235 On a New Type of Electric Furnace , with a Redetermination of the Melting-Point of Platinum .
By J. A. Harker , D.Sc .
, Joule Student of the Boyal Society , Assistant at the National Physical Laboratory , Teddington .
( Communicated by R. T. Glazebrook , D.Sc .
, F.R.S. , from the National Physical Laboratory .
Received March 21 , \#151 ; Read April 13 , 1905 .
) CONTENTS .
PAGE I.\#151 ; Preliminary discussion .
Use of oxides at high temperatures , etc ... .
235 II.\#151 ; Previous work with solid electrolytic conductors ... ... ... ... ... ... .
236 III .
\#151 ; Description of new type of furnace ... ... ... ... ... ... ... ... ... ... . .
237 IV .
\#151 ; Use of same for melting-point determinations ... ... ... ... ... ... ... .
239 V.\#151 ; Thermo junctions used and their standardisation ... ... ... ... ... ... . .
239 VI.\#151 ; Method of experiment ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... 240 VII.\#151 ; Summary of results and Tables I and II giving details of experiments ... 241 VIII.\#151 ; Data regarding the thermojunctions and Table III ... ... ... ... ... ... . .
243 IX.\#151 ; Properties of the thermoj unctions ... ... ... ... ... ... ... ... ... ... ... 243 X.\#151 ; Result of the determinations ... ... ... ... ... ... ... ... ... ... ... ... ... . .
246 XI .
\#151 ; Doubt as to validity of extrapolation .
Confirmation of formulae at melting-point of nickel , with Table IV showing results ... ... ... 246 XII .
\#151 ; Final value for melting-point of platinum ... ... ... ... ... ... ... ... ... 247 XIII .
\#151 ; Use of the new type of furnace in various physical investigations .
248 XIV .
\#151 ; Conclusion ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... 249 XV.\#151 ; Bibliography ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... 250 I. An investigation is now in progress in the thermometric department of the National Physical Laboratory , having for its object the design of an electrical method of measuring temperatures from 1200 ' C. upwards in some such way as temperatures below this value may now be determined by an appropriate thermocouple or resistance thermometer .
It is , of course , common knowledge that metals such as platinum and alloys of the platinum group are unable for prolonged periods to withstand , without alteration of their structure and properties , the effects of temperatures above about 1200 ' C. , particularly if in presence of even small quantities of certain gases , such as are very difficult to ensure shall be completely absent , especially at the higher ranges .
This is the case with nearly all methods of heating , whether electrical or otherwise .
The brilliant researches of Moissan on the stability of all kinds of substances at very high temperatures , and the recent work of Nernst culminating in the invention of his well-known lamp , in which the light236 Dr. J. A. Harker .
New Type of Electric [ Mar. 21 , giving filament is formed of a complex mixture of oxides of rare earths , directed the attention of the author to this class of bodies .
From a study of their behaviour he has been able to work out a method of attaining by very simple means very high temperatures , which may be controlled with great ease .
A preliminary account of this new type of electric furnace , with some results of investigations undertaken with it , forms the subject-matter of the present paper .
II .
The author was unaware at the outset of the work that the fairly obvious idea of making large tubes of solid electrolytic conductors had been previously applied , and had come across no accounts of any previous work in the usual scientific journals , but during the progress of this investigation he has found , by searching in the patent literature of England , Germany , \#166 ; and the United States , that more than one investigator has been working on the subject .
Nernst himself mentions the idea of using a tube , and H. N. Potter , in 1902 , took out American patents for a tube furnace on this principle , designed for the baking of Nernst glowers in lamp manufacture .
He appears to have had difficulties in getting uniformity of specific resistance throughout the tube , and suggests building up the wall in insulated sections , each fed from an appropriate circuit with series resistance .
Mr. Solomon , of the British Nernst Light , Limited , also used small tube furnaces of the same kind for the baking of glowers .
His furnace tubes , about 9 mm. interior diameter and 100 mm. long , after being formed and treated to render them initially conducting , were mounted in a small copper trough in pure powdered zirconia , and heated by a gas burner .
Such a tube when cold has , after drying , a resistance of at least 1000 megohms , but from about 400 ' C. upwards this falls off so rapidly that on a 250-volt circuit it will light up at a low red heat , the gas being then extinguished .
