]\gt ; [ 823 ] XIII .
On the Reflection and Refraction of Light .
G. A. SCHOTT , , B.Sc. , Scholar College , mbridge .
by R. T. GLAZEBROOK , M.A. , F.R. Received November 29 , 1893 , \mdash ; Read Jannary 18 , 1894 .
Introdnction .
object of the following paper is to exsmine how far the hypothesis of a thin fair of transi ion between transparent media will explain in debail the phenomena connected with the elliptic polarization by reflection at the boundary of two such media .
This problem has been approached by the following writers : .
LORENZ , ePOGGENDORFF nalen , , p. 460 ; RYN VAN ALKElfAADE , 'WIEDEMANN , p. 23 ; and P. DRUDE , ' WIEDEMANN AnnaleI 34 and 36 .
LORENZ starts on the basis of elastic solid theory , assuming that FRESNEL'S formulae hold for a very small change of refractive , and deduces expressions holding for a finite change of refrangibility , which are slight modifications of FRESNEL'S formulae , and clearly unsound , since a rigid elastic solid theory must lead to formulae , and not to FRESNEL'S , as a first approximation .
FRESNEL'S formulae ought not without examination to be assumed to hold even for a very small change of refractive index , for the rate of change of refrangibility in crossing the boundary ust be vely rapid in order to produce a finite change , in a distance of the order of a -length .
VAN RYN VAN ALKEMAADE treats only of the electromagnetic theory of night\mdash ; successive approximation .
His expressions the change of phase are the same in the following paper , namely ( with notation changed from his ) , and the amplitudes he gets RI 28 .
12 .
94 which are incomplete , taking no account of other terms of the same order \mdash ; ( see p. 849 , seq P. DRUDE treats the subject from the of VOIGT'S elastic solid and obtains analogous ] .
He uses KIRCHHOFF'S boundary conditions , and these are at best hypothetical , his method is not perfectly satisfactory .
In the following paper the employment of more or less hypothetical conditions is avoided by supposing the medium continuous , the transition taking in a variable layer of small but finite thickness , and solutions of the equations vibration are obtaitJed in ascending powers of the thickness , which expressions are least as convergent as the geometric progression whose ratio is , where the thickness of the variable layer , is the greatest value of the refractive occurl.ing in it , and is the wave-length of light .
Expressions are then found the intensities and phases of the reflected and refracted light , taking into terms of order The consequences are examined ) of a rigid elastic solid theory , the theories of VOIGT and K. PEjARSON , and of the electrolnagnetic theory Lord KBLVIN'S contractile ether theory , which lead to the same result .
The elastic solid theory gives modifications of GREEN'S expressions , even when refi.active index of the pressural wave differs from that of light , and cannot be to agree with experiment .
The eIectromagnetic and contractile ether theories lead to CAUCHV'S expression , the ellipticity being variable , and these agree very well with S1 .
Genered of Vibration .
It will be well briefly to recapitulate the systems of equations which have proposed to represent the periodic disturbances to which light is due .
gnetic Theory.\mdash ; Let represent components of electric and magnetic force for a periodic disturbance at the point of the medinm , where its specific inductive capacity is K\mdash ; the real parts of complex expressions being taken in the usual way .
Also let the velocity of tion of electromagnetic disturbance in vacuo be .
Then the equations of are ( and two similar pairs ) , whence ( and two others ) These are given by HERTZ Ueber die Grundgleichungen der MR. G. A. SCHOTT ON THE REFLECTION AND REFRACTION OF LIGHT .
82 .
Ann 40 ) , as according with experiment for heterogeneous media in the absence of free electricity .
But following Lord RAYLEIGH Electromagnetic Theory of Light ' Phil. Mag 1881 ) , I have put the magnetic permeabiliby in HERTZ'S equations , so as to make them give results agreeing with experiments on reflection of light and on the scattering oflight small particles .
There are also electrical experiments to justify this course , due to HERTZ , and showing that the phenomena of , at any rate , quick vibrations , are independent of the magnetic permeability of the medium .
Elastic Solid Theory.\mdash ; Let represent components of displacement at the point of the medium , where the effective density is , the rigidity is , and the bulk-modulus is .
Following Lord RAYLEIGH On the Scattering of Light by Small Particles 'Phil .
1871 ) , we shall suppose the same in all bodies , and therefore constant throughout the variable medium considered .
Then the equations of vibration in LAME'S form are ( and two others ) . . . . .
( II .
) .
These equations include the results of the more general bheories of and of K. PEARSON .
VOIGT Theory des Lichtes fur durchsichtige Medien 'WIED .
Ann 19 , p. 873 ) neglects the first pressural term , and replaces respecbively by and , that is , makes the effective density and rigidity depend on the period .
K. PEARSON Generalized Equations of Elasticity 'Proc .
Lond. Math. Soc vol. 20 , p. 291 ) replaces , and by and by Thus , in the general case , are functions of the period .
There are two principal forms of elastic solid theory\mdash ; First.\mdash ; GREEN'S Theory\mdash ; which attempts to get rid of the longitudinal ( pressural ) waves by a kind of total reflection at all but very small angles of incidence , whilst at nearly normal incidence their effect is inappreciable owing to the smallness of the normal component .
The bulk-modulus is made very large , the expansion very small , the pressure is finite\mdash ; it is not necessary that be greater than make the effect of the longitudinal wave inappreciable .
( GLAZEBROOK , .
Report on Optics , ' 1885 , p. 192 .
