20° to -20° C., we get a terrace in the cooling curve, as shown in Fig. 5. This tells us that some change has taken place in the nature of the substance at 0°. We

see directly that this change corresponds with the passage of water from the liquid to the solid state of aggregation.

Now draw the cooling curve of molten sodium thiosulphate. We know that the molten liquid "ought" to freeze at 56° (Fig. 6). But the cooling curve goes

on quite normally below that temperature until, at length, there is a great evolution of heat, and the liquid

solidifies. The temperature may even rise above 56°. The cooling curve of the solid is quite normal. The amount of heat evolved as the molten liquid solidifies corresponds with the "latent" heat absorbed as the solid melts.

§ 8. Surfusion and Recalescence

The molten sodium thiosulphate as it cools down the "surfusion" curve is at a lower temperature than its normal point of solidification or freezing. The liquid is then said to be in a state of superfusion or surfusion. The system is in unstable equilibrium. We may get a similar state of things when a saturated solution of a substance is slowly cooled. More salt may be in solution than the true solubility of the salt. The result is a supersaturated solution. Agitation, or the addition of a trace of something which will serve as a nucleus for crystallization, will generally suffice to start the system on its passage to a state of stable equilibrium. But when the transformation does set in, it usually takes place very rapidly, and is accompanied with a rise of temperature. The cooling curve is distorted in a corresponding manner (Fig. 6).

It is interesting to put a little ether in a small bulb blown at the end of a piece of glass tubing, placed in supercooled sodium thiosulphate (Fig. 7). Drop in a crystal of sodium thiosulphate. The evolution of heat as the liquid solidifies raises the temperature high enough to vaporize the ether. The vapour of ether will burn at the mouth of the tube with a steady flame when ignited.

When a steel bar is cooling, an evolution of heat

occurs at about 690°. The amount of heat evolved is so great that the metal visibly brightens in colour.

The phenomenon is called recalescence. The cooling curve is shown in Fig. 8.

§ 9. The Cooling Curve of Pure Iron

The cooling curve of iron from the molten condition is shown in Fig. 9. The iron was practically pure. It

only contained 0.01 per cent. of carbon. F. Osmond,

a celebrated French metallurgist, maintains that the existence of the transition-points, or discontinuities, Ars and Arg, in the cooling curve of the solidified metal, points to the existence of three allotropic modifications of solid iron :

i. Alpha Iron.-Below Ar2, that is 750°, we have what he calls a-iron, or alpha iron.

ii. Beta Iron.-Between Ar, and Arg, that is between 750° and 860°, we have what he calls ẞ-iron, or beta iron. Beta iron is non-magnetic. Heat is evolved when iron passes from the B- to the a-state, and magnetic properties are developed at the same time.

iii. Gamma Iron.-Above the Ars critical point, namely 860°, we are supposed to have y-iron, or gamma iron. This variety is non-magnetic.

Each critical point is found to be associated with a change in the mechanical properties, the microscopic appearance, the electrical conductivity, the magnetic properties, and the specific gravity of the metal.1

The changes which occur during the cooling of a substance are reversed when the substance is heated. The cooling curve of steel, with 1.2 per cent. of carbon, shown in Fig. 10, is reversed on heating, as shown by the heating curve in the same diagram. There is only one critical point at about 690°, called the Ar1 critical point.

The critical points Ac1, Ac2, Acg on the heating curve of mild steel are generally a few degrees higher

10. Boudouard, Journ. Iron and Steel Inst., 63. i. 229, 1903; H. le Chatelier, Compt. Rend., 128. 1444, 1899; 129. 299, 331, 497, 1899; Metallographist, 2. 334, 1899; 38. 38, 152, 1900; G. E. Svedelius, Phil. Mag., [5], 46. 173, 1898; G. Charpy and L. Grenet, Compt. Rend., 124. 540, 598, 1902; Metallographist, 6. 240, 1903; S. Curie, ibid., 1. 107, 229, 1898.

than the corresponding points Arı, Arg, and Arg respectively. There seems to be a kind of molecular inertia, or lag, which prevents the y to ẞ, the ẞ to a, or the reverse changes taking place sharply. The critical

points on the cooling curve are, in consequence of this lag, a few degrees below the true critical point. The lag induces a state somewhat analogous to surfusion in molten sodium sulphate. The critical points on the heating curve are a little too high, and for a similar reason.1

The critical points of iron really represent ranges of temperature, although, for the sake of inconvenience, we call them points. The Ar, with soft steel commences at 845°, and finishes at 800°; it is most marked at 820°. The Ar2 extends from 755° to 710°; and the Ar1 from 680° to 645°.

The "r" of "Ar" comes from the French word refroidissant, for cooling; the "c" of "Ac" from chauffant, heating. This notation is due to D. Tschernoff, the Russian metallographist.

1 F. Osmond, Metallographist, 1. 270, 1893; 2. 169, 1899; H. M. Howe, ibid., 2. 257, 1899; M. Aliament, La Electricien, 49, 1903. 2 Otherwise spelt "D. Chernoff."