]\gt ; S 522 Drs. J. A. Harker and G. W. C. Kaye .
Electrical [ May 6 , : 4 .
When the mirror is held vertically the displacements are small , but the support cannot be so arranged as to get rid of them , as was the case with the normal component .
It is shown , however , that the critical size of mirror is comparatively large , very much larger in fact than any mirrors at present in use , the limiting radius being somewhere in the hbourhood of 12 feet in the two cases which are discussed in detail , when the telescope is so constructed that its focal length is five times its aperture and the aperture ten times the thickness of the mirror .
For a given mode of support a larger aperture may be used by increasing the focal length .
On the Electrical Emissivity and Disintegration of Hot Metals .
By J. A. HARKER , D.Sc .
, F.RS .
, and G. W. C. KAYE , B.A. , D.Sc .
, National Physical Laboratory .
( Received May 6 , \mdash ; Read June 19 , 1913 .
) Introductory .
In February , 1912 , the authors communicated to the Royal Society am account of some experiments on the emission of electricity from carbon at high temperatures .
* In this investigation all the experiments were concted at atmospheric pressure , and some evidence was brought forward that under these conditions the carriers of electricity appeared to consist almost wholly of ' sputtered \ldquo ; matter , and that corpuscles\mdash ; the carriers of negative electricity in high vacha\mdash ; played in these experiments but a minor or at any rate an indirect part .
It was shown that , even in the absence of any applied potential , it was possible for charged particles to escape in sufficient number from a hot carbon surface to give rise , at high temperatures , to curoents of ammeter rather than electrometer magnitude .
The particles appeared to be emitted with considerable velocity and the emissivity of the radiating surface sending the particles Qeemed to depend primarily on the temperature alone .
It was , however , influenced to some degree also by the nature of the surrounding gas .
Depending on this emissive property of carbon , kind of ionic dynamo was constructed , capable of lighting intermittently a small group of glowlamps ( 2 volts , 3 amperes ) .
* Harker and Kaye , ' Roy .
Soc. Proc 1912 , , vol. 86 , p. 379 .
reproduced wther substanoesP aarbon ceincentive cent interesting researctles oiamquesCion osputtering ottention seems tophenomenon onvestigated , thehave bbservers tccompanying etake furth rexperiments with aview tscertaining txtent tEmissivity ookes , atinum group , interest ohigh temperature phenomena except in high vacua .
It would appear also that of the experiments on metals at high temperatures , a very large proportion have been made on platinum , and , so far as we are aware , nearly all were carried out on a someil what small scale , generally with fine wires .
The present experiments , made at the National Physical Laboratory during the early part of last year , have been interrupted and are still incomplete , but nevertheless we have been led to believe that , there is much involved in the work which is obscure and demands further investigation , enough facts have been elicited to warrant the publication of a preliminary paper .
More especially does this seem desirable as considerable attention is now being given to the nature and properties of the emissions from hot bodies .
Thermal Sputtering .
The volatility of metals at temperatures well below their melting points is a familiar phenomenon to most workers .
A homely example is provided by the blackening which a common enough feature of carbon filament lamps and is occasionally displayed by tungsten lamps , especially when over-run .
Sputtered images of a definite outline depending on the shape of the filament can sometimes be traced on the bulbs of carbon lamps .
The abnormal volatility of iridium has been known to high temperature workers , in connection with its employment in thermocouples , furnace tubes and gas thermometer bulbs ; the extent of the effect is suffioient to prohibit the use of iridium or its alloys at temperatures much above 1000o C. Platinum is also known to exhibit a similar though much less marked * Crookes , ' ) .
Soc. Proc 1912 , , p. 461 .
A detailed account and discussion of the phenomena of ionisation by incandescent solid bodies is given in ) IX of Sir J. J. Thomson 's 'Conduction of Electricity through Gases , ' second , 1906 .
A more recent summary in considerable detail ma be found in H. A. Wilson 's ' The Electrical Properties of Flames and Incandescent Solids , ' 1912 .
It has therefore not been deemed necessary to give here any account of the work or a bibliography of the subject .
VOL. LXXXVIIL\mdash ; A. 2 524 Drs. J. A. Harker and G. W. C. Kaye .
Electrical [ May 6 , volatility ; examples of sublimed platinum crystals are often found on the : tubes of platinum-wound furnaces which , ha.ve been subjected to prolonged use at temperatures up to 1400o or so .
