]\gt ; [ 691 ] XIX .
On the Electromagnetic Theory of the Reflection of Light .
By .
FRAS .
FITZGERALD , M.A. , Fellow of College , ComPnunicated by G. J. STONY , M.A. , F.R.S. , SecretaJ'y of the Queen 's of Received October , 1878\mdash ; Read January 9 , 1879 .
IN the secolld volume of his 'Electricity and Magnetism ' Professor J. CLERK MAXWELL has proposed a very remarkable electromagnetic theory of light , and has worked ou the results as fa as the transmission of light through ystalline a magnetic A media are concerned , leaving the questions of reflection and refraction untouched .
These , however , may be very conveniently studied from his point of view .
If we call the electrostatic energy of the medium , it may be expressed in terms of the electromotive force and the electric displacement at each point as is done in Professor MAXWELL'S 'Electricity and Magnetism , ' vol. , part .
I shall adopt his notation and call the electromotive force and its components , and the electric displacement and its components .
As several of the results of this paper admit of a very elegant expression in Quaternion notation I shall give the work and results in both Cartesian and Quaternion form , confining the German letters to the Quaternion notation .
Between these quantities then we have the equation dxdydz Similarly the kinetic energy may be expressed in terms of the magnetic induction , , and the magnetic force , , or their components and by the equation , I shall at present assume this to be a complete expression for and returll to the cabe of magnetized media for separate treatment , as Professor MAXWELL has proposed additional terms in this case in order to accom ) for their property of rotatory polarisation .
I shall throughout assume the media to be isotropic as regards magnetic induction , for the contrary supposition would enormously complicate the question and be , besides , of doubtful physical applicability .
For the present I shall not assume them to be electrotatically isotropic .
Hence is a linear vector and self-conjugate function of , and notation and if we call the general symmetrical quadratic function of we may assume and consequently Udxdydz As the medium is magnetically isotropic we have or where is the coefficient of magnetic inductive capacity , and consequently the electro kinetic energy may be written dxdydz Now I shall assume the mediums to be nonconductors , and although this limits to some extent the applicability of my results , and notably their relation to reflection , yet it is a necessity , for otherwise the problem would be beyond my present powers of solution .
Witb this assumption , and using NEWTON'S notation of for ' we have the following equations Elect. using for the operation ( / 1 .
the same in terms of its components , namely , OF REFLECTION AND REFRACTION OF LIGHT .
Assuming now a quantity with components , such that and consequently in terms of the components we may evidently write so that we have dxdydz LAGRANGE'S equations of motion may often be very conveniently represented as the conditions that should be a minimum , or in other words that and this method , from its symmetry , is particularly applicable to the methods of Quaternions .
Proceeding on this method we obtain immediately the equation @ .
or in Cartesian notation IIJ Now we may evidently integrate the terms in and with reference to the time , and the terms depending on the limits of the time must vanish parately , and we are not at present concerned with them , so that the equation reduces to dxdydzdt or dxdydzdt NDCCCLXXX .
MR. G. F. FITZGERALD ON TfiE ELECTROMAGNETIC THEORY I shall now proceed to integrate this by parts relatively to , and in order tg express the result conveniently I shall assume to be an element of the surfaoe of the medium and , with components , to be a unit normal to this element of surface , when we evidently obtain or It is evident that the superficial and general integrals must vanish separately , and as @ or @ , @ and S , is arbitrary in the general we must evidently have or As it will not take long I will deduce the ordinary equations for the transmission of plane-waves from these before proceeding to discuss the superficial condition .
In the first place , as our axes of coordinates are perfectly arbitrary I shall assume to be normal to the plane of the wave , and consequently to be functions of and only , so that and , and consequently and and so that if and consequently Now if we had originally assumed we should have so that we obtain H. so that if we assume and parallel to the axes of the section of by the waveplane we evidently have got , and consequently and and as A and are inversely proportional to the squares of the axes of the section of or by the wave-plane , we get M'CuLLAGH'S result that each component in these directions is propagated independently and with a velocity inversely propor- tional to t , he axis of the section of by the wave-plane .
This comes out at once from the Cartesian equations , for as reduces to and by choosing and parallel to the axes of this section , we have , so that and and and the result of putting these in is evidently the same as before .
Returning now to the superficial conditions , I shall follow M'CULLAGH , and assume that at each point of the surface of separation of two media the values of the elements of the integrals must be equal for the two media , and , indeed , I think it is pretty evident that if the original integrals are to express the whole state of in the case of a motion propagated from one medium into another , the superficial tegrals must vanish when the limits introduced in them are the functions corresponding to the two contiguous media .
