took, in the beginning of the year 1872, a similar investigation of various alloys. The present memoir contains the results, the investigation having been carried out in the physical laboratory of the Royal Swedish Academy of Sciences, Stockholm.

1. M. Edlund's method is based on the proposition, deduced from the mechanical theory of heat, that a galvanic current going through an electromotor calls forth in it an absorption or production of heat proportional to its electromotive force, according as the current is in direction the same as or opposite to that generated by the electromotor itself*. Hence the relative quantity of the electromotive force is obtained by measuring the amount of heat absorbed or produced. The phenomenon also indicates, as is readily seen, the direction of the current excited by the electromotor.

The air-thermometer constructed by Professor Edlund for the measurement of the quantities of heat in question is fully described in his memoir; consequently it will be sufficient here to mention only the principal parts of this apparatus. Two perfectly equal cylinders of thin sheet copper communicate through a horizontal glass tube of 2.5 millims. internal diameter. Into both cylinders, through open tubes which enter the middle points of their ends, the same combination of two metals in the shape of wires or rods is inserted in such wise that the place of contact is in the centre of the cylinder. The tubes are lined internally, to half their length, with wood, to prevent contact between the cylinder and the wire. Communication between the outer air and that of the cylinder is then completely cut off by pouring into the tubes a melted mixture of equal parts by weight of wax and rosin. The two wires are inserted in the circuit of a galvanic current in such a manner that, when the current in one of the wires goes from metal A to metal B, the second wire is passed through by the current in the opposite direction, from B to A. Hence arises at the one place of contact a production, at the other an absorption, of heat. The trifling difference hereby produced between the temperatures of the air in the cylinders occasions the displacement of a column a few centimetres long of liquid in the tube, a mixture of alcohol and water of sp. gr. 0.9 which acts as an index. If the current be reversed, a cooling ensues in the cylinder in which there was previously a heating, and vice versa, by which the index is moved towards the opposite side. The displacement towards the one side or the other ceases as soon as equilibrium is restored between the heat given off by each cylinder through radiation and conduction to

*This proposition was first advanced by Professor Edlund (Efv. af K. Vet.-Akad. Förh. 1869, p. 457; Pogg. Ann. vol. cxxxvii. p. 474; Phil. Mag. S. 4. vol. xxxviii. p. 263).

the outer air and that developed by the current in the wires. The positions in which the index becomes stationary are read off a millimetre-scale fixed to the glass tube. The magnitude of the deflection is given by the difference between the numbers read off. A displacement of the liquid one scale-division corresponds to a difference of temperature of 0°-002 Cels.

Each combination was tried with three different intensities of current. From the deflections the quantity a was calculated, which is proportional to the quantity of heat absorbed or produced with the unit of intensity (tan 45°), according to the equation

in which ẞ is a constant proportional to the galvanic resistance of the wire in the cylinder, s the current-intensity, and t the deflection. The constant B is obtained from the deflections t and t1 at the intensities s and s1, by the formula

The combination of each two of the three current-intensities gave a value of ß; in the calculation of a the arithmetic mean of the three values were inserted in equation (1).

Professor Edlund, in nearly all his experiments, made use of two copper cylinders silvered on the outer side. These cylinders proved now to be no longer air-tight; and therefore in the present investigation two others, unsilvered, also employed by Edlund in some experiments, were used.

Waiting till the displacement ceases, before reading-off the position of the column of liquid, makes the labour very tedious. This method is not always practicable, because the index has mostly a motion of its own, independent of the temperaturevariation to be measured, the cause of which will be subsequently stated. Professor Edlund therefore made his experiments after the following method. The position of the liquid column was read off at the end of each three quarters of an hour; and then the current was reversed. The effect of the peculiar motion (which could be taken as uniformly accelerated or retarded) was eliminated in the following manner. The correct deflection, corresponding to the quantity of heat developed or which had vanished at the place of contact, was obtained from three deflections observed in immediate succession, by taking the arithmetic mean of the first and third, and then the mean of this mean and the second deflection. In order to save time, Professor Edlund performed the reading in some experiments every quarter of an hour. Although the temperature of the cylinders had not in so short a time arrived at equilibrium, yet the deflections thus ob

tained may be employed, because the values of a which were calculated from them are, as Edlund has shown, proportional to those furnished by the readings after 45 minutes. By special experiments, which shall in due time be given, it was ascertained that there was this proportionality even when the unsilvered cylinders were employed. Hence in the present investigation, except where otherwise stated, the readings were everywhere taken every quarter of an hour.

