]\gt ; INDEX SLIP .
LEATHEM , J. G.\mdash ; The Theory of the Magneto-Optic Phenomena of Iron , Nickel , and Cobalt .
Phil. Trans. , ) , vol. 190 , pp. 89-127 .
Kerr Phenomena , Mathematical Discussion of .
Leathem , J. G. Phil. Trans. , , 1897 , vol. 190 , pp. 89-127 .
Magnetic Reflection , Experiments of Sissingh , Zeeman , and Wind , com- pared with Theory .
Leathem , J. G. Phil. Trans. , , 1897 , vol. 190 , pp. 89-127 .
Hall Eflect , for llapidly Alternating Currents .
Leathem , J. G. Phil. Trans. , , 1897 , vol. 190 , pp. 89-127 .
Metallic Films , Transmission of Light through .
Leathem , J. G. Phil. Trans. , , 1897 , vol. , pp. 89-127 .
89 ] V. On the Theory of the Magneto-Optic Phenomena of Iron , Nickel , and Cobalt .
J. G. LEATHEM , B.A. , Fellow of St. Jolm 's , and Isaac Newton Student in the ) Commumicated ROBERT .
BALL , F.R. Received May ll , \mdash ; Read June 17 , 1897 .
Introduction .
1 .
IN his 'British Association Report ' ( 1893 ) , on " " Action of Magnetism on \ldquo ; Mr. LARMOR points out that there are two possible ways in which the netic field may regarded as affecting the phenomena of light propagation .
" " The imposed magnetisation is an independent kinetic system of a vortical character , which is linked on to the vibrational system which transmits the light waves and from the first point of view " " the kinetic reaction between the two systems will add on new terms to the electric force and so there would be a " " magneto-optic term\ldquo ; in the expression for the kinetic energy .
This type of theory includes MAXWELL'S hypothesis of molecular vortices , and has been analytically treated by FITZGERALD and BASSET ; it also includes the theory developed by DRUDE in his paper\ldquo ; Magnetooptische Erscheinungen in ' Wiedemann 's Annalen , ' ] . .
The great difficulty arises when one comes to consider the boundary conditions , as a discontinuity of electric force cannot be avoided ; apparently the only satisfactory way of meeting this difficulty is to be found in LARMOR'S suggested modification of FITzGERALD'S analysis , involving the supposition that in the case of reflection of light at tlIe surface of il magnetised metal the constraint introduces an irrotational or compressional way of the her set up at the reflecting surface and travelling with very or infinite velocity through the space occupied by the metal ; a satisfactory system of equations of propagation nnd boundary conditions is thus obtained by applying the principle of Least Action .
I have worked out ) mathematics of this theory and obtained the general so1ntion of the problem of reflection from a magnet ; on comparing this with the results of several German and Dutch physicists , it appears that the of the theory with expeliment is at best very doubtful , even when allowance is made for the possibility of large errors of .
There is moreover one phenomenon , recently discoyered , which the theory quite fails to account for , , an effect of the component of the magnetic field perpendicular to the plane of incidence .
2 .
The second type of tbeory supposes that the imposed magnetisation " " sli htly V0L . .
10.97 .
alters ) structure of the medium which conveys the light vibrations , but does not exert a direct d.ynamical effect on these vibrations\ldquo ; the isotropy of the medium is , as it were , destroyed , and rotational terms appear in the elastic relations between and the corresponding forces .
This theory , in its valid form as regards boundary conditions , has quite lately been formulated for transparent 1nedia by BASSEI ican Journal thematics , , and the principles underlie the very rent analysis of GOLDHAMMER in his memoir of ( ' Wied .
Ann vol. 46 ) .
LARMOR also independently this theory in his ' British Association Report , ' 1893 , and more explicitly in ' Proc. Lond. Math. Soc April , 1893 .
In his exposition it was shown that the rotational terms in the equations connecting electric displacement and electric force are not open to the objectiou that they would imply erpebual motions , as they involve only the rate of change of the force .
The boundary conditions in this theory are of the standard form , nely continuity of the tangential components of electric and magnetic force , and of the normal components of magnetic induction and total current .
It has been shown by BASSET how the whole scheme may be formulated from a single energy nction by the pl.inciple of Least Action .
3 .
In the present paper it is proposed to the fundamental equations of this type of theory in ] a general form on the lines of Mr. LARMOR'S recent papers on EJectrodynamics , and to develop them so as to obtain the solutions of the problems of the reflection of at the surface of a magnet , of the tran smission of tbrough normally magnetised metallic films .
The formulae so obtained will be compared the available experimental results , with a view to asccrtainlng to extent the theory is in reement with the facts .
The theory inyolves a single magneto-optic constant which in ] may be assumed complex ; we shall try whether it is possible , by giving suitable numerical values to the modulus vector lgle of this constant , to make the theoly account for all the observed phenomena ; if so , we shall ascertain what these numerical values are .
If successful we shall a formulation of the phenomena in a mathematical scheme , which ought to serye as a guide in the e]aboration of physical t , heory .
In carrying out this programme , I am ware that I shall be going over ground which has already been covered to some extent by GOLDHAMMER , and by DRUDE , but my method will be entirely different from theirs , and I shall be able to use important experimental results which had not been published at the time their papers were written .
