]\gt ; [ 243 ] VII .
On ofHot HAROLD A. WILSON , , B.A. , of C. .
R. WILSON , F. R. S. 1{eceived June 3 , \mdash ; Read June 18 , 1903 .
THE following } ) aper contains account of the results in of experiments on the loss of icity f hot in several various pressures .
I shall by giving a brief account of ) of the results by previous WOl.kers on this subject , co1lfinin myself to those esults w to bear or less directly on the investigation .
Further mation o11 the results of the stigations may be ' vol. , in J. J. THOMSON'S ches , ' in STARK'S ) in ' and in the ) to .
ELSTER and Wied .
found latinum , in air and ) sitive eification of the surrounding bodies immel.sed in it , except at low when ative electrific tion is obtained .
In hydrogen they found the electrification is always negative .
J. J. THOMSON Phil. , ] 890 ) measul.ed the ent due to a constant P.D. between two atinum 1 ) lates i ] in various gases and vapours at a bright red heat .
He that the tent d ) ends very the nature of the ases which dissociate into JIlS when such as HC1 and salt , conduct very lluch better gases such .
and ammonia , which do not dissociate into atoms .
Unless both electrodes are hot ellough to glow , little current can be obtained .
Wied .
Anll .
measured the curl.ent between parallel platinum plates in , by which they were kept bright red heat .
He that the with P.1 ) .
eases first proportionately to the P.D. , but wheI the P.1 ) .
is more than one or , the curve connecting the CUl.rent and P.D. bends over , and then becomes lleally ) al.allel to the axis of P.D. The is eatly ieased when an alkali salt is introduced into the flame .
He investigated the elative conductivities due to ious salts , but it is not necessary to desol.ibe here the esults he obtained , since they not bear directly on the present investigation .
The results obtained by were extended in by ( 352 .
) Professol ' A. .
H. , and the writer ( ' Phil. Trans vol. 193 , pp. 89-128 ) .
Using as two concentric cylinders of platinum , it was found that the CUlTent was much greater when outer cylinder was made the kathode than when it was made the anode .
Also when the electrodes were simply co1mected to the galyanometer a current was observed the inner to the outer cylinder through the flame .
Phenomena of this kind have been known for a long tinle to occur when metal electrodes are immersed in a flame .
They are eyidently closely analogous to the effects described by LSTER and GEITEL , referred to above .
The discharge of electricity from platinum immersed in a Bunsen flame was further ated iter Trans .
In this paper it is shown that the ions in the discharge between hot platinum electrodes immersed in a flame nearly all at the surface of the platinum , and that the negative ions have a much greatel velocity , due to an electric field , than the positive ions .
Consequently , nearly all the current is carried by negative ions which start at the negative electrode .
These results enable the polar properties of the conductivity of flames to be very simply explained .
Hot platinum in a flame consequently discharges negative electricity very much more readily than positive electricity .
In the ' Phil. NIag .
' December , 1899 , J. J. THOMSON gave an account of some on the atio of the charge ( e ) to the mass ( m ) of the negative ions emitted by a hot carbon filament in a vacuum .
He found that it is impossible to obtain results precautions taken to ensure that the filament is in an atmosphere of pure hydrogen .
The cause of this is the evolved by the filalneut when it is heated .
He found that by heating the filament a long time , and repeatedly ) out the atus letting in pure hydrogen , regular esults could be obtained .
When this had been done , the filament only lost negative electricity , whel.easbefol.e it lost sometimes only positive and sometimes only negative , and sometimes both .
He found that was equal to about , which is nearly the same as for kathode rays , so that it is clear that the filament emits negatiyely corpuscles or electrons .
The importance of taking such precautions as those described by Professor THOMSON , in experiments on the leak from hot wires at low pressures , is very great .
Unless this is done , the gases evolved by the wire fill the apparatus , and the leak observed is very irregular in its behaviour .
It is probable that the gas evolved by wires is mainly composed of hydrogen , which , as will be shown below , has a very on the leak , even when present very small quantities .
A further investigation of the leak from hot platinum was published the writer in 1901 Phil. Trans .
In this paper it is shown that hot platinum in air at atmospheric pressure in general discharges positive more eadily than negative electricity , and that the rate of discharge at temperature falls off gradually with the time and ultimately beco1nes very small compared OF RICITY FROM HOT with its original value .
The current between platinull odes in with constant P.D. increases rapidly with rise of temperature .
If is the cul.rent ( due to a small P.D. ) at the absolute , and that the enel.gv required to produce amme nloleculal weight of and negative ions , then it is shown that ; and is found to be nearly independent of the we llay ) , where A. is a constant tity .
is de assumption that the air close to the surface of the platinum dissociates into The value found is 60,000 small gives volt } , for the P. I ) .
, through which the ionic must fall to obtain enougb enel.gy to ionize one molecule of air in contact with hot platinum .
[ When a large P. I ) .
is used , so that current is the maxinuum , the appropriate formula is , p. 25 ] It is also shown this paper that the maximum between hot electrodes immersed in a stream of an alkali salt is equal to the current to electrolyse in a solution the amount of salt between the electrodes .
It is therefore cleal ' that the leak is due to the ionization of the salt contact with platinum .
Professor , in 1902 , the results of some , on leak from hot platinum in ir at vely low ( ' Soc vol. , Part IV .
, p. 296 ) .
A wire ounded I ) electrode could be heated by passing a current ougI it .
The the wire to the e]ectrode was with various differences of potential ) veen them .
It was found that , at pressures below 1 millim. , the current is when the wire is negatively charged than when it is positively current with the wire positively charged is nearly independent of the P.D. between limits 40 and 200 volts , but above 200 volts it ises rapidly with the ) .
the wire is negatively charged , the current at millim. always ises 1 with the P.D. , but at millim. the rate of rise di1ninishes as the P.D. increases .
These results can be by that the iollS coming the ) others by colliding with the gas molecules when the electric intensity is This was by the ions in by Rontgen rays , it was shown by him that the tive ions produce ) ' collisions when the electric intensity is tively s. The intensity to make the positive ions produce others by collisions probably nearly required to start an ordinary discharge through the gas , but * No assumption need be made to the nature of the they moleeules , touls U corpuscles carrying the ionic charge .
unit .
See also " " The Laws of Electrolysis of Alkali Salt ( Phil. 'Phil .
Mag February , 1901 .
experiments the intensity near the fine ) this value eyen when P. D. is .
less .
LALLAN D states in this per that the negative leak the wire is nearly independent of .
pressure from 1Ylillim .
to millim. , and he suggests that the ative leak at low is caused by the escape of negative ions the metal . .
W. 1{1CHARDSON Proc. Camb .
Phil. Part ublished the results of series of measul.ements of the negative leak from hot platinum in gas at low pressures .
He that the yariation of the leak with the tenlpel.ature call by a mula of the type .
This formula the assumption that the platinum contains negative col.puscles which can uly get the surface layers when their velocity exceeds definite value .
