The Bakerian Lecture.—On the Electro-dynamic Qualities of Metals

XXXI. THE BAKERIAN LECTURE.—On the Electro-dynamic Qualities of Metals*. By Professor William Thomson, M.A., F.R.S. Received February 28,—Read February 28, 1856. 1. An electrified body may be regarded as a reservoir of potential energy, and any material combination in virtue of which bodies can receive charges of electricity is a source of motive power. The development of mechanical effect from the potential energy of electricity, or through electric means from any source of motive power, may take place in a great variety of ways. For instance, electro-static attractions and repulsions may become direct moving forces (as in "Franklin's Spider"), to do work in the discharge of an electrified conductor or in the continued use of a continuous supply of electricity; or the forces of current electricity may, as in any kind of electro-magnetic engine, become working forces on bodies in motion; or the whole energy of the discharge may, as discovered by Joule, be converted into heat, which again may be transformed into other kinds of energy; or the heat evolved and absorbed by electricity, in a circuit of two different metals, at the places where it crosses the junctions from one metal to the other, being a thermal result of dynamic moment† when the junction at which heat is evolved is at a higher temperature than the junction at which heat is absorbed, may be used in a thermo-dynamic engine. Again, a thermo-electric current is a dynamic result derived from a definite absorption of heat in one locality and a definite evolution of heat in a locality of lower temperature. 2. Of these various kinds of action, all except the first mentioned, depend essen- * The author has to acknowledge much valuable assistance in the various experimental investigations described in this paper, from his assistant Mr. McFarlane, and from M. C. A. Smith, Mr. R. Davidson, Mr. F. Maclean, Mr. John Murray, and other pupils in his laboratory. † Either an evolution of heat at a temperature higher, or an absorption of heat at a temperature lower, than that of the atmosphere, may be taken advantage of to work an engine giving mechanical effect from heat; by using the atmosphere in one case as a recipient for discharged heat, in the other as the source of the heat taken in. Or an evolution of heat at any temperature and an absorption of heat at any lower temperature, may be taken advantage of for the same purpose, in a limited material system, neither taking heat from nor parting with heat to any external matter. Hence such a double thermal effect may be said to possess "dynamic moment." See the author's "Account of Carnot's Theory of the Motive Power of Heat," §§ 4 to 11, Trans. Roy. Soc. Edinb. Jan. 2, 1849; also his "Dynamical Theory of Heat," §§ 8, 13, 23 to 30, Trans. Roy. Soc. Edinb., March 17, 1851, and "Dynamical Theory of Heat, Part VI. Thermo-electric Currents," § 102, Trans. Roy. Soc. Edinb., May 1, 1854. The series of articles under the general title "Dynamical Theory of Heat," have been republished in a succession of Numbers of the Philosophical Magazine, viz. §§ 1 to 80, Vol. July to Dec. 1852; §§ 81 to 96, Vol. Jan. to June, 1855; §§ 97 to 181, Vol. Jan. to June, 1856. tially on certain definite properties of matter in regard to which different metals have remarkably different qualities. Thus in electro-magnetic engines the electric conductivity of the coils through which the current passes, and the magnetic inductive capacity and retentiveness of the iron cores of the electro-magnets, are essentially involved; and as essentially, when permanent magnets are used, the magnetic properties of steel, loadstone, or other bodies possessing strong retentiveness for magnetism. In the simple conversion of any kind of energy into heat by means of electric currents in metals, their electric conductivities are essentially and solely concerned. The inverse thermo-electric transformation of energy into an evolution and absorption of heat, at localities of different temperature, in quantities differing from one another by the thermal equivalent of the work spent in maintaining the current*, depends essentially on certain distinct properties of metals in regard to which their various qualities are shown by the differences of their positions in the thermo-electric series at different temperatures; and the accessory circumstances of such operations are influenced by the electric and thermal conductivities of the metals used. The same properties are involved in the direct thermo-electric transformation of energy in which electric currents, sustained by the communication of heat in a hot locality and the abstraction of a less quantity of heat in a locality lower in temperature, either produce any mechanical action, or are allowed to waste all their motive power in the frictional generation of heat†. * See "Dynamical Theory of Heat, Part VI. Thermo-electric Currents," §§ 105, 110. † " . . . a current cannot pass through a homogeneous conductor without generating heat in overcoming resistance. This effect, which we shall call the frictional generation of heat, has been discovered by Joule to be produced at a rate proportional to the square of the strength of the current; and, taking place equally with the current in one direction or the contrary, is obviously of an irreversible kind" (Dyn. Th. Heat, § 104). This definition was given merely to render circumlocution unnecessary in frequently referring to a mode of electric action which bore an obvious analogy to the action of a common fluid generating heat by friction among its particles as a dynamical equivalent to work spent upon it from without in forcing it to circulate in a tube, or otherwise keeping it in motion. It appears to me highly probable, however, that what I have, with reference only to recognized electric currents, defined as the frictional generation of heat, is precisely the mode of action by which all the heat is generated in every case when two solids are rubbed together. Certainly when two bad conductors of electricity are rubbed together, a portion of the heat of friction is generated in visible electric flashes; and a charged Leyden battery contains, in potential energy, a dynamic equivalent for a portion of the heat of friction between rubber and glass never made till the battery is discharged. As certainly a portion of the heat of friction between a metal and a bad conductor of electricity is invisibly generated by electric currents through a very minute depth of the metallic substance beside its rubbed surface. The first effect of chemical forces of affinity, as Joule has so powerfully demonstrated in a variety of cases, is to press electricity into motion; which motion may either subside into heat close to the locality of the combination (as when rough zinc is dropped into dilute sulphuric acid), or, reactively resisting the chemical combination, may transmit the work to a locality distant from the source, and may there either generate heat in a permanent metallic or other undecomposable conductor, or may, without any generation of heat at all, be wholly spent in effecting decompositions against chemical affinities infinitely little less powerful than those from which it proceeds, or in raising weights. So it appears highly probable that the first effect of the force by which one solid is made to slide upon 3. All properties, then, of electric and thermal conductivity, of magnetic inductive capacity and retentiveness, and of thermo-electric rank and its variations from one temperature to another, may be characterized as electro-dynamic; and the degrees to which these properties are possessed by different substances may be called their electro-dynamic qualities. Again, the variation which absolute magnetic inductive capacity, and magnecrystallic axial differences, experience with change of temperature may obviously be made the means of a transformation of heat into common mechanical energy, and we have thus a set of magneto-dynamic properties of matter which may almost in the present state of science be regarded as intrinsically electric, but which at all events (when we consider that the motions contemplated, taking place as they do under magnetic force, cannot but be accompanied by electric currents) may be fairly classed under the general designation of electro-dynamic. The variations of intrinsic magnetism, of magnetic inductive capacity, and of magnecrystallic properties, produced by variations of temperature, are therefore included among the electro-dynamic qualities of metals which I propose to investigate, although I have as yet made no progress in this branch of the subject. PART I. ON THE ELECTRIC CONVECTION OF HEAT, §§ 4 to 77. §§ 4 to 18. Theoretical Indications. §§ 4 to 9. Origin of the Investigation. 4. In first attempting an application of the principles of the Dynamical Theory of Heat to show the mechanical relation between cause and effect in thermo-electric currents, I supposed the effects thermal and mechanical that can be produced by a thermo-electric current in any part of its circuit to be, as first suggested by Joule, due to the heat absorbed, according to Peltier's discovery, at the hot junction in virtue of the current crossing it, and I pointed out that the current crossing the cold junction must evolve a quantity of heat which, were this supposition true, would be less than that absorbed at the hot junction, by an amount precisely equivalent to all the effects, produced by the current in the rest of the circuit*. 5. Introducing Carnot's principle, as modified in the Dynamical Theory of another is electricity set into a state of motion; that this electric motion subsides wholly into heat in most cases, either close to its origin and instantaneously, as when the solids are both of metal, or at sensible distances from the actual locality of friction and during appreciable intervals of time, as when the substance of one or both the bodies is of low conducting power for electricity; and that it only fails to produce the full equivalent in heat for the work spent in overcoming the friction, when the electric currents are partially diverted from closed circuits in the two bodies and in the space between them, and are conducted away to produce other effects in other localities. Still, no hypothesis need be implied by using the expression "the frictional generation of heat by an electric current," as defined in the passage quoted, and it is introduced into the present paper with no other justification than its convenience. * Dynamical Theory of Heat, March 17, 1851, § 17. Heat*, I found a relation between the quantities of heat absorbed or evolved by currents crossing metallic junctions at different temperatures; which led immediately to a general expression for the electrical condition of a circuit of two metals with their junctions kept at any stated temperatures. 6. From this it appeared that the electro-motive force should follow the same law of variation in every case, being expressed by a constant, (representing the thermo-electric difference between the two metals,) multiplied into an absolute function of the temperatures of their junctions, namely, the difference of their temperature on the absolute thermometric scale since proposed by Mr. Joule and myself, and demonstrated by our experiments† to agree very approximately with their difference of temperature as indicated by an air-thermometer. Finding this conclusion contradictory to the statements made by experimenters, that the electro-motive force does not vary with the temperature of the junctions according to the same law in circuits composed of different metals, I perceived that Peltier's discovery did not afford a sufficient explanation of the source whence a thermo-electric current derives its energy, but that electric currents must possess the previously undiscovered property of producing different thermal effects in passing from cold to hot and from hot to cold in the same metal, and must possess this property to different amounts in different metals. 7. Taking this new property of electric currents into account along with that discovered by Peltier, and introducing an application of Carnot's principle, I arrived at expressions for the relations between the heat absorbed and evolved in various parts of a circuit of any different metals, and between the electro-motive force and the temperatures of the junctions, which appear to be in complete accordance with the facts. These investigations were communicated in December 1851, to the Royal Society of Edinburgh‡. 8. Still simpler theoretical considerations (§§ 10 to 18 below) regarding the source of energy drawn upon in a thermo-electric current, make it certain that the phenomena of inversion discovered by Cumming could not exist, unless the metals presenting them had the property of experiencing, when unequally heated, unequal thermal effects from electric currents passing through them from hot to cold, and from cold to hot. Having satisfied myself, both by an examination of the evidence afforded by Becquerel's experiments (the original investigation on the subject by Cumming being at that time unknown to me), and by actual observation, in experiments of my own §, * This, the true form of Carnot's principle, was first published by Clausius in May 1850 (Poggendorff's 'Annalen'). It had occurred to myself, and I had used it in discovering the true expression for the duty of a perfect thermo-dynamic engine shortly before that time. It was not, however, until the beginning of the year 1851 that I thought on a demonstration which would probably be admitted as conclusive in establishing the principle, and my investigation on the subject was only communicated in March 1851 to the Royal Society of Edinburgh. See Trans. Roy. Soc. Ed. of that date, "Dynamical Theory of Heat," § 14. † "On the Thermal Effects of Fluids in Motion," Transactions, June 1854. ‡ See Proceedings of that date, and Philosophical Magazine, June 1852. § See below, Part II, §§ 79, 80, 81, 83, 84, &c. that the doubts which various writers had thrown on the existence of thermo-electric inversions were groundless, I concluded with certainty that the newly conceived thermal effect of electricity in unequally heated metals really exists. But the theory left it undecided what the absolute nature and amount of this effect may be, and only showed how, by observations on thermo-electric currents, its difference in different metals may be determined. 9. I therefore had recourse to direct experiment on the thermal effects of electric currents in unequally heated conductors, not to demonstrate the existence of the peculiar effect anticipated, but to ascertain its nature, with moreover a view of ultimately determining its absolute amount, in some particular metal or metals. Before proceeding to describe experiments, by which I have now discovered the quality of the new effect in several cases, I shall, without entering on the mathematical details of the theory, or the full application of Carnot's principle, repeat in a few words so much of my first communication on the subject to the Royal Society of Edinburgh, as to show the reasoning, founded on incontrovertible mechanical principles, which made me commence the experimental research with the certainty that the property looked for existed, whether I could find it or not. §§ 10 to 15. General inferences regarding the Electric Convection of Heat from Dynamical Principles. 10. Cumming has discovered that in many cases when one of the junctions of a thermo-electric circuit of two metals is kept at a fixed temperature, if that of the other be elevated gradually from equality, an electro-motive force is produced, which first increases to a maximum, then diminishes, vanishes for a certain temperature of the junction, and acts in the contrary direction with gradually increasing strength as the temperature is further raised. It is clear that, at exactly that temperature of the hot junction for which in any such case the electro-motive force is a maximum, the two metals must be thermo-electrically neutral to one another, and must present reverse thermo-electric relations for temperatures below and above this point. Hence the thermal effect depending on the direction of a current crossing the junction of two such metals must be for temperatures above, the reverse of what it is for temperatures below, the neutral point, and must vanish when the metals are exactly at this temperature. 11. For although Peltier himself supposed the effect he had discovered to depend on the conducting powers of the two metals for heat, and remarked as an anomaly the case of bismuth and copper, for which his supposition was violated, his own experiments show the truth to be, that in a circuit of two metals an absorption of heat at the junction where the temperature is higher, and an evolution of heat at the other, must be produced by the thermo-electric current which is caused by the maintaining of the difference of temperature between the junctions. That this is universally true when the temperatures of the two junctions are on the same side of the neutral point, cannot, in the present state of science as regards the theory of heat, be reasonably doubted. 12. If, therefore, a circuit of two metals have one junction kept at the neutral point, and the other at some lower temperature, the current excited will cause the evolution of heat at the cold junction, but neither absorption nor evolution of heat at the hot junction; and in the rest of the circuit there will be effects either purely thermal, or thermal and mechanical or chemical, according to the nature of the resistance against which the electro-motive force is allowed to work. The source from which the electro-motive force derives its energy to produce these effects cannot be at the hot junction (§ 10), where heat is neither absorbed nor evolved, nor at the cold junction (§ 11), where heat is evolved, nor of course in any uniformly heated part in either metal, through all of which, provided the metal has no thermo-electric crystalline characteristic, there can be nothing but a frictional evolution of heat; that is, it is nowhere but in those portions of the circuit where the temperature varies between that of the cold and that of the hot junction. In those portions, therefore, there must be as much heat absorbed, in virtue of the current, as is equivalent to the aggregate mechanical value of the heat evolved at the cold junction, and all the effects, thermal, mechanical, and chemical, produced in the rest of the circuit. 13. If, for example, an electro-magnetic engine be introduced into the circuit, and be allowed to work at such a rate as to reduce, by its inductive reaction, the strength of the thermo-electric current to an infinitely small fraction of what it is when the engine is at rest, the heat absorbed in virtue of the current in the unequally heated parts of the two metals will be equal to the heat evolved at the cold junction, together with the thermal equivalent of the work done by the engine, and will be simply proportional to the strength of the current. On the other hand, if the engine be forced to work a little faster, so as to overbalance by an infinitely small amount the thermal electro-motive force, and cause a reverse current in the circuit, there must be heat evolved in virtue of this current in the unequally heated parts of the two metals to an amount equal to the heat absorbed at the cold junction, together with the thermal equivalent to the work done against electro-magnetic forces in the engine. It follows that in the unequally heated portions of the two metals, the current passing from cold to hot in one, and from hot to cold in the other, must produce a thermal effect, in simple proportion to its own strength, constituting on the whole an absorption of heat when the thermal electro-motive force is allowed to produce a current, and an evolution of heat when a current is forced by other means in the contrary direction. 14. Hence, for any two metals which are thermo-electrically neutral to one another at a certain temperature, and which possess reverse thermo-electric properties for temperatures above and below the neutral point, we conclude the following propositions:— (1) In one or other of the metals (and most probably in both) there must be a thermal effect due to the passage of electricity through a non-uniformly heated portion of it, which must be an absorption of heat or an evolution of heat, according to the direction of the current between the hot and cold parts, and proportional in amount to the whole quantity of electricity that passes in a stated time. (2) The amount of this effect, with the same strength of current and the same difference of temperatures, must differ in the two metals to such an extent, that the effect of a current in passing from cold to hot in one metal, together with the effect of an equal current passing from a place equally hot to a place equally cold in the other, may amount to the absorption or evolution, the existence of which has been demonstrated. 15. The reversible thermal effect* of electric currents in single metals of non-uniform temperature, which has been thus established, may obviously be called a Convection of Heat by electricity in motion. To avoid circumlocution, I shall express it that the Vitreous Electricity carries heat with it, when this convection is in the "nominal direction of the current." On the other hand, when the convection is against the "nominal direction of the current," it will be said that the Resinous Electricity carries heat with it. §§ 16 to 18. Dynamical Theory applied to draw, from thermo-electric data, inferences regarding the Electric Convection of Heat in Copper and in Iron. 16. The application of the preceding theorem to the particular case of copper and iron is a consequence of Cumming's discovery, that, if one junction in a circuit of two arcs of those two metals be kept cold, and the other be heated gradually, a current at first sets from copper to iron through the hot junction with increasing strength; but begins to diminish after a certain temperature, which Becquerel found to be about 300° Cent., is exceeded; falls away to nothing when a red heat is attained; and sets in the reverse† direction when the elevation of temperature is pushed higher. * See an article by the author, entitled "On a Universal Tendency in Nature to the Dissipation of Mechanical Energy" (Proceedings Roy. Soc. Edinb., Feb. 16, 1852, and Phil. Mag., Oct. 1852), where all natural operations are divided into two great classes, "reversible" and "irreversible." See foot-note on § 2 above, for an example of the second class. † Having myself experienced some difficulty in obtaining the reverse current in the manner described by M. Becquerel, in which one junction was heated in the flame of a spirit-lamp, while the other was kept at the atmospheric temperature, I found that it could be obtained so as to be observed with the greatest ease by means of a very ordinary galvanometer and an iron wire with copper wires twisted round its ends, by keeping the lower junction at a temperature considerably above that of the atmosphere, at 100° Cent. for instance; and I ascertained that when both junctions were kept at a very high temperature, in the flame of a spirit-lamp for instance, and one of them cooled a little below the temperature of the other, the current produced was the reverse of that which the same difference occasioned when both junctions were at ordinary temperatures. See Part II. below for further developments on this subject. Some observations of Regnault's having appeared to indicate $240^\circ$ Cent. as, more nearly than $300^\circ$, the temperature of the hot junction which gives the current its maximum strength, I concluded the following proposition: 17. "When a thermo-electric current passes through a piece of iron from one end kept at about $240^\circ$ Cent.*, to the other end kept cold, in a circuit of which the remainder is copper, including a long resistance wire of uniform temperature throughout, or an electro-magnetic engine raising weights, there is heat evolved at the cold junction of the copper and iron, and (no heat being either absorbed or evolved at the hot junction) there must be a quantity of heat absorbed on the whole in the rest of the circuit. When there is no engine raising weights, in the circuit, the sum of the quantities evolved, at the cold junction, and generated in the 'resistance wire,' is equal to the quantity absorbed on the whole in the other parts of the circuit. When there is an engine in the circuit, the sum of the heat evolved at the cold junction and the thermal equivalent of the weights raised, is equal to the quantity of heat absorbed on the whole in all the circuit, except the cold junction†." 18. Hence, if the reversible part of the effect of a current from hot to cold in iron is an evolution of heat, the corresponding effect in copper must be a greater evolution of heat. But if, on the other hand, a cooling effect be produced by a current from hot to cold in iron, there must be either a less effect of the same kind, or a reverse effect, in copper. It is left to experiment to determine which of the two hypotheses is true regarding iron; and should it turn out to be the latter, to ascertain which of the two remaining alternatives regarding copper must be concluded. With this object I commenced the experimental researches which I now proceed to describe. §§ 19 to 77. Experimental Investigation of the Electric Convection of Heat in Copper, in Iron, and in some other Metals. §§ 19, 20. Unsuccessful attempts, and first result. 19. I began, more than four years ago, by observing carefully the ignition produced in short wires of copper, iron, and platinum by electric currents alternately in the two directions, thinking that some of the effects described by various experimenters, as showing a superior heating power in the positive electrode, might possibly be dependent on the convective agency which I was endeavouring to discover. But I never observed the slightest variation in the position of the incandescent part of the * I have since ascertained (see Part II. below), by keeping the ends of an iron wire, with copper wires from the galvanometer soldered to them, in separate vessels of hot oil, and determining different temperatures of the two which give no current, that the neutral point for the particular specimens of iron and copper which I used must be about $284^\circ$ Cent. I should therefore, at present, substitute $284^\circ$ for $240^\circ$ in the proposition quoted in the text; without further research, however, it is impossible to pronounce upon the limits between which the neutral point of various specimens of copper and iron wires may be found to lie. † Proceedings of the Royal Society of Edinburgh, Dec. 15, 1851, republished in Phil. Mag., June 1852. wire, with a sudden reversal of the current. Sometimes the incandescence was assisted by a spirit-lamp flame applied to the middle part of the wire, and the ends were kept cool by wet threads. Sometimes in a long wire with a current through it not quite strong enough to keep it at a red heat, a small part was made incandescent by a slight application of heat as nearly as possible at one point, by a spirit-lamp flame. Still there was never observed the slightest motion of the incandescent part, when the current was suddenly reversed, and I concluded that whatever had been observed in the way of different heating effects of the positive and negative electrodes, must have been owing to peculiar agencies of the current in passing between metal and rarefied air, or to some other cause than thermal convection in metals; and I saw that more powerful tests would be required to bring out the result I looked for. 20. I next made experiments on a conductor of bar iron bent into two equal upward vertical branches on each side of the horizontal part, which was kept immersed in a vessel of hot oil, while the upper ends of the vertical branches were kept cool by streams of cold water. Vessels of water were applied round the two vertical branches, as calorimetric arrangements to test heat evolved or absorbed in them by the agency of a current sent down one and up the other from a nitric acid battery of sixteen small iron cells, arranged as a single element. The current was sent first for half an hour in one direction, then half an hour in the contrary direction; and so on, with a reversal every half-hour. The water round the two vertical branches was kept constantly stirred, and thermometers in fixed positions in them were observed at frequent intervals during the experiments, which were each continued for about two hours. A comparison of all the readings taken showed a rather higher mean temperature in the branch down which the current was passing than in the other; indicating, differentially, a cooling effect in the branch through which the current passes from the hot middle, and a heating effect in the other. This experiment appeared to show that "the resinous electricity" carries heat with it in an iron conductor; but the irregular variations of temperature in each thermometer were so much greater than the differential effect deduced, that I could not consider the conclusion satisfactorily established. §§ 21 to 29. Unsuccessful attempts with large bar conductors. 21. There were difficulties connected with the arrangements of the calorimetric vessel, which made me judge that it would be better, instead of testing the average temperature of two portions of the conductor, each extending the whole way from the hot middle to the cold ends, to simply test the temperature of as nearly as possible one point midway between the hot and cold on each side; and it appeared that the heating could be more easily applied and better regulated by a source of heat at the middle of a straight horizontal conductor, than by the plan I had followed in the arrangement just described. I therefore got bars of copper and iron, with holes to admit the bulbs of sensitive thermometers, made to the following dimensions:— Copper conductor. Iron conductor. | Whole length | 16 | 24 | |--------------|----|----| | Breadth | 1 | 2 | | Depth | 2\(\frac{1}{3}\) | 3 | | Depth of hollows | 2\(\frac{1}{10}\) | 2\(\frac{1}{2}\) | | Diameter of hollows | \(\frac{1}{3}\) | \(\frac{1}{5}\) | These relative dimensions were chosen so that the conducting powers of the two bars for electric currents, and consequently for heat also, might be not very unequal. A vessel of tin-plate, perforated to admit the bar through its sides, was soldered round the middle of each conductor, and two others so as to leave about 2 inches at the ends of the conductor projecting beyond them. The parts of the conductors within these vessels were about 3 inches long in the copper and 4\(\frac{1}{2}\) inches in the iron, and the parts between the middle vessel and the vessels at the two sides were 2\(\frac{1}{2}\) and 2 inches respectively. The bores for the thermometer bulbs were exactly in the middle of the last-mentioned parts. In experimenting on either conductor, the central vessel was generally filled with oil or water, and kept hot by a gas-lamp below it. Streams of cold water from the town supply-pipes were kept flowing through the two lateral vessels. 22. To make these streams constant, whatever variations of pressure might occur in the supply-pipes, a cistern in a fixed position above the conductor was kept full (overflowing), and the coolers were supplied by pipes from this cistern. The supply often failed for several minutes, and sometimes for much longer; and after an experiment (Nov. 19, 1853) was nearly lost from this cause, a plan was arranged to lift water up from a larger cistern (into which the exit-streams from the coolers were discharged), and to pour it into the smaller cistern above, so as to keep the stream constant in quantity (although not quite invariable in temperature) even when the proper supply failed. 23. A galvanic battery for exciting a current through these conductors was prepared, consisting at first of four, and ultimately of eight, large iron cells, each measuring internally 12 inches deep, 10\(\frac{1}{2}\) inches broad, and 2\(\frac{1}{2}\) inches from side to side; eight porous cells, each 12 inches deep, 10 inches broad, and 2 inches from side to side; and eight zinc plates, each 9\(\frac{1}{2}\) inches by 10 inches. The iron cells were charged with a mixture of nitric acid two parts (bulk), sulphuric acid three parts, and water two. The porous cells were charged with dilute sulphuric acid. In each of the cells there were 1\(\frac{1}{3}\) square feet of zinc surface exposed to 2\(\frac{1}{2}\) square feet of iron, and the electro-motive force was not far from double that of a single cell of Daniell's. 24. After preliminary experiments in which, with oil in the central vessel kept hot by a gas-lamp, the temperatures were too unsteady to allow any results of value to be obtained, water was substituted for oil in the central vessel, and was kept boiling briskly by the gas, the place of the water evaporated being frequently supplied by small quantities of boiling water poured in, so that ebullition never ceased. The irregularities having been found to be much diminished, experiments were made in the following manner. 25. Four of the large iron cells, arranged as a single galvanic element, were used to excite the current. The experiment lasted about two hours, during which the current was sent through the conductor for twenty minutes at a time, alternately in the two directions, twice in each direction. Several minutes were spent in changing the direction of the current; the stiffness of the electrodes, and the clamps used for the connexions which had to be changed, rendering the process very troublesome. Readings of the thermometers were taken at intervals of five minutes during the flow of the current in both directions, as well as for some time before the current commenced and after it ceased. 