effect of electric strain is a positive or negative retardation of the one component light whose plane of polarisation is perpendicular to the line of force, the sign of the retardation being, of course, the same as the nominal sign of the dielectric. Therefore, of two vibrations which are (on Fresnel's hypothesis) perpendicular and parallel respectively to the line of force, it is only the latter that is immediately affected by the electric strain, this vibration along the line of force having its velocity of transmission retarded or accelerated according as the dielectric is of the positive class or the negative.

I venture to regard this result as a general law of double refraction in electro-optics, though the proof extends only to four different dielectrics. As the best proof that I can offer, I will merely give a condensed historical sketch of the experiments. It will be seen in this way how the law was first suggested and then confirmed by the phases of a new electro-optic effect. It will be seen also that the proof of the law is independent of all hypotheses, independent even of everything previously known in electro-optics.

The Plate Cell is a piece used in all the experiments. There is an

Earth

P.C

Fig 1.

end-view of it given in the adjacent figure. It consists of five slabs of plate glass, each 10 in. by 7 in., arranged face to face in one block. The inner rectangle represents a tunnel (5 in. by 34 in.) which passes right through the block. Inside are shown the conductors with supporting frame, the shaded pieces being of brass, and the unshaded of plate glass. The lengths of the conductors, at right angles to the plane of the figure (and parallel to the light), are 6 in. and 7 in., the thickness of the cell being nearly 8 in. By means of wires, which pass through the wall of the cell, the internal conductors are con

nected with prime conductor and earth, as indicated in the figure. It is understood, of course, that the surfaces of the two conductors are well planed and polished, all corners and edges rounded off, and the two fronting faces accurately parallel. The cell is closed, in the usual way, by panes of plate glass laid against the ends, and the whole block is kept together by a screw-press. Two borings in one of the plates provide for the filling and emptying. When the cell is put in order and charged with CS, and examined according to the old method (with a pair of crossed nicols), it gives a very pure double refraction, and acts well in all respects, except that (from deficiency of insulation) the largest effect is less than might be expected, hardly amounting to one average wave-length of relative retardation. But this defect is of no great consequence.

The First Experimental Arrangement is shown in the next diagram, in horizontal section through the lamp L and the observer's eye E, but without strict regard to scale.

Two -in. plates of glass are represented in section by the rectangles PQ, RS. Their function is the same as that of the two plates in Jamin's interference refractometer.* The plates are, therefore, parallel-surfaced, and of accurately equal thickness, and are silvered on the back as mirrors; and in their working positions they are almost exactly vertical and parallel, and at 45° to the light. A pencil of light, LB, which passes through a vertical slit in front of the lamp, is incident on the first plate at B, and is divided, in the manner shown in the diagram, into two pencils, BDCG and BFHG; and from G the light proceeds anew as one pencil, and passes through a narrow circular diaphragm,† which is fixed at E in front of the observer's eye. *Preston's Theory of Light,' p. 157.

† Or, otherwise, through a telescope.

The result of the arrangement is that, when the pieces are properly placed, the bright vertical slit L, as seen from E in the direction EG, is crossed by a set of interference-fringes. These are well defined in position by reference to a constant black line, the image of a fine wire which is fixed across the slit L. It may be assumed, without argument, that any small increase or decrease of velocity of one of the pencils BF, CG, will produce a positive or negative displacement of the fringes, at the rate of one fringe-width of displacement for every wave-length of relative retardation. As far as the assumption is required, it is easily verified by the introduction of thin plates of glass into the course of the light, anywhere between the two thick plates; and I find in this way, definitely, that (as the pieces actually stand in the diagram and in all the experiments) an ascent of the fringes indicates a relative retardation of the pencil BF.

There are two essential pieces that remain to be noticed, of which the first is the electro-optic cell. It is shown in the diagram how the laterally separated component pencils pass through the cell, BF through the electric field, and CG through the space electrically screened by the second conductor, this conductor being always to earth. The last piece is a Nicol's prism N, which is placed in the path of either of the single pencils GE, LB, with its principal section laid (1) horizontally and (2) vertically. The design of the apparatus will now be apparent, which is to give the means of detecting electrically generated changes of velocity of the light BF in two successive cases, when the plane of polarisation is (1) perpendicular to the lines of force, and (2) parallel to the lines of force. But in actual experiment there is a difficulty encountered at once, which appears at first sight to be insurmountable.

