]\gt ; [ 43 ] II .
On the of some Boiling and Freezing Poi , by means of the B.y E. H. GRIFFITHS , , Sidney Sussex College , Cambridge .
Communicated by R. T. GLAZEBROOK , F.R. Received May June 19 , 1890 .
PLATES 4 1 .
LENDAR , in his exhaustive comparison of the platinum and air thermometers Phil. Trans , suggested the introduction of the term " " platinum te1nperature\ldquo ; ( denoted by pt ) .
Pt is obtained by assuming the temperature to vary as the resistance .
Thus A curve is then constructed , having being temperature as indicated by the air thelmometer ) as the ordinate and pt as abscissa .
I have adopted this method of expressing my results throughout this paper .
If CALLENDAR'S conclusion is correct , , that whatever be the coefficient of increase in resistance from to of the platinum wire used , then for the same all wires will give the same , the difficulty of determining high temperatures with accuracy has been overcome .
Unfortunately my experiments have driven me to the conclusion that the pt given by platinum thermometers made from different samples of wire differ considerably for the same value of .
However , in my thermometers the wire throuolJouto .
is in contact with or asbestos , or both , and its behaviour is probably much influenced thereby .
In CALLENDAR'S thermometers the coil was in contact with the glass only at its extremities .
* Later on I give particulars of the thermometers used in these experiments .
I was anxious to adopt a form of thermometer which would ( 1 ) quickly assume the surrounding temperature , ( 2 ) occupy little space , and 3 ) bear reasonable amount of * See note in Appendix ( p. 70 ) .
11.3.91 .
rough usage .
I believe that the referred to hereafter as , and fulfil these requirements .
2 .
I proposed to graduate the thermomebers by exposing them to the vapour of certain substances under known pressures , the substances selected being as foUows:\mdash ; TA BLE I. ( 1 ) Melting ice ( 2 ) Steam at ?
60 mm. .
Aniline at 760 mm. * ( 5 ) Methyl salicylate at 760 mm ( 4 ) Naphrhalene at 760 ( 6 ) Benzophenone at 760 mm. ( 7 ) Mercury at 760 mm. .
( 8 ) Sulphur at 760 mm. The change in temperature of the vapour of these substances due to changes in pressure appeared also to have been determined .
In the following pages I endeavour to show how the adoption of the numbers given for ( 1 ) , ( 2 ) , ( 4 ) , ( 6 ) , and ( 8 ) in the above table was forced upon me to the exclusion of the others .
The Apparat us Used .
3 .
The resistance box ( of the four dial pattern ) was one constructed for me by Messrs. ELLIOTT in 1886 .
I have had occasion to test this box severely during some investigations on which I have been engaged for the ] two years , and I believe it to be a most reliable instrument .
Messrs. ELLIOTT were good enough to ascertain for me , the coefficient of alteration per 1o C. for each set of dials .
Althoug.h all the wire used in the box was from one casting , these coefficients differ slightly and are as follows : TABLE II .
10 ohm per right at *Most of these substances were suggested to me by Messrs. HEYCOCK and NEVILLE .
I wish that I could adequately express my indebtedness to these gentlcmen , who have nob only assisted me in my observations , but have freely placed at my disposal the apparatus used by hem in their investigations on the melting point of alloys .
All the apparalus used in these experi1ments , with the exception of the galvanometer and resistance box , was constructed by Nr .
F. THONAS , of Jesus Lane , Cambridge , and , like all other work that he has done for me , loaves nothing to be desired either in design or workmanship .
AND FREEZING POINTS BY THE PLATINUM THERMOMETER .
The accuracy of the coefficients will best be uuderstood by reference to the Tables in the Appendix .
The chief difficulty has been that of ascertaining the real temperature of the coils , for the mere suspension of a thermometer within the box afforded but little information as to the temperature of the wires , unless the temperature of the box was steady .
The rapid changes of temperature in the air of the room in which the experiments were conducted made it necessary to place the resistance box , etc. , in an oining room , which , Jfortunately , had only an unprotected iron roof .
Attempts were made to keep the temperature constant by means of a gas stove , but with little success .
A large copper bath with double walls was constructed , the bottom of the resistance box was cub out , and the top , with the exception of openings for the dials , was covered with a board , the under part of was lined with swansdown .
A gas jet beneath the bath was controlled by a mercury gas regulator of the usual form , but having a very large bulb about a half-inch broad and eight inches long , reaching from nearly the top to bhe bottom of the bath .
This maintained the bath night and day at a temperature which was never observed to vary by However , the radiation from the top of the box could not be prevented , and during some of the frosty nights near the end of March its temperature fell as low The thermometer in the box was coated with paraffin of about the same thickness as that on the coils .
Later , Messrs. ELLIOTT constructed for me a in all respects similar to those in the box , and the thermometer was placed with its bulb in the centre of this coil .
It appeared probable , therefore , that its rate of change would be about equal to that of the bobbin .