26. The results of this experiment manifested, among great irregularities in the indications of the thermometers, a very decided differential variation between the two every time the direction of the current was changed; and appeared so promising, that a series of further experiments on the same copper conductor, and on the iron conductor similarly arranged, were immediately commenced, for the purpose of testing decisively the conclusion which had been indicated, and for discovering the corresponding effect in iron. To avoid the loss of time and the derangement in the position of the conductor by the shifting of the heavy clamps and stiff electrodes between its ends, in changing the direction of the current, a commutator, by which the change could be effected nearly instantaneously, was constructed on the following plan. 27. Four square holes, each of 1 inch, in a square block of mahogany, were fitted Fig. 1. with bottoms of thick copper slabs, passing through the mahogany, and cemented with red lead so as to hold mercury, which was poured into each hole. The copper slabs projected outside to distances of about an inch, and each bore a bundle of 100 No. 18 copper wires soldered to it, two of which, connected with diagonally opposite copper slabs, served as battery electrodes, while the other two were clamped to the ends of the conductor to be tested. The four slabs have only to be connected by two conducting arcs parallel to one pair or to the other of the sides of the square, to send the current one way or the other through the conductor; which was done by means of two heavy brass castings, as shown in the diagram (fig. 1). This commutator has been used in a considerable variety of experiments, and has been found very convenient. It gives the means of reversing almost instantaneously a very powerful current, without the necessity of bending any of the electrodes or deranging any part of the apparatus, and the conductors involved in it are so strong that it occasions very little resistance. 28. As the supposed differential effect had appeared not to be increased after the first five minutes of the flow of the current in either direction, shorter periods of various lengths were tried, and more frequent observations of the thermometers were made, for the purpose of discovering the gradual variation of the temperature in the conductor, towards its final distribution as affected by the current. Four more large iron cells were added to the battery, which made it consist in all of eight cells, arranged as a single galvanic element, exposing 20 square feet of iron to $10\frac{1}{2}$ square feet of zinc surface. As the strength of current thus produced would be nearly double of that given in the previous experiment, any true effect of the kind sought would be augmented in the same ratio; and might be expected, both on this account and because of the improved system of observation, to become much more decided. These expectations, however, were not borne out by the results. The irregularities certainly became much diminished, but with these the differential effect on the thermometers, following the reversals of the current, either quite disappeared, or became very much less considerable than that which had been observed in the first experiment, and which I afterwards was led to attribute to some derangement in the position of the conductor occasioned by shifting the heavy clamps and stiff electrodes from between its two ends, causing the thermometer bulbs to alter a little in their positions in the hollows. 29. Many experiments, both on the copper and iron conductor, were made, from October 1852 to March 1853, and the results of the observations (on each of the two principal thermometers either every half-minute, or every quarter-minute, during an experiment of about two hours) carefully reduced; with much labour at first when arbitrary scale thermometers were employed, but afterwards with far greater ease when centigrade thermometers, constructed for this investigation at the Kew Establishment, were received and brought into use. In the months of September and October 1853 the investigation was taken up again. The thermometric observations which had been made in the previously com- pleted experiments, were all reduced, on the plan of the Tables given (§§ 47 and 56) below for subsequent experiments, and, when thus tested, they appeared to contain some indications of the effect looked for. Several more sets of observations of the same kind were therefore executed, but with various modifications of details. Still no decided result could be obtained, and I concluded from all the experiments which had been made, that the anticipated effect must be too small to be discovered without either increasing the sensibility of the test or diminishing the irregularities. I therefore prepared new apparatus, by which the former, and as much as possible of the latter object, would be attained. §§ 30 to 34. Improvements and Modifications of Apparatus. 30. Instead of increasing the power of the battery, which I reserved as a later resource, if necessary, or of increasing the length of the conductor between the heater and the coolers on each side, which, while it would increase in the same ratio the amount of the effect looked for, would increase in a duplicate ratio the time that would have to be given to allow it to reach a stated proportion of its limiting value, I had conductors made of about the same length as the others, but of considerably less section. 31. With a view to perfecting and testing the action of the heater and coolers, each conductor was made up of a number of slips of flat sheet metal, bent and placed together, as shown in the accompanying diagram (fig. 2). The slips were held firmly together by a vice, while collars of sheet copper, separated from them by vulcanized india-rubber, were soldered round them in the places for the sides of the heater and coolers. Tin-plate vessels, as shown in the diagram, were then put together, and soldered to these collars. The interstices between the slips and the india-rubber, and the metal collars round the india-rubber, were stopped with red lead, and after some trouble were made water-tight. Thus the heater and coolers, without any metallic communication with the conductor, served the purpose of keeping the required supplies of hot and cold water round it in the proper places. The spaces for the thermometers were firmly stopped below with corks fitted to support the lower ends of the bulbs in perfectly fixed positions (fig. 3). Little collars of cork were put round the tubes just above the bulbs, and pushed down into the upper end of the hollow so as to hold the thermometers firmly and prevent all motion of their bulbs. 32. Various methods of heating the central part of the conductor were tried. First, as in previous experiments, water in the central vessel was kept boiling either by a gas-lamp under it, or by steam blown into it from a separate boiler; then a complex system with a boiling fountain, by which I attempted to get a perfectly uniform stream of water at a constant temperature, as little short of boiling as possible, to flow through the open spaces between the different slips within the central vessel, was used during several experiments. Lastly, water filling the central vessel was kept at a very constant temperature, near the boiling-point, by a gas-lamp below it, regulated by a person watching the indications of a thermometer with its bulb fixed in the middle space between the slips, as nearly as possible in the centre of the compound conductor. This last I found to be by far the best plan, and I used it in all subsequent experiments in which any external application of heat to the conductor was required. 33. Each cooler was divided into four compartments by partitions of tin-plate, stopping all communication from one to another, except through the spaces between the different slips composing the conductor. A constant stream of cold water (§ 22), introduced by the compartment nearest to the middle of the conductor, and drawn off by an overflow pipe, from a compartment next the end, was thus forced to flow all through among the different slips, and, as I found by placing thermometers in various positions in each compartment, gave a very satisfactory effect in fixing the temperature of the whole section of the conductor. 34. The experiments were made in other respects exactly as described above (§§ 28 and 29); the electric current, however, not being often again kept up for a longer time than ninety-six minutes, since the fumes, which always began to rise from the battery after the current had been flowing for about an hour, began after half an hour more to occasion great irregularities and inconvenience by causing the liquid (which sometimes became very hot,) to foam and overflow in some of the iron cells. The atmosphere had been in previous experiments sometimes rendered intolerable for the observers, by the acid vapour; but this evil was done away by covering the battery with cloths kept moist with ammonia and water, and by moistening other surfaces in the neighbourhood in the same way, so that the fumes never got far without meeting vapour of ammonia and combining into white clouds, which were perfectly innocuous. §§ 35 to 38. First Experiments with Multiple Sheet Copper Conductor. 35. The copper conductor on the new plan was first used in an experiment on the 28th of October, 1853; with the central vessel heated by steam, and a current from the eight large iron cells kept flowing for seventy-two minutes, alternately in contrary directions, six times six minutes each way. The thermometers were noted every half-minute. The observations thus recorded, when thoroughly examined, indicated a slight differential cooling effect in the part of the conductor in which the nominal current was from cold to hot, and a heating effect where it passed from hot to cold; that is to say, a convection of heat in the nominal direction of the current, or as I shall call it to avoid circumlocution, *a convection of heat by vitreous electricity*. 36. A second experiment with the same conductor was made on the 2nd of November, 1853, in which the current was kept flowing for ninety-six minutes, eight times six minutes each way, and the thermometers were noted every quarter-minute. An examination of the recorded results indicated still the same kind of effect, but to a much smaller extent. Thus the final average, for the alteration of difference, between the temperatures at A and B due to the flow of the current for six minutes in one direction, after it had been flowing for six minutes in the contrary direction, amounted to '039° Cent. in the first experiment, and to only '0143° in the second experiment. A full analysis of the progress of the differential variation of temperature during the flow of the current is given in Tables I. and II., § 56 below, and shows through what fluctuations the final alterations are reached. The temperatures, at the ends of the successive times of flow in one direction or the other, and the evaluation of the mean final effect, are shown, for each experiment, in the following abridged Tables. 37. The observations made during the first period (that is the time from starting till the second reversal of the current) are rejected from the average in every case of experiments on the new conductors, because they were found to show so great absolute elevations of temperature (due to the frictional generation of heat by the current) that no alteration of difference between the thermometer observed during them could be relied on as an effect depending on the direction of the current. 38. Conductor composed of thirteen slips of sheet copper. **Experiment I. October 28th, 1853.** | Periods | \( T_A - T_B = D \) | \( T_A - T_B = D' \) | Augmentations of differences from middles to ends of periods. | |---------|---------------------|---------------------|-------------------------------------------------------------| | I. | 2°18 | 2°09 | -°09 | | II. | 2°07 | 2°10 | -°03 | | III. | 2°15 | 2°11 | -°04 | | IV. | 2°16 | 2°13 | -°03 | | V. | 2°01 | 1°99 | -°02 | | VI. | 1°80 | 2°02 | °22 | Means for five periods... 2°038 2°070 -°032 Augmentation of difference during periods included -°07 Deduct average augmentation per half-period -°007 Effect due to reversal of current 0°039, in favour of Vitreous Electricity. **Experiment II. November 2nd, 1853.** | Periods | \( T_A \) | \( T_B \) | \( T_A - T_B = D \) | \( T_A \) | \( T_B \) | \( T_A - T_B = D' \) | Augmentations of differences from middles to ends of periods. | |---------|-----------|-----------|---------------------|-----------|-----------|---------------------|-------------------------------------------------------------| | I. | 53°79 | 52°43 | 1°36 | 54°00 | 52°70 | 1°30 | -°06 | | II. | 53°68 | 52°42 | 1°26 | 53°51 | 52°23 | 1°28 | °02 | | III. | 53°86 | 52°57 | 1°29 | 53°79 | 52°63 | 1°16 | -°13 | | IV. | 53°90 | 52°60 | 1°30 | 53°99 | 52°63 | 1°36 | °06 | | V. | 54°09 | 52°72 | 1°37 | 54°05 | 52°68 | 1°37 | °00 | | VI. | 55°10 | 54°00 | 1°10 | 54°15 | 52°86 | 1°29 | °19 | | VII. | 53°99 | 52°80 | 1°19 | 54°10 | 52°80 | 1°30 | °11 | | VIII. | 54°00 | 52°50 | 1°50 | 54°00 | 52°60 | 1°40 | -°10 | Means for seven periods... 54°0886 52°8014 1°2871 53°9414 52°6329 1°3086 -°02143 Augmentation of difference during periods included °10 Deduct average augmentation per half-period -°00714 Effect due to reversal of current 0°01429, in favour of Vitreous Electricity. §§ 39 to 43. Decisive Experiments with Multiple Sheet Iron Conductor. 39. These experiments seemed therefore on the whole to establish a probability in favour of the convection of heat by the so-called positive electricity, when a current is kept up through an unequally heated conductor. The convective effect, if of this kind, ought (§ 18) to be less in iron than in copper, I therefore had little expectation of finding an indication of it in the iron conductor which (§ 31) had been in the course of preparation; but as soon as it was ready for use I made the following experiments, and was much surprised by the result, which became manifest before the first of them was finished. 40. Conductor composed of thirty slips of sheet iron. Experiment III. November 12th, 1853. | Periods | \( T_A \) | \( T_B \) | \( T_B - T_A = D \) | \( T_A' \) | \( T_B' \) | \( T_B' - T_A' = D' \) | Augmentations of differences from middles to ends of periods. | |---------|----------|----------|------------------|----------|----------|------------------|--------------------------------------------------| | I. | 51°43 | 53°56 | 2°13 | 51°48 | 53°49 | 2°01 | -°12 | | II. | 51°62 | 53°30 | 1°68 | 51°41 | 53°21 | 1°80 | °12 | | III. | 51°73 | 53°26 | 1°53 | 52°03 | 53°87 | 1°84 | °31 | | IV. | 52°01 | 53°80 | 1°79 | 51°32 | 53°42 | 2°10 | °31 | | V. | 51°30 | 53°00 | 1°70 | 51°00 | 52°95 | 1°95 | °25 | | VI. | 51°14 | 52°98 | 1°84 | 50°69 | 52°80 | 2°11 | °27 | | Means for five periods. | 51°56 | 53°268 | 1°708 | 51°29 | 53°25 | 1°96 | -°252 | Augmentation of difference during periods included ... °10 Deduct average augmentation per half-period ......................... -°010 Effect due to reversal of current ........................................... 0°242, in favour of Resinous Electricity. Experiment IV. November 19th, 1853. | Periods | \( T_A \) | \( T_B \) | \( T_B - T_A = D \) | \( T_A \) | \( T_B \) | \( T_B - T_A \) | Augmentations of differences from middles to ends of periods. | |---------|----------|----------|------------------|----------|----------|----------------|---------------------------------------------------------------| | I. | 57°50 | 59°30 | 1°80 | 58°02 | 59°84 | 1°82 | 0°02 | | II.* | 48°20 | 51°15 | 2°95 | 46°82 | 49°79 | 2°97 | 0°02 | | III. | 46°49 | 49°13 | 2°64 | 47°01 | 49°95 | 2°94 | 0°30 | | IV. | 48°41 | 51°69 | 3°28 | 48°31 | 51°99 | 3°68 | 0°40 | | V. | 48°36 | 51°74 | 3°38 | 48°18 | 51°80 | 3°62 | 0°24 | | VI. | 48°00 | 51°20 | 3°20 | 48°00 | 51°49 | 3°49 | 0°29 | | VII. | 48°31 | 51°51 | 3°20 | 48°06 | 51°60 | 3°54 | 0°34 | Means for five periods... 49°3243 52°2457 3°14 49°20000 52°35143 3°454 0°314 Augmentation of difference during periods included... 0°57 Deduct average augmentation per half-period.......................... 0°57 Effect due to reversal of current ........................................ 0°257, in favour of Resinous Electricity. 41. A full analysis of the differential variations throughout each of these experiments, derived from observations of the thermometer taken every quarter of a minute, was made in each case immediately after the conclusion of the experiment (see Tables I. and II. § 47 below), and was sufficient to convince me that the true effect in the iron conductor is of the kind indicated by the preceding summary of the effects apparent at the ends of the periods. 42. To try whether or not the very considerable effect thus discovered depended on some inequality in the conductor itself, I made an experiment on the 25th of November 1853 exactly like the two preceding, with the exception that the middle vessel previously used as a heater was filled with cold water at the commencement. The current was sent six times eight minutes in each direction; the thermometers were noted every quarter of a minute; and the observations were reduced and compared in the usual way. The result gave no effect of the kind observed in the preceding experiments, but (probably because of a temporary failure in the water-supply for the coolers) showed, on the contrary, a deviation in the mean difference of temperature amounting to 0°029 Cent., being about a tenth part of the amount of that effect, but in the opposite way according to the direction of the current through the conductor. Before the experiment was concluded boiling water was poured through the central vessel and left filling it, but with no lamp below. The two thermometers (A and B) being thus raised to about 27° Cent., the current was again started and was sent through the conductor for three times four minutes in each direction. The thermometers rose each nearly 2°, but fell again by nearly 5½° before the conclusion. The mean differential result, whether from these three periods (amounting to 0°05 Cent.), * Rejected because of a failure in the water-supply through the coolers during the whole of Period I. or from the last two of them without the first (°025 Cent.), was of the same kind as in the first two experiments. This experiment then conclusively demonstrated that the effect previously discovered was really owing to the heat in the central part of the conductor, and not to any inequality in the metal of the conductor itself, nor to any accidental disturbing agency. 43. It was thus established, that the Resinous Electricity carries heat with it in an iron conductor. §§ 44 and 45. Experiment with Copper Conductor, repeated. 44. The very small effect I had discovered of the opposite kind in the copper conductor required confirmation; and indeed the analysis of the progress of the variation (see Tables I. and II. § 56 below) was so unsatisfactory, that I felt it quite an open question, whether it was the true effect, or merely an accidental coincidence of irregularities; and I thought it improbable that contrary effects should really exist in copper and in iron. I made on the 26th of November another experiment on the copper conductor, with the current flowing six times eight minutes each way (instead of eight times six minutes, as before, because the analysis seemed to show that the effects which had chanced to appear in the average results of the six minutes might disappear with longer periods*); but I got still a very small result of the same kind. The full analysis (Table III.) was equally unsatisfactory with those of the two preceding experiments on the same conductor. The following numbers show the temperatures at the reversals, and the final result, as in the previous abridged tables. 45. Conductor composed of thirteen slips of sheet copper. Experiment V. November 26, 1853. | Periods | (Current six times eight minutes in each direction.) Temperatures and differences of temperature after eight minutes of current entering | Diminutions of differences from middles to ends of periods. | |---------|-----------------------------------------------------------------------------------------------------------------|----------------------------------------------------------| | | By end next A. | By end next B. | | | $T_A$ | $T_B$ | $T_B - T_A = D$. | $T_A$ | $T_B$ | $T_B - T_A = D'$. | $D' - D$. | | I. | 50°88 | 52°78 | 1°90 | 50°88 | 52°72 | 1°92 | °02 | | II. | 50°64 | 52°49 | 1°85 | 50°53 | 52°32 | 1°79 | +°06 | | III. | 50°38 | 52°29 | 1°91 | 50°01 | 52°00 | 1°99 | -°08 | | IV. | 50°14 | 52°08 | 1°94 | 49°90 | 51°83 | 1°93 | +°01 | | V. | 49°60 | 51°61 | 2°01 | 49°48 | 51°52 | 2°04 | -°03 | | VI. | 49°11 | 51°22 | 2°11 | 48°80 | 50°92 | 2°12 | -°01 | | Means for five periods | 49°974 | 51°938 | 1°964 | 49°744 | 51°718 | 1°974 | -°01 | Augmentation of difference during periods included... °20 Add average augmentation per half-period ......................... +°002 Effect due to reversal of current .................................. -°01, in favour of Vitreous Electricity. * I now believe that a true effect, amounting to from °01 to °02, was really reached in three or four minutes, and that in the latter parts of the half-periods there was no sensible augmentation of this effect, but MDCCCLVI. 46. Another experiment was also made on the new iron conductor, and results as decisive as those in the first two experiments were obtained. The following abridged Table shows sufficiently the character of the effect demonstrated; and the analysis of the progress of variation is given in the full table (Table III. § 47) below. **Experiment VI. December 2, 1853.** Conductor composed of thirty slips of sheet iron. | Periods | \( T_A \) | \( T_B \) | \( T_B - T_A = D \) | \( T_A \) | \( T_B \) | \( T_B - T_A = D' \) | Augmentations of differences from middles to ends of periods. | |---------|----------|----------|------------------|----------|----------|------------------|---------------------------------------------------------------| | I. | 54°76 | 56°33 | 1°57 | 54°80 | 56°66 | 1°86 | 0°29 | | II. | 54°97 | 56°68 | 1°71 | 54°89 | 56°80 | 1°91 | 0°20 | | III. | 55°01 | 56°70 | 1°69 | 54°93 | 56°86 | 1°93 | 0°24 | | IV. | 55°22 | 56°90 | 1°68 | 55°08 | 57°10 | 2°02 | 0°34 | | V. | 55°31 | 57°08 | 1°77 | 55°07 | 57°12 | 2°05 | 0°28 | | VI. | 55°12 | 57°00 | 1°88 | 54°84 | 57°03 | 2°19 | 0°31 | | Means for five periods | 55°126 | 56°872 | 1°746 | 54°962 | 56°982 | 2°020 | 0°274 | Augmentation of difference \( T_B - T_A \) during included periods 0°33 Deduct average augmentation per half-period .................................. 0°033 Effect due to reversal of current .............................................. 0°241, in favour of Resinous Electricity. 47. The following Tables show the progress of variation of the difference between the temperatures of the two tested localities (A, B) of the iron conductor, during each of the three regular experiments referred to above, as derived directly from the quarter-minute or half-minute observations actually made in the course of each experiment. only irregular fluctuations, sometimes counteracting and reversing the true effect, but generally only diminishing it and increasing it alternately, and always maintaining, during the whole latter half of the aggregate of the half-periods, an average deviation of the kind noted as the final result. A careful consideration of the Tables I., II. and III. given below, § 56, for the copper conductor and of their graphical representation (see Diagram, § 57), is, I think, sufficient to establish this view. [April 9, 1856.] ### Table I. November 12th, 1853. Conductor composed of thirty slips of sheet iron. Middle of conductor in water kept hot by steam blown into it. Streams of water at temperature 6° running through the coolers. Temperatures at middle points A, B of the parts between heater and coolers: Initial $T_A = 51°23'$, $T_B = 52°30'$; Final $T_A = 50°69'$, $T_B = 52°80'$. | Augmentations of difference $T_B - T_A$ (in hundredths of a degree Cent.) during periods | |-----------------------------------------------| | First half-minute of current entering by end next | | Second half-minute of current entering by end next | | Third half-minute of current entering by end next | | Fourth half-minute of current entering by end next | | Fifth half-minute of current entering by end next | | Sixth half-minute of current entering by end next | | Seventh half-minute of current entering by end next | | Eighth half-minute of current entering by end next | | Ninth half-minute of current entering by end next | | Tenth half-minute of current entering by end next | | Eleventh half-minute of current entering by end next | | Twelfth half-minute of current entering by end next | | Thirteenth half-minute of current entering by end next | | Fourteenth half-minute of current entering by end next | | Fifteenth half-minute of current entering by end next | | Sixteenth half-minute of current entering by end next | | Periods | A | B | A | B | A | B | A | B | A | B | A | B | A | B | A | B | A | B | A | B | A | B | A | B | A | B | A | B | A | B | A | B | |---------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---| | I | 0 | -2 | 1 | 1 | 2 | 0 | 0 | 0 | 2 | 0 | 0 | 1 | 0 | 3 | 5 | 1 | 5 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | | II | -4 | 1 | -4 | 2 | -7 | 6 | -4 | 1 | -2 | 1 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | | III | -2 | 1 | -2 | 1 | -3 | 2 | -3 | 2 | -2 | 1 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | | IV | -2 | 1 | -2 | 1 | -3 | 2 | -3 | 2 | -2 | 1 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | | V | -2 | 1 | -2 | 1 | -3 | 2 | -3 | 2 | -2 | 1 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | | VI | -2 | 1 | -2 | 1 | -3 | 2 | -3 | 2 | -2 | 1 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | 2 | -1 | | VII | -3 | 6 | -12 | 7 | -14 | 15 | -23 | 19 | -16 | 14 | -8 | +11 | -11 | 15 | -3 | 13 | -6 | 4 | -3 | 2 | -16 | 11 | -4 | 3 | 2 | -1 | 7 | -4 | 2 | 11 | 10 | 8 | | VIII | -3 | 6 | -12 | 7 | -14 | 15 | -23 | 19 | -16 | 14 | -8 | +11 | -11 | 15 | -3 | 13 | -6 | 4 | -3 | 2 | -16 | 11 | -4 | 3 | 2 | -1 | 7 | -4 | 2 | 11 | 10 | 8 | | IX | -3 | 6 | -12 | 7 | -14 | 15 | -23 | 19 | -16 | 14 | -8 | +11 | -11 | 15 | -3 | 13 | -6 | 4 | -3 | 2 | -16 | 11 | -4 | 3 | 2 | -1 | 7 | -4 | 2 | 11 | 10 | 8 | | X | -3 | 6 | -12 | 7 | -14 | 15 | -23 | 19 | -16 | 14 | -8 | +11 | -11 | 15 | -3 | 13 | -6 | 4 | -3 | 2 | -16 | 11 | -4 | 3 | 2 | -1 | 7 | -4 | 2 | 11 | 10 | 8 | | XI | -3 | 6 | -12 | 7 | -14 | 15 | -23 | 19 | -16 | 14 | -8 | +11 | -11 | 15 | -3 | 13 | -6 | 4 | -3 | 2 | -16 | 11 | -4 | 3 | 2 | -1 | 7 | -4 | 2 | 11 | 10 | 8 | | XII | -3 | 6 | -12 | 7 | -14 | 15 | -23 | 19 | -16 | 14 | -8 | +11 | -11 | 15 | -3 | 13 | -6 | 4 | -3 | 2 | -16 | 11 | -4 | 3 | 2 | -1 | 7 | -4 | 2 | 11 | 10 | 8 | | XIII | -3 | 6 | -12 | 7 | -14 | 15 | -23 | 19 | -16 | 14 | -8 | +11 | -11 | 15 | -3 | 13 | -6 | 4 | -3 | 2 | -16 | 11 | -4 | 3 | 2 | -1 | 7 | -4 | 2 | 11 | 10 | 8 | | XIV | -3 | 6 | -12 | 7 | -14 | 15 | -23 | 19 | -16 | 14 | -8 | +11 | -11 | 15 | -3 | 13 | -6 | 4 | -3 | 2 | -16 | 11 | -4 | 3 | 2 | -1 | 7 | -4 | 2 | 11 | 10 | 8 | | XV | -3 | 6 | -12 | 7 | -14 | 15 | -23 | 19 | -16 | 14 | -8 | +11 | -11 | 15 | -3 | 13 | -6 | 4 | -3 | 2 | -16 | 11 | -4 | 3 | 2 | -1 | 7 | -4 | 2 | 11 | 10 | 8 | | XVI | -3 | 6 | -12 | 7 | -14 | 15 | -23 | 19 | -16 | 14 | -8 | +11 | -11 | 15 | -3 | 13 | -6 | 4 | -3 | 2 | -16 | 11 | -4 | 3 | 2 | -1 | 7 | -4 | 2 | 11 | 10 | 8 | --- **Note:** Differences of augmentation during corresponding half-minutes of first and second halves of a period, summed for five periods. Differences of augmentation in equal intervals from beginning and minute of period, summed for five periods. Mean augmentation of difference in favour of thermometer next entering current after reversal and flow for min. ### Table II. November 19th, 1853. Middle of conductor in water kept Streams of water at temperature Temperatures at middle points A, B of the parts between heater and | Augmentations of difference $T_B - T_A$ (in hundredths of a degree Cent.), during Periods | 1. | 2. | 3. | 4. | 5. | 6. | 7. | 8. | 9. | 10. | 11. | 12. | 13. | |-----------------------------------------------|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----| | First quarter-minute entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Second quarter-minute entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Third quarter-minute entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Fourth quarter-minute entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Fifth quarter-minute entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Sixth quarter-minute entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Seventh quarter-minute entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Eighth quarter-minute entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Ninth quarter-minute entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Tenth quarter-minute entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Eleventh quarter-minute entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Twelfth quarter-minute entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Thirteenth quarter-minute entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | After the conclusion of this experiment the current was started in one direction and broken, and effect, when the heater and coolers were kept regular, on either thermometer amounted to about $1^\circ 4$. * Period II. rejected because the water-supply, which had failed during the whole of Period I., ductor composed of thirty slips of sheet iron. boiling by steam blown into it. 6° running through the coolers. coolers:—Initial $T_A = 56°44$, $T_B = 58°28$; Final $T_A = 48°06$, $T_B = 51°59$. | 14. | 15. | 16. | 17. | 18. | 19. | 20. | 21. | 22. | 23. | 24. | 25. | 26. | 27. | 28. | 29. | 30. | 31. | 32. | |-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----| | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | | 4 | 3 | 4 | 2 | 1 | 5 | 2 | 3 | 2 | 3 | 1 | 1 | 0 | 1 | 5 | 2 | 2 | 2 | 0 | 0 | | 6 | 3 | 4 | 2 | 1 | 5 | 2 | 3 | 2 | 3 | 1 | 1 | 0 | 1 | 5 | 2 | 2 | 2 | 0 | 0 | | 0 | 0 | 3 | 0 | 3 | 0 | 1 | 0 | 1 | 0 | 1 | 0 | 1 | 0 | 1 | 0 | 1 | 0 | 1 | 0 | | 4 | 0 | 3 | 0 | 2 | 2 | 1 | 1 | 5 | 3 | 0 | 2 | 2 | 1 | 1 | 5 | 3 | 0 | 2 | 2 | 1 | | -2 | 2 | 1 | 3 | 2 | 1 | 3 | 2 | 1 | 3 | 2 | 1 | 3 | 2 | 1 | 3 | 2 | 1 | 3 | 2 | 1 | | 1 | 0 | 2 | 0 | 0 | 1 | 0 | 1 | 2 | 1 | 0 | 1 | 2 | 1 | 0 | 1 | 2 | 1 | 0 | 1 | | 0 | 2 | 1 | 2 | 1 | 0 | 2 | 3 | 1 | 1 | 1 | 1 | 0 | 1 | 0 | 1 | 0 | 1 | 0 | 1 | | 3 | 4 | 10 | 1 | 4 | 4 | 4 | 15 | 2 | 1 | 12 | 2 | 4 | 5 | 2 | 6 | 1 | 7 | 2 | 8 | 4 | 1 | | 1 | 11 | 8 | 19 | -3 | 13 | 6 | 7 | -8 | 8 | 6 | 3 | -1 | 11 | 9 | -3 | 5 | 10 | 4 | | 152 | 163 | 171 | 190 | 187 | 200 | 206 | 213 | 205 | 213 | 219 | 222 | 221 | 232 | 241 | 238 | 243 | 253 | 257 | | °152| °163| °171| °190| °187| °200| °206| °213| °205| °213| °219| °222| °221| °223| °241| °238| °243| °253| °257| | 3½ min.| 3½ min.| 4 min.| 4½ min.| 4½ min.| 5 min.| 5½ min.| 5½ min.| 6 min.| 6½ min.| 6½ min.| 7 min.| 7½ min.| 7½ min.| 8 min.| started in the other direction and broken, several times; and it was found that its absolute heating only commenced giving a stream through the coolers at the commencement of the second period. ### Table III. December 2nd, 1853. Middle of conductor at Streams of water at temperature Temperatures at middle points A, B of the parts between heater and | Augmentations of difference $T_B - T_A$ (in hundredths of a degree Cent.), during Periods | 1. | 2. | 3. | 4. | 5. | 6. | 7. | 8. | 9. | 10. | 11. | 12. | 13. | |---|---|---|---|---|---|---|---|---|---|---|---|---| | First quarter-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Second quarter-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Third quarter-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Fourth quarter-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Fifth quarter-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Sixth quarter-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Seventh quarter-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Eighth quarter-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Ninth quarter-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Tenth quarter-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Eleventh quarter-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Twelfth quarter-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | | Thirteenth quarter-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | **I.** | 0 | 0 | 1 | 2 | 0 | 1 | −2 | 2 | −1 | 1 | 3 | 1 | −1 | 2 | 0 | 2 | 2 | 2 | 0 | 2 | −2 | 2 | 3 | 3 | −1 | 1 | **II.** | 4 | 0 | −1 | 0 | −2 | 1 | −1 | 3 | −2 | 1 | −1 | 1 | −2 | 2 | −3 | 1 | 0 | 1 | −1 | 0 | 1 | 0 | −4 | 2 | **III.** | 1 | 0 | 0 | 0 | 0 | 0 | −1 | 1 | 0 | 1 | −1 | 0 | 0 | 2 | 0 | 3 | 0 | 3 | −1 | 3 | −1 | 2 | −2 | 1 | **IV.** | 0 | 1 | 0 | 1 | −2 | 0 | 1 | 0 | −2 | 1 | −4 | 1 | −3 | 3 | −2 | 2 | 0 | 3 | −2 | 2 | 1 | −2 | 1 | 0 | **V.** | 0 | 1 | −1 | 3 | 0 | 3 | −1 | 4 | −1 | 0 | 0 | −1 | 4 | −3 | 5 | −1 | 1 | −2 | 2 | −1 | 1 | −1 | 2 | −1 | 1 | **VI.** | 0 | 0 | −1 | 0 | 0 | 1 | −1 | 1 | −1 | 0 | −1 | 0 | 0 | 0 | 0 | 2 | −1 | 1 | −1 | 2 | 0 | 2 | 0 | 3 | −2 | 0 | **Augmentations during quarter-minutes, summed for five periods** | 5 | 2 | −3 | −2 | −4 | 5 | −3 | 9 | −6 | 3 | −7 | 1 | −6 | 11 | −8 | 13 | −2 | 9 | −6 | 9 | −4 | 7 | −3 | 9 | −11 | 2 | **Differences of augmentation during corresponding quarter-minutes of first and second halves of a period, summed for five periods** | −3 | 1 | 9 | 12 | 9 | 8 | 17 | 21 | 11 | 15 | 11 | 12 | 13 | **Differences of augmentation in equal intervals from beginning and middle of a period, summed for five periods** | −3 | −2 | +7 | 19 | 28 | 36 | 53 | 74 | 85 | 100 | 111 | 123 | 136 | **Mean augmentation of difference in favour of thermometer next entering current** | °-003 | °-002 | °-007 | °-019 | °-028 | °-036 | °-053 | °-074 | °-085 | °-100 | °-111 | °-123 | °-136 | **after reversal and flow for** | ¼ min. | ½ min. | ¾ min. | 1 min. | 1¼ min. | 1½ min. | 1¾ min. | 2 min. | 2¼ min. | 2½ min. | 3 min. | 3¼ min. | ductor composed of thirty slips of sheet iron. temperature 95° Cent. 7°5 running through the coolers. coolers:—Initial $T_A = 54°20$, $T_B = 55°70$; Final $T_A = 54°84$, $T_B = 57°03$. 48. The gradual augmentation of the difference $T_A - T_B$ from its value at a time when the current had been flowing for eight minutes entering by the end next B, consequent upon reversing the current and letting it flow continuously entering by the end next A, is shown by the numbers at the foot of each table, as a mean result derived from a single experiment. The mean of the results of the three experiments is shown by the following numbers, and is exhibited by a curve in the Diagram (fig. 4) of § 57 below. **Current entering by end next A.