By means of a preferential etch technique in conjunction with transmission electron microscopy, a stress‐annealed pyrolytic graphite has been shown to be a compact of crystallites withLavalues typically lying in the range 1–40 &mgr;. The boundaries defining these crystallites varied from the low‐angle tilt boundaries &thgr;∼1′, found in Ticonderoga crystals, to high‐angle boundaries denoting misorientations of up to 30°. In contrast to the low over‐all boron concentration of the material, 1–2 ppm wt/wt, the concentration of this element residing in the intercrystallite boundaries was calculated at ∼103ppm wt/wt. This provides confirmation of the premise that the lattice disorder existing across a boundary trace supports a vacancy concentration far in excess of that pertaining to more perfect regions of the graphite. It is postulated that at temperatures >2100°C this enhanced vacancy concentration will allow a dynamic interchange mechanism for crystallite growth, having an activation energy given byE*= ½(Efv+Efi) +Emai. Nonbasal screws have a deleterious effect on the rate of graphitization for two principal reasons. First, the nonbasal screws residing in intercrystallite boundaries effectively stabilize the excessive vacancy concentration at the latter sites. Second, if the distance of closest approach of these dislocations becomes less than some critical value, they may be responsible for creating new boundaries. As it appears unlikely that many formal boundaries, which can be described in terms of lattice defects, will exist between neighboring crystallites in pyrolytic graphite, the initial stages of graphitization must occur by an alternative mechanism. The recent work of Murty would seem to imply that the latter is in fact an interstitial mechanism, having an activation energy of 9.8±0.7 eV as compared to the value of 8.3±0.3 eV predicted for the dynamic interchange mechanism.