thermo-electrical energy. With the arrangement described (9), the rate of decomposition per diem by the battery was taken at the end of 90 days' continued action, when I was certain, from previous experiment, that the joints had greatly changed; they were then resoldered, and the voltameter attached, in as nearly as practicable the same condition as for the first rate. This experiment shewed no alteration in the rate, from which we may infer, that thermo-electrical batteries are capable, at moderate temperatures, to generate their currents for almost an indefinite length of time.

Two single pairs of antimony and bismuth bars, fig. 5, decomposed sulphate of copper, and Argento-cyanide of potassium, for two months respectively.

a antimony bar, b bismuth bar, weight of the pair 19 grains, f projecting point made to touch a blue flame, ee the poles, de decomposing cell same as fig. 4.

Fig. 6.

A thermo couple of antimony and bismuth bars, soldered by pure bismuth, and excited like battery described (9) for 120 days, separated in the joint, at the end of this period, by a light touch; and a small quantity of the bismuth at the part where it has been in contact with the antimony, rubbed off in a pulverulent form. The antimony at the soldering was unaltered. This experiment appears to me to be of value, to shew the nature of the change in metals attendant on thermo-electrical actions.

31. I now entered on a series of experiments, to ascertain the alterations in the texture of metals, and their changes in density from annealing, as shewn by the balance, their fractures, and the galvanometer; of these it will be sufficient here briefly to notice the results.

Cast bars of antimony, bismuth, and zinc, never changed their fractures, although annealed for four months at a temperature little lower than their melting heat. The grain of some of the specimens tried, was of a fine unnatural texture, occasioned by casting in a cold metal mould, yet the annealing made no alteration in this grain. Rolled zinc, where the crystals have been bruised, undergoes a marked change by annealing at temperatures 30° higher than our summer weather; but the extent of the change to a larger grain is much governed by the annealing temperature; thus, with a temperature of 150°, it is impossible to produce the same size of grain as 250° gives, although the time of annealing the first is prolonged ever so much beyond the period allowed the second; but annealings at low temperatures required more time for the full development of their change than was necessary for higher temperatures.

With iron I have found the same order to prevail in the results; but it requires high temperatures for all those changes, to be recognised by the balance or the fracture. Iron softened at 400° for 3 months, did not sensibly change the specific gravity. Soft steel retains its density with remarkable uniformity; although, in hardening, the loss exceeds

1 per cent. in density, this process was repeated fifty successive times, without any change in the specific gravity, when tested in a soft state. Antimony and bismuth bars, when often heated and immersed in cold water, lose density on each repetition of the heating and immersion; they become hollow in the centre, and full of fissures; zinc bars continue solid under the same treatment. The following metals lose density by annealing :-Bismuth, zinc, lead, tin, copper, silver, iron, gold, platinum. Antimony cast in a cold mould, and hard steel, gain density by annealing. Zinc, cadmium, lead, and tin, gain density when poured on ice, compared with the same pieces poured on a flat surface, and slowly cooled. Antimony and bismuth shewed no change in density when treated like zinc, cadmium, &c.; but when cast in a confined cold metal mould, in bars similar to those used for battery, fig. 1, they lose about th of their specific gravity. Copper and silver lose density when poured on ice, or on cold metal surfaces.

In these experiments, bismuth and antimony resemble hard steel, in so far as the process of hardening is concerned; but in the annealing, antimony only follows the same change as the steel, and gains density. Bismuth, when cast in a metal mould immersed in dry snow, loses, part of its density in the lower portion of the bar, yet a still further loss of about part is produced by annealing; while antimony, cast under similar circumstances, suffers the same loss of density; but with annealing always regains part of the loss.

In comparing the changes in the densities of the metals for equal intervals of temperature, in the foregoing experiments, bismuth passes through the widest range of variations, then follow zinc and antimony.

