Zinc crystal wires in which the basal plane was nearly parallel to the wire axis were found to twin at abnormally high stresses. No single critical resolved shear stress exists; resolved shear stresses near 3⋅7 Kg/mm
2
were common for short crystals, while longer crystals (which often fractured at the instant of twinning) had an average critical stress of 2⋅9 Kg/mm
2
. All crystals experienced detectable plastic deformation before twinning, by slip on the plane (112¯2) in the direction [1¯1¯23], for which the critical resolved shear stress in tension was 1⋅0 to 1⋅5 Kg/mm
2
. The existence of this slip system was placed beyond doubt by observations on slip lines and lattice rotations. The twinning stress is determined by the work-hardening accompanying the slip on (112¯2), which continues until the lattice has rotated sufficiently for slip to begin on the (0001) planes, originally impeded by the specimen grips. The observations suggest that, quite generally, slip is the necessary prelude to the nucleation of a twin. A model for twin nucleation is developed in detail; it postulates the creation, by homogeneous lattice shear, of a twin nucleus about 250 Å in diameter, at a site where the applied stress is magnified by an array of dislocations. Twins are nucleated at a stress which depends on the
slip
characteristics of the particular specimen; hence the great spread in published twinning stresses. Crystals which were indented under stress twinned immediately if the resolved stress exceeded about 2 Kg/mm
2
, but indentations made before the test were ineffective, probably because of a strain-ageing effect. Ageing also led to a slow reduction of the high stresses needed to produce basal slip in the twinned crystals. Immediately after the twin had been created, a shear stress, resolved in the basal plane, of 550 g/mm
2
was required; this fell to 200 g/mm
2
two years later, which is still six times the normal basal slip stress. Basal slip in twinned crystals was accompanied by considerable work-softening. Stresses needed to thicken existing twins were very variable but much lower than the stresses for twin creation; thickening of twins always took place smoothly and was never accompanied by stress relaxation.
A Metallurgical Approach to the Cracking Resistance of Hot-Dip Galvanized Zinc Coatings