]\gt ; [ 175 ] VI .
The Scattering of Plane Electric Waves by Spheres .
T. J. .
BROMWICH , Sc. D. , F.R. Received April \mdash ; Read November 23 , 1916 .
INTRODUCTORY NOTE .
THE problem which gives its title to the present paper has been handled by various writers , notably by Lord RAYLEIGH , Sir J. J. THOMSON , and Prof. LOVE .
In most cases the solutions have been expressed in a Cartesian form ; but it appears to me that a marked simplification is introduced by using spherical polar co-ordinates .
The preliminary analysis becomes shorter , and the conclusions are easier to interpret ; in fact , the analysis is nearly as simple as in the analogous problem of electrostatics , when an electric field is disturbed by the presence of a dielectric sphere .
To obtain the requisite solutions a new general solution of the electromagnetic equations in Cartesian form is given in S1 , and is then transformed to the spherical polar form ; SS2 , 3 contain a summary of certain analytical results required in the sequel .
S4 contains the general solution of the problem of finding the scattered waves when a plane simple harmonic wave strikes a sphere ; and in S5 the solution is applied to the case of a small sphere .
These formulae ( all of S4 and part of S5 ) were originally worked out in 1899 , but publication was postponed in the hope of completing the problem of the large sphere .
In S6 the problem of a large sphere is considered by applying to the formulae of S4 a method of approximation devised by Prof. H. M. MACDONALD* for dealing with waves incide11t from a Hertzian oscillator on a conducting sphere .
The formulae of S6 were worked out early in 1910 and were given in my University lectures at Cambridge in that year .
At the same time I succeeded in obtaining a different treatment ( given in S7 below ) which confirmed the other results , and gave an easier process for dealing with * Phil. Trans. Roy .
Soc , vol. 210 , 1910 , p. 113 .
Prof. MACDONALD tells me that he worked out ( at about the same results in reference to the problem of S6 ; but these have not been published .
An alternative solution was obtained by Prof. J. W. NICHOLSON at about the same time ; his solution starts from Sir J. J. TIIONSON'S formulae .
Prof. NICHOLSON'S results originally differed from those of S6 ; but on revision agreement was obtained Proc. Lond. Math. Soc vol. 9 , 1910 , p. 67 ; vol. 11 , 1912 , p. 277 ) .
von .
ccxx .
[ Published February 2 , 1920 .
points behind the sphere .
The method of this section is similar in some respects to one used by Prof. MACDONALD in a later paper .
* The formulae of SS6 , 7 have been delayed in publication for two reasons : in the first place I wished to obtain some confirmation from direct numerical calculation .
This has 1low been carried out by Messrs. PROUDMAN , DOODSON and KENNEDY ( of Liverpool University ) .
It appears that the agreement with the formulae of S6 is quite close from to , and for the -component up to about .
The formulae of S7 also give good results in a cone of about behind the sphere that from .
It is clear , however , that an approximation suitable from to 17 ( for Y ) and from to 17 ( for Z ) has still to be obtained .
But nevertheless the present approximations proved a valuable auxiliary in checking and testing the numerical work .
[ The paper in its original form was presented to the Society on Apri113 , 1916 ; owing to the difficulties in regard to labour and paper during the war , I was asked to condense the introductory matter of SS1-3 .
This proved to be impossible until now , on account of pressure of war-work of kinds .
In the present version S1 has been re-written so as to reduce its bulk ; in SS2 , 3 certain formulae have been omitted which were not used in the applications of SS4-6 .
In re-arranging the paper it proved convenient also to number the formulae differently .
The decimal system has now been adopted ; here the figure before the decimal point indicates the section of the paper in which the formula occurs .
The figures folowing the decimal point are to be regarded as following the same order as ordinary decimal fractions .
Thus and fall between and , and all these formulae occur in S5.\mdash ; Added March 18 , 1919 .
] S1 .
A GENERAL SOLUTION 0F THE FUNDAMENTAL ELEGPROMAGNETIC The fundamental equations of electromagnetic waves may be written ( E ) ( M ) 'Phil .
Trans. Roy .
Soc , vol. 212 , 1912 , p. 299 .
The two methods are not identical ; but they appear to yield equivalent results in all the cases to which they have been applied .
'Phil .
Trans. Roy .
Soc , vol. 217 , 1917 , p. 279 .
The calculation was originally undertaken by Dr. PROUDMAN in consequence of a suggestion made in my lectures of 1912 ; the work , however , proved to be longer than had been anticipated and was completed by Messrs. DOODSON and KENNnDY .
