]\gt ; The Velocity of the Negative lons in Flames .
By E. GOLD , B.A. , Fellow of St. John 's , Cambridge ; Wheatstone Laboratory , King 's College , London .
( Communicated ) Professor H. A. Wilson , F.R.S. Received December 4 , \mdash ; Read December 13 , 1906 .
) The paper contains an of a series of experiments on the electrical conductivity of a coal-gas flame , the principal object being to determine the yelocities of the negative ions in the flame produced eithel by the ionisntion of -vapours or by the ionisation of the flame gases .
Two nlethods have been used , and the results arrived at concordant .
The first metlJod depends on a determination of the ratio between the potential gradient in the flame some distance from the electrodes and the current flowing through it .
This method was used to determine the velocity of the ions of the flame gases .
The second method depends on finding the potential gradient necessary to cause the ative ions due to salt-vapour to move across the flame from one electrode to the other while rried upwards with the stream of gas .
The results obtained show that the velocity of negative ions in the flame due to a potential gradient of 1 ] per centimetre is about 10,000 cm .
per second .
The calculated velocity for a negative electron or corpuscle is about 13,000 cm .
per second , so that it appears that the negative ions in flames probably electrons .
Previous determinations of the velocity of the negative ions of various in flames have been made by several other observers , and all aglee in finding the velocity to be about 1000 cm .
per second only .
It is shown that this discrepancy is due to the assumption made by the previous experimenters that , for small potential difi'erences , the potential gradient between the electrodes is uniform .
The experiments to be described that this is very far from being the case , even when the potential erence between the electrodes is a fraction of a volt .
A discussion of the results previously obtained is contained in a ecent paper by H. A. Wilson .
* J. J. Thomson has calculated the of ion in a flame , and suggested that the negative ions in flames are electrons initially , but af.terwards get loaded with electrically-neutral ' The Velocities of the Ious of Alkali Salt Vapours } ) \ldquo ; ' Phil. Mag June , 1906 .
" " Conduction of Electricity through Gases p. 204 .
Mr. E. Gold .
[ Dec. 4 , molecules .
This view has also been put forward as possible by H. A. Wilson and by Lenard .
The results obtained in the present paper show conclusively that the negative ions are electrons and not atoms .
Conductivity Measurements .
The apparatus for the first series of experiments consisted of the quartztube burner previously described by H. A. Wilson.* The tube was 25 cm .
long and 2 cm .
in diameter , and had a row of 40 holes , each mm. in diameter , uniformly distributed along a line parallel to the axis of the tube .
FIG. 1 .
V , voltmeter .
, galvauometer .
, key .
, commutator .
, quadrant electrometer .
to source of E.M.F. gas and air from a large Bunsen burner were admitte to this tube and burnt at the holes .
In this way a , fairly uniform , flame was obtained and , as the quartz insulates well , there was no fear conduction down flame and along the tube .
'Phil .
June , 1905 , p. 476 .
1906 .
] The Velocity of the Negative Ions in Flames .
The electrodes used consisted of a pair of circular platinum discs cm .
in diameter .
They were supported in the flame so that they were clear of the green cones , and they could be adjusted to any desired position .
is a diagram of the apparatus .
In most of the experiments one electrode was kept fixed near one end of the flame while the other was moved horizontally along the flame .
The current through the flame was measured by a moving-coil galvanometer and the potential difference between the electrodes by an electrostatic voltmeter .
The potential gradient in the flame was measured by means of two fine platinum wires immersed in the flame and connected to the quadrants of a 's electrometer .
H. A. Wilson*has stated that for distances etween the electrodes the effect of putting salt on the tive electrode appears to be to increase the conductivity of the flame .
Experiments were made to test this effect , and for currents from to ampere it was found not to exist .
The above table gives some of the results obtained .
The galvanometer 'Phil .
Mag 1905 , p. 476 .
large increase in the conductivity of the flame between the electrodes was obtained ater by putting salt on both electrodes ; see p. 63 , infra .
Mr. E. Gold .
[ Dec. 4 , had a resistance of 220 ohms , and gave a deflection of one scale divi sion for a current ampere .