) Secondly.\mdash ; Lord KELVIN'S Contractile Ether is made zero , so the longitudinal wave is not propagated from any place where it may arise .
Putting aero for in equations , they become of exactly the same form ns quations ( I. ) for the electromagnetic theory .
These two theories will , therefore , considered together .
shall suppose the medium in which the disturbances take place to be perfectly continuous , though its qualities may vary from place to place .
It follows must be continuous functions of ( xyz ) , as well as their first coefficients , and this condition must replace boundary conditions at places where ture of the medium changes , however rapidly it may do so .
S2 .
in Variable between two Media .
For our purpose it is only necessary to consider the very special case when heterogeneous medium is arranged in plane ]ayers , perpendicular to Ox suppose , we shall further suppose the variable portion to be a thin layer separating two of different but constant quality , into each of which the layer passes continuously .
We shall suppose plane waves incident in the first medium , which will give rise plane reflected and refracted waves .
Take Oz perpendicular to the plane of incidence ; then Oy will be parallel to the intersections of the plane of incidence with the layers , and since the traces of all the waves on the plane layers must move along layers at the same rate , the coefficient of must be the same in the expressions the dffierent waves .
Let now be the wave-lengh in vacuo of the light its refractive index vacuum into the variable Iayer , the values for the two media on either Then we have ? .
If be the angle the wave-normal makes Ox , the coefficient of in the expression of the wave will be being everywhere the same , we shall write Also write , so that the velocity of the pressural wave is times of the transverse wave .
By what has been said above , are fuIJctions of only .
And the displacements , \amp ; c. , are perldent of , and proportional to .
First medium .
This is of constant quality and extends from to ; equations ( II .
) become Vibrations parallel to of incidence .
MR. G. A. SCHOTT ON THE CTION AND OF .
827 Vibrations perpendicular to plane of incidence .
These are satisfied by . . . .
Here the first term represents an incident wave , the second the reflected , and the third a pressural wave , which last travels along the boundary , and rapidly diminishes away from that boundary .
is the angle of incidence ; is a constant which is found to .
are complex constants ; are the amplitudes , the tions of phase , of the reflected and pressural waves .
The pressure is proportional to ; thus vanishes for the electromagnetic theory and for Lord theory , which vanishes .
Second Medium .
This is of constant quality and extends to ; equations ( II .
) are to incidence .
Vibrations perpendicular to plane of cidence .
They are satisfied by The first term represents the refracted wave , the second the pressural wave , only exists close to the boundary is the angle of refraction , which obeys SNELL'S law , is a constant found to .
are complex constants are the amplitudes , the retardations of phase , of the refracted and pressural waves .
The pressure proportional to The signs of are chosen so that at normal incidence shall each shown in the figure , when the signs of are Variable It extends from to and is continuous with the media bounding it .
The displacements and their first differential coeflicients with respect to must have the same values at the boundaries in the variable layer and in the media beyond , iving in all twelve boundary conditions , six of which determine the motion in the variable Iayer , nd the six determine the constants for vibrations arallel and perpendicular to the plane of incidence .
We may write the displacements in the form write also ; then are functions of alone , and is proportional to the pressure , which vanishes for those theories which make zero .
The equations ( II .
) become of incidence .
brations parallel to plane MR. G. A. SCHOTT ON THE REFLECTION AND FRACTION OF LIGHT .
829 .
Vibrations perpendicular to plane of incidence .
We shall choose , so that , when , and when , where \ldquo ; The six conditions determining . .
will be When .
When S3 .
of the cements fthe Variable It is in general impossible to solve the equations in finite terms ; in the physical problem the transition layer may be considered thin even in comparison with the wave-length , and the equations can be solved in very convergent series , proceeding in ascending powers of some small quantity depending on the thickness of the variable layer .
This quantity we shall take to } ) @2d/ .
Putting also for value of will lie between and 1 ; and the equations become \mdash ; @ .
Vibrations parallel to plane of incidence .
( III .
) , Vibrations perpendicular to plane of incidence , and when ) ; and when It will be necessary to treat separately the cases of the netic and contractile ether theories on the one hand , for which zero , and of the elastic solid theory on the other hand , for which is very large and is finite .
this latter case we shall nsglec in the ] terms , which are themselves only corrections due to the finite , though small , thickness of the transition layer .
vibrations to the plane of incidence ( theories ) .
The equation to be solved is , p. 829 , with the conditions that when , when Put .
, where ) , . . .
and when . .
, when These give , . . .
whence , . . .
Let a bar-written over a quantity denote its greatest numerical value between and , e.g. , the maximum refractive index .
is given by , or by .
Hence where hes between and 1 , and therefore .
Now , and therefore and .
It lows that which is finite as long as S or .
We shall neglect powers of above the second , so that we may write Then MR. G. A. SCHOTT ON THE REFLECTION AND REFRACTION OF LIGHT .
831 Hence ( V. ) .
Vibrations parallel to of incidence Contractile Ether Theories ) .
The equations to be solved are ( III .
) , p. 829 , with zero , \mdash ; @ whence eliminating with the conditions that when , and when Put where , . .
and when , when Write so that increases as long as and maximurn of .
We have at once .
whence , .
with the same notation as before .
hence , and which is finite as long as or As before we neglect and higher powers , and find , since , and thus .
Hence ( VI .
) Vibrations to the of incidence ( Elastic Solid Theory ) .
The equations to be solved are ( II1 .
