63. This latter line is one of the lines of greatest energy; the corresponding line of greatest energy being parallel, but on the opposite side of the cylinder (56).

64. Similar phenomena will be displayed by making any other part of the convex surface near to the ends of the cylinder the point of heat. The current uniformly flows over the surface on the heated side from the point of heat, expands into two distinct tides which sweep the surface of the metal, and re-uniting on the opposite side, recurves into itself at the heated point of the cylinder.

65. The general distribution of the electric force on the surface of the cylinder, by heating it as directed (60, 61, 62), will be pretty accurately indicated by the arrows in Fig 13. The cylinder is supposed to be divided into halves, and its convex sides upwards, as in Fig. 12, (56). The straight arrows indicate the lines of greatest energy; and the edges a b, a b, which coincide when the halves are replaced, are in one of the neutral lines: the other neutral line is ed in the centre of the figure.

66. The thermo-magnetic energies can hardly ever be traced more than four inches from the point of heat: they are, however, excited to a certain extent by the slightest disturbance of temperature near to either of the ends of the cylinder.

67, When any point is suddenly made pretty hot, without elevating the temperature of the opposite side, which can easily be done when a cylinder is employed of more than two inches diameter, the electric force is very considerable, and will deflect the needle to an angle of 20° or 30°; and by dexterously turning it the other side up before the returning needle arrives at the magnetic meridian, another impulse is given, and the angle increased on the other side. Two or three turns of the cylinder in this way will cause the needle to sweep a considerable arc; but the arc over which the needle passes will be very much increased if the needle be followed up by the active sides of the cylinder, still keeping the one parallel to the other.

Remarks on the preceding Experiment.

68. There is a peculiarity in the phenomena displayed in this experiment, which has not been observed in any of the rest:-When a cylinder of antimony is cavernous on one side, I have shown (57) that the electric current invariably flows over the same parts of the surface; but in cylinders of uniform density on every side of the axis, the law of thermo-magnetic action is very different, and the route of the electric current over the surface of the metal entirely depends upon the situation of the point of heat.

69. When a cylindical bar of antimony is uniformly dense on every side of its axis, it will invariably present a regular crystalline form at every transverse fracture. The general contour of the section is that of a series of exceedingly thin, concentric crystalline laminæ, of which the whole face of the fracture seems to be composed from the centre to the surface of the cylinder. Aided by a

magnifier, the eye is enabled to trace apparent radiating veins, which by close inspection are observed to separate the laminæ into distinct parcels or tall narrow bundles, with their edges inclined to each other at various angles, both salient and re-entering and the apparent veins, which are frequently nothing more than an angle at which two bundles of laminæ unite, give to the fracture a beautiful glittering appearance. Some of those radiant veins, however, are absolutely the flat facets of laminæ, or more frequently the sloping edges of bundles of them, which have a brilliancy far superior to any other part of the fracture. The general position of the laminæ, however, is, that their flat surfaces are presented to the axis of the cylinder; and although there are certainly objections to this position being uniformly determined by the crystalline laminæ, because of several of the piles or bundles being posited at various angles, yet, the major part of those piles are absolutely set in that position, and not a single crystalline film has its plane determined at right angles to the axis of the metal. Hence it is that the edges of the greatest part of the bundles of laminæ are presented to view at every transverse fracture, and may be compared to tall narrow bundles of thin metallic leaves, or slips of paper, placed round a central nucleus, with one of their narrow ends presented to the centre, and the other towards the surface of the cylinder; which position, together with others which some of those bundles assume, give to them the appearance of radii, with various degrees of splendour. Fig. 14 will assist in giving an idea of the general disposition of the strata of crystals in a transverse section of an uniformly dense cylinder of antimony.

70. If the sharp edge of a hammer be applied in the direction of the axis, the cylinder may be completely dissected from its surface to its centre; or the crystalline layers may be peeled off one after another with very great accuracy, as far as the dissection is required to be carried on. When a cylinder of antimony is thus disrobed, it presents an exceedingly beautiful appearance: the refulgent facets of its crystals are exposed to view, which stud its surface as if it were decked in a most brilliant coat of mail; whilst the multitude of spangles which those facets display are now seen to be disposed in the crystalline arrangement already described.