The electrometer gave a deflection of one scale division for a potential difference of volt , the deflection in all cases being a double one .
In the table the different values of the current , obtained with the same E.M.F. and salt and distance between the electrodes , are due to changes in the position of the salt-bead in the flame just below the negative electrode .
The mean value of deduced from this set is in the units used .
The diameter of the electrodes was cm .
, so that the ratio of the potential gradient to the current per square centimetre is approximately , which in C.G.S. units is or The conductivity of the flame is therefore C.G.S. unit .
This conductivity has been obtained on the assumption that the current flowed uniformly through a cylinder from electrode to electrode .
This is not the case , since the current will spread out .
and flow through the whole of the flame .
In the experiments the flame was not quite so wide as the electrodes , but a part extended above and below the electrodes .
These parts were , however , colder than the part where the gradient was measured , and the assumption that the coolness of the edges of the flame and these .
portions which project beyond the cylinder compensate for each other cannot be far from the truth .
We can from this determine approximately the velocity of the ions ; for if is the } on an ion and the number of ions per cubic centimetre , then the curTent density is given by the formula\mdash ; , where potential gradient , velocities of positive and ative ions respectively due to unit potential gradient .
Thus the conductivity of the flame is ne Now may be estimated roughly from the results obtained by Professor Wilson and the writer .
* It is shown in the paper referred to that the number of ions per cubic centimetre in a flame is given by where , in the experiments described , electrostatic units of potential , cm .
, and is the " " apparent specific inductive capacity\ldquo ; of the flame ' ' On the Electrical Conductivity of Flames containing Salt Vapours for Rapidly Alternating Currents 'Phil .
' April , 1906 .
1906 .
] Velocity of the Negative Ions in gases for rapidly .
currents .
For the flame without salt was found then to be approximately 2 Another value of can be got by comparing the results obtained wich currents and those previously obtained with steady currents .
* is shown in the paper referred to above that , where is the steady current due to an E.NI .
F. of volt .
For the free flame this in the units used , so that Taking to ) the mean of these two values , we get static umits , electromagnetic unit .
Therefore we may write or nearly , which gives a value 8000 under 1 volt per centimetre .
Now , the velocity of the positive ions , is only about 60 cm .
, so that cm . .
approximately .
Owing to the uncertainty about the value of , this result can only be regarded as a approximation .
second series of experiments was undertaken in order to obtain the velocity of the negative ions by finding minimum electric force ( ( some of them from one electrode to the other , assuming an upward velocity equal to that of the flame oases .
This method used by H. A. and G. Moreau , velocity deduced the obseryations on the assumption of a uniform potential gradient the flame .
The present investigation includes measurements of this radicnt for applied 's of the same magnitude as necessary to the ions across the flame .
The following is the arrangement adopted : Gas and air were led into a jar fitted with ( cork which three equal glass tubes passed ; the mixture frolU these was bnrnt at three equal small ) cyliuders fitting thin glass tubes , placed vertically over the tubes the jar and joined to them by short pieces of wide .
The flames so obtained were in every way similar and very steadily , * Smithells , Dawson , and Wilson , ' Phil. Trans , 1900 , p. 89 .
'Phil .
Trans , 1899 , p. 499 .
'Annales de Chimie et de Physique , ' 1903 .
Mr. E. Gold .
[ Dec. 4 , the of the green cones remaining constant throughout the experiments .
The flame tubes passed through three holes in a wooden stand , which served also to support tubes carrying the electrodes .
These were plane .
platinum rectangles cm .
and cm .
wide ; they were made rigid by thick platinum wires welded diagonally across them .
The flame and the electrodes were surrounded by a metal cylinder which shielded the flame from draughts .
The electrodes were connected through the galvanometer to .
a potentiometer .
is a diagram of the apparatus .
To obtain the minimum potential gradient necessary to drive some of the ions across the flame , the currentE .
M.F. curves were obtained for the free flame and for the flame with vaporised beneath the negative electrode , with various distances between the electrodes .
FIG. 2 .
, flames .
, metal shield .
, electrodes .
, wooden stand .
A , glass tubes .
, thermo-couple .
Figs. 3 and 4 give two pairs of such curves for distances and cm .