Notaiion .
4 .
The notation is nearly the sa1ne as is electromotive force , the total current , magnetic induction , specific conductivity , specitic inductive capacity taken as a pure ratio , the velocity of radiation ; corresponds to MAXWELL'S total electric displacement ; its components MAGNETO-OPTIC PHENOMENA OF IRON , NICKEL , AND COBALT .
and are the vectors of LARMOR'S theory Phil. , namely is the displacement involved in the aether strain , and that involved in the polarisation of the matter .
mental E 5 .
It being as usual assumed that for oscillations so rapid as those of light } effective magnetic permeability is unity , the fundamental equations of the theory are as follows:\mdash ; The two circuital relations ( 1 ) , ( 2 ) .
( ii .
) The equations of the current ( 3 ) , where the vector represents the Hall effect .
( iii .
) The displacement relations , and the ] astic relations between force and the corresponding polarisation , . . . . .
( 4 ) , .
( 5 ) , and the vector representing , in transparent matter , the whole magneto-optic effect .
The restriction of the relation to this form isjustified as follows : \mdash ; When there is no matter present the polarisation is null , and there is tory effect .
When there is matter present the action of an electric field on the risation induced by it gives rise to a pondero-motive force which does mechanical vork in a small displacement of the matter , equal , for an element of volume S , to S. When there is no conduction therefore no dissipation , this quantity must be an exact differential , otherwise mechanical work could be gained in a complete cycle of displacement of material ediunl , which would imply the possibility of perpetual motions .
This restriction must also hold universally , because the nature of the molecular polarisation is independent of whether conduction is present or not .
It requires that J Sh S and therefore shall be the variation of , where is some function of and diffsrential coefficients , terms at the time limits being left out of account .
The expression for in terms of can therefore involve no rotatory terms in itself , as Lord first showed , but it may have rotatory terms in , which the characteristics of the m-optic property .
Rotatory of a certain type in the spacial fluxions O are also adrnissible ; these to optical rotation of the structural kind ; they foreign to the present problem because they are isotropic , instead of being related to an imposed vector , the intensity of magnetisation .
In either .
differentiations of odd order 1night also come into the expressions : these would affect the relations of the phenomena to optical rsion , but not the questions ] ) treated .
of Propagation .
6 From the fundamental iollS we readily obtain ( 7 ) .
For brevity we may put ( 8 ) -OPTIO PHENOMENA OF IRON , NICKEL , AND and . . . . . . .
( 9 ) , so that equations ( 7 ) become . . .
( 10 ) .
Now and are supposed to be exceedingly small quantities , so that are also extremely small .
If we squares and products of and solve equations ( 10 ) for , we get ( 11 ) .
To get the equations of propagation differentiate with respect to the the first of equations ( 1 ) by ( 2 ) hence , if for brevity we put .
( 12 ) , our equations of propagation become ( 13 ) .
J. ATHE l ON OF THE a Mediurn .
7 .
In the ) of waves in a ] medium , let us assume ( 14 ) where represents , and write . . . . . .
2 .
( 15 ) .
Substituting these values in equations of propagation , we get ( 16 ) .
Addition of times the of these to times the last gives , as was to be ( 17 ) , and hence if we eliminate , and ( !
, we get which reduces to . . .
( 18 ) .
This equation gives possible values of corresponding to given values of and .
It is a quartic ) therefore has four roots , of which two have their imaginary parts negative and their real parts positive ; let us denote these roots by , and the responding values of by and respectively , so that ( 19 ) .
MAGNETO-OPTIC PHENOMENA OP IRON , NICKEL , AND T. In the particular case of , both zero , the equation to determine would be so that if be the value of given by this equation , and the value of and therefore ---H . . . .
( 20 ) and the values coinciding in pairs .
The sign of is uous ; we determine it by requiring that shall have its imaginary part negative , in which case we shall find that its real part is positive .
As we neglect second and higher powers of and , the equations ( 19 ) of the general case may now be , introducing this quantity ( 21 ) so that ( 22 ) , ( 23 ) , ( 24 ) , ( 25 ) , results will be of use later on .
Corresponding to the two values of there are two sets of constants ; these we distinguish by the suffixes ( ) and ( 2 ) .
Equation ( 17 ) shows that and . . . . . .
( 26 ) .
This taken in conjunction with the second of equations ( 16 ) gives whether the suffix be ( 1 ) or ( 2 ) .
Hence } virtue of equations ( 19 ) 1 .
or ( 27 ) .
8 .
In the case of air ( .
any medium in which tbere is no magneto-optic rotation ) are zero .
For air also ) .
The substitution of the of ) in the equations of propagation gives or , as was to be expected .
Problem of ReflectioJl .
9 .
We now in position to attack the problem of the reflection of light at the surface of a magrnetised metal .
Let the interface between the two media be the plane ; the air occupying the space positive , and the metal the space negative .
The plane of incidence is taken as the plane We assume that , the ( 28 ) , where resent the incident wave , and the reflected wave .
MAGNETO-OPTIC PHENOMENA OF IRON , N , AND COBALT .
\ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; \ldquo ; ln the metal , ( 29 ) .
