]\gt ; The Scattering of X- and ) .
395 thallium .
Prof. Lindemann has been enough to examine these plates , and he is able to confirm the reality of the difference in rpness , but it may perhaps be due to slight valiations in pressure or the huixture of other elements , and it would not be safe to draw any conclusions from it .
The results in case of lead SGGI1 so definite that the spcctroscope may yet prove to be a most lable weapon the problem of isotopic elements .
The of b Test of the Ring of Atom .
By G. A. SCHOTT , } , D.Sc .
, Professor of Applied , University College of lVale Aber.ystwyth .
Communicated armor , Rcceived September 1 .
It is generally dnli t the cctron R Theory of on the classical electrodyna1nics and mechanics , not explain certain results of recent estigations respecting thermal radiation , photoelectric emissions , transformations between corpuscular and X-ray energies and spectrum series .
All these phenoltlen a involve the iltelaction of radiation in one form or another , not merely with electrons , but also with matter , not so much with tter iu as with material atoms , and the failure of the Electron Ring Theory to account for them may after all be due rather to our ignorance of atomic structure than to defects in the fundamental principles of the theory itself .
Moreover the hypotheses proposed for the explanation of the phenomena in question , such as the hypothesis , themselves fail to account for other phenomena , such as interference , diffraction and pol}risation , which so familiar to us that the necessity oi them afresh is apt to ] ) overlooked when a new theory is introduced .
In view of the present unsatisfactory state of our theories and the indecisive character of our experimental knowledge a crucial experimental test of the Electron Ring theory is very ltlllch to be desired , especially one little of atomic structure as possible .
object of the p1esent investigation is to show that the phenomenon of the scattering of and -rays by matter a test of this kind and , moreover , if the issue of the test be favourable , supplies a means of investigatin , nentally the arrangement of the electrons in at any rate the hter atoms .
By scattering meant the redistribution in space without change of frequency of a portion VOL XCVI . .
2 Prof. G. A. Schott .
of the radiation incident on matter .
For our purpose any radiation of altered frequency , generated by processes akin to fluorescence or phosphorescence , must be carefully distinguished from scattered radiation .
2 .
Many years ago Sir J. J. Thomson propounded his well-known theory of the scattering of X- and -rays by matter , based mainly on three assumptions : ( 1 ) X- and -rays consist of extremely thin and intense electromagnetic pulses ; ( 2 ) scattering is due entirely to the comparatively mobile electrons of the atom , that due to the relatively inert residue being neglected ; and ( 3 ) the netic forces exerted by the pulses on the electrons are so intense that the due to other electrons and to the residue of the atom may be left out of account .
Two of the results of this theory are important for the purpose of comparison with those of the present investigation : ( 1 ) The fraction of the incident energy scattered in a direction making an with the direction of propagation of the incident beam is proportional to ; and ( 2 ) the energy scattered every second in all directions together is times the intensity of } incident radiation for electrons all of wlJich scatter effectively , and it is independent of their arrangement .
With the usual values , , the coefficient is equal to per electron .
If we assume that the hydrogen atom has a mass grm. and contains one mobile electron , we find that the mass scattering coefficient of hydrogen is .
If furthel ' we assume that the number of electrons in atoms other than the hydrogen atom is one-half of atomic weight , we find that the mass scattering coefficients of elements other than hydrogen are all equal to 3 .
Recent experiments contradict this theory of scattering in three respects : ( 1 ) although the distribution of scattered radiation actually observed agrees with the formula for directions nearly to the beam , the amount of energy scattered in the forward direction exceeds , and in the backward direction falls short of that given by the formula , so that there is considerable asymmetry fore and aft , increasing with the hardness of the incident rays ; ( 2 ) the atomic scattering coefficient is not equal to for all elements other than hydrogen and for all rays , but diminishes to a very much smaller value for the -rays for aluminium , iron and lead ( Ishino ) ; and ( 3 ) scattering always appears to be accompanied by a change of type to a softening of the radiation ( Florance ) , but whether this effect is to an admixture of softer fluorescent radiation or not , does not appear to be determined .
4 .
view of this contradiction between Sir J. J. Thomson 's theory and experiment I have endeavoured to develop a more complete theory of the scattering of-and -rays , } ) lacing the simple pulse theory adopted in his The of X- by Rings of Electrons .
397 first assumption by the wave theory , which supe ] sede it , but his last two assumptions .
The third assumption requires us to assume that the frequency of the incident radiation is large compared with the frequencies of the free vibrations of the mobile electrons in the atom ; the limits , within which this assumptiou satished with a ) roximation , can only be determined exactly by experiment in our present state of as to the structure of the atonl , but rough estimate be got for rings of electrons by the angular velocity in place of the , since it is of the order of magnitude according to the various theories of atomic structure proposed hitherto .
In addition to these three assumptions , we shall assume , for the rpose of the present investigation , that ( 4 ) the mobile electrons of the atom ) are arranged in coaxal rings of equidistant electrons , which are rotating with uniform angular velocities , usually different for differe1lt , as a particular case , one or more of the may be at .
and the of each electron is so small compared with of light that the effect of the magnetic forces of the waves , as well as the reaction due ) radiation from the electron and the variation 01 its with speed , may be neglected .
The fourth assumption is needed to ensure the permanence of the in spite of radiation , and is justified by the experiments of O. W. Richardson and of Einsteiu and de Haas on the gyrostatic ntomentnm of magnetised substances .
The fifth is made for the Qake of simplicity , to avoid confusing .
the main lines of the tbeory with unnecessary detail , but there is some justification for it in the fact that theories of atomic structure , such as that of Bohr , assign values to , the ratio of the velocity of electron to that of light , which are only of the order for ogen and other light atoms ; it may possibly cease to afford a sufficient approximation for heavy atoms in which the variation of mass with speed has to be into account in calculations resl ) ecting the determination of their constitution from their -ray ( Vegard ) .