The curve for the free flame is very nearly a straight line .
When the salt bead is put in the flame beneath the ative electrode , the curve still remains nearly raight and in the same direction for small E.M.F. 's , but it gradually curves upwards until at a certain stage the rise has become .
1906 .
] Velocity of the Negatire Ions in Electrodes cm .
apart .
FIG. 3 .
Electrodes 3 cm .
apart , cm .
long .
VOL. LXXIX .
FIG. 4 .
Mr. E. Gold .
[ Dec. 4 , much more rapid , and the curve is again approximately a straight line , but has a much greater inclination .
It is assumed that this latter stage is reached when the E.M.F. applied becomes just enough to carry some of the rising ions entirely across the flame , from the bottom of the negative to the top of the positive electrode .
In the curves given the critical stage is reached for E.M.F. 's of and volt respectively .
The variation in the potential taken up by a platinum wire in the Bunsen flame depended on at least four things : ( 1 ) temperature changes in the flame , including changes in ionisation due to temperature alone ; ( 2 ) changes in ionisation arising from vaporisation of salt in the flame ; ( 3 ) changes in the E.M.F. applied to the electrodes ; ( 4 ) changes in position relative to the electrodes .
, battery .
, galvanometers .
, potentiometer .
, quadrant electrometer .
, voltmeter .
Pt Rh , thermo-couple wires .
, water bath .
The object in this part of the experiments was to determine that part of the effect of ( 4 ) which was due to applied E.M.F. alone , and as it was impossible to eliminate ( 1 ) and ( 2 ) , it was necessary determine separately the effect due to these and to correct for it .
For this purpose it was decided to use a thermo-couple as explorer , the couple consisting of two fine platinum and platinum-rhodium wires twisted together , the twisted part being about 1 mm. long , and the wires then bent back at right angles so that the distance between the parallel parts was about 6 mm. 1906 .
] The Velocity of the Negative Ions in Flarnes .
The thermo-couple was supported on the stand belonging to a reading microscope , and could be adjusted , by sliding along the raduated stand , to any position in the flame .
The tube passed a rectangular window in the metal shield , and to keep the flame as steady and free from draughts as possible , the tube carried a metal shutter which closed up the window while .
the couple to be moved across the flame .
The thermo-couple was standardised by finding the alvanometer deflection when a small bead of was just melted at the junction .
The temperature corresponding to this was taken to be 1066o C. , and that corresponding to any other deflection was calculated by first finding the platinum temperature\ldquo ; and then reducing to the Centigrade scale by the ctlrve given by Callendar .
* At first an attempt was made to find how the potential taken up by the thermo-couple depended on its temperature ; it was found that , when there was no applied E.M.F. , the couple was always at a potential than that of the electrodes , and that the potential difference between them varied approximately as the difference of temperature .
Fig. 6 shows the variation of temperature as the couple moved from electrode to electrode .
FIG. 6 .
Fig. 7 shows the way the potential difference between the couple and the electrodes changes as the couple is moved , there being no applied E.M.F. 'Phil .
Mag vol. 48 , 1899 , p. 533 .
Mr. E. Gold .
[ Dec. 4 , and no salt , and the electrodes being connected together , so as to be at one potential .
Fig. 8 shows the corresponding variation when is vaporised beneath the electrode ( B ) .
FIGS .
7 and 8 .
It will be seen that the Curves 7 and 8 are very similar , with gradients the same on the right and left sides , but that the maximum potential difference in 8 is nearly double what it is in 7 .
Thus , putting salt on one 1906 .
] The yelocity of the Ions in Flames .
electrode makes the difference of potential between the thermo-couple and the electrodes greater than before .
In each case electric equilibrium cannot supervene until ions have accumulated at the electrodes .
From these results corrections were applied to the values obtained for the gradient with an applied E.M.F. , and from the results deduced in this way it was seen that the major part of the fall still took place at the electrodes , at the negative electrode for the free flame , and at the positive electrode when salt is vaporised beneath the cathode .
It was not possible , however , to obtain in this way accurate results for the radiant in the body of the flame .
