122 The Consumption of Carbon in the I. with Current and Arc-length .
upon the Luminous Radiation from the Arc. By W. Geoffrey Duffield , D.Sc .
, Professor of Physics , and Dean of the Faculty of Science in University College , Beading . .
( Communicated by Sir Ernest Rutherford , F.R.S. Received July 16 , 1915 .
) CONTENTS .
PAGE I. Variation with Current and Arc-length\#151 ; 1 .
Preliminary , and Method of Experimenting ... ... ... ... ... ... ... ... . .
122 2 .
Experimental Results .
Tables ... ... ... ... ... ... ... ... ... ... ... ... . .
123 3 .
Relation between Loss of Weight and Arc-length ... ... ... ... ... ... ... 124 4 .
Relation between Loss of Weight and Current-strength ... ... ... ... ... 127 i. Long Arcs ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... .
12 ?
ii .
Short Arcs ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... . .
129 5 .
Experiments with Heavy Currents ... ... ... ... ... ... ... ... ... ... ... . .
131 b\ General Conclusions from the Preceding Experiments ... ... ... ... ... . .
132 7 .
Theories of the Electric Arc ... ... ... ... ... ... ... ... ... ... ... ... . .
132 8 .
The Cathode Stream ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... .
133 9 .
The Bearing of the Present Research upon Existing Theories ... ... ... 134 10 .
The Mechanism of the Very Short Arc ... ... ... ... ... ... ... ... ... ... .
137 11 .
Previous Experiments ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... . .
139 II .
Influence upon the Luminous Radiation from the Arc ... ... ... ... ... ... . .
141 I. Variation with Current and Arc-length .
1 .
Preliminary , and Method of Experimenting .
Experiments have been carried out to determine the amount of material lost by the poles of a continuous current carbon arc under different conditions of current and arc-length .
The results indicate that when the arc-length is extremely small , the loss from the cathode of one carbon atom is accompanied by the transfer between the poles of a quantity of electricity equivalent to four electronic charges , a surprisingly simple conclusion when the complex changes of shape of the carbon poles are taken into account .
The experiments were conducted with poles which had been previously burnt to shape by running the arc for some minutes before beginning the weighings .
This prevision is of the utmost importance , and is responsible for the remarkable consistency of the observations .
Method 1.\#151 ; The carbons were held in special clips which could be removed as a whole from an ordinary vertical arc lamp and weighed with the ?
electrodes in position .
They could be replaced in the lamp exactly as before .
i The Consumption of Carbon in the Electric Arc. 123 Method 2.__The knife-edge and scale-pan were removed from one end of the beam of a balance , and a small brass tube fixed vertically to its upper surface to serve as a holder for the lower carbon of the arc .
The current was led to this by a wire which was soldered to the centre of the beam and which dipped into a mercury cup on a level with the central knife-edge .
The beam could be lifted clear of the mercury by the ordinary device for swinging the balance and weighings conducted in the usual way .
This arrangement obviated the necessity for touching the lower pole at all .
The upper pole could be raised vertically by means of a rack and pinion , so that after the arc had been burnt to shape , and the first weighing made , the positive and negative poles could be brought back to their original places and an exact fit obtained .
The current used was derived from a secondary battery ; suitable resistances enabled the current to be kept constant .
An enlarged image of the arc was formed by a lens upon a screen ruled with lines which represented millimetres at the corresponding magnification , and the arc length was maintained constant by watching this screen and adjusting the top electrode .
The first experiments dealt with arcs whose lengths varied from 0*5 to 15 mm. or more , and the current from 1 to 10 amperes .
The carbons employed were initially all of the same diameter ( 1 cm .
) , but , after having been burnt to shape , their diameters in the neighbourhood of the arc gap * were much smaller , the poles having so adjusted themselves that there was a characteristic contour for each value of the current-strength and arc-length .
The carbons were solid , and were freed from gross impurities by washing in acid and water ; for undertaking this work the writer is indebted to Mr. J. W. Dodgson , of the Chemical Laboratory .
The electrode under examination was the lower one , unless the contrary is specifically stated .
The experiments lasted for 5 minutes as a rule , but sometimes longer .
2 .
Experimental Residts .
In Tables I and II details of the experiments upon the anode and cathode are given .
Many of the weighings given in the following Tables were made for me by Mr. Southam , an Honours Student in the Physical Laboratory .
These results are plotted in Diagrams 1 and 2 , which show that consistent results have been obtained , except perhaps in the case of the 1-ampere arc , where the accuracy is less on account of the small quantity of material lost VOL. xcn.\#151 ; A. r Prof. W. G. Duffield .
and the difficulty in maintaining the arc , and where the small error introduced in striking the arc is likely to be appreciable . .
^pfoU.___The diagrams include individual observations ; the Tables , the average values for given arc-lengths .
] Tables I and II.\#151 ; Rate of Consumption per Coulomb .
I. Anode .
II .
Cathode .
Current ( amperes ) ... 2 .
4 .
8 .
10 .
1 .
2 .
4 .
6 .
8 .
10 .