) , p. 829 , adding the third multiplied by to the second for a new second , \mdash ; \mdash ; @ MR. G. A. SCHOTT ON THE REFLECTION AND REFRACTION OF LIGHT .
with the conditions that when , and lvhen Put . .
, . .
, . .
, where and when , when We have at once 1 \mdash ; \mdash ; .
In the same way as before denoting maximum value of ' have , and since . .
or if whence and this is finite , and so also are , as long as the denominator does not vanisb .
Writing this denominator in the form , we have Now is large , at least 10 , and is at least , hence is very small negative , in which latter case is small ; thus , , the larger of the two , .
Hence are finite as long as or We shall , as before , neglect , . . .
, but it will be necessary to go to order in in order that the result should be correct to ; we shall also neglecb when iplied by , since is about 100 ( ' B. A. Rep 1885 , p. 192 ) .
We have being multiplied by .
G. A. SCHOTT ON THE REFLECTION AND REFRACTION OF LIGHT .
835 neglecting the terms in involving \mdash ; \mdash ; , neglecting Thus , we have finally @ ( II ) .
The conditions determining , . .
are Vibrations perpendicular to of incidence , when , Vibrations parallel to incidence Electromagnetic , when , Elastic Solid Theory , when when when ; when MR. G. A. SCHOTT ON REFLECTION AND REFRACTION OF LIGHT .
Write I Then there are the following relations between these constants\mdash ; \mdash ; Hence so that for And Using this notation we have for to Plane Incidence\mdash ; , 838 .
G. A. SCHOfT ON REFLECTION AND REFRACTION OF LIGHT .
gnetic Theory .
Vibratiom Parallel to Plane of Incidence .
, .
( VI ' .
) .
Elastic Solid Parallel to the Plane of Inc , idence .
S S .
( VII ' .
) .
where We found also that the series converge ] east as rapidly as the geometrical progression .
The series ( VI ' .
) converge at least as fast as and the series at least as fast as where denotes the greatest value has in the variable layer .
The greatest refractive index for transparent substances ( excluding metals ) in Greenockite , and has the value .
Taking this value for the three ratios MR. G. A. SCHOTT ON THE REFL CTION AND REFRACTION OF LIGHT .
, and 20 .
If for we take they are , ; if we take , the value for water , the ratios are REINOLD and RUCKER found for the thickness of a black soap-film to cms .
, for red of first , order , blue of second order ; these are in wave-lengths of yellow light , .
It follows that for tho series to converge at all , the thickness of the film must beless than that of a soap-film giving the red of the first order .
S 5 .
Equations Determining the onstan t , .
Vibrations Perpendicular to the Plane of Incidence .
The equations ( V ' .
) ( p. 837 ) give , on substituting for or , whence or @ - , since .
Similarly \mdash ; S Since , we get , by changing into \mdash ; , multiplying and dividing and and Hence \mdash ; 2 or using the equations and ( p. 837 ) or using the same conditions as for , neglecting , \amp ; c. , neglecting MR. G. A. SCHOTT ON THE CTlON AND RACTION OF LIGHT .
These give at normal incidence Parcdlel to of jidence ( Ether Theories ) .
The equations ( VI ' .
) p. 838 , give on substitution for , since , in this case , are zero , or , , whence @ Sa @ @ , or , CCXCIV. .
51 ' @ @ and , since ( p. 837 ) .
Since , changing into , multiplying and dividing , we have .
MR. G. A. SCHOTT ON THE CTION AND REFRACTION OF LIGHT .
843 Hence .
since \mdash ; ( p. S37 ) , where 844 lf G. A. CHOTT ( REFLLCTION AND FRAUTION OF LIGHT .
When , we have the same as for vibrations perpendicular to the plane of incidence .
When , we have according as to of incidemce ( Elastic Solid Theory ) .
The equations ( VII ' .
) p. 838 , give on substitution for , .
- \mdash ; 1 \mdash ; - ( 1 \mdash ; r ) \mdash ; S. S. .
icos } \mdash ; S \mdash ; S. 1 \mdash ; .
\mdash ; here we lmve throughout neglected except in of orders MR. G. A. ON LCTTON AND FRACTTON LIGHT .
We get , collecting the terms , S(F \mdash ; . . . . . . .
( 1 ) , . . . . . . . . . .
( 2 ) , S \mdash ; S \mdash ; ( 3 ) , ( 4 ) .
In working with these equations we shall throughout neglect S , \amp ; c. , \amp ; c. We may thus in terms of order , interchu , nge and Multiplying ( 1 ) by and subtracting ( i3 ) we Ssin ( 5 ) .
Similarly , from ( 2 ) and ( 4 ) \mdash ; S \mdash ; S Ss ' Multiply ( 1 ) by \mdash ; S \mdash ; ( 2 ) \mdash ; S and add , nsing the relations \mdash ; , , lsnce terms in . .
, and neglecting terms of order 1 , , . .
we have SA ' .
( 7 ) .
Again , multiplying ( 1 ) by \mdash ; @ ( 2 ) by and subtracting , we find MR. G. A. SCHOI'T ON Rh'F CTION AND RACTlON OF 847 ( 8 ) .
So . .
from ( 7 ) and ( 8 ) for in ( 5 ) the , we find \mdash ; 1 - cos2 S \mdash ; In the same way we get from ( 6 ) \mdash ; Ssin - @ S Solving these equations we get , after some algebraic transformations , using the values in terms of order , and finally discarding a common factor , S. @ \mdash ; @ ; and , in the same way , MR. G. A. SCHOTT ON REFLECTION AND REFRACTION OF LIGHT .