See the paper last quoted , p. 292 et seq. S Revised March 18 , 1919 ; see note at the end of the introductory remarks above .
SCATTERING OF PLANE ELECTRIC WAVES BY SPHERES .
Here denotes the electric force , the magnetic force , is the dielectric constant , is the magnetic permeability , and the axes of reference are a Cartesian right-handed system .
The units adopted are those of the electromagnetic system , and is the fundamental constant generally identified with he velocity of radiation in free space ; the equations are those derived from AMPERE'S law , and the equations ( M ) are similarly derived from FARADAY'S law , the two together constituting the circuital relations of the electromagnetic field .
It has proved possible to obtain a solution of a very general type , by assuming that ( 11 ) ; then equations ( M ) yield ( 12 ) Substitute from equations in the first equation ( E ) and we obtain ( 13 ) where ' denotes LAPLACE'S operator On comparing equations and they will be seen to be consistent provided that ( 14 ) and that ( 1.5 ) Thus must satisfy the fundamental wave-equation , which is satisfied by any component of electric or magnetic forces or .
For our purpose it is more convenient to express the above solutions in terms of spherical polar co-ordinates ; these are supposed to form a right-handed system , when taken in order , so as to avoid changes of sign in introducing the new co-ordinates .
We write for the components of electric force in the directions of respectively ; and for the components of magnetic force .
Equations then become ( 111 ) * This is done to avoid confusion with the Cartesian components used in equations ( E ) and ( M ) ; but in the sutnequent sections we shall use and for the spherical polar components here denoted by respectively .
DR. T. J. .
BhOMWICH ON TllE As regards the transformation of to , it is sufficient to note that the gradient of has the spherical polar components and that corresponds to the Cartesian vector .
To obtain the formulae corresponding to we observe that the vector on the right is equal to the vector-product of the two vectors , and ; and that these two are represented by and Thus the .
vector-product has the spherical polar components .
Consequently equations now become ( 121 ) while and ( 15 ) give ( 141 ) .
( 151 ) A consideration of these form lae suggests that further simplifications can be obtained by writing ( 16 ) which together satisfy equation ( 141 ) ; and then equation leads to the equation for ( 17 ) SCATTERING OF PLANE ELECTRIC WAVES BY SPHERES .
Substituting from in and we obtain the final expression for the field in terms of ( 18 ) ( 18 ) ( 18 ) ( 18 ) , .
In like mannel we obtain another set of solutions by making an assumption similar to for the components of magnetic force .
This gives the field : ( 19 ) 1 ( 19 ) 1 ( 19 ) 1 ( 19 ) 1 in where V is a second solution of equation .
It can be proved* that gives the most general field in which the radial magnetic force is zero , while gives the most general field in which the radial electric force is zero .
It can also be shown that the field is uniquely determined by the value of and ; and accordingly the most general solution can be obtained by the superposition of and .
S2 .
FURTHER SPECIALIZATION 0F THE SOLUTION 0F S1 .
If we superpose the fields , , and now utilize , to denote the spherical polar components of the field , we have the general ( 21 ) , , * See a paper in the ' Philosophical Magazine , ' July , 1919 ( 6th ser. , vol. 38 ) , p. 143 .
Originally worked out in 1899 , and first published as a question in Part II .
of the 'Mathematical Tripos , ' 1910 .
( 22 ) , where are any two solutions of the equation ( 23 ) A solution of which is sufficiently general for the applications in view may be found by assuming that and V can be expressed as sums of terms of the type .
It is easy to see that then leads to the equation ( 24 ) and since the two sides equation are functions of and of respectively , it is clear that each side must be a mere constant .
If we write the constant in the form , it is evident that must be a surface-harmonic of order Accordingly in problems ( such as those with which we shall be concerned in the sequel ) where the whole of angular space is considered , the value of must be a positive integer ; for ( except when is an integer ) there are no surface-harmonics which are everywhere continuous and single-valued .
Thus we may reduce our solution to the form For , , 1 , 2 , 3 , , where F. is a solution of the equation and .
SCATTERING OF PLANE ELECTRIC BY SPHERES .
The general solution of equation is well known , and it is given F. where the functions and are arbitrary .
In the special case of divergent , the function can be omitted in ; and , if the region considered includes the origin , then , so as to make F. continuous at It will be convenient notice that in consequence of equation the radial components of force can be written in the simpler forms X or .
