]\gt ; 496 R. T. Glazebrook on Velocities of Normal [ June 20 , we previously observed in our iron tubes , and described in our last communication , was a reversal of part of these lines , though the latter extend much further towards the blue than we had observed the absorption to extend .
In fact , the bright lines extend somewhat more than half the distance between and , from 45 to 50 being visible , and placed at nearly equal distances from each other .
They also commence close to the up , i.e. , with a wave-length nearly 5,164 , but the first two or three lines at that end are not so bright as those which immediately succeed them .
The light giving these lines does not extend to more than a short distance from the electrodes , and is generally most conspicuous at the negative electrode .
There is a difficulty in coIlsequence of the flickering character of the discharge in any accurate measures of them , though they are bright enough , especially at the less refrangible end , to be easily seen .
The comparative faintness of the from the iron tubes appears to us almost sufficient to account for our not having seen the reversed lines so completely as the bright ones ; ] , it is quite in accordance with what we in other cases observed , to suppose that some of these Iines may be ors easily reversed at the temperature of the iron tubes than others .
XII .
" " An Investigation into the Velocities of Normal Propagation of Plane Waves in a Biaxal Crystal , with a ison of the Results with Theory By R. T. GLAZEBROOK , B.A. , Fellow of College , Cambridge .
Communicated by J. CLERK MAXWELL , M.A. , F.R.S. Received June 19 , 1878 .
( Abstract .
) In his report to the British Association in 1862 , Professor Stokes called attention to the desirability of accurate measurements of tlIe velocity of normal propagation of plane in a biaxal stal , with a view to testing by the results Fresnel 's theory of double and suggested then a method to determine this velocity .
Let the crystal to be examined be cut into the form of a prism , two or more natural faces .
left to determine accurately the position of the cut faces with reference to the axes of eIasticity .
' ' Let us consider a plane wave of light passing ough the cry stal ' ' Let be the velocity in air , in the , let their usual meanings , let be the angle of the prism , the deviation of the wave normal after passing through the prism Let us observe the of , and the deviation D. 1878 .
] Propagation of Phne Waves in a Biaxat Crystal .
Then is given by the formula ( 1 ) .
' Bub without making any other supposition as to the law of ] refraction , or assuming anything beyond the truth of Huygt ) 's principle , which , following at once from the superposition of small motions , lies at the base of the whole theory of , we may at once deduce from the directions of incidence and emergence the direction and velocity of ation in the \ldquo ; For we have\mdash ; ' ( 2 ) , ' ( 3 ) , . . .
( 4 ) .
Adding and subtracting ( 2 ) and ( 3 ) , Dividing and , This gives Combining with ( 4 ) we can get , and then find the of from either ( 2 ) or ( 3 ) .
In practice the valuc of or was used .
In accordance with these estions I undertook a sorics of obseriltions at the endish boratory , I pose to describe in the paper , adding moreover a comparison of the results ith the theories of Fresnel and Lord yleigh .
Fresnel is the only experimenter who has attempted to verify his theory by experiment , and his attempt affords no veritication , for in ) it ] used approximate resuIts , and to degree of tion to which he went the theory developed by Lord ( Phil. Mng .
, vol. Scries iv , 1871 ) leads to exactly the same equtitions to determlne t velocity of propagation as were used by Fresnel , so that his resuIts form equally a verification of Lord Rayleigh 's theory .
R. T. Glazebrook on of Normal The work was carried out on two pieces of aragonite supplied by A. Hilger .
Tottenham Court Road .
In both cases two of the faces marked 's gave the best reflexions , and were therefore reserved to afford means for the of the position of the artiticial faces .
We will consider the two separately .
The first crystal was cut at Professor Stokes ' suggestion , so as to form two prisms .
The edge of one of these was nearly parallel to the axis of the crystal , the mean axis of elasticity , so that the incipal plane of the prism almost coincided with the principal plane AOC of the wave surface , and I was able to show that the error from supposing the coincidence to be exact would never amount to as much as 00001 .
The axis OA of the crystal almost bisected the angle of the prism , which was SO ' ' .
The observations of devi tion and incidence were made with a goniometer by Grubb , lent me by Professor Stokes .
The circle , about a foot in diameter , was graduated on silver , and was read to by verniers .
Each observation was repeated two or three times on different occasions , and mean of the results taken .
It was rarely that two measures of the same tntity differed by 20 Careful .
precautions were taken to ensure the light passing in a principal plane of the prism .
The observations with this prism extended from about on one side of the axis OC to on the other , passing almost through the extremity of an optic axis .
Observations were taken at angles of incidence , increasing uniformly by , thus forming an progression .
The values of for one wave were nearly constant , and varied but little from Let , be the principal refractive indices .
The series of values given above corresponds to the circle of radius .
So that The values col.responding to the other wave varied considerably .
According to Fresnel they ought to be radii vectores to an ellipse axes and was determined by passing light along the axis OC , which wa possible , the crystal having been cut with this object in view .
I found To find recourse was had to the second prism , which had its nearly parallel to the axis of .
I found 1878 .
] Propagation of Plane a stal .
The value for the veen the optic lxes , seen in air through a face ormal to is , from these values Kirchhcff found by The reement is fairly close , much closer than given by the values of the principal indiceq as determined by Rudbcrg .