At the highest safe-working temperature such a tube drops about 100 volts with a current of 5 to 6 amperes .
After having had the benefit of Mr. Solomon 's experience , with numerous excellent hints from him as to experimental details , the author built and studied the behaviour of a number of these furnaces .
Experience showed that the point in them most likely to fail was the contact between the platinum electrode and the tube .
Further , that the material had a very large coefficient of expansion , and that , probably largely on this account , severe strains were set up between the outer and inner layers on fairly rapid heating and cooling , thus tending to make the life of such a furnace very uncertain .
1905 .
] Furnace , and the Melting-Point of Platinum .
In order , therefore , to avoid the necessity of having contacts on the tubes capable of carrying relatively large currents , and the consequent severe strain on the parts , the author conceived the idea of reaching the high temperature desired by a " cascade " arrangement such as is familiar in all the older methods of obtaining low temperatures .
To reach a temperature of 2000 ' C. , it was , therefore , arranged to only put through the tube itself sufficient energy to raise its temperature 1000 ' C. above its surrounding , and to keep the surrounding itself at about 1000 ' C. by means of a protecting jacket heated in any convenient way by an independent electric circuit .
III .
Fig. ( 1 ) shows the arrangement adopted in the small type of vertical furnace used in the experiments detailed later .
AB is the central conducting tube , which in this case was about 10 mm. internal diameter and 60 to 70 mm. long .
CD is a tube of hard porcelain or other suitable refractory Fireclay tube .
Conducting tube .
Pure zirconia powder .
Nickel heating spiral .
Fig. 1.\#151 ; Diagrammatic sketch of furnace for platinum melting-point determinations .
238 Dr J. A. Harker .
N Type of Electric [ Mar. 21 , material , 30 to 40 mm. internal diameter , on which is wound the heating spiral of nickel wire , protected from oxidation by some suitable means .
The space between is filled with powdered zirconia , which should not contain any appreciable admixture of any other substance , otherwise it gradually becomes a conductor at very high temperatures .
When the furnace is in action , in this layer of zirconia there is a falling gradient of temperature from the centre outwards , which suffices to prevent the nickel spiral from becoming unduly heated by radiation from the inner tube .
The current is led into the tube by platinum fiexibles , enveloping it at A and B and joined by autogen soldering to nickel or plantinum wires leading to suitable terminals .
To a similar pair of terminals on the nickel wire heating circuit is connected an appropriate source of current with regulator and ammeter .
To enable the tube to be heated fairly uniformly to above the upper electrode , it is sunk 20 or 30 mm. below the level of the top of the nickel spiral , and a mouth-piece of rather wider unglazed porcelain tubing is slipped over its upper end , which can be closed if necessary by a stopper .
A capillary tube , made of the same mixture as the furnace tube proper , may be fitted below for the introduction of any gas desired .
To light up such a furnace current sufficient to dissipate 100 to 300 watts is passed through the nickel spiral , the safe maximum amount of energy supplied during the heating up of furnace depending on the details of construction and the nature of the protecting medium in which the nickel is buried .
As a rule , less than half-an-hour suffices to bring the central tube to a red heat .
The terminals of the tube are meanwhile connected to the supply of current at from 200 to 500 volts.* Since the temperature coefficient of practically all solid electrolytic conductors is negative and very large , the current in them becomes unstable at very high temperatures , unless a sufficient steadying resistance is always left in series with them .
Tor furnace work it suffices if 20 to 25 per cent , of the whole volts of the circuit are dropped on the series resistance .
Care should be taken with furnaces run for long periods at very high temperatures that the temperature of the electrodes does not rise beyond the melting-point of platinum .
Though this is easily arranged for in the design of small tube furnaces , it becomes more difficult to ensure in those of larger size .
In these the conductor is arranged in the form of a reel , the electrode wires wound round the wider part at the ends being thus more easily kept relatively cool .
* Short furnaces can be lit up on a 100-volt circuit .
1905 .
] Furnace , and the Melting-Point of Platinum .
IY .
In the type of furnace just described , when about 150 joules are generated in the jacket circuit , a current of about 2 amperes suffices to take the central portion to about 2000 ' C. , the volts dropped on the tube depending on the distance between the electrodes and the condition of the tube material , but rarely exceeding 100 volts at the higher temperatures , and often being only 60 or 80 .