5 .
In accordance with our first assnlnption , shall the incident X- or -radiation as consisting of undamped trains of plane unpolarised waves ; it he stated at once that the theory based on this ) remaining assumptions above is one of diffraction by the electrons , coupled together kinematically in a ring , and as it were , electronic grating .
For bydrogen it leads to the same result as the simple pulse theory ; for atoms containing rings of more than one ] ectron it fore and aft asymmetry of scattered radiation and a scattering coefficient , which assumes the simple pulse theory 's value for waves of infinitely high frequency , but increases , as the frequency diminishes , series Prof. G. A. Schott .
more and more accentuated minin ) and lnaxima to a value not much less than times the simple pulse theory 's value , where is the number of electrons in the ring ' .
Thus the theory does not account for the exceptionally small scattering by lshino , nor for the softening of the rays found by Florance ; but there is to that both of these effects may be explained on the assumption X- and -rays consist of damped wavetrains .
modification of the theory fated at present .
SpcciJication of the Radiation .
6 .
thu of generality , shall assunle that incident radiation consists of a nber of independent simultaneous , or , alternatively , successive trains of undaJnped , elliptically polarised , parallel , plane wave distributed in such a wily in ) , or in time , ) the case may be , as to produce on whole ] ) arised rtion .
constiment wavetrain is chal.actelised by five The frequency , all .
( 2 ) epoch , ) so the phase is for the wave which tin happens to pass through ) centre of the electro ] ) ling .
We shall assume that is on the distributed miforlnly over the whole from to ) and ( 4 ) The es , vibration lipse .
The polarisation azimuth , 1\ldquo ; measured in the wave front some convenient initial line to the .
We shall suppose that also is on the average distributed uniformly oyer the whole range from to .
The assumptions the distri btion of and are obviousl needed to ensure the average unpolal.ised character of the radiation as a whole .
The electric gnetic fULCCS at the centre the electron time have components , in the directions of the semiaxes OA , OB of the vibration ellipse , which are given by the equations where the summation is to ) taken for all possible sets of values of and The average energy flux , I , is the of the several component flnxes on account of the independence of the curTesponding wavetrains and the average distribution of their epochs and polarisation azimuths .
Hence we have by 's theorem ( 2 ) where the bar denotes time averaging , and is the velocity of light .
The Scattering of X- -Rays by Rings 399 fication of th .
Let the radius of the ring be , the nulllber of equidistant electrons ?
their charge e.s.u. , their mass and their uniform angular velocity .
Also let ; then according to assumption ( 5 ) is so small th may be treated as constant and used to represent the transverse or the longitudinal mass a pleasure .
The relative positions of the incident and scattered wavetl.ains and the electron are show1l in represents the surface of the sphere whose centre is the centle of the electron ring , and whose radius is unity .
FIG. 1 .
The great circle .
lnarks the plane 01 the ' the direction of revolution being from to ) its axis .
marks the radius vector to the electron at time ; we take the arc to be equal to where is any integel from 0 to The great circle has been to , which marks the direction of fation of the wavetrains of the incident radiation ; marks that vave front of one of the trains , which happens to pass through the of the ring at time , and nark the semi-axes OA , OI3 of the vibration ellipse , so that marks the direction in which scattering takes place ; the great circle MN scattered wave front which passes through the centre of electron ring at , and narks the external bisector of the WOU \mdash ; it is required for purposes of calculation later .
EP is an arc of the great circle drawn through perpendicular to the incident wave front ; similarly EQ is an arc of the great circle UEQ drawn perpendicular to the scattered wave front MN .
Prof. G. A. Schott .
8 .
We shall use the following notation:\mdash ; LNM , .
( 3 ) We shall require the perpendicular distances , and , of the electron from the incident and scattered wave fronts and MN respectively ; these will he reckoned positive when the electron is in front of the respective wave front .
Thus we obtain by lleans of fig. 1 Ey QE si EN LNM .
( 4 ) We shall require the value of below ; we find from ( 4 ) siu } .
( 5 ) From ( 3 ) and fig. 1 we obtain since we have Hence we find from ] ( 5 ) , .
( 6 ) This result will be required presently .
The the Scattered Waves .
9 .
In order to find the scattered radiation we must determine the electromagnetic field at a distant point due to the disturbance of the electron generated by the incident waves .
Let field point be in the direction of the unit vector at a reat distance from the of the electron ; its distance from the electron is equal to very approximately .
Also let denote the acceleration of the electron ; then we know that to the same approximation the electric and magnetic forces at the field point at time are given by f-(Uf ) .
( 7 ) The Scattering of X- and by Rings of Electrons .
401 Terms of the relative order have been neglecled in calculating inten- sities , but they must be letained in estimating phases , so that the function inside the bracket in the expression for must be taken for the time , not for the time , in order to calculate the diffraction elfect .
10 .
In order to calculate the acceleration of the electron due to the lncident waves we must bear !
mind our assuDlptio1ls ( 3 ) , S2 , and ( 5 ) , S4 .
In virtue of ( 3 ) we need only retain the effective force due to the incident on the left of the vector equation of motion , the cfiective force due to steady revolution as well as the controlling forces due to the electrons of the ring and the residue of the atom .
In virtue of ( 5 ) we may on the left take the mass of the electron to be the constant and on the right ueglect the elfect of the crnetic forces of the incident waves .
Denoting the electric force of the incident wave at the position of the electron at the time of nission by for the moment , we may write .
( 8 ) In writing down this equation we have neglected the effect of the disturbance on the mass of the disturbed electron and have used the ?
mdisturbed instead of the disturbed position of the electron at the time of emission in calculating the electric force which generates the disturbance .
The error is of the order of the squal.e of the disturbing force in each case ; we shall errors of this order throughout for the sake of simplicity .