The assumption was , therefore , made that when the temperature , registered by the thermo-couple at a particular point in the flame , was unchanged and the vaporisation of salt at the electrodes remained steady , the change in the potential , taken up by the thermo-couple when a potential difference was applied to the electrodes , represented accurately the change in the potential due to the applied potential difference .
The temperature at a point in the flame oscillates about a mean value , and it is assumed that when temperature at a particular point has this mean value , the temperatures at other points are also mean-value temperatures .
To obtain results , with this condition holding , the following method was adopted:\mdash ; The thermo-couple was set at a certain height and moved across the flames from one electrode to the other in steps of a millimetre near the electrodes and cm .
in the middle of the flame .
For every different position of the thermo-couple the galvanometer and electrometer readings were taken ( 1 ) with the electrodes connected and no E.M.F. applied ; ( 2 ) with applied E.M.F. of , and volts .
These readings were all taken without any change in the condition of the flame or in the position of the thermo-couple or electrodes .
The difference between the electrometer deflections ( for a given position of the thermo-couple ) with no applied E.M.F. and with an applied E.M.F. was taken as the true potential difference between the point in the flame and the electrodes due to the E.M.F. applied .
In this way the true fall of potential across the flame , due to the applied E.M. was found , errors due to the thermo-couple not taking up the potential of the flame being eliminated .
The temperature of the thermo-couple in any given position was not quite constant , owing to draughts which it was found impossible entirely to prevent .
Thus the time-temperature curve of the thermo-couple in a given position in the flame is a continuous curve with maxima and minima oscillating about a mean value ; the electrometer readiugs .
this ) osition Mr. E. Gold .
[ Dec. 4 , were all taken when the thermo-couple indicated this mean value of the temperature .
The following tables give some of the readings taken in this way .
In Column I are the deflexions of the thermo-couple galvanometer ; in II the current through the flame ; III gives the distance in centimetres of the couple from the anode ; , the deflections of the electrometer ; , the potential difference applied to principal electrodes in volts .
April 4\mdash ; Curve 12 .
No Qalt .
Thermo-couple 2 cm .
below top of electrudes .
Distance between electrodes , 5 cm .
' April beneath B. Thermo-couple cm .
below top of electrodes .
volts on potentiometer .
Distance between electrodes , cm .
Mr. E. Gold .
[ Dec. 4 , beneath B. Thermo-couple cm .
beneath top of electrodes .
volts on potentiometer .
Distance between electrodes , cm .
over 9 III .
IV .
The current with an applied E.M.F. of volts here reached a value too large to be read off on the scale .
The mean current for this E.M.F. has been taken to correspond to a reading Figs. 9 , 10 , 11 are the curves obtained by plotting the results given in these tables .
It will be seen that the fall of potential consists of a large drop at the electrodes , and of a nearly uniform gradient in the body of the flame .
Fig. 9 gives the gradient for the free flame , and the fall is seen to be mainly at the cathode .
1906 .
] The Velocity of the Ions in Flames .
FIG. 9 .
give the fall of potential with salt on the cathode taken along horizontal lines from electrode to electrode at distances and cm .
below the top of the electrodes .
The curves are similar , so that the potential gradient does not change appreciably as we move up in the flame so long as we take points within the space included between the electrodes .
The fall of potential is now mainly at the anode , and the fraction of the total fall which occurs there increases as the E.M.F. applied diminishes , while the cathode fall disappears .
The following table gives the results obtained for the potential gradient with an applied potential difference of volts : Remarks .
Mr. E. Gold .
[ Dec. 4 , It is not possible from the readings taken to estimate accurately the potential gradient for the smaller applied E.M.F. .
We can see , however , that the gradient is very small and that there is no change in the nature of the curve .
FIG. 10 .
onKathode .
Thermocouple below top of electrodes .
Now , in the first part of the paper it was proved that the ratio of the current to the potential gradient in the body of the flame was independent of the applied E.M.F. and of the vaporisation of salt on the electrodes .
In the present case we find from the experiments the potential gradient for an applied potential difference of volts , and , knowing the values of the current for the other applied potential differences , we can deduce the gradient corresponding to them .