** | Time from instant of reversal in quarter-minutes | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | 32 | |-----------------------------------------------|---|---|---|---|---|---|---|---|---|---|----|----|----|----|----|----|----|----|----|----|----|----|----|----|----|----|----|----|----|----| | Mean Augmentation of difference $T_A - T_B$ | .000 | .006 | .0185 | .032 | .044 | .0695 | .091 | .1167 | .138 | .1590 | .175 | .1925 | .2183 | .2461 | .2767 | .3092 | .3455 | .3843 | .4265 | .4725 | .5229 | .5773 | .6368 | .6995 | .7667 | .8383 | .9161 | .9981 | .9829 | .9634 | .9403 | .9167 | 49. That Vitreous Electricity carries heat with it in copper is indicated by each of the three experiments on the thirteen slip conductor adduced above, but by so narrow an effect; amounting on an average to only 0°.02 Cent., which corresponds to a reading of half that amount, or $\frac{1}{100}$th of a degree, being $\frac{1}{10}$th of a division on the scale of each thermometer; with such discrepancies among the results of the different experiments (Oct. 28th, effect '039, Nov. 2nd, '0143, Nov. 26th, '01); and with so great fluctuations in the course of each experiment (see Tables I., II., III., § 56 below); that I did not venture to draw from them so seemingly improbable a conclusion, as that the convective effects in copper and in iron should be in contrary directions. The dynamic theory (§ 18) was fully satisfied by the demonstration which the experiments gave, that the convective effect is undoubtedly in iron a conveying of heat in the direction of the Resinous Electricity, and that it is less in amount in copper, whether in the same direction as in iron or in the contrary direction. But it was still an object of great interest, (in fact an object of much greater interest than any verification of conclusions from the dynamic theory, which were in reality as certain before as after the experiments directly demonstrating them,) to ascertain the actual nature of the convective effect in copper, and I therefore endeavoured to make more decisive experiments for discovering it. 50. The three experiments which had been made were quite sufficient to prove that the convective effect, whatever its true nature might be, was nearly insensible to my thermometers without either more powerful currents or a more sensitive conductor. To work with more powerful currents would have increased immensely the labour of carrying out the experiments, and would besides involve a large addition to the battery which had been used hitherto. I preferred therefore to make the conductor more sensitive, which I saw could be done by diminishing the body of metal in the tested parts, and so preventing the looked-for thermal effect from being so much conducted away from the localities of the thermometers as it had been. I accordingly had several slips of copper cut away from each side of the conductor in the parts between the heater and the coolers, leaving the parts within these vessels unchanged. 51. Several experiments were made on the conductor thus reduced successively to smaller and smaller numbers of slips; but the results did not appear much more decided than they had been in the experiments on the unreduced conductor, until it was tried with all the slips but two cut away. Thus with four slips left, the following results were obtained: 52. Copper conductor reduced to four slips. Experiment VII. February 1854. | Periods | \( T_A \) | \( T_B \) | \( T_A - T_B = D \) | \( T_A \) | \( T_B \) | \( T_A - T_B = D' \) | Augmentations of differences from middles to ends of periods. | |---------|--------|--------|-----------------|--------|--------|-----------------|--------------------------------------------------| | I. | 46°00 | 45°24 | .76 | 45°92 | 45°27 | .65 | .11 | | II. | 45°82 | 45°19 | .63 | 46°01 | 45°43 | .58 | .05 | | III. | 46°00 | 45°50 | .50 | 46°02 | 45°54 | .48 | .02 | | IV. | 46°20 | 45°77 | .43 | 46°21 | 45°76 | .45 | .02 | | V. | 46°18 | 45°80 | .38 | 46°09 | 45°71 | .38 | .00 | | VI. | 46°09 | 45°72 | .37 | 46°05 | 45°68 | .37 | .00 | Means for five periods: 46°058 45°596 .462 46°076 45°624 .452 .010 Diminution of difference during periods included ...... .28 Add average diminution per half-period ........................................ +.028 Effect due to reversal of current .................................................. .018, in favour of Vitreous Electricity. The effect shown here is of the same kind as had been found in all the previous experiments, but was still too small to be very satisfactory. Some unknown cause made the difference \( T_A - T_B \) to diminish so much through the whole experiment as to overpower the apparent tendency of the current from B to A to increase it, and the abridged table has on this account a very unsatisfactory appearance as regards the conclusion drawn from it after the proper correction for their diminution is applied: but the full examination of the progress of variation in the course of the experiment shown in Table IV. below is much less unsatisfactory, and shows undoubtedly the true convective effect in copper. 53. The following results, derived from the first experiment made on the conductor reduced to two slips, show a very marked increase in the effect, and make the result quite apparent even without the full analysis given below in Table V. Experiment VIII. February 23rd, 1854. Copper conductor reduced to two slips. | Periods | \( T_A \) | \( T_B \) | \( T_B - T_A = D \) | \( T_A' \) | \( T_B' \) | \( T_B' - T_A' = D' \) | Diminutions of differences from middles to ends of periods | |---------|----------|----------|------------------|----------|----------|------------------|--------------------------------------------------------| | I. | 54·81 | 56·31 | 1·50 | 58·47 | 56·88 | 1·41 | 0·09 | | II. | 55·90 | 57·32 | 1·42 | 56·37 | 57·63 | 1·26 | 0·14 | | III. | 56·70 | 57·99 | 1·29 | 57·16 | 58·29 | 1·13 | 0·16 | | IV. | 57·49 | 58·70 | 1·21 | 57·80 | 58·91 | 1·11 | 0·10 | | V. | 57·80 | 58·98 | 1·18 | 58·18 | 59·20 | 1·02 | 0·16 | | VI. | 58·29 | 59·80 | 1·51 | 58·37 | 59·80 | 1·43 | 0·08 | | VII. | 58·27 | 59·80 | 1·53 | 58·08 | 59·50 | 1·42 | 0·11 | | Means for six periods... | 57·4083 | 58·7650 | 1·3567 | 57·663 | 58·888 | 1·2283 | 0·1283 | Augmentation of differences during periods included... 0·01 Add average augmentation per half-period.................. 0·008 Effect due to reversal of current.......................... 0·01291, in favour of Vitreous Electricity. 54. The effect here shown is, as regards amount, unmistakeable, and it appeared quite to establish the conclusion I had not ventured to draw from the previous experiments on the copper conductor. As it was clear that diminution in the power of the conductor had now begun to augment its sensibility for the effect under investigation, I had it further reduced by cutting away from its breadth, above and below, between the heaters and the bends for the thermometer-bulbs on the two sides, and on the other sides of those bends as far as the coolers. An experiment was then made which led to the following results: Experiment IX. March 7th, 1854. Copper conductor diminished in breadth. | Periods | \( T_A \) | \( T_B \) | \( T_B - T_A = D \) | \( T_A' \) | \( T_B' \) | \( T_B' - T_A' = D' \) | Diminutions of differences from middles to ends of periods | |---------|----------|----------|------------------|----------|----------|------------------|--------------------------------------------------------| | I. | 56·61 | 58·00 | 1·39 | 57·56 | 58·98 | 1·42 | 0·03 | | II. | 58·19 | 59·69 | 1·50 | 58·74 | 60·01 | 1·27 | 0·23 | | III. | 59·12 | 60·70 | 1·58 | 59·61 | 61·08 | 1·47 | 0·11 | | IV. | 59·91 | 61·53 | 1·62 | 60·18 | 61·75 | 1·57 | 0·05 | | V. | 60·50 | 62·18 | 1·68 | 61·15 | 62·53 | 1·38 | 0·30 | | VI. | 61·71 | 63·02 | 1·31 | 61·87 | 62·91 | 1·04 | 0·27 | | VII. | 61·07 | 62·80 | 1·73 | 61·07 | 62·62 | 1·55 | 0·18 | | Means for six periods... | 60·083 | 61·653 | 1·57 | 60·4367 | 61·8167 | 1·38 | 0·19 | Augmentation of differences during periods included... 0·13 Add average augmentation per half-period.................. = 0·01083 = 0·0083 After this experiment I considered it quite established that the Vitreous Electricity carries heat with it in copper. 55. The conductor was still further diminished in breadth (so as to be only an inch broad in the parts between the heater and coolers on each side), and an experiment was made before my class on the 19th April, 1854, leading to the following results, shown as in the abridged tables of the preceding experiments. Experiment X. April 19th, 1854. Copper conductor of two slips, further diminished in breadth. | Periods | T_A | T_B | T_B - T_A = D | T_A | T_B | T_B - T_A = D' | Diminutions of differences from middles to ends of periods | |---------|-----|-----|---------------|-----|-----|----------------|----------------------------------------------------------| | I. | 74·30 | 76·60 | 2·30 | 74·81 | 77·50 | 2·69 | -0·39 | | II. | 73·80 | 76·48 | 2·68 | 75·32 | 78·10 | 2·78 | -1·10 | | III. | 76·25 | 79·42 | 3·17 | 76·17 | 79·18 | 3·01 | -1·16 | | IV. | 76·33 | 79·51 | 3·18 | 76·28 | 79·37 | 3·09 | -0·9 | | V. | 75·60 | 78·69 | 3·09 | 75·20 | 78·07 | 2·87 | -2·2 | | VI. | 74·80 | 77·70 | 2·90 | 75·00 | 77·75 | 2·75 | -1·5 | | VII. | 74·10 | 76·84 | 2·74 | 75·42 | 78·20 | 2·78 | -0·4 | Means, Period I. off ... 75·147 78·107 2·96 75·565 78·445 2·88 -0·08 Augmentation of differences during periods included ... 0·09 Add average augmentation per half-period ......................... 0·0075 Effect due to reversal of current .................................. 0·0875 Means, Periods I. and VII. off ............... 75·356 78·36 3·004 75·594 78·494 2·90 -0·104 Augmentation of differences during periods included ... 0·06 Deduct average augmentation per half-period ......................... 0·006 Effect due to reversal of current .................................. 0·110 The effect here obtained, although of quite a decisive character, does not appear to show any increased sensibility resulting from the further diminution in the breadth of the conductor. 56. The following Tables [printed after a part of § 58] show a complete analysis of the results of the seven experiments on the copper conductor which have been adduced. 57. The average progress towards the final effect of a reversal, as indicated by the numbers at the ends of these Tables, is exhibited graphically for the copper conductor in the different states in which it was used in the experiments, in the following Diagram, along with the curve exhibiting the corresponding reverse effect in the iron conductor. The uppermost curve represents the results of three experiments with the Iron conductor (thirty slips), the points marked * representing the mean of three days' observation, and the points marked ‡ that of two days' observation. The lowest curve represents the results of three experiments with the Copper conductor (thirteen slips), the points marked ¥ denoting three days', and the simple dots • two days' observation. The middle curve represents three experiments with the Copper conductor (two slips), the points marked ⊙ denoting the mean of three days' observation. Fig. 4. 58. The diminution of the conducting power in the copper conductor had so markedly augmented the looked-for indication of a convective effect, that it was to be expected a corresponding augmentation might be obtained by treating the iron conductor similarly. Instead, however, of cutting up the iron conductor, which, as it stood, possessed sensibility enough to give a very decided result, I prepared a new iron conductor on a much smaller scale. It appeared that the smaller the conducting power for the same strength of current, and the same difference of temperatures between hot and cold, the greater would be the indication of convective effect; and the greatest indication would therefore be obtained by [Continued after Table VII.] ### Table I. October 28th, 1953. Conductor composed of thirteen slips of sheet copper. Middle of conductor in water kept boiling. Streams of water at temperature $6^\circ$ running through the coolers. Temperatures at middle points A, B of the parts between heater and coolers:—Initial $T_A=54^\circ70$, $T_B=52^\circ68$; Final $T_A=54^\circ70$, $T_B=52^\circ68$. | Augmentations of difference $T_A-T_B$ (in hundredths of a degree Cent.) during periods | 1. | 2. | 3. | 4. | 5. | 6. | 7. | 8. | 9. | 10. | 11. | 12. | |------------------------------------------|----|----|----|----|----|----|----|----|----|----|----|----| | First half-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | | Second half-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | | Third half-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | | Fourth half-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | | Fifth half-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | | Sixth half-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | | Seventh half-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | | Eighth half-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | | Ninth half-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | | Tenth half-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | | Eleventh half-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | | Twelfth half-minute of current entering by end next | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | Differences of augmentation during corresponding halves of five periods, summed for five periods: | Differences of augmentation in equal intervals from beginning and middle of a period, summed for five periods | 37 | -14 | 24 | -20 | 3 | -34 | 20 | 22 | 23 | -17 | 0 | -5 | | Differences of augmentation in equal intervals from beginning and middle of a period, summed for five periods | 37 | 26 | 47 | 27 | 30 | -4 | 16 | 38 | 61 | 44 | 44 | 39 | | Mean augmentation of difference in favour of thermometer remote from entering current... after reversal and flow for 1 min. | <037 | <023 | <047 | <027 | <030 | <004 | <016 | <088 | <061 | <044 | <044 | <069 | | Mean augmentation of difference in favour of thermometer remote from entering current... after reversal and flow for 1½ min. | <037 | <023 | <047 | <027 | <030 | <004 | <016 | <088 | <061 | <044 | <044 | <069 | | Mean augmentation of difference in favour of thermometer remote from entering current... after reversal and flow for 2 min. | <037 | <023 | <047 | <027 | <030 | <004 | <016 | <088 | <061 | <044 | <044 | <069 | | Mean augmentation of difference in favour of thermometer remote from entering current... after reversal and flow for 2½ min. | <037 | <023 | <047 | <027 | <030 | <004 | <016 | <088 | <061 | <044 | <044 | <069 | | Mean augmentation of difference in favour of thermometer remote from entering current... after reversal and flow for 3 min. | <037 | <023 | <047 | <027 | <030 | <004 | <016 | <088 | <061 | <044 | <044 | <069 | | Mean augmentation of difference in favour of thermometer remote from entering current... after reversal and flow for 3½ min. | <037 | <023 | <047 | <027 | <030 | <004 | <016 | <088 | <061 | <044 | <044 | <069 | | Mean augmentation of difference in favour of thermometer remote from entering current... after reversal and flow for 4 min. | <037 | <023 | <047 | <027 | <030 | <004 | <016 | <088 | <061 | <044 | <044 | <069 | | Mean augmentation of difference in favour of thermometer remote from entering current... after reversal and flow for 4½ min. | <037 | <023 | <047 | <027 | <030 | <004 | <016 | <088 | <061 | <044 | <044 | <069 | | Mean augmentation of difference in favour of thermometer remote from entering current... after reversal and flow for 5 min. | <037 | <023 | <047 | <027 | <030 | <004 | <016 | <088 | <061 | <044 | <044 | <069 | | Mean augmentation of difference in favour of thermometer remote from entering current... after reversal and flow for 5½ min. | <037 | <023 | <047 | <027 | <030 | <004 | <016 | <088 | <061 | <044 | <044 | <069 | ### Table II. November 2nd, 1853. Middle of conductor in water kept Streams of water at temperature Temperatures of middle points A, B of the parts between heater and | Augmentations of difference $T_A - T_B$ (in hundredths of a degree Cent.), during Periods | 1. First quarter-minute entering by end next | 2. Second quarter-minute entering by end next | 3. Third quarter-minute entering by end next | 4. Fourth quarter-minute entering by end next | 5. Fifth quarter-minute entering by end next | 6. Sixth quarter-minute entering by end next | 7. Seventh quarter-minute entering by end next | 8. Eighth quarter-minute entering by end next | 9. Ninth quarter-minute entering by end next | 10. Tenth quarter-minute entering by end next | |---|---|---|---|---|---|---|---|---|---|---| | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | | I. | -6 | -7 | 5 | 1 | 0 | -4 | -3 | -2 | 1 | -5 | -5 | -9 | 0 | 22 | -1 | -5 | -1 | 0 | 2 | -11 | | II. | 0 | 13 | 0 | -1 | 8 | -3 | 4 | -6 | 10 | -3 | -4 | -4 | 1 | 2 | -4 | 7 | 2 | 9 | -5 | -10 | | III. | 7 | -1 | 6 | 0 | -2 | 3 | -1 | 7 | -2 | 0 | -7 | -8 | 7 | -1 | 11 | -9 | 3 | 7 | -7 | 11 | | IV. | 1 | 1 | 3 | 4 | 0 | -1 | 1 | -2 | 7 | -1 | -6 | -2 | 9 | -2 | 9 | 7 | -3 | 10 | -2 | 3 | -5 | | V. | -5 | 9 | 2 | -5 | 13 | 0 | -5 | 7 | -2 | -1 | 3 | -6 | -6 | -1 | 3 | -8 | 4 | 2 | -7 | | VI. | 4 | 10 | 1 | -15 | 1 | 18 | 2 | -2 | 9 | -3 | -5 | 0 | -3 | 3 | -3 | 8 | -4 | 2 | -7 | | VII. | 4 | 5 | 9 | -1 | 3 | -5 | 10 | -4 | -1 | 7 | 5 | 4 | -9 | 5 | 4 | -8 | 15 | -4 | 1 | -6 | 7 | | VIII. | 7 | -9 | 2 | -1 | 5 | 0 | -1 | 2 | -8 | 4 | 6 | 5 | 12 | -1 | -10 | 5 | 2 | -6 | 1 | 0 | Augmentations during quarter-minutes, summed for seven periods | 9 | 30 | 9 | -15 | 9 | 7 | -5 | 5 | 2 | 4 | -7 | -20 | 23 | 15 | -4 | 12 | 13 | 11 | -35 | -18 | Differences of augmentation during corresponding quarter-minutes of first and second halves of a period, summed for seven periods | 21 | -24 | 16 | 10 | 2 | -13 | -8 | 16 | -2 | 17 | Differences of augmentation in equal intervals from beginning and middle of a period, summed for seven periods | 21 | -3 | 13 | 23 | 25 | 12 | 4 | 20 | 18 | 35 | Mean augmentation of difference in favour of thermometer remote from entering current... after reversal and flow for | °-015 | -°-002 | °-009 | °-016 | °-018 | °-009 | °-003 | °-014 | °-013 | °-025 | |---|---|---|---|---|---|---|---|---|---| | ¼ min. | ½ min. | ¾ min. | 1 min. | 1¼ min. | 1½ min. | 1¾ min. | 2 min. | 2¼ min. | 2½ min. | ductor composed of thirteen slips of sheet copper. boiling-hot by steam blown into it. 10°4 running through the coolers. coolers:—Initial $T_A = 54°10$, $T_B = 52°80$; Final $T_A = 54°68$, $T_B = 53°38$. ### Table III. November 26th, 1853. Temperatures at the middle points A, B of the parts between heater and | Diminutions of difference $T_B - T_A$ (in hundredths of a degree Cent.), during Periods | 1. First quarter-minute entering by end next | 2. Second quarter-minute entering by end next | 3. Third quarter-minute entering by end next | 4. Fourth quarter-minute entering by end next | 5. Fifth quarter-minute entering by end next | 6. Sixth quarter-minute entering by end next | 7. Seventh quarter-minute entering by end next | 8. Eighth quarter-minute entering by end next | 9. Ninth quarter-minute entering by end next | 10. Tenth quarter-minute entering by end next | 11. Eleventh quarter-minute entering by end next | 12. Twelfth quarter-minute entering by end next | 13. Thirteenth quarter-minute entering by end next | |---|---|---|---|---|---|---|---|---|---|---|---|---| | I. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | A. | B. | | | −1 | −3 | 0 | 2 | 3 | 0 | 0 | 1 | 1 | 2 | 1 | 1 | −1 | 1 | | II. | 0 | 0 | 0 | 0 | 1 | 0 | 1 | 0 | 0 | 1 | 1 | 0 | 0 | −1 | −1 | | III. | 1 | −1 | 2 | −2 | 0 | 1 | −1 | 0 | −1 | 0 | −1 | 0 | 1 | 0 | −2 | −2 | | IV. | 0 | 0 | 0 | −1 | −1 | 0 | 0 | 1 | 0 | 0 | 0 | 1 | 0 | 1 | 1 | 0 | 0 | | V. | 1 | 0 | 0 | 0 | −1 | 1 | 0 | −1 | −2 | 0 | −2 | −1 | 0 | −3 | −1 | −2 | −2 | | VI. | 0 | 0 | 0 | 2 | −1 | −1 | 0 | −1 | −2 | 1 | −1 | 0 | −1 | −1 | 0 | −2 | −2 | Diminutions during quarter-minutes, summed for five periods | 2 | −1 | 2 | −1 | −3 | 2 | −1 | 0 | −5 | 1 | −3 | 0 | −1 | −3 | −1 | −1 | −3 | −3 | −5 | 2 | −1 | 8 | 6 | 1 | 4 | −4 | Differences of diminution during corresponding quarter-minutes of first and second halves of a period, summed for five periods | −3 | −3 | 5 | 1 | 6 | 3 | −2 | 0 | 0 | 7 | 9 | −5 | −8 | Differences of diminution in equal intervals from beginning and middle of a period, summed for five periods | −3 | −6 | −1 | 0 | 6 | 9 | 7 | 7 | 7 | 14 | 23 | 18 | 10 | Mean augmentation of difference in favour of thermometer remote from entering current ... | −0.003 | −0.006 | −0.001 | −0.000 | −0.006 | −0.009 | −0.007 | −0.007 | −0.007 | −0.014 | −0.023 | −0.018 | −0.010 | ... after reversal and flow for | ½ min. | ¼ min. | ¾ min. | 1 min. | 1¼ min. | 1½ min. | 1¾ min. | 2 min. | 2¼ min. | 2½ min. | 2¾ min. | 3 min. | 3¼ min. | ductors composed of thirteen slips of sheet copper. ductor at 99° Cent. coolers:—Initial $T_A = 50^\circ 96$, $T_B = 52^\circ 89$; Final $T_A = 48^\circ 80$, $T_B = 50^\circ 92$. | | 14. | 15. | 16. | 17. | 18. | 19. | 20. | 21. | 22. | 23. | 24. | 25. | 26. | 27. | 28. | 29. | 30. | 31. | 32. | |--------|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----|-----| | A | B | A | B | A | B | A | B | A | B | A | B | A | B | A | B | A | B | A | B | | 1 | 1 | 0 | 0 | 0 | 2 | 1 | 0 | -2 | -1 | -1 | 1 | 1 | 0 | -1 | 0 | -1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | | 0 | 1 | 0 | 1 | -1 | -1 | 1 | 2 | 1 | -1 | 0 | -3 | -1 | 2 | 0 | 1 | 1 | 2 | 2 | 1 | -2 | 1 | -1 | 0 | 0 | -1 | 2 | 1 | 0 | | 1 | 1 | -4 | 1 | -1 | 0 | -1 | 1 | 0 | -1 | -1 | 0 | -3 | -2 | -1 | -2 | 1 | 0 | -1 | 2 | -2 | -1 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | | 3 | 1 | 1 | 0 | -1 | 0 | -3 | 0 | -2 | 0 | -1 | -2 | 0 | -1 | 1 | 1 | 0 | 0 | -2 | -1 | 1 | 1 | 0 | 0 | -1 | 0 | -1 | 1 | 0 | | -2 | -2 | 0 | -1 | 1 | 1 | 0 | 1 | 1 | 0 | 0 | 1 | 2 | -1 | 0 | 3 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | -1 | -1 | 0 | 0 | 1 | 0 | | 1 | -2 | 0 | 0 | 0 | 2 | -4 | 1 | -1 | 0 | 0 | -2 | 1 | 0 | 1 | 0 | 0 | 1 | 0 | -1 | 0 | -1 | -3 | -1 | 0 | -1 | 1 | 4 | 0 | -1 | 1 | 0 | | 3 | -1 | -3 | 1 | -2 | 2 | -7 | 5 | -1 | -1 | -2 | -7 | 2 | -4 | 1 | -1 | 3 | 3 | -1 | -1 | 2 | 1 | -1 | -7 | -1 | -1 | -1 | 5 | -1 | -2 | 2 | -2 | 0 | 0 | 4 | 1 | | -4 | 4 | 4 | 12 | 0 | -5 | -6 | -2 | 0 | 0 | -1 | -6 | 0 | 0 | -6 | 4 | 2 | 4 | 0 | | 6 | 10 | 14 | 26 | 26 | 21 | 15 | 13 | 13 | 13 | 12 | 6 | 6 | 6 | 0 | 4 | 6 | 10 | 10 | | °-006 | °-010 | °-014 | °-024 | °-026 | °-015 | °-013 | °-013 | °-012 | °-006 | °-006 | °-006 | °-004 | °-006 | °-010 | °-010 | | 3½ min. | 3½ min. | 4 min. | 4½ min. | 4½ min. | 4½ min. | 5 min. | 5½ min. | 5½ min. | 5½ min. | 6 min. | 6½ min. | 6½ min. | 7 min. | 7½ min. | 7½ min. | 8 min. | MDCCCLVI. ### Table IV. February 1854. Copper conductor reduced to four slips. Stream of water at 60° Cent. through middle of conductor. Streams of water at temperature 5°9 running through coolers. Temperatures at middle points A, B of the parts between heater and coolers:—Initial $T_A = 44°7$, $T_B = 45°3$; Final $T_A = 45°68$, $T_B = 46°05$. | Diminutions of difference $T_B - T_A$ (in hundredths of a degree Cent.), during periods | |-----------------------------------------------| | I. | II. | III. | IV. | V. | VI. | | First half-minute of current entering by end next | 1 | 2 | 3 | 4 | 5 | | Second half-minute of current entering by end next | 1 | 2 | 3 | 4 | 5 | | Third half-minute of current entering by end next | 1 | 2 | 3 | 4 | 5 | | Fourth half-minute of current entering by end next | 1 | 2 | 3 | 4 | 5 | | Fifth half-minute of current entering by end next | 1 | 2 | 3 | 4 | 5 | | Sixth half-minute of current entering by end next | 1 | 2 | 3 | 4 | 5 | | Seventh half-minute of current entering by end next | 1 | 2 | 3 | 4 | 5 | | Eighth half-minute of current entering by end next | 1 | 2 | 3 | 4 | 5 | | Ninth half-minute of current entering by end next | 1 | 2 | 3 | 4 | 5 | | Tenth half-minute of current entering by end next | 1 | 2 | 3 | 4 | 5 | | Eleventh half-minute of current entering by end next | 1 | 2 | 3 | 4 | 5 | | Twelfth half-minute of current entering by end next | 1 | 2 | 3 | 4 | 5 | | Thirteenth half-minute of current entering by end next | 1 | 2 | 3 | 4 | 5 | | Fourteenth half-minute of current entering by end next | 1 | 2 | 3 | 4 | 5 | | Fifteenth half-minute of current entering by end next | 1 | 2 | 3 | 4 | 5 | | Sixteenth half-minute of current entering by end next | 1 | 2 | 3 | 4 | 5 | Differences of diminution during corresponding half-minutes of first and second halves of a period, summed for five periods: - Differences of diminution in period intervals from 1st to 5th, and middle of a period, summed for five periods: -3°008, -3°008, -3°008, -3°002, -3°006 Differences of diminution in period intervals from 6th to 10th, and middle of a period, summed for five periods: -3°001, -3°010, -3°019, -3°023 Mean augmentation of difference in favour of thermometer remote from entering current: -3°008 After reversal and flow for 1 min., 1½ min., 2 min., 2½ min., 3 min., 3½ min., 4 min., 4½ min., 5 min., 5½ min., 6 min., 6½ min., 7 min., 7½ min., 8 min. ### Table V. February 23rd, 1854. Copper conductor reduced to two slips. Water kept at temperature 99°.5 Cent. in heater. Streams of water at temperature 5°6 through coolers: Temperatures at middle points A, B of the parts of conductor between heater and coolers: Initial $T_A = 51°.25$, $T_B = 52°.48$; Final $T_A = 58°.10$, $T_B = 59°.61$. | Diminutions of difference $T_B - T_A$ (in hundredths of a degree Cent.) during periods | |-----------------------------------------------| | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | First half-minute of current entering by end next | | Second half-minute of current entering by end next | | Third half-minute of current entering by end next | | Fourth half-minute of current entering by end next | | Fifth half-minute of current entering by end next | | Sixth half-minute of current entering by end next | | Seventh half-minute of current entering by end next | | Eighth half-minute of current entering by end next | | Ninth half-minute of current entering by end next | | Tenth half-minute of current entering by end next | | Eleventh half-minute of current entering by end next | | Twelfth half-minute of current entering by end next | Differences of diminution during corresponding half-minutes of first and second halves of period, summed for six periods. Differences of diminution in equal intervals from beginning and middle of a period, summed for six periods. Mean augmentation of difference in favour of thermometer remote from entering current after reversal and flow for: - $\frac{1}{4}$ min. - 1 min. - 1$\frac{1}{2}$ min. - 2 min. - 2$\frac{1}{2}$ min. - 3 min. - 4 min. - 4$\frac{1}{2}$ min. - 5 min. - 5$\frac{1}{2}$ min. - 6 min. Values: - 0.015 - 0.0283 - 0.06083 - 0.0825 - 0.09416 - 0.0983 - 0.107 - 0.105 - 0.11175 - 0.1291 ### Table VI. March 7th, 1854. Copper conductor to two slips, and diminished in breadth to 1½ inch. Water kept at temperature 99° Cent. in heater. Streams of water about temperature 6° through coolers. Temperatures at middle points A, B of the parts of the conductor between heater and coolers: Initial $T_A = 50°33'$, $T_B = 50°70'$; Final $T_A = 61°07'$, $T_B = 62°62'$. | Diminutions of difference $T_A - T_B$ (in hundredths of a degree Cent.) during periods | |-----------------------------------------------| | I. | | II. | | III. | | IV. | | V. | | VI. | | VII. | | First half-minute of current entering by end next | |--------------------------------------------------| | A. | | B. | | Second half-minute of current entering by end next | |---------------------------------------------------| | A. | | B. | | Third half-minute of current entering by end next | |--------------------------------------------------| | A. | | B. | | Fourth half-minute of current entering by end next | |---------------------------------------------------| | A. | | B. | | Fifth half-minute of current entering by end next | |--------------------------------------------------| | A. | | B. | | Sixth half-minute of current entering by end next | |--------------------------------------------------| | A. | | B. | | Seventh half-minute of current entering by end next | |----------------------------------------------------| | A. | | B. | | Eighth half-minute of current entering by end next | |----------------------------------------------------| | A. | | B. | | Ninth half-minute of current entering by end next | |---------------------------------------------------| | A. | | B. | | Tenth half-minute of current entering by end next | |---------------------------------------------------| | A. | | B. | | Eleventh half-minute of current entering by end next | |------------------------------------------------------| | A. | | B. | | Twelfth half-minute of current entering by end next | |-----------------------------------------------------| | A. | | B. | | Mean augmentation of difference in favour of thermometer remote from entering current | |---------------------------------------------------------------------------------------| | after reversal and flow for 1 min. | | 1 min. | | 1¼ min. | | 2 min. | | 2¼ min. | | 3 min. | | ¾ min. | | 4 min. | | 4¼ min. | | 5 min. | | 5¼ min. | | 6 min. | Table VII. April 19th, 1854. Copper conductor two slips further diminished in breadth. Water kept at temperature 99° Cent. in heater. Streams of water at about temperature 6° through coolers. Temperatures at middle points A, B of the parts of the conductor between heater and coolers: Initial $T_A = 63°65$, $T_B = 63°81$; Final $T_A = 72°90$, $T_B = 75°05$. | Periods | Diminutions of difference $T_B - T_A$ (in hundredths of a degree Cent.), during | |---------|----------------------------------------------------------------------------------| | I | | | II | | | III | | | IV | | | V | | | VI | | | VII | | Diminutions during half-minutes, summed for six periods | Diminutions during half-minutes, summed for five periods | Differences of diminution during corresponding half-minutes of first and second halves of a period, summed for six periods | Differences of diminution in equal intervals from beginning and middle of a period, summed for five periods | Mean augmentation of difference in favour of thermometer remote from entering current... after reversal and flow for... | |----------------------------------------------------------|-----------------------------------------------------------------------------------------------------------------|---------------------------------------------------------------------------------------------------------------|----------------------------------------------------------------------------------------------------------------| | $-4$ $-1$ $9$ $29$ $-9$ $27$ $-3$ $9$ $-9$ $4$ $-18$ $5$ $3$ $-6$ $-21$ $-5$ $-8$ $-3$ $1$ $-2$ $-1$ $-3$ $2$ $-4$ | $7$ $26$ $38$ $16$ $15$ $27$ $-18$ $28$ $-64$ $-3$ $-8$ $41$ $-1$ $-2$ $-1$ $-3$ $2$ $-4$ | $5$ $35$ $61$ $73$ $86$ $100$ $100$ $116$ $118$ $116$ $110$ $-9$ $16$ $5$ $-3$ $-2$ $-6$ | $°005$ $°025$ $°091$ $°073$ $°086$ $°109$ $°100$ $°116$ $°121$ $°118$ $°116$ $°110$ | Periods I. and VII. cf. | Diminutions during half-minutes, summed for five periods | Differences of diminution during corresponding half-minutes of first and second halves of a period, summed for five periods | Differences of diminution in equal intervals from beginning and middle of a period, summed for five periods | Mean augmentation of difference in favour of thermometer remote from entering current... after reversal and flow for... | |----------------------------------------------------------|-----------------------------------------------------------------------------------------------------------------|---------------------------------------------------------------------------------------------------------------|----------------------------------------------------------------------------------------------------------------| | $-4$ $-1$ $9$ $29$ $-9$ $27$ $-3$ $9$ $-9$ $4$ $-18$ $5$ $3$ $-6$ $-21$ $-5$ $-8$ $-3$ $1$ $-2$ $-1$ $-3$ $2$ $-4$ | $7$ $26$ $38$ $16$ $15$ $27$ $-18$ $28$ $-64$ $-3$ $-8$ $41$ $-1$ $-2$ $-1$ $-3$ $2$ $-4$ | $5$ $35$ $61$ $73$ $86$ $100$ $100$ $116$ $118$ $116$ $110$ $-9$ $16$ $5$ $-3$ $-2$ $-6$ | $°005$ $°025$ $°091$ $°073$ $°086$ $°109$ $°100$ $°116$ $°121$ $°118$ $°116$ $°110$ | reducing the conductor so much that the current through it would generate heat enough to keep up the required difference of temperatures without any external heater. 59. The new conductor was therefore made of just two slips of sheet iron broad enough to admit the whole length of the thermometer-bulbs in the same manner as in the conductor previously used; these slips were bent in the places for the thermometer-bulbs, but were kept straight and bound close together elsewhere. Gutta-percha pipes were cut and cemented upon the iron slips near their ends, so as to lead streams of cold water across them. The part of the conductor between these coolers was packed round with a large mass of cotton wool, the thermometer-bulbs being steadied in the apertures prepared for them by means of corks, as before (§ 31). The breadth of the conductor was $2\frac{1}{2}$ inches, the length between the coolers only $3\frac{1}{2}$ inches (instead of 10 inches, as in the iron conductors used previously), so that too great a time might not elapse before such a nearly permanent state of temperature as depended on the heating effect of the current would be reached. 