The value he obtains for corresponds to a fall of the charge through volts , which does not differ Jnuch from the value volts fOund by the writer in .
at nospheric pressure .
ding to theory , is the equired to drive a ] weight of the corpuscles out of the metal ough tsurface layers .
It is shown below that the formula be deduced without thel .
assumption than that ions are produced at the surface of the metal , so that the agreement of the experimental results obtained with mula can1lot be regarded as evidence in favour of any particular eory with regard to the way in which the ions are produced .
RICHARDSON also showH on it is possible to de from the constant A in the above tl numbel of COl.pusc ] in a cubic centimetre of the platinum .
He obtains , which is of nearly the same of nitude as the values of deduced ) on the variation the electrical resistance of in a etic f RICHARDSON recently the Society on this Hubject , in he examined negative leak some othel lnetals besides in at low .
The results with platinum described in that ) the same as those just to .
In all these expel.iments the ) leak was very small lnpared with the negative leak .
W. Anllalen Physik , described some on the leak hot ) wires in high vacua .
He finds that even in good cuum a hot call still rge electricity to distant small if it is to a potential .
In best the P.D. lecessary was 800 volts for negative electricity and volts positive .
ding to C. 1 ) .
CHILD eview , the negative ] a hot ) ] atinum wire , using a , increases rapidly with dimin pressure one atmosphere down to small fraction of a millimetre of mercury .
It will be shown below that this increase is due to ionization by collisions with , 'Phil .
Mag June , 1902 .
See ' Phil. , vol. ) ) DISCHABGE OF TCTTY FROM HOT I. the gas molecules , and that below a cel.tain pressure the leak in air diminishes with the pressure unless the P.D. used is too small to produce ions by In the following paper the results are given of experiments on the variation of the leak from hot platinum wires with the temperature of the wire , the pressure of the gas , and the nature of the gas .
Some results also given on leak bof paLadium .
The rest of the paper is divided nto the sections : ( 1 .
) Description of apparatus , \amp ; c. ( 2 .
) The leak in air , nitrogen and water ( 3 .
) The yariation of the negative leak with the temperature .
( 4 .
) The leak in hydrogell .
( 5 .
) The leak from palladium in hydrogen .
( 6 .
) Summary of principal results .
( 7 .
) Conclusion .
( 1 .
) tion of The apparatus used is shown in fig. 1 .
leak was measured from a platinum wire PP to a thin cylinder of platinum AB surrounding it .
The cylinder was Fig. 1 .
PP , platinum wire ; , voltmeter ; , commutators ; , rheostat ; , { lnmetc ] ' ; AB , cylinder ; , key ; , resistances ; ometers ; centim. in diameter and 5 centims. long .
wire PP was in the of ging vertically downwards .
The cylinder was contained in a glass tube also made measurements in hydrogen and found the leaks greater in hydrogen than in itir .
He did not measure the temperature of his wire , so that his results purely qualitative .
DR. ROLD A. WILSON ON THE fitted loosely and was welded to two platinum wires sealed through the end of the tube at F. The loop was welded to two platinum wires and , which were sealed through the other end of the bube as shown .
These wires were each 1 millim. thick centims. long , and the current used to heat the loop entered through them .
cylinder was connected through a reyersing commutator to an Ayrton-Mather anometer G , and then through a liquid resistance of about ohms and another eversing commutator to one pole of a battery of small secondary cells , the other pole of which was connected to loop .
The P.D. between the loop and cylinder was measured by means of an electrostatic yoltmeter V. The resistance of the loop was Jneasured by means of a Wheatstone 's bridge arrangement .
The loop was connected in series with a resistance coil of thick silver wire spirals immersed in a jar of paraffin oil , the temperature of which was indicated by a normal thermometer .
The coil and the loop formed two arms of the bridge , and for the other two arms the coils of a P.O. box were used .
The coil had a resistance of about ohms , 1000 or 200 ohms was usually used for the adjacent arm in the box .
The fourth then had a value , usually several thousand ohms , depending on the resistance of the loop .
The current through the loop , which nearly all passed through the coil , was supplied by a battery of from 4 to 10 secondary cells , and was measured by means of a Weston ammeter .
Another Ayrton-Mather galvanometer was used with the bridge arrangement , and was connected to the wire and the opposite corner of the bridge .
A key was used with this galvanometer , which closed first the two arms of the bridge in the P.O. box and then the galvanometer circuit .
In this ) the current only passed through the coils in the box when the key was pressed down .
The resistance of these coils was so large compared with that of itnd the loop that no appreciable change in the current through the loop occurred when the key was pressed down .
A rheostat in the battery circuit enabled the current through the loop to be varied , and the method used was to give the resistance of the variable arm of the bridge the desired value , and then by altering the current through the loop by means of the rheostat balance the bridge .
At high temperatures the bridge was very sensitive and the resistance of the loop could easily be kept constant to one part in 5000 by adjusting the rheostat .
When it was desired to measure the resistance of the loop cold , a single Leclanche cell in series with a resistance of 10 ohms was substituted for the secondary cells , and the variable arm of the bridge adjusted till a balance was obtained .
This measurement could be made to one part in a thousand .
The temperature of the loop was got in this case by a thermometer placed close to the tube , and whenever possible this measurement was made with the apparatus filled with gas at atmospheri-c pressure , because at low pressures a very small is sufficient to heat the loop appreciably .
The current through the loop in this case , therefore , was only kept on for a few seconds at a time , in order to prevent the loop from being heated above surrounding temperature .
OF CTRICITY FROM HOT ATINUM .
24 !
) The resistance of C. was culated from the resistaIlce at the nperature of the room , usin the known value of the temperature coefficient at low temperatures of the wire used .
The loop and wires and were made } ) platinum supplied by Messrs. JOHNSON , MATTHEY and Co. The temperature of the loop was deduced from its resistance in the following manner .
temperature coefficient of the wire used was determined by measuring the resistance of a piece of the wire cold and at the melting-point of pure which , according to HEYCOCK and NEVILLE , is 1.066c C. The wire was hung horizontally , and screened from air currents , and a minute particle of put on it and obseryed through a microscope .
It was then heated by passing : current through it until the particle of just began to melt , and its resistance when this occurred was measured by the bridge arrangement just described .
This was done with a particle at several points along the wire , and the mean of the resistances when the particles melted , which were all very nearly same , taken to be the resistance at 1066o C. The following are the results of one experiment of this kind : Resistance in variable of bridge required to balance resistance of 18 centims. of wire , millim. in diameter , at C. , 1014 ohms .
Resistances required at C. : ( 1 ) 4098 , ( 2 ) 4100 , ( 3 ) 4102 ; Mean 4100 ohms .
Resistance at C. Hence mean temperature coefficient of the wire C. and 1066o C. is ohm per ohm at C. The " " platinum temperature\ldquo ; of the wire loop in any experiment was first calculate from its resistances at C. and at the required temperature by means of ) , where is the at C. , that the cquired temperature ( pt ) , the coefficient .