It was found that with a well designed regulating resistance , the control of the temperature of the inner walls of these surfaces was so perfect between 800 ' and 2000 ' C. , that well-defined melting-points could be obtained in them on very small quantities of substance .
Using a suitably supported thermojunction of bare platinum , with platinum-rhodium or platinum-iridium , it was found possible , even with fairly rapid heating , to obtain very concordant values for the melting-point of the platinum wire forming the junction .
It was further obvious that , if the melting-point of platinum were accurately known , this method would furnish an excellent and easily determined high temperature fixed point for thermojunction standardisation .
Y. Having in the laboratory a considerable number of specimens of platinum and platinum alloys of different composition from different sources , some of which had been accurately studied as thermojunctions over a large range , a selection was made from these of a number of characteristic specimens , giving widely different electromotive forces , and the formula connecting the thermoelectromotive force with temperature was carefully determined for each .
In most cases this was done by direct comparison in a specially arranged electric furnace with the laboratory standards , from somewhat below 400 ' C. to over 1250 ' C. In certain other cases the standardisation was made by means of selected fixed points .
The details of these methods of standardisation and of the potentiometer used , which was specially designed for thermoelectric work , were described in the author 's earlier paper on the " High Temperature Standards of the National Physical Laboratory , " published in the 'Phil .
Trans. ' for 1904 , vol. 203 , pp. 348 to 384 .
As a measure of the accuracy attained it will suffice to say that the coefficients for the type of formula E t = \#151 ; a-f fi + iy were calculated by least squares for each junction from the observations made , and that in a comparison of six junctions with two standards embracing 10 to 12 temperatures spaced fairly regularly between 400 ' and nearly 1300 ' C. , it was seldom that the difference between the observed values of the E.M.E. of any junction at any point , and the value of the E.M.F. for that temperature 240 Dr. J. A. Harker .
New Type of Electric [ Mar. 21 , calculated from the formula , differed by as much as 1 ' C. , and in no single case exceeded 2 ' C. The type of formula chosen is the same as is given in the paper previously referred to , and the scale of temperature adopted is the one there defined , which was shown to agree within the limits of error with the scale of temperature used at the Reichsanstalt , as established by the experiments of Holborn and Day .
Below 300 ' C. a formula of this kind having a term which does not vanish when t = 0 , does not represent the E.M.F. exactly , since from about this temperature downwards there is a marked change in the curve of E.M.F. for all junctions made of the platinum metals , and on this account , therefore , it is not permissible to use any of these formulae for downward extrapolation except over quite a limited range.* YI .
The experiments with each junction were made as follows:\#151 ; The two wires selected were melted together in a small oxidising oxyhydrogen flame , and after cooling the whole was mounted on a convenient insulating stand with fine adjustments .
The " hot " junction was brought to a central position in the heated furnace at an appropriate height , and the two " cold " junctions were placed in glass tubes in ice .
The observer then followed the the rise of E.M.F. on the potentiometer while an assistant gradually increased the furnace current as required .
When the melting-point of platinum was reached , if the rate of rise were not very much too rapid , an exceedingly well marked halt was obtained .
During the halt , the duration of which depended of course on the rate of heating of the furnace , one of three things usually happened , the particular phenomenon observed depending on the relative position , diameter and state of tension of the wires forming the thermocouple .
Either , first : The wires sprang away from one another , one of them usually coming into contact with the side of the tube .
The moment when this occurred was always easily recognisable on the potentiometer .
Or secondly : A globule of molten platinum formed on the end of the wires .
This sometimes attained such dimensions as to drop off , thereby breaking contact .
Or possibly , third : The globule of platinum formed gradually climbed up * Professor Callendar in his article on thermoelectricity in the supplementary volumes of the Encyclopaedia Britannica says " Holborn and Day have gone back to Tait 's method at high temperatures , employing arcs of parabolas for limited ranges .
But since the parabolic formula is certainly erroneous at low temperatures , it can hardly be trusted for extrapolation above 1000 ' C. " The results given later seem , however , to render it highly probable that , contrary to expectation , upward extrapolation is justifiable .
1905 .
] Furnace , and the Melting-Point of Platinum .
the rhodio-platinum or iridio-platinum wire , to a higher position in the furnace .