In this assumption we take account of only two of the four possible waves in the metallic medium ; the othe two are omitted because they are waves which travel in a direction that makes an acute angle with the axis of ; and as in our present problem all waves in the metallic medium are originated at the plane , only those can actually occur whose direction of propagation makes an obtuse ] with the axis of 10 .
The surface conditions which have to be satisfied are the continuity of across the interface ; and , as usual , the continuity of involves that of , while the continuity of involves that of Thus the conditions are four , namely continuity of ( 1 ) ( 2 ) , which .
( 3 ) , which ( 4 ) , which leads to the continuity of or We shall denote the of the metallic medium by to distinguish it from that of air .
In the air .
If be the periodic time , and .
the wavelength in air of the light considered , so that since , and therefore p , is the same for the metal as for the air , we see by ( 20 ) that and therefore .
( 31 ) , VOL. CXC.\mdash ; A. is the ( iv irldex the metal ( J. J. THOMSON , 'Recent ' p. 419 .
us substitute ) ssumed exponential expressions for in the surface conditions ; we thus get HL HL HL ( 32 ) .
make the form of the tirst of these equations analogous with that of the others by llultiplying it across by or four equations may then be itten as follows : When we ubstittlte in these values of found above , nd neglect quantities of the second and higher orders , we get MAGNETO-OPTIC OF IRON , , ANT ) From the second and third of these equations we readily get HL , and .
Substituting these for and in the first and fourth of ( 34 ) , remembering ( 31 ) , and for brevity denoting by , we get , , which may be written , and solving these for A and , we get which is the complete mathematical solution of the problem of reflection .
Relation of to the gnetic F 12 .
Before going further , we must consider how depend on the imposed magnetic field ; and first it should be noticed that , though appear as 00 MR. J. G. LEATHEM ON THE OF THE operators , in the case of light vibrations they algebraical quantities , because In theories of this nature it is usual to assume that are proportional to the components , parallel to the axes , of the imposed magnetic force ; the constants of the HALL effect are also usually assumed to vary as the magnetic force ; and therefore in this case would do so also .
But in experiments on the transmission of light through magnetised metallic { ilms it is found that the rotation of the plane of polarisation certainly does not vary as the magnetic force , but very probably varies as the intensity of magnetisation , a quantity very difficult to determine .
We shall see later that in the mathematical solution of the problem of transmission the rotation varies as , and hence we are driven to the assumption that vary as the components of tisation .
I shall also suppose that vary as the components of mngnetisation , which necessitates the assumption that do so likewise ; the question whether the HALL effect varies as the magnetic force or as the magnetisation has not , I think , been put to an experimental ; the latter supposition seems more probable .
These assumptions can be readily justified from physical considerations .
For in vacuo there is no -optic rotation , though there is magnetic force ; it is-therefore not the magnetic force , but matter , or some property of matter when under the influence of magnetic force , that causes the rotation ; and the property of matter under the influence of magnetic force is not force , but magnetisation .
Denoting the components of the imposed magnetisation by , we assume ( 87 ) , where is the complex magneto-optic constant of the theory .
any particular metal the values of and may be determined by experiment , and if we find that the numerical values of these constants , as determined by all the different sorts of experiments , are the same , we shall conclude that the theory can account for all the observed facts , and therefore constitutes a complete mathematical explanation of the phenomena .
The Optic Constants of 13 .
The constants and are different for different metals , and also for light of different colours .
Their values have not been directly tabulated , but they are easily obtained from the tabulated values of DRUDE'S optic constants ; these latter are denoted by and , and are connected with and by the relations so that , . .
, .
MAGNETO-OPTIC PHENOMENA OF IRON , NICKEL , AND COBALT .
The values of and for iron , steel , and llickel will be found in a paper of DRUDE'S ( ' Wied .
Ann vol. 39 , p. 481 ) , quoted in THOMSON'S 'Recent Researches , ' p. 421 .
The constants for cobalt are given by DRUDE in ' Wied .
Anu vol. 46 , I ) .
407 .
These values are shown in the following table:\mdash ; Steel Nickel .
Cobalt Steel Nickel .
Cobalt Steel Nickel .
Cobalt wherein , for red light , centim. , and for sodium we take centim. Hence we find the corresponding values of and Iron Steel Nickel .
Cobalt The KERR 14 .
We shall first compare our theory with the results of the KERR experiments , which are so well that they need not be here described .
In .
KERR'S second experiment the magnetisation is parallel to the refle , cting , and to the plane of incidence ; and the incident light is polarised perpendicularly to plane of incidence .
Thus , in our notation , and ; and the reflected light is specified by A and , whose values as by formula ( 36 ) are ( 88 ) .
If the incident ray be as represented in the figure , and be the angle of incidence , and and be positive , then The incident ra being plane polarised , is ] .
But A and are both complex , and have not necessarily the same vector angle ; hence the reflected light is elliptically polarised .
If be the angle through which the major axis of the ellipse of polarisation is rotated round the reflected ray ( in the direction from the axis of towards the sxis of y ) from the plane , since the modulus of is veIy mall cpared with the of is given by real part of in circular measure .
Tn the case of iron , KERR found that when is negative , is negative if be than about ; while if be greater than is positive .
