]\gt ; The ersion of Double cxction in to By T. H. HAVELOCK , M.A. , D.Sc .
, Fellow of St. John 's College , Cambridge , Lecturer in Applied Mathematics in Arnlstrong College , Newcastleon-Tyne .
( Communicated by Professor J. Larmor , Sec. R.S. Received July 5 , 1907 .
) In a previous communication , writer discussed the double refraction of a medium composed of similar particles arranged in rectangular order , the results })plied especially to the produced in colloidal solutions by mechamcal stress and by a magnetic field ; the present paper is the discussion of a similar problem , a ) method of procedure suggested by Professor Larmor .
The object is to investigate to what extent it is possible to consider double , whether produced artificially or occurring in natural crystals , as due simply to an oeolotropic distribution of similar particles ; the questions which arise when one considers whether the aeolotropy occurs naturally or is produced by the action of mechanical , electric , or mag1letic force are not specially considered , but the various cases are classified in the section .
In the next sections we considel a lnediul1l ) osed of a homogeneous assemblage of optically isotropic molecules and obtain an expression giving the dispersion of the double acCion .
This is applied first to artificial double refraction produced by mechanical stress and by an electric field ; then on the same basis the dispersion in quartz is examined .
Certain divergences in natural crystals and cially such lnomalous cases as ylite and vesuvian to an exCenfflon of theory .
This is , bl'ieiiy , all aeolotropic ibution of optically cnles , each of which disperses regularly ; it is exalnined first uniaxal crystals and finally in general for biaxal cryfitals , and is to contain the possibility of all the varieties of dispersion of double tion which occur in natural crystals .
1 .
Thjorics of The ordinary theory of double refraction .
in as it is ecular , amoumts to considering the as a collection of crystal molecules in ical order , the vckrieties of ) ) ostulated of the single particle .
Thus with the principal inate { be the .
Proc , vol. , 170 , 1906 . .
180 .
Daspersion ofDouble in Relation to Crystal Struct ure .
29 polarisation of a typical particle at which the ective electric force , we have equations of the form .
( 1 ) These equations mean that the aeolotropy is assumed to inate w in the individual crystal molecule , and that it affects the quasi-elastic force under which the polarisation electrons are supposed to vibrate .
The same method has usually been followed in attempts to electron theory to the double refraction due to mechanical stress .
It we consider the equations for an isotropic medium , to ( 1 ) with a single constant , then Voigt*lmes that is altered by amounts proportional to the stresses in the medium ; this is a direct effect on the internal forces of the molecule , while on the other hand a theory which confines the effect to a rearrangenlent of the molecules in space will the result by a modification of the effective force in the previous equations .
The latter theory appears to comprise the double refraction effects obtained by Majorana by placing certain colloidal solutions in a field .
These effects appear to be quite distinct those involved in the Zeemann effect .
The double refraction obtained in the latter way is explained by a direct action of the magnetic field on the electron , and a result of the is that the extraordinary and ordinary indices both from the index of the isotropic medium under no netic a the variation of the two indices is in the same direction , one three tlmes the other , and the difference varies sely as the of the In the Majorana effect , on the other hand , the of the two indices are in opposite directions , one ately tvice the other .
The same distinction occurs in the theory of the effect , that is , of double refraction produced by an electric field .
has extended his theory of the Zeemann effect to cover an effect by an elpCtriC field , the results being of the same character .
Now Kerr 's experiments appeared to prove that only the index for vibrations parallel to the { ield was altel'ed ; , in some experiments , claims to have observed in Voigt , ' Annalen der ysik , ' vol. 6 , p. 461 , 1901 .
Voigt , ' Annalen der ysik , ' vol. 4 , , 1901 .
.eckerlein , 'Physikalische hrift , ' vol. 7 , ) , 1906 ; also , ?
bicl .
, p. 811 , Dr. T. H. Havelock .
Dispersion of Doubfe [ July 5 , certain liquids a change in both indices , the variations being in opposite senses and one apparently twice the other .
It is , of course , possible that there may be some direct effect of the electric field upon the periods of the electrons , in addition to a merely mechanical effect of rearrangement of molecules .
For the present we are concerned more with working out the possibilities of the latter supposition ; that is to say , although Cauchy 's theory of optical dispersion , as caused by simple discreteness of the medium , was insufficient to account for ordinary dispersion , it is gested that it may be capable of application to the differences of dispersion occurling in different directions in a doubly refracting medium .
2 .
Optical Properties of a Homogeneous Assemblage of Molecules .