The thermal disintegration of the platinum metals was first investigated gravimetrically by a group of workers\mdash ; Berliner , Elster and Geite ] , and Stewart in 1887-9 .
Holborn , Hemming , and Austin , *took up the matter at the Reichsanstalt in 1903 , from the point of view of the high temperature : worker .
In Sir William Crookes'experiments ( loc. cit samples of the platinum metals were heated still air at atmospheric pressure .
His esults indicate a decreasing volatility in the order Ru , Ir , Pd , , and Bh ; the effects were unexpectedly large .
The volatilisation of a metal increases rapidly with the temperature .
In the case of platinum , iridium , and rhodium , the disintegration lessens as the pressure of the surrounding gas is reduced , and the effect would not there .
fore seem in these instances to be one of true subhmation .
Roberts has recently brought forward evidence that with these metals the volatilisation is effected through the intermediary of endothermic oxides more volatile than the metals themselves .
It may be added that almost all workers are agreed that the presence of oxygen is essential to bring about the disintegration of most metals .
Hydrogen and nitrogen are in general unfavourable to the effect .
Experimental .
After one or two preliminary trials , the apparatus described below was constructed and used in practically all the experiments .
It is shown in fig. 1 .
Through a round iron plate , forming the support of a large glass belljar , were fitted near the circumference two insulated vertical brass electrodes capable of carrying for short periods some hundreds of amperes without undue heating .
Projecting inwards some 15 cm .
from each electrode toward the centre of the jar was a stout iron strip with its inner end bent upwards .
To these iron strips were clamped adjustable jaws of the same material .
The strip of metal ( S ) to be heated was gripped between these jaws in a horizontal position , edge upwards .
The strips employed were of uniform thickness and width but varied somewhat from sample to sample .
The currents employed to heat them varied from about 50 to 250 amperes and the voltage across the strip from to a maximum ( rarely * Phil. Mag 1904 , vol. 7 , p. 388 .
PhiL Mag February , 1913 , , p. 270 .
When necessary to prevent contamination of the metal of the strip touched the jaws , which sometimes became considerably heated , a platinum lining was fitted to the part where the metals in contact .
other oetalscylinder ( provided woles ( iewing ttrip barger specimens water itudy oomewide.long between tlamps ; mitting aboutreached)bout 6Most otrips wpproximately 5mlnl 9ssivity.ounding ttlip ansulated cetely fmetal means of an optical pyrometer .
with suitable adjntments from an insulated support ( r ) passing through the base-plate of the apparatus .
One pole of the strip was connected through a currentmeasurer to the cylinder , the whole of the circuit being everywhere carefully insulated earth .
Eeat Supply .
The cylinder carried on an arm ' In order to avoid any local effects due to a different polarity at different parts of the strip , alternating current was always used for heating purposes .
This was taken from a step-down transformer iving a number of variable ratios and voltages from upwards .
The primary current to the transformer was supplied from a motor-alternator , usually at a frequency of about 90 cycles .
The field windings FIG. 1 .
of the alternator were excited from a direct-current 100-volt supply .
In series with the usual field regulator on the switchboard was arranged a convenient portable rheostat having a large number of segments .
This could be manipulated by the observer at the optical pyrometer , who regulated the temperature of the strip as required .
The control of the temperature was thus rendered extremely easy and accurate , and it was possible to maintain a strip at any given temperature to within a few degrees or even on the verge of melting for a considerable period .
In the later experiments the heating current and volts , and hence 526 Drs. J. A. Harker and G. W. , Kaye .
[ May 6 , approximately the expended in the hot strip , were measured by suitable alternating instruments .
Measurement of the Ionic Currents Produced .
ments caken osame gvanometer owing tanydifferent ranges oensitivity , arge aeflection wways aoyed designed study oheabout 1npere Fower ranges aresistance movin.ivan gvanometer igher .
ITangementThe ionic current.enerated vfrom afractiou omicroampere t means of this instrument , continuous and accurate current and volt measureif required , at any value of the current or voltage to be measured .
On account of the very rapid changes in the currents obtained in most cases with rising temperature , some device of this kind was essential to convenient work .
Employed .
In practically the whole of the experiments now to be detailed the gas employed was nitrogen of the ordinary commercial quality supplied by the British Oxygen Company and made from liquid air by the Linde process .
