The key feature of a reverse roll coater is the fluid flow in the gap between two parallel rigid cylinders rotating with opposed surface velocities. A simplified model system studied by Greener and Middleman has the two rolls side-by-side and half-submerged in the coating liquid. Over a certain range of roll surface speed ratio, their experimental measurements of the flow rate through the gap deviated from predictions of a simple lubrication theory and recirculations were observed. This model configuration is examined experimentally and by finite element solutions of the Navier-Stokes equations in two dimensions. The results indicate that the film-transfer free surface and the recirculations under it do not significantly influence the flow rate through the gap. Deviations from lubrication theory occur only under conditions of low speed ratio and large gap when the effect of gravity becomes appreciable. Under such conditions, experimental error is introduced if the end effects of a third component of flow off the ends of the rolls is not accounted for or is not eliminated.