Abstract
A sequential method of estimating the minimum of a random variable achieves the desired level of confidence only for interval widths that are large relative to the density of the distribution at or near the minimum. A modification of the procedure is suggested.
Keywords: