A number of techniques using discrimination reversal for construction of phyletic scale were discussed, and some theoretical modeJs which attempted to explain the behaviour shown in discrimination reversal were discussed in terms of the generality of each theory and its consistency with the data. Subjects of two species were tested on SDR under a number of conditions in an attempt to throw a little more light on aspects of discrimination reversal learning, and it was concluded that improvement in SDR is determined by the characteristics of the individual subject and the particular schedule employed rather than the species of the subject, so that SDR is not of use for phylogenetic scaling. An ORE was found in conditions such that it would not be predicted by any of the theories considered except that of Capaldi and Stevenson (1957 ). The results of this research indicate that where improvement is found in SDR, the effect of training level is to increase discriminative performance in all subsequent reversals.