"I hope, then, a result of the Committee's action may be to carry out an attempt of this kind for every class of variations for which the data give even the narrowest foundation. It might be applied, I believe, with success, as regards the main conclusion, to every case in which each of the three components has been well determined for even only THREE stations widely apart from one another.

"It seems probable that an individual deflection of a magnetic storm cannot be identified in localities at very great distances from one another. This must certainly be the case if an individual deflection, and individual flash or flicker of aurora, are simply related to one another, because the individual auroras are certainly local in the sense of being only seen at once over a very limited area of the earth, being in fact actually situated at some distance of not more than 150 miles (which I believe is the highest estimate) from the surface. Hence it is probable that it will be found whether the seat of the disturbing action, producing an individual deflection in a magnetic storm, is above or below the surface, by comparing observations made at stations within a few hundred miles of one another, and endeavouring to identify a single disturbance in the three components at all the localities. If the three components could thus be determined at three localities so wide apart as to show considerable differences in the amounts, but yet not so wide as to render the identification of the disturbance difficult, the question whether the seat of the disturbance is in the earth or the air would be answered with high probability. "I remain, yours very truly, "WILLIAM THOMSON."

(Signed)

On Thermo-electric Currents in Circuits of one Metal.
By FLEEMING JENKIN, Esq.

LAST year I had the honour of directing the attention of the Association to the fact, that an electric current of considerable intensity may be obtained in a circuit of one metal by the application of heat to one or the other side of an interruption in the wire composing the circuit. The experiment is most simply performed by looping together the two ends of two perfectly similar wires connected to the terminals of a galvanometer, and heating one of the loops to a white or red heat in a spirit-lamp, or Bunsen's burner. If the one loop rests very lightly on the other a current will be obtained, which in the copper wires will flow from the hot to the cold loop across the joint with sufficient intensity to deflect a moderately sensitive galvanometer, even with a resistance in circuit equal to 1000 miles of No. 16 copper wire.

The electromotive force of the combination is about one-tenth that of a Daniell's cell. With two iron loops a permanent current in the opposite direction is obtained, flowing from cold to hot across the joint, but the electromotive force in this case is very much smaller.

When the loops are drawn tightly together the current ceases, but reappears as soon as the strain is slackened.

I was at the time unable to show the connexion between these singular currents and other electrical phenomena, but I am now, in consequence of further experiments undertaken for the Association, able to point out that connexion.

The currents were clearly not due to chemical action on the wires; for, in the first place, currents of considerable strength were obtained from two perfectly homogeneous platinum wires, flowing from hot to cold across the loose contact; and in the second place, the direction of the current was different in copper and iron, whereas the chemical action undergone by the wire was alike in the two cases.

The researches of Becquerel, Pouillet, Buff, Hankel, and Grove were examined, to see whether the electricity produced during combustion, or the properties of flame, would account for the currents, but it was found that all the electrical effects produced by flame could be divided into two classes: first, phenomena depending on the relative position of the two wires in the flame; and secondly, phenomena depending on the voltaic couple formed by the metals used, and the hot vapour acting as an electrolyte between them. My results were independent of the position of the wires in the flame, and could not be accounted for by supposing these wires to form a voltaic couple, inasmuch as though in some cases, where wires of two metals were looped together as described, the current flowed from the metal most attacked across the imaginary electrolyte to the other wire, in other cases it flowed in the opposite direction.

It remained to be seen whether the currents might not have a thermoelectric origin. Last year I imagined that the effect observed might be directly due to discontinuity, but that idea was dispelled by some experiments with loose contacts between wires of different metals, which have thrown great light on the question.

Loops of iron, silver, platinum, gold, and copper wires were combined two by two in all the possible arrangements, and the currents measured which were obtained when one or the other or both loops were heated with loose and tight contacts between them.

A Table was thus formed, which is appended to the present paper. The resistance of the circuit was so large (2050 × 10, Weber's absolute foot

) that the inherent resistance of the joint and of the different short seconds wires used in each experiment could be neglected, and the deflections obtained on a reflecting galvanometer could be taken as approximatively proportional to the electromotive force of each combination. The common thermo-electric currents produced by the metallic contact between dissimilar wires almost vanish in comparison with those produced by the loose contacts. I need not present a complete analysis of the Table, but will speak only of the combination of iron and copper with which the most remarkable results were obtained. When the usual tight metallic contact was made between these two wires and the two loops equally heated, the current first flowed from copper to iron across the joint, and then as the temperature rose ceased altogether, and finally, at a red or white heat, flowed from iron to copper. The maximum deflection obtained in either direction was three divisions. These deflections showed the celebrated inversion discovered by Cumming.

If the pressure between the loops was relaxed, the current ceased altogether; but when the loops were moved, so that the copper became red-hot while the iron was cool, a current flowed from the copper to the iron, or from hot to cold across the joint, giving a deflection of 100 divisions; whereas if the iron was heated red-hot and the copper cooled, a current giving 90 divisions flowed in the opposite direction, or from iron to copper, but from hot to cold as before. Thus in these two cases the loose-contact currents given when one or the other loop was heated, flowed in the opposite direction be

3

tween the metals, but in both cases from hot to cold across the joint, and were in each case about thirty times as great as the currents given by the thermo-electric difference between the metals.