During its movement , small temperature oscillations are observed on the potentiometer , amounting to perhaps 2 ' or 3 ' .
The molten platinum did not appear to sensibly attack the other wire , any irregularities on its surface being practically as sharply marked after the experiment as before it and its total length remaining unaffected .
Provided the furnace wall near the junction were nowhere allowed to rise beyond the melting-point of the platinum-rhodium or platinum-iridium wire ( which with 10 per cent , alloys is in each case much above that of the pure metal ) , the temperature attained by the drop of molten platinum , whichever of the above alternatives happened , was very steady and independent of the current in the furnace .
There was usually no difficulty in repeating the setting of the potentiometer-slider for the same junction , without looking at it , to less than 1 ' C. , if the conditions are not greatly changed .
Provided the immersion in the furnace is sufficient , and that there is an ascending gradient of temperature from the furnace mouth downwards , varying the immersion considerably makes no difference in the point of balance attained .
After having had the method once explained , an observer not familiar with the apparatus , who was quite unprejudiced as to what value of the thermal force or temperature to expect , was able at once to take an observation , agreeing to within about 2 ' C. of the mean value previously obtained from a number of experiments with the junction in question .
VII .
In Table I ( p. 242 ) a summary is given of the values obtained in 66 determinations made on the melting-point of platinum , no observations recorded in the note-book having been arbitrarily rejected .
Three different furnaces were used\#151 ; a few experiments having been made in one arranged horizontally\#151 ; and for many of the junctions perfectly independent determinations were made on different days .
For the sake of better showing the degree of concordance obtained , the individual observations made with junction T9 , a commercial 10-per-cent , iridium of medium thickness\#151 ; the first set taken\#151 ; are given in Table II ( p. 242 ) .
This set were all made on one day , but by three different observers , one of whom had had no previous experience of the method .
42 Di .
J. A. .
Harker .
N Type of Electric [ Mar. 21 Table I\#151 ; Summary of Determinations of the Melting-Point of Platinum J unction .
Date .
Observers .
No. of experiments .
Highest value found .
Lowest value found .
Mean .
T9 1904\#151 ; September 16 ... J. A. H. 7 1713 1709 1711 m9 September 16 ... W. H. H. C. H. C. J. A. H. 4 1709 1707 1708 N. P. L. 3 September 20 ... R. t. a. 9 1714 1709 1712 October 13 October 25 J. A. H. J. A. H. 14 1707 1703 1705 Tl5 1905\#151 ; January 16 January 28 1904\#151 ; October 25 J. A. H. 10 1712 1703 1707 m4 October 27 1905\#151 ; February 6 January 16 H. C. H. C. 3 1705 1702 1704 M5 January 16 J. A. H. H. C. H. C. 3 1693 1691 1692 Tso January 28 J. A. H. J. A. H. 13 1698 1694 1696 t21 January 31 January 31 J. A. H. 3 1713 1711 1712 .
Table II.\#151 ; Observations on Melting-Point of Platinum with Junction T9 .
( Commercially Pure 10-Per-Cent .
Iridium from J. , M. and Co. ) No. Date .
Observer .
Value found .
Mean of group .
1 September 16 , 1904 J. A. H. 1709 2 55 55 ... ... 55 1712 \#151 ; 3 5 5 55 1713 \#151 ; 4 JJ J5 W. H. 1711 1711 5 J ) 55 J. A. H. 1709 \#151 ; 6 55 55 55 1712 \#151 ; 7 55 55 H. C. H. C. 1713 * 1905 .
] Furnace , and the Melting-Point of Platinum .
VIII .
In Table III ( pp. 244\#151 ; 245 ) are summarised further data regarding the different junctions , which are here grouped according to their composition .
Column I gives the distinguishing number of the junction .
II its composition .
III the diameter of the wires forming the junction .
IV its formula directly obtained by comparison with standards or at fixed points between 400 ' C. and 1250 ' C. V the error of the formula at 0 ' C. VI and VII the sensitiveness or " thermoelectric power " of the junction at 400 ' C. and 1700 ' C. respectively .
VIII.\#151 ; Percentage of the t~ term of the E.M.F. at 1700 ' C. of the whole E.M.F. at the same temperature .
IX.\#151 ; Mean value of the E.M.F. in microvolts given by the junction at the melting-point of platinum .