11 .
We must now find the value of the disturbi1lg electric force at the undisturbed position of the electron at the time of emission .
We commence by its components at the centre of the electron ring at the time by means of ( 1 ) , S6 , and fig. 1 , S7 .
Resolving the electric force in the directionls OK , and O we obtain ( 9 ) where the conjugate terms have been omitted for the sake of brevity .
Clearly ( 9 ) give the components of the electric force at the uudisturbed position of the electron at the time , provided that we replace in ( 9 ) by , because every incident wave front reaches at a time later by the amount , where iven by ( 4 ) , S8 , for the time Also ( S ) , S10 , gives the corresponding acceleration at time .
Hence Prof. G. A. Schott .
find that the acceleration due to the disturbance of the electron at time is given by si1l si1l si1l si1l si1l si1l si1l si1l si1l si1l with ) .
In calculating the electric and magnetic forces of scattered wave by means of ( 7 ) , S9 , we must not take the components of for the time when the wave arrives at the field point , but the time when it is emitted by the scattering electron , so that we must replace in ( 10 ) , S11 , by , and calculate by ( 4 ) , .
S8 , with in place of in place of ( 10 ) we must use the following equations:\mdash ; 13 .
At this a difficulty requi1es notice , which arises from the occurrence of the term in the argument of the cosine in .
With the usual notation of Bessel Functions , bearing in mind that is equal to we Inay write part of parG of ( siu ) ving to the smallness of we ) all Bessel Functions except and replace this by unit ) ; this amounts to neglecting the term in the argument of because it is of the order Obviously we cannot do this with the term in the argument of of ( 11 ) , because is assumed to be large in comparison with , but The Scattering of X- and -Rays by Rings of lectrons .
403 we may , S8 , in place of ( 4 ) , with in place of , so that we may write where ( 12 ) This equation , ether with the first three equations ( 11 ) , S8 , determines the values of the components of to be substituted in ( 7 ) , S9 .
14 .
By means of the well-known Bessel Function expansion we obtain from ( 12 ) , S13 , .
( 13 ) Substituting from ( 13 ) in ( 11 ) , S12 , we find , a .
( 14 ) Since the components of the unit vector , and respectively , whilst in fig. 1 , S7 , and obtain from ( 14 ) These explessions , ( 14 ) and 15 ) , when they are rled in ( 7 ) , S9 , the electric ynetic forces at the distant point due to the aves scattered by the electron .
To obtain those due to the whole we ltllst sum the resulting expressions for all values of from , but since the forces are linear functions the quantities and ( Uf ) ; obtain the result by first performing the summation on the } ) ressions ( and ( 15 ) , and then substituting the sums in ( .
We notice that the imen terms given in ( 14 ) and ( 15 ) involve the factor ; the with spect to from to of each such factor vanishes identically for all values of , with the exception of those of the form , whers is an , and in these cases the exponential reduces to unity for every value of i , and the sum to .
Thus the required sums are obtained front ( 14 ) and ( 15 ) Prof. G. A. Schott .
changing into and multiplying each expression by .
It is hardly worth while to write down their actual expressions , as the indications just given suffice for our purpose .
Th Radiation .
15 .
To find the amount of the radiation scattered by the ring in the direction of the unit vector , we apply Poynting 's theorem .
From ( 7 ) , S9 , we obtain for the Pointing flux due to the whole , ( 16 ) where the sunnn ) ation indicated by the symbol is with respect to and is performed in the manner indicated at the end of S14 .
The radiation scattered per unit solid angle in the direction , to be denoted by the symbol in future , is obtained by multiplying the coefficient of in the expression ( 16 ) for by and the result with respect to the time and the ation azimuth .
In the time averaging all squares and products of exponentials vanish identically with the sole exception of products of conjugate exponentials , so that the epochs and , as well as the angle disappear in the process .
In averaging with respect to the product vanishes , and the squares and reduce each to one-half on account of the unpolarised character of the incident radiation as a whole .
Bearing in mind what was Qaid at the end of S14 as to the sums with respect to of and ( Uf ) , we obtain by means of ( 14 ) , ( 15 ) , and ( 16 ) , in the first instance , .
The summation is for all values of and , since the remaining parameters of the incident radiation , , and , have all disappeared ; thus enters into ( 17 ) as a factor , and by ( 2 ) , S6 , its value is Moreover , we see from , S7 , that WU Hence ( 17 ) siu .
( 18 ) 16 .
The infinite summation with respect to can be replaced by a finite one with respect to , for we have the well-known equation The Scatbering of X- -Rays by Rings of Electrons .
405 Hence we obtain in succession by }Jeans of Dirichlet 's discontinuous integral siu .
Substituting this result in ( 18 ) , separating the term from the rest and replacing the ratio by , we obtain for the scattering coefficient in the direction the expression .
( 19 ) Thus , the scattering coefficient per unit solid is determined by ( 19 ) as a function of two variables only , viz. , , the angle between the incident and scattered , and , the between the axis of the scattering electron ring and the external bisector of the , which is shown in fig. 1 , S7 .
The radius of the ring and the wave-length of the incident vibration only enter into the expression for implicitly through the para- meter , which is equal to by ( 12 ) , S13 .
The form of the result depends essentially on , the number of electrons in the ring .
When there is but one electron , , the sun ] disappears , and ( 19 ) ives .
In this case , and this case alone , the scattering coefficient is independent of , which determines the position of the ring to the incident beam , and is simply proportional to , in oreement with the simple pulse theory .
We have every reason for that this is the cnse of the hydrogen atom .
Thus scattering nents on hydrogen , either under pressure or liquid , with a sufficiently extended range of wave-lengths would enable us to test ( 20 ) , but such expetiments lone could not decide between the sinlple pulse theory and the present one .
17 .
