reappears immediately after each of the sparks, increasing regularly from zero as the potential rises, and then increasing and decreasing quickly or slowly as the potential rises and falls quickly or slowly. Even when the potential falls most rapidly, as in spark-discharge, the direction of the backward jump is always evident to the eye; otherwise the disappearance of the dislocation in that case is so very quick that one would call it instantaneous.

Very little need be said upon the optical theory of these phenomena. What must be remembered is, that each of the sets of fringes (*) and (e) is due to the interference of two such pencils as BF and CG reunited in GE, the vibrations being horizontal in one pair of interfering pencils, and vertical in the other pair. With regard to changes of refringent power which are due to mechanical disturbance, it may be assumed that these are independent of the direction of the vibration: both pairs of pencils are therefore similarly and equally affected at each instant, and the corresponding displacements of the two sets of fringes are at each instant similar and equal, however irregularly they may vary from one instant to another. It is otherwise with the bi-refringent action of the medium; for here the two pairs of pencils are differently affected at each instant, and the difference is determined solely by value of potential, so that the corresponding effect comes out steadily in the midst of all the irregular changes which are produced by mechanical disturbance of the dielectric.

I think it must be admitted that in this regular dislocation of the fringes there is a new and clear presentment of the double refraction which is produced by electric strain. I think also that the new effect is made all the more suggestive by the regularity and perfect steadiness with which it comes out in the midst of the disturbance.

First Appearance of the Law.-Before leaving the present experiments I must notice one or two facts observed, but not yet mentioned, that go towards a solution of the question with which we started. The phenomena to which I refer were seen clearly enough in some of the earlier experiments; but it was only at a later stage that they were well understood, and they were then obtained more regularly.

Beginning with the last form of the experiment-that with the rhomb of Iceland spar as eye-piece. The spar, I should mention, was always so placed that the plane of polarisation in the set of fringes (e) was vertical. What I have to notice now is a peculiar feature of the jump of the fringes at the instant of discharge. To a carefully strained, as to an unstrained, attention, this jump appeared as a movement of the set of fringes (e) down to the level of the set (8), never as a movement of the set (ô) up to the level of the set (e). I must say, however, that the accuracy of this perception or judgment was to myself in some degree doubtful, not because of any expectation that could have led to it, but because of the very fugitive character

of the phenomenon, and its partial obscuration in many cases by disturbance.

Returning, therefore, to the first form of experiment, that with the nicol N as eye-piece. When the principal section of N was horizontal, and the vibration directed therefore along the line of force, there was a perfectly regular jump of the fringes downwards at the instant of discharge; and at high potential the effect was large and strikingly distinct. When the principal section was vertical, there was nothing regular of this kind seen in any of a large number of observations : there were disturbance-movements at or about the instant of discharge, as before and after, but nothing that could be accepted as a regular jump of the fringes at that instant, always in one direction or always in the other. The interpretation of these results is obvious. I have already stated, as a matter of observation, that a rise of the fringes indicates a relative retardation of the pencil BF which passes through the electric field. From the downward jump of the fringes in one of the two cases, we infer therefore that the pencil BF is in that case relatively accelerated in consequence of discharge. But in the present experiment, and with reference to the pencil BF in relation to the pencil CG, it is evident that relative acceleration and absolute are equivalent; because it is only in that division of the cell through which the pencil BF passes that there is any sudden physical change at the instant of discharge. It appears, therefore, that to relieve the liquid of electric strain, is to relieve one of the vibrations (that along the line of force) of an absolute retardation, leaving the perpendicular vibration unaffected.

In several of the later sets of these experiments with CS, as dielectric, and with nicol N as eye-piece, I got what appeared to be a perfectly clean liquid. The potential also was made to vary regularly and very slowly; and from both causes the disturbance was very much reduced. The effects then were these :-Principal section of N horizontal: a slow ascent of the fringes during the process of charge, pretty regular, but often obscured and sometimes overpowered by disturbance; the contrary jump seen always at the instant of discharge. Principal section of N vertical: irregular, and generally very small oscillations of the fringes during the process of charge; but no regular motion in one direction or the other exclusively, either during the process of charge, or at the instant of sparkdischarge from high potential.

From all these experiments with CS2, it seems to follow that of the two principal vibrations, the only one immediately and regularly affected by electric strain is that along the line of force. This conclusion requires and well deserves to be verified; and I proceed to verify it by another method, or rather by the use of new means.

The Second Experimental Arrangement. The optical instrument

here used is known as Jamin's Interference Refractor for polarised light. For a description of it I might refer to a paper already published; but I think I ought rather to describe the apparatus here again. The essential pieces are shown in horizontal section in the following diagram.

R and S are large blocks of Iceland spar, of equal thickness, their principal sections horizontal, and their faces parallel. A pencil of light from a vertical slit L, passes through a Foucault's prism H, and is polarized by it at 45° to the vertical, and then enters the rhomb R. The two pencils emergent from R pass immediately through a half-wave plate P, so placed as to interchange the two planes of polarisation. Ordinary pencil and extra-ordinary in the crystal R become thus extra-ordinary and ordinary in S, and the bi-refringement action of R is neutralised by that of S. The light enters R and leaves S as a single pencil, but between P and S it passes as a couple of pencils, BF and CG, about 14 mm. apart, and polarised in planes vertical and horizontal. The pencil emergent from S is received at E through a Nicol's prism N, which is laid as for extinction with the Foucault H. When all the pieces have been properly placed, the slit L is seen crossed by a set of interference-fringes, and these are modified at pleasure by fine screw movements of the spar S.