X.\#151 ; Number of experiments made .
XI.\#151 ; Mean value given by the junction for the melting-point of platinum in degrees Centigrade .
Junction N. P. L. 3 was part of the stock of 10-per-cent , plat.-rhod .
alloy obtained from Heraus in 1901 .
Full particulars regarding junctions N. P. L. 1 , 2 , and 3 made of this wire are given in the author 's paper on high-temperature standards previously alluded to .
Some of the observations in the second set included under this heading were made with junction N. P. L. 2 , the actual wire used in the direct gas-thermometer comparisons .
The three wires coincide to within the limits of accuracy attainable .
T15 is a sample of extremely-carefully prepared wire from Messrs. Johnson , Matthey and Co. , made under the personal direction of Mr. George Matthey , F.E.S. , to whose kindness the laboratory is indebted for the care spent on the preparation and analysis of a number of specimens of these platinum alloys .
Analysis revealed no trace of any other metal present but platinum and rhodium , the figures obtained for these elements being Pt = 89-9 , Eh = 8-98 both being directly determined .
IX .
Tu is a unique specimen in that it consists in what is probably the first wire investigated thermoelectrically , made from the large mass of exceptionally pure platinum-iridium prepared by Mr. Matthey for the construction of the " 6talons prototypes " of the kilogramme and metre for the Dr. J. A. Harker .
New Type of Electric [ Mar. 21 , Table I , II .
III .
IV .
No. of junction .
Composition of junction .
Diameter of wires in mm. Formula for E.M.F. of junction from 400 ' C. upwards . .
N. P. L. 2 and 3 t15 Pure rhodium alloys\#151 ; G-erman rhodium 10-per-cent .
alloy from Heraus , 1901 Purest English rhodium 10-percent .
alloy from Johnson , Matthey and Co. , 1904 0-62 0/ 51 -304 + 8 '165* + 0 -001663#2 j -250 + 7-953#+ 0-001842#2 j M4 Ta , Commercial rhodium alloys\#151 ; Commercial 10-per-cent , rhodium from J. , M. and Co. , 1903 Ditto , a second sample formerly called T6 0-48 0*31 -692 + 11 -55 t + 0 -001245#2 -446 + 11 '697# + 0 -001416#2 Tu .
Pure iridium alloy \#151 ; Purest English iridium alloy drawn from " etalon " wire , J. , M. and Co. , 1904 0*54 -409 + 15 -7635# + 0 '0007339#2 .
t9 m9 m5 Commercial iridium alloys\#151 ; 10-per-cent , commercial iridium , J. , M. and Co. , 1904 Ditto , another sample , 1904 " " 1903 0*33 0*50 0*33 -442 + 15 -6957# + 0 '0007349#2 -544 + 15 -2403# + 0 -001749#2 -781 +15 *5043# + 0 '001186#2 t21 Rhodio-platinum against iridio-platinum\#151 ; Iridium alloy of same sample as T9 against rhodium alloy of same sample as T20 .
0*33 and 0*311 +4 + 4 '0259#\#151 ; 0 '0006809#2 I .
III .J International Commission of Weights and Measures in 1886 .
The analysis .
A made by Stas and by St. Claire Deville of two samples of this alloy gave a j mean value Pt = 89-841 Eh = 0-135 Eu = 0-034 Ee = 0-066 Ir = 9-880 99-956 all the constituents being determined directly , and not by difference .
This alloy serves as a standard of comparison for all junctions of platinum-iridium wire .
Furnace , and the Melting-Point of Platinum .
III .
Y. Error of formula at 0 ' C. VI .
dE .
dT\amp ; t 400 ' C. VII .
\lt ; *E .
TTat 1700 ' c. VIII .
t- term x 100 whole E.M.F. at 1700 ' C. IX .
Mean value of microvolts at the melting-point of platinum .
X. No. of experiments .
XI .
Mean value of melting-point of platinum in degrees C. ' C. -37 9-49 13 '82 Per cent. + 26 18580 9 1712 -31 9*43 14 '21 + 29 18693 10 1707 -60 12 -55 15 '78 + 16 22590 3 1704 -38 12 -81 16 '51 + 17 23415 13 1696 -26 16 -35 18 '25 + 7 28581 13 1705 -28 16 -28 18 '19 + 7 28532 7 1711 -36 16 -64 21 -18 + 17 30600 4 1708 -50 16 -46 19 '52 + 12 28840 3 1692 + 1 3'40 1'8 -40 4905 3 1712 T9 and Mg are specimens supplied commercially to the laboratory at different dates , the products being stated to be of high commercial purity .