When the tent of the Bessel Function in ( 19 ) , S16 , is zero , that function reduces to unity and obtain .
( 21 ) Prof. G. A. Schott .
This occurs in three cases:\mdash ; ( i ) When vanishes , , when the , of the incident radiation infinitely great in collparisolJ with , the radius of the electron .
This case is never actually realised , because resonance sets in when becomes , and assumption ( 3 ) is violated ; but , owing to the flatness of the first maximum of the Bessel Function , it can be approached very , and we may therefore expect an increase of the scattering coefficient the wave-length increases to values considerably greater than the value given } ) the siml ) pulse theorJ7 , and in fact not much less than times that val ue .
( ii ) When is or , i.e. , when the plane of the electron ects the angle between the incident and scattered beams .
This case conceivably be realised with an of magnetic atoms by applying a strong ynetic field in the direction of the external bisector of the incident and scattered beams ; at any rate , an increased scattering might be obtained , if not the full amount of times the simple pulse theory 's value .
( iii ) When is , in the forward direction of the incident beam , a case which merits special consideration .
and Scattering .
18 .
Putting in ( 21 ) , S17 , we obtain for the forward scattering , ( 22 ) which is exactly / ?
times the simple pulse 's value for every wavelength and all orientations the electron ring .
yain , putting , S16 , we obtain for the backward scattering .
( 23 ) The effect of sum in ( 23 ) is most pronounced when .
This case might be realised approximately by using a magnetic iubstance in a strong netic field perpendicular to the incident beanl , and observing the scattering in a direction as nearly as possible opposite to the direction of tion of the incident beam .
The simplest example is boiven the value , which probably corresponds to the atom ; we obtain from ( 22 ) and ( 23 ) siu ( 24 ) The hrst and lowest minimum of is equal to and is given ] ) , or .
For this value of is less The Scattering of X- and -Rays by ltings of .
407 than 30 per cent. of , so that the asymmetry and aft is considerabIe .
For reater values of it may become still more marked .
Qualitative experiments with magnetic substances in strong magnetic fields of the kind indicated in this and the preceding section wo1Ild enable us to decide between the simple pulse theory the present , but not between the latter and a theory based on the Parson electron , because the ring eIectron gives rise to analogous effects .
Scattering 19 .
In order to calculate the scattering of an assenblage of electron rings , must make it further ssnmption ( : each electron the incident radiation independently of all the other .
Actually the radiation scattered by any one on others is scattered by assumption amounts to neglecting the proportion of radiation which double , triple , in general multiple .
It is obvions that the effect of multip ] will be to the radiationl and partially destroy the asynlmetry produced by a single scattering , and will the denser and } the is .
Its existence be detected experimentally with dillelenG pressures ; its effect made ively h by thin layers of material , by layers of varying thickness and thickness in the ulanl]er .
We shall only consider the pleRt case , the axes ) the ) rings are on the equally in all directions .
the average scattering in the direction luultiply , S 16 , by chance that the lie veen the limits and S , and rate with respect to from 0 to .
Thus we obtain , most simply by ex Funclion and integrating terJn by terln , .
( 25 ) This expression leads to .
results as regards asylnmetry , S16 , , as is to be expected , they are not quite so marked .
direction .
, , the scattering is times the simple pulse value before , but in the backward , the Bessel FiterlIl in ( 23 ) , S18 , is leplaced by is osoillatory ; its and is equal to a to the two-electron ring of hli } catterper cent. of the forward Prof. G. A. Schott .
Scatt of an Irregular Assemblage of through a Finite Sold A ngle .
20 .
We shall calculate the total scattering through the front half of a cone of semivertical angle its axis to the direction of the incident beam , and for this purpose multiply ( 25 ) , S19 , by and from to .
The integration is ) performed by changing the variable from to the total scattering through the cone by , we obtain .
( 26 ) where Adopting the notation of ' B. A. Report , ' 1914 , p. 87 , we write .
( 27 ) Then we write ( 26 ) in the form .
( 28 ) shall consider two nporbant particular cases of this formula .
Scatteriny .
21 .
The total scattering from the is obtained by putting , or , in ( 28 ) , S20 .
In this case we shall omit the argument of as unnecessary and obtain where .
( 29 ) For hydrogen we assume that , so that the sum disappears .
Denoting the corresl)onding value by , we find , ( 30 ) here the numerical values are the same as those used in S2 .
The result is precisely that of the simple pulse theory , so that on both theories the The ttering of X- and -Rays by Rings of Electrons .
409 scattering of hydrogen is independent of the , and its distribu- tion is symmetrical fore and aft .
It is convenient to take hydrogen as the standard to which all other substances are referred .
Atomic and Mass Scattering 22 .
An is assumed to contain coaxal rings of mobile electrons ( S3 , assumption ( 3 ) ) .
We shall neglect the possibility that two rings may have a constant phase relation between them , and so interfere with each other ; this occurs when they have the same angular velocity , a contingency which is extremely unlikely on any of the current theories of atomic structure .
Then the scattering of an atom is the sum of the scatterings of its several electron Let the atomic number be so that , ( 31 ) where the summation is for all the rings .
We obtain from ( 29 ) and ( 30 ) , S21 , ( 32 ) Experimenters often use the llass scattering coefficient , , the scattering per unit mass , in place of , which denotes the scattering .
atom , or atomic scattering coefficient .
We shall denote the mass coefficient by , and that of hydrogen by ; to obtain them we must by the mass of the atom .
Taking the mass of the hydrogen atom to be , we find from ( 30 ) , S21 , .
( 33 ) Since the mass of an atom of atomic bt A is A times the mass of the atom , we obtain and ( 33 ) .
( 34 ) molecules of compound substances we must add the tomic scattering coefficients of its constituents , and divide b , in older to obtain the mass scattering coefficient .