Disturbance of the Fringes.-Suppose all the pieces placed as in the diagram, the cell nearly filled with carbon disulphide, the second internal conductor put permanently to earth, and the fringes obtained in good form and position. When connexion is made between the first internal conductor and the knob of a charged Leyden jar whose outer coating is to earth, there is an immediate disturbance of the fringes, a set of large and irregular movements, with deformations, ending in the disappearance of the whole system in one or two seconds. The effects are seen better when the first internal conductor is connected permanently with the prime conductor and an attached Leyden jar, for the potential can then be raised regularly and very slowly from zero, and the full course of the disturbance takes a longer time; but in other respects the phenomena are the same as before.

When the fringes have been extinguished in this way by the electric action, it is easy to recover them, either by putting the prime conductor to earth, or by keeping the potential at a sensibly constant

VOL. LV.

T

value, high or low, for a little time. If with this view the machine be kept working at a constant rate throughout the experiment, the extinguished fringes return gradually into the optical field, and in a little time (twenty to eighty turns of the plate) they are as clearly visible as they were before disturbance; their forms also are good, and their positions approximately constant, though they do not often continue quite motionless in such circumstances, even for a fraction of a second. If the prime conductor be now put to earth for a little, and the experiment be then repeated, the disturbance passes through all the same phases as formerly, though it is more violent at starting as the preceding interval of rest is longer. All these effects come out equally well with common light, and with light polarised in the two principal planes.

This optical disturbance is evidently a remote effect of the electric action, produced immediately-not by electric strain-but by irregular changes of density in the medium. We know that in the present cell, as in every like arrangement, the electric action throws the liquid into currents, which pervade all parts of the cell and are very intense at high potential. These material currents explain the changes of density; for, at starting, they give rise to a rapid process of mixture, forcing denser masses upward into the course of the light, &c., and, afterwards, when the mixture is completed, they are still accompanied by irregular variations of pressure in the liquid. It should be easy, therefore, to imitate the effects by means purely mechanical; and of this I can give an example from actual observation.

A plate cell, about an inch thick and open at the top, was charged with water, and placed in the course of the pencils BF, CG, immediately behind the electro-optic cell; and the fringes were obtained in good form and position. The stirring of this water gave a set of optical effects that could not be distinguished from the former disturbance. And when the fringes, extinguished in this way mechanically, were well restored and made moderately steady by regular stirring kept up for a time, I found that a disturbance of the same kind could be obtained at pleasure, either by an interval of rest (the longer the better), or by the addition of a little warm water. But leaving this and returning to the electro-optic experiments, I proceed to show how, in spite of these irregular movements of the fringes, and in the midst of them all, it is possible to obtain a steady effect, which corresponds perfectly to the known bi-refringent action of the medium.

Regular Dislocation of the Fringes.-The electric arrangements are the same as formerly, the two internal conductors being connected permanently, the first with the prime conductor, and the second with earth. There is only one change made in the apparatus; the nicol

N is withdrawn, and a small rhomb of Iceland spar (about 3 cm. long) is put in its place at E, with principal section horizontal. In this way the two systems of fringes which were given by the nicol N in succession are now given simultaneously, side by side, and each the exact prolongation of the other; the successive systems (a) of the next diagram are changed into the double system (ẞ, y).

The machine is now set in motion. The system (B, y) is disturbed as was the system (a) formerly; but in the midst of the disturbance, and as long as the fringes are clearly visible, the sets (B) and (7) are

Fig 3.

seen to be relatively displaced, the system (B, y) being changed into the system (d, e). The extent of the dislocation increases as the potential rises; that shown in the diagram, which is about threefourths of the fringe-width, is not much below the highest that can be got with the apparatus. The direction of the dislocation is constant, and indicates a relative retardation of that vibration in the electric field which is parallel to the line of force; and this agrees with the known character of the medium CS, as a positive dielectric.

It is very interesting to watch the two sets of fringes () and (e), and to see them sometimes moving rapidly and very fitfully, but moving always as one system, with its two parts dislocated unchangingly, except so far as the extent of the dislocation varies with varying potential. It is equally interesting to see the effect of spark-discharge of the prime conductor, especially from high potential. At the instant of the spark there is a sudden disappearance of the dislocation, an extremely quick jump of the fringes. into line with each other, and this without perceptible check or sudden change of any kind in the disturbance-motion common to the two sets at the time. The best way of observing the effect is to take sparks from prime conductor to earth at stated intervals, while the machine is kept working at some constant rate. The dislocation then