In a classic paper, Levitt (1971) described an adaptive procedure for
estimating points on the psychometric function known as the transforme
d up-down method. Levitt discussed the assumptions of the method and p
resented a brief table with simple rules that converge to a few differ
ent points on the psychometric function. Levitt's original table conta
ins only the simplest rules, and sparsely covers the range of the psyc
hometric function. This paper provides a table with previously unpubli
shed rules which cover the range of the psychometric function at 5% in
tervals. There is a brief review of the major issues in adaptive testi
ng. Technical issues such as the mean length and logical construction
of the new rules are discussed.