120 Mr. H. A. Wilson .
On Electrical " On the Electrical Conductivity of Flames containing Salt Vapours/ ' By Harold A. Wilson , B.Sc. ( Lond. and Vic .
) , 1851 Exhibition Scholar .
Communicated by Professor J. J. Thomson , F.RS .
Received March 10 , \#151 ; Bead April 27 , 1899 .
( Abstract .
) The experiments described in this paper were undertaken with the object of following up the analogy between the conductivity of salt vapours and that of Bontgenised gases , and especially of getting some information about the velocities of the ions in the flame itself .
They are to some extent a continuation of the research of which an abstract has already been published in the 'Proceedings of the Boyal Society.'* The paper is divided into the following sections :\#151 ; ( 1 ) Description of the apparatus for producing the flame .
( 2 ) The relation between the current and E.M.F. in the flame .
( 3 ) The fall of potential between the electrodes .
( 4 ) The ionisation of the salt vapour .
( 5 ) The relative velocities of the ions in the flame .
( 6 ) The relative velocities of the ions in hot air .
( 7 ) Conclusion .
The apparatus used for producing the flame was similar in principle to that used in the investigation referred to above .
Carefully regulated supplies of coal gas and air were mixed together along with spray of a salt solution , and the mixture burnt from a brass tube 07 cm .
in diameter .
The flame thus obtained was steady , and measurements of its conductivity , when a particular salt solution was sprayed , did not differ more than 1 or 2 per cent , on different days .
The height of the inner sharply defined green cone was 1'5 cm .
, and that of the outer cone 7'5 cm .
The current between two gauzes of platinum wire , each 14 cm .
in diameter , and placed horizontally one above the other in the flame , was measured for E.M.F. 's up to 800 volts , and with various distances between the gauzes .
The current with a large E.M.F. was found to be independent of the distance between the electrodes when the upper electrode or gauze was positively charged , provided that the distance between the electrodes was not so great that the upper one was in the cooler parts of the flame near its point .
When the upper gauze was comparatively * " The Electrical Conductivity and Luminosity of Flames containing Yaporised Salts , " by Arthur Smithells , H. M. Dawson , and H. A. Wilson , ' Roy .
Soc. Proc. , ' vol. 64 , p. 142 .
Conductivity of Flames containing Salt Vapours .
121 cool the current was much smaller , but if the upper gauze was kept hot by passing a current through it , then the current with a large E.M.F. was independent of the distance between the electrodes , even when the upper electrode was above the point of the flame .
If both of the electrodes were hot , then the current , as the E.M.F. was increased , attained a nearly constant value .
Cooling the positive electrode by raising it in the flame caused the current to increase towards this saturation value much more slowly than before , while cooling the negative electrode , the positive one being hot , caused the current to show no sign of arriving at a maximum value .
The current was much greater when the negative electrode was hot , and the positive electrode cool , than when the negative electrode was cool , and the positive one hot .
The fall of potential in the flame between the gauzes was examined by putting in an insulated platinum wire , and finding the potential it took up .
When both the electrodes were hot , the fall of potential closely resembled that observed in gases at low pressures .
That is to say , near each electrode there was a comparatively sudden fall of potential much greater near the negative electrode than near the positive , with a small and nearly uniform gradient in between .
If either of the electrodes was cooled , then the fall of potential near that electrode became much greater , and often was nearly equal to the total drop of potential between the electrodes .
This effect was usually much more marked in the case of cooling the negative electrode than with the positive electrode .
If the positive electrode was uppermost and somewhat cool , then with small E.M.F. 's practically all the potential fall occurred near to the positive electrode ; but if the E.M.F. was sufficiently increased , then a drop of potential appeared at the negative electrode , and with a still greater E.M.F. this became greater than that at the positive electrode , as it is in gases at low pressures .
Some of the results obtained pointed to the conclusion that nearly all the ionisation of the salt vapour takes place at the surfaces of the glowing electrodes , and not throughout the volume of the flame .
A variety of experiments were tried to test this view , all of which confirmed its correctness .
Thus , with two platinum foil electrodes opposite one another in the flame , no increase in the current between them occurred when a bead of salt was put between them , so that the salt vapour from it passed between them without touching either electrode .
If the vapour came in contact with the negative electrode , there was a great increase in the current , and a considerable but smaller increase when it came in contact with the positive electrode .
The relative velocities of the ions of alkali metal salts in the flame were estimated by finding the potential gradient necessary to make the ions travel down the flame against the upward current of gases .
122 Electrical Conductivity of Flames containing Salt Vapours .
This was done by putting a bead of salt between the two gauze electrodes , and finding what E.M.F. was necessary to produce an increase in the current between the electrodes when the bead was put in .