71. Assuming, then, that the general crystalline arrangement is that of concentric laminæ, two hypotheses may perhaps present themselves for an explanation of the thermo-magnetic phenomena elicited in an uniform cylindrical bar of antimony, one of which, it appears to me, will ultimately be found to be the true theory.

72. First, then, it may be supposed that the opposite faces of each metallic film are in different states of electricity, or at least that they have different thermo-magnetic qualities. If it could be satisfactorily proved that this were the case, their concentric arrangement would reconcile the phenomena to all those which are displayed by the juxta-position of any pair or series of pairs of dissimilar metallic plates, and each bundle of films would become an electric column.

In that case the thermo-magnetic character of the inner surface of each film would be to its outer surface as bismuth is to antimony; for the current in a pair of those metals flows, through the point of heat, from the former to the latter; and the rest of the circuit answers no other purpose than that of a conductor. When the point of heat is close to the edge of a transverse fracture of a cylinder of antimony, two, or a very few more of these plates or crystalline films may possibly be the only parts excited, and the rest of the bar assume the character of a conductor only; in which case the current would flow, at the point of heat, across the films from the internal to the external parts of the cylinder; the direction which experiment discovers it to proceed in: besides, it is possible that the crystalline laminæ may, individually, have different electric powers.

73. The other hypothesis supposes, that as the crystalline strata are only in juxta-position, and not very firmly united, it is possible that the heat applied at any point on the surface of the cylinder, would meet greater obstacles in its progress whilst passing from film to film, than any which it would fall in with whilst flowing over the surface of those films, or over the general surface of the metal: and as heat is well known to affect electrical phenomena generally, and as it is the exciting agent in this particular class, it may be supposed that, by its travelling at different rates in those directions, the electric powers of the metal may also be put into motion, and assume certain uniform directions, as regards the directions in which the heat flows, with the greatest and least facility.*

74. Experiments with solid cones of antimony.-When a solid cone of antimony, uniformly dense on every side of its axis, is made the subject of experiment, the surface near to its base displays thermo-magnetic phenomena of precisely the same character as those which have been described in the experiments with a cylinder (60, 61, 62).

75. The cones which I have employed were 4.5 inches high, and the diameter of the base 2.25 inches.

76. When any of these cones were heated at any point of the side near to the base, the current uniformly proceeded from the point of heat over the surface towards the apex, and returned on the opposite part of the surface to the base. This was the direction of the lines of greatest energy, but, like the cylinder, the surface of the cone becomes generally thermo-magnetic by this process, and the direc tion of its forces are easily traced by the compass needle.

77. Fig. 15 will represent the surface of a cone of antimony in a state of thermo-magnetic action: the cone is supposed to be divided into halves from its apex to its base, and in the plane of the neutral lines. The same explanation will apply to this figure as to the cylinder, fig. 13 (65).

• These hypotheses are offered merely as conjectures, without any intention of insisting on either of them, until experiment affords more data in their favour. If I mistake not, however, some of those which I have yet to describe will bear directly on the subject.

T

78. It is not necessary that the point of heat be exactly in the edge of the base to produce the greatest effect, for the direction of the electric force is still the same, and quite as energetic, when the point of heat is at some short distance from the base. Neither is it necessary that any point be made very hot, unless it can be done very suddenly; for the powers excited are decisively exhibited when the selected point of heat is held only for a few moments in the apex of the flame of the spirit lamp, and the cone immediately applied to the compass needle, before the heat has time to spread, to any great extent, over the conical surface.

79. When the apex of the cone of antimony is heated, the electric force is exceedingly feeble, and its direction quite uncertain. In general, the thermo-magnetic forces displayed by heating any point nearer to the apex than to the base are comparatively insignificant, and their directions not easily predicted.