One important result of ( 34 ) should be noticed ; when the of the incident radiation becomes very great , the argument of the function becomes very small , and the function itself becomes very nearly to unity , as we shall see below .
Prof. G. A. Schott .
Hence we find .
( 35 ) The simple pulse theory gives for all wave-lengths , so that the present theory gives a very much larger value for long waves , although it gives the same value for very short waves .
The long wave limit cannot be reached owing to the intervention of resonance , but it can be approached pretty closely , say within 10 or 20 per , long before resonance becomes appreciable , so that measurements of scattering for waves should enable us to decide between the two theories .
Before considering particular examples of the theory , we shall first consider the second particular case of for1nula ( 28 ) , S 20 .
Eg i Scattered .
Florance defines emergent scattered adiation as that which is scattered forwards on the whole , i.e. , between the limits and and returned scattered radiaetion as that which is scattered backwards on the whole , i.e. , between the limits and .
We shall denote the corresponding atomic and nlassscattel.ing coefficients by , and respectively .
Putting in ( 28 ) , S20 , and using ( 30 ) , S21 , we obtain for a singIe ring where and is given by , S21 , as before .
The value of is clearly , but for our purpose it is more convenient to use the difference , which we shall denote by E. From with ( 29 ) and ( 30 ) , S21 , we obtain where ' The Scattering of X- by Rings of Electrons .
411 Proceeding just as before in S22 we obtain for the lnass coefficients of atom of atomic number and atomic weight ' , where is given by ( :34 ) , The coefficient may be taken as an absolute of the aft asymmetry of the atom on the same scale on which ] measures the total scattering ; a relative measure would be aflorded by the ratio , but the variation of this ratio with the wave-length differs so little from of itself that it is hardly worth while to introduce it in addition .
asymmetry is reckoned positive when is positiye , i.e. , when the exceeds the returned scattered radiation , which is ] the case .
umerical Vof the Coefficients .
24 .
For conyenience sake , we collect the principal expressions which are of use for the purpose of comparison with experimental results , viz. , ( 34 ) , S22 , for the mass scattering coefficient , and ( 37 ) and ( 38 ) , S23 , for the asymmetry coefficient .
They are , 0 .
where , 0 .
where , 0 .
where , 0 .
where The functions and are fundamental for the theory , whilst the function is useful as an intermediary for purposes of calculation .
Tables of these functions are }iven at the end of this paper for values of from to 10 , and graphs of and are given in fig. 2 between the same limits .
We see that has a sharp maximum equal to unity when , a low minimum equal to when , a comparatively low maximum equal to nearly when , a very low minimum equal to when , and so on , fluctuating between ninishing positive limits , and ultimately vanishing when becomes infinite]y great .
On the other hand , vanishes when , has a fairly sharp ntaximun } equal to nearly when , a small negative minimum equal to when , a small maximum equal to when , and so Prof. G. A. Schott .
on , fluctuating between diminishing alternately positive and negative limits and ultimately vanishing when becomes infinitely great .
We notice that negative asymmetry occurs for certain ranges of wave-length .
Comparison with Experiment . .
In order to test the correctness of the theory by comparison with results we must work out numerical values for particular arrangements of electrons , especially those that are likely to occur in nature .
From the theoretical point of view the best examples to choose are the simplest , not only because they easily lead to definite numerical results , capable of experimental verification or otherwise , but also because the fundamental assumptions of the theory , given in 5S2 , 4 , 19 and 22 are probably most nearly fulfilled in these cases .
Of course , experimental difficulties may arise in realising the conditions needed for a test , such as the production of X- and -rays of sufficient intensity within the ranges of wave-length corresponding to the rapid changes of occurring near its principal maximum or to the maximum of .
These possible difficulties however do not concern us here , so that we shall choose the comparatively simple cases of helium and lithium as examples .
We shall ignore the possibility that all their mobile electrons are arranged in separate rings , for this arrangement is extremely improbable from the theoretical point of view , and in any case would only give values for the scattering coefficients identical with those for hydrogen and therefore needing no further discussion .
26 .
Helium.\mdash ; We make the usual assumptions .
In ( 39 ) , S24 , there is but one sum , and this reduces to the single term ; hence we have .
( 40 ) There is no need to give graphs , .
obviously those giveu in fig. 2 will serve , if the vertical scale be reduced to one-fifth , and the constant amount be added to give the graph of We see that for very short waves , for which vanishes , approaches the value , one-half that of hydrogen , whilst for long waves , for which is nearly equal to unity , approaches the value , that of hydrogen .
For or is a minimum and equal to , for or , it is a maximum and equal to , so that for wavelengths equal to or less differs from the simple pulse theory 's value by 7S per cent. at most .
On the other hand for , or and differs from its theoretical maximum by only 15 per cent. In order to estimate what these wave-lengths correspond to in absolute measure we must form some idea of the value of , the radius of the electron tering of and -Rays by Rings of Electrons .
413 ring of helium .
If we take Bohr 's theory as a guide and assume that the angular momentum of each electron is , where denotes Planck 's constant , we { ind that AU , , whilst , the ratio of the velocity of the electrons to that of light , is .
Thus for the first minimun ] AU .
, and for the 15 per cent. deficiency below the maximum .
AU .
Although our theory is quite independent of Bohr 's , or any other particular theory of atomic structure , these numbers may serve to erive some idea of suitable values of the wave-length forexperimental tests .
More over , since , the corresponding values of are 122 and 39 respectively ; these are so that appreciable vitiation of the results owing to resonance is not to be feared .
Again , as regards the asymmetry , the maximum value of is and occurs for , or .
This value of corresponds to and by ( 38 ) , S23 , gives and , so that the returned is about 58 per cent. of the scattered radiation .
Thus the greatest asymmetry is in the direction enerally observed and is considerable even for so simple an atom as that of helium .
ative a is also possible , but it is very small , as we see from fig. 2 , Its greatest value corresponds to and , and occurs when , or .