It will be noticed that we can at once determine the form for or V when the radial forces have been expressed in the form ; this agrees with the general conclusion stated at the end of S1 , that ( in spherical polar co-ordinates ) the remaining components of force are completely determined when the two radial components are known .
S3 .
SPECIAL CASE 0F SIMPLE HARnONIC WAVES AND THE APPROPRIATE FUNCTIONS .
We assume in future that the waves are simple harmonic , of wave-length in free , space ; we can then suppose the time to occur only in the form of a time-factor , with the usual convention that finally only the real ( or the imaginary ) parts of the formulae will be used .
The functions occurring in equation above are then exponentials of the types and where is given by , or .
Thus ( if we now suppress the time-factor ) the functions given by are of the types ( 31 ) We shall be concerned with two special types only : divergent waves ; waves which are continuous at .
The former of these corresponds to the * See , for instance .
LAMB'S 'Hydrodynamics , ' 1906 , art .
295 ; an alternative method of solution given in S3 of my paper in the ' Philosophical Magazine , ' quoted on p. 179 above ; compare also A. E. H. LOVE ( ' Phil. Trans. Roy .
Soc , vol. 197 , 1901 , pp. 9 , first expression in ; while the latter is found by combining the so as to yield ( 32 ) Using the notation explained in and below , the standard functions are for divergent and for waves within a spherical boundary .
, for waves inside a spherical boundary , and can now be replaced by the forms ( 33 ) oror , for divergent waves function S must be replaced by .
Definitions and Properties of the Two Functions , .
We write for brevity ( 34 ) ' In terms of the known Bessel function we can write ( 341 ) , and accordingly the function is the same as that denoted by in one of papers .
* In the notation adopted by LAMB , and those writers who have used LAMB'S solutions as the fundamental forms , we have the identity ( 342 ) .
'Phil .
Trans. Roy .
Soc , vol. 210 , 1910 , p. 113 .
See in particular p. 116 .
'Hydrodynamics , ' 1906 , Art .
287 .
SCATTERI OF PLANE WAVES BY SPHERES . .
183 Similarly , we write ( 35 ) , where ( 36 ) C. In terms of the function ( the modified Bessel function used by MACDONALD ) .
we have the relation ( 351 ) Thus in terms of the notation used by MACDONALD in the paper last quoted , and in LAMB'S notation .
In consequence of the equation we see that both and are solutions of the differential equation ( 38 ) The functions , and have been tabulated from to 10 , and for values of ranging from to 22 , by Mr. DOODSON , these tables have formed the basis of the numerical calculations mentioned on p. 176 above .
It will be convenient to collect here the simple relations amongst the functions , which correspond to the known results for Bessel functions , or to those given by LAMB for the equivalent function .
Difference Functions From ( 3 4 ) we see that ( 3 and by using ( 3 5 ) we see that the same relation holds for Again , it will be found that British Association Report , ' 1914 .
PROUDMAN , DOODSON and KENNEDY , ' Phil. Trans. Roy .
Soc , vol. 217 , 1917 , p. 279 .
VOL. CCXX.\mdash ; A. 2 and using this in equation ( 3 4 ) we deduce that ( 3 82 ) Combining ( 3 81 ) nnd ( 3 82 ) we have also The relations , ( 3 83 ) hold equally for En ( z ) and , as may be seen from and .
As , are independent solutions of the equation ( 3 , it is evident that const .
Now when is small , it is easy to verify from ( 3 4 ) and ( 3 6 ) that , and accordingly we have ( 3 84 ) In the discussions of S6 , when are both large , it will be convenient to adopt the following notation:\mdash ; , so that Substituting from in we deduce that ( 386 ) Before leaving these preliminary formulae it will be convenient to quote the formula for in terms of our standard fun.ctions ; nameIy , where and ) is LEGENDRE'S polynomial of order This result follows at once from the formula given in LAMB'S ' Hydrodynamioe , ' Art .
291 , on using the relation between and , already quoted .
It is of course evident that an expansion of the type might be anticipated , since each side satisfies the wave-equation , is symmetrical about the axis of , and is continuous at ; the determination of the numerical coefficients may be carried out quickly by comparing the terms in on the two sides of the equation .
S4 .
PLANE ELECTROnAGNETIC WAVES INCIDENT 0N A SPHEmICAL OBSTACLR Suppose that the incident wave-train is travelling the negative direction of the axis of that from towards ; and that it is polariged in the plane of ( that is , in the plane .