849 @ \mdash ; 8 .
\mdash ; \mdash ; We easily find Q NDCCCXCIV .
850 MR. G. A. SCHOTT ON THE ]]iCTION AND REFRACTION OF LIGHT .
, the equations become , dividing out common factors , , and neglecting , .
as before , -@ S. \mdash ; @ @ and \mdash ; Now ; hence changing into - , multiplying and dividing corresponding equations , and , as before , neglecting ( and , therefore , \mdash ; we find MR. G. A. SCHOTT ON THE FLECTION AND RACTlON OF .
851 @ .
\mdash ; S. .
\mdash ; S. .
M. \mdash ; 2S \mdash ; @ MR. G. A. SCHOTT ON THE REFLECTION AND RIIFRACTION OF LIGHT .
Hence , we have finally , using the values of ( X. ) .
and And here as long as , and MR. G. A. SCHOTT ON THE REFLECTION AND REFRACTION OF LIGHT .
853 These give , when the same as for vibrations parallel to the plane of incidence , as should be the case .
At the polarizing angle , when , we have , since then , \mdash ; 2 S6 .
of Results .
We shall shortly summarize those results that are of use for comparing with experiment .
Vibrations to plane incidence .
Plane of polarization parallel to plane of incidence .
These give the sine-formula of , which holds for parallel polarized light .
( VIII .
, p. 840 ) . . . .
MR. G. A. SCHOTT ON THE REFLECTION AND REFRACTION OF LIGHT .
parallel to plane of incidence .
Plane of polarization perpendicular to of incidence .
These correspond to FRESNEL'S tangent formula for perpendicularly polarized light .
Electromagnetic and Contractile Ether Theories .
Elastic Solid Theory . .
.sin2 .
where provided si There is a point here which calls for remark , viz. , as to the quadrant in which are to be taken .
Neglecting @ , in the equations ( V ' .
) of S5 , p. 839 , we find MR. G. A. SCHOTT ON THE REFLECTION AND REFRACTION OF LIGHT .
large part of to be \mdash ; \mdash ; \mdash ; which is negative , so that for parallel-polarized light at normal incidence the vibration in the reflected light is opposite to that in the incident at the reflecting surface , so that there is a retardation of phase Similarly , the equations give for the important , part of for perpendicularlypolarized ight - which is negative at normal incidence , but positive for ' incidences greater than the polarizing angle .
We shall suppose be taken equal to the absolute values of the above ratios .
Then will lie between and , or between and , according as , and will differ from by an nount of the order S. The same will apply to , whose difference from , however , does not remain of order S , but which increases t , hrough to , or decreases through to , according to the sign of .
The difference \mdash ; the retardation of phase of the perpendicularly-over the parallel-polarized light\mdash ; is positive or negative according as is positive or negative , and increases 'numerically from at normal incidence at polarizing angle to at grazing incidence .
And the reflection is said by JAMIN to be positive or negative as the case may be .
If a ray of elliptically-polarized light be reflected normally from a surface , then the difference of phase of he components , and the position of the axes of the vibrational ellipse , as well as the direction of its description , are all unchanged in spa , ce , but with reference to the direction of propagation , and , therefore , also to an observer viewing both rays , the position of the axes has changed into one symmetrical to the former one , with respect to the plane of incidence , and the ray from being right-handed has become left-handed , or vice versed .
Thus , there is an apparent change of phase of which is called by JAMIN de retournement and causes him to give he measured diffel.ence of phase as lying between and , instead of between and We must also consider the effect of a finite , though large , velocity for the pressural wave in the Elastic Solid Theory .
We have made no supposition as to the values of the , the ratios of the pressural-wave velocity to that of light the different media , except that these ratios are large .
The ratio may have any value , so that the refractive index for the pressural-wave between the two media may also have any value .
The effect of the pressural-wave is to add to a quantity , for moderately-large values of , such as are used in most of the experiments ; in , also , there is an additional , which at no angle of incidence is of magnitude more than comparable with Now is large , perhaps 100 , as above , S3 , p. 834 .
The term in ) may always be neglected ; and at but very small angles of incidence be put equal to .
This result is in agreement with GREEN and in opposition to HAUGHTON , who proposes to make ' where , ascribing it to a difference between the refractive index for the pressural-wave and that for light ; but it hss been shown above that no such difference could diminish and therefore M. The formulae VIII .
, IX .
, and X. can be put into a more suitable form for calculation ; the quantities experimentally determined are usually and .
In doing so we powers of S above the second and make use of SNELL'S law and the equations and With the same notation for the constants of the variable layer as before , viz. , thickness , Srefractive index , and mean value of since these enter into the expressions in different combinations , we shall introduce lifferent set of constants , involving , and together with , and defined by the following equations\mdash ; Then the expressions for , and become\mdash ; MR. .
A. SCHOTT ON THE REFLECTION AND REFRACTION LIGHT .
857 Electromagnetic Contractile , Ether XI .
These expressions are true as far as order , provided greatest value of occurring in the variable layer .
Except in the neighbourhood of the polarizing angle , reduces to % Elastic Solid Theory .
Here we introduce subsidiary angles defined by the equations Mtan , where Then we have D. .
( XII .
) or * The expression for inclusive of involving is of the form ' , . . .
, . .
being constants of order .
Since is large only in tho neighbourhood of the izing angle , we may put in the small terms , thus obtaining the expression in the .