The electro-optic cell is not given in the diagram. It is the same piece as that shown in the diagram of the first arrangement, and is placed here exactly as there, so that the two laterally-separated component pencils pass normally through it, BF through the electric field, and CG behind the second conductor.

The only other optical piece employed in the experiments is a Jamin's Glass Compensator, which is placed immediately in front of the spar S; it enables the observer to specify small differences of retardation of the pencils BF and CG.

The results obtained formerly (with nicol N as eye-piece) were fully verified with the new apparatus. The method finally adopted as the best was so similar to the former, and the effects also, that any long description of the experiments would be superfluous. But to give a fair view of the results I will describe one day's work.

* “On the Bi-refringent Action of Strained Glass," 'Phil. Mag.' for October,

1888.

+ Preston's Theory of Light,' p. 159.

Final Experiments with CS.-The first internal conductor connected permanently with prime conductor without Leyden jar, the liquid quite clean, and the conditions of electric work perfect all day. The observations were taken in five successive sets.

First Set.-Plane of polarisation of the pencil BF (through the electric field) vertical, or perpendicular to line of force: Rise of fringes indicates relative retardation of that pencil. When the fringes were obtained in good form and position, the machine was started, and kept working at a constant rate throughout the experi ment. As formerly, the first effect was a large disturbance, the fringes being displaced and deformed, and disappearing altogether at the second or third turn of the plate; but in a little time (thirty or forty turns) they reappeared in good form and approximately constant position. For distinctness of effect the central fringe was brought back to the line of reference (generally downwards) by a small screw movement of the spar S; and then, at every spark from prime conductor to earth, there was a quick downward jump of the fringes, the effect being as distinct as possible from the irregular and slow and generally small movements that went on before and after the spark. As the experiment proceeded the liquid was more thoroughly mixed, the disturbance decreased, and the effect came out much more purely. Sparks were taken repeatedly at every 3rd, 5th, 10th, 15th turn of the plate, and the jump was there in every instance, and beautifully distinct. The extent of the jump varied from about one-third of the fringe-width at every 5th turn of the plate to about three-fourths at every 10th turn. I should add that the disturbance movements, though they were greatly reduced at last, were still such as to prevent any good static observation of the fringes.

It is proved clearly by this set of observations that when the plane of polarisation is perpendicular to the line of force the light is absolutely retarded by electric strain. The spars R and S were now turned round LE through 180°, and the pieces were moved across the optic bench into good position.

Second Set.-Plane of polarisation of the pencil BF (through the electric field) horizontal: Rise of the fringes indicates a relative retardation of that pencil. The method was the same as in the first set, sparks being taken from prime conductor to earth at regular intervals, long and short. When the initial disturbance was over, movements of the fringes were still seen, sometimes in one direction and sometimes in the other, but not exclusively or specially at the instant of discharge. These disturbance movements were slow and generally small; and as the experiment proceeded they became very faint, and were occasionally not seen at all for a little time. As to the effect specially looked for, I need only say that in several scores

of observations, taken at different potentials, high and low, there was not a trace observed of a regular jump of the fringes at the instant of discharge. It appears, therefore, that when the plane of polarisation is parallel to the line of force, the light is neither retarded or accelerated by electric strain. The spars R and S were now turned back through 180°.

Many observations were regularly; but they were

Third Set.-The same again as the first. taken, and the former effects were obtained now more striking, because of the strong contrast with the negative results of the set of observations immediately preceding. The action appeared also to be stronger than before, probably because of improved insulation. The extent of the jump, taken at every 5th turn of the plate, was now half the fringe-width; and at every 10th or 15th turn it was clearly four-fifths. I find in my notes that this large jump of the fringes impressed me here, again and again, as a thing peculiarly beautiful.

Fourth Set.-The same again as the second. The only question in this case was, whether it might still be possible, by the most careful work and under the best conditions attainable, to detect a very small jump of the fringes at the instant of discharge. Many observations were taken at high potential, some at the highest, but without a trace of effect of that kind.

Fifth Set. The same again as the first. The results of the first and third sets were recovered regularly. Sparks were then taken, sometimes at every turn of the plate, sometimes oftener. At every spark there was a very small downward jump of the fringes, so small sometimes as to be barely caught, but quite regular and beautifully distinct.

Remarks. The jump of the fringes was chosen as the principal object of observation, because it was never quite concealed, nor even much obscured, by the mechanical disturbance of the liquid; but I should add that the contrary motion-the gradual ascent of the fringes during the process of charging-was generally evident enough in the experiments, though not often undisturbed or quite regular in its

course.

The best observations were got when the fringes happened to continue at rest through a sensible interval of time, including the instant of discharge. The contrast between the two cases was then very remarkable, especially at high potential; in the one case, the beautifully clear jump so often mentioned; in the other case, no trace of a jump in either direction, generally not even a perceptible shiver of the fringes at the instant of strongest discharge. Instances of this kind occurred not very rarely in the experiments; and there could be no contrast more striking than that between the phenomena in the two cases.