Same data as in the previous figure, now recalculated with the assumption that the halo eccentricity is e = 0.54 (upper panel). The new fitted power-law line is given in the text. The dashed line gives the best-fitting de Vaucouleurs r^1/4 profile to the same data. Residuals from the power-law solution in the upper panel, plotted vs. azimuthal distance from the isophotal major axis (lower panel). Significant scatter remains, but there is no remaining trend with θ_a.