80. A cone of antimony which had exhibited the phenomena already described, was cut in two by a saw, at about 1.5 inches from the apex, and parallel to the base. The small cone operated precisely as the original one, of which it was a part; but the energies were by no means so powerful.

81. The frustum presented phenomena as if it had been a complete cylinder, and the electric currents were as decidedly traced when the point of heat was near to the section as when it was near to the base.

82. When the cylindrical bars or cones of bismuth are experimented with in the manner I have described with antimony in those shapes, the thermo-magnetic phenomena are precisely of the same character, and are regulated by the same laws; so that whatever phenomena be displayed by the one metal will also be displayed by the other, provided the cylinders or cones be well cast, and of uniform density on every side of the axis.

83. In bismuth, however, it sometimes happens that, in consequence of an irregularity of crystallization, which it is prone to assume, there will be one point, and sometimes two, which, when heated, will display thermo-magnetic phenomena very different to those I have before spoken of; but these are irregularities which have nothing to do with the general character of the phenomena, and but seldom occur.

84. Observations.-Whatever peculiarities there may be in the crystallization of antimony and bismuth when in masses of other forms, they exhibit arrangements exceedingly similar to each other when cast into cylinders, which are regularly and uniformly cooled on every side; and there is so little difference in the general aspect of a transverse fracture of the two metals, that were it not for the dif ference of colour, it would require some practice to distinguish the one from the other. From this circumstance it appears highly probable, that the same cause, whatever it may be, is the fountain of the thermo-magnetism in both metals.

85. It has been intimated to me by some very scientific gentle

men, that impurities in the metal may possibly be the cause of all the thermo-magnetic phenomena, which I have attributed to homogeneous bodies; and I must confess that, for some time, I had entertained a similar opinion: experience and observation, however, by no means sanction the concession. Some other cause than that of impurity in the metal is unquestionably in active operation; and to some other cause we must direct our attention before we can accomplish an explanation of the phenomena in question. A very small portion of tin added to bismuth, not only dispossesses it of its magnificent crystalline ramifications, but also of the superlative display of its natural innate thermo-magnetism: moreover, that small morsel of tin, not only paralyses the the thermo-magnetism natural to bismuth as a homogeneous metal, but absolutely transfers its thermo-magnetic character as regards other metals, from one extremity of the range to the other; so that if pure bismuth be regarded as the most positive metal, its alloy with tin will be the most negative substance, either simple or compound, with which we are acquainted; and antimony, which has hitherto claimed the negative extremity of the range, is highly positive to this simple alloy.

86. The thermo-magnetism natural to antimony becomes completely stagnated when mixed with tin or lead, and the crystals of the metal become insignificant shapeless specks. Zinc also, which when in larger masses, displays its innate thermo-magnetism in a degree superior to any other metal except antimony and bismuth, becomes comparatively inert by a mixture of tin or lead.

And what perhaps may appear a more convincing fact than all the rest is, that antimony and zinc, which separately, as homogenous bodies, display fine crystalline forms, and also active thermo-magnetism, will, if mixed together as an alloy, become robbed of both those distinguished characters at once, and the resulting metal appears as compact as the finest steel.

87. Whatever may be the notions entertained as regards the mass or quantity of metal employed in heterogeneous thermo-magnetic combinations, I find that in the display of the thermo-magnetic phenomena of homogeneous bodies, the quantity employed is an essential consideration; for in several of the metals, although no trace of thermo-magnetism can be detected in small pieces, its powers are promptly developed in masses of considerable dimensions, and the laws of its phenomena may be determined with precision. Zinc, when in large masses, displays thermo-magnetic phenomena in a very exalted degree, but in small pieces hardly any trace of that power is to be found. Copper is a still more striking instance of the superior thermo-magnetic powers of large masses. Those powers could not be detected in a few ounces of the metal; but in a mass weighing sixty or seventy pounds, they would become very conspicuous. But a mass of copper of a hundred weight, however heated, would not deflect a needle half so far as it would be deflected by a single ounce of