]\gt ; takes the simple form the solution of which is , ( 2 ) being the concentration of the ions initially , and the concentration after an interval of time .
In what has been said above the effect of diffusion is supposed to be negligible .
The coefficient of recombination is of fundamental importance in the theory of the ionisation of gases , and an experimental verification of equation ( 2 ) serves as a general test of the theory .
Methods of measuring the coefficient have been devised by Rutherford , *Townsend , Langevin , and others , and many determinations have been made by different observers .
S The results obtained present some rather wide divergences , both as regards Butherford , ' Phil. Mag 1897 , vol. 44 , p. 422 .
Townsend , ' Phil. Trans 1899 , vol. 193 , p. 129 .
Langevin , ' Annals de Chimie et de Physique , ' 1903 , vol. 28 , p. 433 .
S McClung , PhiL Mag 1902 , vol. 5 , p. 283 ; Hendren , 'Phys .
Rev 1905 , vol. 21 , p. 314 ; Retschinsky , ' Ann. der Physik , ' , 4 , vol. 17 , p. 518 .
values are liable to present wide divergences , and there is some difficulty in weighing up the results .
In the second place , the distribution of the ionisation in the vessel is of no small importance .
On account of the secondary radiation which is always present in these experiments , and which may produce intense local ionisation , it is impossible to know in an exact manner the distribution of the ionisation in the vessel .
Finally , the effects of the diffusion of the ions to the walls of the vessel are important , and there is some di[ficulty in correcting for this effect .
Langevin*has shown the importance of diffusion in McClung 's experiments .
At ordinary 'Journ .
de Physique , ' 1906 , vol. 4 , p. 322 .
Rontgen Rays in Gases Vapours .
479 pressures diffusion accounted for about 10 per cent. of the observed recombination , and , at a pressure of one-eighth of an atmosphere , the values obtained were probably five or six times too large .
In Langevin 's method these are avoided , and this was the'one used throughout these experiments .
Before giving a short account of the method , it may be useful to consider the significance attached by Langevin to the term ' number of colIisions.\ldquo ; Suppose that round each negative ion in an ionised gas a surface is described , the dimensions of which are arbitrary , except that its largest dimension is large compared with the mean free path of an ion , and small compared with average distance between two ions .
The number of positive ions which , in a time , penetrate into the interior of these surfaces , as a result of the attraction of the negative ions inside , is independent of the dimensions of the surface and of the eJectric field existing in the gas due to the neighbouring conductors .
This number Langevin defines to be the number of collisions between ions of opposite sign in the time The number of these collisions is by the expression in which and are the mobilities of the positive and negative ions respectively , and are the concentrations of the positive and negative ions , and is the charge on an ion of either sign .
This expression contains neither the dimensions of the surface supposed to surround the ion nor the strength oectric fdchange oositive aegative itokinetic energy oemented bnergy dheir meparate again after havingapproached , ravitated tround ther .
Afraction necess inity , umber oisions w If each collision results in a recombination of the two ions , the rate of But recombination does not necessarily follow a collision .
The initial Comparing this with the expression ; we have and .
( 3 ) in which and Experiment gives the relative values of the quantities and : : are proportional to the quantities of electricity received by the electrode in the two cases after one discharge from a Crookes tube , and and are propor : uantities olectricity induced olectrode aProduced bontgen R when tectrio fangevin.onvenieIhce oanation 1ater , onvenientto describe triefly hhich Qatios Qgiven bationx aurve irawn weing The points corresponding to and have a known rence of abscissae . .
The corresponding ordinates are and and their difference is also a known quantity In order to determine and , it is sufficient to determine on the curve two points and , the abscissae and ordinates of which differ respectively by the known quantities and .
By ving a transparent pa:per over the curve it is possible to find these points with both ease and accuracy .
The abscissae and ordinates of these points are and , and we have An example will illustrate this .
In an experiment on dry carbon dioxide at a pressure of 614 mm. of culy and a temperature of C. , th6 following results were obtained:\mdash ; ' necessary modifications were introduced .
A momentary discharge was produced in the Bontgen-ray bulb by the breaking of the contact at , which was in the primary circuit of a Marconi induction coil .