Suppose further that the electric force in SCATTERING OF PLANE ELECTRIC WAVES BY SPHERES .
the wave-train has uni amplitude ; then , in terms of the Cartesian specification , the incident wave is defined , the remaining components of force being zero .
We must first express this wave in the standard forms of and ; we therefore introduce polar co-ordinates , and then proceed to find the radial components of foroe , which will suffice to determine the functions , V. These radial components are given by where the time-factor is now omitted .
Now from we have the formula .
So , differentiating with regard to , we find that Accordingly , on substituting in , we find that in the incident wave ( 43 ) by comparing the two formulae and .
The corresponding waves in the interior of the sphere will be given by the two functions ( 44 ) , and , where and are the fundamental constants of the spherical obstacle .
is*sumed that .
in the wave we may take Similarly the scattered waves will be given by the two functions .
The boundary conditions are given by the continuity of the tangential components of electric and magnetic force at the sphere It is evident from the form of equations , that these conditions will be satisfied if we take ( 46 ) and Thus we find that and ( the coefficients in the scattered waves ) are given by ( 47 ) .
The special case of a fectly conducting sphere is given by making the tangential electric force zero at the sphere ; and this condition is satisfied if ( 461 ) Thus we find the simpler formulae .
which may be regarded as limiting forms of , when and The formulae and lead at once to those quoted by Dr. J. PROUDMAN* in calculating the pressure of radiation due to a plane wave incident on a smaU conducting sphere .
In all the applications with which we shall be concerned at present the point at which the disturbance is to be calculated be at a distance large compared with ' Monthly Notices of the Royal Astronomical ' vol. 73 , 191$ , p. SCATTERING OF PLANE ELECTRIC WAVES BY SPHERES .
the wave-length .
Then we can simplify the general formulae by observing that may be replaced by the approximation if , \amp ; c. , are neglected .
Further , in the final formulae for the forces , and occur only in the two combinations where terms of the relative order have been rejected .
On substituting from we find , to the same degree of accuracy , ( 48 ) Then , substituting in the general formulae and , we find that ( to our present order of approximation ) the radial components of force are zero , and that the transverse components are given by ( 49 ) Accordingly the and gnetic forces in the scattered right angles to each other to the radius , their mwnitudes are related in the as in a plane This conclusion might very well have been anticipated ; and for the case of small obstacles of any shape with constants differing little from unity ) the conclusion is contained in a paper by Lord RAYLEIGH .
* But I cannot find that it has been noticed for the case of spheres of any size , and of any electrical and magnetic constants .
This may serve to indicate one advantage of the formulae in spherical polars over those in Cartesian co-ordinates .
The formulae , , with the values of given by , were those used by Messrs. PROUDMAN , DOODSON , and KENNEDY in their numerical calculations quoted in the introduction to this paper .
* Scientific Papers , ' vol. 1 , pp. 522-536 .
For a small perfectly conducting sphere the same conclusion is given by Sir J. MSON , 'Recent Researches , ' p. 448 .
S5 .
SPHERES SMALL COMPARED THE . .
: .
, The fundamental assumption is that is small enough to justify us in ting all but one or two terms in the power-series for and .
It has been usual to assume further that is correspondingly small but Dr. PROUDMAN has remarked that in the case of a dielectric sphere with a large value of the second assumption need not follow from the first .
It seems rth while therefore to simplify the formulae by expanding in powers of , while retaimng the general forms for ; it will be seen moreovel that the resulting formulae form a link between the results for dielectric spheres and those for conductors .
Remembering that is of order and that is of order , it is easy to see from that in general and are both of order .
Thus in the first approximation it will be sufficient to deal only with the coefficients and ; and for these we need the mulae for and .
Now from we have , and so .
Hence where now .
In the second , for the functions of , from and we find the first approximations Substituting , it will be seen that for the first equation in gives where After a little reduction the last equation gives ( 51 ) The- second equation in ives a similar formula for with taking the SCATTERING OF PLANE ELECTRIC WAVES BY SPHERES .
place of K. However , in most cases , it is sufficient to write , and then the formula simplifies further and becomes ( 52 ) The two formulae and are due to Dr. PROUDMAN , who has pointed out that they connect the results found by Lord RAYLEIGH for the case of dielectric spheres , and by Sir J. J. THOMSON for conducting spheres .
To deal with the case of spheres we do not regard as large , so that may be regarded as small ( of the same order as ) ; and then the approximations , may be used .