32. Experiments with antimony and bismuth couples were now made, to ascertain if, while they were engaged to develope thermo-currents, the same changes in density took place; the mean density of three specimens of bismuth changed during twenty-one days' action from S.G. 9.853 to 9.838, three specimens of antimony for the same period S. G. 6.645 to 6.670. The direction of these changes agree with the former experiments where the metals are operated on singly. At first sight I expected they would exercise much influence on the energy of thremo-electrical currents developed by antimony and bismuth; but from a number of clear experiments with bars in different states combined in thermo couples, I have found the more nearly the bars approach to their natural density, when in a soft state, the better they are adapted for energetic action in combination with another metal. The experiment which first shewed me this fact, should sufficiently illustrate the whole of the class. A bismuth bar, combined with a bar of soft steel, and attached to the galvanometer, was heated by oil to 95°, the deflection in the galvanometer was 67°. The bar of steel was then untied to harden it; after hardening, the arrangement was now replaced, to be in every respect the same as before, when the couple of bismuth and hard steel shewed a diminution of 2° in the deflection of the galvanometer needles. This experiment was many times repeated with other elements, all of which agreed in shewing a result similar to the above, and to develope the greatest amount of electricity from a given couple. Both sides must be thoroughly annealed; hard bismuth,

in connection with hard antimony, are not so effective for thermo-bars, as these metals in a soft quiescent state.

33. When thermo-electrical currents developed in wires or thin pieces of one metal are examined, an opposite order prevails; for the further a portion of the metal can be forced from its natural density, the better will the arrangement answer as a thermo-couple; a hard and soft piece of any metal form a pair of elements, and their point of junction corresponds to an ordinary thermo-joint. A slip of watch-spring with a part in'its centre hardened by heating and quenching in water, is a good specimen of this kind of thermo couple; for by heating the point of junction of the two densities only 10. above the adjoining parts, distinct evidence of a thermo-electrical current is given.

M. Becquerel has shewn, that, when a portion of wire is hammered or twisted, a current of electricity is found passing when the part is heated; here the density of the metal has been changed, and the galvanometer merely tells, that, in all probability, the texture of the wire is slowly equalizing; however this may be, the experiment suggests an application of the galvanometer, which renders it a most delicate test of any change in the density of metals. For this purpose, a wire of the metal which the operator desires to test, take for example iron, is repeatedly drawn through the draw-plate to harden it; this hardened wire is then filed into two halves'; the contiguous parts where it has been cut are now tied together without twisting, and the necessary connections of the other extremities made with the galvanometer. On heating their points of contact, no electricity is detected. But when one of the halves of the wire has been annealed for three weeks at 400°, then again tied to the hard half, and the foregoing experiment repeated, the galvanometer immediately shews the passage of a current from the soft half to the hard, thus proving that the temperature of 400° has effected a change in the iron, which so far as I have been able to test, cannot be detected either by examining the fracture or trying the specific gravity.

34. The direction of the thermo-electrical current developed betwixt hard and soft portions of the same metal, either in two separate pieces, or, what is more convenient, in one continuous piece, where the nature of the metal examined admits of this, was now tested in bismuth and 7 of the malleable metals, and for all of them found to be from the side of the joint which was losing density to the hard part of the wire or bar. Steel, hardened and made dense by hammering, agreed with the above 8 metals. But steel, hardened by quenching in water, and antimony cast in a cold mould, and formed into couples, with soft steel for the former, and antimony cast in a mould heated nearly to redness, for the latter, the direction of the electrical current, with reference to hardness, appeared to be reversed; yet there is no change, when examined, with reference to the action in the joint, for in the two last cases the hard portions are gaining density, while in the 8 other metals the hard portions lose density.