The usual procedure , after mounting the strip in position , was to exhaust , by means of a rotary oil-pump , the large -jar , which served to cover the whole apparatus , and rinse out twice with nitrogen direct from the cylinder before the final filling and pressure adjustment .
No special precautions were taken to secure either dryness of the gas used or great constancy of pressure during a run .
Owing to the energy dissipated in the strip , amounting in some cases to a kilowatt or more , the upper parts of the apparatus usually became quite warm after five or ten minutes ' run , and the pressures given are therefore only to be considered as approximate mean values .
Pressures below a few millimetres of mercury were intentionally not employed , as it was desired to avoid pressures so low that the vative ion became the unencumbered negative electron .
The experiments were conducted in nitrogen with the idea of using a gas which would be chemically inert to most of the metals .
We were aware from the outset that it was more than probable , from the experiments of Strutt and other workers , .
that the same kind of effects as we were seeking could be obtained at relatively low temperatures if more delicate means of detecting minute currents had been adopted .
We resolved , however , to study only the phenomena from the point of view of the currents generated of ammeter rather than electrometer order , and therefore more sensttive cuIxent measurers were intentionally not adopted .
1913 .
] Emissivity and of Hot Metals .
Measurement of Temperature .
The temperatures were measured by an optical pyrometer of the HolbornICurlbaum type made by Messrs. Siemens Bros. This instrument consists of a telescope of rather short focal length , iu the focal ] of which is fixed the filament of a small incandescent lamp .
In series with the lamp are a four-volt accumulator , a regulating resistance and an accurate ammeter .
In taking an observation , the central portion of the horseshoe-shaped filament is matched in brightness with the object whose temperature is to be measured and the current through the lamp noted .
Previous calibration of the arrangement serves as the basis of a table showing the connection between the current through the lamp and the temperature of the radiating object , applying , if necessary , the appropriate correction for want of " " blackness\ldquo ; in the surface of the radiator .
The observations are taken through an eyepiece consisting of one or more red glasses .
The absorbing device supplied with the instrument , consisting of two black glass mirrors , was used before the object glass in all cases , a few of the observations at the lowest temperatures .
In this arrangement the beam of light from the hot body is reflected twice at an angle of incidence and thereby weakened to about of its original intensity , thus allowing continuous observations to be made up to about C. without risk of over-running the pyrometer lamp .
Observations even to the highest temperatures are easy and a practised observer can follow continuously the temperature of the strip to within a few degrees up to its melting point .
Experience showed that , except the last two or millimetres at each end , the temperature of the strips was generally very uniform .
The temperatules recorded are , of course , black-body temperatures and are to some extent a function of the predominant wave- length in the red glass used in the pyrometer .
In some of the curves below these black-body temperatures are reduced to true temperatures .
As a rule , the error ( of the order of or so at ) due to absorption in the glass walls of the bell-jar was unimportant .
Employed .
We happened to have by us a number of suitable specimens of the platinum metals in the form of strip .
On account of their relative chemical inertness , high melting-point and , as Crookes showed , their large volatility , we decided to use these for the earlier work .
In the present experiments , platinum and iridium were employed ; and , later on , the metals iron , tantalum , copper , nickel , and brass .
The behaviour of the different metals is dealt with below .
\mdash ; The platinum strip when heated showed small positive currents 528 Drs. J. A. Harker and G. W. ( at about 110 : at higher temperatures th the relation between temperature and experiment ; the pressure was about 40 current increased rapidly as the pressure example of this ; the temperature of the Te ( Bl body ) FIG. 2.\mdash ; Platinum in Nitrogen at 40 mm. Pressure .
Relation between Negative Ionisation and Temperature .
1460o B.B. : there was practically no sputtering either on the bell-jar or elsewhere .
Iridium.\mdash ; At atmospheric pressure the iridium strip , when heated to high temperatures , emitted a black cloud of smoke , which ceased to be noticeable as the pressure was lowered .
The volatility of iridium under such conditions is course well known .
In one experiment with nitrogen at 4 mm. pressure iridium gave a maximum positive current\mdash ; about ampere\mdash ; at 1200o C. , which , on cooling and reheating , did not attain 1nore than a quarter of this amount .
At 1300o the negative current was plainly evident : it increased to 20 microamperes at , and steadily rose until when the trip burnt out 1913 .