Then 858 A. SCHOTT ON FLHiCTTON AND RACTION OF LIGHT .
These expressions are true as far as order , provided greatest value of To get some of the thicknesses of the film , let us compare them with soap-films ; INOLD and RUCKHiR estimate the thickness of a black soap-film at aoout centim. , that of a film showing red of the 1st order at about ' centim. Hence for black soap-fihn is , line A 6 red of 1st order Since the refractive indices of transparent substances lie between 1 and 3 , it follows that a transition to which the above analysis is to be applicable must certainly be less than necessary to show even a red of the 1st order .
of \mdash ; Elastic Solid The expression found for the change of phase is by ( XII.)\mdash ; where , , , and is a disposable constant .
denominator of may be writte1r and this cannot vanish even to order unless Now , gives , or , whence instead of BREWSTER'S angle .
In order that we should obtain BREWSTER'S angle it is necessary that should be only a small fraction of , which would giye .
This , as is wel ] known , was pointed out by HAUGHTON , who thought it possible that a small ] effective refractive index for the pressural-wave would lead to such a value of , but the rigid theory developed above , which includes the most general theory possible , ccording to VolGT , without absorption , shows that any alteration in the refractive index for the pressural-waves consistent with keeping their velocity of propagation large could only produce a very slight change in the value of M\mdash ; and that an increase\mdash ; except at very small angles of incidence .
It is clear then that a rigid Elastic Solid Theory cannot explain the change of phase at reflection .
MR. G. A. SCHOTT ON RE CTION AND ' HT .
85 gnetic cEther Theories .
Here the expressions XT .
) for the amplitude and change of phase at contain four constants , of which the effective ones , E. The constant A in the expression for the ratio of the amplitudes is multiplied by , and thus is without effect at the polarizing angle , at which the deviation from FRESNEL'S formula is most marked .
A cannot therefore be determined with any great accuracy , seeing that a considerable change in its value produces only a very slight effect on the result .
In some cases it may be put zero without impairing the accuracy of the formula .
The same considerations apply to the constant in the expression for the phases .
The other two constants ought to satisfy the condition , .
B. As regards accuracy of determination the order of the constants is , A. The experiments discussed are those of JAMIN on solids and liquids ( see his two papers , ' Ann. de Chimie et Physique , ' se'rie III .
, 29 ( 1850 ) and 31 ( 1852 ) ; a series flint-glass by KURZ ( ' POGG .
Ann. , ' 108 ) , and some of ( ' POGG .
Ann 128 Of these the experiments of JAMIN are much the best , are almost as well represented by the empirical formulae of CAUCHY as by theoretical formulae bove .
This might excite surprise\mdash ; seeing that CAUCHY'S formulae volve only one independent constant , the ellipticity did we not relnember that of the three independent constants ( of course is not independellb ) , two , and , do not have much influence on the result .
The experiments of QUINCKE are the most irreoularb , but they are of interest because QUINCKE in yestigates the reflection in each othel from the bounding of pairs of media .
Of these I have only taken those in which there are ten or more different detel.tninations , where there is some chance of the constants being accurately termi n. The of HAUGHTON ( ' Phil. Trans 1863 ) I have not had time to consider , but , with but one or two exceptions , his series consist of too few determinations to allow of an accurate deterlninatio of the constants .
In all the above cases measurements were of the difFerence of se , by of a BABINET'S compensator , directly , and of the ratio of the intensities , ectly .
The polarizer was placed at a large angle with the plane of incidence , so in the incident beam the component polarized perpendicularly to the incidence is of great intensity relative to the parallel .
The azimuth of the reflected light was determined .
Then is giveo by the equation .
By this means the determination of is endered more accurate , firstly , because the absolute error in is made much less than of owing to the largeness of , and secondly , because the determination of is itself more accurate , the intensities of the components in the refiected more nearly equal .
In combining the experiments I have assumed as a first approximation that the accuracy is the same for all values of , the of phase , and likewise for all values of .
In strictness this is not true , since the accuracy of the readings ia greater the more nearly equal are the intensities of the two components of the reflected light .
But as in most cases ranges from above to below , the ssumption will be sufficiently true to give values of the constants not far removed from their most probable values .
The sets of constants A and , and and have in each case been determined independent]y by making the sum of the squares of the errors in and respectively , a minimum .
We have by ( XI .
) @ Let S , be the true , S ' , the observed values , and let be approximate values , given tentative values of the constants .
Let being small quantities to be determined by the conditions 2 minimum , minimum .
, substituting for S , their values in the equations and measuring S ' \mdash ; S in degrees , we obtain , neglecting squares of small quantities\mdash ; , MR. G. A. SCHOTT ON TfON A ) LIG and here we may , in the coefiicients , replace , by S ' , wherever convenient .
We thus find This is the method used in cases , but the more it was Qier to find the sums of the squares of for values of the constants , and thence , by a kind of interpolation , to find best values of the constants .
Since the values of are determined independently , the nearness with which they satisfy the relation will serve in some measure as a test of formulae .
I have for comparison given the deviations from formulae , calculated the given value of by the imenter himself .
These roughly parallel with the deviations from the theoretical values , and where there seemed any very great deviation from parallelism , I have recalculated the results of CAUCHY'S formula .
For instance , JAMIN , for fire-opal , gives incorrect values for his .
On recalculating from the given values of , some of his values found to be the square roots of what they should be .
As an index of the accuracy of agreemeut , I have given the probable error of single observation , as calculated by the formula , where is number of observations , the sum of the squares of the errors .