The " " platinum telnl)eratures\ldquo ; so determined were corrected to the centigrade scale by means of a table of corrections .
These corrections were obtained by determining the resistance of the wire at the melting-point of platinum , which was be 1820o C. This done by a wire loop in the apparatus described above until it just melted .
The wire heated in a yacuum and the current increased ( very dually when close to the melting-point ) , and the resistance measured until wire just melted .
This 1nethod gave sufficiently concordant results .
The ving are the numbers obtained in two experiments of this kind:\mdash ; ( 1 ) Balancing resistance at C. for wire millinl .
in diameter and about 13 centims. long , 1320 ohms .
Ditto whenjust on point of melting , 7615 ohms .
VOL. CCIf .
\mdash ; A. Hence ] ohms , on the scale used .
( 2 ) resistanct1 , at C. for wire millim. in diameter and about I3 centims. long , 694 ohlYlS .
Ditto melting , 4000 ohms .
Hence ohms , on the platinum scale used .
The mean of these two results , 1638o , was taken to be the melting-point of platinum on the atinum scale used : The correction to the centigrade scale at the melting- point of platinum is therefore The correction at other temperatures was calculated from this by means of a parabolic difference mula as used by CALLENDAB and GRIFFITHS , where temperature centigrade , temperature on platinum scale used , constant .
The constant A is given by equation , which gives .
The following table giyes ) values of the correction at several temperatures :\mdash ; A curve was plotted from these numbers giving the correction in terms of the temperature on the platinum scale .
The following table gives some values of the correction taken from this curve : DISCHARGE OF ELECTRICITY FROM HOT PLATINUM .
The temperatures so mined in this ) subject to an due to the ends of the wire being cooled by the thicker electrodes to which they are welded .
There is also an error due to the resistance of these electrodes being measured along with the wire and changing with the temperature .
These errors are , however , almost entirely eliminated by the method of standai.dising the wire adopted , for the wire was standardised under arly the same conditions as it was under in the experiments , consequently these erl'ors are reduced to almost zero at 1066o and at 1820o , and are therefore small at the termediate temperatures .
The numbers obtained consequently represent very the temperature of the wire at some distance from the electrodes .
There was no error due to thermo-electric effects .
It was found that after a wire had been just ffised , as descl.ibed above , whole length of the wire , except about 1 ntim .
each end , showed of fusion , although as soon as any part of the wire fused completely it broke , and so stopped the current .
The wires used were always about 13 centims. long , but in calculating the leak per sq .
centim. of platinum surface , their length was taken as 1 centim. less than their actual to allow for the cooling of the ends near the electrodes .
This amounts to counting the 1 centim. of wire at each end as leaking half as much per unit area as the rest of the wire .
The tube containing the platinum wire was sealed on to a tube which communicated with a Topler pump and McLeod gauge .
The connecting tubes between the different parts of the apparatus were made short to enable the gas to get into equilibrium as quickly as possible .
The methods used for preparing and purifying the different gases used will be described in the following sections .
( 2 .
) The Leak from Hot in Air , Nitrogen , and Water The results obtained with air will first be described .
The air was passed slowly through tubes containing soda-lime and calcium chloride and then let into the apparatus through a good mercury-sealed tap .
A bulb containing phosphorus pentoxide was sealed on to pparatus near the tube containing the hot wire .
The used a good sample , which was found to be quite free from lower itnd no luml ) of meta-phosphoric acid .
In some of the earlier tube packed tightly with gold leaf was put in between the hot wire tube and the rest of the apparatus to absorb mercury pour .
About 50 gold leaves were packed into a tube 1 centim. in diameter and 3 centims. long , and it was found that the gold near the end of the tube next the pump and gauge became amalgamated , whereas the rest of the gold remained for months perfectly bright .
I believe that such a tube entirely prevents the diffusion through it of mercury vapour , and that previous failures in using this method of absorption are due to too few gold leaves having been used and not sufficiently tightly packed .
Of course , such a tube does not allow very free communication between its ends , so that some little time has to be allowed to elapse after changing the pressure in the apparatus before making any measurements .
It was found that the leak was very little affected by the absorption of the mercury vapour , so that in the later experiments this gold-leaf tube was abandoned .
Some of the experiments in which the gold leaf was used will be described in the section on the leak in hydrogen .
In the earlier experiments very inconsistent results were often obtained .
This was to be due to gas , probably mostly hydrogen , given off by the wire when heated .
A very small quantity of has a very large effect on the leak , as will be shown in Section ( 4 ) .
When the platinum wire is welded to its electrodes it is heated in a blow-pipe flame , and probably absorbs some hydrogen , which is afterwards given out when the wire is heated in a .
A new wire was generally found to give very large leaks , which varied in an irregular manner with the time .
Longcontinued heating in a good yacuum nearly gets rid of these irregularities , but long continued heating also causes the wire to disintegrate , and its surface becomes roughened and covered with a network of deep cracks .
It was found that the irregular behaviour of the wire could be entirely got rid of without long continued heating by the simple process of boiling it in strong nitric acid , after weldin it to its electrodes and sealin them into the tube .
The tube was about half filled with pure nitric acid and boiled over a spirit lamp for a few minutes , the acid then poured out , and the process repeated two or three times .
The tube was then washed out with distilled water .
If the platinum loop had been bent by the boiling process , it was straightened by a clean platinum wire hook introduced through the side tube , and then the tube was sealed on to the pump and gauge , \amp ; c. The apparatus was then pumped out and filled up with air several times , and then on heating the wire the negative leak at once took up an almost steady value .
The positive leak was very small , and did not fall off with the time after the first few seconds .
The process of filling the apparatus with and pumping out several times with the wire kept hot all the time was repeated at frequent intervals to ensure that the apparatus should only contain pure dry air .
ELECTRICITY FROM HOT PLATINUM .
Fig. 2 shows the way in which the leak from the hot wire yaries with the P.D. at pressures of 766 , 211 , and 27 millims. of mercury , the being gatively c When the P.D. is small , the current is nearly independent of the P.D. , but with Fig. 2 .
larger P.D. the current rises rapidly with the P.D. , the rate of rise being greater the lower the pressure .
The temperature of the wire in this case was 1648o C. , and the wire was 13 centims. long and millim. in diameter .
Fig. 3 shows the variation of the ative leak with ] pressures of Fig. 3 .
, 0.18 , 0.29 , 0.49 , 0.81 , and 121 millims. of mercury .
At these pressures rise of current with the P.D. is more rapid the higher the pressure .
The teml ) el.ature of the wire in this case was 1483o C. Fig. 4 shows the variation of the negative ] with the , using constant P.D. of 250 and 40 volts .
With 40 volts the current is quite independent of the pressure , but with 250 volts it rises rapidly to a maximum value at nearly 3 millims. , and then ] with pressure .