Arc-length .
mm. 0-5 10 -o 10*4 7 T 4*8 4*5 41 1 *0 25 *3 17 *6 12*8 11 *0 20 -2 \#151 ; 9-2 6*5 5*5 5 *4 2*0 26 *6 18 T 1ST 12 *0 27 0 19 *2 10 -7 7*7 61 5*9 3*0 27 *6 19 *4 \#151 ; \#151 ; 35 T \#151 ; 13 -6 \#151 ; \#151 ; \#151 ; 4-0 29 *6 20 -0 13*8 \#151 ; 37 *4 21 *7 14 *5 10 *3 \#151 ; 7 '4 5-0 \#151 ; 21 *3 \#151 ; \#151 ; 37 *7 \#151 ; 15 *5 \#151 ; \#151 ; \#151 ; 6*0 34 *6 \#151 ; 15 *2 \#151 ; 36 *3 23 '9 15*7 11 *4 10 T 8*5 7*0 \#151 ; 21 -4 \#151 ; \#151 ; \#151 ; \#151 ; 15*8 \#151 ; \#151 ; \#151 ; 8*0 34 *5 22 *5 15 *5 \#151 ; 35 T 24 *6 15 -7 12 *0 9-7 8*5 10*0 \#151 ; 22 *2 16*6 \#151 ; \#151 ; 25 T 16 T 12 T \#151 ; 8 9 12*0 \#151 ; 22 *8 15 *9 \#151 ; \#151 ; \#151 ; 16 *0 12*3 \#151 ; \#151 ; 15 *0 \#151 ; 22 *7 16 -0 \#151 ; \#151 ; \#151 ; \#151 ; \#151 ; \#151 ; \#151 ; \#151 ; 10 -1 \#151 ; 17 *0 \#151 ; \#151 ; \#151 ; \#151 ; \#151 ; 25 *8 16*3 \#151 ; \#151 ; \#151 ; 18*0 \#151 ; \#151 ; 16 *4 \#151 ; \#151 ; \#151 ; \#151 ; \#151 ; \#151 ; \#151 ; 20 *0 \#151 ; 22 *9 The above figures , multiplied by 10 5 , express the loss of weight in grammes per coulomb of electricity .
3 .
Relation between the Loss of Weight and the Length of Arc. The loss per coulomb for a given current-density increases with increasing arc-length until a nearly constant value is reached at about 8 mm. This is true both for the anode and cathode , but there is a difference in the initial rates of increase in the two cases , the anode loss per coulomb being at first gradual and then increasing more rapidly as the poles are further separated , whereas the cathode slope is very steep at the outset and shows no point of inflexion .
The general shape of the curves is explained by the increasing oxidation of the hot poles as the arc is lengthened , a limit being reached when increasing the length does not augment the amount of air which has access to them .
As the existence of a crater would suggest , the loss from the anode is greater than the loss from the cathode under similar conditions of arc-length and current^the ratio of anode loss to cathode loss for currents of 2 , 4 , and 8 amperes being 1*36 , 1*40 , and 1*64 respectively when the arcs are long , a slightly increasing ratio .
The Consumption of Carbon the Electric Arc. 2 Amperes ANODE 8 Amperes 100 Amperes Diagram 1 .
ARC LENGTH 20 M MS The ordinates , multiplied by 10"5 , express the loss of weight in grammes per coulomb of electricity .
L 2 Prof. W. G. Duffield .
Loss per Coulomb X 10 grammes ^ o Amp CATHODE 2 Amps .
6 Amps .
8 Amps .
10 Am Diagram 2 .
3-1 x 10 grarpmes ARC LENGTH 20 The Consumption of Carbon in the Electric Arc. 4 .
Relation between Loss and Current-strength .
An important distinction exists between the behaviour of long and short arcs .
( i ) Long Arcs.\#151 ; From Diagrams 1 and 2 we see that the loss of material per coulomb for long arcs decreases with increasing current .
It was expected that with large currents the combustion would be greater on account of the larger amount of heat generated , but this is evidently not the case , probably because with small currents the time taken for the passage of a given CATHODE Long Arc Diagram 3 Short Arc CURRENT IN AMPERES Prof. W. G. Duffield .
quantity of electricity is longer , so that more extraneous combustion occurs .
It means that when the arc is long doubling the time causes more carbon to burn away than doubling the current .
This feature is further illustrated by the upper curve in Diagram 3 , in which the loss per coulomb is plotted Diagram 5 .
Diagram 4 .
Loss OF Weight XIO grammes Loss OF Weight XIO grammes CATHODE 100- 2mms .
Shortest Shortest ANODE Amperes against current-strength .
For long arcs the rate of loss at first decreases rapidly with ^increasing current , but subsequently more slowly until some minimum value is reached for heavy currents .
To examine the rate at which the total loss for each electrode increases The Consumption of Carbon in the Electric Arc. with the current , Diagrams 4 and 5 have been drawn from the data supplied in Tables I and II .
For the anode they show that for currents between 2 and 10 amperes the loss in an arc 10 mm. long is approximately a lineal-function of the current , but that the cathode curve differs in being slightly concave to the axis of current-strength .
Neither curve passes through the origin , indicating that for long arcs the mechanism of the transport of the electric current is only responsible for a fraction of the whole loss , the hulk of which is due to combustion or evaporation .
