atly coarsened, reflecting the condition of the austenite
crystals from which they were born.

It maybe noted in passing that the coarse crystals of cast metal
cannot generally be refined by heat treatment unless some forging
or rolling has been done in the meantime. Heat treatment alone does
not seem to be able to break up the crystals of an ingot structure.

HARDENING

Steel is hardened by quenching from above the upper critical. Apparently
the quick cooling prevents the normal change back to definite and
sizeable crystals of ferrite and cementite. Hardness is associated
with this suppressed change. If the change is allowed to continue
by a moderate reheating, like a tempering, the hardness decreases.

If a piece of steel could be cooled instantly, doubtless austenite
could be preserved and examined. In the ordinary practice of hardening
steels, the quenching is not so drastic, and the transformation of
austenite back to ferrite and cementite is more or less completely
effected, giving rise to certain transitory forms which are known
as "martensite," "troostite," "sorbite," and finally, pearlite.

Austenite has been defined as a solid solution of cementite (Fe3C)
in gamma iron. It is stable at various temperatures dependent upon
its carbon content, which may be any amount up to the saturated
solution containing 1.7 per cent. Austenite is not nearly as hard
as martensite, owing to its content of the soft gamma iron. Fig.
49 shows austenite to possess the typical appearance of any pure,
crystallized substance.

In the most quickly quenched high carbon steels, austenite commonly
forms the ground mass which is interspersed with martensite, a large
field of which is illustrated in Fig. 50. Martensite is usually
considered to be a solid solution of cementite in beta iron. It
represents an unstable condition in which the metal is caught during
rapid cooling. It is very hard, and is the chief constituent of
hardened high-carbon steels, and of medium-carbon nickel-steel
and manganese-steel.

Troostite is of doubtful composition, but possibly is an unstable
mixture of untransformed martensite with sorbite. It contains more
or less untransformed material, as it is too hard to be composed
entirely of the soft alpha modification, and it can also be tempered
more or less without changing in appearance. Its normal appearance as
rounded grains is given in Fig. 51; larger patches show practically
no relief in their structure,