60. On the 25th of March, 1854, an experiment was made with this conductor in the following manner:—A constant stream of cold water was maintained through each of the coolers; a current from the full nitric acid battery of eight large iron cells was sent through the conductor for twelve times four minutes in each direction, that is for ninety-six minutes in all, and the thermometers were noted every half-minute. The actual observations of temperature are required to show the circumstances of this experiment, and I therefore give them as follows; instead of an analytical table, such as those by which the results of the preceding experiments were exhibited:— | Time in minutes and half-minutes from commencement of current. | Current. | |---|---| | Started. | Change. | | Entered by end next B. | Entered by end next A. | | Entered by end next B. | Entered by end next A. | | Entered by end next B. | Entered by end next A. | | Entered by end next B. | Entered by end next A. | | Entered by end next B. | Entered by end next A. | | Entered by end next B. | Entered by end next A. | | Entered by end next B. | Entered by end next A. | | Entered by end next B. | Entered by end next A. | | Entered by end next B. | Entered by end next A. | Observing temperature at A by Kew, No. 114, $T_A$. Observing temperature at B by Kew, No. 91, $T_B$. Difference of observed temperatures $T_A - T_B = D$. The differences of temperatures here tabulated, and the half sums of the same temperatures, are graphically represented in Plate XXXV. The result is obvious, either with or without the graphical representation, and affords a striking confirmation of the conclusion first arrived at by so different an apparatus (§ 31), that the Resinous Electricity carries heat with it in iron. 62. About the same time another form of the experiment was tried on a copper tube, with a vessel of oil fitted round it in the middle, and kept hot by a lamp below it, and with gutta-percha tubes fitted to conduct streams of cold water round it. A current from the battery was sent alternately in the two directions through it, as in the previous experiments, and it was attempted to observe the thermal effects by means of two open thermometer-tubes with small spherical bulbs, pushed into the copper tube from each end, and bent down at right angles outside it, with their lower ends immersed in two cups of spirits of wine. The want of any sufficient regulation of temperature to keep the liquid column of these air-thermometers within range, made it impossible to get any clear indication of a result by this experiment; but on the whole, there appeared to be an effect of the same kind as had been previously discovered in copper. 63. A few weeks ago, I began again to make direct experiments on electrical convection with a view to obtaining additional evidence in support of the conclusion which I had arrived at previously, and to investigate methods by which the nature of the quality in other metals could be discovered more readily, and the specific heat of electricity in any metal determined in absolute units. I had determined to give up the use of the nitric acid battery in consequence of the inconveniences which had been alluded to above (§ 34), and accordingly I had constructed a large Daniell's battery: consisting altogether of eight wooden cells lined with gutta percha, and fitted with sheet copper, suitably arranged with shelves to bear crystals of sulphate of copper; sixteen porous cells, some of which had served previously in the iron battery; and sixteen zinc plates of the same dimensions as those previously used. Each wooden cell had sheet copper not only round its interior, but also a portion of the same sheet carried across it so as to divide it into two spaces, each completely surrounded by the metallic surface. A porous cell is put into each of these spaces, and a zinc plate into each porous cell; the two zines in the porous cells contained in the same wooden cell being always united. The ordinary liquids of a Daniell's battery, acidulated solution of sulphate of copper and dilute sulphuric acid, are used. The whole battery power thus consists of eight independent cells, which, with the connexions in ordinary use, may be arranged either in one or in two elements, but which may also, should there be occasion, be readily enough set up in four or in eight elements. Any power may of course be used down to the lowest, with only a single porous cell and a single zinc plate in one of the wooden cells. The sulphate of copper solution is kept constantly in the wooden cells, which remain in a fixed position on a shelf. Electrodes from the large commutator (§ 27), which is fixed to the wall in an adjoining apartment as near as possible to the middle of the wooden cells, are brought through the partition between the two rooms, and kept always ready to be put in communication with the two poles of the battery, however arranged. This battery, or parts of it, have been used in nearly all the experiments described below in Parts IV. and V., and it has been found very convenient. Some of the wooden cells have contained the acidulated solution of sulphate of copper now for more than a year [for more than two years now, Nov. 1856], and as yet their gutta-percha linings have shown no signs of injury. 64. The first of the recent experiments on electrical convection was made with an iron conductor prepared as follows: The conductor, XY, consists of two pieces of thin sheet iron $8\frac{1}{2}$ inches long and $\frac{3}{4}$ of an inch broad, and bent so that when put together they form three tubular spaces, A, B, C, fig. 6. The iron is cut so as to make prolongations of these tubes of about an inch beyond one side of the conductor. The slips thus put together are soldered so as to make the tubes perfectly air-tight, one end being closed, and the other left open to receive the thermometer-tubes $a$, $b$, $c$, which were cemented air-tight with wax. In soldering, great care was taken to prevent the solder from spreading between the iron slips. Copper electrodes were now soldered to the ends of the conductor, and the junctions were enclosed within pieces of gutta-percha tube, $g$, $h$, through which a continuous stream of cold water was made to flow. The distance between the coolers was $7\frac{1}{2}$ inches, and they were placed so that the four spaces between them and A, B, C were all equal. Divided scales were attached to the tubes, of which the lower ends were immersed in small vessels, $k$, $l$, $m$, containing spirits of wine. The conductor between the coolers was wrapped in a large quantity of cotton wool represented by the space within the dotted line. To send a current through the conductor thus prepared, the whole battery, arranged, as described below, in two elements, each exposing ten square feet of zinc surface to seventeen square feet of copper, was employed: **Description and Drawing of Battery with Connexions.** R, R and S, S, two series of cells, each containing eight porous cells and eight zinc plates. K, K and D, D, thick copper supports for the zinc plates, the zinics of R, R being firmly clamped to K, K, and those of S, S to D, D. E, E, a thick conductor connecting the coppers of series R, R together. F, F, a similar conductor connecting the coppers of S, S. H, H, a bundle of wires connecting the coppers of S, S with the zinics of R, R. M, M, a wooden partition separating the battery-room from the experimenting-room. L and M are two bundles of wire which pass through holes in the partition and connect the commutator G respectively with the coppers of series R, R, and the zinics of series S, S. The bundles of wires O and P complete the circuit through XY, the conductor to be tested. The dotted spaces round the porous cells represent shelves for holding crystals or sulphate of copper. 65. After about an hour and a half, the thermometer at the middle of the conductor indicated 170° Fahr. (76°.7 Cent.) ; and one of the brass bridges of the commutator was then lifted so as to break the circuit. Immediately the liquid mounted rapidly in each of the three glass tubes of the air-thermometers, and it was prevented from rising above a certain point in the middle one by completing the circuit again. The column of liquid was kept as steady as possible at this point in the middle air-thermometer by a person observing it, and making and breaking the circuit by means of the brass bridge, while two other persons noted the indications of the two lateral air-thermometers. The current was reversed every three or four minutes, and the liquid in the middle air-thermometer brought back to the same point, and kept as nearly as possible to it. The imperfection of the regulating system was such as to make it very difficult to prevent great oscillations in the thermometers, but the instantaneous manner in which their indication followed the operations of the break made it certain that the plan would be perfectly successful when a continuously acting regulator should be introduced. 66. As it was, the result afforded a most striking and immediate confirmation of the conclusion previously arrived at regarding the electrical convection of heat in iron. Every time the current was reversed, the liquid fell rapidly (showing a rise of temperature) in the thermometer next the end, by which the current nominally entered, and rose rapidly (showing a fall of temperature) in the other. Mr. Joule assisted in this experiment, and was satisfied with the evidence it afforded in favour of the conclusion that the Resinous Electricity carries heat with it in iron. 67. Unsuccessful attempts were next made with tubular conductors of different metals; and in endeavouring to get decisive results regarding the qualities of copper and brass, I again had recourse to the form of conductor used in the preceding experiment. The new conductors were, however, made of much thinner sheet metal than those of the iron, to admit of a less powerful battery being used; and consequently, in each case, a frame-work had to be arranged to hold the conductor steady. Great difficulties were met with in continually repeated failures of the air-thermometers. It was therefore found necessary to have metallic tubes continued downwards several inches from the bulbs, so as to prevent the wax by which the glass was cemented from being melted by the heat. The battery, however, had also to be reduced to a single zinc plate in one of the wooden cells, as with more of the battery than this, the heating action had been found to be so sudden in the thin copper and brass conductors, as almost immediately to melt the solder about some of the bulbs, and so make one or more of the thermometers fail before the regulating action of the break was applied. Notwithstanding all precautions, the central thermometer failed in each case, and the action of the lateral thermometers was very unsatisfactory both in the copper and in the brass conductor. The central thermometer could, however, be well dispensed with by regulating by the break one or other of the lateral thermometers; and thus, after many unsuccessful attempts, experiments were made on copper and brass conductors, which, although still unsatisfactory, showed decidedly the looked-for convective effect. In each case, the thermometer which was not kept to one point by the regulator, always showed an increase of temperature, both in the copper and in the brass conductor, when the current was reversed so as to enter by the end remote from it, and showed a diminution of temperature when the current was again reversed so as to enter by the end next it. Hence it appeared that the Vitreous Electricity carried heat with it in both copper and brass. 68. The lateral metallic tubes branching down from the conductor to carry the glass tubes of the air-thermometer, constituted a great defect in the plan of apparatus used in the experiments just described; and the only way of avoiding it appearing to be to make the glass tubes pass through the body of the conductor itself, so as to admit of their being cemented air-tight at its cool ends, I again had recourse to the tubular form of conductor which had been tried unsuccessfully before. 69. A tube made of very thin sheet platinum, soldered with gold, was arranged in the following manner:—A glass rod, $2\frac{1}{8}$ inches long, wrapped closely round with thin cotton-thread, was pushed into the central part of the tube, in which it fitted closely, and was carefully luted with red lead. After keeping it for several days heated by a stove, gutta-percha coolers, A, A, were fitted on it, leaving a length of 6 inches of tube between them. Wooden troughs, B, B, were then fitted on outside the coolers, and fastened to the ends of a piece of wood, C, C; straps of thick copper, about an inch broad, were bent to form conducting linings for the troughs, their ends turned round, firmly fastened to C, C, and brought together at D, D, thus forming connexions with the electrodes of the commutator (for this part of the arrangement see also fig. 9). Two pieces of thermometer-tube, bent to right angles, had their short arms rolled with thread, and were pushed into the tube from its ends, as far as b, b, leaving spaces ab, ab, each two-thirds of an inch, between them and the stopper aa in the middle of the tube, and made air-tight by cement applied at E, E. The dotted line represents the space round the tube and its wooden stand C, C, filled with cotton-wool. A conducting communication was established between the platinum tube and D, D, by pouring mercury into the troughs (see also fig. 9). 70. The system of regulating the temperature in one part of the conductor by breaking and making the circuit, had been adopted only as a temporary expedient in the experiment on the iron conductor (§ 65), in consequence of the failure of a continuous regulator which had been fitted up for that experiment. It had the advantage of requiring no other apparatus than the commutator, in regular use in all applications of the battery, and it had been found to answer the purpose tolerably well in the first trial. It proved, however, very inconvenient with the finer conductors, from the too great abruptness of its action. Besides, it was open to this very serious objection, that it kept up the required heating effect by an intermittent current, and therefore by the passage of a much less quantity of electricity than would be required to produce the same heating effect if flowing in a nearly constant current (the rate of generation of heat being proportional to the square of the strength of the current at each instant, while the looked-for convective effect is proportional simply to the strength of the current at each instant, and is therefore, on the whole, proportional to the whole quantity of electricity that passes). In order, therefore, that the current might be kept as nearly as possible constant at the particular strength required to maintain the heating effect used, I had the following regulator constructed. 71. Two iron tubes, AB, CD, 20 inches long and \( \frac{3}{4} \)ths of an inch in diameter, open at the top but closed at the bottom, are bound firmly together with insulating blocks of wood, AC and BD, so as to be parallel to one another. Pieces of thin sheet copper are bent into cylinders; to their tops pieces of thick copper, E, E, are soldered, and the copper cylinders are put into the iron tubes. To each end of a piece of thick copper wire, shown separately at F, two pieces of No. 18 iron wire are fixed, one of the same length as the iron tubes, and the other less than half that length, and the two branches are parallel, and at such a distance that when their ends are introduced into the two tubes, they move along their axes. To use the regulator, the tubes are filled with mercury, the apparatus is put into the circuit by connecting with EE, and the requisite amount of resistance is introduced by raising G, which is kept in any position by having one end of a cord fixed to its upper part, carried over a pulley, and stretched by a counterpoise hung at its other end. [Great improvement has been since made in the regulator, by using, instead of No. 18 iron wire, thick copper wire tapering to points at the lower ends; and by attaching cups of gutta percha to the tops of the iron tubes allowed to communicate with the interior by small holes, to serve as overflow cisterns for the mercury. By this arrangement the tubes were kept always full of mercury, and irregular contacts between the connecting conductor and the interior of empty parts of the tubes were prevented.—Nov. 1856.] 72. The apparatus was set up as shown in the accompanying view. The battery connexions were completed with the regulating break partly up, so as to check the current somewhat, and prevent injury from sudden overheating in any part of the conductor. After a few minutes the break was raised further so as to reduce the current very much, and the liquid began to rise in the stem of each of the glass tubes, showing that both air-thermometers at first acted perfectly. One of the thermometers was then steadied with great ease to a small fraction of a scale-division by using the regulator. The liquid in the other thermometer was observed, and its position occasionally noted. The direction of the current was reversed every few minutes, as before, by means of the ordinary commutator. Fig. 11. 73. Slight differences were observed in the free thermometer after the reversals, but as yet no very decisive indications of the looked-for effect appeared. The mercurial thermometer beside the central conductor indicated less than $80^\circ$ Fahr. ($27^\circ$ Cent.), its column of mercury having not yet become visible, after the experiment had been continued in this way for several reversals. 74. The regulating break was then pushed down until a somewhat further elevation in the temperature of the platinum was indicated by a considerable escape of air in bubbles from the open ends of the thermometer-tube. The break was again drawn up until the liquids again mounted in the stems. One of the thermometers was again steadied by the regulator, and, the other being observed, the experiment was continued as before. A decided effect now appeared almost immediately after each reversal. The free thermometer regularly indicated a higher temperature when the current nominally entered by the end next it, and a lower temperature when the current nominally entered by the remote end. After four reversals this part of the experiment had lasted about twenty minutes, and the mercury thermometer beside its middle showed $104^\circ$ Fahr. ($40^\circ$ Cent.). 75. The break was again pushed down for some time, and again raised till the liquid rose in each thermometer-tube, and the experiment was continued as before. for four reversals, the central mercury thermometer rising to about 150° Fahr. (66° Cent.). The free thermometer rose and fell alternately through several scale-divisions almost immediately after each reversal, and showed the same convective effect as had previously been observed by smaller indications. 76. The bridge was again pushed down and air again escaped copiously from the thermometers, but very soon beads of liquid began to appear following one another rapidly down the capillary tubes from the interior of the conductor. As the spirits of wine had not once been allowed to run up into the bulb of either thermometer, these beads of liquid could be nothing but products of the distillation of the oil which had been used in the luting of the central plug; and on taking away the cups of spirits of wine from below the tubes, the smell and taste of the small quantities of liquid which continued to descend gave unmistakeable evidence of their origin. After this it was scarcely possible to get any satisfactory indication from either of the air-thermometers; but the experiment was continued, and one or other of them, when by any means the beads of disturbing liquid could be sufficiently got rid of for a time, was steadied to a constant temperature; the other thermometer being observed when possible, and the reversals repeated as before. The same result was still obtained; and on the whole, notwithstanding the defect which caused so much inconvenience, it was very decidedly established by the experiment that the Resinous Electricity carries heat with it in platinum. 77. [Added Dec. 1856.]—After many unsuccessful trials on short brass tubes, first with air-thermometers of the metal itself and capillary glass tubes arranged as in the platinum tube (§ 69), and latterly with glass air-thermometers (§ 62) having very small cylindrical bulbs, the following conclusive experiment was made a few days ago. Four of the large double cells, connected to form a single Daniell's element, exposing 10 square feet of zinc to 17 square feet of copper, were used to send a current through a piece of brass telescope tube 6 inches long, \( \frac{1}{4} \) of an inch diameter, and ground as thin as it could be without breaking it up by emery-paper, over the length of \( 3\frac{1}{2} \) inches which was left between the near sides of gutta-percha coolers, fitted to it in the manner represented above (see fig. 11, § 72). Streams of water being, as in other experiments, kept running through the coolers, and the regulating break (§ 71) being used to keep the liquids within range in the tubes of the air-thermometers, a small mercurial thermometer pressed against the middle of the brass tube, with its stem and scale projecting out through the cotton wool, indicated from 190° to 195° Fahr. (90°6 Cent.). The regulator was not used so much as it might have been with advantage; but, notwithstanding great unsteadiness in the indications of the two air-thermometers, the observations showed decidedly, after each reversal of the current, a cooling effect on the thermometer next the entering stream, in every case in which the irregularities were not so great as to make a comparison impossible. This effect is manifest from the following four cases, selected merely as being those in which one of the thermometers was most nearly steady during a few minutes of flow of the current, first in one direction and then in the other. | Current entering by end next | Readings, in arbitrary scale-divisions, of Thermometer A. | Thermometer B. | |-----------------------------|----------------------------------------------------------|---------------| | I. {A | 43 | 57½ | | B | 42 | 44½ | | II. {A | 41 | 41 | | B | 41 | 34½ | | III. {A | 31½ | 26½ | | B | 29½ | 16½ | | IV. {A | 27½ | 22 | | B | 20½ | 12½ | Hence the conclusion (see below, §§ 102 and 103), that the Vitreous Electricity carries heat with it in brass, which I anticipated three years ago from the mechanical theory*, is now established by a direct experimental demonstration. PART II. ON THERMO-ELECTRIC INVERSIONS. 78. Cumming's discovery of thermo-electric inversions having afforded the special foundation of that part of the theory by which I ascertained the general fact of electric convection in metals, and every observation of a thermo-electric inversion being a perfect test as to the relative positions of the two metals between which it is observed in the Table of Convections (see below, § 103), I was induced to make experiments with a view to finding new instances of inversion, and to determine in each case, with some degree of precision, the temperature at which the two metals are thermoelectrically neutral to one another. 79. In the experiments on thermo-electric inversion described by Cumming, and by Becquerel, the only other experimenter, so far as I am aware, who has published observations on the subject, one junction between the two metals is generally kept cool, while the other is raised until the current indicated by the galvanometer, instead of going on increasing, begins to diminish, comes to a stop, and then sets in the reverse direction†. 80. In this way Cumming found that "if gold, silver, copper, brass, or zinc wires be heated in connexion with iron, the deviation [indicating the current], which is at first positive, becomes negative at a red heat‡." Many other experimenters have professed themselves unable to verify these extraordinary results, and have attempted * See "Dynamical Theory of Heat," § 132. † Cumming's 'Electro-Dynamics,' section 104, p. 193. Cambridge, 1827. ‡ Cambridge Philosophical Transactions, 1823, addition to p. 61. to explain them away by attributing them to coatings of oxide formed on the metals, or to other causes supposed with equally little reason to exercise sensible disturbing influences; but the descriptions, given by the original observers, of their experiments leave no room for such doubts. It is certainly not easy to get the inversion between copper and iron (with such specimens as I have tried) by the heat of a spirit-lamp, applied as described by Becquerel to one junction while the other is left cool; but I readily obtained it by raising the other junction somewhat in temperature, with the first still kept at a red heat. Probably if the atmospheric temperature had been higher, or if a somewhat more intense red heat had been obtained from the spirit-lamp, I should at once have obtained the result simply in the manner described by the previous observers. 81. The easiest way to verify the thermo-electric inversion of iron and copper is to take a piece of iron wire a foot or two long, and twist firmly round its ends two copper wires connected with the electrodes of any ordinary astatic needle-galvanometer. Then first heat one of the junctions with the hand, or by holding it at some height over a flame; and note the deflection, which will be found such as to indicate a current from copper to iron through the hot junction. Again, heat both junctions in flame, or in sand at any temperature above 300° Cent., and withdraw one a little from the hottest place, so that while both junctions are at temperatures above 300° Cent., that which was heated in the first experiment may be still decidedly hotter than the other. The deflection will now be found to be the reverse of what it was before, and will be such as to indicate a current from iron to copper through hot. The reversal of the current may be very strikingly exhibited by allowing the two junctions gradually to cool, while ensuring that the same one remains always somewhat above the other in temperature. When the mean of the temperatures of the two junctions falls below 280° Cent. or thereabouts, the primitive deflection will be again observed. All these phenomena are observed indifferently whether the copper wires be simply twisted on the ends of the iron wire, or brazed to them, or tied to them by thin platinum or iron wire. 82. Similar phenomena may be observed without the necessity of going to so high temperatures, by soldering galvanometer electrodes of copper to the ends of a double platinum and iron wire, and treating this compound circuit in the manner just described, only with a more moderate application of heat. If the platinum wire be very thin in comparison with the iron one connected along with it, the circumstances will be but little altered from those observed when iron simply is used. By taking a thicker platinum wire, or several thin ones together, in connexion with the same iron wire, or by using a thinner iron wire and the same platinum, the neutralization and reversal may be shown with temperatures below the boiling-point. Most specimens of platinum wire thus applied reduce the neutral point of copper and the compound platinum and iron MDCCCLVI. wire much below the temperature of melting ice, when the proportion of platinum to iron in the bundle is sufficiently increased (the limit, of course, being the neutral point of copper to the platinum itself. See below, §§ 83, 84). 83. A certain specimen of platinum wire in my possession, when tested by such elevations of temperature as could be produced by the hand, was found to lie in the thermo-electric series, on the other side of copper from the position in which platinum is placed in all statements of the thermo-electric qualities of metals previously published. That is to say, when connected by copper electrodes with the circuit of a galvanometer, and when heated at one junction up to ten or twenty degrees above the atmospheric temperature, a current set from copper to platinum through hot. On raising the temperature of the hot junction towards the boiling-point of water, the strength of the current began to diminish; came to a stop when a temperature I suppose little above that of boiling water was reached; and set in the reverse direction with increasing strength when the temperature of the hot junction was further raised, the other junction being kept all the time at the atmospheric temperature. I afterwards found that this specimen of platinum wire (referred to under the designation $P_1$ in what follows) became neutral to ordinary copper wire at the temperature 64° Cent. 84. Of two other specimens of platinum wire which I tried with copper, one (marked $P_2$) gave indications of a neutral point about the zero of Fahrenheit's scale, but the other ($P_3$) remained, for the lowest temperatures I reached, always on the same side of copper as that on which platinum appears, at ordinary and at high temperatures, generally to lie. When these three platinum wires were tried with one another thermo-electrically, they gave, as was to be expected, the mutual thermo-electric indications of different metals lying in the order Bismuth, $P_3$, $P_2$, $P_1$, Iron, Antimony. They retained all the same qualities after being heated to redness; and in a great many experiments performed upon them, in which I have found them extremely convenient as thermo-electric standards, have exhibited perfect constancy in their thermo-electric bearings. I have not yet discovered on what their differences depend, but in all probability it is on the degrees to which they are alloyed with other metals. 85. The fact of copper changing in the thermo-electric series from below the position of the platinum specimen $P_2$ to above that of iron, when the temperature is raised from −30° or −20° Cent. to 300°, proves that every metal which lies between $P_2$ and iron for any intermediate temperature, must become neutral to either $P_2$ or copper, or iron at some temperature between these limits. Now nearly all the common metals, for instance, lead, tin, brass, zinc, silver, cadmium, gold, lie between platinum and iron in the thermo-electric series at ordinary temperatures, and no doubt many of the rarer metals (I have found aluminium to lie between $P_3$ and $P_2$) are to be ranked within the same limits. Hence at temperatures easily reached and tested, neutral points may be looked for with the certainty of finding them, between each of those metals and one or other, if not several, of the metals and metallic specimens (P₂, P₁, Copper, Iron) referred to above. Taking then the platinum specimens P₁, P₂, P₃ as standards, and using besides ordinary copper and iron wires, I commenced investigating their thermo-electric relations to as many other metals as I could obtain. 86. In experiments to determine temperatures of neutrality, the first apparatus which I employed for regulating the temperatures of the two junctions, consisted of copper vessels placed side by side in which oil could be raised by gas-burners as high in temperature as the mercurial thermometer can be used, that is to 340° or 350° Cent., or somewhat above the boiling-point of mercury. To do away with irregularities from the flame and cold air playing unequally on the sides of these vessels, smaller ones were placed on wire stands within them, and were completely filled and surrounded with oil. In each experiment a wire or slip, about 18 inches long, of one of the metals to be tested, had somewhat longer wires or slips of the other soldered to its ends. The compound conductor thus constituted was bent into such a shape that the two junctions of the metals could be placed near the centres of the oil-baths; it was supported in this position, carefully insulated from touching the copper vessels and from all other metallic contacts; and thermometers were put with their bulbs in the oil as close to the junctions as possible. The gas-furnaces were applied below and round the sides of the large copper vessels, so that they could be regulated to any desired temperatures. 87. After this apparatus had been used in several experiments, and neutral points between copper and iron, copper and P₁, lead wires and P₁, and brass and P₁ had been determined, I saw reason to alter the arrangements in various respects, and had another apparatus constructed, according to the following description. 88. Two small oil-baths were made, each of an outside partly cylindrical and partly plane sheet of copper, and a concentric copper tube 5 inches long and \( \frac{6}{10} \)ths of an inch diameter brazed to it by ends of sheet copper, shaped as shown in the diagram. The space round the inner tube and within the outer sheet and ends was filled with oil completely covering the inner tube, and, when heated, rising into the space between the upper parallel plane parts of the outer sheet. A narrow ring of sheet metal with a long slip projecting from one side for holding it by, was put in the inner tube before the other parts were brazed on, and during an experiment was kept as constantly moving from one end of the bath to the other and back as was required to keep the whole mass of oil at one temperature. The second drawing represents, on the actual scale, a section of either bath through the position occupied by its stirrer. This diagram also shows a section of an external case of sheet metal which supports the bath, and serves as a flue to carry the flame and products of combustion round its sides. The rows of gas-burners for the two baths were fixed in a line, and each burner regulated by a separate stopcock. The outer cases are screwed to the same stand, and the copper vessels holding the oil are pushed into them and rest with their axes in a line over the burners. The ends of the baths and of the outer cases are kept about one-fourth of an inch apart, and their supports are also made quite separate, which was found to be necessary to allow one of the baths to be kept cool, while the other is raised to a high temperature. (In the third drawing, the stems by which the stirrers are held are accidentally omitted.) 89. When the baths and their furnaces are all fixed in their proper positions, a tube of thin glass, about $10\frac{1}{2}$ inches long, just small enough to enter easily, is pushed into the inner tubes of the baths, and is left resting there, with its ends projecting a little outside their remote ends. In recent experiments I have substituted a simple roll of paper for the glass tube, and have found it answer quite as well. 90. A compound conductor, to be tested thermo-electrically in this apparatus, con- sists of a wire or a thin bar of one metal, or a bundle of wires of two different metals 5½ inches long, with wires of from 18 to 30 inches long of another metal soldered to its ends. To avoid circumlocution I shall call the former the mean conductor, and the wires soldered to it the electrodes, of the thermo-electric arrangement. The connexions between the mean and the electrodes are generally made by brazing, or by hard silver solder, when temperatures much above the boiling-point of water are to be used. 91. A conductor thus prepared of two metals to be tested is drawn through the glass tube till the mean occupies a position, lying on the glass or paper tube, with its centre under the centre of the tube, and consequently with its ends about the middle of the hollow spaces surrounded by the oil-baths. 