The behaviour of the 2 mm. arc is similar , but the absolute value of the loss is smaller .
( ii ) Short Arcs.\#151 ; The curves given in Diagram 1 for the anode did not suggest any regularity in the behaviour of very short arcs , but interest was aroused by the shape of the cathode curves ( Diagram 2 ) when the arc-length was small .
In this case all of them appeared to converge towards a point which was not the origin , but rather higher up , representing a loss of about 3 x 10-5 grm. per coulomb for all values of the current-strength .
The coincidence of this number with the value for the electrochemical equivalent of carbon ( 3-109 x 10-5 ) , on the assumption that this element is quadrivalent , appeared too striking to be accidental , so further experiments were carried out ; extremely short arc-lengths were employed , because under these conditions the amount of subsidiary burning is reduced to a small order of magnitude .
Method 'No .
1 ( p. 122 ) was adopted , in which the carbons were held in special clips in an ordinary arc lamp .
The actual length of the arc could not be measured , but it was rarely , if ever , more than one-tenth of a millimetre .
The results of this investigation have been recorded in Table III .
The observations were made by the writer . .
Table III.\#151 ; Very Short Arc-length .
Variation of Loss with Current .
Anode and Cathode .
Current .
Arc- Volts .
Total loss .
Duration of Loss per second .
Loss per coulomb .
length .
Anode .
Cathode .
experi- meant .
Anode.* Cathode .
Anode.* Cathode .
amperes .
2 -0 mm. 0-0 *2 19-32 grm. 0 *198 grm. 0*040 secs .
600 33 *0 x 10"5 6 *7 x 10"5 1 16 *5 x10-5 3 *34 x 10"5 4 0 0-0*2 23 0*284 0*073 540 52 *8 13 *5 13*2 3*38 5*75 0-0 *2 20 0*335 0*120 610 54 *9 19*6 9*5 3*42 8'0 0-0*1 25 0*300 0T05 420 71 *2 24 *9 8*9 3 *12 10 -o 0-0 *2 28 0*552 0*190 600 92 *0 31 *6 9*2 3 *16 * The anode was the upper pole in these experiments .
The anode losses are included in Diagram 1 , the cathode losses in Diagram 3 .
Prof. W. G. Duffield .
Details of the experiments are given in successive columns , and in the final column the loss per coulomb from the cathode is recorded .
It at once became evident from this investigation that each reading , as well as the mean value , 3'28 x 10~5 grm. , was in remarkable agreement with the figure already arrived at from Diagram 2 .
But as the curious forms assumed by the poles when the arc was small made it doubtful whether so regular a rate of loss could be real , the series of observations was repeated .
The second or balance method of conducting the experiments was adopted on this occasion and the observations were carefully carried out by Mr. A. H. Davis , B.Sc. , a Research Student in the Physics Laboratory , and the results are set forth in Table IV .
During this set of readings , after burning the electrodes to shape and starting with a very short arc , the observer watched the voltmeter and ammeter and adjusted the arc-length and resistance to keep their readings constant .
The method was not quite so satisfactory as in the previous series , in which attention was devoted to maintaining as small an arc as possible .
The readings are consequently slightly higher than before , but nevertheless the mean loss per coulomb , 3-47 x 10~5 grm. ( omitting the first observation , where the amount dealt with is very small ) , is in excellent agreement with that previously obtained .
Table IV.\#151 ; Cathode Loss for Very Short Arcs .
Amperes .
Yolts .
Time .
Loss .
Loss per coulomb .
Remarks .
( 1 *0 secs .
grm. grm. 30 300 0*005 1 *67 x 10~ ' ) 1 -o 30 300 0*009 3 *00 x 10~5 2*0 27 300 0*027 4*50 2*0 27 300 0*026 4*33 Carbons touched momentarily .
3*0 22 *5 300 0*024 2*67 3*0 22 *5 300 0*035 3*80 Mushroom growth burnt off , re-formed , and again burnt off .
4*0 20 300 0*044 3*67 4*0 20 300 0*047 3*92 Mushroom growth burnt off once .
5*0 20 300 0*046 3*07 7*0 20 205 0*039 2*72 9*0 20 300 0*092 3*30 Mushroom growth burnt off once .
10 *0 21 330 0*099 3*00 Mushroom growth burnt off once .
* Mean 3 -47 x 10-5 The Consumption of Carbon in the Electric Arc. Notes were made when the mushroom growth dropped off the tip of the lower ( negative ) pole , and it is obvious that part at least of the observed loss is due to this occurrence .
We shall see , in a subsequent paragraph , that the loss of these growths introduces an error which is of the order of , but rather less than , 10 per cent. If this correction be introduced , the above determination approximates still more closely to the theoretical value .
5 .
Experiments with Heamj Currents .
The largest current employed in the above experiments was 10 amperes\#151 ; it seemed desirable to extend the work to higher values in order to see if the same rate of loss held good .
An experiment was therefore undertaken with 100 amperes in the Electrochemical Laboratory of the University of Manchester , through the kindness of Sir Ernest Rutherford , and the writei is indebted to Dr. Pring for interrupting a series of his experiments to allow the use of the heavy current for this work .
