]\gt ; 1902 .
] jrential Eqnations of , -vector .
Shift -metres .
4046.98 1902 .
] jrential Eof -vector .
\mdash ; Shift -metres .
4046 .
6 4045 .
\mdash ; 1902 .
] jrential Eof -vector .
\mdash ; Shift -metres .
4046 .
6 4045 .
\mdash ; 1902 .
] jrential Eof -vector .
\mdash ; Shift -metres .
4046 .
6 4045 .
\mdash ; 1902 .
] jrential Eof -vector .
\mdash ; Shift -metres .
4046 .
6 4045 .
\mdash ; 1902 .
] jrential Eof -vector .
\mdash ; Shift -metres .
4046 .
6 4045 .
\mdash ; 1902 .
] jrential Eof -vector .
\mdash ; Shift -metres .
4046 .
6 4045 .
\mdash ; We find then that the known direct effect of pressure on the radiation or absorption lines is the same , in quality , in water as in air , that is , we get displacements in the ] to that we observe the dark lines to occupy in the of Novae , and we find further that the amount of shift observed in the spectra of new stars differs llot only in this respect in degree , thus:\mdash ; Spark in water .
1 .
Absorption lines least shitted .
] inesmo st hifted .
2 .
Radiation ines most shifted .
Radiation lines least ifted .
3 .
Absorption shift small .
Absorption shift enolmous .
It would bhus appear that the pairs of bright and dark lines shown in the spectra of new stars do not arise from the cause which produces the appearances presented in the spectrum of the spark in water .
thanks are due to .
C. P. Butler , who obtained and discussed the photographs of the spark spectra , and who , together with Dr. Lockyer , assisted me in the preparation of the paper , and to .
F. E. Baxandall , who checked the wave-lengths of the lines discussed and Qtudied the behaviour of the lines representative of the different phenomena .
" " The Differential Equations of Fresnel 's Polarisation-vector , with an Extension to the Case of Active Media By JAMES WALKER , M.A. Communicated by Professor CLIFTON , F.R.S. Received February March 6 , 1902 .
1 .
In many problems of physical optics it becomes necessary to know the differential equations that the polarisation-vector of a stream of light has to satisfy , and the boundary conditions that subsist at the interface of media possessing different optical properties .
Mr. J. Walker .
The Dafferential Equations [ Feb. 8 , These are usually obtained by formulating some theory respecting the character of the ether in the media and the nature of the vibrations in a train of waves , but there is an obvious advantage in directly sing our lnvestigations , if possible , on the known experimental laws of the propagation of a luminous disturbance .
This has been done by in the case of an ordinary isotropic medium by using the principle of interference that lies at the very asis of the sclence of physical optics , combined with the fact that the propagational speed of light is independent of the direction of the ; from the equations thus obtained he then forms an expression that be regarded as representing the energy of the luminous disturbance , and generalising this he deduces by the principle of least action the equations that refer to other classes of homogeneous media .
There is , perhaps , something artifical in this of the expression for the energy , and it is therefore better , when this can be done , to apply to each separate case the method employed by isotropic media .
2 .
This plan of procedure presents no difficulty the case of ordinary stalline media .
According to Fresnel 's laws of double refraction , the polarisation-vectors of the waves that can be propagated in any given direction , are parallel to the axes of the central section of a certain ellipsoid\mdash ; the ellipsoid of polarisation\mdash ; parallel to the of the waves , and the propagational speeds of the corresponding waves are given by the inverse of the lengths of these axes .
If then the equation of the ellipsoid be ( 1 ) , we obtain by the ordinary methods of determining maxima and minima ( 2 ) , where ( 3 ) , being the direction-cosines of the normal , the propagational speed of the wave , and the direction-cosines of its polarisation- vector .
Now if be the components of the -vector , the principle of interference is expressed by A. , * Kompendium Theoretischen Physik , ' vol. 2 , part V , SS 6 , 7 .
1902 .
] of 's Polarisation-vector .
39 where being the wave-length , and the differential equations of the vector are obtained by eliminating the exponentials and the direction-cosines from these expressions by the aid of equations ( 2 ) .
This gives at once ( 4 ) , where ( 5 ) .
If inlroduce a new vector , the time-gradient of which is defined by , equation ( 4 ) may be written ( 7 ) .
Thus Fresnel 's theory of double refraction leads vo the consideration of three vectors\mdash ; ( 1 .
) The polarisation-vector with components ( 2 .
) A vector with components ( 3 .
) A vector , such that curl and and are connected by the relation , \mdash ; curl Also the vectors and are perpendicular to one another and in the plane of the wave , and the vector is perpendicular to the vector and in the direction of the normal to the ellipsoid of polarisation at the point in which the -vector meets it\mdash ; that is , it is perpendicular to the ray .
