A recent paper in this journal by G. Barton proposed an efficient algorithm for locating locally optimal alignments between two sequences. Although the paper claims that all such alignments are found, the approach frequently fails to detect some of the significant matches. This note explains the deficiency.