The carbons employed were 27 mm. in diameter and about half a metre long , each weighing about 400 grm. A horizontal arc was used , the carbons being first burnt to shape under a pressure across their terminals of 33 volts and a current of 97 amperes ; initially the end of the positive pole was flat and the negative pointed , but after about 15 minutes running with a very short arc the cathode was found to have accumulated a mushroom growth of about the size of an almond , which exactly fitted into a cavity in the anode which had been eaten away .
* The poles were then removed , weighed , and re-inserted in their clamps in the same position as before , the negative growth being fitted into the positive crater , and nearly , but not quite , touching it .
The experiment proper was then begun , the voltmeter and ammeter being read every two minutes .
There was very little tendency for the arc , which was very noisy , to grow longer , but it was kept as constant and as small as possible .
The actual arc was hidden from view , but it was of the order of 0*1 mm. The experiment lasted 15 minutes , and the carbons were then removed , cooled and weighed .
The results of this experiment are shown in Table V and Diagram 3 , the rate of loss , 3*2 x 10"5 grm. per coulomb , agreeing so excellently with both the theoretical value and that previously obtained for smaller currents , that there can be little doubt that within a very close approximation and over a wide range of current , from 2 to 100 amperes , a mass equal to one electrochemical equivalent of carbon , on the supposition that that element is quadrivalent , is liberated from the cathode for each coulomb of electricity which passes between the poles .
' [ A further experiment was carried out with the same apparatus with a Prof. W. G. Duffield .
long arc which increased from 10 to 30 mm. in length during the running of the arc .
Details are given in Table Y below and the readings are plotted on Diagrams 1 and 3 .
] Table Y. Current .
Arc- Volts .
Total loss .
Duration of Loss per second .
Loss per coulomb .
length .
Anode .
Cathode .
experi- meant .
Anode .
Cathode .
Anode .
Cathode .
amperes .
mm. grm. grm. secs .
grm. grm. grm. grm. 100 o-o-i 34 6*980 2-880 900 774 x10"5 320 x 10"5 7 *7 x 10"5 3 -2 x 10~5 100 10-30 52-61 \#151 ; \#151 ; \#151 ; 1316 \#151 ; 13 -2 .
100 10-30 52-61 1020 10 -2 In these experiments the arc was horizontal .
6 .
General Conclusions from the Preceding Experiments .
We may express the results of the preceding investigation by the following general laws:\#151 ; 1 .
The loss of an atom of carbon from the cathode of a very short carbon arc is accompanied by the transfer between the poles of a quantity of electricity equivalent to four electronic charges .
2 .
In long arcs the loss is due to this essential carbon disappearance plus a quantity due to combustion or evaporation .
The first of these is based upon the agreement between the individual results of each set of readings comprised in the first part of this paper , between the mean results of each set , between the values for high and low current densities , between these values and the point to which the curves drawn with decreasing arc-length tend ( Diagram 2 ) and between all these observations and the theoretical value for the electrochemical equivalent of carbon .
In view of the complex structure of the short arc one would hesitate to accept the evidence if it were less overwhelming .
7 .
Theories of the Electric Arc. Before discussing the mechanism whereby the above law can be satisfied in the arc a brief r6sum6 of the existing theories is given :\#151 ; In 1890 Fleming* put forward the view that " the negative carbon is projecting off a torrent of negatively electrified carbon molecules , and these impinging against the positive carbon wear out a crater in it by a sandblast-like action .
Further he says " th$ electromotive force is thus able to * Fleming , 'Roy .
Soc. Proc. , ' vol. 47 , p. 123 ( 1890 ) .
The Consumption of Carbon in the Electric Arc. 133 keep up a projection of negatively charged carbon molecules from the end of the negative carbon , which molecules are loosened from the mass by heat , and then move away from the surface in virtue of the electric charge which they retain .
" Dealing with the potential difference necessary to maintain an arc , * Fleming suggested that a certain fraction of the ^working electromotive force might be employed in detaching carbon molecules from the mass of the poles .
Sir J. J. Thomson* put forward a different view : " The cathode is bombarded by positive ions , which maintains its temperature at such a high value that negative corpuscles come out of the cathode ; these , which carry by far the larger part of the arc discharge , bombard the anode and keep it at incandescence ; they ionise also either directly by collision or indirectly by heating the anode , the gas or vapour of the metal of which the anode is made producing in this way the supply of ions which keep the cathode hot .
It will be seen that the essential feature of the discharge is the hot cathode/ ' DuddelTsf research upon the electromotive forces within the electric arc opened the way for an interpretation by Pollock^ of its mechanism .
According to the latter the forward electromotive force at the surface of the cathode is due to the projection of electrons from that surface , and in a similar way the back electromotive force at the anode is due to the liberation of electrons from that pole .
The falls of potential near the electrodes are due to the distribution of ions within and at the end of the mean free path of the electrons projected from the poles , in which region ions are produced by collision with the molecules of the vapour .
8 .
The Cathode Stream .
The important part played by the high temperature of the cathode can be simply illustrated by the following experiment:\#151 ; Two carbon rods A and B form the positive and negative poles of an arc ; when the arc springs from the end of B to the rounded surface of A , B can be moved rapidly up and down the length of A without extinguishing the arc , whereas if the poles are reversed the arc at once goes out when the positive is moved .