The temperature in this case was 1483o C. Fig. 4 .
The esults on the variation of the CUlTent with the P.D. and pressure can ined if it is assumed\mdash ; ( 1 .
) That the number of gative ions coming from the sul.face of the platinum per second is independent of the air pressure and so remains constant during each series of measurements , and ( 2 .
) That the tive ions can produce ionization by collisions with the air moving with sufficiently great velocities .
It has been shown by Professor TOWNSEND* that negative ions produced in air and other gases at low pressures produce ionization by collisions with the molecules when moving under the influence of a sufficiently large electric intensity .
He finds that if is the number of negative ions produced by one negative ion in going 1 centim. , then where is the pressure , X is the electric nsity , and denotes a function whose values he has determined .
The following table gives the values of for air found experimentally by TOWNSLND ( ' Phil. Mag vol. 5 , 1903 , p. 395 ) for several values of 'Phil .
Mag February , 1901 , June , 1902 , April , 1903 , and ' Electrician , ' April 3 , 1903 .
TOWNSEND has determined a large number of values of with various values of and X , and he finds that they can be very accurately represented by an equation of this form .
DISCHARGE OF ELECTRICITY FROM HOT PLATINUM .
When is than about 200 , is given shows ) by the formula , where and yolts .
The yalues of are given in the third column of tbove table , and it can be seen that they agree with the determined values of .
X is in volts per centimetre and in millimetres of mercury .
Thus when is large , is neal .
lie equal to If is the P.D. between wire loop and cylinder , and the distance centre of the wire , then approximately near the wire , where l ) is radius of the cylinder and that of the wire .
In the experiments just described centim. and centim. , so that and Hence X at the surface of the wire is equal to 40 V. Let , denote the numbel of ative ions , a cylinder of radius coaxial with the hot wire , per second .
Then we have V ; also , and , so that Integrating , we ) } .
This equation should therefore be capable of representing the variations of the current with the pressure and P.D. at constant temperature , when the P.D. is enough to drag all the corpuscles which escape away from the wire .
The valuc of or the maximum number of negative ions produced by a in going 1 in air at 1 millim. pressure is equal , according to TOWNSEND'S results ( loc. cit to the number of collisions with air molecules made by the corpuscle .
should , therefore , be propoltional to the number of molecules present at 1 mill-im .
pressure , that is , it should be inversely proportional to the absolute temperature of the air .
The .
table contains values of taken from the curve at 27 millims. pressure and those calculated by the formula taking and , which values from the observed values of ) , at 150 and 300 volts .
If varies inversely as the absolute temperature , then , the value of at C. , at 1648o C. should be .
The number 3 obtained is rather this , as was to be expected , because the air near the wire must the : .
HAROLD A. WILSON ON THL wire itself It seelns therefore that the results obtained are consistent with the view that varies inversely as the absolute temperature .
: The dotted curve in was calculated by means of the expression taking and .
These values were obtained by making the nlaximum on the calculated curve coincide with the observed maximum .
Below 4 millims. pressul.e the observed and calculated curves agree fairly well , but at the higher ureS the calculated current is too large .
At high pressures most of the ionization by collisions takes place near the wire where the gas is very hot , so that the value of ought really to be taken smaller at the higher pressures , consequently the calculated currents are too large .
Fig. 5 shows the variation of the negative leak with the pressure using 40 yolts only , so that practically no ions are produced by collisions .
The current falls off as the pressure increases .
This is probably due to the presence of the gas hindering the escape of the negative corpuscles from the surface of the wire .
Some measurements of the negative leak were also made in and in watervapour .
The nitrogen was from and and passed over before admitting it into the apparatus .
The last traces of oxygen were removed from it by sodinm electrolysed into the apparatus } hot lass .
The details of this ocess will } ) described in Section ( 4 ) .
The leak in at pressures up to several millims. was nearly equal to that in a good vacuum from similarly treated wires , provided .
used was not large enough to produce ionization by collisions .
The tects with large P.D. were similar to those obtained in air .
The leak in water-vapour at temperatures below 150 C. with a pressure of about 15 millims. also very nearly equal to that in a good vacuum at the same temperature .
The tube the wire was connected to a three-way tap , which was also connected to a tube leading to the pump and to a small bulb containing water .
The tube and the bulb were exhausted separately and then connected together , several times , so that the tube containing the wire was filled with water .
The current obtained was 25 divisions at 1446o C. The corresponding current in a vacuum was 27 divisions , using a wire which had been treated in a similar way .
The positive leak in air and other gases is small at low pressures .
When a platinum wire has not beeu specially cleaned is first heated in a vacuum or in OF ELECTRICITy FROM HOT PLATINUM .
air at low pressures , it can discharge positive electricity to an extent appreciable on a galvanometer , but the current rapidly falls off and in a few seconds becomes inappreciable .
A wire which has been cleaned by boiling usually only shows this effect to a very slight extent or not at all .
In air at high pressures the permanent positive leak is , of course , appreciable on a galyanometer .
It increases with the pressure at constant temperature .
In pure air at atmospheric pressure , using a wire cleaned with in the way described , the positive leak is about double the negative leak when the temperature is about The two leaks are nearly equal at a pressure of about 200 millims. at the same temperature .
( 3 .
) The of the Negative Leak from Hot Platinum with the The results in the preceding section show that the number of negative ions or corpuscles which ] ) start from the platinum is independent of the air pressure , provided that this is not re than a few millims. , or that if the pressure is large the P.D. used is sufficient to drag away all the corpuscles which get out .
In the present section the variation of the number of these corpuscles which escape from the platinum with the temperature will be considered .
The wire used was centims. long and millim. in diameter , and it was well cleaned with after it had been mounted as described above .
The apparatus was filled with pure air and pumped down to a low pressure several times with the wire red hot .
When it was thought that all gases evolved by the wire got rid of , the apparatus was pumped down and the measurements made .
If these precautions were not taken , the leaks were always much larger and very irregular .
The following table contains the results obtained with a particular wire treated in this way .
Balancing resistances in P.O. box at C. before experiments 1000 and 1669 obms , and after the experiments 1000 and 1669 ohms .
millim. scale dvolts .
* In making a series of measurements , it is necessary to measure the resistance of the wire when cold from time to time , because at high temperatures the wire disintegrates and its resistance consequently rises with long continued heating . .
CCIT .
\mdash ; A. The leak is the same at millim. as at millim. , and it was verified carefully that no change occurred even on pumping down to millim. The numbers in brackets refer to the order in which the measurements were made .
The following gives the values of the leaks in amperes and the temperatures deduced from the above numbers:\mdash ; It will be observed that as the temperature increases by nearly equal increments the current increases by nearly equal factors .
A theoretical proof of the formula will now be given , and then it will be shown that the above results can be represented by this formula .
The emission of negative corpuscles by hot platinum is analogous to the evaporation of a liquid , and whether the corpuscles come out of the metal or are produced at its surface , the number produced per second per unit area of platinum surface may be regarded as analogous to the number of molecules emitted per second by unit area of a liquid .
