Abstract
It is shown that Sleeswyk's (1974) contention that Venabie's (1961) twinning model contains a topological error is incorrect. A further type of twinning model is explored, in which the ‘pole' dislocation remains in the twin-matrix interface. It is considered that models of this type may explain the experimental observations in f.c.c. (and possibly also in b.c.c.) lattices in a more satisfactory manner than pole mechanisms of the conventional type.