M5 is representative of several alloys of a similar character , not definitely called " pure , " but sold simply as 10-per-cent , alloy , the quality of the platinum wire forming the other side of the junction being also unspecified .
M4 and T20 are 10-per-cent , rhodium alloys of similar quality bought at different times , and T31 is a junction composed of the same alloy of iridium as the sample used in junction T9 , against the rhodium alloy used in T20 .
It will be noticed that the sensitiveness of this junction is only about a quarter of that of a pure platinum-rhodium alloy against platinum , and that it diminishes fairly rapidly with increase of temperature .
In the experiments with this junction the melting-point was taken of a small piece of pure platinum , which was twisted several times round the junction of the two wires .
After melting , this formed VOL. lxxvi.\#151 ; a. s 246 Dr. J. A. Harker .
Type of Electric [ Mar. 21 , into a large drop , which surrounded the point of contact of the two alloy wires .
The specimens of platinum used in making up the junctions were not specially analysed at the laboratory .
Five separate samples were utilised .
It is , however , certain that the specimens used in the first four junctions and in the experiments with T31 were of very high quality , but it was found that the wire used in M4 and M5 , when tested thermoelectrically against that of Tu and T15 , gave an appreciable thermoelectric force at the higher ranges , and it is probably owing to this cause , and not to any lowering of the freezing-point of the platinum by impurities , that the two low values 1692 ' and 1695 ' are due .
X. Rejecting , therefore , these two determinations , the agreement between the remaining seven junctions is of such a character that it appears highly probable that the value 1710 ' C. represents to within 5 ' C. the melting-point of platinum as determined by the thermoelectric method .
It may be objected , however , that in spite of the good agreement of determinations made with so many junctions , whose curves of E.M.F. against temperature differ so widely from one another , both as regards slope and degree of curvature , the extrapolation of a formula , which is only known to hold over a range of about 800 ' 0 .
, to cover an additional 500 ' C. , is unjustifiable , and that it is quite possible an intermediate point on the extrapolated part of the curves might not show such good agreement .
XI .
It was , therefore , decided to determine the freezing-point of nickel with some of the junctions as control .
Table 1Y gives the result of these determinations summarised .
The nickel used was a very pure sample of nickel berries from Brunner , Mond and Co. , made by the Mond process and found by analyses made at the laboratory by Dr. Carpenter to contain 99-6 per cent , nickel before fusion .
Three careful experiments made in an electric furnace in a reducing atmosphere by the ordinary crucible method with junction N. P. L. 2 .
gave 1428 ' , 1429 ' , and 1427 ' C. ; a previous determination made on the same material by Dr. Carpenter on a much larger scale , using a wind furnace and a junction from the same stock of wire as T9 , gave 1427 ' C. as the temperature of the commencement of solidification .
In both these determinations the junctions were protected by thin fire-clay tubes .
The two perfectly independent results agree within far less than the probable error .
Some further observations showed that it was quite easy to obtain a well- 1905 .
] Furnace , and the Melting-Point Platinum .
defined melting-point of a small granule of nickel round which the junction was wrapped , without sensible oxidation of the metal , in the new type of furnace used for the platinum points .
Preliminary experiments with different conditions of immersion , etc. , showed that the melting-point thus obtained agreed satisfactorily with the standard method even when no special gas was passed through the furnace , and that any nickel volatilised or diffused into the thermo-junction wires only affected these for a few millimetres of their length over the part actually in contact with the metal .
The results obtained with the six junctions used are shown in Table IY .
XII .
It will be noticed that the junction T30 , which reads 14 ' C. low at the platinum point is low by a similar amount at the nickel point , but that the first four junctions given in the table agree fairly well , those which are low at the platinum point being also low near 1400 ' C. These data furnish an answer to the possible objection which might be urged against the method that the comparative agreement at 1700 ' C. is really fortuitous .
Table IY.\#151 ; Melting-Points of Nickel and Platinum Compared .
( Most probable value for nickel 1427'\#177 ; 3 ' C. , for platinum 1710'\#177 ; 5 ' C. ) No. of junction .
