The purpose of the study was to examine the influence of a nearby wall on the lift experienced by spheres fixed in a fluid stream, and the decay of this influence with increasing separation from the wall. For gaps less than one quarter of the sphere diameter, wall effects were found to dominate all dependencies other than that on Reynolds number. At larger gap ratios the lift vector rotates around the flow axis, sometimes this way and sometimes that but only occasionally changes direction abruptly. The rotation seems to be associated with free-stream turbulence and is inhibited by shear or by uneven surface roughness. These effects are not influenced by sphere mounting, and sphere rotation does not seem important.
Three series of experiments are described, each using different apparatus. In a Reynolds number range 3000–50000, calculated from sphere diameter and relative local velocity, repeatable lift values were observed in zones close to a flow boundary (the wind-tunnel wall), and remote from a boundary for smooth spheres. For rough spheres, results were erratic remote from the boundary.
Large persistent lift forces away from the boundary were found when the gap was less than 0·02 diameters and when it lay between 0·1 and 0·2 diameters (all these fractions being approximate). For larger gaps the lift force was strongly time-dependent and time-averaged values were smaller than those for gap ratios less than 0·2.
Lift forces are related to localized downstream displacements of the separation line round the sphere whose origin seems to lie in the response of the sphere's boundary layer to turbulent structures in the free stream. Weak regularities of the lift record are related not only to the incidence of such features but also to a natural time-scale of response of the boundary layer. Separation line disturbances do not propagate in a predictable fashion but do develop sharp boundaries as a result of recirculation in the wake.
In one series of experiments considerable pains were taken to allow the sphere to rotate freely about an axis parallel to the local tunnel wall and normal to the flow direction. No significant rotation occurred and the effects reported are not associated with sphere rotation. In shear flow remote from a boundary mean lift acted down the gradient of relative velocity.