The observations made after March , from which date the regulator was constantly at work , showed a greater agreement than those taken before that date , and as the greatest divergence from the normal temperature was , the residual errors due to temperature could not in any case be very great .
4 .
The galvanometer was a -resistance one , made for me by the cientific Instrument Company .
Its total resistance was between 8000 and 9000 ohms ; however , its coils being connected in parallel arc .
, its resistance was about 2000 ohms .
Although this was far too high a resistance , as compared with the other resistances of the bridge , I considered it advisable to use it , an important consideration being that any thermal effects were unlikely to affect it .
The galvanometer was dead beat , and was so arranged as to be ensitive .
The scale was about two yards from the galvanometer , and thus a very small deflection was easily noted .
* For some time commutators were used , all connections reversed , and the observations repeated to eliminate the effect of thermal currents .
In no case , however , could any such effect be traced , so that in my later experiments I only reversed the connections occasionally .
A double contact key of the usual kind was used .
* A low-resistance galvanometer was tried and rejected .
was found necessary not only to place the galvanometer on an insulating stand , but also the batteries and key .
All connecting wires were run through india-rubber or glass tubes .
Until this was done the galvanometer showed a permanent , deflection after the wires were attached to the batteries .
The whereabouts of this leakage I never discovered .
5 .
The apparatus was aCed in rooms adjoining the Sidney Laboratory , and I was much troubled by the vibration caused by the traffic .
In fact it was only possible to take some of the observations ( such as those in which I wished to trace the process of surfusion during the freezing of metaJs ) late night .
One great difficulty I met with at the outset , i.e. , the warming of the thermometer-coil by the current used when determining its resistance , and since it was necessary , during some of the observations on surfusion , to maintain the current for several seconds at a time , the point was one of considerable importance .
CALLENDAR mentions that he used six Leclanche cells without any trouble of this kind .
I found it impossible to use two unless I lowered the current by introducing extra battery resistance .
The effect was clearly traceable if too strong a current was used .
If the arms of the bridge were in equilibrium and the key held down , the spot would gradualy slide away in the direction indicating a steady increase in the resistance of the coil .
A few experiments showed that t , his rate of increase was nearly as the square of the current , thus indicating the cause .
Again , the stouter the wire in the thermometer , the greater the E.M.F. that could be used without producing this effect .
For instance , the thermometer , in steam under 755 mm. , gave a reading of when used with one Leclanche cell , and of when used with two Leclanche cells .
The same difference was observable throughout all the readings of , and , at different temperatures .
This effect was arly marked in ( whose coil was formed of exceedingly fine wire ) , particulal.s of which I have given later on .
I found it convenient in my later experiments to work with a large number of introducing into the battery-circuit sufficient resistance to reduce the current to harmless dimensions .
On the determination of the 'fixed points 6 .
-point of .
The method described by GUILLAUME ( see 'Traite Pratique de la Thermometrie de Precision ' ) was adopted with this addition , that the vessel with finely powdered ice was surrounded by a larger one packed with large lumps .
In some of the experiments the ice in the inner vessel was drained , in others , a mixture of ice and water was used .
Snow was also tried ( see Table XVIII In all cases the ice had been thoroughly washed with distilled water .
Boiling-points .
A hypsometer of the usual form was used water .
gauge connected with the instrument showed thal the internal pressure rarely exceeded the external by more than 1 mm. of water .
AND FREEZING POINTS BY THE PLATINUM The remaining boiling points were all determined in glass vessels .
Round bottomed , long necked flasks , having a capacity of about litres , were used for anilin , naphthalene , and sulphur ; somewhat smaller ones for the remaiuing substances .
At about from the toP , a side tube was fitted into the neck , and through this tube the vapours passed into the condenser .
The bulb of the thermometer just projected into the body of the flask ( in no case touching the liquid ) , and the vapour on its way to the condenser had to pass the whole of the immersed portion of the thermometer .
The top of the flask and the mouth of the condenser were closed with cotton wool or asbestos cloth , in which the vapours while the risk of increased internal pressure was shed .
7 .
The difficulty with regard to super-heating was overcome as follows : An oven was built up with asbestos cards and in the roof a circular hole was cut in which the base of the flask was placed .
The surface of the liquid in the flask was always maintained at a higher level than the rim of this ] , and it was , therefore , impossible for the flame to play on any portion of the glass above the liquid .
To prevent radiation through the liquid a smoky flame was used until the exposed portion of the flask was covered with carbon .
Asbestos cloth was then wound round the flask , completely covering it from the roof of the oven to the exit tube , and on removing the cloth after distillation no signs of condensation were observable on the sides of the flask .
From the time that these precautions were taken no indications of super-heating presented themselves , the temperature during distillation being independent of the size of the flame .
Every endeavour was made to secure pure specimens of the selected mces .
In most cases the observations were repeated from the commencelnent of distillation until about two-thirds of the sample had distilled over , the process was , therefore .
practically , one of fractional distillation .
