shown that a piece of soft iron at a bright red heat loses all ordinary signs of magnetism; and Faraday has shown* that wrought iron retains only traces of its ordinary magnetic capacity at that temperature. He has further shown† that nickel first loses its distinctive magnetic power at about 635° F. (=335° C.), and that the temperature of boiling oil is sufficient to render large masses of that metal insensible to the action of common mag

-also that on raising the temperature of iron and nickel

2

0° F. to 300° F. (=-17°7 C. to 149° C.) the magnetic arty of iron remains constant, whilst that of nickel dimiumradually. According to Matteuccit, the magnetism of ironenses up to a certain temperature, then decreases rapidly; * retais, kowever, even at a white heat, a very minute degree of it's que le conacity, calculated at only '000015 of its ordinary amount, fra globule melted in a lime spoon was still attracted a powerful electromagnet.

2

I apper "On a Momentary Molecular Change in Iron Wire, published in the Proceedings of the Royal Society, No. 108, p. 260, January 28, 1869, I described a singular phenomenon which I observed in the cooling of iron wire which had been heated to full redness whilst under a suitable degree of longitudinal tension by means of a spring attached to one of its ends; the iron during cooling, and whilst still red-hot, gradually diminished in length, then suddenly elongated by diminution of cohesion, and finally contracted gradually nearly to its original length during the remainder of the cooling process; a corresponding but reverse phenomenon did not occur during the process of heating the wire. Various other metals were similarly examined, but no such peculiar phenomenon was found. In another paper, "On the Development of Electric Currents by Magnetism and Heat"§, I showed that, by cooling a brightly red-hot iron wire (under the influence of a permanent magnet) within the axis of a coil of thin insulated copper wire, an electric current was induced in that wire; and during the first few seconds in the process of cooling, and at apparently the same temperature at which the aforesaid elongation and loss of cohesion occurs, an irregular action took place in the induced current, which "was probably connected with the momentary molecular change." In the present paper I have employed a different method of examining these molecular movements and magnetic changes of iron and nickel under the influence of heat, and have obtained some

*

Experimental Researches in Electricity, 2344-2347.

† Phil. Trans. vol. cxlvi. (1856) pp. 178-180.

Bibl. Univ. de Genève, [Arch, des Sc. xxiii.] xxiv.

Proc. Roy. Soc. January 28, 1869, p. 265 [Phil. Mag. July 1869, pp. 59, 64.).

further results showing the existence of additional molecular changes.

I took a well-annealed and straight bar of wrought iron 32 inches (81.3 centims.) long and of an inch (=95 millims.) in diameter, supported it in a horizontal position by two wooden clips, placed one end of the bar in a coil of thin insulated copper wire connected with a distant galvanometer, and the other end within a coil of thick insulated copper wire for attachment to a voltaic battery; the battery consisted of ten large Smee's cells. A row of five Bunsen's burners was placed beneath the middle part of the bar for the purpose of heating it to redness; and the bar at that part was between notched plates of firebrick to increase the heat. With the whole of the bar at 60° F. (=15°•5 C.), on connecting the thick coil with the battery a deflection of 14° or 16° was obtained of the needles of the galvanometer; but with the middle of the bar at a red heat a deflection of only 4° could be obtained. In a similar experiment with a bar 2 feet (=61 centims.) long and an inch (=12·7 millims.) thick, the same battery, and more suitable coils of wire, a deflection of 20° or 22° was obtained with the middle of the bar red-hot, and a powerful swing of 90° with the whole of the bar cold, the needles striking strongly against the stops of the galvanometer. With the bar of iron entirely absent, no perceptible electrodynamic induction took place. Similar but much less powerful results were obtained on substituting a permanent bar magnet for the battery and thick wire coil. A red heat, therefore, in the middle part of a wroughtiron bar largely diminishes, but does not entirely prevent the transmission of magnetism along the bar.