Using the suffix for one medium and 1 for the other , and assuming the normal to the surface as , we get , as SR is evidently arbitrary and same for both media ) the surface of separation , or the two equationS which are the same as can be got at once by putting in the Cartesin superficial equations , and taking @ and @ as arbitrary and independent , when we If now we assume that the axis of is the line of intersection of the plane of incidence and the surface , we may evidently assume to be the resultant of an incident and reflected ray , and to be the resultant of the two refracted rays , so that we may write when are respectively unitvectors parallel to the direction of magnetic placement in the incident , reflected and each refracted ray .
Similarly the unitvectors normal to these wave-planes , and the variable dlstaI ) along these , evidently we may assume , \amp ; c. , and substituting these we at once obtain In these I shall now assume as it is convenient to suppose this medium to be isotropic and the velocity of propagation in it unity , and and are the direct ons in the wave-plane perpendicular to the magnetic displacement ; so that if be the direction angles of these lines referred to the superficial axes and if be that axis of the section of by one of the refracted wave-plana direction of magnetic displacement , en dently which is perpendicular to the velocity of propagation of this wave in this medium , and we may write when is the perpendicular on the corresponding tangent plane , and consequently TvTs .
OF THE REFLECTION AND REFRACTION OF LIGHT .
when is the angle between and the axis of , and evidently represents the velocity of this wave , while is the direction of the ray .
Hence our equations may be written These may be readily deduced from the Cartesian equations as follows : being isotropic it can be written , and is evidently unaltered by transformation , and if : and : : be the direction cosines of and to any arbitrary axis , we get and now , as these are linear , we may suppose the superficial disturbance in one medium to be due to an incident and reflected wave , and in the other to two refracted rays , so that we can write these as and as and are supposed to be referred to arbitrary axes , we may suppose to be referred to such that is the only component\mdash ; i. , to such that the ection of the magnetic force is the axis of , that of being normal to the wave-plane , and similarly being the only component in while the axis in this case is normal to its wave-plane , and of course this will be a function of only and of only , these being the corresponding ordinates .
We thus obtain so that and the equations become Now if we consider we see that they are proportional to the direction cosines of the perpendicular on the tangent plane to where it is met by \mdash ; i.e. , it is the direction conjugate to the electric displacement corresponding to Now if we call the velocity of propagation of this wave and of the other , wg see at once that is an axis of the section of constant by one wave-plane , and is an axis of the section by the other wave-plane , and if and be the corresponding conjugate directions , and and and the angles between these latter and the superficial and axes , and if be the corresponding angles between and and these same superficial axes , we get exactly the same equations as before , and write In order to reduce these further we assume and if our superficial axes are so placed that the ax of is the intersection of plane of incidence and the surface , and if be the angles the respect ve normals make with the normal to the surface , we evidently may write i\fnof ; m \fnof ; m lvin would be inadmissible , as it is easy to convince oneself that any terms nvo vng lvin the time could not they could not by hypothesis occur in , and the terms nvoH vngthese wave normA vanish out of our equations if they occurred in the others ence be true inde are all in the same plane .
When our equations must dently pendently of the time and , from which we see that no change of phase is possible in OF THE REFLECTION AND FRACTION OF LIGHT .
reflection .
Hence these equations cannot explain metallic reflection .
Indeed , this question of change of phase seems to be one of a higher order than I am here dealing with , and requires a discussion of the nature of the transition from one medium to another , which , of course , cannot be abrupt , as our equations suppose , nor indeed probably , in these cases , even very sn1all compared with the vibrations .
In order terms involving the time should disappear from the equations when we lmust have which involves that the angle of incidence should be equal to the angle of leflection ; and if the second medium were an ordinary one , so that we should have that .
the ratio of the sines of the angles of incidence and was constant .
Putting in these values , our equations reduce to together with the condition that when the superficial displacements should be the same to whichever medium they belong , namely , As each of these is a resolved part of the vibrations and we get three additional equations , the last of which , however , is the same as the second of the former ones and there result , consequently , but four equations from which the four quantities , namely , the three intensities , and the azimuth of are to be determined .
It is remarkable that whether we assumed or no that it is here introduced .
That is , however , no proof that it is wrong to omit it , as in FRESNEL'S method of obtaining the intensities of the reflected and refracted rays , for the fact of its turning up independently shows that there is something at least debateable about it , and as I shall have cause to omit this equation as leading to Ilient results in a ubsequent part of my paper , I thought it well to mention that there is something curious about it even here .