MR. G. A. SCHOTT ON THE REFLECTION AND REFRACTION OF LIGHT .
877 The expressions for are given on p. 856 .
Considering the values of , given on the same page , we see that do not when the two media on either side of the variable layer are merely challgessign , and does the same\mdash ; this is evident from the physical meaning of ; we have to take an element and another element , form the expression , multiply by the product of the elements , and then sum , first , for elements on that side of from which the light comes , and , lastly , for all the elements of the film ; callin the result , according as the comes from one or the other side of the layer , clearly in forming the sum we must sum for all elements and for all elements vanishes , since the two elements of the integral for two points destroy each other .
An inspection of the values of that they should have the same values , whether the light comes from one side or the other\mdash ; provided , of course , the layer remains the same .
As for \mdash ; ecomes becomes . .
It has already been stated that are not independent constants ; by theol.y we have A comparison of the values of and , as given in the table , p. 876 , shows that this last condition is , with few exceptions , very nearly fulfilled .
The chief exception is in the case of essence of lavender , where is 000027 , whilst is 000065 , but this is sufficiently accounted for by the smallness of , and the consequent smalIness of and , which makes a small error in the determination of important relatively to the magnitude of B. The large differences in the last four pairs in the table on p. 876 may be due to terms of the third order in , but these sets of experiments not very accurate , the contact of liquids solids being irregular in character .
Of the two constants , is determinable with much the greater accuracy , since the variations from FRESNEL'S formulae , which are given by all the constants zero , are much greater for the phases than for the intensities , but it is not to say weight should be attached to each determination .
I myself should prefer to rely solely on the value of , and thence calculate ; this will llot very much alter the values of , which chiefly determined by the values of This is confirmed by the experiments of KURZ on flint-glass in air , where formula is seen to give nearly good a of the inteusities theoretical formula and that of CAUCHY .
The only experiments bearing on the relations between the constants for reflection from either of several of media are those of QUINCKE for , flintglass\mdash ; water , and crown-glass\mdash ; air .
These experiments are very irregular , as shown by the very large " " probable errors\ldquo ; occurring in all except the first .
QUINCKB himself admits that he could not attain to the accuracy of JAMIN and even of KURZ , and , as already stated , in order to make CAUCHY'S formulae fit at all , he has to use a different value of from ] which determined in the ordinary way .
instance , for flint-gJass\mdash ; air he uses in place of , for flint-glass\mdash ; water in place of , and for crown-glass\mdash ; air instead of .
He gives seve , ral other sets of experiment , in addition to these , but they consist of few observations and are very much more unreliable still .
As ated ( p. 877 ) should be the same for the two sets of riments on each pair .
In the case of A this is certainly not true .
For flint-glass\mdash ; air flint-glass\mdash ; water they are of opposite sign .
The determination of A depends almost entirely on the extreme terms of the series of observations , for it is multiplied by , which is very small for the middle terms .
Now the extreme observations in these experiments of QUINCKE'S show very large errors indeed , in some cases of lmore than a degree in , and are not to be much relied upon .
The entire extinction of A would not make a difference of more than a few minutes , and if we decide to letain it , little stress can be laid on its not satisfying the theoretical conditions .
* The case of and is much more important , as the deviations from FRESNEL'S formulae depend on them to a first approximation .
*On the the Constants are determined .
The expre sious o p. 860 give Let , cotlbidor tbc effect of small errol.s of 10 ' in , and of in , say for a glass such as used by KUBZ ( p. 871 ) at an angle of of this angle is about is about or We find being surod in The circular leisure of is 0029 , that of of or 1o 8 ' \mdash ; is 0200 .
Thus , cc uld be by , or by and of could be produced by by } Thus , .
G. A. SCHOTT ON THE CTION AND FRACTTON OF LIGHT .
879 The values of for the three are Reflection from flint-glass in air air in flint-glass 0050 , ] , , flint-glass in water 0120 , , , water in flint-glass 0100 , , , crown-glass in air 00040 , \ldquo ; air in crown-glass , , Here , of course , is the absolute refractive index of the second medium .
It will be seen that the relations and are satisfied with fair accuracy for the first two pairs , whilst for the third they are only of same order of magnitude .
According to CAUCHY and JAMIN the ellipticities ' in such cases ought to satisfy the relation - .
These ratios are , and instead of , and , and the agreement is less than for our theoretical formula .
Of course it has throughout been assumed that the nature of the film of transition is the same in both sets of experiments .
The outstanding difference from agreement may possibly be due to a change in the film .
DRUDE ( ' WIED .
Ann 38 , p. 35 ) by observations on cleavage faces of calc-spar has shown that there is in that case a gradual change in the elliptic polarization during exposure , so that part of the effect at least must be ascribed to condensed air or dust , it is quite possible that such ayer would be affected by atmospheric conditions .
Without some assumption as to the law of variation refractive index in the layer , there is no relation between the constants for sets of media other than those given above .
Theoretically CAUCHY'S constant for reflection from medium ( 1 ) in medium ( 2 ) should satisfy the equation , but this is very A is determined with au accuracy only about 71 that of and In the experiments just quoted , the error\ldquo ; of is about , that of of .
Hence , in this case the accuracy for is only about , or that of But in most cnses the dispariby not so great .
The last of the second order in ' and in most cabcs is from bhe exc.eptions being fluorspar\mdash ; air , essence of flint and -air , .
In the case of essence of lavender is very small .
The last four olve the imaccurate measurements of all those considered .