92. The electrodes are carried from the ends of the insulating tube to the con- nexions required for completing the circuit through the coil of a galvanometer. These must essentially be maintained at the same temperature, unless the electrodes of the thermo-electric arrangement be copper, the same as those of the galvanometer. After trying several obvious, more or less troublesome plans to secure the fulfilment of this condition, I found a perfectly effective way simply to tie the connexions firmly together as close to one another as possible, only separated from contact by a fold or two of paper wrapped round each, and to tie a quantity of paper, or to make up a bundle of cotton wool, or some other bad conductor, round the two, for two or three inches on each side of the junctions. The junctions themselves, except when they are between homogeneous metals, are not made by binding-screws, but either by soldering, or by cleaning the surfaces and then tying the metal firmly together by fine twine. To avoid mistakes and prevent the necessity of disturbing the bundle round the junctions, in tracing the courses of the conductor on the two sides of it, a thread or mark of some kind is attached to one galvanometer electrode, and a corresponding mark on the electrode of the thermo-electric apparatus to which it is joined. This system of electric insulation and thermal connexion between junctions of dissimilar metals, I have found very convenient in a great variety of thermo-electric and other electro-dynamic experiments, and when it was used I have never observed the slightest trace of a current attributable to any difference of tem- peratures in the parts of the circuit to which it is applied. 93. The conductor being thus arranged, two thermometers are pushed into the glass or paper tube from its ends and placed with the centres of their bulbs as close as possible to the metallic junctions, and with their graduated tubes extending nearly horizontally outside the apparatus, but inclined upwards as much as the inner dia- meter of the insulating tube and their dimensions permit, so as to check as much as possible the tendency (in some of the thermometers found very inconvenient) of the column of mercury to divide when sinking rapidly. All the space inside the glass or paper tube left vacant by the thermometers and the conductor is filled with cotton wool, well pressed in to prevent currents of air. 94. This apparatus has many advantages over that first used and described in § 86 above: the temperatures of the baths can be changed with great rapidity, in consequence of the smallness of the quantities of oil which they contain; and by watching the thermometers and adjusting the gas-burners, can be regulated as desired with great ease. I have found it not a small practical advantage to be freed from the necessity of bending the mean part of the conductor to be tested, and of making the arrangements to prevent irregular contacts and to keep the junctions and the thermometers in their proper positions immersed in the oil. When a rare metal is to be tested, or one, such as sodium or potassium, which cannot be kept in air, it will be of great consequence to be able to apply the tests to a little straight bar or slip only a few inches long, or to a small column filling a glass tube. 95. For experimenting at low temperatures a modified apparatus was made, consisting of a double wooden box, each compartment, nearly a cube of 4 inches side, fixed to a common base with a space of about \( \frac{1}{4} \) inch between their sides, and a glass tube running through them and cemented at the apertures in the sides so as to hold water-tight and resist the action of acids which might be employed in freezing mixtures. The conductor to be tested and the thermometers are arranged in this glass tube, as in that of the other apparatus (§ 89 to 93); and while a freezing mixture is kept in one compartment, the other is either allowed to take the atmospheric temperature, or is heated by hot water or steam. 96. The way of experimenting which I followed, was to raise the temperature of one bath until a deflection of the galvanometer-needle became sensible; then to go on raising it, and letting that of the other follow, so that the two thermometers may indicate as nearly as may be a constant difference of temperatures; and to watch the needle until a reversal is observed, or until the limit of temperature which the arrangement admits of is reached. As soon as a reversal is obtained, the two thermometers are allowed to sink until the needle begins to return from its reverse deflection. When it approaches zero the thermometers are kept from any rapid changes, but allowed to sink very slowly, with always the same difference, or at least with a quite decided difference of the same kind as that raised between them at the beginning. The last readings of the sinking thermometers which give a sensible deflection before the original deflection is recovered, several readings when the needle appears perfectly at zero, and the first readings when the needle is discovered to deviate again in the original direction, are carefully noted. The arithmetical mean of the temperatures of the two thermometers for each of these simultaneous or nearly simultaneous readings is taken; and it is generally found that the means derived from the readings taken when no deflection can be discerned, lie within a fraction of a degree of the mean of the last sinking mean temperature of the junctions which show one deviation, and the first which shows the deviation in the other direction. The mean of either the readings which give no deviation, or of the last and first which give the contrary deviations, or of all these readings together, according to the nature of the memoranda made by the observer, is taken as a determination of the neutral point of the two metals, that is, the temperature at which they are thermo-electrically like one metal, or thermo-electrically neutral to one another. In the course of one experiment several such determinations, both with descending and with ascending mean temperature, are made, and if possible also, with first one and then the other junction higher. 97. Either in one experiment, or with the same apparatus on successive days, determinations are sometimes made with as considerable a variety of differences of temperature between the two junctions as is attainable. Sometimes the difference of temperatures used is so small as to give very slight indications of electro-motive force, even when the mean of the temperatures of the junctions differs widely from the neutral point, in which cases, of course, the test is deficient in sensibility. The best determinations are generally those derived from observations showing the galvanometer at zero, with the widest difference between the temperatures of the junctions, to which the thermometers are applicable with trustworthy indications; as, for instance, 100° or 150° Cent., which are attainable in the most favourable cases, being those in which the neutral point is at about midway between the temperatures of freezing and boiling water. The differences between these determinations sometimes amount to a degree or two, and even to several degrees when zinc was one of the metals; but generally the final mean for the neutral point does not differ by more than a degree from any single determination considered as satisfactory at the time it was made. 98. The mutual interchanges of thermo-electric order observed in various specimens of zinc, gold and silver, occasioned considerable perplexity, which has only been cleared up by observations made subsequently to the communication of this paper. The following determinations were made at different times and by different observers, as noted: | Observer | Date | Metals | Neutral points | |---------------------------|--------------------|-------------------------------|---------------| | Mr. C. A. Smith | Sept. 27, 1854 | P₁ mean | - 3·06 | | | | Gold electrodes | | | Mr. C. A. Smith | Aug. 18, 1854 | P₁ mean | - 1·5 | | | | Silver electrodes | | | Mr. C. A. Smith | Sept. 8, 1854 | P₁ mean | + 8·2 | | | | Zinc electrodes | | | Mr. C. A. Smith | Sept. 20, 1854 | Silver mean | 43·9 | | | | Zinc specimen (1) electrodes | | | Mr. G. Chapman and Mr. J. Cranston | Jan. 29, 1856, and Feb. 5 | Silver electrodes | 51·5 | | Mr. G. Chapman and Mr. J. Cranston | Feb. 1856 | Silver mean | 46·55 | | Mr. G. Chapman and Mr. J. Cranston | Feb. 1856 | Zinc (1) electrodes | 58·18 | | Mr. J. Murray | Aug. 1856 | Silver mean | 56·95 | | Mr. G. Chapman and Mr. J. Cranston | Feb. 1856 | Zinc electrodes | 71 | | Mr. G. Chapman and Mr. J. Cranston | Feb. 1856 | Gold mean | 69·76 | | Mr. G. Chapman and Mr. J. Cranston | Feb. 27, 1856 | Silver electrodes | 70·8 | | Mr. J. Murray | Aug. 21, 1856 | Silver mean | - 5·7 | | | | Gold electrodes | | Of the two results for the neutral point between silver and gold, only the last can be reconciled with the indications derived from the previous results as to the relative positions of these and the other metals tried along with them; and accordingly $-5^\circ$ has been taken as the neutral point of gold and silver in the thermo-electric diagram given below (§ 101). The first result, $70^\circ$, was found as the mean of several determinations, from none of which it differed by more than $0^\circ$, and the discrepancy can scarcely be attributed to errors of observation, but is probably due to slight differences in the specimens of gold and silver used in the different experiments. That very slight chemical differences in specimens of gold and silver wire may make great alterations in the temperature at which they become thermo-electrically neutral to one another, is readily understood by glancing at the diagram given below (§ 101), and observing how close together the lines for gold and silver lie. 99. The question, *Does the difference between the specific heat of electricity in two metals vary with the temperature*? may be answered by experiments showing the law according to which the means of widely different temperatures of the junctions giving no electro-motive force deviate from the true neutral point, which is the mean of any infinitely small difference of temperature giving no electro-motive force. I have not yet obtained indications of such a deviation in any case, having been prevented from prosecuting the inquiry by delays in the construction of a suitable air-thermometer. The examination I have been able to give the subject is only sufficient to show that the arithmetical mean of the temperatures of the two junctions giving no current, is probably in general within a degree of the true neutral point, when the difference between those temperatures does not exceed $100^\circ$ Cent. The following summary of a series of experiments made on two consecutive days may serve as an example of the degree of consistence of the results obtained by the method which has been explained, in a case in which the two metals deviate rapidly from one another above and below their neutral point. Sheet-lead Electrodes; $P_1$ mean. Determinations by Mr. C. A. Smith, May 17 & 18, 1854. | Difference of temperatures | Half sum of mercurial thermometer temperatures giving no current | |---------------------------|---------------------------------------------------------------| | $77\frac{1}{2}$ | $121\frac{1}{4}$ | | $71$ | $121\frac{1}{4}$ Mean | | $71\frac{1}{2}$ | $121\frac{1}{4}$ | | $71\frac{1}{2}$ | $121\frac{1}{4}$ | | $70$ | $122$ | | $185\frac{1}{2}$ | $120\frac{3}{4}$ | | $158\frac{1}{2}$ | $121\frac{3}{4}$ Mean | | $143$ | $121\frac{1}{2}$ | | $133$ | $121\frac{1}{2}$ | | $125\frac{1}{2}$ | $121\frac{1}{2}$ | | $68\frac{1}{2}$ | $122$ Mean | | $50$ | $123$ | Mean of temperatures by mercurial thermometer giving no current. Differences from $50^\circ$ to $77^\circ$...... $122^\circ 15$ Differences from $125^\circ$ to $185^\circ$... $121^\circ 4$ * See "Dynamical Theory of Heat," § 115, equations (15) and (17). These results seem on the whole to show that the mean of apparent temperatures giving no current is rather less for the wide than for the narrow ranges, in the case of the two metals concerned; that is, that the mean of the apparent temperatures giving no current is somewhat below the true neutral point. I need scarcely remark, however, that even if this indication could be relied on, it would be necessary to compare the actual mercurial thermometers which were used, with an air-thermometer, before any conclusions of value could be drawn from it regarding the constancy of the difference of specific heats of electricity in lead and platinum. 100. The following Table shows the results of observations leading to actual determinations of neutral points between various pairs of metals: | Temperature | P₃ Brass | P₁ Cadmium | Silver | P₁ Gold | P₁ Silver | P₁ Zinc | P₁ Tin | P₂ Lead | P₂ Brass | P₂ Tin | Lead | Different specimens of Silver | Different specimens of Zinc | |-------------|----------|------------|--------|---------|----------|--------|-------|--------|---------|-------|------|--------------------------------|----------------------------| | -14° C. | | | | | | | | | | | | | | | -12° 2. | | | | | | | | | | | | | | | -5° 7. | | | | | | | | | | | | | | | -3° 06. | | | | | | | | | | | | | | | -1° 5. | | | | | | | | | | | | | | | 8° 2. | | | | | | | | | | | | | | | 33° | | | | | | | | | | | | | | | 36° | | | | | | | | | | | | | | | 38° | | | | | | | | | | | | | | | 44° | | | | | | | | | | | | | | | 44° | | | | | | | | | | | | | | | 47°...71° | | | | | | | | | | | | | | The number at the head of each column expresses the temperature Centigrade by mercurial thermometers, at which the two metals written below it are thermo-electrically neutral to one another; and the lower metal in each column is that which passes the other from bismuth towards antimony as the temperature rises. It was also found that Aluminium must be neutral to either P₃ or Brass, or P₂, at some temperature between —14° C. and 38° C.; that Brass becomes neutral to Copper at some high temperature, probably between 800° and 1400°; Copper to Silver, a little below the melting-point of silver; Nickel to Palladium, at some high temperature, perhaps about a low red heat; and P₃ to impure mercury (that had been used for amalgamating zinc plates), at a temperature between —10° and 0°. P₃ appears to become neutral to pure mercury at some temperature below —25° Cent. 101. The following Diagram exhibits graphically the relative thermo-electric bearings of the different metals, and may in fact be regarded as a series of tables of the thermo-electric order of metals at different temperatures from —30° to 300° Cent. * This determination has been added in consequence of information given by Mr. Joule (December 1856), that hardened steel at ordinary temperatures differs thermo-electrically from copper by about one-tenth of the thermo-electric difference of iron from copper. Explanation of Thermo-electric Diagram. The orders of the metals in the thermo-electric series, at different temperatures, are shown by the points in which the vertical lines marked with the temperatures Centigrade, are cut by the horizontal and inclined lines named for the different metallic specimens. Fig. 18. The object to be aimed at in perfecting a thermo-electric diagram, is to make the ordinates of the lines (which will in general be curves) corresponding to the different metallic specimens be exactly proportional to their thermo-electric differences * from a standard metal (P₃ in the actual diagram). §§ 102, 103. Theoretical inferences regarding Electrical Convections of Heat, from facts of Thermo-electric Inversion. 102. The thermo-dynamic reasoning adduced above (§§ 10 to 15) leads to the conclusion (§§ 14, 15), that the convective power of the vitreous electricity is greater, or (which is the same thing) the convective power of the resinous electricity is less, in each metal for which the line in the diagram cuts the line for another metal from below it on the left to above it on the right, than in this other metal. Now it was established * See "Dynamical Theory of Heat," § 140. in Part I., that the vitreous electricity carries heat with it in copper (§ 54), or, as it may be expressed, the electric convection of heat is positive in copper. From the diagram we infer that it is greater, and consequently positive, in Brass. That it is positive in brass has been proved also by direct experiment (§§ 67 and 77). We infer also with certainty from the diagram, that the electric convection of heat (whether positive or negative) is greater in Zinc than in Gold, and greater in Gold than in Silver; that it is greater in Brass, Tin, Lead, Copper, Zinc, Gold, Silver, and Cadmium than in Platinum; that it is greater in Brass, Copper, Gold, Silver, and Cadmium than in Iron; that it is greater (that is to say, since it has been proved, § 76, to be negative, less negative) in Platinum than in Mercury; and that it is greater in Nickel than in Palladium. In Cadmium, as we may judge by the eye from the diagram, the convection is probably greater than in Copper; and in Palladium probably less (that is, greater negatively) than in Platinum. 103. These conclusions, certain and probable, are collected in the following Table of Convections, in which the different metals are arranged in order of the amounts of the electric convection of heat which they experience, or in the order of the values of "the specific heat of electricity in them." Electrical Convection of Heat | Metal | Convection | |----------------|------------| | In Cadmium | Positive | | Brass | Positive | | Order doubtful.| | | Copper | Positive | | Lead | equal | | Tin | Positive | | Zinc | Positive, Zero, or Negative | | Gold | Positive, Zero, or Negative | | Silver | Positive, Zero, or Negative | | Order doubtful.| | | Iron | Negative | | Platinum | Negative | | Nickel | Probably Negative | | Probably nearly| | | Palladium | Probably Negative | | equal | | | Mercury | Negative | PART III. EFFECTS OF MECHANICAL STRAIN AND OF MAGNETIZATION ON THE THERMO-ELECTRIC QUALITIES OF METALS. 104. Physical agencies having directional attributes and depending (as all physical agencies we know of except gravitation appear to do) on particular qualities of the substance occupying the space across or in which they are exerted, are transmitted or permitted with different degrees of facility in different directions if the substance is crystalline. The phenomenon of crystallization, exhibiting different chemical affinities on different bounding planes, between a growing crystal and the fluid from which it is being formed, and the cleavage properties (different specific capacities for resisting stress in different directions) afford the primary illustrations of this statement. It is probable that the proposition asserted is a universal proposition in the sense, that there is no kind of physical agency falling under the category referred to, which does not meet with different capacities for receiving it in different directions in some crystals. There certainly may be, and probably are, crystals which transmit certain physical agencies equally in all directions. Crystals of the cubical system, for instance (unless possessing the conceivable dipolar rotatory property*, from which some, if not all, are certainly exempt), conduct heat and electricity equally in all directions, and have equal magnetic inductive capacities and equal thermo-electric powers. But thermal and electric conductivity, magnetic inductive capacity, and "thermo-electric power†" are undoubtedly different in different directions in many, if not in all, crystals not of the cubical system. Many crystals have not shown any marked difference in their absorption of light according to the direction of its propagation through them; but some undoubtedly do show a difference of this kind, to such a degree as to give sensibly different colour to light passing short distances through them in different directions‡. Faraday had good reason, after making the discovery of the induction of electro-polarization in non-conducting substances§, to try the specific directional qualities of crystals used as dielectrics; and although he found no sensible differences in the inductive capacities of the crystals (rock crystals and Iceland spar) which he tried for this kind of action, in different directions, it appears highly probable that induced electro-polarization will sooner or later be ascertained to be no exception to the general rule. 105. Another very general principle is, that any directional agency applied to a substance may give it different capacities in different directions for all others. Whether or not this is true as a universal proposition, events have proved that the probability of its being true in any particular case is quite sufficient to warrant an experimental inquiry. Brewster discovered that mechanical stress induces in glass directional properties with reference to polarized light, which are lost as soon as the stress originating them is removed. These properties were shown by Fresnel to be of the same kind as the property of double refraction possessed by a natural crystal. Experiments made by Sir David Brewster and Mr. Clerk Maxwell prove that isinglass and other gelatinous substances dried under stress, thin sheet gutta-percha * See "Dynamical Theory of Heat," § 168; also §§ 163, 166, 167, 169 to 171, Transactions of the Royal Society of Edinburgh, May 1854. See also Professor Stokes "On the Conduction of Heat in Crystals," Cambridge and Dublin Mathematical Journal, Nov. 1851. † Or thermo-electric difference from a standard metal. See "Dynamical Theory of Heat," § 140. ‡ Most crystals not of the cubic system, even when nearly colourless, exhibit difference. See Haidinger's 'Researches.' § Experimental Researches in Electricity, Series XIV. §§ 1688, 1689, 1692 to 1698. June 1838. permanently strained by traction, and probably all non-brittle (or plastic) transparent solids when permanently strained otherwise than by uniform condensation or dilatation in all directions, possess double refraction as a property of the molecular alteration which they acquire under the stress and retain after the stress is removed. Again, magnetization, as Joule discovered, causes an elongation of iron in one direction (that of the magnetization) and a contraction in all directions perpendicular to it, with no sensible change of volume. Faraday discovered the wonderful dipolar optical property of transparent bodies in a magnetic field (the first and only case known of any dipolar qualities, other than those of magnetic and electric reactive forces, called into existence by induction): Maggi discovered that magnetized iron conducts heat with a greater facility across than along the lines of magnetization*. 106. In applying the dynamical theory of heat to thermo-electric currents in conducting crystals, I was led to consider the probable effects of mechanical strain, and of magnetization on the thermo-electric properties of non-crystalline metals, and in consequence entered on the investigation, of which the results, so far as I have yet advanced in it, are now laid before the Royal Society. 107. To find the effect of longitudinal tension on the thermo-electric quality of a metal, I first took eight thin copper wires each capable of bearing about 10 lbs., and, attaching their upper ends to a horizontal wooden arm at distances of about \( \frac{1}{4} \) of an inch from one another, allowed them to hang down, each kept stretched by a weight of about \( \frac{1}{2} \) lb. They were connected with one another in order, and the first and last with the electrodes of a galvanometer, by nine wires soldered to them, as shown in the diagram; the junctions between the successive wires being alternately in the upper and lower of two horizontal lines 4 inches apart. Every alternate wire was then stretched with a weight of about 3 lbs., and a slip of hot plate glass was applied, sometimes to the upper and sometimes to the lower row of junctions. A deflection of the galvanometer needle was observed in one direction or the other, according as the glass heater was applied to one set of junctions or the other. The deflection was also reversed when the weights were changed to the alternate set of wires, and the heater kept applied to the same set of junctions. In every case the deflection was such as to indicate a current from stretched to unstretched through hot junctions. The uniform and consistent nature of the indications was such as could leave no doubt as to the result; and I concluded that copper wire stretched --- * Doubts have been thrown on this result, I believe, by other experimenters, who have not succeeded in verifying it by their own observation, but its close correspondence with a result I have recently discovered by experiments on the electric conductivity of magnetized iron, have diminished the impression such doubts produced on my own mind; and I look with much interest to a repetition of Maggi's experiment. by a longitudinal force bears to copper wire of the same substance unstretched, the same thermo-electric relation as that of bismuth to antimony. 108. I next made a similar experiment on iron wire, varying the arrangement so that the weights could be rapidly shifted; and again so that equal sets of forces could be applied to one or to the other of the two sets of wires, merely by pressing with the foot upon one or another of two levers. A perfectly decided result was at once obtained; and I ascertained that the thermo-electric effect was induced and lost quite suddenly on the pressure being applied and removed. In this case the nature of the effect was the reverse of that found in the experiment on copper, the deflections being always such as to indicate a current in the iron wires from unstretched to stretched through the hot junctions. 109. The thermo-electric effect which these experiments demonstrated to accompany temporary strain produced by a longitudinal force, was, in each of the metals, the reverse of that which Magnus* had previously discovered in the same metal hardened by the process of wire-drawing, and which I ascertained for myself to be produced in each case when the metal is hardened by simple longitudinal stress without any of the lateral action inseparable from the use of the draw plate. I thus arrived at the remarkable conclusion, that when a permanent elongation is left after the withdrawal of a longitudinal force which has been applied to an iron or copper wire, the residual thermo-electric effect is the reverse of the thermo-electric effect which is induced by the force, and which subsists as long as the force acts. 110. I have made a single experiment demonstrating this conclusion for iron by means of a multiple tension apparatus, similar in principle to that described above (§ 105). But with a somewhat more sensitive galvanometer than the one I used, the result may be shown in a perfectly decided manner (for iron at least) without any multiplication of the thermo-electric elements; and a very striking experiment may be made on the following plan:—A thin iron wire is wrapped three or four times round a wooden peg held firmly in a horizontal position, and again two or three times round another parallel peg, about 4 inches lower. A frame is rigidly connected to this second peg, so that it may remain stably in a horizontal position; hanging from the wire and pulled down by the frame with either a light or a heavy weight attached to its lowest point. To keep the wire from slipping, the parts of it running from the pegs towards the ends are kept stretched by light weights tied to them; and the slack parts below these weights are carried away to the galvanometer electrodes, with which they are connected in the manner described above (§ 92). Any convenient source of heat is applied to the part of the wire bent round either peg, so as to keep it at some temperature, perhaps about as high as that * Poggendorff's 'Annalen,' Aug. 1851. of boiling water. If the wire be well annealed at the commencement of the experiment, and if weights be gradually added to the lower side of the frame, the galvanometer needle gradually moves to one side, indicating a current from the unstretched to the stretched round the hot peg; and the deflection goes on increasing as long as weights are added, up to the breaking of the wire. If, however, before the wire breaks, the weights are gradually removed, the needle comes back towards its zero-point, reaches zero, and remains there when a certain part of the weight is kept suspended. If this is removed the needle immediately goes to the other side of zero, and remains, indicating a current from the strained part into the unstrained part of the iron wire round the part wrapped on the hot peg; that is, from strained to unstrained through hot, or as Magnus found, "from hard to soft through hot." 111. If weights be added again, as at first, this deflection is done away with, and the deflection that first appeared is regained, when the weight which previously allowed the needle to return to zero is exceeded. We thus conclude that iron wire hardened by longitudinal tension, may, by the application of a certain longitudinal force, have its thermo-electric quality reduced to that of unstrained soft iron, and by a greater force may be made to deviate in the other direction; or that hardened iron under a heavy stress, of the kind by which it has been hardened, and hardened iron left free from stress, are on different sides of unstrained soft iron in the thermo-electric series. There can be no doubt but that the same property holds for copper wire, being in fact demonstrated by the experimental results described above in §§ 107 and 109. 112. I have not yet investigated the thermo-electric effects of stress (that is, the effects accompanying temporary strain) in other metals than iron and copper; but it appears probable that the same law of relation to the thermo-electric effects of permanent strain without stress will be found to hold in each case, since it has been established for two metals in which the absolute thermo-electric effects are of contrary kinds. I hope, however, before long to be able to adduce experimental evidence which will supersede conjectures on the subject. [Since this paper was read I have verified the same law for platinum wire.] 113. The object which was proposed in entering on the investigation, being to test the thermo-electric properties of a strained metal, in different directions with reference to the direction of the strain, was not attained by comparing the thermo-electric properties of a longitudinally strained metal with those of the same metal in its natural state; but it would certainly be promoted by discovering the effect of lateral pressure on a wire in modifying its longitudinal thermo-electric action. I therefore made the following experiments on the thermo-electric effects experienced during the application of a moderate lateral pressure, and of permanent strain after the cessation of excessive lateral pressure, in various wires. 114. Experiment to discover the temporary effect of lateral pressure on the thermo-electric quality of iron wire:—A rectangular bar of iron (1\(\frac{2}{3}\) inch square), with pieces of thin hard wood placed on two opposite sides, had fine iron wire laid in a coil of about twenty turns round it. The wood perfectly insulated the wire from the iron bar, and the different turns of the wire were kept from touching one another, by little notches cut in the edges of the pieces of wood. The whole coil was made firm, and its extreme turns tied down to the wood to prevent slipping. The ends of the wire, extending a foot or two on each side of the coil, were connected in the usual way (§ 92) with a galvanometer. The bar bearing the coil was laid with its two wooden faces horizontal, and one of them supported on a thin piece of hard wood lying on the stage of a Bramah's press. Another thin piece of hard wood was laid upon the top of the coil, to prevent the upper part of it (when, in the course of the experiment, it is forced upwards,) from touching the roof of the press. Blocks of iron were placed on the ends of the bar, so that when the stage is pushed up they may be resisted by the roof, cause a heavy stress to act on the bar, and press the lower horizontal parts of the wire coil between the two pieces of hard wood touching them above and below. The same blocks are afterwards shifted to rest on the stage and bear the ends of the bar upon them, so that, when the stage is forced up, the upper parts of the wire coil may be pressed against the piece of hard wood above them, which will then be resisted by the roof of the press. Pieces of plate glass highly heated were applied to the vertical parts of the wire on one side of the bar, those on the other side being left cool, and the galvanometer was observed. Some slight deviation of the needle was generally noticed. Then the press was worked, and immediately a strong deflection took place, indicating a current in the iron coil, from the uncompressed portions through the heated vertical portions, into the compressed portions. The pressure was relieved, and the galvanometer needle returned nearly to zero. It was reapplied, and the same powerful deflection was observed. The glass heaters were shifted to the other side, and, the pressure being continued, the deflection of the needle became reversed. The pressure was removed, and by shifting the iron blocks, and working the press again, was applied on the other horizontal side of the coil. The heating being kept unchanged, a reverse deflection was observed, powerful as at first. The current indicated was in every case from free iron wire to pressed iron wire through hot, as is illustrated in the diagram, for a case in which the upper parts of the wire are compressed. 115. From this, in conjunction with the result regarding the effect of longitudinal stress previously obtained, we may nearly conclude that a longitudinal strain in iron developes reverse thermo-electric qualities in the axial direction and in directions perpendicular to it; for there can be little doubt but that a lateral traction would produce the reverse effect of a lateral pressure, or that a portion of a linear conductor of iron pulled out on two opposite sides in a direction at right angles to its length, would acquire such a thermo-electric quality as to give rise to currents from stretched to free through hot. But in the former experiment (§ 108) it was demonstrated, that when part of an iron conductor is pulled out longitudinally, the thermo-electric effect gives currents from free to stretched through hot. The crystalline characteristic is therefore established for the thermo-electric effect of mechanical stress applied to iron, if it be true that traction produces the reverse temporary effect to that of pressure in the same direction. There seems so strong a probability in favour of this supposition, that it may almost be accepted without experimental proof; but I intend, notwithstanding, to make experiments, for the purpose of explicitly testing it, as soon as some preparations at present in progress enable me to do so. In the mean time I have made the following decisive experiment on the difference of thermo-electric quality in different directions in iron subjected to stress. 