The weak-beam method of electron microscopy (Cockayne, Ray & Whelan 1969, 1970) has been used to investigate the dissociation of dislocations in silicon. Total dislocations with a/2<110> Burgers vectors were found to be dissociated into Shockley partial dislocations, with a separation of 7.5 +0.6 nm (75 + 6 Å) for the pure edge orientation and 4.1 +0.6 nm (41+ 6 Å) for the pure screw orientation. The intrinsic stacking-fault energy, calculated from the measured dissociation width using anisotropic elasticity theory, is 51 + 5 mJ m
-2
(51 + 5 erg cm
-2
). The method has also been used to image partial dislocations at threefold dislocation nodes in silicon. All nodes in the specimens examined were found to be extended, and of about the same size, indicating that the intrinsic and extrinsic stacking-fault energies are comparable. Measurements of the radii of curvature of partial dislocations at the nodes gave a value of 50+15 mJ m
-2
(50+15 erg cm
-2
) for the intrinsic stacking fault energy, using the method of Whelan (1959) as modified by Brown & Thölén (1964). Dislocations in silicon specimens annealed at a high temperature were found to be constricted along segments of the dislocation line. Evidence is presented which suggests that the constricted segments have climbed out of the slip plane.