The general effect is much as though the hot spot upon the negative pole acts as a nozzle through which a stream of negative electricity is discharged , the arc * Thomson , 'Conduction of Electricity through Gases , ' p. 613 .
t Duddell , 'Phil .
Trans.,5 A , vol. 203 , p. 305 ( 1904 ) .
+ Pollock , 'Phil .
Mag.,5 vol. 17 , p. 361 ( 1909 ) .
Prof. W. G. Duffield .
developing whenever this stream falls upon another conductor connected to the opposite terminal of the source of current supply .
The experiment may be modified by mounting one pole so that it can be rotated about its long axis , and by forming an arc between its rounded surface and the end of a 'second carbon rod .
When the latter is the cathode , the former can be rapidly rotated without affecting the arc , but not when it is the anode unless the motion is extremely slow , wdien it is possible to wrap the arc completely round the circumference of the cathode , the hot spot upon it remaining fixed and turning with the pole .
9 .
The Bearing of the Present Research upon Existing Theories .
In connection with the theories already discussed we have now to take into account the additional fact provided by this research , namely , that associated with the passage of a definite quantity of electricity through the arc there is a definite consumption of carbon at the cathode .
The most obvious view to take , but one which is not free from objections , is that in the immediate neighbourhood of the cathode the entire current is carried by carbon atoms , each of which as it leaves the cathode takes with it four electrons derived from the source of current supply ( No. 2 in the following summary , p. 136 ) .
The electrons may be subsequently liberated by such means as the incidence of ultra-violet light , or interaction with other ions ; if the latter , it occurs at the end of the mean free path of the carbon atom , where presumably it comes into contact with the other vapour in the arc .
The electrons thus produced play the part assigned to them by Thomson 's theory .
It is , however , difficult to see how we are to escape from the conclusion that some of the molecules of air must penetrate to the surface of the negative pole ( probably conveying positive charges to it ) , and there interact with carbon atoms instead of waiting till they have proceeded some distance from it , and that it is at the pole surface that electrons are ejected by virtue of such collisions and also by virtue of thermionic* and photo-electric agency ( one does not know to what extent the incidence of ultra-violet light from the arc upon the poles is an essential feature of its maintenance ) .
I am inclined to regard the interaction of the gas molecules with the carbon as playing a very important role in the mechanism , because it is recognised that an arc can only with great difficulty be run in an atmosphere which does not readily form a chemical compound with the material of the pole .
Metallic arcs , for instance , burn with difficulty in an atmosphere of hydrogen unless the poles be of magnesium or zinc , or some metal which * Richardson , 'Phil .
Trans.,5 vol. 201 , p. 516 ( 1903 ) , etc. The Consumption of Carbon in the Electric Arc. 135 forms a hydride , and even then the nature of the are is modified , as we can tell by examining its spectrum , which , instead of being confined to metallic lines , is now rich in those of the surrounding gas .
As another instance Arons found it impossible to form an arc between silver poles in an atmosphere of nitrogen , and ascribed this to the absence of chemical combination between the silver and the nitrogen .
Without entering into the question whether thermionic emission requires the presence of a second element , there is considerable experimental evidence that the nature of the second element , if it be present , exerts an important influence upon the amount of ionic ejection , at any rate at atmospheric pressure .
Harker and Kaye , * for instance , dealing with the passage of ions from a hot carbon tube to a cooler brass one longitudinally disposed within it , found that the current flow was affected when hydrogen was substituted for nitrogen between the tubes .
I consequently regard the presence of the gas molecules as an essential feature of the phenomenon of the arc .
It does not appear probable that chemical compounds of a permanent character can be formed within the arc itself on account of its high temperature , but there is reason to believe that there is momentary interaction between the carbon atom at the pole and the visiting gas atoms ( probably , but not necessarily , oxygen ) which may render the electronic content of the former atom unstable ( especially if it possesses a surfeit derived from the current supply ) , and cause it to yield up some of its store .
' In the following I have enumerated categorically the alternative actions which comply with the first of the generalizations given in Section 6 .
Where it is suggested that a chemical compound is formed it does not require that it should be stable ; all that is proposed is that reactions of such a nature should occur as are involved in the formation of that compound ; it is immaterial to the theory , as far as the immediate neighbourhood of the cathode is concerned , whether the product splits up immediately afterwards or not .
The idea common to all is that each carbon atom before being detached from the cathode is furnished with four negative electrons by the current supply , but the process whereby the atom and electrons leave the pole is different in each case ; they may leave together or separately , attached or unattached:\#151 ; 1 .
The ejection of four electrons from each carbon atom under the influence of purely thermionic or photo-electric action , and the subsequent liberation of that atom , uncharged .
* Harker and Kaye , 'Roy .
Soc. Proc. , ' A , vol. 86 , p. 379 ( 1912 ) .
Prof. W. G. Duffield .
This is included in view of the largely accepted view that purely thermal emission occurs , but it is open to the serious objection that an atom , should be less easily detached after having been deprived of its surplus electrons .
2 .
The carbon atom evaporates from the hot pole , being loosened by heat agitation , and carries off the four electrons with it .
The velocity acquired by virtue of its temperature is augmented by the electric forces in the arc .