; the correspondin returned is 104 per cent. of the emergent scattered radiation , but the difference is probably too small to be detected by experiment .
27 .
\mdash ; For this element two alternative arrangements of the three mobile electrons in the atom may be suggested : ( 1 ) all three in one ring , ( 2 ) two in one and one by itself in a co-axial ring .
In each case we put ; further we put in ( 1 ) , and for the ring of radius in ( 2 ) , the radius of the other ring being immaterial , as its contribution to the scattering coefficients is independent of the wave-length and radius , as with the single electron ring of hydrogen .
Thus we obtain from ( 39 ) , S24 , .
( 41 ) The graphs given in fig. 2 will again serve , provided that in case ( 1 ) we reduce the vertical scale to of its value and add the constant amount to give , whilst in case ( 2 ) we reduce the vertical scale to of its value and add the same constant amount to give We see that for very short waves , and in both cases , in agreement with the simple pulse theory .
But for long waves approaches the maximum value , considerably greater than the value of for Prof G. A. Schott .
hydrogen , whilst only approaches the maximum value , much less than that for hydrogen .
Again , the maximum value of is ives per cent. , and occurs when or , whilst the maximum value of is only per cent. , and occurs for or Thus we see that both the scattering coefficient and maximum symmetry are so much greater for the first ement than for the second that experiment should enable us to ] ecide between them .
In the case of lithium , timates of wave-lengths are rendered uncertain by the failure of Bohr 's theory to provide a satisfactory explanation of the spectrum : nevertheless they are not likely to be so far rong as to be without value .
For the first alternative Bohr 's theory gives , so that the wave-length for maximum asymmetry is The first minimum of is given by , or whilst 15 per cent. deficiency from the maximum corresponds to , or .
The corresponding values of are 140 and 39 , so that resonance can hardly be appreciable .
The second alternative is much more uncertain , for we do not know whether the two rings are in the plane or not , nor do we know which has the greater radius .
Consequently we may content ourselves with the rough approximation obtained by neglecting the action of the outer ring on the inner one , and treating the latter as part of the nucleus in estimating its action on the former .
Then the radius of the outer ring will be underestimated , that of the inner overestimated , because the repulsion between them is underestimated .
In subcase , with the ring of two electlons outside , this ring is prac ically the helium ring , so that we have U. , , as in S26 .
The maximum asymmetry is given by , the first minimum of scattering by , and the 15 per cent. deficiency below maximum scattering by In subcase ( b ) , with the two-electron ring inside and practically subject to the attraction of three positive charges , we find that .
The maximum asymmetry is given by the first minimum of scattering by AU .
, , and the 15 per cent. deficiency below maximum scattering by for lithium the of wave-lengths may be regarded as extending from .
to , and the values of are so great that appreciable vitiation by resonance is hardly likely .
The Scattering of .
415 28 .
The examples of helium and lithium , discussed in SS26 and 2 are sufficient to indicate how expenments on the scattering .
and -rays may be used with the twofold object , first , of testing the correctness of the theor ) of scattering here developed , and with it the truth the ental principles of the theory on which it is based , and , secondly , of determining the arrangement of the electrons in the atom , if and when the first test has proved satisfactory .
It is important to bear in mind that the estimates of iven above are only intended to serve as rough guides to the experimenter ; their correctness , or otherwise , has no imp ortauce for the theory of scattering as such , but can only serve as a test of Bollr 's theory of atomic structure .
If the theory of scatte1ing survives the test of experiulents on the simplest atoms , such as those of hydl.ogen and helium , further measurelnents of the scattering by more complex atoms , such as those of lithim , carbon , and so on , will enable us to obtain information respecting the arrangement of the electrons in these atoms .
The character of the variation of scattering and with the wave-length upon the number of rings and the number of electrons in each ring , whilst the position of the maxima and minima depends upon the radii of the Very few experimental results are available for comparison as yet , and no absolute measuremen ts at all on hydrogen and helium , which are so necessar .
for a test of the theory .
iven a very valuable series of measurements relative to water , from which it appears that the atomic scattering coefficient of lithium is about four times that of hydrogen for This gives , with an error which may be as much as 10 or 20 per cent. , because of the uncertaint , in Auren 's value for the coefficient of hydrogen ; this standard number has to be calculated from water and carbon compounds and necessarily appears as the difference between nearly equal numbers .
The value of is much greater than the simple pulse theory 's value , , and is well within the limits given in S27 , but it is too unreliable to furnish a reliable estimate for or to decide between the two angements discussed there .
29 .
We must now consider certain experimental results obtained by for very hard -rays scattered by aluminium , iron , and lead , bearing in mind that our theory can only be applied to atoms as complex as these with a very rough approximation to the comparatively values of that are to be expected for some of their rings .
In the first place , Ishino and Madsen before him , found very ooreaC asymmetry coupled with a very small coefficient , viz. , a ratio Prof. G. A. Schott .
equal to and a mass scattering coefficient of the order for alurninium and for lead .
Thus we obtain for aluminium , and for lead , so that is about three-quarters of for each element .
The asymmetry is greater than any we have discussed , but is not so great that it cannot be accounted for by the presence of rings of many electrons .
In S27 we found that a ring of three electrons gave a value of one-half that of , so that a ring of seven electrons , such as Vegard assumes for aluminium , may easily give the lequisite asymmetry .
But the smallness of the mass scattering coefficient remains unexplained by the present theory , which gives the simple pulse theory 's value as a limit .
The exceptionally small scattering coefficients observed by Ishino are , however , accompanied by a change of type , a softening of the rays , pointing to an emission of rays accompanying the scattered rays , but of greater wave-length , a fluorescence in other words .
May not these be due to vibrations of the electrons , generated by the incident radiation and accompanying the forced vibrations to scattering is due ?