This gives , in place of ( 61 ) , ( 6 the simpler forms due to Lord RAYLEIGH* ( 6 21 ) On the other hand , Sir J. J. THOMSON'S case corresponds to the assumption that is of the form where is very large ; then may be regarded as large , and as complex , with a negative imaginary part .
Thus approximately , and so , which ( although large ) is small compared with .
Hence we find from ( 5 1 ) and ( 5 2 ) the approximate results , ( 6.22 ) as given by Sir J. J. THONSON .
Of course this pair of formula follow at once from ( 4 , on inserting the approximations for and given on p. 188 above .
Dr. PROUDMAN makes the further remark that , under the conditions assumed in and , variations in the wave-length may produce very considerable changes in the magnitudes of and , on account of the presence in of , which m'ay vary very fast .
It is of course supposed that the sphere is dielectric , otherwise could be replaced by , as already stated .
It is worth while to note the simple formulae for the scattered wave , derived from ; these give , to the present order of approximation ( 53 ) ; * Scientific Papers , ' vol. 4 , p. 321 ( 106 ) ; see also vol. 1 , p. 526 .
Provided that the imaginary part of exceeds in numerical value , the error in this approximation is less than one half per cent. 'Recent Researches , ' p. 448 .
190 DR T. J. I'A BROMWICH ON Tlffl and so ( to this order ) the scattered wave is zero in the direction given provided that is numerically greater than .
Thus in Lord RAYLEIGH'S case , the direction is given by and in Sir J. J. THOMSON'S by It may be here that , if the sphere has a sufficiently large dielectric constant , it may happen that is numerically less than ; and then the direction given by ( 5 31 ) is no longer real .
Taking to be real ( the case of a conductor having been already considered p. 189 ) , it is easy to see that gives assumption K .
Now the function adily dases from 3 to as varies from to and a rough calculation shows that is slightly less .
than unity .
Also in order to justify the approximations used for and , we must suppose that Hence the possibility contemplated may occur if , say , , ' and giving The direction in which the scattered wave vanishes will be given by the final formula being simplified by remembering that is large .
I am not aware that there is any experimental evidence showing traces of this phenomenon ; in fact all the evidence shows that is not from the truth .
Thus the circumstances in actual experiments cannot have been such as to introduce the reversal of magnitude between and ( ii .
) Second Approximations .
We proceed next to find second approximations , assuming that is not large ; it will be necessary to retain the second terms in the series and for and , but the first terms will suffice for and It is easy to see that then terms of order occur in the coefficients and , but that no other coefficients can contain terms of order lower than Using now the series for we have , retaining the second terms only in.each series .
Thus we find , to the same order , Also gives similarly .
closer approximation is worked out on the next page ; see formulas ( 6 6 ) below .
It would be possible , of course , to obtain second approximations to and ( 5 2 ) , but , glance at the formulae shows that the work is so laborious as to be most impracticable .
and , where , for brevity , we have written On reducing these equations , the results are ( 54 ) To determine , the series for and will be required to the first terms only ; these are and on substituting in , we find ( 55 ) .
The field of the scattered waves is then given by ( 56 ) The field is accordingly zero ( to same degree of approximation ) in the direction given by This conclusion is apparently new ; but it confirms an approximate result due to Lord RAYLEIGH , *according to which the scattered wave is zero in the direction given by when ( K-1 ) is treated small .
But on the other hand , our result contradicts 'Scientific Papers , ' vol. 1 , p. 631 , formula ( 61 ) .
VOL. CCXX .
a statement made by Prof. LOVE* that there is no direction in whioh the mrttmed wave is completely cut out ; however , on a oSer examination of Prof. LOVB'S formulae , they appear to confirm the present conclusion .
The formulae in question are ( 42 ) and ( 43 ) of the paper just quoted , but apparently there is a slip in ( 43 ) .
In the last line of ( 43 ) the factol given as should really be ; the source of the inaccuracy being apparently in the passage from the formula ( 39 ) to ( 41 ) .
On introducing this adduional term in the magnetic force , it appears that the electric and magnetic forces are zero in the direction given by in Prof. LOVE'S notation ; of rse " " zero\ldquo ; means that the forces are really oi order at most .
It is not difficult to prove that the formulae agree with those found by Lord RAYLEIGH and Prof. LOVE ; S the method to be adopted is similar to that used in S4 of my paper in the ' Philosophical Magazine ' ( quoted on p. 179 above ) .