As essential parts of the apparatus the double cylinders made of metal sheet and filled with water must be mentioned, the mantles, which protected the copper cylinders against the variations of temperaturc in the observation-room. Those employed by M. Edlund were made of zinc. During the experiments the oxidating action of the water had covered the inner sides of the mantles with a thick layer of hydrate of zinc, which obstructed the passage of the heat developed in the air-thermometer into the water. Thereby was produced an elevation of temperature in the annular layer of air between a cylinder and its mantle; and because this rise of temperature was not equally great for both cylinders, the column of liquid in the glass tube was put in the above-mentioned continuous motion towards one side. To remove this a means was tried which is indicated in M. Edlund's memoir. New mantles were procured, of the same dimensions as the old, but of sheet brass and electroplated with silver on the side presented to the water. These were filled with distilled water. With this arrangement the continuous motion almost entirely disappeared; only towards the end of the investigation did it show itself in a slightly higher degree. The reason for it, however, did not lie in the mantles; for the displacement in question vanished again when an experiment was made with the silvered cylinders. Probably one of the unsilvered cylinders had undergone a change, so that it gave up the heat with greater or less velocity than the other. On account of the continuous displacement, the deflections given in the following are in every case calculated, according to the method beforementioned, from three deflections observed in succession.

2. The galvanic current was generated by a pile of from two to four Bunsen's elements; and in the circuit, besides the wires in the cylinders, a rheostat, a tangent-compass, and a commutator were inserted. The alloys, except the German silver, were cast in a mould of gypsum, in cylindrical rods of about 16 centims. length and 2 millims. thickness. These were soldered with tin to copper wires 1 millimetre thick, which had been drawn out of chemically pure, galvanically precipitated copper.

* The entire apparatus in this improved form is delivered by M. Sörensen, of Stockholm, at the price of 110 thalers Prussian currency.

In the alloys commercial (impure) metals were employed. During the investigation it was observed that both the electromotive and the thermoelectric force diminished as the time proceeded. This diminution, especially remarkable immediately after casting, depended doubtless on physical (perhaps also chemical) molecular changes in the alloys. As the electromotive force was usually determined immediately after the casting, but the thermoelectric not till some days subsequently, one could not be sure that both determinations corresponded to the same molecular state of the alloys. It was consequently necessary with most of the alloys to repeat the experiments. This later. investigation was effected perhaps six weeks after the first; and the same pair of wires was employed. As it could be assumed that at the end of this time the molecular changes had for the most part ceased, and since also the determination of the thermoelectric force was now taken immediately after the ascertaining of the electromotive force, the values of the two forces thus obtained correspond to the same molecular state and are comparable with each other.

3. The determination of the thermoelectric force was effected

in the following manner. The copper wire which was soldered to the alloy was bent, near the soldering, twice at right angles, and so made parallel to the alloy at a distance of about 2 centims. The soldering was inserted in a test-tube, which was immersed in cold water, whereby the place of contact was cooled 7-10° C. below the temperature of the room (15-18° C.). The free ends of the wires were joined to the ends of the galvanometer-wire by means of binding-screws; these were surrounded by a wooden case, the ends of which were perforated for the wires to pass through. The difference of temperature of the places of contact was ascertained by two exactly similar very sensitive thermometers. The bulb of one of them was put close to the place of contact in the test-tube, that of the other near the binding-screws in the box; and then the mouth of the testtube and the perforations of the box were stopped with cottonwool, in order to prevent currents of air. After the contacts had acquired the temperature of the air surrounding them, the two thermometers and the deflection of the galvanometer were read off simultaneously several times at brief intervals. The galvanometer was furnished with the usual mirror-reading. The conducting-power, measured by means of an inducing magnet, could be modified by a rheostat interpolated in the circuit.

4. We come now to the narration of the observations. The experiments made at the later determination of the electromotive and thermoelectric forces will be given in full. Of the experiments first made the results only will be stated, in order to render

conspicuous the influence of the molecular changes. In the denomination of the combinations we shall always place first the name of that substance to which the positive galvanic current at the place of contact must go in order to produce a cooling. The electromotive force of the combination generates therefore a current which at the contact-place goes from the last-named to the first-named body. The same direction always resulted also for the thermoelectric current at the hotter place of contact. We may therefore regard the first-named body as the electropositive, and the last-named as the electronegative constituent of the combination.

Iron-copper.

For the purpose of being able to compare the electromotive force of the alloys with that of the unmixed metals, the combinations iron-copper and copper-bismuth (in fact, the wires of these combinations already used by M. Edlund) were first tried. The determination of the electromotive force of the combination iron-copper gave the following result:

Experiment 1. Current-intensity stan 29° 6'5.

The mean value of ß=2.911. With this we obtain for a the following values:

At the end of the entire investigation the iron-copper combination was tested afresh to ascertain what changes might have Phil. Mag. S. 4. Vol. 47. No. 309. Jan. 1874.

D