116. A piece of sheet-iron 36 inches long and 16 inches broad, was rolled round two thick iron wires (\(\frac{1}{4}\)-inch diam.), along its breadth at its two ends, and soldered to them. It was cut into narrow slips, each about \(\frac{1}{4}\) of an inch broad and of different lengths, as shown in the diagram, so as to prevent electric conduction, except along a band about half an inch broad running across the sheet at an angle of 45° through its centre. The ends of the slips on each side of this band were clamped (as shown in the annexed sketch) between two flat iron bars, but insulated from them by thin pieces of hard wood*, and from one another, where necessary, by pieces of cotton cloth. These bars were each \(\frac{1}{2}\) an inch thick, 3 inches broad, and 30 inches long; and the two at each side clamped together upon the pieces of hard wood, with the iron slips between them, formed a firm beam, by means of which a considerable stress would be brought to bear on the sheet iron to stretch it in the direction of the slips. The upper of these beams was laid resting with its two ends on the tops of stout wooden pillars, supported below on a very strong wooden bar laid on the stage of a Bramah's press. The lower double iron beam hanging down and straightening the sheet iron by its weight had strong iron links put over its ends, and an iron bar of about \(1\frac{3}{4}\)-inch square section slipped through them below, so as to hang down a small distance below the roof of the press. Thus, when the press is worked, the upper double iron beam is forced up, and the sheet iron is stretched between it and * The thinner the better, I believe, as a partial failure was experienced from these pieces of wood breaking at one side and allowing the ends of the iron slips to get drawn in between the iron bars. the lower double iron beam, which is held down by the links and the bar under the roof of the press. Before working the press, the rectangular wooden frame with its iron cross- head is steadied by cords from hooks in the ceil- ing, and the following arrangements are made:— Two slips of sheet iron, each about 18 inches long, are soldered to the upper and lower ends of the oblique conducting channel, and their other ends are soldered to copper wires and put into the circuit of a galvanometer, with the usual precau- tions (§ 92) to ensure equality of temperature and electrical insulation between the two junctions of the dissimilar metals. Four tin-plate tubes, of semicircular section, each about \( \frac{2}{3} \)-inch diameter, and coated with a single fold of paper pasted round it, are pressed with their flat sides on the two sides of the sheet iron against the upper and lower edges of the oblique conducting band; and are connected by india-rubber junctions, so that steam may be blown through two of them to heat one edge of the conducting band, and cold water sent through the other pair to keep the other edge of the band cold. The arrangements being thus made, a small boiler, heated by a common wire-gauze gas-lamp, is used to send steam through one pair of the tubes, and the town-supply water-pipes give a continued stream of cold water through the other pair. When the galvanometer was observed, there was at first no sensible indication of a current. The press was then worked, and the galvanometer immediately exhibited a slight deflection. The press was released, and a careful observation gave again little or no evidence of a current. Then, by an arrangement of double-branched stop-cocks, the steam and cold water were quickly reversed, so that the edge of the conducting band which was hot became cooled, and the other one became heated. Still the galvanometer showed no sign of current until the press was worked, when a reverse deflection to the former was manifested. While the press was kept up the steam and cold water were again sent along the same edges as at first. After a short time the deflection of the needle was reversed, and the same current as at first was indicated. The deflections were very slight in each case, but were unmistakeably demonstrated by the use of the reversing break (com- mutator) connected with the galvanometer. Had it not been for the accident noted above, a much more powerful stress would have been applied to the iron, and I have no doubt but that conspicuous deflections of the needle would have been produced. 117. The current in every case was down the inclined channel of sheet iron when the upper edge was heated, and up the incline when the lower edge was heated. That is, if we imagine a rectangular zigzag, from side to side of the bar, instead of the true rectilinear course of the current, the current would be from transversely stretched to longitudinally stretched through hot. Hence it is established by this experiment, that iron, under a simple longitudinal stress, has different thermo-electric qualities in different directions. Knowing, as we do, from the first experiment on copper, described above (§ 107), that iron is not the only metal thermo-electrically affected by stress, we may conclude with much probability that, in general, metals subjected to stresses not equal in all directions will acquire the crystalline characteristic of having different qualities, as regards thermo-electricity, in different directions. 118. The qualitative investigation of the thermo-electric effects of stress, unaccompanied by permanent strain, that is, the elastic thermo-electric effects of stress, would be complete for iron if the thermo-electric effect of a uniform dilatation or condensation in all directions had been ascertained. I hope before long to be able to carry into effect various plans I have formed with this object in view; but in the mean time it would be the merest guessing to speculate as to the result. 119. The establishment of the crystalline characteristic for the thermo-electric effects of stress not equal in all directions, would make it probable that any thermo-electric effects which a metal permanently strained by such a stress can retain after the stress is removed, must also possess the crystalline characteristic. That this is really the case I had in fact proved, before performing the decisive experiment, just described, regarding the nature of the elastic effect, which was only made a few weeks since. The following experiments on the thermo-electric effects of permanent strains in metals were all made more than a year ago. 120. Well-annealed iron wire was rolled in a coil of about twenty turns on a flat bar of iron \( \frac{1}{4} \)-inch thick and 2 inches broad. The bar was laid on an anvil, with little pieces of thicker wire laid upon it to support the iron core and prevent the lower parts of the coil from being pressed. The upper parts of the coil lying on the upper flat side of the core were hammered till they were all very much flattened. The coil was then a little loosened and drawn off the bar of iron, and a similar wooden core was pushed into it. The ends of the iron wire were arranged, with the usual precautions (§ 92), in connexion with the electrodes of a galvanometer. A piece of hot glass (not above the boiling-point of water) was laid along one edge of the coil, so as to heat the iron wire at one set of the points separating hammered from unhammered portions. The galvanometer showed by a great deflection of its needle a current through the iron coil from hammered to unhammered through hot. When the heater was applied at the other edge of the flat coil, the deflection soon became reversed; still, and always in subsequent repetitions, indicating a current from the strained to the soft metal through the hot junctions. 121. The coil was next replaced on its iron core, heated to redness in the fire, and cooled slowly. It was then insulated by slipping in paper between it and the iron bar, or by putting it once more on its wooden core; and it was tested in the galvanometer circuit with the application of glass heaters as before. Not the slightest trace of a current was now found; a result verifying the conclusion arrived at by Magnus, that it is not peculiarities of form in different parts of a circuit of one uncrystallized metal, but variations in its quality as to mechanical strain, that can ever give it continuous thermo-electric action. 122. It has thus been proved that a circuit of iron permanently strained by pressure across the lines of conduction acquires the same kind of thermo-electric quality as that which Magnus first discovered to be produced by the lateral pressure compounded with longitudinal traction, which the process of wire-drawing calls into play, or as that which I had myself found to result from a simple traction, leaving a permanent elongation after the force is removed. In all these cases the iron is found to be harder than it was before acquiring the strain, or than it becomes again after being annealed. Hence the nature of the thermo-electric effect in each of the three cases falls under the designation "current from hard to soft through hot," by which Magnus stated his result as regards iron. This is just as is to be expected from the crystalline theory; since longitudinal extension has a common characteristic with lateral condensation in the theory of strains, and only differs from condensation uniform in all transverse directions, by a certain degree of absolute dilatation which accompanies it, instead of the slight absolute condensation accompanying the lateral condensation as an effect of pressure all round the sides. In fact the agreement between the characters of the thermo-electric effects due to longitudinal traction and lateral pressure, and again between the reverse characters of the effects of permanent longitudinal extension and those of permanent lateral compression established by the experiments which have been described, proves that these effects are due to distorting stress, and to permanent distortion, in the main, and leaves it quite an open question, only to be decided by further experimental investigation, what may be the effects of uniform pressure and of permanent uniform condensations or dilatations. 123. The crystalline theory is really unavoidable when it is thus established that the effect discovered is due to distortion; but still, as the one designation "current from hard to soft through hot" applies to all the cases of permanent strain in iron as yet experimented on, I thought it necessary, for removing the possibility of objections, that an iron conductor giving a current from soft to hard through hot, should be constructed. I therefore took twenty-four small soft iron bars turned in a lathe to a cylindrical form \( \frac{1}{4} \)th of an inch diameter, and each an inch long, with flat ends; and compressed twelve of them longitudinally in a Bramah's press, so as to permanently shorten each by about \( \frac{1}{8} \)th of an inch. They were then set in a wooden board cut to hold them firmly lengthwise in two rows, those hardened by compression and those left soft, being placed alternately with their ends in contact. The end pieces towards one side were connected with one another by a little slip of iron touching each, and the other ends of the rows were connected with the electrodes of a galvanometer by slips of iron touching them. Each row was firmly wedged up between its terminal iron slips to ensure metallic contact; but after several attempts, and with all care in cleaning the surfaces meant to touch, no sufficient completeness of contact throughout the circuit could be obtained until mercury was introduced as a liquid solder to connect the pieces of iron. This was done simply by pressing them together as at first, pasting paper round the junctions, and pushing little drops of liquid mercury or small quantities of soft mercurial amalgam into apertures in the tops of these paper coverings. Twelve hollows were cut in the board under and round the junction of the iron bars, each except the last including a pair of ends of the bars in contact in each row, and the last including the ends of the extreme bars on that side and the slip of iron by which they are connected. These hollows were filled alternately with hot sand and cold sand, which was everywhere piled over the junctions; and the galvanometer gave slight indications of a current, the direction of which through the iron appeared to be generally from uncompressed to compressed, through hot. 124. The result, however, was not satisfactory; and it was obvious that the plan which had been adopted for heating and cooling was quite insufficient to sustain the required differences of temperature through so considerable masses of iron; I therefore had an apparatus constructed for the purpose, consisting of two main pipes of tin-plate, each carrying six smaller pipes and leading to small cells, also of tin-plate, with cylindrical passages through them to admit the iron bars, and with short discharge pipes attached to them on the other side from that by which the former enters. These cells were fitted into the hollows cut for the sand in the board formerly used, the main pipes occupying parallel positions above them on each side several inches from one another. The iron bars, each coated with paper and united as before one to another with mercury solder, were pushed through the hollows of the cells, and were fixed in two rows, with a junction in the centre of each of these hollows, and with the terminals adjusted as before. Cold water from the town supply-pipes was then run into one of the main pipes, so as to flow through the branch pipes and cells connected with it; and steam from a boiler heated by an ordinary wire-gauze gas-burner was sent through the other system, so as to cool and heat alternately in their order of position the twelve cells with the junctions which they surround. A deflection of the galvanometer needle, amounting to about 4°, was now observed; and when the cold water and steam supplies were interchanged in the two sets of tubes, an equal reverse deflection almost immediately took place. The current indicated was always in many trials from uncompressed to compressed through hot in the iron of the circuit. 125. Here then we have a case of thermo-electric action in iron giving a current from soft to hard through hot; not as found before, “from hard to soft through hot.” Hence it is not pieces of hardened iron in general, but the direction of extension or directions perpendicular to the direction of compression, in iron hardened by extension or by compression, that have the thermo-electric quality of deviating from soft iron towards bismuth; and a line of compression, or (as we may now safely conclude) lines perpendicular to a line of extension, have the reverse deviation, that is deviate from soft iron towards antimony, in the thermo-electric series. [Addition, Dec. 1856.—Subsequently to the reading of the paper, I have, in verification of this conclusion, found, by a direct experiment, that a conductor of sheet iron, hardened by lateral extension and softened in parts, has the thermo-electric property of giving a current from soft to hard through hot.] The crystalline theory being thus fully established for the thermo-electric effects of mechanical strain in iron, whether temporarily induced during the application of stress, or remaining with molecular displacement after the stress is removed, we may readily suppose it will be found to hold equally for all thermo-electric effects any metal can experience from mechanical action, except the hitherto undiscovered effects of condensations or dilatations equal in all directions. The experiments I have already made on other metals than iron, do not go further in verifying the crystalline theory than to show for copper and tin wires what I had previously shown for iron, that the same thermo-electric effect in a linear conductor is produced by permanent longitudinal extension and permanent lateral compression. 126. The process of raising to a high temperature and then cooling very suddenly, produces a marked effect on the mechanical qualities of most metals, especially on their hardness; and generally all that is necessary to do away with this effect and restore the metal to its primitive condition, is to keep it for some time at a high temperature and let it cool slowly. This process being called annealing, I shall for brevity designate as unannealed, any substance which has been subjected to the former process (sudden cooling) and which has not been subsequently annealed. It is not easy to judge exactly of the relation of the strains in the different parts of an unannealed piece of metal, to simple mechanical strains; but some thermo-electric effect, whatever its exact nature and explanation may be, is to be anticipated, with so great a change of other qualities as many metals experience in the process of sudden cooling; and it may be readily supposed that different thermo-electric qualities will be found in unannealed pieces of different shapes. I have therefore made experiments on the thermo-electric differences between unannealed and annealed linear conductors. consisting of round wires, of wires flattened by hammering, and of flat slips, of one metallic substance. 127. Twisting a wire beyond its limits of elasticity hardens it perhaps as much as traction or hammering, and certainly in every case, when continued far enough, makes the metal very brittle. The nature of the mechanical strain here operative is easily expressed and explained in the theory of elasticity in terms of simple strains different in magnitude and direction in different parts of the wire; but it is not very easy to judge by theory from the effects of simple strains supposed known, what kind of thermo-electric effect, if any, is to be expected in a metallic wire, with strain thus heterogeneously distributed through it. I have therefore made experiments to determine this effect in various metals. 128. For experimenting on the thermo-electric differences between annealed and unannealed metallic conductors, a wire, round or flattened, or a slip of the metal was wrapped in a coil of from ten to thirty turns on a wooden core, about 2 inches broad and \( \frac{1}{4} \) of an inch thick, or sometimes only an inch broad, with a flat slip of thin sheet-iron laid on one side of it. The wooden core was then drawn away, and the coil, held in form by the thin iron core, was heated to redness in the fire, or to some temperature short of its melting-point, in hot oil, and was then suddenly plunged in cold water. After that, one side of the iron core was held over a flame, so as to heat the parts of the coil next it, while the parts of the coil on the other side were carefully kept cool, by the constant application of cold water with a sponge. The wooden core was then slipped in and the sheet-iron removed; and the coil was ready for testing by the galvanometer. 129. The preparations for an experiment on the thermo-electric effect of permanent torsion, were commenced by bending a short portion at each end of a length of two or three yards of the wire to be examined, holding these end portions so as to keep the wire between them firmly stretched, and twisting it till it became brittle. It was then wound on a flat iron core (unless it was too brittle, as often proved to be the case, and then another wire was similarly prepared but not twisted quite so much); the parts of the coil on one side were carefully annealed by flame or hot oil, while those on the other side were kept cool by sponging with cold water. The iron core was then drawn out and the wooden core slipped into its place; and the coil was ready for testing by the galvanometer. 130. In making the thermo-electric experiments on the coils prepared in these various ways, glass heaters were first used, but I afterwards substituted two tubes of horseshoe section made of tin-plate and coated with paper, which were applied with their concave parts touching the coil round its two edges. Steam from the small boiler was sent through one of these, and cold water from the town supply-pipes through the other. 131. The wires used, with the exception of the iron, steel and brass, were all supplied by Messrs. Matthey and Johnson, as chemically pure. The results of the experiments (made as described in §§ 120 and 121) on the effects of lateral hammering were, in every other kind of wire tried, the reverse of those found for iron. Thus in steel, copper, tin, brass, lead, cadmium, platinum, zinc, the current was always found to be from the unhammered to the hammered portions through hot. All the wires except zinc were carefully annealed by myself, before they were coiled and hammered (§ 120); but the process of annealing by heating in oil and cooling slowly made the zinc very brittle and crystalline, instead of softening it as in the other cases, and it was therefore taken as supplied by the manufacturers, and coiled on the core and hammered in the manner described. 132. The experiments on the coils differently tempered in their different parts (§ 126), in the cases of tin and cadmium, gave only doubtful galvanometer indications; zinc wire proved so brittle in the annealed parts as to defeat some attempts to test the thermo-electric effects of temper. I have little doubt but that results may be obtained in all these cases by a careful repetition of the experiments, with perhaps some modification to meet the peculiarity of zinc. Slips of sheet iron and of sheet copper were tried without any thermo-electric indication being noticed. [Addition, Dec. 1856.—I have recently found in slips of sheet iron the same thermo-electric effect of temper as in round and flattened iron wires.] All the other conductors tried gave very decided results. In the cases of round iron wires of very different diameters, of iron wire flattened through its whole length by hammering, of round steel wire, and of steel wire flattened through its whole length by hammering, and of steel watch-spring, the thermo-electric effect of annealing portions of the coil after the whole had been suddenly cooled, was a current from unannealed to annealed through hot. In round wires of copper and brass, the thermo-electric effect of the same process was a current from annealed to unannealed through hot. 133. The effects of permanent torsion were decisively tested only for iron and copper wires; and they proved to be in each case the same as the effects of hardening by longitudinal extension, by lateral compression, or by rapid cooling, being quite decidedly from brittle to soft through hot in the iron, and from soft to brittle through hot in the copper. 134. The views explained above (§ 105), by which I was led to look for the thermo-electric qualities of a crystal in a non-crystalline metal subjected to mechanical strain, show the probability of finding such properties also developed along with magnetism, by external magnetic force, especially in the few metals, iron, nickel and cobalt, which have high capacities for magnetic induction. Towards verifying this idea I tried first the following simple experiment, analogous to the first experiment (§ 107) which I had made on the thermo-electric effects of tension. A little helix about 3 inches long, consisting of 220 turns of thin covered copper wire laid on in three strands on a cylindrical core of pasteboard, about \( \frac{1}{4} \) of an inch internal diameter, was slipped upon a piece of thick straight iron wire about 2 feet long, which was supported in a horizontal position by its ends, and through them put in the circuit of a galvanometer. A spirit-lamp was held under the middle of the wire so as to raise it to a high temperature, and then a current from a few of the iron cells was sent through the helix, which was kept a little on one side of the middle of the wire. Immediately the galvanometer needle, which was not at first disturbed by the application of the spirit-lamp, experienced a deflection. The little helix was slipped rapidly through the flame of the spirit-lamp to the other side of the hot part of the wire, and a reverse deflection was immediately produced. It was easy, by moving the helix alternately to the two sides of the hot middle of the wire, to make the needle of the galvanometer to swing through an arc of $10^\circ$ or more. When the needle was brought to rest there was always a most sensible permanent deflection, on one side or the other, according as the helix was left on one side or other of the heated parts. When the circuit of the galvanometer was broken, none of these effects followed from the motions of the helix. They were therefore not due to the direct force of the magnetism in the helix and iron wire, but to that of a current through the galvanometer coil. This always took place in such directions as to indicate a current from unmagnetized to magnetized through hot. 135. The decided character of the result of this experiment established it beyond doubt, that the thermo-electric quality of iron is altered by magnetization. Immediately the question arose (from the general considerations referred to above, §§ 104 and 105), are the thermo-electric qualities equally or even similarly affected in all directions? and the crystalline hypothesis suggested the answer:—no; probably even the reverse thermo-electric effect may be found across their lines of magnetization. As theory could give no more than a conjectural answer, I tried to find the truth by experiment; and, after various fruitless operations, obtained a very decided result, in the following way. 136. A piece of thin sheet iron was cut into the shape shown in the diagram, the breadth everywhere being about $\frac{1}{4}$ of an inch, the length of the longer branch 45 inches, and that of the shorter 6 inches. The longer branch was rolled into a plane spiral, on a cylindrical core $\frac{1}{2}$ an inch diameter, the different successive turns being prevented from touching one another by a piece of narrow tape wound on along with the iron slip. The shorter branch, which stood out from the inner end of the coil at right angles to the plane of the spiral, was bent round into this plane, and carried out along one side of the spiral several inches beyond its circumference. Along with it, a portion of the slip next the other end which was left uncoiled, was carried out from the outer part of the spiral, and cut to such a length as to let the two ends be brought close together. Copper wires, to lead to the galvanometer electrodes, were soldered to these ends, and the junctions of dissimilar metals thus formed were arranged with the usual precautions (§ 92) to ensure equality of temperature and electrical insulation. Contrary poles of two steel bars, each about 3 feet long and of rectangular section, 4 inches by \( \frac{1}{2} \) inch, were placed pressing on each side of the spiral, as shown by the dark shading in the diagram, but insulated from it of course. Four rectangular pieces of thick plate glass, two of them very hot (perhaps about 300° Cent.) and two cold, were applied, touching the coil on each side, and symmetrically arranged on the two sides of the steel magnets. The galvanometer showed a current in the direction indicated by the arrow-heads. The pieces of hot and cold plate glass were interchanged, and the current became reversed. The magnets were removed, and their effects became scarcely perceptible, or altogether ceased. On repeated trials a current was found always in the direction, from parts of the coil between the magnets towards parts touched by the hot glasses. The experiment was repeated with a powerful electro-magnet, and gave the same result, but not with the same ease, because of difficulties in applying the heaters, &c. 137. The very strong tendency iron has to assume longitudinal rather than transverse magnetization, when of any form extended in one direction more than in others, was partially done away with by the mutual influence of the different turns of the spiral used in the experiment which has been described; and the symmetrical arrangement of the heaters was such as to nearly exclude all thermo-electric action, except what is due to the thermo-electric difference between that part of the coil touched on each side by the steel magnets, and the part diametrically opposite. Any thermo-electric effect there may have been from longitudinal magnetization in the parts of the iron ribbon on each side of the steel magnets, must, so far as I could judge, have been contrary to the effect observed. The result obtained, therefore, demonstrates an electro-motive force urging a current from transversely magnetized parts of the iron conductor, through hot parts, to comparatively unmagnetized parts. Hence a transversely magnetized iron conductor deviates from unmagnetized iron towards bismuth, or in the reverse direction to that of the deviation discovered in wire longitudinally magnetized, in the first experiment on the thermo-electric effects of magnetism. It may be concluded, à fortiori, that in uniformly magnetized iron, directions transverse to the lines of magnetization differ thermo-electrically from directions along the lines of magnetization; and differ in such a way, that if we could get an iron conductor of the shape indicated in the diagram magnetized, with perfect uniformity everywhere, in the direction shown by the lines of shading, and if, when the two ends kept at the same temperature are put into the circuit of a galvanometer, the corner is heated, a current would be found to set in the direction shown by the arrow-heads, that is, from transversely magnetized to longitudinally magnetized through hot. 138. To test and illustrate this conclusion, I took a piece of sheet iron, cut to the shape shown in the diagram, and wound it spirally on a wooden cylinder, prepared with spiral grooves and pipes for steam and cold water, as described below. The oblique edge of the iron, shown on the left boundary in the diagram, being cut at angles of $45^\circ$ and $135^\circ$ to the long edges conterminous with it, was bent in a plane perpendicular to the axis of the cylinder, and thus the long edges of the iron, and the cut separating it into two branches, formed spirals, each at an angle of $45^\circ$ to the axis of the cylinder. The two long edges themselves came very nearly to coincide, the circumference of the cylinder being a little greater in length than the oblique edge of the iron which thus nearly met round it. These two edges, as well as the two edges on each side of the cut between the branches, were prevented from touching one another by being, one at least in each of the contiguous pairs, bound with cotton tape. The projecting slips (shown on the right in the diagram) came to positions parallel to the axis of the cylinder, through two diametrically opposite parts of its circumference. Their ends had copper wires soldered to them, and were arranged with the usual precautions ($§92$) to ensure electric insulation and equality of temperature between them. The wooden cylinder had two diametrically opposite spiral grooves, each at the same inclination of $45^\circ$ to the axis, and spiral sheet copper tubes, prepared of the proper shape, were slipped into these grooves, and nearly filled up the spaces to the surface of the cylinder. The outsides of these tubes were coated with paper, so as to maintain electric insulation between them, and the sheet iron wound on outside. The wooden cylinder bearing the spiral tubes, and the sheet iron arranged in the manner described, was slipped into the hollow of an electro-dynamic helix, steam was sent through one of the spiral tubes and water through the other, and the copper wires soldered to the ends of the iron slips were connected with the electrodes of a galvanometer. No current was at first indicated. The galvanometer circuit was broken by its own commutator, and a current was sent through the magnetizing helix. The galvanometer circuit was completed again, and immediately a strong indication of a current through it was manifested. The galvanometer circuit was broken, the magnetizing current reversed, and the galvanometer circuit again completed; again the same current as before was observed. The steam and cold water were interchanged in the spiral pipes, and the galvanometer current soon set in the reverse direction, with about the same force as before. The magnetizing current was stopped (the galvanometer circuit being broken for the time and closed again), and only slight traces of the current that had been so powerfully indicated could now be observed. 139. In this experiment the action of the electro-dynamic helix caused the double slip of iron to receive magnetization in lines nearly parallel to the axis of the cylinder (only a little disturbed in consequence of the gaps between the adjacent edges), that is to say, magnetization as nearly as may be in directions at an angle of 45° to its length. The sources of heat and cold applied along the two spirals, gave either heat along each of the outer edges of the double slip, and cold along the inner edges between the two branches, or cold along the outer edges and heat along the inner edges. When the ends were connected with the electrodes of the galvanometer, in the case illustrated in the diagram, the current was in the direction indicated by the arrow-heads; and it was always in such a direction, that if a zigzag line be traced through the two slips from side to side of each, on the whole in the same direction as the current, the changes of direction at the sides of the slips are from transversely to longitudinally magnetized through hot, and from longitudinally to transversely magnetized through cold; which is the conclusion that was anticipated. 140. I also experimented on the thermo-electric effects of retained magnetism in steel after the magnetizing force is removed, and obtained very decided results, showing that at least in the case of magnetization along the lines of current, the effect is of the same quality as in soft iron or in the steel itself while under a magnetic force which induces such a state of magnetization. 