The angle of dence for which changes sign ( and therefore vanishes ) has been obselved by different erinlenters , whose results differ considerably .
They aoe as follows:\mdash ; KERR KUNDT to RIOHI SISSINGH DRUDE from ( 38 ) so that , where and , therefore , since ( 39 ) , NETO-OPTIC NOMENA O I ) , NICKC , AND 10.3 cbanges sig for that value of which nakes the vector } of to an odd number of right angles .
Let us assume that must lie between ving the sign of to be determined afterwards .
To clngle for given angle of incidence we must calculate the vectol angles of the various complex factors which occur in numerator denominatol of the actio in equation .
This involves troublesome work ; but it is to the approximation on the } ) position that is large , used by J. J. THOMSON ( ' Recent Researches , ' p. 498 ) in a silnilar investigation , as that method in troduces error of quite a large number of degrees .
Using the constants for yellow light , I get the following values :\mdash ; whence are derived the following : Vector a ' So if ohanges sign when If when , then If when , then And probably if changed sign when , the corresponding of would be about The uncertainty as to the exact value of the allgle of incidence which , and the large difference , caused by a small error of observation , in ths resulting value of , render this experiment unsuitable as a means of arriving at the exact value of .
It will , however , be useful in testing a value of determined in some other way .
iment will us the sign of ; for , in accordance with KERR'S observations , when the incidence is very nearly normal , is of the same sign as ; when the angle of incidence is nearly is of the opposite sign to .
Now the table of values given above indicates that as passes through that value ( be it or for which vanishes , from a less to a greater value , the cosine of the vector angle being negative becomes positive , so that for very great angles of incidence is of the same sign as .
Hence is negative .
15 .
In KERR'S first experiment the magnetisation is parallel to the reflecting the incident light is polarised in the plane of incidence .
If be the otation of the major axis of the eIlipse of polarisation in the same sense as before , KERR found that hns the same sign for all angles of incidence , and that this sign is opposite to that of Tn this case , and real part of .
From result ( 36 ) we readily deduce that of which tho vector angle ( including the minus sign ) is \mdash ; sum of ngles of , .
If , vector angle of is vector angle is , vector angle is And evidently for angles of incidence the vector angle lies between and that if have value between and , the cosine of the ngle of is negative for all angles of incidence .
Thus has always the same sig as , that is , the opposite sign to Hence any value of lying between and satisfies all the conditions of first experiment .
16 .
In another of KERR'S experiments the magnetisation is normal to the reflecting surface , nd the incident light is polarised in the plane of incidence .
Here ; , having the same is found by KERR to be of opposite sign to angles of incidence .
Now is real part , and we readily deduce froIl that MAGNETO-OPTIC PHENOMENA OF IRON , NICKEL , AND COBALT .
of which complex the vector angle is found in the usual way to lie , for all angles of incidence , between the values , corresponding to , and corresponding to If have any value between and , the cosine of this vector angle is always negative ; and so has the same sign as , or the opposite sign to Hence this experiment of KERR'S is satisfied if have any value between and 17 .
In KERR'S fourth experiment the magnetisation is normal to the reflecting surface , and the incident light is polaxised arly to the plane of incidence .
Here ) , and is the real part of where , from ( 36 ) , ' of which complex the vector angle l.ies between the values , corresponding to , ond , corresponding to Now KERR found that is , for all angles of incidence , of opposite sig to , that is , of the same sign as .
Hence the cosine of the vector angle of is always positive .
Tbis is in accordance with our theory , provided the alue of lie between and Obviously this conclusion is at variance with that derived from the preceding experiment , unless happen to have exactly the value But this experiment of KERR'S was repeated by KIJNDT , who found that has not the same sign for all angles of incidence , but that it vanishes and changes sigll for an angle of incidence which he estimated at about I have calculated the values of the vector angle of for several ) gles of incidence in the neighbourhood of ; they are as follows : \mdash ; So that if we denote by that angle of incidence for changes sign , the values of corresponding to various hypothetical values of are as VOL. CXC .
MR. J. G. ON THE THEORy OF THE And if KUNDT'S observation be accurate , the value of is about 18 .
From the preceding paragraphs it appears that any value of lying between and will account very for the four KERR experiments ( in the case of iron ) , except that the agreement with result in the fourth experiment would be imperfect to the of four or five degrees .
19 .
The KERR experiments were also tried on mirrors of nickel of cobalt but the observations made were so that they are of little use as a test of the present theory .
In the case of polar tion from nickel , when the incident light is polarised perpendicularly to the plane KUNDT found that the rotation changes .
for an angle of incidence somewhere between and I have calculated ( for yellow light ) the values of the vector of for these angles ofincidence , and find them to be for for so that any of lying between and will give a satisfactory explanation of this experiment .
Also would be negative .
When the reflection is polar and the incident light is polarised in the plane of incidence , KUNDT finds that has the opposite sign to 70 for all angles of incidence .
The vector angle of is found to be when and when The cosine of this vector angle will be negative for all angles of incidence , provided the value of lie between and The two experiments indicate that for nickel has a value intermediate between 14 and The Experiments of SISSINGH ZEEMAN .