The contact at was between two pieces of platinum which could be separated very suddenly .
For effective action at this break it was necessary to have the platinum pieces held together by a steady , firm pressure .
A capacity of about 8 microfarads was inserted in order to suppress .
the spark which would occur when the circuit 1913 .
] Produced by Rontgen Rays in Gases and .
483 was broken .
A small spark here gave great irregularities in the -lay flash ; however , when this was entirely eliminated , and with a constant as measured by the ammeter , in the primary , the ionisation produced in the chamber by a single flash of rays was very constant , and readings could be repeated with great ease .
The ionisation chambers are denoted by AB and .
The upper electrodes A and , by means of a key , could be connected , either separately or together , with the insulated pair of quadrants of the electrometer .
Two exactly similar ionisation chambers were used and the Rontgen-ray bulb was adjusted so that the ionisation produced in each chamber by a single flash of the rays was the same .
With electric fields of opposite sign in the two chambers , the electric charge received by the electrometer when connected with both chambers was zero .
The quantity received in either of these chambers gives the charge If the electric field in one chamber be now increased , one electrode receives a quantity of electricity QL and the other a quantity of opposite .
If the two electrodes are both connected to the electrometer , the latter will suffer a deflection proportional to the difference of these charges , that is to .
This difference is a very small fraction of either of the charges or ( \amp ; and the error in is approximately proportional to the error in .
It is better , therefore , to measure and rather than and separately .
may be only 1 or 2 per cent. of , and some special device is necessary if both quantities are to be measured with the same electrometer .
Langevin used an electrometer with a condenser of variable capacity .
similar method was first tried in these experiments but was not found to be very satisfactory .
The variable capacity employed consisted of two sets of 10 parallel plates fixed to two axes .
One set , which was connected with the electrometer , remained fixed ; the other , which was kept at zero potential , could be rotated into or out of the fixed set , thus increasing or decreasing the capacity .
In measuring the quantity , the condenser was The accumulation of the ions in the neighbourhood of the electrodes affects the electric field , which is no longer uniform but is of greater intensity in the vicinity of the electrodes .
In the case of a uniform ionisation between the electrodes , Langevin has shown that the relative error resulting from the modification of the fieJd by the ions has for its principal term and , at the same time , its upper limit , the value .
The error , therefore , will be less than 1 per cent. so long as does not exceed .
With a non-uniform ionisation the modffication of the field by the ions becomes more important .
At low pressures with small values of and becomes very small , and it is necessary to have arger values of the ratio .
In general , however , voltages were ohosen so as to render the effect of the nisation on the field practically negligible .
In fig. 2 , A and are the ionisation chambers .
The plates 1 and 4 can be connected , either separately or together , to the electrometer , and the potential of the lower plate 2 can be varied .
The outer case of the ionisation chambers , receives auantitiesLet qoefficients oapacity anduction oystemsto beasured aatios ( where 4ductors , uffixes referl iumbers iigurethe eectrostatic fetween tates 1irst chamberexperiment .
Suppose tiash oontgen rthemeter , part oonductor kthroughout thethe guard tubes Tuadrants , oduced biontgen Rectrometer bated a disconnected ; the charge on the conductor 1 is given by the relation in which and , are the potentials of 1 , 2 and 3 , respectively .
Initially , and are both zero , so that the equation may be written .
( 4 ) When the charge is given to 1 , let the potential of the electrometer become .
Then we have .
( 5 ) Suppose now that the potential of 2 is changed to , and that , in consequence , the potential of the electrometer becomes , then , since there is no increase in the charge on 1 , we have , whence , subtracting ( 4 ) , If be adjusted so that the potential of the electrometer is brought back to zero , then .
( 6 ) Similarly , when 1 and 4 are both connected to the electrometer and therefore , since is zero .
( 7 ) When the charge is given to 1 , let the potential of the electrometer become .
Then we have Suppose now , that the potential of 2 is changed to , and that the electrometer , in consequence , assumes a potential .
We have and subtracting ( 7 ) from @ method it was unnecessary to have the plates charged for any appreciable time , and therefore the effects of slight insulation leaks were reduced to a minimum .
The manner in which the changes were brought about can be seen from the figure .