The number of molecular movements and magnetic changes which occur in wrought iron by change of temperature at a moderate red heat are quite remarkable, and were gradually revealed by the following experiments :-An iron bar 32 inches (=81·3 centims.) long and 3 of an inch (=9.5 millims.) thick, in a horizontal position and diagonal to the magnetic meridian, was surrounded at one end by a coil of thin insulated copper wire connected with a distant galvanometer. On gradually heating the middle part only of the bar to redness, when the temperature acquired a low red heat a slight sudden deflection of the needles occurred; and on gradually cooling the bar a similar slight and sudden deflection in an opposite direction took place, apparently at the same temperature. The directions of the induced current during the heating agreed with what would be produced by a decrease of magnetism, and during the cooling with an increase of magnetism. With a bar 2 feet (=61 centims.) long and an inch (12.7 millims.) thick, and the north pole of a small permanent bar magnet lying in contact with one end of the bar, the

other end of the bar being enclosed within a coil 5 inches (12.7 centims.) long, containing 20 layers or 2674 turns of "No. 27" (=0·25 millim. thick) insulated copper wire, connected with the galvanometer, a slight deflection only of the needles was produced during the process of heating, and a quick motion of the needles 3° in the opposite direction during the cooling. The directions of movement agreed with the previous ones, and showed that the bar suddenly increased in magnetic capacity during cooling at a particular temperature of moderate red heat, and apparently at the same temperature at which it undergoes the anomalous diminution of cohesion and increase in length already referred to. With the same bar and fine-wire coil, and 10157 inches (=13.3 centims.) long containing 8 layers or ren of " No. 16" (=1.5 millim. thick) insulated copper

the opposite end connected with the battery of ten large Smce's clesiats, on gradually applying the heat the following effccts took place:-When the bar became red-hot in the middle, a smail and in regular deflection of the needles of 110 occurred, and the needles then returned steadily to zero, and no other deflection took place as the bar became somewhat hotter. The gasflames being now suddenly extinguished, in less than half a minute the needles moved slowly 1° in the opposite direction, then stopped, and then went rapidly 18° further and quickly back again, swinging nearly equally on each side of zero, and soon settled near that point, and remained there during the whole of the remainder of the cooling process. The directions of the currents agreed with those previously obtained.

In the next experiment an iron bar 2 feet (=61 centims.) long and of an inch (=19 millims.) thick was employed, with the same battery, the thick-wire coil being 6 inches (=15.2 centims.) long and containing 536 turns of "No. 17" (=1.5 millim. thick) copper wire, and the other coil being 6 inches long and containing 1960 turns of " No. 27" (=0.25 millim. thick) copper wire. The order of procedure was the same as in the last experiment. As soon as the bar had become red-hot, several minute molecular changes and increases of magnetism in succession in the bar took place as the bar became hotter, indicated by corresponding small deflections of the needles. The bar being larger than the previous one, it did not become quite so highly heated; therefore on stopping the gas the needles were instantly deflected 25° in the opposite direction; the current which produced this deflection lasted in its full strength only a few seconds, and ceased entirely in about one minute; it was succeeded, however, by another, transient and feeble current in the same direction. By immersing about 6 inches (=15.3 centims.) of the middle part of the bar (which was at about 50° F. =10° C.) in a freezing-mixture

at -26° F. (=-32°.2 C.), composed of 4 pounds of ice and 6 pounds of crystallized chloride of calcium, an irregular deflection of the needles, indicating another feeble molecular change and increase of magnetism was observed; the general deflection, how. ever, obtained by this artificial cooling was in accordance with a decrease of magnetism, contrary to that which resulted from cooling at higher temperatures: a repetition of this experiment gave similar results. With a bar 3 feet (91.4 centims.) long and 1 inch (25.4 millims.) thick none of the electrical effects were obtained by the heating process, owing to the heat of the gasburners being not sufficiently powerful.