In a dielectric medium , if we write for the electric force , for the magnetic force , and for the total electric displacement , we have the usual equations curl , curl The total displacement is given by equations of the form , ( 3 ) where refers to the polarisation of a typical particle and is the number of such per unit volume .
For we have an equation , ( 4 ) where is the component of effective electric intensity considered as acting on the given particle .
The determination of the mean value of is the problem in question , and in particular , the estilnation of the influence of neighbouring particles .
In the ordinary procedure , we consider round the given particle a sphere whose radius is small compared with the wave-length of the radiation medium , but such that it contains a large number of particles ; then we evaluate separately the influence of the particles outside and within this spherical region .
1907 .
] to Structure .
The first part is taken to be the force within a sphel.ical cavity cut a medium uniformly polarised Go the value at the point in question of the vector ; this , then , contributes a force .
The second part is due to the average effect of the immediately ing particles within the sphere , and is enerally written as , where are constants depending upon the arrangement of the particles .
For a simple cubical arrangement , we the -constants to be all ; thus , as in the similar argument in the case of liquids , we are to eliminate the consideration of the purely localpart of the hbouring action .
Now , for any other regular ement of particles , the -constants of this method will not all be equal and zero ; but suppose that of a sphere we can choose some other form of surface , such that we can ) the local effect of the particles within , then we are left with the evaluation of the force within a cavity in a medium uniformly polarised to the value of the material polarisation vector at the point in question .
In particular , it is clear that for a approx tion for a ular arrangement of particles ering slightly from a simple cubical systeln , we may assume to be an ellipsoid whose axes are nearly equal .
If we take co-ordinate axes in the ections of the principal axes of the ellipsoid , the force due to the particles outside is giyen by ( 5 ) where ( 6 ) with similar expressions for and C. Hence , if we put , we have to the first power in , .
( 7 ) .32 1 Dr. T. H. Havelock .
Dispersion of Dispersion of Writing these for the moment as have , from ( 4 ) and ( 5 ) , ( 8 ) If the quantities vary as , we find from this equation an expression for ; this can be substituted in equation ( 3 ) to give a relation between and X which must necessarily be of the form , where is refractive index .
Hence we obtain a value whioh is expressed most in the form here is the refiactive index of the same substance if isotropic and of the same mean density .
Now we substitute from ( 7 ) the values for and expand from ( 9 ) to the first power in the small quantities ; we find the following values for the refractive indices of the medium:\mdash ; ( 10 ) The mean value is given by .
The medium , then , is doubly , and we shall consider first some consequences when it behaves like a crystal .
In this case we may take and as zero , and we have the extraordinary index and the ordinary index given by\mdash ; where .
The mean index , that is the index for the same medium if isotropic and of the same density , lies unsymmet rically between the ordinary and extraordinary indices , being twice a , s far from the latter as from the former .
Further , if we measure the double refraction by the difference between the two principal indices , we have 1907 .
] Refraction in ltelc , to , then , that the ellipticity of the eflective ellipsoid is constant , if we know varies with the wave e , this tiun gives of dispersion of the double refraction ; we shall this with some experimental data in the following sections .
3 .
Donble Dnc The most suitable test for the bove formula would ) in the of artificially produced double refraction , as , .
instance , when isotropie medium behaves like a uniaxal crystal if ected to mechanical stress .
However , in these cases t , he effect is extrelnely small , and the pl.incipal indices and ?
have not usually been measured separately , but ) } ) their difference .
If the dispelsion of the unstrained isotropic known either by a curve or a dispel'sion formula , then it would be possible to calculate the dispersion of the difference for instauce , if for the isotropic medium of the same density , were given by then , ince we should have given pproximately by a relation of the .
( 14 ) Further , the separations of and from should be in opposite directions , with the former approximatel .
twice the latter .
consequences rest on the initial assumption of the theory , namely , that the isotropic medium is , in fact , an semblage of optically isotropic molecules , and not a collection of crystalline molecules , their distributed equally in all possible directions .
4 .
Electric Refraction .
For the double refraction produced in lnedia by the action of all electric tield , Kerr observed a dispersive effect , and supposed that the double refraction varied inversely as the square root of the , but this has beeu shown to be erroneous by Blackwell .
* His experiments deal with carbon disulphide , and curves are given for variation of the double refraction with the wave-length for two ferent temperatures with a field strength of 56,000 volts per centimetre .
These esults give the only data suitable for the present purpose .
* H. L. Blackwell , ' American Academy Proceedings vol. 41 , p. 647 , 1906 .
