If a martensite containing less carbon than that represented by the point G is cooled down from a temperature of, say, 900°, then when the temperature has fallen to that, represented by a point on the curve IG, [beta]-ferrite will separate out, and, as the temperature falls, the composition of the solid solution will alter as represented by IG. On passing below the temperature of HG, the [beta]-ferrite will be converted into [alpha]-ferrite, and, as the temperature falls, the latter will separate out more and more, while the composition of the solid solution alters in the direction GF. On passing to still lower temperatures, the solid solution at F (0.8 per cent. of carbon) breaks up into pearlite. If the percentage of carbon in the original solid solution was between that represented by the points G and F, then, on cooling down, no [beta]-ferrite, but only [alpha]-ferrite would separate out. We see, therefore, that when martensite is allowed to cool _slowly_, it yields a heterogeneous mixture either of ferrite and pearlite (when the original mixture contained up to 0.8 per cent. of carbon), or pearlite and cementite (when the original mixture contained between 0.8 and 2 per cent. of carbon). These heterogeneous mixtures constitute soft steels, or, when the carbon content is low, wrought iron. The case, however, is different if the solid solution of carbon in iron is _rapidly_ cooled (quenched) from a temperature above the curve IGFE to a temperature below this {228} curve. In this case, the rapid cooling does not allow time for the various changes which have been described to take place; so that the homogeneous solid solution, on being rapidly cooled, remains homogeneous. In this way hard steel is obtained. By varying the rapidity of cooling, as is done in the tempering of steel, varying degrees of hardness can be obtained.