But it should be observed that in the specification of the incident wave adopted by Lord RAYLEIGH and Prof. LOVE , the electric force is to the axis of ; but here the electric force is parallel to the negative direction of .
Thus if denotes the azimuthal angle corresponding to the former specification , it is evident that corresponds to ; and accordingly we shall have in general the relation because both angles are measured in the right-handed sehse about the axis of Lord RAYLEIGH'S paper contains tables and graphs from which it is easy to determine the variation of the field with ; and in order to connect his tables with our formulae , let us write in the form Consider , first , points in the plane given by , in Lord EAYLEIGH'S notation ; this gives , or , omitting the factor , the electric force is equal to ( in the direction of decreasing ) ; and accordingly is resented by the graph of the Cartesian component , given by Lord RAYLEIGH .
Secondly , consider the plane ; that is , or .
Here the electric * Proc : Lond. hIath .
Soc vol. 30 , 1899 , p. 318 .
A numerical slip in the first line of each of the formulae and ( 43 ) to be corrected ; the correction was given in the Errata , vol. 31 , ' Proc. Lond. Math. Soc. ' 'Scientific Papers , ' vol. 6 , p. 669 , ( u ) and ( v ) .
S Formulae ( 42 ) and ( 43 ) of the paper just quoted ( allowing for the corrections just mentioned ) .
SCATTERING OF PLANE ELECTRIC WAVES BY SPHERES .
force is equal to , perpendicular to the plane ; thus is represented by the graph of the Cartesian component In general , the resultant electric force is represented by .
S5 .
Second for Conducting Spheres .
The foregoing algebra needs no alteration beyond replacing by the appropriate complex refractive index associated with the particular metal and wave-length considered .
This of course assumes that is not so large that the convergence of becomes too slow to justify the approximation made above ; and then the formulae to provide the solution .
It will be noticed that when is complex , the equation will not usually give a real value for : and so there is usually no direction in which the scattered wave is zero* ( or of order ) .
It may be of interest to note here that experimental work on the scattering of light by fine particles has been carried out with silver particles suspended in water .
The corresponding values of seem to vary from to 2 , and the value of is taken as ; thus the approximations in ( ii .
) are not sufficient to calculate either or with any accuracy .
In actual fact it proved necessary to use Lord RAYLEIGH'S exact formulae , equivalent to above , and to go as far as in the series .
S S6 .
CASE 0F LARGE PERFECTLY CONDUCTING SPHEmES .
Before proceeding to the final formulae , it will be convenient to state certain results given by MACDONALD for the values of the functions , , when both and are large .
* For the case in which K-l is small this conclusion is given by G. W. WALKER , 'Quarterly Journal of Mathematics , ' vol. 30 , 1899 , p. 217 .
The formulae given on that page agree with ( 5 6 ) , when K-l is small ; but the more general formulae on the preceding page do not agree with ( 5 6 ) completely .
I have not succeeded in tracing the discrepancy on account of the fact that G. W. WALKER has omitted some of the details of his preliminary calculations .
E. T. PARIS , 'Phil .
Mag vol. 30 ( Ser. 6 ) , 1915 , p. 459 .
To obtain an accuracy of 1 per cent. in by retaining two terms of the series only , it must be supposed that does not exceed 1 3 .
S .
T. PARIS , .
cit. , p. 472 .
'Phil .
Trans vol. 210 , , 1910 , : the formula are due to L. LORENZ originally .
A very interesting method of deriving the results is given by DEBYE Math. Annalen , .
Provided that is of an order higher than the formulae are ( 61 ) We shall need also the corresponding formulae for and ; it will be seen that , because gives Hence ( 62 ) Similarly , we find that .
The formulae to be used finally are those for and , given in , thus we take where now de1lotes .
It follows from above that , and using and we see that It is now an easy matter to writ down an approximation to the functions and defined in , provided that is not near to or .
Under these conditions we can take the approximate value * Uuder our conditions is uot near to zero and is large , so that X is small .
SCATTERING OF PLANE ELECTRIC WAVES BY SPHERES .
giving a corroeponding approximation P. Thus , where and we have replaced by unity , because is large .
Thus , on putting , we find that gives the approximation ( 64 ) .
Similarly , we get the formula .
With series of this type , the leading part is found by making the index of the exponential stationary ( regarded as a function of ) .