The electrode ( fig. 1 ) was connected to the point When and and and were connected , was at the potential of the plate of the battery of cells .
By breaking the contact between and , and joining and , the potential of could be changed to that of the point N. was a potentiometer consisting of 100 equal resistances .
The current from a small number of accumulators passed through this , and , by moving 1913 .
] Produced by Rontgen in Gases Vapours .
487 round the pointer , any fraction of this small potential could be added to or subtracted from that of the battery .
By connecting to instead of to , the potential of the electrode could be changed by a small known amount .
This was used in measuring the quantity .
The potential of the electrode was determined by that of the battery .
The small change of potential required to balance was measured by means of a potentiometer and a Weston cell .
An experimental difficulty may be noticed here .
If the gas between the plates was ionised while the plate A was insulated and the plate was connected to earth , a deflection corresponding to a positive charge was produced in the electrometer .
This effect was not due to a difference in the coefficients of diffusion of the positive and negative ions because an effect of a similar order of magnitude was obtained in carbon dioxide , in which the coefficients of diffusion of the two kinds of ions are very nearly the same .
The effect was probably due to the emission of electrons from the upper electrode under the influence of the Rontgen rays .
It was shown , however , that this effect did not influence the results obtained .
The charge received by the upper electrode , when a flash of Rontgen rays was used , was measured for increasing electric fields , first in one direction and then in the opposite direction .
For small values of the electric field the curves obtained were unsymmetI'ical , but this lack of symmetry decreased very rapidly as the potential increased , and for the large fields used in these experiments the two curves for electric fields of opposite sign were identical .
4 .
Preparation of ihe Gases The atmospheric air was thoroughly dried before being admitted to the apparatus by allowing it to remain in contact with phosphorus pentoxide for some time .
The carbon dioxide was liberated by the action of hydrochloric acid on calcium carbonate and was dried by allowing it to remain in contact with phosphorus pentoxide for some time .
The carbon monoxide was liberated by the action of sulphuric acid on formic acid and was dried over phosphorus pentoxide .
The sulphur dioxide was obtained from a cylinder of the liquid and was dried over phosphorus pentoxide .
The nitrous oxide was obtained from a cylinder of the gas and was dried over phosphorus pentoxide .
In some cases the gases were passed a spiral tube immersed in liquid air .
The ionisation chambers were exhausted by means of a Topler pump , and the remaining gas was absorbed by means of charcoal immersed in liquid air .
The dry gas was then admitted to the apparatus .
In every case great care was taken to make the gas or vapour dust-free .
variations , however , were within 6 or 6 per cent. of the mean value .
The values apply to a mean temperature of about C. The temperature at the time of the observations was always recorded , but no correction was made as it was considered that such a correction would almost certainly be within the limits of error of the experiment .
6 .
Discnssion of Results .
this table it will be seen that the coefficieni is very nearly proportional to the square of the pressure over a fairly wide ange of pressures .
At higher pressures this rate of increase becomes smaller .
1913 .
] Produced by Rontgen in Gases .
489 Table I. Pressure in mm. Air .
0.272 0.272 0.272 0.272 0.272 0.272 0.272 0.272 0.272 0.272 0.272 0.272 0.272 0.272 0.272 630 Carbon Dioxide .
0.151 0.078 0.038 0.151 0.078 0.038 0.151 0.078 0.038 0.151 0.078 0.038 0.151 0.078 0.038 0.151 0.078 0.038 0.151 0.078 0.038 0.151 0.078 0.038 0.151 0.078 0.038 0.151 0.078 0.038 0.151 0.078 0.038 0.151 0.078 0.038 0.151 0.078 0.038 0.151 0.078 0.038 0.151 0.078 0.038 Carbon Monoxide .
364 Sulphur Dioxide .
0.466 0.019 0.466 0.019 0.466 0.019 0.466 0.019 0.466 0.019 0.466 0.019 0.466 0.019 0.466 0.019 0.466 0.019 0.466 0.019 0.466 0.019 Nitrous Oxide .
Such a departure is to be expected if we accept Langevin 's interpretation of , for it would tend to approach a limit when every collision is accompanied by recombination .
Under these circumstances would be unity , and a further increase in the pressure would produce no increase in the value of .