Taking the mean value of to be from the above table , the following the gradients deduced in this way for an applied E.M.F. of and volt : 1906 .
] The Velocity of the lons in Flames .
The currents for a given applied potential difference were nob quite constant during an experiment , but the ratios were nearly so , and corresponding currents have been taken throughout .
The mean gradient for an applied E.M.F. of volt with on the negative electrode is volt per centimetre .
Mr. E. Gold .
[ Dec. 4 , The conductivity of the flame deduced from the mean value is , which is in C.G.S. units , where A is the area of the cross-section of the conducting part of the flame .
The area A cannot be foumd accurately , but for the flame used , which was about 12 cm .
high above the green cones and cm .
wide , it is estimated that A would be 20 sq .
cm .
, allowance being made for the coolness of the edges and upper parts of the flame .
This gives for the conductivity in C.G.S. units 8 nearly , a result not differing much from the value found for the quartz-tube flame in the earlier experiments .
E.M.F. at which the current-E.M.F .
curve begins to rise rapidly with salt on the negative electrode for a distance between the electrodes of cm .
is volt nearly .
Thus , if is the of the electrodes , the upward velocity of the flame gases and therefore of the ions , the distance between the electrodes , and the distance the salt vapour extends from the negative electrode , then the velocity of the negative ions for unit potential gradient is given by where X is the gradient corresponding to an applied E.M.F. of volt , each side being the reciprocal of the time required for the first salt ions just to cross the flame .
Now , the flame containing the vapour extended cm .
from the electrode to its junction with the middle flame , and we may take this to be the value of .
Hence nearly , since volt per centimetre , or for a gradient of 1 volt per centimetre , To determine , the following method , suggested by Professor H. A. Wilson , was adopted .
A plane mirror was attached to one of the of an electrically-driven tuning-fork .
The reflection of the flame in this mirror was viewed through a telescope .
The reflected images of the paths of bright particles moving up in the flame were approximately sine curves when the mirror was vibrating .
If is the periodic time of the fork , the wave-length of the curves seen , the velocity of the particles upwards , then it is easy to see that .
The particles moved very quickly across the field of view and , although the curves could be plainly seen , it was not possible to 1906 .
] The Velocity of in Flames .
accurately measure the wave e. The paths of the particles were therefore photographed .
Lycopodium dust was fed in with the gas and air and was found to give out enough light .
The photographs were taken with a camera kindly lent by Mr. H. S. Allen , some on ordinary rapid plates and some on isochromatic plates ; the latter give the better results .
The negatives show a number of intersecting curves , which it is not easy to disentangle , but occasionally one shows up more distinctly than the rest and can be traced through two or three waves .
A mmimetre scale the same position as the flame was also photographed after reflection at the mirror , and this enabled the actual wavelengths to be obtained .
The photographed scale gave 36 divisions for 1 cm .
The measurement of the curves them a wave-length in the photographs of cm .
, or an actual wave-length of cm .
The fork nsed complete vibrations per second .
Hence the velocity of the particles is 187 cm .
per second , or cm .
per second .
Now , the velocity of the negative ions was found to be for a gradient of 1 volt per centimetre , thus cm .
per second nearly .
The rough estimate obtained in the previous section agrees with this result as well as could have been expected .
The velocity of corpuscles in an electric field may be calculated on the assumption that the corpuscles after impact with molecules act on the whole as though the velocity gained owing to the electric field vanished at impact .
This will at least furnish us with a lower limit to the velocity .
Let be the mean free path of a corpuscle among the gas molecules , and let be its mean velocity of agitation .
Then the number of impacts in unit time is , and we may assume that during this time the corpuscle traverses paths of all lengths .
The number of these paths whose lengths lie between and is The time of describing this path is if we the due to the velocity added by the electric field .
In our case this is permissible , provided the field does not exceed 10 volts per centimetre and we neglect changes produced in paths longer than 100 times the mean free path .
The distance moved in the direction of the field during this time due to the acceleration produced by the field , where X is the intensity Mr. E. Gold .
[ Dec. 4 , of the field and and the charge and mass of a corpuscle .
Therefore adding all the paths described in unit time the distance moved will be This will be the mean velocity of the corpuscle due to the field , and , its velocity in unit field , will be Now very nearly .
