In work before 1991, the papers of Hunt et al. [1988a,b] suggested that mass-transfer limitations would render groundwater extraction ineffective in removing NAPL trapped at residual saturation by capillary forces, recommending steam injection to remedy the situation. Analysis, supported by sand-tank experiments, predicted that a low fraction of solubility would be achieved under typical pump-and-treat conditions. Parker [1989] and Sleep and Sykes [1989] postulated that LEA could fail to be satisfied at field scale even if it held at pore scale, because heterogeneity could cause phases to be out of contact at many locations. Miller et al. [1990] experimentally studied conceptual mass-transfer models based on a driving force (the difference between a component's concentration in a phase and its equilibrium concentration in that phase) and the interfacial area between phases. The models correlated the Sherwood number (mass-transfer coefficient times square of flow diameter, divided by molecular diffusivity) to the Reynolds number (water velocity times water density times flow diameter, divided by water viscosity), the saturation of NAPL, and the Schmidt number (water viscosity divided by product of water density and molecular diffusivity). In homogeneous glass-bead columns, they observed fairly high transfer rates and suggested that LEA could be valid.