]\gt ; Tloeory of the Reflection of Laght the Polarising Angle .
By RICHARD C. MACLAURIN , M.A. , LLD .
, Fellow of St. John 's College , Cambridge , Professor of Mathematics , Wellington , New Zealand .
( Communicated by Professor J. Larmor , Sec. R.S. Received January 17 , \mdash ; Read February 2 , 1905 .
) Supposing the phenomena of light to be due bo displacements in a rotational ether , we have the following expressions for the kinetic energy and function in a transparent isotropic medium:\mdash ; V. Here denotes an element of volume , is the density , the displacelnent , curl , a constant that will prove to be the velocity of light in free ether , whereas will be identified with the refractive index .
The dynamical equations and boundary conditions are most simply obtained the Principle of Action , which makes We have The term containing in the variation of is ( ng-mh ) wIlere ( are the direction cosines of the outward normal to the bounding surface S. Picking out the coefficient of in the variation , we get , ( i ) and the surface conditions require ng\mdash ; , mf\mdash ; , to be continuous .
The displacement must also be continuous to avoid VOL. LXXVI.\mdash ; A. Dr. R. C. Maclaurin .
Theory of the [ Jan. 17 , rupture of the medium .
We shall take to be continuous , and if the surface of separation between different media be constant , the boundary conditions are satisfied when are continuous .
These boundary conditions combined with three dynamical equations of ths type ( i ) will suffice to solve completely the problem of the reflection and refraction of light at the boundary of two transparent isotropic media .
We take to represent the light vector , and if we wish to interpret results in the language of electric theory , we identify with the electric displacement , and take proportionalto the neCic force .
We shall first consider the case of an abrupt transition from one medium to the other .
The surface of separation is , the plane of is that of incidence , so that everything is independent of ( 1 ) Vibrations parallel to the plane of incidence ; so that is at right angles to this plane .
Thus ; ( incident ) ( reflected ) ; ( refracted ) .
The boundary conditions require the exponential factor to be the same for all values of and when ohus whence , which is the law of refleotion ; and 1905 .
] Reflection of Light the Polarising Angle .
which is the law of refraction for in order that the dynamical equation ( i ) for may be satisfied .
; ; O. Thus ; .
; .
; . .
These represent waves whose amplitudes are in the ratios 1 : The boundary conditions give ; whence , remembering that , we and 2 .
Vibrations to the plane of incidence .
In this case it is convenient , introduce a new vector which is the curl .
; ; ; ; ; Take Then the amplitude of the incident , reflected , and refracted waves are in the ratios 1 : : ; ; ; The boundary conditions give ; whence and If the incident light is plane polarised at an azimuth of to 'he plane of incidence , then the amplitu of the incident light are equal for the vibrations parallel and perpendicular to the plane of incidence .
After reflection and refraction the amplitudes will no longer be equal .
Let be the ratio of Dr. R. C. Maclaurin .
Theory of the [ Jan. 17 , the amplitude of the reflected light polarised parallel to the plane of incidence to that polarised at right angles , and let be the same ratio for the refracted light .
Then and are the " " coefficients of ellipticity\ldquo ; of the reflected and refracted light , and we have : and The formulae for and with those of .
They make vanish when , i.e. , when the polal ising a ; and they indicate a sudden change of phase of ls ( half a wave-length ) at this angle .
Experiments show that these formulae represent the facts very well as far as the ellipticity is concerned , except in the neighbourhood of the polarising angle .
It is found , however , is sensible at all angles , although smallest at the Brewsterian angle , and that the of phase does not occur suddenly , but enters by degrees .
explanation of this departure from Fresnel 's formulae was long ago ascribed to a gradual rather than an abrupt transition from one medium to the other .
This gestion was strongly supported by 's experiment on reflection from water having its surface artificially cleansed , and by the observation made in 1899 by Drude that the ellipticity of the polarisation of the reflected in the case of a freshly split surface of rock salt is very small , but that it rapidly increases on standing .