We thus see brought out clearly the connection between Fresnel 's theory and the electromagnetic theory for crystalline media .
The boundary conditions that must be satisfied at the passage between two crystalline media follow at once , if we assume that the transition takes place by a rapid but continuous change of the properties of the one medium imo those of the other , and that the above equations hold within the region where this variation occurs .
Taking the interface as the plane , we see that these conditions are the continuity of and since the curl of a vector 40 Mr. J. Walker .
The Differential Equations [ Feb. 8 , no divergence anywhere , we may add to these the continuity of and ; but these conditions are clearly not independent of the former .
3 .
In the case of active crystals we are on less sure ground : we are without the guidance of Fresnel , who only considered the passage of light along the axis of an active uniaxal crystal , and our knowledge of the laws of propagation of lighti in such media is less definite .
, however , established that in any direction within an active niaxal crystal there are two streams of permanent that ype at are oppositely polarised their planes of maximum arallel and perpendicular respectively to the principal section , and Gou para shown that , neglecting small terms of tlJe second order , the existence as of these " " privileged\ldquo ; streams may be accounted for position of the effects of ordinary double refraction and of an ind a supere- pendent rotary power possessed by the medium .
If this be so , it is easily shown that to the same degree of approximation we have the following extension of Fresnel 's theorem respecting the ellipsoid of polarisation:\mdash ; In any direction within an active crystalline medium two oppositely polarised streams can be propagabed with their planes of maximum polarisation parallel respectively to the axes of the central section of the ellipsoid of polarisation parallel to the plane of the waves .
and the propagational speeds of these waves are respectivel in excess or defect of the speeds represented by the reciprocal of the length of either of these axes by an amount directly proportionaI to the period of the vibrations of the polarisation-vector and to the ratio of the axes of the elliptic vibrations perpendicular and parallel to that axis , of the section .
Thus if be the axes of the section , the axes of the elliptic vibration parallel respectively to these axes , and the propagational speeds of the waves , , whence , approximately , , where * ' Journ. de Phys ( 2 ) , vol. 4 , p. 142 1885 ) .
By taking the excess or defect inversely , instead of direotly , proportional to the period , is introduced in front of the terms in ( 12 ) .
1902 .
] of Fresnd 's ation-vector .
41 Now the components of the polarisation-vector of a stream of elliptically polal.ised light may be represented by the real parts of A. bars over the letters representing that they are complex , provided that the ratio is not real , and if we so choose the origin of time that then , are the direc ion-cosines of the axes of the ellipse traced by the extrenlity of the polarisation-vector and are the length of the axes in these directions .
Taking again equation ( 1 ) to represent the ellipsoid of polarisation , we obtain in place of equations ( 2 ) the two sets of equations ( 8 ) , and ( 9 ) , where is given by ( 3 ) and obtained from it by writing for Whence we have ( 10 ) , and two similar equations , being obtained from by writing for Now , being the direction-cosines oi three vectors at right-angles to one another , we have , and whence ( 10 ) becomes .and two similar equations .
42 Mr. J. Walker .
The Differential Bquations [ Feb. 8 , Hence , from the principle interference expressed by we obtain , as in S2 , ( 12 ) , which may be written in the form \mdash ; curl curl , and having the sams significance as in S2 .
The boundary conditions , obtained as in the former case , are continuity of , the interface being , together with the continuity of and , since within the transition-layer the two latter conditions are not inde- pendent of the previous four , as .
4 .
When we come to the consideration of magnetically active media , our position is still more uncertain , but the following is suggested as an extension of Fresnel 's theorem , being a'generalisation of results that appear to be established for isotropic media .
In any direction within a magnetically active crystal two oppositely polarised streams can be propagated that have their planes of maximum polarisation parallel respectively to the axes of the central section of the ellipsoid of polarisation parallel to the plans of the waves : and the propagational speeds of these waves are respectively in excess or defect of the speed represented by the reciprocal of the length of either axis of the section by an amount that is inversely proportional to the period of the vibrations and directly proportional to the length of the axis , to the ratio of the axes of the elliptic vibration perpendicular and paralle,1 to the axis , and to the compoperpendicular to the section of a vector dependent upon the intensity of the magnetic field .
Thus if be the axes of the section , the directioncosines of its normal , the components of the vector deter1902 .
] of vector .
43 mined by the magnetic field , the propagational speeds of the waves are given by Whence , approximately , Proceeding as in the last case we have instead of ( 11 ) , ( l3 ) , and two similar equations ; and applying the principle of interference we obtain the equations , which may be written in the form Ctlrl curl The interface being the plane , the boundary conditions are the continuity of , to which we may add the continuity of and , since within the .
transition-layer : the number of independent conditions is , however , only four , as required for the treatment of magneto-optic reflection and the Kerr effect , since .