Date of experiments .
No. of experiments on nickel .
Mean value .
Divergence from probable value .
Platinum point on same i junction .
Divergence from probable value .
N. P. L. 3 1905\#151 ; January 25 , J. A. H 3 1428 + 1 1712 + 2 t9 1904\#151 ; November 21 , H. C. H. C. 1 1427 4 0 1711 + 1 Tls ( crucible method ) 1905\#151 ; January 28 2 1420 - 7 1707 - 3 t14 " 28 5 1422 - 5 1705 - 5 T20 " 31 4 1412 -15 1696 -14 T21 February 6 5 1419 - 8 1712 4- 2 The value given by these experiments for the melting-point of platinum , 1710 ' \#177 ; 5 ' C. , is very considerably lower than the previously accepted numbers .
Of these earlier determinations undoubtedly the most important is the one made by Yiolle , which was a calorimetric estimation , depending on the extrapolation of a value determined for the specific heat of platinum from 1200 ' C. upwards .
Yiolle 's value is usually given as 1780 ' 0 , In his own memoir he says that 248 Dr. J. A. Harker .
New Type of Electric [ Mar. 21 , if this value is in error it will probably be found to be too high , as platinum is a metal which softens gradually , and when in this state the specific heat will probably be intermediate between that of the solid and the liquid , thus leading to an error in this direction .
XIII .
A few further words may be said regarding the furnaces , one type of which is here described .
Preparations are now being made for building them on a much larger scale , and it is hoped to publish shortly a further account of their construction and uses .
In addition to their use for work such as is here detailed , the type of furnace appears to be capable of numerous applications , both scientific and technical .
Of the scientific applications might be mentioned its use in the determination , at steady high temperatures in the absence of noxious gases , of the general physical properties of bodies , such as their coefficient of expansion , density , and specific heat in both liquid and solid states , and also vapour density and dissociation .
A horizontal form could easily be arranged for softening and annealing of long lengths of continuous wires , particularly such as cannot be directly heated electrically by current .
It is quite easy to arrange such a furnace to work vacuo , and in this form it might be of use in many chemical investigations .
In the preparation of metals like silver for such work as the determination of its electro-chemical equivalent , where the highest possible purity of the product is desirable , it would be quite easy , with an appropriate form of furnace , to refine by distillation considerable quantities of material , as was done by Stas in his classic researches .
The method of distillation in the oxyhydrogen blast employed by Stas suffers from an obvious disadvantage in that he says , after describing the process\#151 ; " Je dois avouer toutefois que dans les operations que je viens de decrire la moitie au moins de l'argent employ a ete perdue .
" In addition to avoiding this great loss of material , the liability of the metal to occlude gases would in the new furnace probably be much diminished .
For work on radiation , and as a " black body , " some rough preliminary experiments have shown the great advantages of this type of furnace over the carbon tube type , which lasts only a very short time , takes a very large current , and gives off large quantities of poisonous carbonic oxide gas For the realisation of the Yiolle Standard of light , in which the unit is the amount of light given off by 1 sq .
cm .
of pure platinum at its freezing-point , it is likely a furnace of this type might be very convenient , since it might easily be designed to render possible the desired result with a much Furnace , and the Melting-Point Platinum .
smaller quantity of platinum than has hitherto been thought necessary , and at the same time greatly increase the time of solidification , the only period in each experiment during which measurements can be made .
XIV .
In conclusion , I have to thank Mr. Swinburne for several suggestions made at the beginning of the work ; Sir William Crookes for kindly replying to queries , and for sending valuable references to the chemical literature of the rare earths forming the tubes ; my friends , Mr. E. S. Hutton , lecturer in electrochemistry at the Manchester University , for references to the American literature , Mr. Sheppard , of the British Nernst Light , Limited , for advice on practical points , and especially Mr. Maurice Solomon , for placing at my disposal valuable knowledge gained in his experience of Nernst filament making .
Thanks are also due to Mr. Gr .
Matthey , F.B.S. , who provided the very pure metals and alloys used as thermo-junctions ; and to the Director of the laboratory for provision of special facilities for this research , including the addition to the thermometric department of a specially-designed switchboard and safe regulating resistances for the 250 and 500-volt circuit of the local electric supply .