If is the vapour pressure of a liquid at the absolute temperature , and its latent heat of evaporation per gramme molecular weight , then where volume of vapour and volume of liquid .
Neglecting and putting we get .
Let the internal work done in evaporating the liquid be and suppose , where is some constant , then , so that we have or Hence if is the vapour pressure at a temperature , and that at .
Now bmVN , where is the mass of a molecule , V the square root of the mean square of the velocities of all the molecules , and the number leaving each square centimetre of the liquid surface per second and a constant .
But is proportional DISCHARGE OF CTRICITY FROM HOT PLATINUM .
to , hence we may write , where is a constant .
Substituting this above equation we get In the case of hot platinum in an atmosphere of its own ions , if is the current unit area due to ions leaving the platinum surface , then where e is the chal'ge carried by one .
Hence we get , putting , which is its value in small calories for one gramme molecular weight of any gas , The following table gives the values of calculated by means of this equation from the observed currents and taking .
The numbers in brackets indicate which pair of experimental numbel .
S taken from the last table above was used .
The variations in are not greater than can be ascribed to experimental errors .
The value of obtained corresponds to a fall of the ionic charge through volts .
RIOHARDSON'S result for this quantity was volts , but since the leaks which he obtained are about 1000 times gl.eatel .
than those given above , it was to be expected that his value of would be smaller than that which my expel.iments lead to .
The value of the constant A in the formula , calculated from the value of the current per square centimetre at 1580o C. , is , so that the equation becomes The following table contains a comparison of the currents calculated by meaDS of this formula and those found experimentally:\mdash ; Current found .
Current calculated .
7 .
The calculated and observed currents agree very well , so that we may conclude that the formula is really ero or nearly so .
It is easy to show that the value of cannot be greater than unity in order for the observed and calculated results to agree as well as they do .
Consequently , since does not vary as much as 1 part in 100,000 per degree centigrade .
The value found for is slightly more than double that found previously ( ' Phil. Trans , vol. 197 , p. 430 ) for the work required to ionize a gramme molecular weight of air in contact with hot platinum .
All that it is necessary to assume to deduce the equation used to calculate either for the positive leak in air at atmosl ) heric pressure , or for the negative leak in a yacuum , is that ions are produced in some way at the surface of the hot platinum .
The resnlts just described cannot , therefore , be regarded as any additional evidence in favour of the view that the negatiye leak from hot platinum is due to negative corpuscles coming out of the metal ; they are , however , consistent with this view .
The positive leak in air at atmospheric pressure can be represented by the formula The value of may be obtained from vation that the positive leak at 1550o C. is double the gative leak .
This gives .
The temperature at which the two leaks with the particular wire used were equal is , therefore , given by by the formula 6 .
which gives or 1622o C. Above 1622o C. the negative leak with this wire was greater than the positive leak , and below this temperature less .
At low ) essures the positive leak is smaller .
If we assume that it varies as the pressure , then it will be giyen at any pressure by the formula At 1 millim. pressure this gives , and the temperature at which the negative leaks are equal comes out .
These results OF ECTRICITY FROM HOT PLATINUIsI .
afford an explanation of ELSTER and GEITEL'S observation that an insulated conductor near a hot wire becomes positively charged in air at atmosl ) heric pressure , but negatively charged at low pressures .
The results described aboye show that the negative leak from a platinum wire is very little affected by the presence of air , nitrogen , water vapour .
It was found that putting some mercury into the tube containing the hot wire , so that it was filled with mercury yapour , increased the leak considerably at vely high The whole tube , of course , gets hot , so that the pressure of the mercury vapour must have been very considerable in this experiment .
following ) gives the esults obtained : \mdash ; above 1600o C. the presence of mel .
CUl.y in large quantities appears to increase the negative leak , but below this it is without eflect .
The very small amount of mercury yapour ordinarily present can , thel.efol.e , safely be assumed to have no effect below 1600o C. With different pieces of pure platinum wire cut from the same sample and treated in the same way with nitric acid ] concordant results were obtained .
Nevertheless , long continued heating of the wire diminishes the negative leak , provided the air is frequently changed .
If the wire is simply left nding in .
at a low pressure for a long time the leak is often greater than before on heating the wire .
The of cleaning the wire with nitric acid and frequently changing the air in the apparatus diminishes the negative leak something like a thousand times usually .
Thus the leaks given above are about one thousand times smaller than those given by RICHARDSON at the same temperatures .
These facts naturally suggest the question whethel more thorough of the wire and purification of the air would not still further diminish ative leak .
To test this a wire was mounted as before and boiled for about houl .
It was then left in the for 24 hourH and a passed the wire to the cylinder through the , so that the ions were set free on the wire .
The tube was then washed out and the leak measured in pure air at a low pressure , as before .
The following table gives the results obtained:\mdash ; DR. HAfiOLD A WILSON ON THE 1 the treatment of the wire with reduced the negative leak to about of that previously obtained , but the rate of variation of the leak with the emained almost unchanged .
These numbers give and If , then , RICHARDSON'S theory , the number of corpuscles ( n ) in a cubic centimetre of the platinum would be .
PATTERSON ( loc. cit. ) obtained by experiments on the variation of resistance of platinum in a magnetic field .
It appears , , that cannot really be deduced fxom the constant A. On one occasion some phosphorus pentoxide accidentally got into the tube containing the hot wire , and the negative leak was increased enormously by its presence .
The current at 1490o C. was about ampere , or 1 ampere per square centimetre .
The negative leak was therefore increased about one million times by the presence of the phosphorus pentoxide .
Now ordinary phosphorus pentoxide usually contains phosphorus acid and impurities , and may giye off some vapours when kept in a vacuum .
If ally such vapours were present in the tube containing the hot wire , this result shows that they would produce a leak of negative electricity from it .
It is clear , therefore , that minute aces of val.ious substances such as hydrogen , phosphorus pentoxide , and alkali salts produce a large increase in the negative leak from hot platinum , and that the more thorough the precautions taken to remove traces of such substances the smaller the leak becomes .
I have not succeeded in reducin the leak to less in 250,000 of that observed by RICHARDSON , but I have little doubt that , if all traces of impurities could be got of , the negative leak in air would be educed to a still smaller value .
There is considerable evidence in favour of the view that the negative leak in air is due to the presence of traces of hydrogen in the platinum .
It will be shown in the next section that the negative leak in hydrogen is enormous compared with that in air , and that at low pressures it is proportional to the pressure of the hydrogen .
When the pressure of the hydrogen is very small , the value of in the formula is nearly equal to its yalue in air , which makes it very probable that the small negative leak in air or a yacuum is due to the of traces of This view wns suggested long by BERLINhR , lVied .
Ann vol. 33 , p. 289 DISCHARGE OF ELECTRICITy FROM HOT PLATINUM .
hydrogen .