The potential gradient corresponding to this least E.M.F. was then determined .
In this way it was found that the positive ions of salts of Li , Na , K , Kb , and Cs , all have nearly the same velocity in the flame , whilst the negative ions of various salts of these metals also have equal velocities which are about seventeen times as great as the velocities of the positive ions .
The velocity of the positive ions was estimated to be about 60 cm .
per second for one volt a cm .
, and that of the negative ions was about 1000 cm .
per second .
The relative velocities of the ions of various salts was also determined in a current of air at about 1000 ' C. , which was obtained by passing the air through a platinum tube 1*3 cm .
in diameter , and 50 cm .
long , heated in a gas-tube furnace .
The method used was exactly analogous to that used in the flame .
The ions could be divided into three classes , in each of which all the ions had equal velocities , viz.:\#151 ; Velocity .
1 .
Negative ions of salts of Li , Na , K , Kb , Cs , Ca , Sr , and Ba ... ... ... ... ... ... ... ... ... ... ... .1 ... ... ... ... 26'0 cm.-sec .
2 .
Positive ions of salts of Li , Na , K , Kb , and Cs ... ... ... 7-2 " 3 .
Positive ions of salts of Ca , Sr , and Ba ... ... ... ... ... j 3*8 " It thus appears that those ions which in solutions carry equal charges have equal velocities in the gaseous state .
This points to the conclusion that the velocity of a gaseous ion in a given medium depends only on its charge .
The velocities are less than those calculated for ions consisting of one atom , so that each ion appears to be a cluster of atoms .
If we regard this cluster as held together by the charge on it , then it is reasonable to suppose that the size of the cluster will be determined by the charge .
Hence those ions having equal charges will be of equal sizes , and consequently of equal masses , since the atoms forming the cluster probably come from the medium rather than from the small quantity of salt present .
Consequently they all have the same velocity under similar conditions .
The two main results arrived at in this paper , viz. , that the ionisation of the salt vapour in the flame takes place only at the surfaces of the glowing electrodes , and that the velocity of the negative ions in the flame is very much greater than the corresponding velocity of the positive ions , 'enable the phenomena of unipolar conduction to be very easily explained .
For example , if one electrode is much hotter than the other , then if the hot electrode is negative , it will give off negative ions very freely , and there will be a large current ; but if the hot 124 Messrs. E. Threlfall and J. A. Pollock .
has to be twisted or untwisted to bring the lever to its sighted position .
In carrying out the construction of the instrument on the principle thus explained the following conditions have to be fulfilled:\#151 ; The instrument must be portable , and must be able to withstand the rough usage inseparable from travelling , without being put out of adjustment .
It must have a sensitiveness of at least one part in 100,000 of the value of " g , " i.e. , a change in the value of " g " amounting to one part in 100,000 must be shown by the balance .
These conditions have been satisfied in the following manner :\#151 ; The thread supports form part of a girder mechanism which is itself contained in a thermally insulated tube .
During transport the lever is arrested by a mechanism which clamps it with a definite pressure in a definite position .
The end of the lever is observed by a microscope which is always brought into the same relative position with respect to the horizontal plane through the thread by means of sensitive striding levels .
It is shown as a consequence of the mechanical conditions that the lever will be in unstable equilibrium when its centre of gravity rises above the horizontal plane through the thread by about 3 ' .
Consequently the accuracy with which the lever can be brought to its sighted position is very great , for the position selected as the sighted position is within a small fraction of a degree of the position of instability .
As it is necessary either to keep the balance in an atmosphere of constant density or to correct the observations for changes in the barometrical pressure , the former course was decided upon , and consequently the instrument is contained in an air-tight space .
This involves working the vernier axle through a stuffing box which must be practically frictionless , a condition satisfied by a sort of mercury sealing .
The difficulty in making the apparatus arises from the fact that quartz fibres , though infinitely better than any other material , are not really sufficiently perfect in their elastic properties for the present purpose , and it is only by a judicious balancing of advantages that it is possible to arrive at the necessary sensitiveness .
Even after two years ' twisting the thread of the instrument still undergoes a continual , though small , viscous deformation ; this , however , becomes sensibly constant , and can be allowed for .
A further complication arises from the fact that as the temperature rises the quartz becomes stifFer , so that at a given station the circle readings are a function of the temperature .
We have found that the relation between the circle reading and the platinum temperature is a linear one at ordinary temperatures .
An essential feature of the apparatus , therefore , is a platinum wire thermometer placed alongside the thread .
' The following statement shows concisely the effect on a determi-