The eifect of is to the diff ) from angle by an amount radius or ; this for lgnr is about flint-glass\mdash ; wate about , for most other substances or so .
As it unlikcly that ] angle can be determined with an accuracy of 1 minute of arc , it is tk.at is known only far from being the case .
In so far as no such relation exists for onr theory , it has the over CAUCHY'S .
Let us now consider the values of in greater detail ; by p. 836 the values of these constants are , leaving out of account terms in ' where are the mean values of for the variable ayer respectively .
We have vanishes , when or ( of course values of do not occur ) , it has an algebraic minimum for the value of , is negative ) values of between and , is positive for values either less or boreater than both and HeIlce if for the variable layer lie between and is certainly negative , if outside those limits , certainly positive .
In any other case can be said a as to its sign , unless indeed the law of variation of in the riable layer be given .
If then lie between and , will be positive or negative\mdash ; and the same will be the character of the reflection in JAMlN'S sense\mdash ; according as the first medium is the more refractive or the less .
And the reverse holds when is outside the iven limits .
Now JAMIN'S and the other experiments show that the reflection is in most cases ( but not in all ) positive or negative according as is greater or less than In these cases , we are at liberty to suppose that for positive reflection , that is , when i4G , for the film , and that for negative reflection , when ( but ) for the film .
This shows that when the second medium is air ( as is tacitly assumed by JAMIN , otherwise the critical value might be different ) , the refractive index of the films is , for some parts at least , , and than or so , or perhaps we ought more properly to say that the average refractive index is between those limits .
KUNDT has shown that the refractive index of colcothar , or red oxide of iron , which is a common polishing material , is about ; that of chalk , I suppose , would be of the same order of magnitude as for and arragonite , that is , about glass surface , with lumps of such polishing material embedded in it , might be expected to behave as if coated with a film of average refractive index between 1.5 and 2.5 , and thus certainly give positive reflection .
Of course it has not been proved that for every part of the MR. G. A. SCHOTT ON THE REFLECTION AND REFRACTION OF LIGHT .
film must lie between the above limits , this is ] sufficient , not a necessary condition .
There is another point to be considered , the magnitude of .
It was shown ( p. 857 ) that if be less than the reciprocal of the greatest value of for film , the expressions found will be convergent .
It follows that .
These are the limits which the absolute values of must not exceed .
Consider , for instance , the case of water-flint-glass , for which If lies between and , then ; the greatest numerical value of occurs for throughout , and is ( see p. 880 ) or .
The greatest value of is given by and is therefore .
Hence we must have ; .
( 28 ) 190 .
If ( about the gl.eatest value of for a substance ) , the maximum of is ) of ) is .
Hence ; fhese expressions show that it is possible ( or at least probable ) to tisfy the conditions for convergence by conceiyable values of for the layer .
And since thess upper limits for are much wider on the second supposition , close to the actual values of on the first supposition , there is a very distinct presumption in favour of the second , namely , that the average value of the variable layer is greater than ( and less than 2.67 ) .
QUINCKE does not state whether his reflecting surfaces were polished with ferric oxide or not , but this is a common enough material .
Emery also has a higher refractive index than so also diamond dust , and some one of these would perhaps be or silica being hardly hard enough for the purpose .
* The retardation of phase of light polarized in the plane of incidence is finds that this retardation is at most a few thousandths of a wave-length , so that is probably less than 01 , and quite incapable of reasonably accurate measurement .
CXCIV .
It has ah.eady been pointed out ( p. 858 ) that the above supposition would give value of less than that for the red of the first order of thin plates , so that colours of thin plates are to be expected .
The constants , . . .
, of course vary with the colour , but leir effect , in any case , would not equal that due to variation of , .and therefore of and .
[ Owing to the secondary importance of the constants , and the impossibility of suring them accurately , it will be to take account only of in discussing bhe uitations to which any law of variation of the refractive index in the variable layer is subject .
In any particular case , the law must be such as to make continuous in value throughout the layel and to and at the two ; and to give to its experimental value by a sufficiently small choice of the thickness of of the layer to ensure convergence of the series for the displacements .
Besides , must llowhere be less than 1 , and nowhere greater than about 10 , thislast esenting the greatest vadue of known to exist for a nsparent substance .
The of tion must involve at least two disposable constants in addition to If the law is to be a as to include every known case , then it must be capable of positive itnd negative , corresponding to and negative reflection .
That is , must be capable of maxima or minima .
For example , the law of variation discussed by Lord RAYLEIGII ( ' Proc. Lond. Math. Soc XI .
, p. 51 ) will s , atisfy this condition .
In this case , we have being the distance the { irst face of the variable layer ; this gives which is always positive when the second is the more refractive .
Hence , Lord RAYLEIGH'S law will only explain positive reflection .
If the medium a refractive index 1 , then must have a maximum to give negative reflection .
If the second lnedium have acbive index to the ) limit , that is 3 or so , then must have a minimum in order to give negative reflection .
In addition the general law must make vanish , that is , when or so , in order to explain JAMIN'S results .
It follows GLADSTONE and DALE'S experiments , and others of the same kind , that the law of variation of be of the same form as that of density .
The of capillary forces will be to make the vary near the surface of a liquid , possibly also of a solid .
A somewhat problematical investigation of the law of variation of the density in the transition film between a liquid and its vapour is given by J. CLLRK his article on pillary Action , in the ' Encyclopaedia Britannica ' ( 9th Ed which gives the density of the variable portion an exponential furlction of the distance from the surface .