141. In one of these experiments, thirty-nine pieces of steel wire, each about $\frac{1}{8}$th of an inch diameter and 2 inches long, soft tempered, were connected by thirty-eight pieces of copper wire, each an inch long, placed between each two of the pieces of steel, and hard soldered to their ends. Pieces of copper wire of the same length were soldered to the outer ends of the first and last pieces of steel, and several feet of steel wire to the ends of each of these. A little electro-dynamic helix was made, 2 inches long and wide enough internally to slide freely over this compound steel and copper conductor; and by means of it every second piece of the 2-inch steel wires, commencing with the first and ending with the thirty-ninth, were magnetized alternately with their poles in dissimilar directions, while the other short wires, and the longer steel terminals, were left as free from magnetism as possible. The magnetizing helix was then removed, and the compound conductor was made into a flat coil on a wooden core (2 inches broad and \(\frac{1}{4}\)-inch thick), by bending the short copper wires, and arranging the 2-inch steel wires alternately on the two sides of the wood. The terminals were joined, with the usual precautions (\$ 92), to the galvanometer electrodes, and one edge of the coil was immersed nearly an inch below the surface of a vessel of oil at the temperature of about 100° Cent. Immediately a strong deflection of the needle showed a current, of which the direction in the coil was from unmagnetized to magnetized through hot. When the other edge of the coil was similarly heated, a contrary deflection of the needle as decidedly showed the same thermo-electric difference of quality between the magnetized and the unmagnetized steel wires. 142. The object of the peculiar arrangement just described, was to prevent the magnetism from spreading to those of the steel portions of the circuit which were to be kept as free from magnetism as possible in order to be compared with those which were magnetized. The introduction of the connecting pieces of a different metal from steel into the circuit cannot give rise to any thermo-electric disturbances*, provided the two ends of each are at the same temperature, a condition which was nearly enough fulfilled in the way the experiment was made, and which was very much favoured by the shortness and the high thermal conductivity of the little copper arcs. The same result was demonstrated in an experiment made with a homogeneous coil of steel wire, of which parts had been magnetized, by ordinary steel magnets, before it was bent on the core. [\$ 143. Received May 10, 1856.] § 143. Experiment.—On the Effect of Magnetization on the Thermo-electric Quality of Nickel. Through the kindness of Dr. George Wilson, I have been able to experiment on a bar of nickel, about \(\frac{1}{2}\) an inch in diameter and about 8 inches long, in the form of a horse-shoe magnet, belonging to the Industrial Museum of Edinburgh. The accompanying sketch and description show the plan of the experiment. **Description of Sketch.** N, nickel horse-shoe. B B, double tubes of sheet copper, electrically connected with one another by a copper band, and insulated from the nickel by silk paper, laid on with shell-lac varnish; serving to drain all electrical leakage from the magnetizing * Dynamical Theory of Heat, § 138, Cor. 1. coil, without causing the slightest sensible current through the nickel, and serving also to convey a stream of cold water to maintain the lower parts of the two branches of the horse-shoe at as nearly as possible equal temperatures. Fig. 37. A A A, india-rubber pipes to lead a stream of cold water through the coolers. C, magnetizing coil, wrapped on one of the copper coolers. E E, electrodes of magnetizing battery of twenty iron cells, charged with nitric acid, &c. F, commutator for interrupting and reversing the connexion between the magnetizing battery and coil, or reversing the current. M M, mercury cups, in which the extremities of the nickel were immersed (mercury being both very convenient for the purpose, and the metal least thermoelectrically removed from nickel of all that have been tried by any experimenter). m m, mercury electrodes joining copper galvanometer electrodes D D, at G G. K, commutator for interrupting and reversing the connexions of the galvanometer electrodes. Heat was applied at H H by means of a gas-lamp and blowpipe. A current from magnetized to unmagnetized through hot, was indicated by a considerable galvanometer effect, which, by management of the galvanometer break, K, was readily directed to give oscillations of the needle through three or four degrees. The same conclusion had been indicated in several previous attempts, with various defects of arrangement remedied in the experiment just described. In this last experiment the result was made most manifest; and, being completely separated from all effects of induced currents (which were quite insensible), of electrical leakage, and of unequal heating of the junctions of mercury and nickel, and of the junctions of mercury and copper, was set beyond all doubt. I therefore conclude, that longitudinally magnetized nickel in a thermo-electric circuit deviates from nickel not under magnetizing force, in the same direction as bismuth. This is the reverse of the deviation which I formerly found to be produced in iron by longitudinal magnetization. 144. The results of the various experiments which have been described in Part III. are collected in the following Tables. **Table I.—Effects of Stresses and Strains on the Thermo-electric Qualities of Metals.** | Description of Conductor. | Thermo-electric Order reckoned from Bismuth towards Antimony. | |---------------------------|-------------------------------------------------------------| | Iron | Free ........................................................................... Under longitudinal traction. | | Iron | Free ........................................................................... Under transverse compression. | | Iron | Under transverse traction ....................................... Under longitudinal traction. | | Iron | Permanently strained by longitudinal traction, and left free from stress. Soft ................................................ Permanently strained by longitudinal compression, or by lateral extension, and left free from stress. | | Iron | Hardened by transverse hammering ................................ Soft ................................................ Hardened by longitudinal hammering. | | Round iron wires of different diameters. | Made brittle by twisting ........................................ Annealed after being made brittle by twisting. | | Round and flattened iron wires. | Suddenly cooled ...................................................... Annealed. | | Steel wire | Some specimens flattened by transverse hammering. ............ Soft ................................................ Other specimens flattened by transverse hammering. | | Round and flattened steel wires. | Hardened by sudden cooling ...................................... Annealed. | | Steel watch-spring | Hardened by sudden cooling ...................................... Annealed. | | Copper | Under longitudinal traction ..................................... Free. | | Copper | .............................................................................. Soft ................................................ Permanently elongated by longitudinal traction, and left free from stress. | | Copper | .............................................................................. Soft ................................................ Hammered transversely. | | Round copper wire | Annealed after being made brittle by twisting. ................ Made brittle by twisting. | | Round copper wire | .............................................................................. Annealed .............................................. Suddenly cooled. | | Platinum | Under longitudinal traction ..................................... Free. | | Platinum | .............................................................................. Soft ................................................ Hammered transversely. | | Tin | .............................................................................. Soft ................................................ Permanently elongated by longitudinal traction, and left free from stress. | | Tin | .............................................................................. Soft ................................................ Hammered transversely. | | Brass | .............................................................................. Soft ................................................ Hammered transversely. | | Round brass wire | Annealed .............................................................. Suddenly cooled. | | Cadmium | .............................................................................. Soft ................................................ Hammered transversely. | | Lead | .............................................................................. Soft ................................................ Hammered transversely. | | Zinc | .............................................................................. Soft ................................................ Hammered transversely. | Table II.—Effects of Magnetism on the Thermo-electric Qualities of Iron and Nickel. | Description of Conductor | Thermo-electric Order reckoned from Bismuth towards Antimony. | |--------------------------|-------------------------------------------------------------| | Iron | Under transverse magnetizing force ... Free .................. Under longitudinal magnetizing force. | | Steel | ........................................................ Unmagnetized ........ Retaining longitudinal magnetization. | | Nickel | Under longitudinal magnetizing force. Free. | Part IV. METHODS FOR COMPARING AND DETERMINING GALVANIC RESISTANCES, ILLUSTRATED BY PRELIMINARY EXPERIMENTS ON THE EFFECTS OF TENSION AND OF MAGNETIZATION ON THE ELECTRIC CONDUCTIVITY OF METALS. 145. In endeavouring to discover the effects of magnetization and of mechanical strain on the electric conductivity of iron and other metals, I was led, from trying various more or less obvious methods for testing resistances, to use a differential galvanometer of a very simple kind, which I constructed for the purpose. I shall give no description of this instrument, as I now (Nov. 1856) find it in one important quality inferior to the differential galvanometer first constructed and used by M. Becquerel*, and I do not know that its peculiarity has compensating advantages. I mention it only because it was with it that I made nearly the first of my trials to find the effects of magnetism on the electric conductivity of iron, and the very first by which I obtained a decided result. 146. In these experiments I used two covered iron wires, each several yards long, coiled into circles about 4 inches diameter, as the two resistance branches in the divided channel through the two conductors of the galvanometer. Magnetizing one of them tangentially by means of a coil of covered copper wire wound on a copper sheath soldered round it as an electric drain, I ascertained, on the 23rd of April, 1855, that the electric conductivity of iron wire is diminished by longitudinal magnetization. The arrangement however proved, as I anticipated, to be of a very unsatisfactory kind; and the needle kept moving across the field in one direction almost steadily, during the whole time the current was sustained through the tested conductors, which was for several hours. Continually more and more resistance had to be added to the conducting channel containing the iron wire round which there was no magnetizing coil, to keep the needle within range. After the magnetizing current had passed for some time, this variation of the needle went on more rapidly, and called for more frequent adjustment by the additions to the other branch. All this was just as must be expected; and my reason for not introducing currents of cold water round the two iron coils, to maintain them in precisely similar thermal circumstances, was that the tubular systems required for the purpose could not be easily made, and that I thought I might find out the nature of the result in the first * Annales de Chimie et de Physique, tome xvii. 1846. instance, notwithstanding the imperfection of the arrangements. In this hope I was not disappointed. The glass needle (carried by the little suspended magnet, which was only about \( \frac{1}{2} \) an inch long), while moving steadily across its field, would receive an impulse forward and make two or three very rapidly diminishing oscillations, when the current was started through the magnetizing coil: when the current was suddenly reversed, the needle would show little or no indication of any effect: when the current was broken, it would make a start backwards, and after two or three oscillations would continue advancing as before, perhaps rather more rapidly. Traces of induced currents in the iron coil under the influence of the magnetizing helix were exhibited by scarcely perceptible differences in the bearing of the needle, according as the current was made in one direction or the other, and by slight impulses it received when the magnetizing current was suddenly reversed. After the current had been kept up for some hours through the iron wires, and when, partly by the heat developed by the magnetizing current during the periods of its flow, and partly by heat conducted from the iron wire within, the outside of the magnetizing coil had become very sensibly hot to the touch, the variation of the needle in the galvanometer became much less rapid than at first; and tolerably satisfactory indications, amounting to a fraction of a degree of permanent deflection, showed with perfect consistence an increase of resistance in the iron wire under magnetic force when the magnetic current was sustained in either direction, and a diminution of resistance in the same iron wire following immediately a cessation of the magnetizing current. 147. I followed the same method in a first attempt to find the effect of transverse magnetization on the electric conductivity of iron; two spirals made on the plan described above (\$ 136) being used as the resistance branches in the two channels conveying the divided current, and one of them placed between convex poles of a Ruhmkorff electro-magnet. The induced currents in making, reversing, and breaking the magnetizing current were of course most conspicuously indicated by the galvanometer needle, but the needle came to rest after a few oscillations; and then it did not exhibit any deviations of a sufficiently marked character, when the direct effect of the electro-magnet (which by a very troublesome process of shifting the position of the magnet, was reduced as much as possible in preliminary arrangements,) was eliminated by reversals, to allow me to draw any decided conclusion as to the effect of the magnetic force on the conductivity of the iron spiral across which it acted. 148. Before carrying into execution various obvious improvements in the experimental arrangements just described, or applying the system with the differential galvanometer to other investigations, I began to think of Maggi’s experiment* on the relative thermal conductivities of a magnetized iron disc in directions across and along the lines of magnetization. As the electrical analogue, the method which * De la Rive, ‘Electricity,’ vol. i. part 3. chap. iii. (p. 316, English edition, 1853). MATTEUCCI, and I believe KIRCHHOFF and others, have used in tracing equipotential lines on the surface of a conductor traversed by an electric current, occurred to me. Six months later, I thought of the multiplying branch (first used in the experiment described in § 161 below) to render available the sensibility which a powerful current through the body to be tested, with the use of a moderately sensitive galvanometer, must obviously give to that method when applied to the investigation of differential effects on the electric conductivity of a body in different directions; and I succeeded with great ease in making very satisfactory experiments (§§ 161 to 165 below) by means of it, which first decided the question as to whether or not the effects of magnetization give different electric conductivity in different directions to a mass of iron. At first, however, I did not see this or any other way to render the method practicable with galvanometer electrodes, either moveable upon the sheet of metal to be tested (in which case a motion of \(\frac{1}{100}\)th of an inch would drive the needle from an extreme deflection on one side to an extreme reverse deflection), or by electrodes soldered to points on an equipotential line (in which case a slight alteration in temperature in different parts of the plate might drive the needle irrecoverably to an extreme deflection on one side or the other); but the experiments which I knew as having been made by MATTEUCCI suggested to me the following very simple plan, which I immediately commenced trying, and which I have since found applicable with the greatest ease to a variety (I believe now to every variety) of experiments on electric conductivities*. 149. Let \(AB\) be the conductor to be tested, and let \(CD\) be another of nearly equal resistance, either a piece of the same wire continuous with the other through an arc \(BC\), or connected with it by a thicker arc of copper, or of another metal, as may appear convenient for the particular case treated. Sometimes the experiment is arranged to test differential effects experienced alternately or simultaneously by \(AB\) and \(CD\). But when one of them, \(AB\), alone is acted upon, with a view to varying its resistance, it alone may be regarded as the conductor which is tested; and the other, \(CD\), will then be called the reference conductor. Let a wire, \(AOPD\), which will be --- * [Note added Nov. 1856.] An hour before the meeting of the Royal Society at which this paper was read, I learned that a method of testing resistances had been given by Mr. WHEATSTONE which would probably be found to be the same in principle as that to which I had been led in the manner described in the text. I have since ascertained that Mr. WHEATSTONE's "Differential Resistance Measurer" (described in § 15 of the Bekenian Lecture for 1843, see Transactions, June 15, 1843) is an instrument founded on precisely the same principle as all the various arrangements by which, with great and necessary alterations of detail, I have continued the investigation of effects of magnetism and of other influences, on the electric conductivity of metals, to the present time, and of which some are fully described in Parts IV. and V. of the text. Mr. WHEATSTONE refers to "Experimental Determinations of the Laws of Magneto-electric Induction," printed in the Philosophical Transactions for 1833, "as containing the description of a differential arrangement of which the principle is the same as that on which" his own instrument has been devised, and adds, "To Mr. CHRISTIE must therefore be attributed the first idea of this useful and accurate method of measuring resistances." It is worth remarking, that the experiments of MATTEUCCI and KIRCHHOFF, alluded to in the text, are stated to have been first suggested from WHEATSTONE idea of applying the two electrodes of a galvanometer to points in separate channels through which two parts of the whole current from one battery are conducted. called the *testing conductor*, be soldered by its ends to the ends A and D of the conductor to be tested and of the reference conductor, or to strong pieces of metal to which those ends are firmly attached. Let one electrode of a galvanometer be soldered to the connecting arc BC, at its middle, or at any other point of it, Q; and let the other galvanometer electrode be ready to be applied by the hand to any position on the testing conductor. A current is then sent from one or more cells of Daniell's battery through electrodes connected with A and D. This current flows through the divided channel ABCD and AP'OPD, in quantities inversely proportional to the resistances of the two parts. The moveable galvanometer electrode is then applied, first to one point and then to another of the testing conductor (care being taken not to reverse, nor even to diminish, the magnetism of the lower needle in the astatic system of the galvanometer*), until by trial the point O that may be touched without producing any deflection in the needle, is found. The influence to be tested, whether it be magnetization, or tension, or elevation of temperature, is then applied to AB, or the influences to be tested against one another are applied to AB and CD, and the moveable galvanometer electrode is (if it has been removed) again applied at O. If the needle remains undisturbed, no effect is indicated; that is, no alteration in the resistance of ABQ, or only an alteration in the same proportion as an alteration experienced by QCD, has been indicated. If, however, a deflection is observed, in such a direction that the moveable electrode must be moved to some point P in the part OD, it is inferred that the ratio of the resistance of ABQ to that of QCD has been increased; or on the other hand, if such a deflection as requires a motion of the moveable electrode to a point P' in OA, the resistance of AB has been diminished relatively to that of CD. 150. As an example, I shall describe an experiment on the relative effects of tension on electric conductivity in copper and iron wires. Two pieces of stout copper wire, A, D, were each twisted into a loop which was made fast by solder; a couple of inches towards one end of each wire being left free from the twisted part. These loops were put upon a strong hard wood peg about \( \frac{3}{4} \) of an inch diameter, at a distance of about \( \frac{1}{4} \) of an inch from one another; and to their lower ends were firmly soldered fine iron and copper wires (strong enough --- * In the galvanometers which I have used, the two needles of the astatic combination are of similar material (pieces of the same steel wire, tempered brittle), and the lower one is a little longer (perhaps by about \( \frac{1}{6} \)) than the upper. Both are magnetized to saturation, and consequently the lower preponderates and gives its to bear weights of about 8 lbs. and 5 lbs. respectively). These wires were cut to the same length of $4\frac{1}{2}$ feet, and their lower ends were put into slits about $\frac{1}{4}$ of an inch deep, cut in the top of a piece of stout copper slip of the form and dimensions shown in the diagram, and the copper pressed upon them to hold them fast by a pair of pincers. Solder was then applied, to make a complete and compact metallic connexion between the wires and the copper piece. A testing conductor, consisting of seven yards of No. 18 copper wire, was soldered by its ends to the upper copper pieces A, D; and a current from six small cells of Daniell's was sent through the double channel by electrodes soldered a little higher up to the same copper pieces, A, D. One galvanometer electrode was soldered to the lower copper piece, and the other was applied to the testing conductor till the point O, equipotential with the point of attachment of the former, was found. As from previous experiments I knew that an accidental variation of $\frac{1}{100}$th of an inch in the position of the moveable electrode on the testing conductor might lose or overbalance the effect looked for, I added a multiplying branch, TFO'EU, consisting of a yard of No. 18 copper wire, with its ends soldered about half an inch on each side of O. This, of course, when touched by the moveable electrode, gave about thirty-six times the motion that would be required to produce or to correct any effect on the galvanometer if the simple testing conductor were used. The point O', on the multiplying branch, that could be touched without giving any deflection was then found; and weights were hung from the lower end of the lower copper piece, so as to stretch the copper and iron wire equally. Immediately a deflection of the needle in the galvanometer showed a current. This was corrected by sliding the moveable electrode on the multiplying branch towards U, that is, towards the parts conterminous with the copper wire. When the weights were removed, immediately a reverse deflection was observed. The conclusion is, that iron and copper wire equally extended have their resistances altered differently when under the stress; that of the iron wire being more increased, should the absolute effect in each wire be an augmentation of resistance, as other experiments I have made give me reason to suppose it is, or less diminished should it turn out that the absolute effect in each wire is a diminution of resistance. 151. Again, a heavier weight was applied so as permanently to elongate the wires. direction to the system. The strongest current through the coil only confirms the required state of magnetization, provided when it is started the index is either at zero, or on the side of zero towards which the deflection is to be. If by accident a powerful current is admitted through the coil when the index is on the wrong side of zero, the lower needle has its magnetism instantaneously reversed; but it may be as instantaneously put right again by suddenly reversing the current. If at any time, from the lower needle having either lost magnetic moment, or acquired a reverse magnetization, the astatic system is found reversed, it may be put in order with ease either by simply sending a powerful current through its coil, or by doing so and then suddenly reversing the current. The deflection, which was in the same direction as at first, was noted, but not corrected by any motion of the moveable electrode, and the weight was again removed. The needle returned towards zero, but remained deviating in the same direction as it had done to a greater degree with the weight on. By applying the hand instead of weights and gradually pulling down the lower copper piece, at first slowly, and afterwards rather faster, the needle could be made to deviate to $7^\circ$ and kept steadily there. After the wires had been stretched by rather more than an inch, the hand was removed with a gradual diminution of stress, which could easily be regulated to let the needle down without oscillation to whatever position it would rest in, with the stress entirely off. This in several repetitions of the experiment on the same wires was found to be somewhere about $3^\circ$ or $4^\circ$ in the same direction as the deviation which was kept at $7^\circ$ for a few seconds during the stress. Hence it was further concluded, that, as regards electric conductivity of the substance, the effect of permanent elongation, remaining after the stress is removed, differed between iron and copper in the same way as the effect of longitudinal stress during its action; that is, that the galvanic resistance of iron is more increased by permanent elongation than that of copper. Irregular variations to a considerable extent, obviously due to thermo-electric effects from the copper and iron in the compound conducting circuit, made me not attempt to measure with much care, the distance the moveable electrode had to be shifted to counteract the effects of tension; but I intend repeating the experiment and making it for other pairs of metals, with this source of irregularity removed by a modification of the testing conductor. 152. In the kind of experiment which has been described, the channels through the two metals experienced exactly the same elongation, and, it may be said without committing any sensible error, the same narrowing, by the longitudinal extension. The effect observed, therefore, depends truly on variations in the conductivities of their substance. I had made previously various experiments on copper wire alone, and on iron wire alone, in which I attempted to eliminate the effects of elongation and narrowing, and had very nearly established, for the case of iron wire at least, that the augmented resistance due to tension, either temporary or permanent, is a very little more than can be accounted for by the change of form. As, however, I have other experiments in progress, by which I hope to be able to show for a single metal the absolute effect on its specific conductivity separated perfectly from any influence on the resistance of the conductor occasioned by a change of its form, I defer in the meantime giving more details of investigation on this subject. 153. The method which has now been described has many great advantages over that by the differential galvanometer, or any other that I know of for testing or measuring galvanic resistances. In the first place, the irregularities, dependent on the electrodes, connexions, and circular conductors, of the differential galvanometer, are entirely done away with, and only the tested and the testing conductors, all connected by compact solderings, can influence the indication from which the results are to be drawn. In the second place, the galvanometer circuit may be broken and completed, and reversed, as often as is desired, by its own commutator, without affecting to the slightest sensible degree, the strength of the current through the tested and testing branches; while in the former mode of experimenting the indicating needle was always under the action of the divided current, unless the current in one or the other of the branches was broken, which introduced irregularities lasting for a considerable time, by the consequent changes of temperature through the conductors. This was an immense convenience in every experiment, and allowed small deflections, amounting to the tenth of a degree, to be tested with ease by using the commutator of the galvanometer, and getting oscillations. But it was of especial advantage in the experiments on the effects of transverse magnetization, since the galvanometer circuit had only to be kept broken for a few seconds during the making, breaking, or reversing of the magnetizing current, to get entirely rid of all disturbances of the needle due to induced currents; and in all experiments in which the Ruhmkorff magnet was used, since by breaking the galvanometer circuit and using a little steel magnet in the hand, the galvanometer needle could be let down in a few seconds into its position as affected by the direct action of the large magnet, before proceeding to test the current due to the change of resistance under investigation. In the third place, it is possessed of almost unlimited capacity for increase of sensibility. In some of the experiments on the influence of tension on electric conductivity, I have tested with the greatest ease effects amounting to only $\frac{1}{15000}$th of the whole resistance of the wire under examination, and I see no difficulty in testing effects amounting to only the tenth part of that, or even hundreds of times smaller effects, by using more powerful currents, and applying artificial means to keep the wires cool. PART V. ON THE EFFECTS OF MAGNETIZATION ON THE ELECTRIC CONDUCTIVITY OF METALS. 154. The remarkable effects which I found produced in the thermo-electric quality of a metal by magnetization and by mechanical strain, appeared to render it highly probable that the same agencies would also influence their electric conductivities. To demonstrate this if I could, and to discover the nature of the anticipated effects, I commenced an experimental investigation of the subject, and, after various nugatory operations, arrived at a variety of positive results by the following processes. 155. Exp. 1. On the longitudinal electric conductivity of longitudinally magnetized iron wire.—A length of seventy-two yards of silk-covered copper wire was rolled in six strands, or altogether in about 860 turns on a core made up of two concentric brass tubes, connected at their ends by a ring of sheet brass, and arranged to have water sent through the space between them by suitable entrance and exit pipes soldered to apertures in the outer one; the external diameter of the brass tube was about $\frac{5}{12}$ inch, and the internal diameter of the inner one about $\frac{1}{8}$ inch; the metal of both outer and inner tubes being as thin and as well smoothed as it could be got. The piece of iron wire to be tested was soldered at one end to a piece of thick copper wire, and then insulated by a thin coating of writing-paper, wrapped twice round it, and pushed into the inner brass tube, which was just large enough to admit it easily. A second iron wire of equal dimensions was similarly prepared and inserted in a second core, in all respects like the other, except that in this experiment it had no copper wire wrapped round it. The two cores being laid side by side, the free ends of the iron wires were connected as shown in the diagram, by an arc of thick copper wire, C, soldered to them. A current from a single large cell of Daniell's was admitted and carried off by the electrodes A and B. Cold water was kept constantly flowing through the spaces between the concentric brass tubes round the iron wires. The testing conductor (§149) used in this experiment consisted principally of the following parts:—(1) Two pieces of No. 18 copper wire, each sixteen yards long, prevented from touching one another by a piece of twine between them, rolled together on a thin copper cylinder, 12 inches long and 3 inches diameter, from which they were insulated by a coating of two folds of silk cloth sewed round it. (2) Soldered to two of their contiguous ends, a connecting arc of thick copper wire, which was at first intended to be gradu- Fig. 41. ated, and will be called the scale of the testing conductor. (3) Separate short thick wires soldered to the other ends of the wires coiled on the copper cylinder, to bear binding screws for making connexions with the electrodes A and B of the conductor to be tested. One electrode of the galvanometer was soldered to the middle of the connecting arc between the two iron wires, and the other was held in the hand, and applied about the middle of the scale of the testing conductor. A rather troublesome process was then required to bring the galvanometer to zero by adding resistance on one side or the other between the ends of the testing conductor and A or B. When this was done, it was found that great deviations of the galvanometer needle were produced by sliding its moveable electrode a few inches in either way on the scale, and a perfectly sensible deflection by sliding it as much as \( \frac{1}{8} \)th of an inch. The point of the testing scale to which the moveable electrode had to be brought, to give no deflection of the galvanometer, was determined: the circuit of the galvanometer was broken, and a current from six of the small iron cells was sent through the magnetizing coil. Immediately on completing the galvanometer circuit again, with its electrode held on the same point of the testing scale as before, a very considerable deflection was observed. On breaking the galvanometer circuit, reversing the magnetizing current, and completing the galvanometer circuit again, the same deflection was observed; and when the magnetizing current was stopped the galvanometer again gave zero, or nearly so. On repeating the process as regards the magnetizing current, without breaking the galvanometer circuit, the same deflection was always observed, in whichever direction the current was sent through the magnetizing coil; and little or no either instantaneous or permanent effect was produced on suddenly reversing this current. It was found that the deflection occasioned by the magnetization was diminished by sliding the moveable electrode along the scale from its end communicating with B, towards its end communicating with A, and was corrected by such a motion through a space of about $\frac{3}{4}$ths of an inch; equivalent to $\frac{1}{16}$th of an inch of the No. 18 wire, constituting the chief part of the testing conductor. It was concluded that the iron wire had its electric resistance increased by magnetization, and that this augmentation amounted, in the particular experiment, to about $\frac{1}{3000}$ of the whole resistance of the magnetized piece. 