These equations are those long ago given by M'CULLAGH in the ' Transactions of the Royal Irish Academy , ' vol. , to solve the problem of crystalline reflection and refraction , and from which he deduces his beautiful theorem of the polar plane and thus marvellously simplifies an extremely ] icated problem .
Indeed , as the forms into which I have thrown and are identical with his expressions for what are practically the same quantities , my whole investigation so far 18 but a modification of his .
In the simple case of the second medium being also otropic we have that , and if we in the first place suppose to be in plane of incidence , we have at once , and consequently , and so that the two refracted waves coincide .
Also and likewise , as a easily be seen by assuming that the cted and refracted waves have components oui of the plane of incidence and then trying to satisfy the equations .
Hence our equa tions reduce to and becomes being the angle of incidence and of reflection .
The first of these is and solving them we get If be perpendicular to the plane of incidence we obtain , and our equations become and which give Having now deduced the already known laws of reflection and refraction of light at crystalline and ordinary surfaces , I shall proceed to consider the that Mr wonderfully beautiful experiments have made so interest ng\mdash ; name 1 the case of reflection from magnetic surfaces .
In order to do this I shall assume , with Pro Professor J. CLERK MAXWELL Electricity Magnetism , , that the kinetio dium contains a term depending on the displacement of certa energy of the me supposed vortices , and that it may be expressed by an equation of the form ( S826 ) dxdydz where and are now the components of the vortex M. I shall assume the medium to be isotropic , so that taking the elcttaW energ of the medium may be expressed as *When is the refractive index .
OF THE REFLECTION AND REFRACTION OF LIGRT . .
dxdyc dxdydz While the complete value of is dxdydz dxdydz Working as before , and obtaining our equations as the condition that should be minimum , we have to satisfy the equations dxdydzdt or in Cartesian coordinates Now of course the terms depending on are the only additional ones , and it will be necessary to integrate them both relatively to the time and also relatively to The integration with respect to the time is easily performed as it merely consists in removing the dot from one of the terms under the integral to the other and changing the sign , so that , neglecting the terms depending on the limits of the time with which we are not concerned , our equation becomes NDCCCLXXX .
: or in Cartesian coordinates Integrating these now by parts relatively to , and calling the direction cosines of the normal to the surface of separation , and when is the vortex whose components are , we get or in Cartesian coordinates dxdydz OF THE REFLECTION AND REFRACTION OF LIGHT .
I shall not consider the general equations of wave propagation in the Cartesian they are the same as those given by MAXWELL in a more general form , for he assumes that the static energy of the medium is expressed more generally than I have assumed , but as the Quaternion investigation is not long I shall give it .
The equation of motion is and if be the normal to a wave-plane , we may evidently assume , and as is in the wave-plane we have as our equation becomes Substituting for , and equating the coefficients of and separately to zero , we get from which it is easy to see that and that is determined by the equation which gives of course two different values of , one for each circularly polarised ray , or disregarding the solutions for waves going in the negative direction , we have approximately and of course can easily be determined from these by observing the rotation produced , but there is nothing except experiment to prove that may not be a function of , as we know to be , to some extent at least ; so that using these formulae in order to obtain the laws of dispersion of rotatory polarisation seems to approach towards deducing the known from the unknown .
All we can be sure of is that is in general extremely minute .
Returning to the superficial equations , and using a sign of substitution , we ffl As I intend only to work out the results in Cartesian coordinates , I shall confine myself to them , and for simplification assume the axes to be normal to and and in the surface , and consequently , and may also be supposed to vanish for one of the media for which is the dielectric inductive capacity .
I shall also assume that @0 as the vanishing of its coefficient leads to inconvenient results , and the assumption may be to some extent justified by considering that as these S , @ , S are superficial values , no virtual displacement out of the surface , as this would be , is admissible .
From the other two , S and , we evidently get A remarkable point about these equations is that they admit of being integrated with regard to the time as far as their right-hand members are concerned , so that in addition to the values which satisfy them as they stand , and which are the only ones of much interest , there are other periodic values of independently of which satisfy these supel.ficial conditions , and which may consequently be looked upon as a sort of free vibration of the surface of the medium .
But even if this could be propagated into the rest of the ether it is improbable that the resultant vibrationm would be of such a period as to be visible , though some energy might be expended on them .
I shall now further assume that the axis of is the intersection of the surface with the plane of incidence of a plane-wave , and that cQnsequently none of my quant tes are functions of , which reduces these equations to In order still further to simplify the problem I shaH now consider separately tht cases in which the magnetisation of the medium is normal and that hich it it in the surface .