If snch a law represent the actual circumstances , then negative reflection must be ascribed to adventitious iilms of dust .
G. A. SCHOTT ON THE TION AND FRAC I OF LIGHT .
883 or condensed ases .
It is orth n. that wP.ter , which gives strong negative reflection when its surface is covered with grease to even a small amount , when .
clean shows hardly any elliptic polarization by reflection ( Lord RAYLEIG RoNTGEN ) .
Again , various specimens of , whose surfaces have been repeatedly cleaned by a method due to WERNICKE , of removing the polishing material by an adhesive coating of gelatine , show much greater positive reflection than when polished with oxide of iron or oxide of tin ( WERNICKE , ' WIED .
Ann , ' 30 , p. 402 , and K. E. F. qCHMlDT , ' WIED .
Ann 51 , p. 417 , and 52 , p. 75 ) .
It is clear that the effect of highly refractive surface film , either of grease or of polishing material , is to produce negative , reflection which is on the effect due to a transition between ) ether inside a 1oody that outside .
This latter we should expect to depend on the same causes that dispersion ption ( SCHMIDT , loc. cit. , p. 89 ) .
Dispersion is taken account of through the refractive index .
The absorption effect can be conveniently treated by supposing the refractive index everywhere complex of the form .
The distance in which , by absorption , the amplitude is reduced to , of its original value is .
In a metal this distance may be as little as of a wave-length , in a tran-sparent substance such as glass it may be as much as 100,000 wave-lengths .
These ctlues would give about 100 , about ooo respectively .
In the one case is , in the it is very nall compared with must be less than .
In such substances as glass , we may take account of quantities of order , but may neglect all of higher order .
The effect of absorption on the values of , is of order may be neglecbed .
The effect on is of order The new value of is where are the values of for the first and second medium respcctively .
No term of order due to the film itself occurs .
Hence any small degree of opacity in the film changes the retal.dation of phase , if at all , by a whole number of wavelengths .
WERNICKE ( WIED .
Ann 51 , p. 449 ) finds that whilst there is normally retardation of phase of waye-length , when light is reflected in glass from an opaque layer of silver closely adhering to the glass , yet the presence of a trace of dust , or air between glass and silver suffices to produce instead an acceleration of phase wave-length .
If the more refractiye medium be also the more absorptiye , as is generally the case , absorption increases positive reflection ( since ) ; and of two substances having the same refractiye index , the more absorptiye will show greater positive ( or * ' Phil. Mag. ' ( 5 ) , 1ED .
Ann 46 , negative ) reflection , when they are placed in contact with the same third substance .
This agrees with the concJusion arrived at by SCHMIDT ( loc. cit from his experiments on various crown and flint-glasses.\mdash ; July 20 .
] The above experiments are sufficient to test the accuracy of the theory , which merely assumes the existence of a film of transition , without entering into the question of its origin and constancy ; whether it be due to a surface property of the medium\mdash ; a kind 01 capillary effect\mdash ; or merely to an adventitious film of dust or of polishing powder , is of no consequence as far as the theory is concerned , its existence is the crucial ) othesis , and of that existence there can be no doubt .
The theory does however point to the idea f , hat the film may be , in part at least , of adventitious origin .
This is confirmed by the experiments of DRUDE already mentioned , and those of Lord RAYLEIGH on the reflection from pure water surfaces ( ' B. A. Repts 1891 , p. 563 ) , who finds that perfectly clean water reflects only of perpendicularly polarized light found by JAMIN , so that its ellipticity is only about of JAMIN'S value .
The darkness of the band observed in the analyser at the polarizing angle was disturbed by a small of oil applied to the surface and producing a thin film .
S12 .
Conclusion .
We may sum up the results of the preceding discussion as follows:\mdash ; ( 1 .
) A rigid elastic solid theory , proceeding on the assumption that the velocity of the pressural-wave is much greater than that of the light-wave , will not explain the experimental results , whatever be the refractive index for the pressural-wave .
( 2 .
) Lord KELVIN'S contractile ether theory and the electromagnetic theory in HERTZ'S form , lead to the same equations , containing three independent constants ( of two have little effect , except at a distance from the polarizing angle ) ; and these equations agree with the experiment rather better than CAUCHY'S empirical formulae containing , as used by JAMIN , one constant , , and as used by QUINCKE , two constants , and .
At a distance from the polarizing angle FRESNEL'S expression for the intensity is sufficient .
( 3 .
) Whilst CAUCHY'S constants , , have been found not to satisfy the theoretical conditions assigned by JAMIN ( so that CAUCHY'S formula must be regarded as an empirical one ) , the constants of the above theoretical formulae satisfy the conditions theoretically deduced , as nearly as is to be expected , considering that the whole effect under discussion is itself but a small correction .
This last conclusion as to the possibility of a theoretical explanation of the phenomena of reflection based on the existence of a film of transition is at variance with the result arrived at by M. H. BOUASSE from a critical examination of the theories so far proposed .
( See his paper in the ' Annals de Chimie et Physique ' for February , 1893 , p. 145 .
) MR. G. A. SCHOTT ON THE REFLECTION AND RACTION OF LIGHT .
885 of the difference of phase of the components of Light reflected in Diamond ( ccording to experiments ) .
The black line is the theoretical phase curve ; the crosses represent JANIIN'S experimental results .
of of intensities of Light reflected in Air from Diamond to experiments ) .
The black line is the theoretical ; the crosses show JAMIN'S experimontal resuIts .