It is well known that it is almost impossible to entirely get rid of the last traces of hydrogen from a yacuum tube .
This view is confirmed by the effect of nitric acid on the negative leak , for nitric acid oxidises hydrogen .
It is also in agreement with the fact that the negative leak in air at low pressures falls off during long-continued heatin of the wire .
The rate at which the leak falls off diminishes as the leak gets smaller , so that with wires well cleaned with nitric acid the rate of fall is not very great .
Platinum wires not specially cleaned evolve gas when heated in a vacuum , and since hydrogen is the only gas which platinum is known to occlude in any quantity , this gas is no doubt .
It is practically impossible to get rid of this evolution of gas merely by long continued heatin and pumping out .
Thus WIEN ( loc. cit. ) states that his wires continued to evolve gas after veeks of heating in an extremely good vacuum .
My wires , which had been treated with nitric acid and heated in air which was repeatedly changed , did not evolve anything like as much gas as those not so treated .
( 4 .
) The in Hydrogen .
In the experiments in at low pressures the hydrogen was obtained sufficiently pure by allowing it to diffuse through platinum .
A spiral of fine platinum tubing , millim. external and millim. internal diameter , was sealed on the apparatus at one end and the other end fused up .
On heatin this spiral in the flame of a spirit lamp , the hydrogen present in the flame diffused through into the apparatus .
In this way the pressure could be increased several tenths of a millimetre in one minute when the total capacity of the apparatus was about 300 cub. centims. The spectrum of a discharge in hydrogen let in in this way was examined , and appeared to consist almost entirely of the hydrogen lines .
It was thought that faint traces of the CO spectrum could sometimes be seen , but it was not certain that any CO entered through the platinum spiral .
For experiments at pressures above 10 millims. the hydrogen was obtained from pure zinc and hydrochloric acid , and passed into the apparatus through tubes containing soda-lime .
In some experiments the hydrogen was further purified by means of sodium let into the apparatus by electrolysis through hot glass , in the manner described by E. WARBURG .
This method was found to work very well , and a deposit of metallic sodium could easily be obtained .
The sodium , of course , absorbed some ydrogen as well as the oxygen and other impurities possibly present .
The apparatus for this experiment is shown in fig. 6 .
A is the ) containing the hot wire , a ) tightly packed with gold leaf , and the sodium tube .
This latter consisted of a thin glass tube about 15 centims. long and centim. in diameter , with an aluminium wire electrode along its axis , as shown .
It was immersed * 'Wied .
Ann vol. 40 , ( 1890 ) .
sodium amalgam contained in a test-tube .
The test-tube was supported on asbestos wool in a thick tube DE , the lower end of which was closed by iron wire gauze .
The gauze was heated by a small Bunsen flame and the temperature of the amalgam kept at about C. At pressures above about 1 millim. a current from about 400 cells could then be passed from the amalgam to the aluminium electrode , and sodium deposited , forming a metallic mirror in the upper parts of the tube C. The Fig. discharge in was a brilliant yellow colour , giving the line .
If oxygen was present , sodium oxide was formed , and the brilliant yellow light did not appear so strongly at first .
This process did not appear to have any particular effect on the amount of leak obtained from the hot wire in hydrogen at any pressure , so that we may conclude that the increase of the leak due to letting in hydrogen is not due to the presence of impurities in the gas .
DISCHARGE OF ELECTRICITy FROM HOT PLATINUM .
The following table gives some numbers showing the effect on the negative leak at a constant temperature of about 1350o C. , due to letting in hydrogen:\mdash ; of mercury .
Pressure iillims .
volts .
This increase of the leak with the pressure cannot be ascribed to ionization by collisions , for at the highest pressure used , one negative ion can only produce fresh negative ion in going 1 centim. The leak in air at C. from a clean wire would have given about division in the .
above experiment .
Thus the leak in hydrogen at millim. in this experiment was more than 25,000 times the leak from a very well cleaned wire at the same temperature in air .
In the experiment just described , the hydrogen was let in at intervals through the platinum spiral , so that the pressure increased during the experiment .
Another way is to let in hydrogen to the highest pressure desired and then pump out during the experiment .
It was found that the results were not the same in the two cases .
On pumping down , the leak falls , but not so quickly as it rises on letting in the gas .
It was soon discovered that the leak is not always a single-valued function of the pressure , but depends also on the previous treatment of the wire .
The results can be explained on the hypotheses\mdash ; ( 1 .
) That the wire absorbs an amount of hydrogen depending on the pressure and temperature , and that these are changed it takes a considerable time for equilibrium between the gas in the wire and that outside to be established ; and ( 2 .
) the leak at any temperature depends on the amount of hydrogen in the , and not directly on the pressure outside .
Consequently , if the equilibrium is established at a particular temperature and pressure , then if the temperature is raised or the pressure lowered the leak should not be steady at first under the new conditions , but should gradually fall off , in consequence of gas leaving the wire , until equilibrium is again established .
On the other hand , if the temperature is lowered or the pressure increased , the leak should gradually rise until equilibrium is again established .
Experiments showed that these effects occurred exactly in the way expected .
The following tables giye the results of some measurements of the negative leak in hydrogen , showing the effects of suddenly changing the temperature or pressure:\mdash ; VOL. CCII.\mdash ; A. 2 The above numbers show the effect of lowering the temperature at constant pressure .
The current drops and then gradually rises , ultimately becoming steady when equilibrium has been established .
Of course it is necessary to be sure that equilibrium has been reached at the higher temperature before changing to the lower temperature , in order to observe this effect , and to get equilibrium at the higher temperature may take a long time in some cases .
These numbers show the effect of suddenly raising the temperature at constant pressure in hydrogen .
The above numbers show the effect of lowering the pressure at constant temperature .
DISCHARGE OF LECTRICITY FROM HOT The aboye numbers ) the effect of raising the pressure at constant temperature .
The effect of raising the temperature of the wire is most marked when a wire which has been standing at the ordinary temperature in hydrogen is suddenly heated to a high temperature .
In this case an enormous current is obtained for a short time .
Thus a wire at a certain temperature in hydrogen at millim. pressure gave a current of ampere when equilibrium had been established .
It was then allowed to stand cold for 20 hours when , on raising the temperature to the same value , the leak was at first more than ampere , and only fell to its original value after long continued heating .
Some experiments were tried in which the temperature of the wire was raised by equal steps through a series of values and then lowered again through the same series of values , several times .
The following are the results obtained in an experiment of this kind at a pressure of millim. The numbers in brackets are the times in minutes at which the currents were measured .
It will be noticed that when the tenperature just been raised the on waiting , whereas when it has just been lowered the leak rises on waiting .
In spite of this , however , the numbers at each temperature keep fairly concordant towards the end of the experiment , which shows that they must approximately represent the variation of the leak with the temperature for a constant state of the wire with regard to the hydrogen .
This means that the time between the observations was so short that very little change in the amount of hydrogen in the wire occurred during the latter measurements , in spite of the changes made in the temperature .
