“…With any given value of applied pressure gradient, there exist either dual solutions or no solution at all; in particular, with zero applied pressure gradient, one solution branch yields static fluid with no temperature variation, whereas the second solution branch yields "passive convection", in which viscous dissipation provides the heat to maintain (through buoyancy forces) the flow that gives rise to the dissipation. Setting up the flow specified by this second solution branch would be very difficult, and indeed Miklavčič [2] has recently shown that solutions on this branch are unstable. Even on the stable first branch, flow velocities and temperature gradients are only of reasonable magnitude if the applied pressure gradient is of moderate magnitude; furthermore, large applied pressure gradients may violate the Boussinesq approximation (made in all the calculations) which requires any vertical dynamic pressure gradient to be small compared to the hydrostatic pressure gradient.…”