Powers et al. [1991] observed through one-dimensional numerical work that larger NAPL blobs or higher water velocity would lead to greater departures from LEA. Subsequent experimental work [ Powers et al., 1992] with laboratory columns confirmed these tendencies and also found that small NAPL saturation or small dissolved NAPL concentration favored nonequilibrium. Soil grain size was found to affect patterns of NAPL distribution, which in turn affected dissolution rates. The authors developed a correlation of the Sherwood number to the Reynolds number, the median grain size, and a ``uniformity index'' indicative of grain-size distribution. It was suggested that differences in correlations could be due to different techniques of NAPL placement that resulted in different specific surface areas, and speculated that water flushing of layered or lenticular media over extended periods would favor nonequilibrium by reducing NAPL surface area. Simulations of stochastically generated media by Mayer and Miller [1992] found nonequilibrium concentrations at greater distances from a NAPL source than in homogeneous cases. Based on one-dimensional simulations matching a conceptual model to a breakthrough curve, Brusseau [1992a] found that contaminant removal could be strongly rate-limited even with a high mass-transfer rate, with nonequilibrium liquid-liquid transfer and medium heterogeneity being more important than nonequilibrium sorption in causing field-scale nonideality.