atoms is relatively hindered due to stenc reasons. One would expect that 0 2 develops a rounded pit in the early stage which leads to a circular pit. When tungsten carbide is present on the surface, 0 atoms are expected to be present near the surface because tungsten carbide is effective in dissociating 0 2 . It might involve an oxygen spillover process similar to the hydrogen spillover phenomenon (Levy and Boudart, 1974). With 0 atoms the steric factor should not be important and indeed hexagonal pits are formed by 0 atoms from the gas phase (Wong et al., 1983). A spillover mechanism was earlier proposed for chrornia catalyzed carbon oxidation (McKee, 1970 Baker, 1981. A further for the tungsten being in the form of tungsten carbide is that any metallic tungsten would have been oxidized, at least on the surface layers, under our conditions, and tungsten oxide behaves very differently in its catalytic activity. Our results on WOs catalyzed carbon oxidation will be reported shortly.The orientation of the hexagonal etch pits has been determined by in situ 4ectron diffraction using TEM. All sides of the pits are in the (1,010) direction i.e., they are zigzag surfaces, which is the orientation shown in Figure 4, as well as the orientation of the hexagonal pits etched by 0 atoms (Wong et al., 1983). Deep pits created by catalyst particles are mostly circular, but in a few cases hexagonal, e.g., by Fez03 (Thomas, 1965;McKee, 1970). The mechanism proposed here may also be operative in deep pits formation.