] Ernissivity and Disintegration of Hot Metais .
Pressure FIG. 3.\mdash ; Platinum in Nitrogen .
Relation between Negative Ionisation and Pressure .
about 2300o C. ) the current had attained 80 milliamperes .
shows the growth of the current with temperature in this run .
The higher temperature readings are somewhat doubtful owing to the increasing deposit on the bell-jar .
Iron.\mdash ; Transformer iron was used in most of the experiments .
The heating currents ranged from 40 to 150 amperes at voltages up to 2 or 3 .
No measurable ionisation current was detected at temperatures below 1100o in nitrogen at atmospheric pressure .
At lower pressures a small positive current showed itself in some cases at about 1150o to 1250o .
Occasionally for some reason the positive effect would persist right up to the meJting point .
Fig. 5 shows the relation between current and temperature at a pressure of 9 mm. The sputtering of the iron during this particular run was very great .
As will be seen , the negative current increases very rapidly in the neighbourhood of the melting point .
Fig. 6 shows a similar run at a rather pressure and somewhat lower temperatures .
Our experience certainly seemed to be that the greater the volati the greater were the currents recorded .
In those experiments in which for this changed to negative at about .
At .
the negative current had increased to about 220 .
Fig. 7 shows the variation of the ionisation current with temperature .
The tantalum on this ocoasion yielded a very considerable deposit , which when analysed proved to be nearly all tantalum with a trace of iron .
@ 1913 .
] and Disintegration of Hot Metals .
Te pe a.tu re FIG. 6.\mdash ; Iron in Nitrogen at 9 mm. Pressure .
Relation between Negative Ionisation and .
Large Sputtering .
Tem perafure body FIG. 6.\mdash ; Iron in Nitrogen at 12 mm. Pressure .
Relation between Negative Ionisation and Temperaturs .
Slight Sputtering .
C. Kaye .
Electrical [ May 6 , Black -body ) FIG. 7.\mdash ; Tantalum in Nitrogen at 1 mm. Pressure .
Relation between Negative Ionisation and Temperature .
On a second heating at a slightly higher pressure there was very little sputtering , and the highest current recorded at about 1660o B.B. was about 50 microamperes .
Heating currents up to 80 amperes and volts were employed .
Nickd.\mdash ; Nitrogen at 4 mm. pressure .
Heating current up to 80 ampe]es .
There was no sputtering .
Small positive currents were recorded from 1200o upwards ; they increased suddenly to about 1 microampere as the strip melted ( 1440o Copper.\mdash ; Nitrogen at 4 mm. pressure .
Heating current up to 200 amperes .
Small positive currents from 1000o upwards ; a larger positive current flashed on the galvanometer as the copper melted At higher pressures , this final CUl'rent was smaller , e.g. 5 at 55 mm. Hg .
Brass.\mdash ; Nitrogen at 3 mm. pressure .
Heating current up to 110 amperes and volts .
There was no noticeable sputtering .
At temperatures of upwards small positive currents were noticed , which at the instant when the brass melted ( about ) became momentarily much larger .
The fused ends of the strip showed very plainly the copper present .
Carbon.\mdash ; From our experience with carbon at atmospheric pressure we 1913 .
] Emissivity and Disintegration of Hot Metals .
anticipated that the sputtering would be greatly enhanced by a reduction of pressure .
On trying the experiment with a carbon rod of unusually high purity , in a good vacuum , we were surprised with the result , for only a very slight deposit was obtained , and this was whitish and obviously due to traces of silica , etc. , present in the carbon .
The copper cylinder was absent and the temperature employed was about 2000o C. A similar negative result was obtained from an experiment oonducted with Acheson graphite in a hydrogen vacuum , and a temperature probably exceeding 2500o C. Incidentally , evidence was obtained in these experiments of the softening of carbon at temperatures from about 2500o upwards .
special clamps were made , and as they were not quite in alignment the carbon rod was clamped under lateral strain .
At the end of the run it was found that the rod , originally straight , was now crooked ; the carbon had evidently been plastic enough to accommodate itself to the strain .
The result is interesting having regard to a recent controversy on the subject .
Fig. 8 shows the carbon rod before and after the experiment .
FIG. 8 .
We were tempted to return for the moment to an experiment at atmospheric pressure .
