wires, viz. :-for No. 1 the value 313-06; for No. 2, 309-62; mean, 311-34. Hence we can take the thermoelectric forces of the two wires as equal also in the later determination. The difference between the numbers 295.24 and 311-34 is not of importance. Copper and Alloy No. 6.

This alloy was obtained by melting together 32 parts bismuth and 3 parts antimony.

From these experiments 8=4-306 and

a=541.40 with s=tan 18 18,

Eleven days previously the thermoelectric force was determined at 680-93, and the electromotive at 521-23. This alloy appears therefore to be as little variable in relation to its electromotive properties as the preceding bismuth-antimony. This was confirmed by a special thermoelectric experiment. Immediately after casting, the reduced deflection 613-87 was obtained, and five days afterwards the deflection 624-13. The difference does not amount to 2 per cent.; so that they may be regarded as not essentially different.

Copper and German Silver.

The copper wire was soldered to a German-silver wire of the same thickness (1 millim.). This combination was tried only once. The electromotive force.—

5. Results of the Investigation.-If the bodies investigated be arranged according to the magnitude of the electromotive and thermoelectric forces, in such wise that a preceding, electropositive, is opposed to a succeeding one, the following series are obtained, in which the numbers give the relative values of these forces on contact with copper. In the composition of the series, only the values obtained in the later determinations are taken into account.

Consequently a metallic alloy, like the pure metals, takes the same position in the electromotive as in the thermoelectric series. Because the thermoelectric current at the hotter place of contact has the same direction as the galvanic current produced by the electromotive force in action there, this force is augmented by the heating. These results hold only for temperatures below +30° C., above which the temperature did not at any time ascend during the experiments.

If the numbers of the thermoelectric series be divided by the corresponding ones of the electromotive series the following are the quotients obtained :

Excepting the last, these quotients may be regarded as equal

=

The mean value, if the last number be excluded, 1·06; and the deviations from the mean, which at the most amount to only 6 per cent., may be attributed to errors of observation. The last quotient may reasonably be looked upon as not comparable with the rest. The combination copper and 32 bismuth 3 antimony, in consequence of its great electromotive force, produces in the cylinders of the air-thermometer considerable variations of temperature on the reversal of the current; and therefore it is possible that the equations employed in the calculation of a, having been deduced on the hypothesis that the temperaturevariations in the cylinders are only moderate, furnish for this combination a value of a which is not comparable with those for the rest of the combinations. This supposition is confirmed by the circumstance that the quotients appear to increase with the electromotive force. Hence we believe we are justified in assuming that the ratio between the thermoelectric and the electromotive force of the alloys investigated is constant and equal to the ratio for the iron-copper and copper-bismuth combinations.

Lastly, let us mention as another result of this investigation the remarkable constant diminution of the electromotive and thermoelectric forces which was observed in the bismuth-tin alloys. Of the pure metals, bismuth seems also to undergo such a change. Taking into consideration the known fact that the thermoelectrical condition of solid conductors is extremely sensitive to molecular changes (for example, slight impurities or a relative displacement of the molecules by mechanical treatment, as tension, torsion, pressure, &c.*), it seems very probable that the phenomenon in question is connected with some molecular alteration. By rapid solidification in the casting-mould free crystallization of the alloy is prevented, and a state of tension is produced, from which the molecules strive to free themselves. Simultaneously, chemical changes may take place; either the constituents of the alloy may enter partially into chemical combination with one another, or combinations that were formed in the casting may again fall asunder.

Similar molecular changes have been previously observed in some bodies. It is known, for instance, that the freezing-point of an ordinary thermometer rises a little in the course of time in consequence of the diminution in volume of its bulb. A similar phenomenon was observed by M. Edlund in 1857, when readjusting three gauged measures of the Swedish Kanne. The

* Thomson, Phil. Trans. 1856, p. 711. Magnus, Pogg. Ann. vol. lxxxviii. p. 469. Matthiessen found the thermoelectric force in bismuth and antimony very unequal, according as these metals were investigated after casting or in the state of compressed wires (Pogg. Ann. vol. ciii. p. 412).

48 Electromotive and Thermoelectric Forces of Metallic Alloys.

volume of these measures, which were made of glass, was most accurately determined by F. Rudberg in the year 1833 at 100 cubic decimal inches; M. Edlund, however, found the volume of all three vessels perceptibly less, thereby proving a permanent structural alteration in the glass, which in twenty-four years had brought about a sensible lessening of the volume of these measures. Finally we will mention the transformation of the artificially produced monoclinic crystals of sulphur into an aggregate of minute rhombic crystals, the passing of arsenic acid from the amorphous state into the crystalline form, and other phenomena of this kind well known in chemistry.

It is probable that the bismuth-antimony alloys are subject to a change similar to that undergone by the bismuth-tin alloys, although the effect of it on the electromotive force may be too little to become apparent during the proportionally brief time of the present investigation. Whether all solid conductors are thus altered when subjected to mechanical treatment, and to what extent the alteration is accompanied by modifications of their other physical properties (as density, capability of conducting heat and electricity, &c.), may be brought to light by future inquiries.

Supplement. I take leave here to mention that General Freiherr v. Wrede, of Stockholm, has recently discovered very remarkable alterations of the length of a rod, 1 decim. long, of the alloy 2 parts bismuth 1 part tin, as well as variations in its expansion by heat*. With respect to variations of the resistance to electric conduction of some alloys I made some experiments in Bonn at the end of last summer (1872). Through the kindness of Geheim-Rath Clausius I had the opportunity of using in these experiments an excellent universal galvanometer by SieThe results were :-The resistance of the alloy 2 parts bismuth 1 part tin diminished in 21 days 0.064 of its original quantity; that of the alloy 1 part bismuth 1 part tin, 0·07 in 9 days; and that of the alloy 1 part bismuth 2 parts tin, 0.06 in 4 days. The velocity of the change had its greatest value immediately after casting. The alloys 12 parts bismuth 1 part tin, 1 part bismuth 5 parts lead, and 10 parts bismuth 1 part antimony exhibited a pretty constant resistance to conduction, since the variations did not amount to more than 1 per cent.

mens.

*Kongl. Svenska Vet.-Akad. Handl. 1872.