20 .
I now pass to a much more precise and accurate test of present theory , the erials for which are to found in the elaborate series of experiments mnde during the last few years at Leyden , by SISSINGH and by ZEEMAN .
MAGNETO-OPTIC PHENOMENA OF IRON , NICKEL , AND COBALT .
The first of these series is described by Sissingh in a paper in the ives Neerlandaises , ' vol. 27 .
The experiments were made with an mirror , magnetised parallel to the reflecting surface ; the amplitude and phase of what Sissingh calls " " the magneto-optic component of the reflected ligh were measured for various angles of incidence .
One result f.the observations is the conclusion that , and ; that is , that for any given angle of incidence the magneto-optic component has the same amplitude and phase , whether the incident be polarised in or perpendicularly to the plane of incidence .
It is to be noticed that the of this component is defined as its retardation of phase calculated relatively to that component of ordinary metallic reflection which is polarised in the plane of incidence .
The amplitude is reckoned on the supposition that the amplitude of the incident ray is unity .
The accompanying figure shows the relation between the axes of coordinates and the principal directions for the incident and reflected , as defined by The standard ray for phase is \mdash ; B. The standard lay for amplitude , or , according as the incident light is polarised perpendicularly to or in the plane of incidence .
' It will be convenient to denote by the of phase of the magnetooptic component of the reflected ray calculated relatively to that componenb of ordinary ] reflection which is in ths plane of incidence .
lf be calculated from theory , and from experiment , the theory and expel.imentwil ] agree if or .
When the incident light polarised in the plane of incidence , , and in formula ( 36 ) the incident ray is represented , the magneto-optic component of the reflected ray by , and the component relatively to which phase is to be measured , by \mdash ; B. Hence vector angle of vector angle of ) ) ( 40 ) .
When the incident light is trised perpendicularly to the plane of incidence , , and the incident ray is represented by ; the magneto-optic component of the reflected by \mdash ; , that is to say that term in which contains the factor ray relatively to which phase is measured is represented that in which colltains the ( vanishing ) factor .
If be supposed to be only just not zero , then , since the incident ray is supposed to be plane rised , a real quantity .
Hence we have vector angle of .
( 41 ) .
From ( 40 ) and ( 41 ) we see at once that when the reflection is equatorial , that is , when ( say ) , and this agrees with observations .
We also see that when the reflection is polar , that is when Now ZEEMAN , as a result of experiments on polar reflection described by him in the ' Archives ] andaises , ' vol. 27 , came to the conclusion that .
It is very possible that this discrepancy is due to his using a definition of and slightly different from SISSINGH'S .
2 When the reflection is equatorial , we see from ( 40 ) that vector angle In determining from this expression there is an ambiguity to the extent of ; for in defining SISSINGH that it shall not be altered when changes sign .
paper , we see that in eqnatorial reflection the standard case is when is negative .
Hence , remembering that is negative , we find that \mdash ; the sum of the vector angles of , and .
( 42 ) ; and to get accurately for any particular angle of incidence , these three vector must be calculated .
The following table shows the results of SISSJNGH'S observations on the phase for MAGNETO-OPTIC PHENOMEINA OF IRON , NICKEL , AND COBALT .
various angles of incidence , and the theoretical values of the phase for the same angles of incidence , calculated from the present theory .
It is to be observed that the calculation involves only the ordinary optic constants of the metal , and that it is from the comparison with experiment that we derive informa , tion as to the value of iu the magneto-optic constant .
There is thus no question of being able to adjust two coefficients , and , so as to satisfy the observations , as might be supposed ; only one coefficient is involved , and the test is accordingly a severe one .
EQUATORIAL Reflection Iron .
Yellow Light .
C.G.S. The constancy of the angles in the last column is remarkably good ; and the theory accounts for the phenomena with great accuracy if the value assigned to be the mean of the amounts by which these angles respectively fall short of , namely 23 .
When the reflection is polar , we see from ( 41 ) that vector angle of Taking positive as the standard case , and remembering that is negative , we find that \mdash ; the sum of the vector angles of and !
43 ) .
Experiments as to the amplitude and phase of the magneto-optic component of light reflected from an iron mirror , magnetised normally to the reflecting surface , have been made by ZEEMAN .
He gives an account of these in the Archives Neerlandaises , ' vol. 27 ; he confines himself to one angle of incidence , viz. , His result as regards phase compares with theory as follows:\mdash ; POLAR Reflection from Iron .
Yellow Light .
C.G.S. The theory will agree accurately with the experiment if the value of be The values of giyen by ZEEMAN for ] light are ( 1 .
) Derived from " " rotations to zero\ldquo ; ( 2 .
) erived from " " minimum rotations\ldquo ; In determining the phase the method of " " rotations to zero\ldquo ; is preferable to that of ' ' minimum rotations and so in the above table I have used the mean of the values got by the former method .
24 .
Another test of the present theory is afforded by observations of the amplitude of the " " magneto-optic component This is denoted by or , according as the incident light is polarised in or perpendicularly to the plane of incidence .
In the former case magneto-optic component is represented by , and the incident ray ; in the latter case the incident ray is represented and the magneto-optic component by .
Hence ( 44 ) .