AB ( fig. 9 ) is a rod of specially pure carbon Am meter FIG. 9 .
containing only about per cent. of impurity ( chiefly silica ) .
The ends were mounted in graphite blocks , and the rod was surrounded by a carbon cylinder ( CD ) insulated with bushes of mabor brick .
The cylinder served as the " " cold\ldquo ; electrode , and was joined through an ammeter to the inner rod , which was heated by transformer current and made as hot as our resources would permit ( probably not far from 3000o C The current used was of the order of 1000 amperes , the energy about 10 kilowatts .
The outer cylinder rose to about 1100o C. , and under these conditions the ammeter recorded a steady ionisation current of about 3S amperes .
The degree of conductivity akin to that of a metal .
heated bternati otook tcarbon tdown tentre ohich assed aisation oetween tectrod iurse , tsbeen removed beontinued heati rithdrawing theWas pHarker atition omention d metal tube at the end of half an hour or so it was found to be coated with a I deposit of carbon , hard enough and coherent enough to be slid off in short lengths .
Fig. 10 shows two examples of such carbon .
The method is of FIG. 10 .
interest as furnishing a possible means of obtaining by fractional distillation carbon of a purity unattainable by other methods- During experiment ionisation currents up to about 1 ampere were recofded .
It certainly seems that the ionisation effects obtained with carbon at higher pressures are due to the chemical activity of the carbon , for which the presence of gas in quantity is essential .
Positive Currents .
The emission of positive electricity was found to occur between ) and 1400o C. In the case of metals which melt within this range a sudden and striking increase in the positive current at the melting point was sometimes remarked with samples which had not been heated for more than a few minutes .
There were two factors which helped to display this effect to ; one , the nicety of control over the heating , by which it was possible to hold the temperature just below or on the melting point for some time ; the other , the large size of the samples which served both to enhance the effect and to give the metal some ability to hold together at the melting point .
In some instances the slow local fusing of part of the strip was watched in the pyrometer while temperature measurements were being taken .
The positive currents obtained from most bodies at moderate temperatures are now pretty generally attributed\mdash ; at any rate in part\mdash ; to the escape of occluded gas , which leaves the metal in an ionised condition .
The escape of this gas would naturally be greatly facilitated when liquefaction of the 1913 .
] Emissivity Disintegration of Hot Metals .
metal occurred , and it is to this cause we think the momentary increase in the positive current at the liquefying point is due .
En passant , we may note that our general experience with most of the metals we have tried has been that the positive current is augmented by the presence of oxygen .
The positive current readings were not steady or reproducible under specified conditions\mdash ; the currents usually drop considerably after a few minutes ' heating , and this without altering the temperature or pressure in any perceptible way .
Yegative Currents .
As will have been remarked , the negative ionisation increases rapidly with the temperature : the various curves shown above , connecting temperature and current , are all exponential in character .
It should be pointed out that in all cases the currents are unsaturated ; no potential whatever is applied to the hot metal apart from the small variable potential due to the alternating current .
The negative ions must accordingly derive their velocity directly or indirectly from the heat of the strip , or from some chemical reaction between the strip and the gas .
Sir J. J. Thomson*showed that charged particles of metal were among the carriers of current in the positive discharge from a hot platinum wire .
We believe that they are also often present in the negative discharge obtained from metals at higher temperatures , and that while the negative current at moderate pressures is usually due mainly to the negative gas ion , the current is ] augmented if the conditions are such that particles of metal also cross between the two electrodes .
These conditions are probably secured by the presence of a gas to which the metal is chemically responsive \mdash ; with most metals oxygen is so effective .
Thus the ionisation current is to be regarded as a direct manifestation of the chemical energy of combination .
With this view in mind we tried to gerate the effect by burning iron and magnesium wire in oxygen , the wire being surrounded by au outer cylinder , with a galvanometer in circuit between the wire and the oylinder .
We tried various forms of apparatus , but the experiments were interrupted before we were completely assured of their success .
If it should be that , at any rate at higher pressures , the ionisation current from hot metals is in some instances carried chiefly by particles of the metal , we have the means of deriving a mean value for the ratio of the charge to the mass for the particles .
For example , in one experiment with iridium , conducted in nitrogen at a pressure of about 20 mm. and a temperature of 1360o B.B. , the iridium strip lost mgrm .
in 11 minutes ; of this * J. J. Thomson , ' Conduction of Electricity through Gases , ' 1906 , second ed. , p. 217 .
eceived bHarker aKaye.onisation ourrent during this time was microampere .
