that these differing thermo-electric qualities have in each case the thermo-electric quality of soft iron intermediate between them.

These various results show that the character of the effect in each case is decided by distorting stress or by distortion, and leave entirely open, and only to be answered by further experiments, the questions: what is the thermo-electric effect of pressure or traction, applied uniformly in all directions to a metal? and what is the thermo-electric effect of a permanent condensation or dilatation remaining in the metal, when freed from the force by which that condensation or dilatation was produced?

Experiments are also described, by which the author found that in soft iron under magnetic force, and in that retaining magnetism when removed from the magnetizing force, directions along the lines of magnetization deviate thermo-electrically towards antimony; and that directions perpendicularly across the lines of magnetization in soft iron, deviate towards bismuth, from the unmagnetized metal. He illustrates this conclusion by an experiment on a ribbon of iron, magnetized nearly at an angle of 45° to its length, and heated along one edge while the other is kept cool. When the two ends, kept at the same temperature, are put in communication with the electrodes of a galvanometer, a powerful current is indicated, in such a direction, that if pursued along a rectangular zigzag from edge to edge through the band, the course is always from across to along the lines of magnetization through the hot edge, and from along to across the lines of magnetization through the cold edge.

(4) In this part of the communication, attempts made by the author to find the effects of various influences on electric conductivities of metals are described. One of these, with a very unsatisfactory method for testing resistances, led to the conclusion that longitudinal magnetization diminishes the conducting quality of iron wire. The general plan for testing resistances, which he subsequently adopted as the best he could find, and which has proved very satisfactory, is next explained; and as an illustration, a single experiment on the relative effect of an equal longitudinal extension on the resistances of iron and copper wires is described. The conclusion established by this experiment is, that both by extension with the tractive force still in operation, and by permanent extension retained after a cessation of stress, the conductivity of the substance is more diminished in iron than in

copper; or else that it is more increased in copper than in iron, or increased in copper while diminished in iron, if it is not in each metal diminished, as the author is led by a partial investigation of the absolute effect in each metal to believe.

(5) The result previously arrived at regarding the effect of longitudinal magnetization on the conductivity of iron is confirmed; and an experiment that would have been found impracticable by the less satisfactory method, proves the same conclusion for magnetized steel wire, with the magnetizing influence away. Two very different experiments show further, that the electric conductivity of magnetized iron is greater across than along the lines of magnetization. A last experiment, showing that iron gains in conducting power by magnetization across the lines of the electric current, leads to the conclusion that there is a direction inclined obliquely to the lines of magnetization, along which the conductivity of magnetized iron would remain unchanged on a cessation of the magnetizing force.

ABSTRACT II. PART VI.

[From the Proc. Royal Soc., May, 1875.]

Effects of Stress on Magnetization.

Weber's method, by aid of electromagnetic induction and a “ballistic galvanometer" to measure it, which has been practised with so much success by Thalén, Roland, and others, has been used in the investigation of which the results are at present communicated; but partial trials have been made by the direct magnetometric method (deflections of a needle), and this method is kept in view for testing slow changes of magnetization which the electromagnetic method fails to detect.

The metals experimented on have been steel pianoforte-wire, of the kind used for deep-sea soundings by the American Navy and British cable-ships; and soft-iron wires of about the same gauge, but of several different qualities.

I. Steel.

The steel wire weighs about 141⁄2 lbs. per nautical mile and bears 230 lbs. Weights of from 28 lbs. to 112 lbs. were hung on it and taken off, and results described shortly as follows were found:

(1) The magnetization is diminished by hanging on weights, and increased by taking the weights off, when the magnetizing current is kept flowing.

(2) The residual magnetism remaining after the current is stopped is also diminished by hanging on the weights, and increased by taking them off.

(3) The absolute amount of the difference of magnetization produced by putting on and taking off weights is greater with the mere residual magnetism when the current is stopped, than with the whole magnetism when the magnetizing current is kept flowing.

(4) The change of magnetization produced by making the magnetizing current always in one direction and stopping it is greater with the weights on than off.

(5) After the magnetizing current has been made in either direction and stopped, the effect of making it in the reverse direction is less with the weights on than off.

(6) The difference announced in (5) is a much greater difference than that in the opposite direction between the effects of stopping the current with weights on and weights off, announced in (4).

(7) When the current is suddenly reversed, the magnetic effect is less with the weights on than with the weights off.

II. Soft-Iron Wires.

Wires of about the same gauge as the steel were used, but, except one of them, bore only about 28 lbs. instead of 230 lbs. All of three or four kinds tried agreed with the steel in (1).

The first tried behaved (excepting a seeming anomaly, hitherto unexplained) in the reverse manner to steel in respect to (2), (4), (5), and (6); it agreed with the steel in respect to (7). Another iron wire*, which, though called "soft," was much less soft than the first, agreed with steel in respect to (1) and (2), but [differing from steel in respect to (3)] showed greater effects of weights on and off when the magnetizing current was flowing than when it was stopped.

Other soft-iron wires which were very soft, softer even than the first, agreed with all the steel and iron wires in respect to (1),

* It was tested magnetically with weights up to 56 lbs., and broke, unfairly however, when 63 lbs. were hung on.

but gave results when tested for (2) which proved an exceedingly transient character of the residual magnetism, and were otherwise seemingly anomalous.

The investigation is being continued with special arrangements to find the explanation of these apparent anomalies, and with the further object of ascertaining in absolute measure the amounts of all the proved effects at different temperatures up to 100° C.

ABSTRACT III. PART VII.

[From the Proc. Royal Soc., May, 1878.]

Effects of Stress on Magnetization of Iron, Nickel, and Cobalt. This paper commences with a detailed description of a series of experiments on the effects of stress on the magnetism of soft iron, of which some first results were described in a preliminary notice, communicated to the Royal Society on the 10th of June, 1875 and published in the Proceedings (§§ 197-197,' above) of that date. A few months later, the author found that he had been anticipated by Villari* in the most remarkable of those results—that showing increase or diminution of magnetization by longitudinal pull, according as the magnetizing force is less than, or greater than, a certain critical value.

In the first series of experiments described in this paper, the amount of the magnetizing force is varied through a range of values from zero to 900, on a scale on which about 12 is the value of the vertical component of the terrestrial magnetic force at Glasgow, and the effects of hanging on and taking off weights of 7 lbs., 14 lbs., and 21 lbs.† in changing the induced magnetism, are observed. The experiments were made at ordinary atmospheric temperatures, and at temperature 100° C. The results. are shown in curves (Plates II. to XII. below), of which the abscissas represent the magnetizing forces and the ordinates, the change of magnetism of the wire produced by "ons" and "offs" of the weight while the magnetizing force is kept constant. The Villari critical value was found to differ for the two temperatures, and for different weights: thus approximately :

* Poggendorf's "Annalen," 1868.

The wire was of about No. 22 B. W. G. gauge, and weighing therefore about 14 lbs. per nautical mile. It was so soft that it had experienced a considerable permanent stretch by 21 lbs.; it would probably break with 30 or 40 lbs. Steel pianoforte wire of same gauge bears about 230 lbs.

The maximum effect of the "on" and "off" was found in each case with a magnetizing force of from 50 to 60 of the arbitrary scale divisions (or about four times the Glasgow vertical force). Its amount differed notably, though not greatly, with the temperature, and, as was to be expected, greatly with the different amounts of pull; but it was not nearly three times as much with 21 lbs. as with 7 lbs.; thus approximately :

The curves all tend to asymptotes parallel to the line of abscissas on its negative side for infinite magnetizing forces; and they indicate the following ultimate values for the two temperatures, and the different amounts of pull: