fig- 5°> the section at A being symmetrical will be straight when cold.
B, c, and D will become concave along the wider flanges, but in different
degrees, c less than B and D, because its top flange is wider. E, where the
flange is very wide, will not curve. A wide web resists the effect of flange
shrinkage because it is rigid, and it acts as a carrier of heat to the
shrinking smaller flange, delaying its setting. The gutter sections in the
next group, fig. 51, will all become concave on the solid sides. While
A and B will have a
curve in one direc-
tion, c will be curved
A B C both on the bottom
Fig. 51.—Illustrates the Camber of Castings produced by unequal Shrinkage and the Vertical side.
The difficulty which
confronts moulder and pattern-maker is how to counteract the effects of
shrinkage in unequal sections. No possible rule can be stated, and ex-
perience of similar classes of work is the only guide. The greater the
disproportion, the more flimsy the casting; and the greater its length, the
larger will be the departure from lineal accuracy. A moulder will sometimes
uncover a casting or a portion of the same while at a red heat, to hasten the
cooling, and so prevent curving. But that is not always practicable, nor
is it a sure method. Generally, the pattern-maker imparts camber to the
Fig. 52.—Crystallization in Cooling
pattern in the opposite direction from that which the casting would assume.
Uncertainty, when work is repetitive, is avoided by making one trial cast-
ing, noting its amount of camber, and altering the pattern accordingly.
Crystallization.—The needle-like crystals of cast metals arrange them-
selves normally in relation to the surfaces of the mould. In fig. 52 the
strongest form is shown at A, and the weakest at B. The cylinder at c,
terminated with a semi-sphere, is much stronger than one terminated with
a flat end. These are commonplace axioms, but they have infinite applica-