Thus , for torial reflection , we readily derive from ( 36 ) same , and therefore ( say ) , which agrees with SISSINGH'S result .
MAGNETO-OPTIC PHENOMENA OF IRON , NIUKEI AND COBALT .
If brevity we put , we have siu L. mod \mdash ; ) ' and the latter factor may be calculated for any of incidence .
In the following table the theoretical values of for various angles of incidence are compared with the values observed by SISSINGH .
Here again the theoretical value of involves the magneto-optic constant only by being proportional to , and , is not involved : thus we ve not available any adjustment of to improve the agreement , and the test is very severe .
In fact one set of experiments involves the other set only : ) that a complex magneto-optic constant really gives no more opportunity for adjustment than would a real one .
EQUATORIAL Reflection from Iron .
Yellow Light .
\mdash ; 1400 C.G.S. Exact agreement of theory with experiment would be indicated by the numbers ' in the last column being all equal .
Though this is not the case , their tion to equality is , considering the probability of errors in the observations , remarkably good .
The mean of these numbers is about 44 ; and if we assume to have such a value that , the ratios of the calculated to the observed amplitudes for the above angles of incidence taken in order are , and respectively .
The corresponding value is where the units electromagnetic and C.G.S. And hence 25 .
For pofar reflection , we derive from ( 44 ) and ( 36 ) ( say ) .
Comparing this with the amplitude in equatorial reflectiolJ , we find , the value of this calculated from theory , is 321 .
But the values ascribed to , and correspond to the experiments of SINGH and ZEEMAN ; and the latter found experimentally So that here again we have a very fair agreement of the theory with experiment .
Nickel 26 .
In the paper already quoted ZEEMAN gives a few measurements made by himself on polar reflection from nickel .
He also quotes experimental results of KUNDT ( ' Wied .
Ann vol. 23 ) , and DRUDE ( ' Wied .
Ann vol. 46 ) , which he sses in a form similar to his own .
These I have used to form the fcllowing tables , wherein the theoJ ; etical values of the phase and amplitude are in all cases ated for yellow light .
-OPTIC NOMENA OF IRON , , AND COBALT .
EQUATORIAL Reflection Nickel .
( Probably ) White Light .
Fairly good agreement is here indicated ( except in case of first of incidencs ) if the value of be about EQUATORIAL Reflection Nickel .
Whitc Light .
Angle of incidence .
Angle of incidence .
If or thereal)outs a fairly good reement is indicated , except in the case of For this case DRUDE'S ffers widely from KUNDT'S , and is perhapswrong-\mdash ; .
QUATORTAL from Nickel .
The agreement theory with experiment is here more defective .
As the ) VOL. CXC .
of magnetisation is not stated , this series of experiments gives no information as to the value of POLAR Reflection from Nickel .
Yellow Light .
alue o ' showing agreement if The experiments quoted in the two following tables are from a paper of ZEEMAN'S ( ' Communications from the Leiden Laboratory of Physics , ' No. 10 ) :\mdash ; POLAR Reflection from Nickel .
White Light .
C.G.S. Calculated value of Observed value of showing agreement if POLAR Reflection from Nickel .
White Light .
C.G.S. .
ved .
ved wherein The agreement indicated is excellent , provided have such a value that nanlely , 9 .
-OPTIC NOME N OF IRON , [ CKEL , AND COBALT .
The experimental results used in the following table are taken from a paper by Dr. WIND Communications Leiden Physics , POLAR Reflection from Nickel .
Yello Light .
Calculated value of Obsel.ved value of Calculated value of Obsel.ved value of .
mi .
Calculated value of Obsel.ved value of .
mi .
Calculated value of Obsel.ved value of .
mi .
Calculated value of Obsel.ved value of .
mi .
Calculated value of Obsel.ved value of .
mi .
Calculated value of Obsel.ved value of .
mi .
Calculated value of Obsel.ved value of .
mi .
Calculated value of Obsel.ved value of .
mi .
Calculated value of Obsel.ved value of .
mi .
Calculated value of Obsel.ved value of .
mi .
Calculated value of Obsel.ved value of .
mi .
' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' For incidence of ZEEMAN'S result is to be preferred to that of WIND , as he took more precautions to eliminate causes of error .
For the other two angles of incidence agreement is indicated if is about In estimating the consistency of the above results it is to be remembered that the optic constants ( and ) for different specimens of nickel are often sensibly different .
( I have used the same set of values all through .
) Indeed in the case of iron ZEEMAN lound that his observations of the optic constants of a particular mirror , made ] eqpectively before and after an observation of the KERR phenomena , differed considerably .
Cobalt .
27 .
In his paper in the 'Archives Ne'erlandaises , ' vol. EMAN describes experiments made by himself on mirrors of cobalt , and also the results of experiments made by ) .
The comparison of these with the present theory is shown in the following tables:\mdash ; POLAR Reflection irom Cobalt , .
White Light .
Good agreement is indicated if POLAR eflection from Cobalt .
Green Light .
This good agreement if is about .
In the above is calculated from the constants for yellow light .
Reflection from Cobalt .
White Light .
the agreement is not so good ; the last three angles of incidence would indicate bhat is about .