Dr. T. H. Havelock .
Dispersion of Double [ July 5 , We have , then , two sets of values for the double refra.ction of at the temperatures of C. and C. , each set ving values for six differenf , wave-lengths with the value of for and taken as 100 ; absolute values can be obtained by means of a determination of for at the latter point .
For the present theory we require actual values of and iu order to calculate mean value equal ; vever , it is sufficient to take for the refractive index of under the same conditions in regard to density and temperature .
To obtain these values for required , we have a dispersion formula for at obtained by Martens , *which ives values correct to four decimal places at least ; the formula is where ; ; ; The correction for temperature can be made to sufficient accuracy by decrease of index of for } degree rise of teml ) Making these calculations , we have the of used in the following Tables I and II .
The first column gives the , the second the third the observed values of the double refraction ; the fourth column gives the calculated values of , which , according to the theory , ought to be constant .
Table I.\mdash ; Carbon ) isulphide at Mean value of ; greatest divergence per cent. approx. The fifth column in each table was obtained by the mean value of , together with the values of , in order to recalculate the values of comparison of the third and fifth colunlns shows the amount of reement between the ) served and calculated values of ; it may be ) arked that * Martens , ' Annalen der Physik , ' vol. 6 , p. 632 , 1901 .
Landolt , ' ' 3rd edition , p. 671 .
( I Iosr.edmoo j aqt XUB qonm saop J0E aas .
puooas ' .
08 III ' .
ns tSTP 0 aq Iqnop $13 \mdash ; : pug ) } puti xoJl ?
: 6 . .
Z9 9 989 I 86 T9 I 889 TI 7 899 I .
I9 4 8 .
IZT .
T9 .
IZT 9 08 T9 689 Sb ' c- .36 Dr. T. H. Havelock .
Dispersion of Double [ July 5 , calculate the values of some other relation which might suggest itself ; for instance , the quantity decreases ntinuously throughout the whole range , its extreme from a mean value by as much as 16 per cent. Table III.\mdash ; Quartz .
In consequence of the use of Babinet 's } ) sator in determining the double refraction of minerals , the dispersion of double refraction in quartz has been studied extensively .
From experimental observations in the visible and ultra-violet spectrum , Mac6 de found that could be represented very accurately by a formula\mdash ; * Mace de Lopinay , ' Annales de la Faculte des Scieuces de Marseille , ' vol. 50 , p. 1 , 1891 .
to Structure .
The numerical values of the constants were found by to fit also his observations in the infra-red .
This relation is of the form to be expected our formula and , assuming to be explessible ) a dispersion formula of the usual type .
We have assumed in the calculations that the effective cavity was an ellipsoid of small ellipticity .
We might then examine the order of magnitude of the second order terms which have ected , but it is simpler to obtain now a more coeneral relation free from this approximation .
Using the notation of equations ( 5 ) to ( 9 ) , the only assmnption we make about the effective cavity is that the force within it , due to the polarised medium outside , has components , ' , ( 16 ) where are constants .
Then assuming as before that the molecules are isotropic , we a principal index iven by The quantity differs by a constant from its value in any other state ; or , more simply , we have Consequently , if the medium is uniaxal , we have the relation constant , ( 18 ) where and are the extraordinary and ordinary indices .
Using the values given in Table III , we can examine this relation for quartz ; the result is shown in Table .
Mean value Dr. T. H. Havelock .
of Double [ July 5 , 6 .
Some Exceptional The relation obtained in ( 18 ) does not cover all the cases of regular dispersion , as we see from the ures for Iceland Spar given in Table .
But it may be noticed that in this the double refraction is 20 times reater than in quartz ; thus simple distribution of isotropic molecules is apparently not quite sufficient to account for so large a difference between the two indices .
able VIceland Spar .
Further , we have exceptional cases in which decreases with the waveIength , but much more rapidly than in other crystals .
For instance , in strontium hyposulphate , the rate of decrease is about 10 times that of quartz , and we find that the quantity .
the previous calculations is by no means constant , but decreases with the wave-length .
If we have , given by the relation we see that the substance ) be either a sitive or ative uniaxal crystal , according as is ative or positive ; but in either case the absolute value of decreases with , that is , in general , will decrease as increases .
Now this is the rule in most actual crystals , but there are a few for which increases with , although diminishes ; the figures for apophylite are given in Table for the purpose of comparison with a later set of figures .
* B. Trolle , ' Physikalische Zeitschrift , ' vol. 7 , p. , 1906 .
1907 .
