For peak power, 0.Three of the error of measurement; for different measures, 0.2 of the between-topic standard deviation in Desk 2, adjusted for small-sample bias.