Occasionally the distillate was separated after about half the sample had come over , and the residue distillate experimented upon separately ; and in one or two cases the process was repeated on the portions thus separated .
The only substances which showed slight signs of change when thus treated were methyl salicylate and triphenyl methane .
It is worthy of notice that the same specimens of naphthalene and benzophenone were boiled or distilled between twenty and thirty times , and yet their boilin-points showed no signs of change\mdash ; a strong argument in favour of the purity of samples , and of their adoption for the determination of fixed points .
As large a quantity as possible was used in each case .
* * , in the paper referred to in Table I. , writes as follows:\mdash ; ' ' . . .
La naphtaline du commerce no vary plus de dans son point d'ebullition quand elJe a ete purifiee , par un methode quelconque , ce que la substance fond a .
La , meme chose est vraie pour la benzophenone preparee par l'oxychlorure de carbone , son point de fusion est amene par un purification convenable a .
On voit que des limits considerables peuvent etre tolerees dans The aniline , naphthalene , and sulphur ( when liquid ) measured nearly 1 litre ; the mercury about 200 .
( 2600 grms. ) ; the quantities of benzophenone and methyl salicylate were rather smaller .
In those cases in which different specimens were tried , the results were the same ( see Table XIV Comparison of taken under different pressures .
8 .
If be the resistance at a temperature .
C. , then will give the approximate increase per 1o .
at a .
The accuracy of the result depends of course on the nearness of and Observing the pressure ( during my first series of observations with each there mometer ) at the time of the determination of the B.P. of water , naphthalene , benzophenone , and sulphur , I was able ( from the data supplied by CRAFTS and REGNAULT ) to determine the actual temperatures , \amp ; c. , corresponding to observed resistances , \amp ; c. The resistance at being also known , the approximate increase in resistance per 1o at ternperatures , , \amp ; c. , could be deduced .
After plotting the points thus obtained I was enabled to determine the approximate increase per 1o at intermediate points .
On examination of the change in temperature due to increased pressure as given by different observers ( in the of the selected substances ) it was evident that however they might differ in regard to the actual temperature , they were in practical agreement as to the over a range of about 30 mm. above and below 760 mm. For instance , both REGNAULT and RAMSAY agree in givin per 1 mm. as the change in cury vapour through the above range , although they differ by nearly 1o in the actual temperature .
* les points de fusion sans que le point d'ebullition change , ou , en d'autres mots , des impuretes font rier plus vite le point de fusion que le point d'ebullition ; et cette qualite , qui donne un moyen sir de reconnaitre la purete , est indispensable pour l'usage en question On May lllh I determined the meIting-points of the samples of naphthalene and benzophenone used during the experiments .
The thermometer used was a small sensitive one of GEISSLEE'S .
The resuIts were as foilows:\mdash ; Benzophenone .
Melting-point of a portion which had not been boiled or distilled . . . .
( 1 ) ( 2 ) Melting-point of a portion which had been used throughout all the experiments Naphthalene .
Melting-point of a portion which had not been boiled or distilled . . . . .
Melting-point of a portion which had been used throughout the experiments considerable variation in this ooeffioient would have but a small effeot on the fifth figure of the AND FREEZING POINTS BY THE PLATINUM THERMOMETER .
49 Tables were thus constructed , giving in each case the change in resistance of the wire for a difference of 1 mm. pressure in the vapour of the substance .
I give the tables used for comparison of observations with , and ; similar ones having been constructed for the other TABLE III .
resisbance .
For example , if 0780 be taken it would produce a change of 0106 per 10 mm. in the resistanr , of instead of 0105 , the number obtained by assuming it to be 0760 .
Subsequent experiments have shown that the approximate corrections thus obtained are sufficiently accurate .
For example , the resistance of at C. is and , .
, , : , C. is 8.0689\mdash ; 4.9543 approximate increase per at Now ( see Table XIX .
, experiments 6 , '7 , and 14 and 15 ) the true resistance of in steam under mm. , Difference for mm. Now the difference in temperature of steam at and 772 is given by the accepted authorities as Increase per , Here the result as obtained by experiment differs from the oximate 1esult by only 1 in 156 over a range oi mm. , which is the extreme variation in pressure met with during the experimen , and this difference would only cause a change of 1 in the fifth figure of R. The value of used in Table III .
was in no case so large as 200 ( greatest value ) , consequently the above is an extreme case .
The value given by Table .
for at gives the following formulae for the boiling-points of these substances:\mdash ; \mdash ; 760 Naphthalene , Benzophenone CCCXCI .
\mdash ; A. MR. E. H. GRIFFITHS ON THE ] RMINATION OF SOME BOILING Freezing Points of Metals .
9 .
The apparatus used was similar to that described by Messrs. HEYCOCK and NEVILLE in their paper on the melting-point of alloys ( ' Chem. Soc. Trans May , 1890 ) .