Whence E.M.U. , and assuming that each particle can.ies unit charge , Now the mass of the iridium atom is about , so that the average number of iridium atoms in one particle is 1200 .
Apropos of the source of the ionisation current , it may be added we have repeatedly noticed that if the metal is hot enough to be giving out negative electricity , and is allowed to come into actual contact with the surrounding colder cylinder , the galvanometer shows a large current in the reverse direction .
The thermal E.M.F. is thus opposed to the ionisation potential .
This remark holds equally well for a hot carbon rod in an outer cylinder of cold carbon .
Straight Line Projection of Particles .
While there is no doubt that a good deal of the deposit which we obtained in these experiments was due to volatilisation and condensation as ordinarily understood , we think that this is not the whole explanation , and that the effect is partly due to the straight line emission of particles of matter from the heated metal .
* Under most conditions , it is true , this rectilinear propagation is confused by general sublimation , but we have noticed it on several occasions when the circumstances were favourable .
In one experiment a strip of iridium ( S ) was heated , edge upwards , in nitrogen at about 20 mm. pressure .
At the end of the run it was noticed that there were two horizontal bands of deposit on the inside of the copper enclosing cylinder , each one opposite and roughly the same width as the vertical face of the strip .
At points and opposite the edge of the strip there was little or no deposit .
Fig. 12 is a photograph of the iron deposits FIG. 11 .
obtained on the bell-jar , in the proximity of the brass uprights fig. 1 ) .
In this experiment there were two holes siik by side in the cylinder C. The photograph is possibly to be regarded as showing much distorted projections of the uprights and , by particles of metal shot from the heated strip , through the holes in the cylinder .
But this is probably not a complete explanation of the photograph .
The pressure during this run was about 2 mm. , not too low to prevent convection See , in this connection , Reboul and de Bollemont , 'Journ .
de Phys July , 1912 .
1913 .
] Ernissivity and Disintegration of Hot currents controlling the sublimed iron vapour , which , as the deposit all over the bell-jar shows , was everywhere present .
In this experiment , the iron FIG. \mdash ; Iron Deposit .
strip was heated for some minutes and the brass uprights , and iron strips , got appreciably warm\mdash ; warm enough doubtless to cause an upward current of gas to oppose the descending iron vapour , and to cause it to deposit on the glass adjoining .
The successive bands of deposit at each of the crests are certainly interesting and suggestive of a kind of spectrum which might be produced by a magnetic or electric field acting on charged particles of metal with velocities grouped around certain values .
But the only possible source of such a field is the alternating heating current , and even taking the maximum values of the current and potential the strength of the magnetic or electric field which could be produced is entirely inadequate to deflect and disperse charged particles whose masses are as large as we FIG. Iridium Deposit .
believe them to be .
We are not in a position to explain the various pec , uliarities the photograph displays ; fig. 13 is a corresponding photograph in the case of iridium .
538 Emissivity and Disintegration of Hot Metats .
Summary .
Preliminary experiments have been carried out on the volatiIisation and electrical emissivity of a number of metals , mostly in nitrogen at reduced pressures .
The metals were heated by alternating current and no appIied ] potential was employed .
( 1 ) The emission of positive electricity occurs at temperatures from about : ; 1000o to 1400o C. For metals which melt within this range , a sudden and marked increase in the positive current often occurred at the liquefying point\mdash ; due probably to the sudden release of occluded gas .
( 2 ) Oxygen appears to augment the positive current .
( 3 ) At higher temperatures , negative electricity predominates and increases rapidly with the temperature .
The negative current.attained with iridium at the melting point was 80 milliamperes , with 1670o C. 220 microamperes , with iron at the melting point 90 micronmperes .
In the case of carbon in air at atmospheric pressure , an current of amperes was obtained .
( 4 ) The negative current at moderate presures appears to be increased if the conditions are such that considerable sputtering of the metal occurs .
( 5 ) The negative currents are probably a consequence of chemical reaction between the metal and the surrounding gas .
( 6 ) Carbon becomes plastic* in the neighbourhood of 2500o C. At such temperatures it also readily sublimes .
We should like to thank Mr. H. C. Booth , of the National Physical Laboratory , for his sketch of the apparatus ( fig. 1 ) .