Having thus found and , we proceed to the theoreticnl of in different cctions { ompal.with tcory .
' with following esult : from about on one sido of the to on the other , theory and experiment closely .
diffcrence only in two cases mounts to is somctinlcs positive , somctimes ncgative .
[ The error in ucnt is so great But for the ncxt tions extc]-dcd , differences colltinu{lly ) ) ) for tion , the experimcntal values of ulliformly than the theoretical .
So that the results of observation would bc reprcscnted by a circlc of radius , and an ov curve with the same axcs ellipse , hich arees closely with tlIc ellipsc on cifhcr sidc of the axis , and rest of the lies outsidc between the radii vcctorcs of the cves iucl.easlng ) we recedc from The diffcrences between experiment and Lord Rnyleigh 's theory increase much more mount , at the end of the of observed , to , or about times as 011 Lord } frolll tho trtlth by more than Fresnel 's .
The second prism cut so as to its to OC .
The parnllelism , however , was not sufficiently to me to treat the principal of the prism as a section AOB of the } slowness .
fore , to determinc the values of vclocitios of normnl ) ation , from Frcsnel 's construction .
optic axes luect n tlnit spLerc ccntrc at the centre of the surface OO ' cspectively ; let ] ) the part in hich any wave nor1nal meets the Then we may that the valnes of by\mdash ; 500 R. T. Glazebrook on Velocities of Normal [ June 20 , The work extended over an arc of about , with the following results : \mdash ; For one wave the agreement was close throughout .
This section differed but slightly from the circular section of radius the other bhe differences were much greater .
The results of experiment were represented by a curve , which in the lrhood of the lesser axis of Fresnel 's section lies within that section , cutting it at about the middle of the arc considered , and afterwards lying without it .
The excess of experiment over theory changes in the arc considered from to This agrees with the result for the first prism in lying outside Fresnel 's surface as we approach the major axis .
I then proceeded to estimate the effect of any possible errors made in the detcrmination of , or the position of the plane , and showed that no change at all within the limits of experimental would reconcile theory and experiment more closely .
As a test of the accuof the experimental work , ib may be stated that a series of observations , taken at an interval of some three months previously to those ibed above , gave results which rarely differed from those results by more than .
To proceed now to the second crystal .
The measurements on it were made at an interval of nearly a year after those already described .
The crystal was in the form of a hexagonal prism , the base of the prism being nearly perpendicular to .
This base was polished .
The other elld was cut so as to be inclined to the base at an angle of the line of intersection of the faces of the prism thus formed being nenrly parallel to thab of and .
One of the faces , cut so as ) be inclined to the oblique section at about , the line of junction being nearly allel to that of and .
So thab I thus formed two prisms whose principal planes were nearly coincident , having one face in common .
By this means I was able to work over an arc which extended from the neighbourhood of the principal section AOC to more than on the other side of it .
The planes cut the principal plane , AOC , in two points , , such that\mdash ; 1878.of P'lane Wares in a Biaxal Crystal .
501 CL ALP while the position of , the normal to the common face of the two prisms , was given by\mdash ; LP or Let be the angles between ally wave normal and the optic axes .
The formulae used to calculate were\mdash ; , .
The values used for , were\mdash ; These values differ htly from those used in the first part of the paper , not by nearly so large amounts as Rudberg between the values for two different speclmens of aragonitc .
He hatl differences of more than ( ' ' Pogg .
AnnaIen xvii , 1 ) .
For the outer sheet , which differs least from a sphere , the results of theory and experiment agree closely .
But for nner .
sheet the curve given by theory agreeing with given by expcrlment at the extremity of an axis , lics outside of it throughout the wholo of its course away from that axis .
The differcllce is test at about away from the principal scction , being there as reat as 0009 .
From this point the differcllccs , and at the end of the arc considel.ed , or away from the same axis , the value is I found also that no the values of , would produce closer reement , but that if we suppose the , to be increased by 17 ' , and hc angles ALP , AL'P , by about , { hc differences between theory and experimont were uced about 00006 , taken oughout the arc , somctimcs positive , solllotimcs tive .
I have , however , that this 1n tho positiou of involves alterations in the measures by its position tive to tho faces was determined , which are far in cxcess of itny possible expcrimental error , and that only of for them is by supposing the axes of elasticity of arag.onite to be htly variable in position relatively to the faces of the crystal .
502 On Veloci of NormPropagation , .
[ June On tlJe whole , however , I prefer to regard Fresnel 's theory as a close first approximation to the truth .
and to look to the phenomena of dispcrsion to the variation from it .
I may perhaps be allowed to close with a suggestion , which , judging from the results of a fe experlments I have already made , appears to have some basis of truth .
Let us suppose that in a crystal where , \amp ; c. , are functions of the direotions of vibration and propagation .
Let us suppose that for waves of infinite length Fresnel 's construction is true , so that : is a radius vector of Fresnel 's surface of wave slowness , and can therefore be calculated , and suppose we neglect the terms , \amp ; c. Observing the values of in this dicection for different rays , we get , \amp ; c. So that , \amp ; c. The results of experiments on the rays , and , in two different directions give ' These numbers , especially the last , are sufficiently close to it worth while continuing the observations .
The Society adjourned over the Vacation to Thursday , November 21 .