A large iron block , about 8 inches in height , had a circular hole about inch in diameter , bored from the top to within 2 inches of the bottom .
* The capacity for heat of this block was very great , and , therefore , its rate of alteration in temperature very slow .
The block stood on a large FLETCHER'S burner and was surrounded by two circular burners , thus the heating effect was fairly uniform .
A circular ring , having a slight spiral twist , and worked by a water-motor , moved rapidly up and down through the melted metal , stirring it thoroughly .
Down the centre passed the thermometer , its lower end reaching nearly to the bottom of the hole .
The space above the melted metal was filled with paraffin , oxidation was thus prevented , and that portion of the rod of the stirrer which passed out of the metal during the ascent , kept at an even temperature .
It was found possible to use paraffin with all the metals employed , except zinc , although , in some cases , the " " fuming\ldquo ; was excessive .
Zinc was , in consequence , difficult metal to work with .
I endeavoured to prevent oxidation and surface cooling by covering its surface with finely powdered charcoal .
The result was not entirely satisfactory , and I found it impossible to prevent a certain amount of oxidation during the experiments with that metal .
I do not know that this would affect the freezing-point ( unless the oxide was soluble in the molten metal ) , but it prevented any surfusion .
The temperature throughout the process of solidification was , however , very constant , and could be determined without difficulty .
From 400 to 600 grms. of metal were used in each case , and as far as time per , I tested samples from different makers .
It appears to be fairly established that , with certain rare exceptions , the presence of any foreign metal will be shown by the lowering of the freezing-point ; therefore , when different samples give different freezing-points , I have adopted the higher .
The phenomenon of surfusion was , in some cases , very marked .
When taking observations , the key was held down and the resistance of the box lowered as the metal cooled , so as to maintain the spot in its zero position .
When solidification commenced , the process had to be reversed , and the resistance rapidly increased ; for example , in one experiment on Sn the resistance of fell to , and then rapidly rose to .
When making these observations I was in a different room from that in which the metal was cooling , and this rise was my only indication .
It was , however , unmistakable , and , after the first few trials , easily measured .
The stirrer was usually arrested by the solidification of the * The surface of this hole and the stirrer coated with F by raising the block to a high temperature and passing steam into the hole for several hours .
See freezing-point of Pb , able X AND FREEZING POINTS BY THE PLATINUM THERMOMETER .
metal within 30 to 60 seconds after the rise .
In those cases where surfusion was absent , the cessation of cooling was sufficiently marked to allow of accurate determination .
, an 10 .
The platinum wire used for , and had a diamet , er of 005 inch .
Glass tubing was drawn out until of st sufficient bore to admit this wire , and was then cut into lengths of about inch .
The wire was then threaded through these pieces , the ends of which were heated until they adhered to the wire .
About two yards of the platinum wire were thus covered , and a stout silver wire was hard-soldered on to each end of it .
One of these pieces of silver wire was passed through a fine glass tube about ] inches long , placed within a second glass tube of about inch in diameter ; and the second silver wire was passed up the annular space between these two tubes .
The outer tube was widened at its lower end to inch external diameter , the glass-covered pla , tinum wire was folded up at the junctions of the glass threads and thrust into this wider portion , which was then fused at the open end .
The bulb thus formed was heated to a red heat , and allowed to cool slowly .
The whole apparatus had much the size and appearance of a sensitive mercury thermometer , with a bulb about inch long .
The upper end was fixed to a small board , and the ends of the projecting silver wires were soldered to two heavy brass connecting screws , insulated by ebonite .
The same sample of platinum wire was used in all three .
The silver wire in A was " " standard silver in and pure silver .
The total resistance of the silver wires was as follows:\mdash ; ohm at C. From the connecting-screws well-insulated stout copper wires ran to the resistance box , the same wires being used in all experiments with these thermometers .
The resistance of the wires and the connecting-screws was ohm .
The external resistance ( meaning thereby all resistance except that of the platinum wire in the bulb ) .
was of no consequence provided that it remained constant , nor would its variation be of any consequence provided that it was regular , which would probably be the case the same portion of the stem was always immersed .
I scarcely ated that my measurements would be sufficiently to render appreciable the alteration in resistance due to different immersions of the stem , the coefficient of increase of silver being less than that of platinum , and the external resistance only about of the coil .
Herein I was mistaken , and I afterwards ascertained that the consequent error might amount , in some cases , to nearly C. * For example:\mdash ; Thermometer Vapour of ; Barometer at 770 mm. Bulb only just within hypsometer . . .
, .
Resistance , , and 14 in .
of stem within hypsometer . . . .
, , This was an extreme case , and here the difference obm , nearly .
MR. Hi .
H. GRIffFITHS ON THE DETERMINATION OF SOME BOILING However , in nearly all my experiments with these thermometers the same fraction of the stem was immersed , and the error due to this cause inappreciable .
