The authors regret a small number of numerical errors in their paper. The values for the slopes reported in the caption of Figure 14 currently reads: ‘(a) 0.81±0.06, (b) 0.06±0.03, (c) 0.73±0.06, and (d) −0.28±0.06’. These values were derived from earlier calculations of these slopes that were later revised. The correct values (in italics in this corrigendum) are ‘(a) 0.60±0.05, (b) −0.06±0.04, (c) 0.64±0.06, and (d) −0.36±0.06’. Incorrect values of the slopes are also reported in the main text of the paper, and corrections to the main text are specified below. The panels in Figure 14 are correct, and refer to the revised values for the slopes. In the main text, on page 72, on the fifth line of the fourth (bottom left) paragraph (14 lines up from the bottom of the page), the corrected sentence should be ‘A least squares fit of ln (k1) vs. ln (P) yields a slope of 0.60±0.05’. On the same paragraph, two sentences below, the corrected sentence should be ‘For the regression of ln (K¯D,β−1) vs. ln (P), we obtain a slope of −0.36±−0.06, smaller in magnitude than the slope of −0.42 found by Honeyman et al. (1988) from the same field data used to obtain their ln (k1) vs. ln (P) slope’. Finally, on the fifth paragraph of page 72 (the second paragraph from the top right), 13 lines from the top of the right column, the corrected sentence should be ‘We find that for the relationship between ln (k−1) and ln (P), τ=−0.18 (p=0.11) and the slope of the least squares fit is -0.06±0.04 (r=−0.19)’. While the values reported in the paper are incorrect, they do not change the discussion or conclusions of this paper. The authors apologize for these mistakes and any inconvenience they may have caused.