] Refi.action in Relatton to S9 Also , there are crystals for which passes , and some for which the double raction c from positive to \mdash ; the substance being isotropic for some defi1liCe It appears , then , that a theor olotropic d of isotropic molecules which may serve for the al.tificial double lefiRction produced in bodies and for simple crystals is not sufficient to cover the varieties of dispersion found in natural crystals .
This is naturally to be expected ; for , in addition to the varieties of dispersion in biaxal crystals , to be considered later , we see that if such a simple theory were sufficient , then the symmetry of all the physical properties of a crystal would be the as its optical symmetl'y , and this is not the case .
The direction in which to modify the theory is clear ; we shall consider a crystalline as a homogeneous assemblage of crystal molecules .
Whether or no the crystal molecule can be identitied with the chemical molecule need not be considered here .
the general later , we shall consider the case of a uniaxal crystalline medium .
7 .
Modified of The previous theory can be extended in the manner : molecules ( or crystal units ) of the medium are not necessarily to be supposed ellipsoidal in shape , but are optically aeolotropic , so that the subsidiary equations connecting the polarisation of a particle with the effective electric force are aeolotropic , with an axis of symmetr ; the particles are supposed to be arranged in a homogeneous such that the effective cavity may be taken as an ellipsoid of revolution of ellipticity , and having its axis of symmetry coincident in direction with that of the unit .
Then , instead of equation ( 9 ) , we have now .
( 19 ) with a similar equation in and In these , and are the and ordinary indices for the substance , with a regular cubical arrangement of the same molecules in the same density .
Hence , substituting for , and their values in terms of , we have the two principal indices given by , ( 20 ) Dr. T. H. Hayelock .
ispersion of Double [ July 5 , And the double refiaction is given by .
( 21 ) These equations ( 20 ) ) would represent the effect of deformation in a natural uniaxal crystal .
) purpose we apply them to a natural unstrained crystal .
They cannot be vel.ified by actual calculation in this case , but we shall see that they cover all the nomalous cases mentioned in the previous section .
To find how varies with the wave-length , we have , from ( 21 ) , 22 ) Now the object is to account ) anomalous dispersion of in regions free from absorptio ] ) .
Then if and follow the regular law of diminishing as increases , we have both and negative ; and as we can have greater or less than and positive or ative , we see that it is possible to have positive .
In fact , it is possible to have the varieties of change of aa increases : ( i ) slow decrease , ( ii ) decrease , ( iii ) slow increase , ( iv ) passing through a minimuln , and ( v ) changing in sign .
hese cases all occur , typified by quartz , strontium phate , apophylite , and for the latter cases by varieties of vesuvian and melilite .
* The theory can be illustrated best by np from a known crystal of ular dispersion of double refraction one which behaves like apophylite .
Suppose we consider the uniaxal crystal phenakite , for which there are available determinations of the principal in dices for five diHerent wavelengths ; we denote these indices by and we find their difference diminishes as the wave-length increases .
Suppose , now , that the crystal molecules of phenakite are so that the effective cavity has an ellipticity of , and has its axis along direction of the axis of symmetry of the crystal units ; then by equation ( 21 ) we may calculate double refrac , tion of the medium thus built up .
The results are shown in Table .
We see that although the doul ) refraction of the phenakite crystals follows the regular law of decreasing as increases , yet the double refraction of the distorted medium increases as increases , and is of about the same order as that of apophylite ) .
This is , of course , only intended as an illustration of the possibility of explaining the irregular dispersion in regions which are not in the vicinity of absorption bands .
Hlawatsch , ' Tschermak 's Mitteilungeu , ' , p. 415 , 1904 .
1907 .
] Refractio in to Crystal Table \mdash ; Phenakite with 8 .
Dispersion in Bic We consider now the previous theory in the most general form , namely , the double refraction of a medium composed of optically aeolotropic particles with the same orientation and so as to rive an ellipsoidal effective cavity with its principal axes in any three fixed mutually perpendicular directions ; we shall indicate briefly the general character of the results .
We take the principal axes of a typical particle as co-ordinate axes , and have , , and , as the direction cosines of the principal axes of the effective cavity .
Then if is the polarisation of the medium , the -component of the force within the cavity due to the polarisation is given by , ( 2.3 ) where , are the usual constants for the cavity and can be written as\mdash ; Further , if we write for the principal indices of a medium composed of the same units , but in regular cubical order , we three ations of the form We can obtain now three equations connecting the components of electric force with the polarisation ; but if are the direction cosines of a principal axis of the resulting ellipsoid of polarisation and if is the corresponding principal index of refraction for this direction , we must have .
