Nature 394, 561– 563; (1998)

In the fourth paragraph from the end of this Letter, the values associated with the long and short data set differences were switched. The corrected statement should read: “For the ‘long’ data set, maximum surface winds average 3.6 m s−1 slower for observations made on Saturdays than those for Fridays (Fig. 2c). The unbiased ‘short’ data set has Saturday wind observations averaging 5.0 m s−1 slower than those on Friday; these differences are highly significant.”