This difficulty could , of course , be met by usin stouter silver wires .
A large number of determinations were performed with thermometers , and and they have remained quite constant although roughly used .
I have compared them with several large bulb mercury thermometers , and find that they ) , ssume the surroumding temperature more rapidly than some and less rapidly than others .
To illustrate the close correspondence of the experimental results , and also the accuracy of the corrections used for the variations in the temperature of the box and the height of the barometer , I extract the following page from my note-book .
( The resistances in all cases include the exterior resistance .
) TABLE IV.\mdash ; Resistance of A at Temperature 10 C. Although I afterwards discarded this form of thermometer , I have described it , because it is inexpensive , easily made , and may be used in all cases where an accuracy bounded by 1o C. is sufficient .
11 .
The results obtained from these thermometers are as follows ( I consider it unnecessary to give the records of individual experiments):\mdash ; TABLE V. AND FREEZING POINTS BY THE PLATINUM THERMOMETER .
Denoting the platinum temperatures by , and ) , respectively , then Table V. gives us the following results TABLE phur 4 B.P. ( at 760 mm. ) aniline .
12 .
On plotting these numbers it was evident that they were not as concordant as might have been expected , considering the close agreement in the experimental results .
No difficulty was experienced in drawing a smooth curve through the points determined from melting ice , steam , naphthalene , benzophenone , and sulphur , but the curves thus drawn passed above mercury ( REGNAULT ) and methyl salicylate ( RAMSAY ) , but below aniline Again , RAMSAY'S boiling-point for mercury gave a value of , which fell considerably above the curves .
Further , it was found impossible to draw a consistent curve through the points obtained from melting ice , steam , aniline , methyl salicylate , and mercury , even if the remaining points were disregarded .
The experiments were repeated with great care , and all the precautions mentioned in Sections 3 to 7 were adopted ( with the exception of the thermo-regulator to the resistance box ) .
The results , however , were practically the An equation of the form was taken ( where , the ordinate , , and ) , the values of and given by the boiling-points of water , naphthalene , benzophenone , and sulphur were substituted , and some intermediate points then deduced .
Curves , and ( Plate 4 ) were then drawn on a large scale , the vertical spaces ( on the original plan ) representing and the horizontal 2 All these thermometers were constructed from the same sample of platinum wire and glass , and the resistances of the coils are nearly equal .
The results are sufficiently close to lead to the conclusion that the is the same for and at any given temperature , the greatest variation from the mean being By means of curve A the mercury thermometers used by Messrs. HEYCOCK and were The values of the boiling-points of the remaining substances , as obtained from curves , and , are as follows:\mdash ; TABLE On ( jomparing these curves with one plotted from the numbers given by CALLENDAR Phil. Trans , it was evident that they differed considerably : the rise is steeper , and is invariably a larger quantity .
This difference may be in some measure due to the behaviour of the glass with which the wire is in contact throughout .
If , however , this is the case , it will not affect the temperatures deduced from the curve , that the resistance remains constant for each temperature .
I was anxious to see if a thermometer constructed from a different sample of wire would give the same platinum temperatures , and , if these differed , to ascertain if the resulting values of differed .
13 .
Thermometer D.\mdash ; The platinum wire used was inch diameter .
No silver or solder was used within the instrument .
A thick rod of platinum ran through the inner tube , and down the annular space between the two glass tubes was passed a tube of platinum foil , terminating about inch from the lower end .
One end of the fine wire was welded to the rod , and the wire was then wound as a spiral round the lower inch of the inner tube .
This spiral was then entirely covered by fine threads of glass ( laid lengthwise ) , and over them the wire was wound downwards to the end .
This process was repeated until five layers had been ormed , and the end of the was then welded to the platinum tube .
The glass buIb enclosing the coik was about the same size in , and C. This coil contained nearly 9 feet of wire .
Very few expeliments were performed with this thermometer as , owing to the large mass of glass in the bulb , it was very slow in acquiring the temperature of the surrounding vapour .
I , however , obtained one reading at all the previous points\mdash ; with the exception of methyl salicylate .
graduated .
I believe that they have given an acoount of the method adopted in their paper ( ' Chenz .
Soc. Trans May , 1890 ) .
AND FREEZING POINTS BY THE PhATINUM THERMOMETER .
55 TABLE VIII.\mdash ; Thermometer D. Melting ice B.P. of aniline alene .
benzophenone Plotting the same 5 points as before , and deducing from the curve ( curve D ) the remaining temperatures , we get Aniline Mercury Curve ( Plate 4 ) follows quite a different path from Curves , and ( the platinum temperature of the vapour of differing by as much as ) , yet the values of , deduced from the curves , differ by less than .
It must be remembered that the above represent solitary determinations , and , therefore , the resulting values are of little importance in themselves .
Nevertheless , they point to the conclusions ( see Curves , and ) ( 1 .
) That the platinum temperatures obtained by using different samples of wire , differently arranged , are not equal .
