ABSTRACT
By Theorem 2, the average RT distribution function observed under condition (n, r) must be bounded, for all t > 0, from below by that from the corresponding (n + I, r)-experiment, and from above by (n + 1)/n times the lower bound. Consider the proportion, p, of 'short' RT's obtained for list length n, say, RT , to. Theorem 2 says that, not surprisingly, this should exceed the corresponding proportion observed with the longer list, n + 1; however, the 'short RT' property says that the proportion of short RTs for n items, p, must not exceed that for n + I, p' = Gr(n+l)(t), by too much, since p = Gm(t) can never be larger than (n + I)p' / n. An equivalent way of expressing this property is that the distribution functions for nand n + 1 must start from the same point rather than be shifted with respect to each other. Note that the 'short RT property does not depend on the ordering of the conditional distributions, Fk,r(t).