This has already been discussed .
It agrees with Fleming 's early idea .
The whole of the current from the pole face is borne by atomic masses .
The electrons may be ejected from the carbon atom subsequently , but as has already been stated there is no obvious reason for delaying such action until after it has left the pole .
As far as I am aware , carbon has not been found free with four charges .
J. J. Thomson records the existence of a carbon atom bearing one charge , a positive one , in a positive ray tube .
3 .
As in ( 2 ) the carbon atom evaporates from the hot pole , being loosened by heat agitation .
The surplus electrons are liberated at the same instant .
Photo-electric and thermionic action assist , and may or may not be essential .
The free carbon atom is uncharged and the whole of the current in the immediate vicinity of the poles is carried by electrons .
4 .
The carbon atom is detached from the cathode by interaction with the surrounding gases or vapours .
( a ) All four electrons are liberated at the same instant:\#151 ; Assuming that the interacting gas is oxygen , one oxygen atom , uncharged , unites with and detaches one carbon atom to form uncharged CO. As already stated , this may immediately become uncharged C and 0 .
The point is that the presence of the 0 atom is necessary to loosen the electrons .
The whole of the current is carried by the four electrons thus set free .
( b ) Two of the four electrons are liberated at the same instant :\#166 ; \#151 ; Assuming ' that the interacting gas is oxygen , one oxygen atom arrives with two positive charges , uncharged CO is formed which involves two of the electrons , the remaining two are liberated .
The current is partly carried by oxygen and partly by electrons moving in the opposite direction , the total being equivalent to the transfer of four electronic charges .
( c ) No electrons are liberated:\#151 ; Assuming that oxygen is the interacting gas , two oxygen atoms each with two positive charges arrive at the cathode , uncharged C02 is formed which involves all four electrons .
The Consumption of Carbon in the Electric Arc. 137 The current is wholly carried by the two oxygen atoms .
This is the condition which determines true electrolytic action on the lines of the conduction by liquid electrolytes .
The disappearance of carbon at the cathode is thus analogous to the disappearance of copper from the anode in a Cu H2SO4 Cu electiol^tic cell .
... .
It labours under the disadvantage that electronic projection , which on other grounds seems such a vital feature of the action , is the result of a secondary and not of a primary reaction .
The formation of C02 in the arc is even less favoured by chemists than the formation of CO , but it is to be emphasized that it is the nature of the interaction rather than the stability of the compound which concerns us .
Since however , carbon can be removed by the advent of a single oxygen atom we may agree that 4 ( b ) is more probable than 4 ( c ) which requires the advent of two .
The point at issue really is whether electrons or charged atoms are projected from the poles of an arc as bearers of the current .
My experiments at first sight appear to point to the latter , and to agree with Fleming 's early view , but I have been at pains to point out how electronic emission is also compatible with the facts .
One reason why I am anxious to retain the electronic view is that the back E.M.F. found by Duddell at the carbon surface of the anode seems to require , negative rather than positive emission there .
In a subsequent paper I propose to discuss further experiments bearing upon this point .
10 .
The Mechanism of the Very Short Arc. The mechanism whereby the current is transported across a very short arc-l gap is not as simple as the law stated in Section 6 leads one to expect .
The contours of the poles have been examined by projecting an image of the arc upon a screen of transparent paper and tracing upon it a series of profiles at frequent intervals .
There is no characteristic profile for a short arc of given length and current density , but a characteristic cycle of changes , of which the outstanding feature is the growth of a mushroom form upon the I cathode tip ; while this increases in length a constriction develops behind it , and it eventually falls off , but another growth forms and the cycle is repeated .
Though the development of the growth suggests that the cathode is gaining in weight , on no occasion has this been observed , even during the preliminary burning to shape the cathodic consumption of carbon approximates closely to the rate of loss demanded by the above law .
Though the Prof. W. G. Duffield .
formation of the deposit is more obvious it is less important than the loss of a thin shell of carbon from a surface of considerable extent . .
Fig. 1 , a , b , c , d , shows the changing contours of a very short arc using 6-5 amperes .
In order to trace the top of the negative pole , the positive was momentarily raised .
The arc was fed by lowering the anode , the cathode remaining fixed .
In ( a ) the outlines of both anode and cathode are shown ; we note that the diameters of the crater and of the growth remain constant Fig 1 9 Fig. l.t Carbon 1 " deposited Carbon - ' consumed Fig. 1.9 for a given current density , and that the sides of the anode remain nearly parallel to their original directions .
In ( b ) only the cathode tracings are drawn , the intervals of time between A and B , and B and C , respectively , being 4 and 2 minutes .
The increase in the length of the cathode is well shown , and ' so is the development of the constriction , which grows continually narrower with time .
The material deposited at M during AB is consumed in the interval BC .
The observed The Consumption of Carbon in the Electric Arc. 139 cathode loss is the difference between the amount lost from the constriction and that which is piled up on the tip of the cathode ; ( c ) shows that the former is the larger quantity , so there is always a resultant loss of material .
That the weight of the growth is only a small fraction of the material lost was shown by breaking off three , which were found to weigh 0*004 , 0*002 , and 0*002 grm. respectively .
