The formation of longitudinal ripples in the deep benthic environment is one of the important problems discussed by McLean (1985) in his valuable contribution. We would like to discuss further, and comment on this particular aspect of his work. It seems clear that these longitudinal ripples are depositional features (McCave et al., 1984), and it further seems clear that their sharpcrested, symmetrical form and their preferred spacing (of the order of 1 m) can best be explained in terms of longitudinal vortex-pairs near the sea bed, likewise about 1 m apart. The essential problem is to explain the existence, and hopefully the spacing, of these vortices; any explanation must take into account the fine-grained character of the sediments themselves (McCave et al., 1984). Two mechanisms were discussed by McLean, as follows: (1) Secondary circulation, driven by Reynolds stresses resulting from lateral variations in bed roughness. This phenomenon was described and analysed mathematically by McLean himself in an earlier contribution (1981), and was shortly afterwards given the name Differential Roughness Secondary Flow (DRSF; Pantin et al., 1981). DRSF can be selfreinforcing if a rough substrate is overlain by less rough bands of sediment (e.g. sand ribbons over a partly exposed gravel substrate). However, McLean (1985) rejected DRSF as an explanation of the longitudinal ripples, since the relevant bottom photographs (McCave et al., 1984) do not show any significant lateral variations in bed roughness (McLean, 1985, p.262). (2) Lateral variations in bottom shear stress and in sediment content (McLean, 1985, p.262). In this previously undescribed mechanism, McLean assumes a velocity field consisting of streamwise-elongated zones of differing velocity and shear stress. He outlines the probable interaction between sediment content, turbulence, and shear stress, and shows that positive feedback in low-shear zones may cause preferential deposition of sediment, due to the damping effect of density stratification on turbulence. McLean does not specify the cause of the initial perturbations, or their possible wave number, although in situations of this kind a broad initial spectrum is normally assumed. McLean does not explicitly mention secondary flow in the context of this mechanism. The foregoing mechanism alone cannot explain the form of the ripples without making very specific assumptions regarding the velocity structure of the boundary layer. It seems far more likely that some type of secondary flow is involved. Some degree of secondary flow would certainly be caused by lateral density gradi-