mag

These results show that on gradually heating a bar of wrought iron, when it attains a moderate red heat a succession of distinct and separate small movements, all of a similar kind, take place amongst its molecules, and that on gradually cooling such a heated bar, when its temperature has sunk to moderate redness three successive molecular movements and diminutions of netism occur the first being a small one, the second a very large movement, and the third also small; and these movements are all of an opposite character to those which take place during theel the It is singular that there is no sudden large decrease ofagnetism, during the process of heating, to correspond with the sudden large increase of magnetism during cooling; and this precisely agrees with the phenomenon of molecular and cohesive change already referred to (page 171). It is probable that the suuden and momentary increase of length and diminution of cohesion which an iron wire under a suitable degree of longitudinal strain undergoes whilst cooling at a moderate red heat, may be due to the sudden great increase of magnetism which it then acquires, in accordance with the fact discovered by Mr. Joule, that a rod of soft iron at 60° F. suddenly increases in length and decreases in diameter when magnetized.

With a bar of cast steel 27 inches (=68.6 centims.) long and of an inch (22.2 millims.) in diameter, similarly examined, the fine wire coil containing 2850 turns of "No. 23” (=0·7 millim. thick) copper wire, the following results were obtained. During heating, a very feeble sudden molecular change, attended by decrease of magnetism, occurred below a visible red heat; and at a moderate red heat a second sudden, and more extensive, molecular change occurred of the same kind; by further heating to a higher degree of redness no other sudden change took place. The gas-flames being now extinguished, after a period of thirty seconds a slight molecular change and increase of magnetism occurred; and in fifteen seconds more a sudden and more powerful change of the same kind took place, producing a deflection of the needles of 6°; the needles then returned to their original position.

Cold water being now very freely applied to the black hot bar, a similar deflection of 12° occurred; part of this last deflection was probably due to the more rapid cooling.

With a bar of cast iron 30 inches (=762 centims.) long and of an inch (19 millims.) in diameter, and the same wire coils and battery, the effects obtained were similar, but much more feeble than with the bar of steel. At a feeble elevation of temperature, below that at which a similar effect took place in the steel bar, a feeble sudden molecular movement and decrease of magnetism took place. At a red heat a second similar change occured. Soon after this the gas was stopped and the bar allowed to cool; in about twenty seconds a molecular change como,erend, fcable at first and then suddenly stronger, producing a detection of 6 and then ceasing, as with the steel bar.

Varioes for small irregularities in these molecular changes in wrought iron, cast iron, and steel were observed, but are not here recorded snd each different bar gave somewhat different results. If presses of heating and cooling were more rapid, all the clangould be more powerful.

The results I have obtained with bars of wrought iron, steel, and cast iron by the foregoing plan agree in the maine micthose of M. Mauritius*, obtained by quite a different methou, e found that at a bright red heat none of the bars were magnetic; on cooling a red-hot steel bar the magnetism increased at first very rapidly, then for a certain time slowly, and then again followed a second period of rapid increase; cast iron behaved similarly but in a less degree; and with wrought iron the second increase does not exist. He considers the magnetism of iron is developed suddenly at a particular temperature of about 1000°.

With a bar of cast nickel about 18 inches (=45.7 centims.) long and an inch (=12·7 millims.) thick the following effects were obtained :-During heating, a slow deflection of 11° was obtained at a particular temperature much below a red heat; and during cooling, a deflection of 11° in an opposite direction occurred, apparently at the same temperature; no other deflections were obtained by heating the bar gradually to redness and then cooling. With another bar of that metal, 2 feet (=61 centims.) long and of an inch (= 19 millims.) in diametert, a thick-wire coil 6inches (15.2 centims.) long and 24 inches (=5.7 centims.) in diameter, containing 8 layers or 536 turns of "No. 17” (=1·5 millim. thick) copper wire, and a thin-wire coil 6 inches long and and 2 inches in diameter, containing 14 layers or 1960 turns of “No. 27” (=0·25 millim. thick) copper wire, on applying the

* Phil. Mag. 1864, vol. xxvii, P. 399.

+ Obtained by the kindness of H. Wiggin, Esq., of the firm of Messrs. Evans and Askin, nickel-refiners, Birmingham.