Morgan has described a spatial instability in the flow through screens or grids of small open-area ratio. Head and Rechenberg and others have observed large span-wise variations in the thickness and shear stress of nominally two-dimensional boundary layers on flat plates and aerofoils in wind tunnels. It now appears that these spanwise variations are caused by the instability of flow through the screens. The jets of air issuing from the pores of the screen attempt to entrain more air by the usual mixing processes, but can only entrain it from each other, so that groups of jets coalesce in rather random (steady) patterns determined by small irregularities in the weave. The resulting variations in axial velocity are virtually eliminated by the wind tunnel contraction, but variations in flow direction are not so greatly reduced: a theoretical analysis shows that the observed variations of boundary-layer thickness, which often reach ± 10 per cent of the mean, can be produced by directional variations in the working section of the order of ± 1/20 deg, with a spanwise wavelength of the same order as the boundary-layer thickness.