In the first place , it is evident that if it were all in we have , and the coefficient of would vanish in both equations becau OF REFLECTION AND REFRACTION OF LIGHT .
, and hence we get Mr. KERR'S result ( ' Phil. Mag. , ' March , 1878 , p. 174 ) that when the plane of incidence is normal to the lines of magnetic force the magnetising of the mirror produces no change in the reflected light .
Assuming , then , first that the magnetisation is normal to the surface , we have , and , so that calling our equations become ( I ) while if it be all in x\mdash ; i.e. , in the intersection of the plane of incidence and the surface , , and , so our equations are ( II ) I shall now proceed to solve these two systems of equations each for the two cases of waves whose magnetic forces are first in the plane of incidence , and secondly at right angles to it .
From the forms of the it is evident that we cannot assume either the reflected or refracted rays to be similarly polarised , but it is easy to convince oneself that there can be no difference of phase introduced in this case any more than in the former one of ordinary reflection , and , as I then remarked , it is evidently a question of greater complication than to be capable of being deduced from the simple assumption that the alteration of the nature of the medium in going from one into another is abrupt .
Until more is known of the nature and extent of this change I fear we must be content with theories which only partially represents the facts .
As before , I shall assume to be due to the incident and reflected rays , while are due to the refracted ray .
The incident ray may be taken to depend upon the angle calling the velocity in this medium unity and the reflected wave on and here we must have , as before , , and this involves , as before , the ordinary laws of refraction , as is constant and in general Taking then first the case of the magnetisation being all normal to the surface and the incident displacements in the plane of incidence , we must evidently assume ( A ) in order to satisfy the equations As it leads to inconsistent results , in accordance with FRESNEL , omit the third equation , .
It is not easy to justify this omission , I fear , and the result must do so , as well as the consideration that probably if we had a better insight into nature of the change from one medium into another our equations ] be so modified as not to present these anomalies .
From the last equation it is manifest that and as is very small it is obvious that will always be small , so that may .
omitted in the first equation .
Putting in the values of , \amp ; c. , and remembenng that when , these equations evidently reduce to OF THE REFLECTION AND REFRACTION OF LIGHT .
and from these we must solve for , and in terms of .
However , , the component introduced at right angles to the original plane of vibration by reflection , is the only one of much interest , though the alteration produced in the values of the reflected component by assuming to differ from unity are noteworthy .
Calling the period of vibration of the wave we are considering , we may put , and if we call we get for the equation which if reduces to In the second case , when the direction of the vibration of the incident ray is perpendicuhr to the plane of incidence , we must evidently assume ( B ) and , as before , it is evident that is generally very small so that and may be omitted , and our equations become and when the values above are introduced into them we obtain MR. G. F. FITZGERALD ON THE ELECTROMAGNETIC pHEORY Hence we get solving for in terms of which when reduces to Turning now to the case where the netisation is in the surface , I shall firff suppose the incident vibration to be in the plane of incidence , when we assume , as before , equations ( A ) for , and these must satisfy the equations ( II ) , and in addition As before , we evidently get , and from the second equation of ( II ) see that must be small , as is , and consequently we may omit in the first equation , and no the first equation is the same as before , and we have which when solved for in terms of gives which when reduces to Finally , supposing the incident vibration to be perpendicular to the plane of incidence , we have determined by the equation ( B ) , which substituted in equations ( II ) , neglecting as before on account of its smalIness , give which when reduces to OF THE REFLECTION AND REFRACTION OF LIGHT .
It is remarkable that in this case the two components introduced by reflection are same , whether the vibration be in , or perpendicular to , the plane of incidence .
Comparing this method of the effects of reflection from a magnetised surface with that given by me in the 'Proceedings of the Royal Society for 1876 , ' , it is to be observed that my equations ( I ) and ( II ) are unaltered if the signs and are both reversed together , or if those of and and are all reversed , showing that a circularly polarised ray in one direction is reflected according to the laws when the magnetisation is one way as the oppositely circularly polarised would be if the magnetisation were reversed , and hence my former method of .dividing the incident plane polarised ray into two opposite circularly polarised ones , of which was reflected according to its own laws , is justified .
In comparing these expressions with the results of Mr. KERR'S admirable experiments , it is necessary to observe , as I mentioned before , that the introduction of a difference of phase between the reflected components is a question of a different order from that here discussed , and probably to some extent at least depends on the want of abruptness in the change from one medium to the other .