The largest of these is less than 10 per cent , of the total loss of weight from an arc of the same amperage in the experiments quoted , the others are approximately 5 per cent. When , on account of the narrowness of the constriction , the glowing carbon did not extend appreciably beyond the neck , the loss there was smaller than usual , but it was found that the mushroom itself was consumed .
See fig. 1 ( d ) , in which profile C is the profile C already shown in fig. 1 ( c ) , and D refers to an observation 2 minutes later .
When the necessary quantity of carbon cannot be derived from one part of the arc , it can be requisitioned from another ; in this way the law already quoted is satisfied .
The material of the deposit appears to be derived from the anode , both because the advancing cathode follows so closely the receding crater , and because if the carbon anode be replaced by an iron rod the carbon mushroom no longer develops , but a small quantity of iron is transferred across the arc gap .
It is surprising that the bulk of the consumption of carbon takes place at the constriction , because , though it is all glowing brightly there , the most dazzling part of the cathode , is the tip , and a loss occurring elsewhere would seem to have only a secondary connexion with the process occurring in the arc proper .
It is true , however , that the vapours of the arc embrace the cathode down to , if not below , the constriction ; but , on the other hand , if this plays an essential part , the length of the arc is indeterminate and is not of the order of 1/ 10 mm. but of 2 or 3 mm. The drop in the curves near the origin in Diagram 2 is due not so much to the reduction of the arc gap to its narrowest possible limits , as to the reduction in the amount of extraneous consumption occasioned by the whole of the charged vapour being now in a field of force sufficient to prevent its escape from the arc , and perhaps to the complete protection from oxidation of the tip of the cathode .
11 .
Previous Experiments .
In 1825 Silliman* found that the positive lost weight more rapidly than the negative , and observed " little waste of the points exactly where they are opposed but considerable on the laterally ignited portions of the charcoal .
" * Silliman , 'Silliman 's Journal , ' vol. 10 , p. 123 ( 1825 ) .
VOL. XCII.\#151 ; A. * Prof. W. G. DuffielcL He had previously noted* that if a metal were substituted for the positive carbon , the growth upon the cathode did not develop , and ascribed the deposition to a transfer of material from the anode to the cathode .
Matteucci in the ' Comptes Rendus , ' vol. 30 , p. 201 ( 1850 ) , concludes from experiments , which are not quoted in detail : ( 1 ) The loss is chiefly dependent upon elevation of temperature , other factors being constant .
( 2 ) For carbon and iron the positive pole loses more than the negative , the ratio of the losses varying according to the arc-length from 2 :1 to 5 :1 for carbon , the ratio being smaller for iron .
For zinc , copper , tin , lead , brass , and gold the negative loses more than the positive .
( 3 ) Under diminished pressure the quantity of matter that is lost is less .
( 4 ) The loss depends upon the disposition of the arc with regard to the magnetic meridian .
With regard to these conclusions , ( 1 ) appears to be negatived by Waidner and Burgess'sf experiments , who found that the temperature of the arc only increased by 2'3 per cent , when the current strength was doubled from 15 to 30 amperes , a very much smaller change than that which would have been found in the mass of the poles ; ( 2 ) is chiefly due to the fact that carbon forms a volatile oxide , whereas the products of combustion with zinc , copper , etc. , remain upon the poles , sometimes dissolving in the molten metal j ( 3 ) is in accord with the experiments upon the brightness of arcs under increased pressure , ] ; where the extraneous oxidation increased the intensity of the light .
The explanation of the fact recorded in ( 4 ) is probably that the shape of the outer layers of vapour is influenced by the magnetic field to a small extent , and more or less of the poles embraced by the hot vapours .
W. S. WeedonS has measured the losses from the poles of certain metallic arcs .
Using copper poles he found that , if the current was increased 2'5 times the cathode loss was five times as great , whence he concluded that Faraday 's law did not hold good for the arc .
Using water-cooled electrodes the cathode loss was 1/ 450 of the amount required by Faraday 's law .
A copper arc in hydrogen gave only 1/ 1500 of Faraday 's equivalent , When an iron arc was burnt in hydrogen and the poles water-cooled , the anode lost and the cathode gained .
The metallic arcs suffer from the disability already mentioned in connection with Matteucci 's experiments , No. 2 .
A carbon discharge in hydrogen , 3/ 16-inch arc-length , gave a cathode * * * S * Silliman , 'Silliman 's Journal , ' vol. 6 , p. 342 ( 1823 ) .
( In this paper positive and negative poles should be interchanged .
) t Waidne^and Burgess , Bulletin No. 1 , Bureau of Standards , Washington .
] Duffield , 'Phil .
Trans. , ' vol. 208 , p. Ill ( 1908 ) ; vol. 211 , p. 33 ( 1911 ) .
S W. S. Weedon , 'Trans .
Electrochem .
Soc. , ' vol. 5 , p. 171 ( 1904 ) .
The Consumption of Carbon in the Electric Arc. loss which was about 1/ 40 of the rate of loss for the carbon arc in air for the same current and arc-length , or 1/ 6 of the loss if combustion be excluded .
We may doubt whether this can properly be described as an arc in hydrogen because the discharge was narrowed down to a fine line , which crossed from one pole to the other , and the voltage was abnormally high , 180 volts .