Now in both and there is one index only , , which can be stationary : and the condition is Now , and so the leading terms in and arise from taking , or The corresponding value of the index is then To determine the form of the index near to this special value of we take and then we find the approximate formulae ( 651 ) The leading parts of are accordingly given by the where and has the value given in .
The value of is approximately equal to the integral ( 661 ) where .
Thus , approximately Accordingly , to the same degree of approximation , we can take ( 662 ) Then the components of force are given , as in , by ( 663 ) where differential coefficients with respect to are small compared with those with respect to , and so have been rejected .
Thus , usin , we have the approximations to the forces in the scattered waves ( 67 ) assuming that is neither near to nor to When is small , the approximation to must be taken as .
and so SCATTERING OF PLANE ELECTRIC WAYS BY SPHERES .
Thus we now write and , proceeding as before , we are led to the conclusion that is near to ; thus the value of is small , in the parts of series which contribute the principal part of the sum .
Then we can replace by the approximate formulae ( 68 ) Thus Hence the approximation corresponding to is now In like manner the value of is found to differ from only in having as a factor instead of\mdash ; cos In the series the value of may be supposed to vary from to ; and so we obtain the principal part of the sum by using the integral ; and when is very small the value of is approximately equal to .
Thus ( 683 ) Accordingly the components of force are now found to be ( 69 ) l.'hese results in in the approximations , and accordingly the formulae ) vdid right up to the axis When is nearly equal to , the calculation on the present lines becomes more difficult , we shall accordingly obtain the corresponding approximation .
by a different process in the next section ( S7 ) .
It appears that for the special value , the formulae do give the foroes with a fair degree of accuracy up to an angle ; the approximation in fact appears to be better than might have been expected .
[ See p. 176 above .
] S7 .
ALTERNATIVE METHOD , APPLICABLE To AIVY CONDUCTOR WHOSB DIMENSIONS ARE LARGE COMPARED WITH THE WAVE-LENGTH .
It follows at once from GREEN'S theorem that if are solutions of the equations at points within a closed simple surface , then ( 71 ) where the integral is taken over the surface bein the element of outward normal ) , and it is supposed that are both free from singularities in the interior of S. Similarly if has no singularities and is a.solution which behaves like near a particular point ( denoting the distance measured from P ) , we see that ( 7.11 ) provided that is inside the surface S. Equations of similar forms apply when the space considered is outside the surface ; but then the sign of the last equation is reversed , giving ( 712 ) ; it is then necessary to assume also that at infinity , both correspond to divergent waves ( unless it is known that tends to zero more rapidly than ) .
* Compare MACDONALD , lo .
, pp. 120-122 .
The cause of the difficulty is to be found in the fact that now the stationary value may be expected to arise from values of for which is small .
Then is nearly equal to , and in all such cases more is inevitable .
In fact the appronmrtions to and to be modified by different formula lnnding to the cases and to the cases in which is of order or of lower SCATTERING OF PLANE ELECTRIC WAVES BY SPHERES .
, now consider the problem of waves incident from some source ( or sources ) and reflected from the surface S. Let denote any Cartesian component of force in the incident wave , and let the point be outside S. Then applies , and so because has no singularity inside S. If denotes the corresponding component of force in the reflected wave , we have from because has no singularity outside ( and will correspond to a divergent wave at infinity ) .
By addition we have the result where is taken to be Now gives the corresponding component of force in the complete wave ; and this accordingly satisfies certain known relations at the surface ( the exact form depending on the physical properties of S ) .
It must , however , be clearly understood that we cannot usually obtain both and by any simple methods , any more than .
the analogous problems.of electrostatics can be solved by a mere appeal to Theorem .
However , we can obtain approximate solution , suitable to the problem of short -lengths , by assuming that near the reflecting surface , the character of can be determined from that of by the rules of elementary geometrical .
optics .
Thus we treat the reflected wave as derived from the incident by simple reflexion in the tangent-plane at the point of incidence .
Making this hypothesis it is an easy matter to construct both and when the form of is given .
It will be noticed that we shall have at all points within the geometrical shadow ; and so the final integral extends only over the illuminated side of the surface S. now that we consider electric waves incident on a simple convex conducting surface ; and take an origin on the surface such that OP is the reflected ray ( in the sense of geometrical ) .
Take the plane of incidence as the plane of , and the normal at as the axis of CCXX .
Then in the immediate neighbourhood of we can represent the inciffit the components of electric force where , so that is the cosine and the sine of the angle of incidence .
We have also the relation because the electric force is perpendicular to the incident ray .