( 2 .
) That the " " fixed points\ldquo ; ( viz. , those obtained by assuming previous determinations of melting ice , steam , naphthalene , benzophenone , and sulphur ) give , in each case , a consistent curve .
( 3 .
) That the temperatures of the vapour of aniline , methyl salicylate : and mercury , deduced from these curves , are practically identical , but differ slightly from previous determinations .
These were borne out by subsequent experiments .
14 .
A different form of thermometer was now adopted .
I obtained from Messrs. JOHNSON and a thick platinum tube , inch internal diameter , and inches long , having a resistance of about 04 ohm .
Through this a hard glass tube was passed , projecting inch at one end .
Round this projecting portion was wound a slip of asbestos paper* ( obtained from BELL The non-conductivity of this asbestos paper was exposed to severe tests .
As compared with glass , it proved much the better insulator .
A tnbe formed of it was placed within a hard glass tube , and two tightly-fitting wires thrust within until the ends were as close as possible without contact .
This was placed in the circuit between four Leclauche cells and the galvanometer .
It was then slowly heated bo a bright red hea , , but no deflection was observed .
If , however , the wires were allowed to touch the glass , the deflection was considerable .
and Co. ) until the thickness of the roll of paper was equal to that of the platinum tube .
On this roll of asbestos the wire was wound , one end being welded to a projecting slip on the lower end of the platinum tube , the other to a bundle of thick platinum wires passing down the centre glass tube .
The coil consisted of about thirty turns .
The coil and the asbestos were then heated to a red heat and allowed to cool gradually .
A thin hard glass tube , closed at the lower end , was then slipped over the whole .
The top portion of the instrument was filled with paraffin , and the upper inches surrounded by an outer tube covered with asbestos yarn to prevent surface condensation .
Water was kept running through this outer tube during each experiment , the change in resistance of the connections caused by the changes of temperature in the tap water being less than could be measured .
One advantage of this form of thermometer is , that , owing to the diameter being the same throughout , it can be thrust through perforated corks , etc. The asbestos roll being of an elastic nature , presses the coil against the exterior tube , and thus the instrument is extremely sensitive .
If the hand be placed within an inch or two of the coil , when at the ordinary temperature of the room , the spot of light will be deflected off the scale .
In order still further to reduce the external resistance , silver wires , running from the upper end of the platinum tube to within 1 inch of its lower end , were hard-soldered to the platinum connections .
Similar wires were added in the same manner to the platinum wires in the inner tube .
Since the resistance of the platinum is so small , any current caused by differences of temperature within the thermometer will be short circuited , and thus will not visibly affect the galvanometer .
As the silver wires terminated about inch above the coil , the flow of heat along them was not likely to influence its temperature .
Thermometers thus constructed have proved satisfactory in every respect , and can be safely exposed to any temperature up to a dull red heat .
Thermometers , and 15 .
I obtained from Messrs. JOHNSON and MATTHEY some exceedingly fine platinum wire\mdash ; diameter 001 inch , and a resistance of about 48 ohms .
foot .
Great difficulty was experienced both in welding this wire its connections and in winding the coil .
A piece of this wire , about 20 inches in length , was wound into a coil inch in length .
The external diameter of the thermometer was only inch , and was the same throughout its length .
Eight inches projected below the condenser tube .
The total resistance from the coil to the connecting screws at the top of the instrument was ohm .
About two-thirds of the thermometer was kept at a constant temperature by the tap water .
Thus the part exposed to change of tempera- ture was about ohm , and as the greater portion of it was always immersed , any variation caused by difference of immersion ] be safely disregarded .
AND FREEZING POINTS BY THE PLATINUM THERMOMETER .
57 The same remarks apply to , and , which were constructed in the same manner , except as regards the wire used for the coil .
Before using these thermometers the wires running to the resistance box in the next room were replaced by thick copper rods , carefully insulated , the resistance of which , and all connecting screws , was014 ohm .
These rods were only subjected to the same changes of temperature as the air of the room , and any difference thus produced was not measurable .
, however , had one fatal defect .
The capacity for heat of the coil was so small that the difficulty previously referred to ( the raising of its temperature by the current used when determining the resistance ) compelled the reduction of the E.M.F. used to such an extent , that the deflection of the galvanometer was too slight to be accurately measured .
* I determined , however , before destroying the thermometer , to ascertain the resistance at the " " fixed points using seven cells with a battery resistance of 1000 ohms , and using for the arms of the bridge .
The results were as follows:\mdash ; TABLE IX .
Melting ice Steam . .
Aniline ( 760 mm. ) Naphthalene ( 760 mm. ) Benzophenone ( 760 mm. ) , Sulphur ( 760 mm ) $ For example erraaacyynnntion ( 1 ) and ( 2 ) the thermometer in steam under same pressure ( 781 ) .
3 ) and ( 4 ) , , , , , , , , ) .
( 5 ) and , , in melting ice .