It seems probable that Fresnel 's formulae are rigorously applicable to the ideal case of an abrupt transition between two isotropic transparent media , and that the departure from these formulae is due to the fact that in most experiments there is what has been called a ' layer of transition\ldquo ; between the media .
The mathematical treatment of the problem of the layer has been undertaken by various writers\mdash ; amongst others Lorenz , Van Byn , Van Alkemaade , and Drude on the Continent , and in England , in an illustrative way , by Lord Rayleigh , and systematically by G. A. Schott .
* Schott 's * Phil. Trans , 1894 , pp. 82.3 to 885 .
1905 .
] Reflection of Light near the , Polarising Angle .
analysis is long and complicated , and leads him to results which do not coligate the experimental facts any more closely than very much simpler formulae that can be obtained .
Drude 's investigation*is more direct , but from a mathematical point of view his analysis leaves something to be desired , since he treats certain quantities as\ldquo ; approximately constant\ldquo ; merely because they are , continuous , and , further , because in proceeding by approxiation he gives no indication of the magnitude of the terms neglected .
Take the variable\ldquo ; layer of transition\ldquo ; to extend from to and to be continuous ( as regards ) with the media bounding .
In the layer is a function of only .
It will be convenient to put ( 1 ) tions parallel to the plane of incidence ( cf. p. 50 In the layer where is a function of only , The equation for ( see ( i ) .
p. 49 ) is .
Thus Put and , and we get .
( ii ) This equation , of course , cannot be solved completely until we know as a I'unction of ( and therefore of ) , i.e. , until we know the law of variation of in the layer .
However , in all cases to which we shall apply the solution , is a small quantity as will be shown later ) , and we can thus solve ( ii ) by approximations .
We get a first approximation by ecting d , so that ( ii ) becomes , whence and where is a function of " " Lehrbuch der Optik Transl .
Mann and Millikan , p. 288 .
[ somewhat similar analysis has been employed by L. Lorenz.\mdash ; Sec. ] Dr. R. C. Maclaurin .
Theory of the [ Jan. 17 , As a next approximation , put , and the equation for is Integrating , we get where ; ; A second integration then gives where ; ; When all the 's are zero , and when we have ; ; ; ; ; ; ; where E. . .
are constants depending on the law of distribution of in the layer .
Thus , when we have ; ; and when ; The boundary conditions give ; ; Eliminating A and from these , we get two equations to determine and It will appear later that in most cases we can safely neglect terms containing and higher powers of .
If we do this we get ; , to our order 1905 .
] Reflection of Light the Polarising Angle .
where and Thus ( to this order ) the amplitude is the same as in case of an abrupt transition ; but there is a chan of phase .
Similarly where ) , and at the polarising angle , where , and we have and Here again , in the general case , the amplitude is ( to our order of approximation ) the same as that given by Fresnel 's formula , but there is a small of phase depending on .
The most marked departure from Fresnel 's formula , however , occurs polarising angle .
In that case does not vanish as Fresnel gives , but there is a small amount of residual reflected and the change of phase is a quarter wave-length ) .
At the polarising angle ( 2 ) We shall consider in a similar way the case of vibrations to the plane of incidence ( cf. p. 51 ) .
1 the layer ' ; ; : ; Dr. R. C. Maclaurin .
Theory of the [ Jan. 17 , These equations are satisfied if .
Hence , or .
( iii ) .
Solving ( ui ) by approximations , we get first , so that and , as a second approximation , , where Now Hence on integrating ( iv ) , we get and , where When , and when G. Ihus we have when ; ; and when ; The boundary conditions give ; ; Eliminating A and from these equations , we get two equations to determine and .
If , as before , we retain only the first power of , we get 1905 .
] Reflection of the Polarising Angle .
Thus where and Similarly where As before , the amplitude is the same as that given by Fresnel , but there is a small change of phase .
If be the difference of phase between the parallel and perpendicular vibrations for reflection , and that for refraction , we have , since and are small .
except near the polarising angle , where and The above results show that and are the same as given by Fresnsl 's formulae , except in the neighbourhood of the polarising angle when is not , but is very small .