IL Influence of the Consumption of Carbon upon the Luminous Radiation from the Arc. Blondel and Ayrton* have investigated the mean spherical candle-power ( or its equivalent ) of arc lamps , and have arrived at the following conclusions :\#151 ; 1 .
With a constant current the light emitted by the arc increases with increasing arc-length until a maximum is reached ; in some cases there is a slight diminution of the candle-power upon further increasing the arc-length .
2 .
When the length of the arc is kept constant the candle-power increases with the current .
Since these are two of the main facts in the present research if " consumption of carbon " be substituted for " candle-power/ ' the question arises whether the brightness of the arc is not intimately related to the amount of carbon consumed .
A comparison of the curves relating total light and arc-length ( Diagram 6 ) , with those connecting total loss of weight with arc-length ( Diagrams 1 and 2 ) , f suggests that there is a connection , though the low value of the luminosity for short arcs is primarily due to a different cause , namely , the difficulty of getting the light from the arc free from the shadow of the negative pole .
For long arcs the curves are more strictly comparable , and both are characterised by nearly horizontal lines .
If the total loss of weight from both poles is plotted against current ( see Diagrams 4 and 5 ) and compared with curves connecting mean spherical candle-power with current ( Diagram 7 ) , it will be noted that both are approximately linear , but that , whereas the candle-power-current curves point to the origin , the consumption-of-carbon-current curves do not , suggesting that some finite consumption of carbon is necessary before any of its energy can be converted into that of luminous radiation capable of affecting the instruments employed .
It is not improbable that the consumption of carbon should influence the * Blondel and Ayrton .
See ' The Electric Arc/ Mrs. Ayrton .
t Though drawn to represent loss per coulomb these curves represent total loss if the value of the ordinate be suitably altered for each curve .
The shape of the curve is unaffected .
M 2 Prof. W. G. Duffield .
Diagram 6 .
15,00 0 io/ io Siemens 8/ 6 Hard lo/ ioSoft io/ iO Hard Curves connecting Total Light with Length of Arc Current io Amperes Both Carbons Solid .
5000- Length of Arc in Millimetres .
The figures express the diameters of the upper \amp ; lower poles , From Mrs. Ayrton , ' The Electric Arc , ' p. 333 .
Diagram 7 .
5.000 Curves connecting Mean Spherical Candle - Power with Current for Constant Arc-lengths 4,0 00 4mm .
3,000 2,000 Current in Amperes .
5 10 15 Positive cored .
20 25 30 Negative solid .
From Mrs. Ayrton , 4 The Electric Arc , ' p. 366 .
The Consumption of Carbon in the Electric Arc. 143 candle-power , since the consumption of two candles gives twice as much light as one .
The action is exothermic , and the energy thus made available augments the radiant energy produced by electrical agency , to whose amount it bears an appreciable proportion .
To take a specific example :\#151 ; The energy liberated in each second from the anode , which supplies the bulk of the light , by the consumption of the observed amount of carbon ( 120 x 10"5 grm. ) in an arc of 10 amperes ( 52*4 volts ) and 4 mm. in length , upon the assumption that the carbon eventually burns to C02 ( when 97-3 x 103 cals , are produced for each gramme-molecule of C02 ) , is 4*1 x 108 ergs .
The electrical energy supplied per second amounts to 5*25 x 109 ergs , so that the total energy available from all sources is 5*7 x 109 ergs , and of this the energy of combustion is about 7 per cent. If we add to this the energy derived from the cathode consumption , the energy of combustion amounts to 11/ 3 per cent. If this energy is made available at the high temperature of the arc , it adds appreciably to its candle-power .
Increasing the rate of combustion should on this account increase the brightness of the arc .
This explains why soft carbons ( which burn away more rapidly ) have been found to give greater candle-power than hard ones under similar conditions.* The enormous increase in brightness , which the writer found for metallic arcs when burning in compressed air , f is probably due , in part at least , to the same cause ; for a given pressure the brightness was found to be largely dependent upon the freshness of the air supply\#151 ; when the air became contaminated there was a pronounced diminution in the intensity of the emitted light , which could , however , be restored by the admission of fresh air .
Recently it has been announced that increased brightness has been attained by spraying a jet of nitrogen upon carbon poles , which involves an endothermic reaction .
This does not affect our argument , since it is only for the carbon arc in air that it is suggested that the brightness and carbon consumption increase together .
The source of the bulk of the radiant energy is admittedly the current supply .
Mr. A. H. Davis , B.Sc. , and Mr. F. Southam , B.Sc. , whose services have been referred to in the text , assisted efficiently in these experiments , Dr. C. A. Sadler assisted in Section 10 of this paper , and to them and to Mrs. TV .
G- .
Duffield , who undertook the weighings in the early experiments , I express my thanks and appreciation .
To Dr. J. A. Harker , E.R.S. , who read an early draft of this paper , I am grateful for some criticisms of value .
My indebtedness to Sir Ernest Rutherford has been recorded in the text .
* See Diagram 6 .
t Duffield , 'Phil .
Trans. , ' vol. 208 , p. Ill ( 1908 ) ; vol. 211 , p. 33 ( 1911 ) , etc.