The fall in the resistance of the coil due to a diminution in bhe current passing through it , is here very marked .
I omitted to determine mercury B.P. by means of MR. .
H. GRIFFITHS ON THE DETERMINATION OF SOME BOILING This curve Curve Plate differs considerably from the preceding it , however , approaches more nearly to the curve plotted from CAILEIVDAR'S numbers .
Ascertaining from it the temperature of aniIine B.P. at we get , thus bearing out the conclusions arrived at in section 13 .
The thermometer was reconstructed and a wire of diameter007 inch ( the same sample of wire as that used for D ) , and length about 20 inches , was substituted for previous coil .
This thermometer was labelled E. Diameter of wire in inch ( the same sample of wire as that used for and C ) .
was dest.royed during some experiments with melted Zn and was replaced whose coil was constructed of a rather greater length of the same wire .
Both and had an external diameter of a quarter of an inch .
As regards the external resistance , length , etc. , they practically coincide with 16 .
A series of observations led me to the conclusion that the current produced by the following arrangement was the greatest that I could use if I wished to prevent any observable rise in temperature due to the cutTent .
F. 7 Leclanche cells and 500 ohms introduced into battery , arms of bridge E. , , , , 200 , , , , G. ; , .
, 500 , , , These conditions were maintained throughout all the remaining experiments .
TABLE X. -Resistances of , and after correction for temperature of coils and height of barometer.$$ Melting ice B.P. of aniline .
Steam , , naphthalene benzophenone .
mercury , , sulphur From Table X. we get the following results:\mdash ; *Tables of the individnal experiments from which these numbers are obta , ined are given in the Appendix ( pp. 65-69 ) , except those of aniline and mercury which are given in Table XIV .
AND FREEZING POINTS BY PLATINUM THERMOIIETER .
B.P. of aniline mercury 2 naphthalene .
benzophenone These numbers ( with exception of those deduced by assuming the standard numbers for aniline and mercury ) give , as in previous cases , a smooth curve , although differs considerably in the last three columns .
Assuming , as before , , and substituting the values of and given by the above table , we get TABLE XIL From Table XII .
we get following values of Table XIII.\mdash ; Values of .
When Ten points were thus obtained on each curve , five by direct experiment ( Table X. ) and five from the equation .
The curves Curves Plates run without the slightest difficulty MR. E. H. GRIFFITHS ON THE DETERMINATION OF SOME BOILIN($ through these points .
They are plotted on a scale in which a vertical distanoe.of*inch corresponds to a difference of 1o in , and its value can be read with certainty to The marked difference in the coefficients of the equations may be due ( 1 ) to differences in the quality and thickness of the platinum wire used ; ( 2 ) to differences in the total external resistance ; ( 3 ) to differences in the asbestos and glass , different samples being used in each .
That , under the same circumstances , the resistance of each thermometer remains constant is , I believe , clearly shown by the Tables given in the Appendix .
The temperatures of the boiling-points of aniline and mercury ( at 760 mm deduced from curves , and ( using the numbers given in Tables X. and XI are as follows:\mdash ; The greatest variation from the mean value bein My attention was , henceforward , confined to determining various temperatures by means of these curves .
After having used thermometer for the determination of the boiling-point of methyl salicylate and the melting-points of Sn and Pb , it met with an accident , and thereafter I was dependent upon and G. At regular intervals I tested these thermometers by means of a hypsometer , but was not able to detect any alteration in them .
I give particulars of the remaining experiments in full , without omission , as I wish to show how comparatively insignificant the variations are in each case .
I hope that the replacement of glass by porcelain tubes will hereafter enable me to determine the freezing-points of metals melting at higher temperatures .
* Mean of previous results :\mdash ; Amline . . . . . . . .
Meroury . . . . .
AND FREEZING POINTS BY THE PLATINUM THERMOMETER .
AND FREEZING POINTS BY THE PPLATINU THHEROOMETE .
63 MR. E. H. GRIFFITHS ON THE DETERhfINATION OF SOME ]S BOILIN Summary of Results .
17 .
( 1 .
) Different thermometers , whose coils are formed of different specimens of platinum wire , encased in glass or wound on asbestos , do give the same platinum temperature when at the same actual temperature .
( 2 .
) The readings of each thermometer are constant , whenever its temperature is the same .
( 3 .
) The platinum temperatures obtained by assuming CRAFTS ' numbers for the B.P. of naphthalene and benzophenone , and REGNAULT'S for that of sulphur , are consistent with each other , and with and ( 4 .
) Using the substances mentioned in ( 3 ) for the determination of the " " fixed points of platinum thermometers , the thermometers thus graduated agree in assigning the values given in Table XVI .
as the air thermometer temperatures of the boiling and freezing points of the different substances .
The difference between the numbers obtained by using different thermometers on the same specimen in no case affects the fourth figure of the TABLE XVI .
Temp. by air therm. Boiling-points\mdash ; ( Bar .