This limit appears to be reached for compalatively small pressures : air at a pressure of about 4 or 5 atmospheres ; in sulphur dioxide at an even lower pressure .
Pressure 757 690 Coefficient of recombination 1820 1600 Sulphur Dioxide .
Pressure 680 504 444 338 200 Coefflcient of recombination 2740 2260 1900 1510 876 548 Nitrous Oxide .
Pressure 749 596 480 204 200 Coefflcient of recombination 28$0 2400 1690 1110 690 Townsend McClung Langevin Air .
Carbon dioxide .
There is a good agreement between the results obtained by different observers working with air and carbon dioxide at atmospheric pressure .
This is more striking when one considers the great difference between the method employed in the experiments described in this paper for example , and that employed by McClung .
At low pressm.es , however , there are wide divergences .
From the curves * ' Phil. Trans , vol. 209 .
Produced by Rays in Gases and Vapours .
491 Fi it can be seen that , for a considerable range of pressure , depending on the gas , the coefficient of recombination is proportional to the This VOL. LXXXVIIL\mdash ; A. moving particles ejected by the Rontgen rays would disappear almost instantaneously in the presence of the large electric fields used .
This is supported by the agreement between the results of the experiments made under different experimental conditions ; large variations of the electric field and of the intensity of the ionisation do not seem to produce any appreciable effect on the coefficient of recombination .
Although for this reason the results obtained by Plimpton cannot be compared with those here , one point may be noticed .
As would be expected , Plimpton 's values are in every case very large , but these values decrease when the time-interval during which diffusion is allowed to take place is increased .
The final slopes of the curves obtained from Plimpton 's data , however , are too steep to enable one to deduce a limiting value with any degree of accuracy .
This is especially so in the case of carbon dioxide , in which the coefficient of much smaller than it is in air .
The final value obtained is 4880 compared with an initial value of 10,000 , and the value is still decreasing .
In the case of air the lowest value for is 3960 .
This value is about 10 per cent. higher than the result given here .
The values of obtained at low pressures in these ximents may be 'Roy .
Soc. ' 1912 , , p. Produoed by Rontgen Rays in Gases and .
493 briefly mentioned here .
The observations at low pressures are much difficult than those at higher pressures , especially in those gases in which the ionisation is small .
The important quantity which has to be measured is , the difference between the quanli of electricioy which reach the electrode for two different values of the field .
This quantity depends on the value of the coefficient and on the ionisation in the gas .
Both these decrease when the pressure decreases , the former being approximately inversely proportional to the square of the pressure .
At low pressures , therefore , the difference becomes very small .
This difficulty can be obviated to some extent by using more than one flash of the rays , but this is not so satisfactory .
The results obtained do seem to indicate departures from the linear law at lower pressures , but in view of the increased difficulty of the experiments it is perhaps well not to lay too much emphasis on these variations .
It is significant , however , that the changes appear in the region of pressure in which other experiments seem to suggest that some modification takes place in the negative ion .
The constancy of the results that have been obtained in these experiments under conditions which have been very widely varied seems to be in favour of the simple law of recombination or in its more general form , Among the modifications of this law which have been proposed , perhaps ths best known is that of Sutherland It is not always easy to distinguish between these two laws , and the results of some observers seem to fit in almost equally well with both formulae .
Experiments in which ionisation of very widely varying intensity is used seem most promising .
It is hoped to give a more detailed account of some experiments on this point at a later date .
tion .
Regarded from this point of view alone it is of advantage , in the construction of a reflecting telescope , to make the mirror of as large a size aa the mechanical difficulties incident to the construction of large mirrors allow , difficulties , that is to say , such as that of obtaining the glass of the necessary homogeneity throughout , in order to avoid distortion owing changes of temperature , and the difficulties of grinding , shaping , and polishing .
There is , however , another factor to be considered which affects the problem .
The mirror , even if perfect in other respects , will , when partially supported , be distorted by its weight , to a greater or less extent according to the nature of the support .
This distortion will cause a diminution in the resolving power , ffie of which will evidently increase with the size of the mirror , and so will tend to counteract the advantage accruing from the larger aperture .
It is conceive .able even that there will , for any