Townsend has shown experimentally that the mean free path of corpuscles in air is about four times the mean free path of air-molecules at the same pressure and temperature .
Taking the temperature of the flame to be 1800o C. , and the mean free path of a molecule in air to be cm .
at ordinary temperatures , this gives for the mean free path of a corpuscle in the flame nearly .
To get we assume that the equation holds between the corpuscles and the gas .
This gives taking the velocity of hydrogen molecules to be 1844 metres per second and the mass of a corpuscle to be 1/ 1000 of the mass of an atom .
Hence Therefore , substituting , we get for the velocity of a corpuscle under unit electric field , or for 1 volt per centimetre the velocity is 13,000 centimetres per second .
This velocity is very approximately the same as that found experimentally It appears , therefore , that the negatiye ions in the hot parts of flames are free electrons and not atoms .
Determinations of the velocity of the negative ions have previously been made by H. A. Wilson , G. Moreau , S and by E. Marx These writers all agree in obtaining for the negative ions a velocity of about 1000 cm .
per second for a gradient of 1 volt per centimetre .
The results obtained in this paper are of another order , and the difference is probably capable of being explained in the way .
Wilson and Moreau both assumed See Langevin , ' Theses , ' Paris , 1902 , p. 47 .
'Phil .
Mag 1901 , p. 198 ; ' Phil. Mag 1903 , p. 59@ .
Annales dhimie ehysique , Phil. Trans. , p. 30 .
' Annalen der Physik , ' vol. 2 , 1900 , p. 790 .
1906 .
] The yelocity of the Negative Ions in in their determination that the potential fall for small E.M.F. consisted wholly of a uniform gradient right across the flame , and that the electrode fall was negligible .
The present investigation shows that this is not the case and that the greater part of the fall of potential occurs near the electrodes .
The values of the potential gradient in the flame are thus less than they would be if the fall were uniform all across ; the previously deduced velocity would therefore be too small .
Marx measured the potential gradient at the electrodes and deduced the velocity on the assmnption of uniform ionisation .
Near the electrodes the gradient is very large and changes rapidly ; thus it is probable that measurements near the electrodes do notgive accurately the gradient at the electrodes by his theory .
The measurements of the gradient would be likely to give a value too small and , in consequence , a velocity also too small .
The velocities obtained in the present series of experiments have been only those of vapour of potassium carbonate and of the flame gases .
As previous experimenters have shown that the negative ions of all salts have the same velocity , their relative results remain true , although the absolute numbers are subject to correction .
It may therefore be concluded that the carriers of the electricity in salt-vapours in flames are corpuscles and are not atoms nor corpuscles loaded with neutral molecules .
A series of measurements were made with salt vaporised beneath both electrodes .
In this case the current is very much increased , and the fall of potential consists of nearly equal sudden falls at the electrodes and an approximately uniform gradient in the body of the flame .
gives the curves obtained from these measurements .
It appears from these curves that there are ions of the gradient near each electrode .
These are probably due to the excess of positive ions near the cathode and of negative ions near the anode , an inspection of the curves shows that the effect extends for about A cm .
from each electrode .
We may then conclude that there is a marked excess of ions of one sign in a layer extending cm .
from each electrode .
The following table gives the results deduced from these observations:\mdash ; * The equation of the potential , d-V/ d , shows that the volume density of electricity is nearly proportional to the curvature , being positive where the curve is concave upwards .
Mr. E. Gold .
[ Dec. 4 , FIG. 12 .
beneath -Thermocouple cm .
be It will be seen from this that the conductivity is greater than in the preceding .
experiments , although it is practically the same for volts as for volt .
This effect cannot be due to diffusion of the ions ; for if such were the case we should expect a conductivity changing from the electrodes towards the centre of the flame , but the curves show that there is no change for a distance of 3 cm .
, the distance of unilorm gradient .
Neither can it be due to ions driven into the body of the flame by the electric field , for it is unchanged by changing the applied potential difference .
Neither can it be be due to ions shot out from the region near the glowing , for we should then get a similar result with salt on one electrode only , and this we have seen is not the case .