( 25 ) Hence , for the detelmination of the principal axes and indices of the medium we obtain three equations of the type , ( 26 ) Dr. T. H. Havelock .
Dispersion of Double [ July 5 , where the -constants are given by relations of the form ; .
( 27 ) We shall examine now the various types of crystals covered by these equations .
( a ) system.\mdash ; If we take simplest case , we have the principal txes of the crystal unit coinciding with those of the ctive cavity ; then evidently the three principal polarisation axes are in the same directions , and llaintain these directions for all wave-lengths , as in crystals of the prismatic type .
From equations ( 26 ) and ( 27 ) we see that the plincipal refractive indices are give by ; , 2 , 3 .
( 28 ) Thus we have ( 29 ) Hence , as in the theory of S7 for xal crystals , we may have the indices varying differently with the wave-length and consequently the possibility of such crystals as brookite , in the order of magnitude of the indices changes and the plane of the optic axes is turned through a right angle .
( b ) Alonoclinic Suppose , now , that one principal axis of a particle coincides with a principal axis of the cavity , then this will be an axis of the polarisation ellipsoid which is fxed in direction ; thus we obtain the monoclinic system of crystals with a single plane of symmetry .
Without discussing the equations in detail , we consider , as an illustration , the dispersion of the other two isation axes .
These are at right angles to each other and lie in the plane of symmetry of the crystal .
Let be the angle giving their position with reference tu the fixed co-ordinate axes , and the corresponding for the position of the axes of the effective cavity in the same plane .
Then , supposing the plane of symm to the plane , we find , after substitution in ( 26 ) and ( 27 ) and reduction , that .
( 30 ) Now , and are assumed to be of the wave-length ; suppose , then , that indices and for crystal molecules in cubical order decrease arly with the wnve-length , so on the simplest supposition we have dispersion formulae of the type 1907 .
] in to Crystal Then from ( 30 ) we see that the dispersion of the angle is by a rel tion of the form .
( 31 ) Sufficient } ) imental data on the dispersion of the principal axes of monoclinic crystals are not available to test this ) a similar ation without terms in has been suggested on other rounds b and found to well with some cases .
( c ) Anorthic .\mdash ; Finally we have the general case in which none of the axes of the crystal unit coincides with a principal axis of the cavity , and consequently the three principal axes of risation all chan , both in nitude and direction , with the includes , then , all anorthic crystals .
9 .
inning with a theory intended apply to artificial double refraction produced in an isotropic medium , we found that the consequences could be expressed simply in the following mallner:\mdash ; If is a refractive index of the substance , the quantity is altered by an amount independent of the wave-length , and varying only with the direction ; this implies a small which is proportional to , where ?
refers to a standard state of arrangement in ular cubical order .
to natural stals , we find anomalous dispersion of the double refraction in regions away from absorption bands , and a theory modified to account for these cases leads to a consideration of the optical relations of crystal structure .
In this connection it is of interest to compoxe some recent work in two directions .
On the one hand we ttempts to explain crystal structure as a of chemical constitution ; the crystal is described as a homogeneous of spheres which are combined into sub-groups defining the crystal unit or nolecule .
On the other hand , we have work connecting dispersion of double refraction in ronps of minerals with their chemical constitution , some members being found to possess a lninimum of double refraction .
Consequently , it should be possible to explain the varieties of dispersion of double refraction by means of the structure of the Cl.ystal , and we have attempted this in the previous sections , with the mptions : The crystal unit contains electrons , so that their ffect is expressed by principal equations connecting the } ) tion of the Nakamura , ' rift , ' vol. 6 , , 190 Hlawatsch , loc. 44 ofDouble Refraction in Relation to Structure .
unit with the effective electric field ; if , then , these units are in regular cubical order , we have a medium with principal refractive indices along hree fixed directions in space , and we assume that in this case we have regular declease of the double refraction with increasing wavelength in regions away from absorption bands .
considering in general any other homogeneous , we ) the eff'ect by a change in the effective electric field acting on the crystal unit ; this effect we have estimated by supposing , as a cient approximation , that the effective cavity is htly ellipsoidal instead of being spherical .
Thus differences of packing of the crystal molecules are represented opticalIy by variations in the ratios of the axes of the effective cavity and in their directions in space compared the polarisation axes of the individual unit .
Combining these assumptions , we have shown that they are sufficient for a descriptive theory covering the varieties of dispersion of double refraction found in natural crystals .

