Paleocurrent analysis of trough-shaped cosets of cross-bedding in the Brampton esker indicated a high flow variability during deposition. Within-site (w) and between-site (Beta ) estimates of the sampling distribution precision parameter ( k sd ) equalled 1.1317 and 6.4218 respectively, giving k sd = 36.7647. The semi-angle of confidence (psi ) about the grand vector resultant was 24 degrees at alpha = 0.01. As shown by the higher precision of Beta , future sampling in the esker would be increased in efficiency by using more sites than by increasing the number of observations at each site. Bartlett's test for homogeneity-of-variance showed that the large-scale cross-strata of this report had a significantly larger within-site variance than small-scale cross-laminae of an earlier report at alpha = 0.05 and 0.01. Preliminary results suggested that dunes may be more efficient estimators of the overall average flow direction than ripples in the esker, but further rigorous testing is required concerning this point. Analysis-of-variance of paleocurrent vectors demonstrated that the rank of ripple and dune structures as estimators of the grand vector resultant could not be based solely on w because (1) Beta was statistically significant at alpha = 0.01 and (2) the ripple and dune structures did not come from the same site.