The accumulation of ions at both electrodes when salt is supplied at both of these points to a higher proportion of ions all along the flame as infra , which will account for the fact ; further experiments with varying distance 1906 .
] , Velocity of the Negative Ions in between the electrodes and different salts might elucidate the conditions further .
* As previous experimenters have observed , in all the expenments when the temperatures of the electrodes were different , there was a small current when no E.M.F. was applied and the electrodes were connected together .
This effect was increased when salt was put on one of the electrodes .
This current is inappreciable compared with the currents produced with applied E.M.F. greater than 1 volt , but for smaller values of the potential difference this inverse current is comI ) arable with the current due to the applied E.M.F. hroughout these experiments the current due to a small applied E.M.F. has been ured , not from the zero of the galvanometer , but from the scale reading when the galvanometer was connected through the flame .
The current is therefore larger by an amount equal to the inverse current , and as this varies with the salt and the temperature of the electrodes , it was necessary to determine the " " zero\ldquo ; for each experiment .
The explanation of the inverse current seems to be this .
At the surface of the glowing platinum a continual interchange of ions is going ) , the hot platinum both emitting ions and bing them from the flame .
Since the electrodes and galvanometer form an insulated system , a steady state must be reached , when on the whole the emission and absorption are balanced .
Since , however , their rates will not depend in the same way on the temperature , one electrode will be emitting more than it absorbs , and vice versa for the other .
There will thus be a current through the galvanometer from one electrode to the other .
When salt is put on one electrode , the number of ions in the gas there situated will be increased , and the number absorbed by the electrode will also be increased .
The potential of the electrode system will therefore rise until it is sufficiently above the potential of for the increased emission due to the electric force to compensate for the increased absorption .
The emission will increase at both electrodes , while the increase in the absorption will be at one only .
There will , therefore , be an increased current through the galvanometer which is in accordance with observation .
negative ions will be the main factors in this interchange , on account of their greater velocity in the flame , so that there will be an excess of negative electricity at that electrode which is in contact with the salt-vapour .
The current will therefore be through the galvanometer to that electrode , which agrees with the experimental facts .
* A similar effect was obtained by H. A. Wilson toc .
with salt vapour at the negative electrode only .
VOL. LXXIX.\mdash ; A. Sir J. Dewar and Dr. H. O. Jones .
[ Dec. In conclusion , I wish to record my indebtedness to Professor H. A. Wilson for much kind advice and encouragement given during the course of these experiments .
On New Iron rbonyl , on the Action of Light and of Heat on the Iron Carbonyls .
By Sir JAMES DEWAR , M.A. , Sc. D. , .D .
, F.B.S. , Jacksonian Professor in the University of Cambridge , and HUMPHREY OWEN JONES , M.A. , D.Sc .
, Fellow of Clare College , and Jacksonian Demonstrator in the University of Cambridge .
( Received December 19 , 1906 , \mdash ; Read January 24 , 1907 .
) This paper contains an account of results of the continuation of the experiments on the action of light on the liquid iron carbonyl ( iron pentacarbonyl ) and the action of heat on the resulting solid compound , differrononacarbonyl , , which were described in a paper on " " The Physical and Chemical Properties of Iron Carbonyl communicated to the Society in The experiments on the action of on iron pentacarbonyl under varied conditions have resulted in new and interesting observations , and approximate measurements of the velocity of the reaction induced by light have been made and compared with that of other reactions induced by li .
The action of heat on solid iron carbonyl , which we described in our former communication as producing an intense green colour , which colour Dr. Mond stated that he had ) served casually , has been further examined , and the experiments now described have led to the discovery of a new compound of iron and carbon monoxide , iron tetracarbonyl , which forms dark green crystals of remarkable stability , exhibiting some very striking .
properties .
Action of liight on Iron Pjntacarbonyl .
It was shown in the former communication that the action of on iron pentacarbonyl alone or in solution in solvents , with the exception of nickel carbonyl , is represented by the equation Carbon monoxide is evolved and differro-nonacarbonyl , an orange red crystalline is produced and separates out .
When pyridine was- 'Boy .
Soc. Proc , 1905 , , p. 558 .