There is reason for supposing that some of the free vibrations of the electron are but slightly damped and therefore comparatively persistent , otherwise interference with long path differences would be impossible .
Moreover , we know from ordinary dynamics that the initiation of a forced vibration is on the average accompanied by the production of a free vibration , but when the incident vibration is undamped , only a small fraction of its energy is used in generating free vibrations on the average .
When , however , the incident vibration is strongly damped , a large fraction of its energy may be used in this way , and a correspondingly smaller fraction scattered .
This problem is too long to be discussed here , but is under investigation at present .
Electron .
30 .
Just after the completion of the preceding part of this investigation I became acquainted with the full text of the paper on " " The Size and Shape of the Electron by A. H. Compton , which he considers the scattering of the X- and -rays by the flexible ring electron amongst other topics .
Evidently this electron .
may be regarded as the limit of an electron ring when the number of the electrons becomes infinitely great , whilst the total charge and mass remain finite .
Hence we should obtain expressions for the scattering coefficients of the ring electron as limiting values of our expressions for the electron ring , especially ( 25 ) , S19 , and ( 29 ) , S21 .
For this purpose we suppose to become large and replace by by by , and by .
The summations with respect to * Phys. Rev vol. 14 , p. 20 ( 1919 ) .
The ttering of X- -Rnys by Rings of lectrons .
417 i are taken most conveniently from , i.e. , including the first term , previously separated from the rest , and become integrations with respect to from to Thus we obtain from 25 ) , S19 , ( 42 ) where as before .
31 .
The series represents Compton 's ratio , and can be compared with his series ( 20 ) where is used instead of our for the ladius of the ring .
The form of the series is the same , but the coefficients are diff'erent , on account of a difference in the of for the varyig tions of the in an assemblage .
In our fig. 1 , S7 , the pole of the rin \ldquo ; is fixed by its angular distance , or , from the fixed point and by the angle , or , between the great circle and the fixed great circle ; and we have taken the probability of the orientation to be .
On the other .
hand , Compton the position of the ring by the angular distance TV , or , from the fixed point of the intersection of the great circle ( the plane of the ring ) with the fixed great circle and by the angle , or , between the two great circles ; and he takes the probability of the orientation to be .
It is easy to prove from the spherical triangle that The difference amounts to this : with us the poles of the rings are distributed uniformly over the unit sphere , whilst with Compton the point is distributed uniformly over the great circle , and at the same time the plane of the ring is distributed uniformly about the nodal line OT .
The difference is not , however , of fundamental importance .
32 .
To obtain , the whole scattering forwards through a cone of semivertical angle , we multiply ( 42 ) , S30 , by , or , where S , and integrate as before .
Thus we find The total scattering is given by , or ; we obtain .
( 44 ) , p. 42 .
Prof. G. A. ) This be with Compton 's which has the same form as ( 44 ) , but diffel.ent coefficients for the reason above assigned .
As guard y , we obtain Florance 's emergent scattered radiation scattered in the forward direction , or in ( 43 ) ; thus we have The coetficients in the series ( 44 ) and ( 45 ) are arranged so that first term is unity in each hen coefficient of in ( 45 ) is times that in ( 44 ) .
It that the ratio ) , which is one-half when the frequency vanishes , or , increases with to a maximum , and then robably alternates etween ever-narrowing limits as the successive of the predominate in turn , in somewhat the same manner as the mass coefficients in .
The series ( 44 ) and ( 45 ) are convenient when is small , or but for large values of , or small values of , the computation becomes very laborious owing to the reat number of terms required .
In this case it is better to use ( 29 ) , S ) , and ( 36 ) S 23 , or their equivalents ( 32 ) , S 22 , and , S 23 .
We obtain , to the limit for , as in S 30 , , ( 46 ) where as before .
is , of course , the total , coefficient of the ring electron as before , whilst is its asymmetry coefficient , defined as in S 23 , and equal to the excess of the " " emergent scattered radiation , over the " " returned scattered radiation , each reckoned per unit intensity of the incident radiation .
A graphic method of computation could be based on the second type of by making use of the graphs of the functions and CJiven in fig. 2 , but the vanishing factor would cause difficulties near the limit .
On the whole , the besb method is based on the first type of integral : for a given value of , or , the argument is calculated .
a number of equidistant values of from to ; correvalues of and are found from the Tables at the end of this paper , and then one of the usual summation formulae is applied .
When lioc .
, p. 42 .
The Scattering of X- Rings of Electrons .
419 the interval of is , where is an even this ounts to approximately the same thing as calculatin .
, or , as the case ulay be , for a of electrons and diyiding the result by .
The greater is , the better the approxinlation will be .
Put briefly we sa.y : The scattering , or asymmetry , of the ring electron of radius and mass , is the limit , when ] is infinitely freat , of scattering , or asymmetry , of a ring , of radius , of equidistant electrons of the usual type , each of charge and mass 34 .
The results of most importance for a comparison of the three theories discussed in the preceding pages may be summarised in the following Table , where the total ; is for convenience sake expressed in ternls or , the total scattering of the atom 011 the sinlple pulse theory , as a unit .
denotes the atomic number , taken to ] the total number of mobile electrons in the atom , whether rin , ( electrons or ) of the usual type .
For the electron ring theory , as , denotes the number of electrons of the usual type in any one the sign of summation refers to the several rings , so that Ring eectronsi 1ectron R The corresponding values of the mass scattering coefficients relative to hydrogen are obtained from those the relative atomic .
coefficients given in the second and third columns by divicing by , the atomic of the element considered .
The Table enables us to indicate experiments which might be clucial in between the several theories .