A list of the more important references , to work on electrolytic conductivity and furnaces , and to earlier determinations of the melting-point of platinum , is given below .
XV .
LIST OF REFERENCES TO WORK ON SOLID ELECTROLYTES .
Jabloclikoff in 1877 showed that half-baked kaolin conducts when hot .
Nernst .
British Patent , No. 19424 .
1897 .
A minute hollow cylinder of burnt magnesia will run as a lamp .
Nernst .
British Patent , No. 6135 .
1898 .
Deals with differences in behaviour of pure materials and mixtures as lamp filaments .
Pure magnesia only works with great difficulty .
Best materials for permanence at very high temperatures are zirconia and oxides of the zirconia group .
Nernst and Wild .
' Zeitschrift fur Electrochemie , ' vol. 7 , p. 273 , December , 1900 .
" Efficiencies of Yarious Mixtures as Filaments .
" Potter .
American Patents , Nos. 652,640 ( 1900 ) , 715,505 ( 1902 ) , 719,507 ( 1903 ) .
Nernst .
'Gottinger Nachrichten , ' 1903 , vol. 2 , alludes to an attempt to use tubes of solid electrolytic conductors as a furnace for vapour density determinations .
He discards these in favour of the iridium tube furnace described .
Nernst .
'Zeitschrift fur Electrochemie , ' vol. 9 , p. 622 .
Fuller description of same work .
Reynolds .
'Gottingen Dissertation , ' 1903 .
" Resistance of Solid Electrolytes at High Temperatures .
" [ May 3 Prof. H. B. Dixon and Mr. E. C. Edgar .
LITERATURE OF MELTING-POINT OF PLATINUM .
Violle .
'Comptes Rendus , ' vol. 85 , p. 543 , and vol. 92 , p. 866 , 1881 , gives 1775 ' and 1779 ' C. as determinations made by his calorimetric method .
Holborn and Wien .
' Wied .
Annalen , ' vol. 47 , p. 107 , 1892 , and vol. 56 , p. 360 , by thermoelectric method obtain 1780 ' .
Nernst .
' Physikalische Zeitschrift , ' vol. 4 , p. 733 , 1903 , and ' Wied .
Ann. Beiblatteiy vol. 25 , p. 686 .
Obtains by an optical method 1782 ' as melting-point of platinum iu iridium tube furnace .
Older determinations by Becquerel , Yan der Weyde , etc. , vary between 1400 ' and 2200 ' C. The Atomic Weight of Chlorine : an Attempt to Determine the Equivalent of Chlorine by Direct Burning with Hydrogen .
By Harold B. Dixon , M.A. , F.RS .
( late Fellow of Balliol College , Oxford ) , Professor of Chemistry , and E. C. Edgar , B.Sc. , Dalton Scholar , of the University of Manchester .
( Received May 3 , \#151 ; Read Mqy 18 , 1905 .
) ( Abstract .
) Although the atomic weight of chlorine has been determined by Stas and other chemists with extraordinary care , nevertheless owing to the very indirect methods hitherto used in making the comparison between chlorine and hydrogen , it is possible that a , constant error may occur in some link of the long chain of connecting ratios .
To join up the open ends of the chain by a direct comparison between chlorine and hydrogen , if it could be done with reasonable accuracy , would serve not only to detect any such systematic error , but would permit the accidental errors to be distributed and prevent their accumulation at the unconnected end .
According to Professor F. W. Clarke the accumulated " probable error " in his recalculated value for chlorine amounts to + 00048 ; the probable error of the mean of our nine determinations is less than + 0-002 .
It was at the suggestion of Professor E. W. Morley , of Cleveland , U.S.A. , that we have attempted this direct comparison by determining the weight of hydrogen which burns in a known weight of chlorine .
Our method was briefly as follows:\#151 ; Chlorine prepared by the electrolysis of fused silver chloride ( with purified carbon poles in a Jena glass vessel ) was condensed and weighed as a liquid in a sealed glass bulb .
This was attached to a vacuous " combustion globe " and the chlorine allowed to evaporate slow3y into the globe .
The hydrogen prepared by the electrolysis of barium hydrate solution was dried and then absorbed by palladium in a weighed vessel .
The palladium , on being warmed , gave off the hydrogen , which was ignited by a spark and burnt at a jet in the