The consideration of these experiments with metals in different states of density, together with the experiment described (30), where an energetic thermo-current appears to be produced in a vast quantity, from a

slow disintegration of the bismuth, suggest to me the view, that the thermo-electrical current in a joint, as a general principle, passes from the side which is changing towards that which remains more stationary ; and to develope it in its highest energy, the couple should consist of a changeable half (bismuth) and of a fixed half (antimony). To this view it may be objected, that the experiments given above shew that antimony is a metal which can be forced through considerable changes in density, but they also shew that such alterations are injurious to its action as a negative thermo-electric (32); and further, it differs from other metals in gaining density by annealing, so as to approach to its natural statical state. When both sides of a thermo-couple are changing, then the current represents the balance betwixt the two actions. Where a reversal of the current is obtained by varying the temperature, the balance of change is first on one side then on the other; a good experiment of this kind is shewn with an antimony and iron thermo-couple. The iron is the thermo-negative metal for all temperatures below 160°; at this point the two sides of the joint change alike; and for higher temperatures antimony is the negative metal. An extension of this experiment, with iron and antimony elements, shews its great value for explaining the nature of thermo-electrical action; for when the antimony has been cast in a cold mould, the reversal of the current, at a temperature about 160°, is then constant and most decided; but if the antimony is cast in a mould heated nearly to redness and slowly cooled, the iron is on the positive side from the commencement of the heating of the elements, the reversal of the current disappears.

The view I have attempted to trace above, as governing in all thermoelectrical arrangements, appears to me to be applicable to hydro and frictional sources of this agent, where the current passes from the side disintegrating or changing towards the stationary element of the arrangement which continues unaltered; and that the numerous instances of variations in the direction of frictional-electrical currents may be explained on the same principle as the reversal of a thermo-current by a change of temperature, viz., that the greatest amount of change is in the rubber when it excites glass, and, again, when the same rubber excites resin, that the resin then undergoes the most rapid change.

From the last experiment with iron and antimony, it will be apparent, that in batteries, fig. 1, to be excited by the weather, the iron might with advantage supply the place of the antimony; and I would recommend it to be so applied, but for the practical difficulty of soldering pure bismuth to iron. From some other experiments with bismuth, as the thermopositive metal, attached to lead, tin, and alloys of these metals, I could recognise no diminution in the thermo-electrical energy for these couples, when heated in oil to 95°, compared with a couple of bismuth and antimony; so that if the form of the battery, fig. 1, described at the commencement, should prove troublesome, on account of the fragile nature of the long bars, lead may well supply the place of the antimony used for them. Copper and zinc, in connection with bismuth, stand a little below lead. Antimony cast in a hot mould is much more brittle than when cast in the same mould cold. R. ADIE.

ERRATA in the portion of the paper in Volume 35.

Page 347, line 19, for " 80 inches circuit" read" 180 inches circuit"

...

348, the letters c and d omitted at the poles of fig. 1.

348, last line, for "lightest" read " highest"

350, line 7, for "with my" read "with these"

350, line 35, for "pole" read " pile"

351, line 22, for " electrolice" read "electrolite."

N.B.-Authors ought to be more correct in their M.S. in order to prevent lists of errata.-EDIT.

On the Determination of the Index of Refraction by the Sextant, and also by means of an Instrument depending on a new Optical Method of ascertaining the Angles of Prisms. By Mr WILLIAM SWAN, Teacher of Mathematics, Edinburgh. (With a Plate.) Communicated by the Royal Scottish Society of Arts.*

The powers of transparent substances in refracting and dispersing the rays of light, present phenomena of the most interesting description; and the accurate determination of such physical properties has always been regarded as an important branch of experimental science. While these inquiries are valuable in a scientific point of view, they are also of obvious utility in relation to the useful arts, as affording the means of constructing with accuracy many of the most important optical instruments; and, accordingly, the examination of refractive and dispersive powers has occupied the attention, not only of merely scientific observers, but also of the most eminent practical opticians. Newton examined a considerable number of substances; and his sagacious conjecture of the inflammable nature of the diamond, and also of one of the ingredients of water, from their great absolute refractive powers, has been remarkably verified by modern discoveries. affords an example of the importance of the refractive index as a physical character; and it also shews the valuable results that might be obtained by a more perfect knowledge of the connexion between the chemical constitution of bodies and their optical properties.

This

* The paper of which this is an abstract, was read before the Society, 12th June 1843; and the Society's Gold Medal, value 15 Sovereigns, awarded 13th Novem→ ber 1843.