At the polarising angle we have We see from this that is a along with , i.e. , when .
If had this value throughout , we should have .
R. C. Maclaurin .
Theory of the [ Jan. 17 , This enables us to obtain an upper limit for the value of necessary to produce the observed ellipticity at the polarising angle .
Taking Kurz 's experiments on reflection from glass into air , we have , .
This gives and With Jamin 's experiments on reflection romp diamond into air , which gives and .
In the case of otYlass d is about one-hundredth of , while with diamond is about one-tenth of .
In the latter case it would not be satisfactory for accurate work to neglect in the determination of the ellipticity , so that it may be advisable to conduct the calculations to a higher order of approximation , retaining We have ; .
where and 1905 .
] Reflection of Light near the Polarising Angle .
Similarly where and at the polarising , where and .
For vibrations perpendicular to the plane of incidel ] , we have , in like nlanner , Dr. R. C. Maclaurin .
Theory of the [ Jan. 17 , and where and ?
It will be observed that the retardations of phase are the same as to the first approximation , except that at the polarising angle is instead of On examining the various formulae thus obtained , it becomes apparent that it will not be possible to calculate the constants , etc. , so as to fit in with experimental results with much accuracy .
Of the quantities depending on these constants it is only ( the elliptioity ) and the retardation of phase that are large enough to be measured with accuracy , and even with these the departure from Fresnel 's formulae is appreciable only within a few degrees of the polarising angle .
Hence , instead of applying these results directly , we shall derive from them much simpler formulae for and \mdash ; formula which involve only a single arbitrary constant , and which colligate the experimental results well within the limits of elTors of observation .
* In the expression for the factor 2 is equal to .
Now F-E is not very large , for cannot be greater than , nor less than .
Thus for a variation of * See the graphs on pp. 64-5 .
1905 .
] Reflection of Light near the Angle .
a few degrees in , the expression is practically constant .
Pntting , we have these simple formulae for tnn .
( A ) ( B ) If we put , then we have very approximately , ( the of Fresnel ) .
( C ) denominator in will vary very little throughout a considerable range on each side of the polarising angle , so that hout this will be very nearly constant .
Hence , if we draw a raph to represent the amplitude and phase of the reflected light ( the vibrations being parallel to the plane of incidence ) , we very approximately a line parallel to the axis of , and all the points of departure from Fresnel 's formulae are indicated\mdash ; as ards both amplitude and phase\mdash ; by the ; of a line from the axis of to a parallel position .
* We shall apply these formulae to the case of reflection and refraction where the media diamond and air , and compare the theoretical results with in 's imeI ) on reflection with these media .
We have ; the polarising angle is Taking , we get the following table for the ellipticity in the hbourhood of the polarising angle , the theoretical ] calculated from the formula , above:\mdash ; *See graph , p. 65 .
Dr. R. C. Maclaurin .
Theory of the [ Jan. 17 , Outside of this range the ellipticity is given sufficiently accurately by Fresnel 's formula , as the following table will show : The following table ives the retardation of phase calculated from the formula ( A ) of p. 61 .
The column headed " " \ldquo ; gives the retardation expressed as an angle , that headed " " \ldquo ; gives the retardation in fractions of the half wave-length:\mdash ; If , we have the following values of and calculated from the formula ) of p. 61:\mdash ; 1905 .
] of Light near the Polarising Angle .
It has been remarked before thab the constants cannot be determined very accurately .
We have seen that is less than Further , we have , and since and must be positive , we must have , so that .
If , this ooives , so that must lie between and We shall take , and calculate , and from the formulae of pp. 55 and 57 .
following table ives the value of and the difference of phase ( as a fraction of the half wave-length ) between the parallel and perpendicular vibrations , compared with Jamin 's observations on reflection:\mdash ; Dr. R. C. Maclaurin .
Theory of the [ Jan. D7 , Reflection of Light near the Polarising Angle .
VOL. LXXVI.\mdash ; A.