Aniline .
Methyl salicylate 22319 , Bar .
Triphenyl methane ( Bar .
760 ) Mercury . . .
Freezing-points\mdash ; Tin . . .
232 .
Bismuth Cadmium 321.67 Lead .
328.78 Zinc 421.23 I give , in full , tables showing the results of the individual experiments by which the\ldquo ; fixed points\ldquo ; on , and were determined .
* It must be remembered that these determinations were obtained by use of the air thermometer , whose readings agree with the mercury thermometer at and .
It follows that all the temperatures ascertained by my thermometers are expressed in terms of the air thermometer .
If subsequent investigations .
establish new values for the air-lemperature at these points , the figures given in the Tables in Appendix wonld supply sufficient data for the correction of the temperatures deduced from them .
AND FREEZING POINTS BY THE PLATINUM THERMOMETER .
NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI .
\mdash ; A. NDCCCXCI.\mdash ; A. * 68 MR. E. H. GRIFFITHS ON THE DETERMINATION OF SOME BOIhING GO AND FREEZING POINTS BY THE PLATINUM THERMOMETER . .
Note on the Possible of the Differences oj as shown by Different Thermometers , when is the same .
If the external resistance is constant , it cannot the value of whatever its magnitude .
It has been that the discrepancy between my curves and CALLEN , may be accounted for by the difference in the alteration of the " " stem\ldquo ; resistance due to changes in its mean temperature .
If such were bhe explanation , it is probable that the discrepancies would be more marked in , and the whole of the stem would be liable to alteration in temperature ) than in , and ( where early the whole of the immersed portion was kept at the temperature of the tap water .
No such contrast is , however , observable .
On comparison of the pt given by thermometer with that given by CALLENDAR the same temperature , I get the following numbers:\mdash ; Thermometer . . . . .
When then CALLENDAR Table , , , , Difference Now the mean temperature of that small portion of the stem which was immersed and exterior to the condenser must , in all cases , have been somewhere between the temperature of the tap water and that of the coil .
Assuming that , and would be the resistances if the temperature of the stem remained constant at the temperature of the tap water , and , and the observed resistances , then and fortiori , Assuming the mea , n resistance of the stem to be a function of , then Denoting the pt as given by by , and , as given by CALLENDAR by pt , then \amp ; c. , \amp ; c. Now the total stem resistance of was about .
ohm at * Phi ] .
Trans A. , 1887 .
AND FREEZING POINTS BY THE PLATINUM THERMOMETER .
In no case was the portion of the stem immersed and exterior to the condenser more than one-ninth of the whole stem .
Assuming , however , that one-fourth of the whole stem is subject to variation , we get ohm as its resistance .
Of this portion one end was at the temperature of the coil , the other at that of the tap water ; it is probable , therefore , that its mean temperature was about .
Taking the circumference of increase of platinum as 003 , we can deduce a rough approximation ( taking ; see Table XIX .
) for pt when We get , etc. The difference of cannot , therefore , be due solely to the difference in the mean stem temperature .
A more probable explanation is to be found in the decrease in the resistance of glass as its temperature rises .
The platinum tube and rod in the stem were separated by a thin glass tube only , of which the lower six inches were exposed to the extreme changes of temperature .
Again , the coil was in contact with glass throughout its whole length .
A fall in the resistance of this large section of glass sufficient to reduce the total resistance at from to ohms would account for the difference between and pt .
From EVERETT'S ' Physical Oonstants , ' I extract the following numbers:\mdash ; Specific resistance of glass at ] , .
, , Assuming these numbers , the above fall in the resistance would be accounted for on the supposition the total resistance of the glass at 20 did not exceed about ohms , which , bearing in mind the large section , is not an impossible number .
This explanation is borne out by the fact thai Curve is the steepest , showing the greatest falling off in the increase of resistance .
In the section of the glass conductor was much greater than in any of the others , the wire being in close contact with glass on both sides throughout the whole length of nine feet .
Again , slight diffel.ences in the quality of the glass , or in the nature of the contact between it and the platinum would be sufficient ( on this hypothesis ) to account for the differences between the curves .
It is , however , conclusively shown by the experimental results that , whatever is the cause of the discrepancy , it does not affect the determinations of temperature made by these thermometers .
[ January , 1891.\mdash ; In a communication which Mr. CALLENDAR and I had the honour of pl.esenting to the Royal Society on December 18th , 1890 , we gave the results of an investigation into the boiling-point of sulphur .
We found that the temperature by normal air thermometer of the saturated vapour of sulphur under a pressure of 760 mm. of mercury at C. and C.G.8 .
( sea-level in lat. , is C. It follows that the numbers given on page 64 require revision in the sense indicated by the note on that page .
A constant error of 5 mm. in the barometer scale used by me has subsequently been discovered .
The numbers obtained by the revision thus rendered necessary are given by us in the communication above referred to .
] .
A.