For example , for hydrogen we take ; according to the simple pulse and electron ring theories the total scattering has the same value unity for wave-lengths , and there is no asymmetry , but according to the ring electron theory the total scattering only has the value unity forlong waves and diminishes to zero for short whilst the maximum asymmetry is as large as that found Ishino for aluminium , and lead , but probably occurs at a longer If the scattering from hydrogen , whether gas under pressure or caIl be observed at all , even quite rough measurements should be decisive for or against the ring electron theory , but could not decide between bhe other two theories .
Prof. G. A. Schott .
Again , for helium we may put ; according to the pulse .
theory the relative total scattering is 2 for all wave-lengths , and there is no. asymmetry ; according to the electron ring theory the relative total scatter- in is 4 for very long waves and diminishes to 2 for very short ones , and there appreciable asymmetry , but much less than that found by Ishino for aluminium and the other metals ; lastly , according to the ring electron .
theory , the relative total is 2 for very long waves and diminishes to zero for very short ones , and there is large asymmetry , comparable with that foumd by Ishino for aluminium .
Similar results hold for other elements : speaking generally , the simple pulse theory gives relative total scattering equal to the atomic number independently of the wave-length , and no asymmetry ; the electron ring theory gives exceptionally large relative scattering for very long waves diminishing to the simple pulse theory 's value as a lower limit for very short waves ; together with appreciable , but relativelv small , asymmetry for light atoms and large maximum asymmetry for heavy ones ; and lastly , .
the ring ] ectron theory gives the simple pulse theory 's value as an upper limit of the relative total scattering for very long waves and values dimin- .
ishing to zero for very short waves , together with large nlaximum asymmetry of the same order for all elements , whether light or heavy .
35 .
The experimental data available for a comparison are few and contradictory .
The relative measurements of Aure'n on the total scattering for wave-lengths .
between and give values much greater than the simple pulse theory 's value even for elements , e.g. , instead of for lithium , already mentioned in S28 .
Auren 's values include absorption and therefore are subject to vitiation by resonance ; but no characteristic radiation of lithium has been observed near his range of wave-lengths , although the radiation of carbon , originally discovered by Barkla , is not far away .
Since is the upper limit of the total scattering of lithium according to the ring electron theory , this result , as well as others of Auren , is against that theory , but it is not decisive because the possibility of vitiation by resonance is not excluded .
On the other hand , as we saw in SS29 and 35 , Ishino found values of the total scattering of aluminium , iron and lead for hard -rays of a wave-length about , which were only one-quarter of those to be expected according to the simple pulse theory and also the electron theory for very short waves .
Compton shows that values of this order are to be expected for a ring electron of radius .
; they cannot be explained at all by the simple pulse theory , nor by the electron theory The Scattering of X- and by Rings of Electrons .
above for light elements , so long as the -rays are regarded as undamped .
But two possible explanations may be suggested .
In the first place aluminium , iron , and lead are not light elements in the sense of the present investigation , in which the effect of the speed of the electrons has been neglected .
We have some reason for supposing that some of their mobile electrons move with speeds quite appreciable compared with that of light , so that their mass is greater than or , that of a slowly moving electron , and the effect of the magnetic field of the incident radiation also must be taken into account .
The effect of increase of mass is to diminish the total scattering coefficient ; that of the magnetic field requires investigation .
Secondly , damping of the incident radiation may be expected to diminish the scattering for the reasons already given in S29 , but the investigation alluded to there , so far as it has gone , shows that it can have no effect in the case of a ring of one electron ( hydrogen ) .
It may be urged that damping is reconcilable with the production of fine spectrum lines , but it should be borne in mind that the resolving power of -ray spectrographs is far below that of optical instruments , and the phase differences reached are correspondingly smaller , and this seems to be still more true for -rays , such as those used by Ishino .
Thus an amount of damping would be possible for hard -rays , which would be quite out of the question for light .
Moreover , it is suggestive that the deficiencies of the total scattering below the simple pulse theory 's value observed for carbon and aluminium by Barkla and White*are very much smaller than those observed by Ishino , commence at wave-lengths from to , and increase as the wave-length diminishes .
In any case until the investigations on the effects of damping of the incident radiation and of the speed of the electrons have been completed , Ishino 's experiments cannot be regarded as decisive against the electron ring theory , although they are strongly in favour of the ring electron theory .
[ Nof , added December 4.\mdash ; The damping investigation has been completed and gives the following results : values of less than damping diminishes the scattering function , but for greater values it increases it , but the changes are very snlall for any admissible amount of damping .
Large damping almost certainly dinlinishes the scattering to the simple pulse theory 's value , but not below .
Hence cannot explain Ishino 's low values .
The effect electron speeds comparable with that of light is under consideration .
] 36 .
In order that experiments on scattering should lead to definite theoretical results , the following conditions ought to be approximated to as * Phil. Mag Ser. 6 , vol. 34 , p. 270 ( ]917 ) .
Prof. G. A. Schott .
Scales for .\mdash ; Coefficients of He and Li .
nearly as may be consistent with the use of an incident radiation of sufficient intensity to enable measurements of reasonable accuracy to be made : ( 1 ) the incident radiation should be monochromatic ; ( 2 ) the range of The Scattering of -Rays by Rings of Electrons .
423 wave-lengths should be wide ; ( 3 ) the scattered radiation should be sifted from any acco1npanying resonance or fluorescence radiation of different wavelengGhs ; ( 4 ) ) ination of light elements with tvier impnrities should be avoided on account of the disproportionate effect of the heavier elements on the scattering for the longer waves .
Experiments on these lines on the absolute scattering in various directions of and -rays by hydrogen , as well as other light elements , either absolute , or relative to hydrogen , for suitable ranges of wave-lengths , may be expected to give valuable information respecting the nature of ) electron , the arrangement of the electrons in light atoms , and the nature of and -rays , Tables of the Functions , , .
In conclusion , I wish to thank Prof. Sir J. Larmor for his interest in the work , and his valuable criticisms and suggestions .