For instance , my expressions give change of plane of polarisation when light is reflected normally from the end of a magnet , but they would lead one to expect that the only effect was a slight elliptic polarisation , the major axis of the ellipse being in the same plane as the original plane of polarisation .
Now Mr. KERR'S experiments show that there is some rotation of this plane by reflection , and a supposition similar to one long ago proposed to explain the known elliptic polarisation of metallic reflection\mdash ; namely , that the efficient reflecting surface has some depth\mdash ; may easily be shown to lead to Mr. KERR'S result .
On this hypothesis the reflected ray is the resultant of the rays reflected from a small thickness at the surface of separation of the media ; and in the case of normal reflection from the end of the pole of a magnet , each of these components would be slightly turned from its original plane of polarisation owing to having passed through a very small thickness of a very powerful rotatory polarising substance\mdash ; namely , this superlicial layer of the magnet\mdash ; hence it is evident that their resultant would no longer be polarised in the same plane as the incident ray .
I only give this as an instance of how this lestion of a difference of phase affects the results , and how the hypotheses that have been framed to explain it might be used to bring my results into complete accord with Mr. KERR'S experiments .
I hardly think it worth while going into this more fully , as it is treading so closely upon unknown ground\mdash ; namely , the connexion between matter and ether\mdash ; that our hypotheses are to a great extent merely conveniences .
Another question is the extent to which affects ordinary reflection from a powerfully magnetic substance like iron .
I have never come across any experiment tending to show that the reflection iron was at all peculiar .
This may be owing to the electrostatic inductive capacity being a characteristic of the ether in the matter , while magnetic inductive capacity is a characteristic of the matter , and so ] affects the effect of magnetisation on wave-propagation is an expression of the same hypoth* In comparing my equations with Mr. KERR'S results I shaU consequently assu : Mr. KERR'S most elaborate experiments have been made with the lines of magnetj force in the intersection of the sllrface of the medium and the plane of incidence , I shall confine my attention to this case .
My equations give that the principal eff4 of reflection is to introduce a component perpendicular to the original plane of polazj sation whose amplitude is represented by the equation when the amplitude of the incident vibration is unity .
The following table represents the values of the variable part of this for the several incidences mentioned by Mr. KERR in his paper .
This table is calculated on the assumption that is the polarising angle for iron .
This table shows that the intensity of this introduced component vanishes for the two limiting incidences and , and attains a maximum alue of at about .
On comparing this with Mr. KERR'S results there a most striking corre spondence .
Summing up the results of his paper on reflection from a surface hich is magnetised so that the lines of force are in the surface and the plane incidence ( ' Phil. Mag March , 1878 , S 23 ) , he says : ' When the vibration reflected from thetho unmagnetised mirror is either parallel or perpendicular to the plane of reflection , effect of magnetisation is to introduce a new and very small component ra bration in a direction perpendicular to the primitive ation.\ldquo ; This is what Iave ca have called goes on to define its lrect on re ave of and thus direction relativel to the Amperean currents which ich are supposed to produce the magnetic force ; but as the sign of depends on the slgn on that of , which cannot be certainly determined for iron otherwise than by thew very experiments , any confirmation founded upon it such as I meant one my paper is to a certain extent illusory ; but the same arguments as are there used would * ' Phil. Mag March , 1878 .
OF THE REFLECTION AND REFRACTION OF LIGHT .
lffi to same conclusion here\mdash ; namely , that iron , if transparent , would behave like those ferromagnetic substances observed by VERDET , which rotate the plane of polarisation in the opposite direction to the Amperean currents , as is of course most probable .
He further mentions , in S 24 , that the phase of this component is always nearly the same as that of the component of the reflected ray polarised in the plane of 1 incidence .
This question , however , of difference of phase is one I am not at present prepared to explain , and , as I have mentioned already , must await a more comprehensive theory .
Considering next the intensity , of course no very accurate measures were possible on account of the minuteness of the phenomenon ; but he several times remarks that all effects of magnetisation vanished at normal and instances , while his maximum effects were always obtained at incidences of about or The anomalies observed connected with the incidence all belong to the question of the difference of phase between the components , which my investigation does not touch .
On the whole then I that my results , as far as they go , are in complete accordance with Mr. KERR'S experiments .
This investigation is put forward as a confirmation of Professor MAXWELL'S electromagnetic theory of light , in which , though there are some points requiring further investigation , neyertheless the foundation has certainly been laid of a very great addition to our knowledge , and if it induced us to emancipate our minds from the thraldom of a material ether might possibly lead to most important results in the theoretic explanation of nature .