DRUDE'S method that of minimum rotations , wherein errors of observation influence the phase to a much greater extent than in the method of null-rotations .
The following experiments are from a paper of ZEEMAN'S from the Leiden Laboratory of Physics , ' No. ) POLAR Reflection from Cobalt .
White Light .
Augle of Calculated value of incidcnce .
'2092 Augle of CaIculated value of incidcnce .
'2092 wherein NOMhNA IRON , , AN COBALT .
the agreement is good , the indicated value of being about ; if this be so , we have- for cobalt i. e. , Hall ct .
28 .
Before ving this of the subject it is worth to investigate whether the ordinary HALL eflect is enough to contril ) , to any ppreciable extent , to the phenomena we have been considel.ing .
If 9 be HALL'S constant , crtS usually , the equations into which it enters ) of the type , where is the magnetic force .
thie with the ) that equations ( 11 ) would assume if were , it appears that ; so that Now for iron , and if we substitute the values of , corresponding to yellow light , we lind .
where But we see from S 24 that , where .
Also , where is the magnetic permeability of iron and is greater than unity .
Hence the modulus of the fraction has a logarithm less than , so that the modulus itself is less than .
Thus it appears thab ordinary HALL effect is more than two hundred thousand times too small to account for the KERR phenomena .
But , in order that the coefficients should be real , it is necessary that the imaginary parts of the complexes should be supplied by , that is to say by the HALL effect .
Hence it must be concluded that the coefficient of the HALL effect is very much greater for excessively rapidly alternating currents than 1fR .
J. G. LHiATHEM ON THE THEiORY OF THE for steady ones .
There is nothing unn atural in this , for the in cipient conductions which make optical opacity have no relation of continuity whatever with the steady conduction in current ; thus MAXWELL found that the ordinary coefficients of ctivity \ldquo ; are very much smaller in the optical circumstances .
And it may be noticed that , as is proportional to electromotive force divided by current , a greatly ninished conductivity will correspond to a greatly increased value of 9 .
The value which HALL'S stant would , on this supposition , have for yellow light , is obtainable from the equation Ima , in ry part of imaginary part wherein now take the HALL effect to be proportional , not to the magnetic force , but to the ] sity of magnetisation .
This gives for iron , and yellow light .
The real part of is then , so that , if , and hence we find .
of Ma9netisation to the of Incidence .
29 .
A very interesting inference from the presence of in the equations ( 86 ) is that , if the theory be true , the component of magnetisation perpendicular to the plane of incidence will produce an effect nob quite the same as the KERR pbenomenon , but of the same order of magnitude .
On whether such all effect had ever been observed or measured , I found that a few months ago it was predicted from theoretical considerations Dr. C. H. WIND , in a paper which has as yet ppeared only in Dutch .
Acting on this prediction ZEEMAN sought the phenomenon imentally , found it , succeeded in measuring it .
His results are published in the 'Communications from the Leiden Laboratory of Physics , ' No. 29 .
Let us suppose that the is entirely perpendicular to the plane of incidence ; then and , and the reflected light is specified by A and where , from equations ( 36 ) , OF IRON , NICKEL , AND COBALT .
From these essions we see , that , if the incident ray is polarised in the plane of incidence , so that , the expression for the reflected ray does not involve ; and so the magnetisation produces no effect .
This is in agreement with the prediction of WIND .
But if the incident ray be poliarised to the plane of incidence , so that , the reflected ray is given by , which does contain stead of having as would be the case if there were no nagnetisation , we have A equal to this value multiplied by th factor ' which is the same as ' and the effect of this factor , which of course differs from unity by only a very small quantity , is to slightly alter both the amplitude and the phase of still planepolarised reflected ray .
Now the change of phase produced by a factor of the form , where is very small , is , and therefore , in circular measure , is approximately Hence , if for brevity we put the acceleration of phase produced in the reflected ray by the component of magnetisation is , in circular measure , .
In ZEEMAN'S experiment the angle of incidence was and the of was a little over 1100 C.G.S. , the mirror being of iron .
Under these circumstances he found the acceleration of phase to be with a mean error of .
the tbeol.tical value , we notice that itnd if we ( ( which is about the mean of the indicated by the nents o and ZEEMAN ) , then ; we / lso assume With we find that change of phase indicated by the theory is , in circular measure , , or , in degrees , This very well with ZEEMAN'S observations .
Transmission thr.ongh Metal Films .
.3 .
Another effect of the action of magnetism light is the rotation of the plane of ] ) of normally light , on ] passing through very thin films of magnetised .
The principal experiments in this sub .
have been made by KUNDT , DU BOIS , LOBACH , and they found that rotation is always in the direction of the magnetising current , and measured it in special cases .
It is desirable to compare these measucements with the mathematical solution of the problem worked out on the basis of the present theory .
Let the film be bounded by the planes and and let the incident light fall normally on the surface .
The external magnetic field is supposed to be parallel to the axis of , so that and The plane of polarisation of the incident light is taken as the plane of Thus we may assume the followmg expressions to } ) resent the light in the air on the sides of the film , and in the film itself .
MAGNETO-OPTIC PHENONENA OF IRON , NICKEL , AND COBALT .