This may be due to the amount of hydrogen absorbed by the wire varying very slowly with the temperature at the temperatures used in this experiment .
It is stated in Roscoh and SCHORLEMMER'S 'Chemistry ' that platinum at a red-heat absorbs times its volume of hydrogen , and at 10 C. only ) times its volume .
If the volume absorbed were proportional to the absolute temperature , then platinum in equilibrium with hydrogen at any particular temperature would be in equilibrium at any other temperature .
According to ROSCOE and SCHORLEMMER , the volume absorbed increases more rapidly than the absolute temperature up to a red heat , but on the other hand platinum certainly loses some of its hydrogen when heated to near its melting-point ; consequently there must be a region of temperature between a red heat and the melting-point where the actual amount of hydrogen absorbed is nearly independent of the temperature at constant pressure .
The very small variations with time in the latter of the above experiments seem to show that the range of temperature used in this experiment lies in this region .
The latter numbers in the above experiment have been reduced to amperes per square centim. The results are given in the following table .
The pressure in this experiment was millim. The formula } , when these numbers are substituted in it , gives .
A similar series of experiments done at a pressure of millim. of hydrogen gave the following results:\mdash ; OF ELECTBICIT FROM HOT Temperature .
ture .
ture .
ture .
ture .
ture .
ture .
ture .
ture .
ture .
ture .
ture .
ture .
Current per square centimetrc .
These numbers give The variation of the negative leak with the temperature , using 40 volts , was also measured at a pressure of 133 millims. The following table gives the results:\mdash ; Current per centimetre .
Temperature .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc .
Current per centimetre .
Temporaturc . .
These numbers give .
The leak in air or in a vacuum , using a well cleaned with , at C. is only about ampere per squal.e centimetre .
The following table gives the values of found at the different pressures , and also the approximate values of the constant A in the formula Thus the presence of hydrogen diminishes the amount of energy required in the production of the negative ions and consequently increases the leak .
It seems very probable from the above results that the action of the hydrogen depends on its absorption by the platinum , for the leak obtained depends apparently on the amount of gas absorbed by the wire and only indirectly on the gas pressure in the apparatus .
The leak , when there is equilibrium between the wire and gas , is nearJy proportional to the pressure at low pressures , but increases less quickly than the sure at higher pressures to a few lllillinletres .
The leaks the two sets of obseryations aboye at millim. , and at millim. , are nearly in the ratio of the pressures .
If a ] arge P.D. is used ionization by collisions comes in at highel pressures just as with air , and the current rises with the pressure , at pressures of a few millimetres , more quickly with large P.D. than with small .
The following table shows the variation of the leak at a constant temperature of 1340o C. with the pressure up to 760 millims. , using a P.D. of 40 volts , which is too small to produce ionization by collisions :\mdash ; Pressure .
Current .
At high pressures the leak is again nearly proportional to the pressure .
Fig. 7 shows the variation of the negative leak in hydrogen with ] .
at pressul .
CS of 133 and 766 millims. The current at millims. is nearly saturated I00 300 400 Fig. 7 .
with about 200 volts , but then begins to rise rapidly owing to ionization by collisions coming in .
The upper part of the curve can be represented approximately by the formula obtained in the section on the leak in air .
The positive leak , like the negative leak , is much increased by the presence of hydrogen .
With clean wires it is , neyertheless , inappreciable on a galvanometer , DISCHARGE OF ELECTRICITy FROM HOT PLATINUM .
except at pressures of several millimetres .
The followin values of the positive leak were obtained at 1300o C. , usinvolts : The following table gives the values of the negative leak at a temperature of 1400o C. in hydrogen at several pressures:\mdash ; Pressure .
Pressure .
Pressure .
Current per square centimetre .
The number is the leak from a very well cleaned wire in air at a low pressure .
The leak obtained by Mr. RICHARDSON at C. was ampere per square centimetre , with a gas pressure of about millim , ( 5 .
) The Leak from in Hydrogen .
Some measurements were made of the leak from a palladium wire millim. in diameter and 13 centims. long .
The wire was left about 20 hours in hydrogen at a few millimetres pressure ; after mounting it and before it was heated the pressure was reduced to millim. The balancing resistance ( R ) of the wire was S20 ohms at C. On heating to , the negative leak with 200 volts was scale divisions ( 1 division ampere ) and the positive leak 750 divisions .
After five minutes the leaks fell to and 130 divisions respectively , while the pressure .
rose to millim. The positive leak continued to fall with the time , and ultimately , after several hours ' heating , became only 6 divisions with ohms .
It was Fig. found that the positive leak was quite independent of the pressure of the , while the negative leak increased neal .
lie proportionally to the pressure at small pressures .
Fig. 8 shows the variation of the leaks ( P.D. volts ) with the pressure , the resistance being 1400 ohms .
The palladium was found to " " splutter\ldquo ; very much when hot , so that its resistance rose rapidly .
The very large value of the positive leak from it when first heated is no doubt analogous to the positive leak from platinum not specially cleaned when it is first heated .
( 6 .
) of Principal Results .
( 1 .
) The negative leak from a hot platinum wire in air , nitrogen , or water vapour is independent of the gas pressure , unless ionization by collisions occurs , and is the same as the negative leak from the same wire in a good vacuum .
( 2 .
) The variation of the negative leak in air with the P.D. and pressure at constant temperature can be explained on the hypothesis that the negative ions produce !
onization by collisions with the air molecules .
( 3 .
) The number of collisions made by a negative ion probably varies inversely as absolute temperature at constant pressure .
( 4 .
) The variation of the negative leak per square centimetre with the temperature can be expressed by the equation , where is the current in amperes , the absolute rature , and and A are constants .
For a wire very well cleaned with is 155,000 and A is about ( 5 .
) The negative leak in hydrogen is much larger than that in air or in a vacuum .
At low pressures it is nearly proportional to the pressure of the hydrogen .
( 6 .
) The presence of the hydrogen changes the value of the constants and , the following being the values obtained at different pressures:\mdash ; \mdash ; Pressure .
Pressure .
Pressure .
Pressure .
Pressure .
Pressure .
Q. A. 155,000* 120,000 85,900 36,000 155,000* 120,000 85,900 36,000 155,000* 120,000 85,900 36,000 155,000* 120,000 85,900 36,000 155,000* 120,000 85,900 36,000 155,000* 120,000 85,900 36,000 155,000* 120,000 85,900 36,000 155,000* 120,000 85,900 36,000 155,000* 120,000 85,900 36,000 155,000* 120,000 85,900 36,000 155,000* 120,000 85,900 36,000 155,000* 120,000 85,900 36,000 0 0013 0 112 0 0013 0 112 0 0013 0 112 0 0013 0 112 0 0013 0 112 0 0013 0 112 0 0013 0 112 0 0013 0 112 0 0013 0 112 0 0013 0 112 0 0013 0 112 0 0013 0 112 155,000* 120,000 85,900 36,000 ( 7 .