The corresponding reflected wave has the components where and so that ; these results follow by making the tangential components of force zero on the tangent.plane ( instead of at the surface ) .
It is now clear.that , at the point , the components of the total force will be ( 72 ) , and the normal differential coefficients of the total force will be ( 721 ) .
We shall now insert these values for and in the general formula , including also the factor , on of phase-differenoes at points near to .
O. Since reflexion takes place only from the immediate neighbourhood of , the error introduced by this simplification be smah SCAITERING OF PLANE ELECTRIC-WAVES BY SPHERES .
201 Since the co-ordinates of are , the value of is given by where is a point on the surface near to O. Thus , when is very large in comparison with the dimensions of the surface ( as we assumed in the previous investigations , SS4-6 ) , we can use the approximate formula Thus we can write in if is the value of at O. Then the most important term in is seen to be Accordingly the components of electric force in the reflected wave will be given by the approximation ( 7.4 ) To evaluate the integrals in ( 74 ) we must write out the equation to the surface in the approximate form Then* Now is the absolute ( or Gaussian ) curvature of the surface at the point ; and since the surface is supposed convex , we represent this curvature by * This is most easily found by taking the integral as BROMWiCH ON Thus we r- .
Hence , using , we see that the principal part of the reflected wave is given by ( 7.5 ) In order to interpret for any axes of co-ordinates , we need only notice that represents a force numerically equal to the force in the inci\amp ; q wave ; and that the new force is perpendicular to the reflected ray , arranged in such a way that the tangential components are opposite to in the incident wave .
We can apply the formula to the problem of S6 at once ; clearly The point of incidence corresponding to the scattered wave is given by .
Then the incident wave at has the components of electric force !
in the plane of incidence , and perpendicular to the of incidenca Further , the of formula is measured from ; to compare with S6 , we to be the distance , measured from the centre of the sphere .
Thus we are to replace in by .
Accordingly the compon .
of force at , in the reflected wave , are perpendicular to in the plane and - perpendicular to the plane ZCP .
These results agree with and of S6 above .
SCATTERING'OF PLAN WAVES , BY SPHERES .
It to ndify the general formulae so as to cover the case of waves.incident from a point souroe ( say at distance from the point of incidence ) .
Then Thus with the usual approximation of geometrical optics Similarly Now is of the second order in comparison with ; and so we can write Thus here we find where , being the distances of the focal lines ( of geometrical optics from the point of incidence .
Thus now the incipal parts of the reflected wave are given by ; .
usuming that is not close either to , or to The results of the foregoing analysis depend on the tacit assumption that is not zero ; and as a consequence the character of the approximations will change when DR. T. J. ICH ON . .
: is small .
That is ; near the edge of the optics .
In the application to the sphere , the region excluded by this condition corresponds to values of nearly equal to ; and in this region the specification of the scattered wave by means of Cartesian co-ordinates seems simplest .
We consider then the incident wave as specified in Cartesian form by ( 76 ) and consider the approximation to the reflected wave incident at the point on the sphere for which the direction-cosines of the normal are .
The expressions will be of the form ( 761 ) where ( treating the tangent-plane as the reflecting surface ) an is determined by Thus Further , the resultant of and at the point of incidence must be along the normal ; and so Also is perpendicular to the reflected ray ; and so Hence The components at the point of incidence are accordingly equal to We have still to evaluate the normal differential coefficients , which are found to be ( 764 ) The value of is now seen to be given by -t .
SCATTERING OF PLANE ELECTRIC WAVES BY SPHERES .
Thus , when we regard as small , we may take ( 7.65 ) Now , in applying , occurs only in the coefficients and not in the exponential index : thus we can get the first approximation by putting in ; this gives the value ( 766 ) Thus , to our degree of accuracy , .
We can now substitute for and the values given by and : it will be seen that the components allel to give zero ( to this order ) , and that the component paralel to gives Accordingly the reflected wave is given by where is found from and the integral extends over the positive hemisphere .
For the purpose of integration we write Then in we have The integration of with respect to can be carried out at once , because .
Our result accordingly becomes .
X An exact evaluation of the integral would be troublesome ; but it can be transformed by integration by parts .
This process leads to a series of which the first term is ( 77 ) .
: The formula will represent sufficiently accurately if is regarded as smaU ; and this is the correct assumption here , since is supposed to be nearly equal to In the special cases , 10 it appears that the approximation represents the scattered wave sufficiently within a cone extending to about from the axis .