156. Exp. 2. On the effect of permanent magnetization on the electric conductivity of steel wire.—The same apparatus as in Experiment 1 was used, and was in all respects similarly arranged, except that hardened steel wires as free from magnetism as possible were substituted in place of the soft iron cores in the brass tubes. On bringing the galvanometer to zero and sending a current through the magnetizing coil, the same deviation as before was observed, and a much smaller deviation in the same direction remained after the magnetizing current ceased. This experiment was repeated several times on fresh unmagnetic steel cores, and always with the same result. I concluded that steel when subjected to magnetic influence has, like iron, its electric conductivity diminished in the direction of the lines of force; and that it retains some of the same effect with the permanent magnetism subsisting after the magnetizing force is removed. At the same time I was not quite satisfied with the experiment, as the galvanometer needle was never very steady, and, to keep it about zero, the moveable electrode had to be shifted largely along the scale, sometimes quite to one end, when, to get it on the scale again, additional adjustment wires had to be added to the other branch of the testing conductor. This prevented me from using more powerful currents through the wires to be tested and so getting larger indications of the results; but I determined if possible to repeat the experiment afterwards with arrangements better adapted to do away with all variations in the conductivity of the circuit except those under investigation. I still keep it in view to do so, and I have no doubt now of being able to get rid of all the unsteadiness which I had found so troublesome. 157. Exp. 3. Attempt to discover the effect of transverse magnetization on the longitudinal conductivity of a slip of sheet iron.—Two brass cores like those described above, and of the same length (10 inches), but of larger inner and outer diameters, were prepared, and a quantity of covered copper wire rolled on one of them in four strands, or in all 570 turns. Two slips of sheet iron, each 7 feet long and \( \frac{1}{8} \) inch broad, were wound upon single brass tubes coated with paper, and the successive spires of each were kept from contact by a piece of twine wound on between them. A length of 9 inches of each brass tube had 84 inches of the slip iron laid upon it, and therefore the inclination of the helix to a plane perpendicular to its axis was about 6°, being the angle whose sine is \( \frac{9}{84} \). Each of these iron spirals was protected outside with a coating of paper, and pushed into the interior of one of the brass cores. A copper arc, C, was soldered to each of them so as to connect their extremities on one side, and powerful copper electrodes, A and B, were soldered to their other extremities. Then, a stream of water being kept constantly flowing through each of the inner tubes and through the spaces between the concentric brass tubes outside, a current from a large cell of Daniell's (\$ 63) (exposing 2·5 square feet of zinc to 4·4 square feet of copper) was sent through the iron spirals, and a testing conductor (the same one as before) was put in communication with their electrodes, A and B. One electrode of the galvanometer being, as before, soldered to the middle of the copper arc connecting the iron spirals, the other was applied to the scale of the testing conductor. The galvanometer being brought to zero by the insertion of adjustment wires at one end or other of the testing conductor, it was found to be rather steadier than in the former experiments, probably because of the diminution of thermal effects by the stream of water through the cores, and the greater surface of iron exposed outside and inside to refrigeration. When a current was sent and maintained through the magnetizing helix, a very decided permanent deflection was occasioned in the galvanometer; and this the same with each direction of the magnetizing current. If the galvanometer circuit was kept complete, its needle experienced a powerful impulse, sending it through a great many degrees in one direction or the other at the instant of starting, or of reversing, or of stopping the magnetizing current, but quickly in each case showed the nature of the permanent deflection by oscillating about one position, when the current was steadily maintained, in either direction. These impulsive deflections were of course due to induced currents, and were entirely prevented by keeping the galvanometer... circuit broken during the starting, the reversal, or the stoppage of the current through the coil of the electro-magnet. 158. The deflection due to the effect of magnetic force on the substance of the iron was corrected in each case by sliding the moveable electrode towards the part of the testing scale remote from the end connected with the iron spiral which experienced that effect; and it therefore indicated a diminution of conductivity in the iron. 159. If the lines of magnetization had been exactly perpendicular to the lines of electric current through the iron, we should now conclude that transverse magnetization diminishes the conductivity of an iron conductor; that is, that it produces the same kind of effect on the conductivity as longitudinal magnetization. But the lines of current formed spirals inclined at an angle of $84^\circ$ to the lines of the magnetizing force; and the mutual influence of the consecutive parts of the magnetized iron spiral would have an effect (not wholly compensated by the mutual influences between the successive spires because of the thickness of the twine between them,) contributing to longitudinal magnetization; and therefore the lines of magnetization must have been inclined, not at $90^\circ$, but at some angle less than $84^\circ$, to the direction of the lines of current. Hence all we can conclude is, that not only longitudinal magnetization but oblique magnetization up to some angle of obliquity less than $84^\circ$ from the lines of current, diminishes the electric conductivity of iron. 160. It remains to be determined by experiment what is the effect of magnetization right across the lines of current: if a diminution of conductivity, whether a greater or a less diminution than is caused by an equal longitudinal magnetization? or if it is an increase of conductivity, what is the angle of obliquity of the magnetization which gives neither increase nor diminution of conductivity? 161. Exp. 4. To discover the differential effect of magnetization on the conductivity of iron in different directions.—A square of $1\frac{1}{2}$ inch each side was cut from thin sheet metal, and powerful electrodes were soldered to two corners, A and B. A reference electrode ($§$ 149) of No. 18 copper wire was soldered to C, one of the other corners, and the two extremities of a yard of the same kind of wire, to be used as a multiplying branch, were soldered to points D, E, about $\frac{1}{15}$th of an inch from one another on each side of the remaining corner. A current being conducted through the square by the principal electrodes A and B, the reference electrode was used to connect C permanently with the commutator belonging to the testing galvanometer. Another wire used as a testing electrode, was applied to connect any point of the plate, or of the multiplying branch, with the other galvanometer electrode. In the first place, it was found that a powerful current was raised in the galvanometer coil if the testing electrode was applied to any point of the multiplying branch; and it was necessary therefore, as was anticipated, to adjust the distribution of resistance through the square by filing, so that there might be some point on the testing branch which would give no current when touched by the testing electrode. (See below, § 176, where a less troublesome way of managing this part of the arrangement, in an analogous experiment, is described.) For this purpose, in the first place the testing electrode was applied at different places along the edges BD, EA of the square till a point was found which gave no deflection of the galvano- Fig. 43. meter. If this was in BD, the plate had to be thinned in its middle parts parallel to CA and BD, or else to be thinned along the edges CB, AD, so as to increase the resistance to conduction parallel to the last-mentioned edges. Or if the neutral point was in EA, the plate had to be thinned in its middle parts parallel to CB and AD, or along its edges BD, CA. By using the file according to these directions, after a few trials the neutral point was brought upon the testing branch; that is to say, the resistance was so adjusted in the square that the line from C cutting right across the lines of conduction, or which is the same thing, the equi-potential line through C, passed between D and E. A piece of sheet copper as broad as the iron square, but rather longer, was bent as shown in the diagram, so as to give a depressed space in which the iron, insulated from the copper simply by a piece of writing-paper, could rest steadily. This copper cradle was placed resting on the flat poles of a Ruhmkorff electro-magnet, which were pushed together so as to hold it firmly. Any leakage of electric currents from the coils of the electro-magnet was thus effectually drained by the copper, so that a simple sheet of paper was quite enough to do away with all sensible indications of currents in the iron acquired otherwise than through the electrodes A and B. [This electrical drainage would be made more nearly perfect by using paper or some other non-conductor to separate the cradle from the poles of the magnet.] 162. A large single element of Daniell's (§ 63), consisting of seven zinc plates in seven porous cells, contained in four large wooden cells, and exposing in all 8·75 square feet of zinc surface to 15·3 square feet of copper, was then used to send a current through the iron square, insulated between the poles of the electro-magnet, in the manner described. 163. The neutral point on the testing branch being got by trial, it was found to remain tolerably steady, although no doubt during the first minutes of the flow of the current it may have varied much, as the iron got heated, which it soon did to a degree very sensible to the touch. Moving the electrode along the testing branch through a quarter of an inch on either side of the neutral point, gave a very marked deflection of the galvanometer. The galvanometer circuit was then broken, and a current from six of the small iron cells was started through the coils of the electro-magnet. When the galvanometer circuit was again, after a few seconds, closed, with its electrode on the same point of the multiplying branch as before, a very considerable deflection was observed in the needle. To correct this deflection and bring the needle to zero, the testing electrode had to be moved to a position 2 or 3 inches nearer D on the testing branch. 164. The new neutral point was unchanged when the electro-magnet was reversed, and when the magnetizing current was broken there was a permanent deflection in the galvanometer the reverse of that observed when the current was started in either direction. If the galvanometer circuit was completed within a second or two of any of the changes in the magnetizing current, the needle experienced, obviously from induced currents, powerful impulses in one direction or the other, according to the direction of the current made or unmade through coils of the electro-magnet. But in every case, although from various disturbing causes the neutral points gradually shifted largely along the testing branch, the permanent effects of making and of unmaking the electro-magnet were most marked, and were uniformly as stated above. 165. Thus it appears that magnetization shifts the equipotential line through C from its position running across to the opposite corner, to a position (dotted in the diagram) a little nearer CB; so much so that its end is shifted about $\frac{3}{35} \times \frac{1}{35}$, or $\frac{1}{240}$th of an inch from E towards D. This shows that the passage of electricity in the directions AE and CB has become less resisted than it was, relatively to the passage in the directions AC, DB; and it therefore follows that the electric conductivity of magnetized iron is greater across than along the lines of magnetization. 166. Still, as the preceding experiment (Exp. 3) had appeared (§ 159) to show that the absolute conductivity is diminished in all directions by magnetization, it seemed possible that the effect now observed might be caused by inequalities in the distribution of magnetism in the plate. Thus if from the character of the distribution of the magnetizing force, or because of non-uniformity in the plate, the parts between C and B and between A and E were less intensely magnetized, and those between C and A and between B and D more intensely magnetized, than the average, the observed effect could be accounted for without any difference in the electric conductivity of the substance in the different directions. To test this conceivable explanation, pieces of soft iron (cubes and little square bars nearly double cubes) were laid over the square plate, being kept insulated from electric communication with it by paper, so that while the conducting mass remained unchanged, the distribution of the magnetization of its substance might be altered. Before the magnetic force was applied, a great effect on the neutral point of the multiplying branch was observed, taking place gradually during several minutes, and obviously due, in a great measure, to variation of the distribution of temperature in the conducting square. (See below, § 177, for an illustration of this effect.) When a new neutral point was found, the magnet was made, reversed, unmade, &c., and always with the same effects as before. Different arrangements of the little masses of soft iron produced different absolute effects on the neutral point, causing it to shift sometimes as much as fifteen inches on the multiplying branch, but the effects of magnetism were invariably found to be consistent with the first-mentioned result. As the distribution of the magnetism in the square plate must have varied very much under these different circumstances, and in all probability must have been in some of the cases more intense in the quarters towards AE and CB than in those towards AC and DB, the conclusion could scarcely be avoided, that the conductivity of the magnetized substance was greater across than along the lines of magnetization. For the purpose of further testing and illustrating this conclusion I planned the following experiment, to compare directly the resistances of two equal and similar squares of sheet iron, equally and similarly magnetized, arranged in the same circuit to conduct electricity across the lines of magnetization of one and along those of the other. 167. Exp. 5. To compare the conductivities of magnetized iron along and across the lines of magnetization.—A piece of sheet copper, BCHK (fig. 48), 3 inches long, 2 broad and \( \frac{1}{2} \) inch thick, was bent round the line FH into the form shown in fig. 49. Fig. 48. Fig. 49. Fig. 50. A square of thin sheet iron, 2 inches wide (weighing 103 grains), was soldered by one side to the edge CH of the copper in the position shown in fig. 50. The pro- jecting part, FBKL, of the copper slip was bent round its middle line EG, so as to bring its edge, BK, close over the edge of the iron square lying over FC; and to this edge, BK, in its new position (fig. 51), a second iron square, of the same dimensions and weight as the other, was soldered by one side, with its area lying in a position close to that of the former. The relative position of the two squares and the connecting piece of copper will be understood by looking at fig. 52, which represents the  iron squares as if soldered to the piece of copper before it was bent, and the iron square CDMH turned round its side CH, from the position close to the plane of the copper adjoining it, into a position in this plane continued across CH. If, now, we suppose the iron square CDMH to be turned down so as to lie below a square, FL/HC, of the copper; this square of the copper to be bent sharp round its diagonal, FH, till the part FL'H lies over HCF; and, lastly, the part FL'KB projecting beyond FC to be bent downwards round EG with a less sharp bend; the iron square, ABKL, will be brought close under the other one, CDMH, with the edges of the two which are connected to the edges of the copper perpendicular to one another, and the whole compound conductor will have exactly the position shown below in fig. 54. 168. A convenient electrode was soldered along the edge of each iron square parallel to the edge of the same square soldered to the connecting piece of copper; so that a powerful electric current entering by one of those electrodes and carried away by the other would pass through the second-mentioned square of iron in lines exactly parallel to the side AB, through the connecting piece of copper in lines which were parallel to its length, BC, before it was bent; and through the first-mentioned square in lines exactly parallel to its side CD, and therefore perpendicular to the lines along which it traverses the second square. The course of the current will be understood by looking at fig. 52, where the two squares and the copper connecting them are supposed to be opened out so as to throw the course of the current into a straight line. The order followed in constructing the compound conductor was not exactly the order of the description given above; but the connecting piece of copper was first cut and bent, and other pieces, to serve as electrodes (shown in the accompanying views, fig. 52), were prepared, and the iron squares, put in their proper places, were then soldered by their edges to the edges of the connecting piece, and the electrodes were soldered to their opposite edges. A view of the whole thus put together, with the reference and testing wires described below, is given in fig. 54. A testing con- ductor of two yards of No. 18 copper wire was soldered, with its two extremities to the copper electrodes, close to the middle points of the edges AL, DM of the iron squares; and a fixed galvanometer electrode was soldered to the middle point, N, of the copper connecting piece. 169. The squares, their electrodes, the connecting piece, and the testing conductor being then guarded against irregular contacts by a little square of pasteboard pressed between the iron squares, a half-square of pasteboard between the first-mentioned iron square and the portion FCH of the connecting copper (see fig. 49), and fragments of paper and pasteboard elsewhere, the whole was placed, with the second-mentioned square lowest, in a copper cradle lined with paper, and resting between the horizontal edges of the flat poles of the Ruhmkorff electro-magnet used in the preceding experiment. 170. The positions of the magnetic poles of the squares, of the bent connecting piece of copper, of the testing conductor, and of the galvanometer electrodes are indicated in fig. 55, but, to avoid confusion, the principal electrodes are not shown. Fig. 55. A current from the four large double cells, connected so as to constitute in all a single element of Daniell's, exposing 10 square feet of zinc surface to $17\frac{1}{2}$ square feet of copper, was then introduced by the principal electrode soldered to the edge MD of the upper square, and drawn off by the other principal electrode, namely, that soldered to the edge of the lower square lying exactly below the edge MH of the upper. The course of the current into the principal channel between these electrodes would be across the upper square from MD to HC, and across the lower square from the edge below CD to that below HM; also, in the secondary channel between the same electrodes, from T soldered to the first through the testing conductor, to its other end U soldered to the second. 171. A fixed galvanometer electrode being (§ 168) soldered to the middle point, N, of the connecting-copper, the other electrode of the galvanometer was moved along the testing conductor till a point, O, was found at which it might be applied without giving any deflection. By moving it $\frac{1}{10}$th of an inch on either side of O very sensible deflections were obtained, and therefore a yard of copper wire was soldered by its ends to points S and Q a quarter of an inch on each side of O, and was used instead of the "scale" of the testing conductor described as used in the first three experiments. The neutral point, O', on this multiplying branch having been found, the galvanometer circuit was broken, and the electro-magnet was excited by six of the small iron cells. On closing the galvanometer circuit again immediately, a considerable deflection was observed, to correct which the moveable electrode had to be moved through about two or three inches from O' towards Q. On unmaking the electro-magnet a reverse deflection in the galvanometer was observed, and was corrected by bringing back the electrode to O'. The same result was obtained when the magnet was made in the reverse way, and never failed to appear, to an unmistakeable extent and with perfect consistency, after the operation had been repeated many times and varied in every possible way. 172. It showed that the effect of the magnetization was to increase the resistance relatively in the upper square of iron, and to diminish it relatively in the lower square. I concluded with confidence that the electric conductivity of magnetized iron is greater across than along the lines of magnetization. 173. Exp. 6. A double experiment, to test the absolute nature of the two effects of which the difference was shown in the preceding experiment.—A divided current from the battery was made to pass through the two squares by electrodes, of which one was soldered to the middle of the copper band connecting them, and the other clamped to the now united extremities of the bundles of copper wire which had served before to lead in and out the whole current in the preceding experiment. As testing conductor was used the same piece of copper wire which had served as the fixed galvanometer electrode in the preceding experiment, with its end which had been connected with the galvanometer now soldered to the junction of the two copper branches of the divided channel (the resistance of each of which was found to be nearly equal to that of the iron square with which it is connected). The testing wire used in the preceding experiment was cut in two, one part to serve as fixed galvanometer electrode in one, and the other in the other of the two experiments which it was intended next to make. I first attempted to test the effect on the conductivity of the upper of the two squares produced by the magnetization which in it is along the lines of current. I found, however, on fixing the copper wire proceeding from one side of that square to one electrode of the galvanometer, and applying the other to the testing conductor in the usual way, that the circumstances were constantly varying, and that the point to be touched to give no deflection shifted rapidly along the testing conductor. Hence I gave up this part of the experiment, of which the result might be anticipated with certainty from the experiment on the effect of magnetization along the line of current described above (Exp. 1. § 155), and I gave the whole time during which the experiment could be continued, to an examination of the influence of the electro-magnet on the current in the branch leading through the lower square across its lines of magnetization. Accordingly, the galvanometer electrode, which had been united to the part of the old testing conductor terminating in an edge of the upper square, was transferred to the other part of the old testing conductor, that is, to the part terminating in a side of the lower iron square. The same new testing conductor was still used; and as soon as a point could be found on it which gave no current when touched by the moveable galvanometer electrode, points about $\frac{1}{4}$ of an inch on each side of it were taken, and a multiplying branch of one yard No. 18 copper wire was soldered by its ends to them. Before, however, the effect of the magnetism could be decidedly tested, the zero-point had moved off the multiplying branch, which had accordingly to be shifted along the testing conductor to get into range again. The same process had to be gone through a great many times, and at last, after the current had been flowing continuously through the two squares and the divided copper channel for about five hours, the zero-point became sufficiently steady to remain on the multiplying branch when fixed at the right place on the testing conductor, and to allow a decisive experiment to be made. The result was a very slight effect, proving a diminution of resistance in the iron square. 174. The cause of the long-continued variation in the conditions of electric equilibrium between the testing conductor and the fixed point on the edge of the lower square, was clearly the gradual warming of the long copper wires extending up from this point, due to the conduction of heat generated in the iron squares by the electric current; and it would obviously be much diminished by using a simple form of conductor with only one iron square at a time, and with the reference conductor kept near it, so as to acquire quickly whatever temperature it would rest with during the flow of the current. I accordingly made the following experiment, choosing first the effect of transverse magnetization, as the experiment just described had not been of a satisfactory kind, although apparently conclusive, while the first experiment of the series (Exp. 1. § 155) had been less unsatisfactory in point of steadiness, and had led decisively to a conclusion regarding the effect of longitudinal magnetization on the resistance of a conductor. 175. Exp. 7. A square of sheet iron like those used in the last experiment (four square inches, weighing 103 grains, and consequently about \( \frac{1}{78} \)th of an inch thick,) was soldered along one edge to a slip of lead of the same width, about twice as thick and about one-half longer. To the opposite edge of the iron square was soldered a stout copper slip an inch broad and equal in length to the side. The piece of lead was bent round, so as to give a straight part lying about \( \frac{1}{4} \) of an inch from the plane of the iron, and to extend about as far as the copper slip soldered to the other edge of the square. A current from an arrangement of the cells (§§ 63 and 64) constituting a powerful single element of Daniell's was sent through the iron square and the lead band, by electrodes clamped to one end of the lead and to the copper slip fixed to the other edge of the iron. A point in the lead slip having been found, such that the galvanic resistance between it and the edge next the iron was nearly equal to the resistance in the iron square itself, a testing conductor (two yards of No. 18 copper) was soldered by one end to that point in the lead, and by its other end to the middle of the edge of the iron. square to which the copper slip is attached. A copper wire, to serve as fixed galvanometer electrode, was soldered to the lead band, at a point in the middle of its breadth close to its edge of attachment to the iron. A copper cradle was put between the flat poles of the electro-magnet, as before (see above, § 161), and covered with a piece of paper. The iron square was supported upon it in a position with the line joining the poles perpendicular to the line of the current through it. Then, the current being kept steadily flowing through the iron and lead band, a zero-point was found on the testing conductor, and a multiplying branch (one yard of No. 18 copper) was soldered with its ends \( \frac{1}{4} \) of an inch on each side of this point, in the usual way. The zero-point on this multiplying branch was almost immediately found, and continued on the whole very steady from the first. The galvanometer circuit being broken, a magnetizing current from six small iron cells was sent through the coils of the electro-magnet, and the needle of the galvanometer was let settle (as it could be in a few seconds by the aid of a little magnet held in the hand) into its position of equilibrium as affected by the direct force of the magnet. On completing the galvanometer circuit again, with its moveable electrode on the same point of the multiplying branch as before, a current was made sensible by an excessively slight deflection. The galvanometer circuit being broken, and the electro-magnet reversed, a similar deflection was found in the galvanometer on again completing its circuit. It ceased, as nearly as could be discovered, when the electro-magnet was unmade, and was uniformly observed when the magnet was made again either way, in a great many repetitions. The current indicated by the galvanometer when the magnet was made was always such as to be corrected by carrying the moveable electrode from its previous zero-point, along the multiplying branch, towards the part of the testing conductor terminating at the iron square, and therefore indicated an increase of conductivity in the iron. The effect was so very slight, that I could scarcely determine how much the moveable conductor had to be shifted to correct it. I intend to repeat the experiment with similar arrangements, but with two or three times as powerful a current through the electro-magnet, which ought to give about four or nine times the amount of effect. In the mean time, however, I am quite convinced that I have observed the true result, and I conclude that the electric conductivity of iron is increased by magnetic force across the lines of current. 176. Exp. 8. To show the variation of a line of electric equilibrium in a circular disc of iron conducting electricity between two opposite points of its circumference, when subjected to magnetic force in a direction at an angle of 45° to the line joining these points.—A circle 2·3 inches diameter was cut from a piece of sheet iron, and ground down to a thickness which must have been about \( \frac{1}{69} \)th of an inch, as the prepared disc was found to weigh 114 grains. Two stout copper electrodes were soldered to its circumference at opposite points. A point at 90° on the circumference from one of these was taken, and at about \( \frac{1}{46} \)th of an inch on each side of it were soldered the ends of a piece of No. 18 copper wire two yards long, to serve as a multiplying branch. The disc was put on a copper cradle covered with paper, supported between the flat poles of the Ruhmkorff electro-magnet, with the line joining its principal electrodes at an angle of 45° to the magnetic axes of the field, and a current from a large single element of Daniell's was sent through it by these electrodes. One electrode of the galvanometer was applied to the middle of the multiplying branch, and the other was moved about on the opposite parts of the circumference of the disc till a position giving no current was found, where it was then soldered. The moveable electrode applied to different points of the multiplying branch was then found to give sensible galvanometer indications with a motion of a quarter of an inch, and after a very short time the zero-point became tolerably steady. The electro-magnet was then made, with the galvanometer circuit broken, and when it was closed again a decided indication of a current was observed in the galvanometer. This current was checked by sliding the moveable electrode towards the end of the multiplying branch next the equatorial part of the magnetic field; and the conclusion was, that the conducting power of the plate, when magnetized, became greater across than along the lines of magnetization, which was confirmed by every repetition and variation of the experiment. Now it is obvious that the intensity of magnetization must have been on the whole greater in the parts of the disc next the poles: hence a diminution of conductivity across the lines of magnetization, to the same extent as that which we know from Experiment 1. exists along them, would give a contrary effect to that now observed; and it follows that the electric conductivity is in reality greater across than along the lines of magnetization in magnetized iron. 177. This experiment was witnessed by Mr. Joule, and afforded a full confirmation of the conclusion (§ 172) which had been established by Experiment 5. above, and which follows from Experiment 1. and Experiment 6., considered together. The effects of applying pieces of hot wood equatoreally or axially to the disc were very clearly observed, and were always similar to those described above (§ 166), indicating a greater resistance to the parts of the current crossing the hot region than to those passing through the comparatively cool parts of the iron.