In the air In the metal In the air wherein the incident ray is represented by , the reflected ray by , and the mitted ray by .
It should be noticed that in these assumptions multiple reflections are not neglected ; all waves in the film are included in the complex constants , and For surface conditions we may establish continuity of , , , and respectively ; the second of these is the expression of the continuity of the netic force , and is used instead of the continuity of to which it is , as it leads to more symmetrical analysis .
At the surface these boundary conditions lead to the equations , Hm , , At the surface brevity write , the boundary conditions lead to the equations VOL. CXC .
, , , .
If , for brevity , we put , and notice from equations ( 25 ) that in the present problem we readily reduce the boundary conditions to the following:\mdash ; \mdash ; \mdash ; - Eliminating nd from the last six , and representing ?
by , we get -OPTIC OF IRON , NICKEL , AND COBALT .
\mdash ; - The last two equations give ( say ) ( say ) and substitution these in the first leads to whence , Substituting these in the sixth and eighth of the surface conditions , we get where and Now and therefore , if the modulus of be not exceedingly great , .
Substituting these expressions , we readily find , to the first order in -- .
and so whence and if be the angle through which the major axis of the ellipse of polarisation of the transmitted light is rotated from the axis of towards the of is the real part of .
For a given metal , and given values of and , this angle can be calculated exactly from the above ] , but such calculation would be very tedious .
It is well therefore to examine the relative magnitudes of the different terms cases corresponding to the known experiments on this subject , in order to see whether there is any approxiu ) of a simpler character .
It will be sufficient to consider some of the experiments on transmission through films of described by LOBACH ( ' Wied .
Ann. ' vol. 39 , p. 356 ) and by DRUDE ( ' Wied .
Ann. ' vol. 46 , p. 416 ) .
Now MAGNETO-OPTIC PHENOMENA OF IRON , NICKEiL , AN1 ) COBALT .
so that for iron .
Also In DRUDE'S experiments the values o.f lie between and In LOBACH'S experiments the values of lie between and 167 .
Hence , in the two sets of experiments , the greatest value of the of is about , and therefore the greatest value of the modulus of is about 1035 ; this corresponds to the thinnest film , for the thicker films the modulus is very smaller .
Hence if we neglect the fourth term in the numerator and the second term in the denominator , and put we have an approximate formula whose error , for the very thinnest film considered , will not exceed about 10 per cent. , and is very much smaller for the large majority of the experiments .
Putting in numerical values , this becomes ( for iron , and sodium light ) , In the experiments the film is generally magnetised as strongly posbible , but there is no direct way of ascertaining ntens of agnetisation attained .
Thus is to a certain extent indeterminate .
According to EWING the maximuu intensity of magnetisation for some slJecimens of iron is about 1730 C.G. .
units .
I shall therefore assume ; also I take as determined by In one of LOBACH'S experiments the thickness of the film is given by centim. The light used is sodium light , so that centim. ; and the in the direction of the magnetising current is observed to be degrees .
MR. J. G. ON THEORY OF THB1 To compare this with the rotation indicated by theory , we notice that , and So that ( the other term being in this case so small that it may be neglected ) being takell as Ths angle is the real of this complex , and is therefore in circular measure , being positive when is positive , so that it is a rotation in the direction of the magnetising current .
etical value of in is 961 .
As we have seen , the observed value in degrees is ; in comparing these results it should be noticed that one factor of the theoretical of is the cosine of an angle which isjust about greatel than a right angle ; this angle contains , whose value we have had to guess ; a comparatively small error in the value to will therefore a considerable error in the culated value of .
The values of and being also uncertain , the agreement of the theory with may be regarded as good .
In one of DRUDE'S iments h , and the light used is red ; the observed rotation is degrees .
If substitute this value of in the above-obtained approximate formula , we find whence in circular measure , or degTees .
In this case , in addition to the possible causes of error referred to in connexion the previous experiment , it is to be noticed that though the experiment was made with red light , it has been necessary in the calculation to use the values of , and for yellow light , for lack of information as to their values in the case of red light .
When also it is borne in mind that the value of the magneto-optic constant derived from observation of reflection from mirrors has here been applied to test experiments on transmission through thin , with results not only of the same order of magnibude but identical within the limits of uncertainty of the -OPTIC OF IRON , NICKEL , AND intensity of magnetisation , the agreement must ) considered as a very satisfactory vindication both of the theory and of the experiments .
Conclusion .
31 .
The various results obtained in this paper do not , I think , require any detailed comment .
They may be fairly claimed to show a remark , good agreement between theory and experiment , a better agreelnent , believe , than is shown in the papers of GOLDHAMMER and DRTJDE .
The ] considerable discrepancy arises in connexion with the original KERR experiments ; but here it is to be remembered that the experiments of KERR , and those of SISSIN and ZEEMAN , are not measurements of diffel'ent phenomena , but different ways of measuring the same phenomenon .
Hence any theory that agrees with one of these sorts of ought to agree equally well with the other sort ; and if this is found not to be cas it is probably not the fault .
the theoly , ) nrust be attributed to ccuracy in one of the sets of results .
Thus it wouId appear that the original of KERR , who was the pioneer in this , hayc been tively much improved on by later investigations .