) It is extremely probable that the negative leak in air or in a vacuum is due to traces of hydrogen or other substances in the wire .
The leak can be reduced to part of its ordinary value by taking precautions to remove such traces .
( 8 .
) On first heating a positively charged palladium wire in a vacuum it can discharge positive electricity .
The rate of discharge falls off rapidly with time and ultimately becomes inappreciable .
A similar effect sometimes occurs with platinum , but the amount of the leak is very small and it disappears in a few seconds .
The Of the three values of obtained at low pressures ( viz. 155,000 , 131,000 and 120,000 ) the value 131,100 is probably nearest the truth .
The valus 155,000 depends on measurements of very small currents and is probably too large .
See Appendix .
] I HOT I. negative leak in ) with tinuln .
( 9 .
) In air and hydrogen , at pressures above several imetres , there is a permanent positive leak from hot platinum which increases with essure , This leak is larger in hydrogen than in air at the sallle ( 7 .
) The question of most COlll ) ection } the leak of icity f hot platinum is the method of production of the ions .
RICHARDSON loc. has proposed the theory that the negative leak is due to the escape of the corpuscles which , on the ionic theory of metallic conduction , metals contain .
These corpuscles are supposed to moye about freely inside the metal and to have a distribution of velocities the same as the molecules of an ordinary gas .
All corpuscles enterin the surface layers of the metal with a velocity component perpendicular to the surface greater than a definite value are supposed to escape ; and these assumptions a nula of the type can be deduced , as RICHARDSON has .
According to this theory the number of corpuscles in a timetre of metal can be calculated from the value of the constant , and , to this view the leak depends only on the state of the platinum and not the surrounding gas .
The experiments described in this paper show that the phenomena be explained completely by such a simple .
There seem , to be no doubt that the negative leak from hot platinum is due to the emission of negatively corpuscles or electrons , but the number of these emitted depends on the and other substances present at the surface of the platinum .
With clean platinuln in air at a low pressure there is , comparatively speaking , very little leak of electricity eithel positive or negative .
In air at high pressures there is a small leak of positive electricity , while in hydrogen there is a comparatively enormous negative leak .
In Mr. RICHARDSON'S experiments the gas present appears to have been that evolved by the hot etal , and so probably contained This is , no doubt , the of the comparatively very large currents which he obtained .
When occluded hydrogen molecules are present on the sul.face of platinum , lve may suppose that each molecule or atom of has a corpuscle associated it , that these corpuscles ] ] an energy distribution similar to distribution of the molecules of a gas .
If those corpuscles , having than a ce ] tain amount of energy , are able to escape from the surface , these suppositions account for the existence of a negative leak in hydrogen .
* In RICHARDSON'S experiments the gas pressure rose Consid 1helt the temperature of the as raised , showing that gases were ]Jeing evolved by the When a molecule of hydrogen has lost its corpusde it may be supposed to colnbine with one of the corpuscles present in the platinum , so that on this 1 , the hydrogen molecule serves as a sort of stcpping stone to enable the corpuscles to escape from the platinum .
[ Sec Appcndix .
] VOL. DR. HAROLD A. WILSON THE We may suppose also that the energy necessary to enable a corpuscle to escape from a hydrogen atom at the surface falls off as the concentration of the hydrogen atoms on the sul.face increases .
In this way the diminution of by the presence of can be explained .
According to this view the number of corpuscles in cnbic centimetre of the metal cannot be deduced from the constant A. It seems probable that the temporary ) itive leak from platinum and palladium , when they first heated , is due to the presence of a small amount of some decomposable compound or yolatile impurity .
This view is supported by the fact that platinum wires , well cleaned with nitric acid , do not give this effect to an extent appreciable on the galvanometer .
The positive leak in gases at high pressures is no doubt due to ionization of the gas molecules at the surface of the platinum .
In conclusion , I wish to say that my best thanks are due to Professor J. J. THOMSON for advice and kindly interest in these experiments while they were being carried out the Cay.endish Laboratory .
APPENDIX .
Added October 3 , SoIne interesting conclusions can be drawn from the way in which the negative leak from hot platinum or palladium in hydrogen varies with the pressure of the hydrogen .
We have seen that when the pressure of the hydrogen is less than about millim. the leak is nearly proportional to the pressure : Consider two temperatures and the leak at the temperature be represented by the equation , where is the current per square centim. , the pressure , and a constant .
Let the corresponding equation at be If these expressions for and are substituted in the equation ' we see that is independent of the pressure of the hydrogen .
Consequently , when the pressure of the hydrogen is small , the energy required for the liberation of a corpuscle ; independent of the pressure of the hydrogen .
Consider now the formula .
Since is independent of and it that A is pptional to .
It is clear , therefore , that A cannot be regarded as proportional to the number of corpuscles in a cubic metre of the platinum , because this latter number must be independent , or very nearly so , of A is evidently proportional to the number of corpuscles which are in a position to contribute towards the leak , so that we may conclude that this latter number is proportional to the amount of hydrogen present when this amount is small .
Thus we DISCHARGE OF CTRICITY 1 HOT PLATTNUM .
are led to the usion that it is only those corpuscles which are in some way associated with hydrogen molecules or atoms which are able to escape the innm , at the at which it is possible to make observations .
In a note to his papel on " " The Conductivity to a Vacuum by Hot Conductol .
S\ldquo ; Phil. Trans. .
EICHARDSON suggests that the effect of hydrogen on the negative leak which I observed can be explained on his theory by supposing that ] hydrogen diminishes while remains constant .
But , as we have just seen , the fact that the leak is proportional to shows that is constant and A proportional to .
When is greater than about millim. it is found that does diminish as increases , but A does llot emain constant .
Mr. RICHARDSON also suggests that my process of cleaning the platinum with nitric acid may deposit a layer of negatively charged oxygen on it which vould increase and so diminish the leak .
The leak can , however , be reduced without treating the wire with nitric acid .
If the wire is simply heated in a good vacuum the leak falls off with time ; and if the gases evolved by the wire are removed , the leak can be reduced in this way to a small a value as can be obtained by treating the wire with nitric acid .
The nitric acid process was adopted because it enabled the hydrogen to be got rid of without long continued which causes the wire to disintegrate .
[ Although it appears from the expel.iments that the negative leak obtained from hot platinum at low pressures is due to the presence of hydrogen or other substances , it is , of course , possible that with perfectly pure platinum at a sufficiently high temperature some negative leak would be obtained due to the emission of corpuscles by the platinum alone .
Mr. RICHARDSON'S results for carbon , which show that there is a large emission of corpuscles from hot carbon even at very low pressures , make it probable that there would be a similar emission in the case of pure platinum at sufficiently high temperatures .
The author recently made some experiments with carbon which , as far as they have gone , confirm .
RICHARDSON'FI esults for that substance .
] \mdash ;

