiments described in the preceding pages show conclusively thab the platinum resistance thermometer is an extremely self-consistent and trustworthy instrument , and that it is remarkably free from change of zero and from secular alterabion of its coefficients .
It appears certain that , by choosing some particular thermometer as the standard with which other thermometers could be directly compared , it would be possible to attain a very degree of accuracy in the relative measurements of temperature on the arbitrary scale of the standard chosen .
We have found that , by using a good resistance-box ( correct to 1 in 100,000 in the relative values of the coils ) , it is possible to read to between and lo C. By using a suitably-constructed and tmpensated wire bridge , a still higher degree of accuracy may be attained , and the sensitiyeness of the thermometer may readily be increased so as to indicate a difference of one ten-thousandth of a degree at or(iinary temperatures .
The reduction of the platinum-temperatures to the absolute scale is less certain , since it depends on the accuracy of the air thermometer experiments .
But it appears probable that , by assuming the parabolic formula ( d ) , and calculating the appropriate value of the constant @ for each thermometer by means of an observation at the poin of sulphur , as above described ( p. 147 ) , we may attain a degree of accu*GRIFFITHS , " " On Comparison of Platinum and Mercury Thermometers 'Brit .
Assoc. Rep 1890 .
, Also , see the ' Electrical view ' for Sept. 26 , 1890 .
'Mem .
de l'Institut , ' vol. 21 .
1847 , p. 220 .
Three of these thermometers belong to ths B.A. ommittee on Electrical Standards ; see Appendix to their Report , Brit. Assoc. , 1890 .
DETERMINATION OF THE BOILING-POINT OF SULPHUR .
racy of the order of between and 10 C. , or of at 45 C. , on the scale the standard air thermometer .
We believe that thermometers standardised in this way will be found to afford the most convenient and practical method of measuring temperature in all cases where accuracy is desired .
EXPLANATION OF PLATES 7 AND 8 .
PLATE 7 .
Fig. 1 .
General view of the arrangement adopted for supporting the bulb of the air thermometer , and connecting it to the gauge used for determining the variations in the pressure and volume of the enclosed air .
Fig. 2 .
Represents the gauge on a larger scale.\mdash ; A detailed explanation of the various parts will be found on p. 126 .
Fig. 3 .
A slightly enlarged view of the spiral of the platinum thermometers and , showing the method of supporting and insulating the spiral and its electrodes by means of the mica plate AB , and the mica wads shown at A and C. Fig. 4 .
Shows the method of screening the thermometer when placed in the Me.yerTube Appara us for determining the boiling-point of sulphur , as in detail on pp. 142 and 143 .
PLATE 8 .
Fig. 1 .
Iron tube apparatus used in the comparison of the platinum and air thermometers at the boiling-poUnb of sulphur .
It could be maintained at a steady temperature by means of a continuous flow of sulphur vapour for any length of time .
A full description of it is given on pp. 137 and 138 .
The Curves in this plate illustrate graphically the effect of using the lower value , above found , of the -point of , instead of that given by REGNAULT .
See p. 151 .