It was found on examining the Table, that wherever copper appeared in conjunction with any other of the metals named, the direction of the loose-contact current could invariably be determined by the following rule:-When the copper was the hot wire, the current flowed from the copper to the other metal across the joint; but when copper was the cold metal, the current flowed from the other metal to the copper, or in both cases from hot to cold.

Exactly the contrary was found wherever iron appeared in conjunction with any of the five metals but copper; the current then always flowed from cold to hot. Two copper wires alone gave the largest deflection, of about 220 divisions; and two iron wires alone gave the next largest of those obtained where single metals only were used, but of course in the opposite direction to the deflection from copper.

It was then perceived that all these results would be explained if the thin coating of oxide on the copper wire might be regarded as a conductor with a hot and cold junction, and endowed with thermo-electric properties far more positive than the iron, while at the same time the coating of oxide on the iron wire would have to be regarded as far more negative than the copper. It was, however, difficult to suppose that two bodies so similar in some respects as the oxides of copper and iron should be at opposite extremities of the thermo-electric scale, but the following direct experiment left no doubt on my mind.

A little spiral was made of platinum wire, and a small quantity of oxide of copper laid upon it, and held in a flame till white-hot; another platinum wire was then dipped in the melted mass, when a strong current was at once observed from the hot to the cold wire, as if a loose contact had been made between two copper wires. When either of the oxides of iron was tested in a similar manner, a strong current was obtained from the cold to the hot platinum wire, as if a loose contact had been made between two iron wires. I do not yet know positively what the substances are which, interposed between silver and platinum and gold wires, give rise to the loose-contact currents, but I feel no doubt that these are as much thermo-electric currents as those given by the oxides of copper and iron, and are produced in a circuit composed of the metal and a very thin hot film, of which the two surfaces are unequally heated.

There are, however, some good reasons for doubting whether electrolytes can be included in a true thermo-electric series, and I consulted many authorities with reference to this point. Seebeck himself includes many electrolytes in his thermo-electric scale, and places acids below bismuth, a result confirmed lately by Gore (in 1857); he also places certain salts above antimony, a result subsequently confirmed by Andrews of Belfast in 1837. This gentleman observed that the tension produced by the salts between the wires was about equal to that between a platinum and silver plate in dilute sulphuric acid, and that the metals used as electrodes did not influence the deflection. He considered the current certainly due to a thermo-electric action.

Faraday in 1833 discovered what Becquerel subsequently called pyro-electric currents; the currents were in different directions with different substances used, and some, if not all, were of the same nature as those I have described. Leroux and Buff obtained currents where glass acted as the electrolyte. Leroux considered them thermo-electric, and Buff chemical effects. Buff also attributes some of the electrical phenomena connected with flame to a

thermo-electric action in which unequally heated air or gas forms part of the circuit. The currents obtained when a hot and cold platinum wire are dipped into dilute sulphuric acid and other liquids are well known; and finally (in 1858), Mr. Wild published a laborious research, in which he seems to prove the development of thermo-electric currents not only at the junction between metals and various solutions, but also between two different solutions. Thus, although none of the above observers seem to have tested the oxides, there seems little reason to doubt that they may be classed with other electrolytes, and may give rise to currents in the same manner. On the other hand, I cannot yet consider it definitively proved that any of the currents obtained from electrolytes are due to a true thermo-electric action—that is to say, to an absorption of heat only, especially as Mr. Wild could find no trace of the Peltier heating and cooling effect at the junctions of his solutions. Further research, showing the source of the power developed, is most desirable.

While consulting the literature connected with this subject, I found that Gaugain had to some extent preceded me in the discovery of the loose-contact currents, in a paper published in the Comptes Rendus' in 1853. He comes to the same conclusion as I had done independently, that they were due to the unequally heated film of foreign matter, and places oxide of iron below platinum, and oxide of copper above gold and zinc, but below iron, instead of very much above it as I find. He does not appear to have observed the exceedingly high electromotive force to be obtained from these bodies, no doubt owing to the use of a short galvanometer coil of thick wires, such as is commonly used for thermo-electric researches. He introduces a carburet of iron, of which I find no trace, with more positive properties than oxide of copper, to explain some of his results. He gives very few data on which to found his theory, but simply mentions his conclusions, and appears to have made no direct experiment whatever with the oxides. Owing to these circumstances his experiments seem to have attracted little attention. I have endeavoured to contrive a convenient apparatus by which to study the properties of the oxides, but have not hitherto met with much success, owing to the great difficulty in maintaining a constant difference of temperature between the surfaces of the very thin film, which can alone be used with Next year I hope to obtain further results in elucidation of these quasi thermo-electric currents from electrolytes.

success.

I now wish to add a few remarks on the currents which occur when true metallic contact is made between a hot and cold end of a wire of one metal. The existence of these currents was placed beyond all doubt by Magnus's careful experiments, but their connexion with other thermo-electric phenomena has hitherto remained entirely without explanation. Wild has suggested that they might be due to a thermo-electric couple formed with hot air or gas at the moment of junction; but experiments which I have made show this explanation to be founded on a mistaken conception of the duration of the current, which is by no means instantaneous, but lasts at least five minutes with copper or with iron wires, very gradually decreasing in intensity from a maximum to zero.

Another explanation, viz. that the deflection is due to a sort of discharge of a statical effect produced by the unequal distribution of heat, is also negatived by the same consideration, as well as by the fact that a tension of sufficient magnitude to produce such a charge could not possibly have escaped observation by direct measurement.

Professor W. Thomson has shown conclusively, in his Dynamic Theory of