Fifty educable retarded children who were determined to be nonconservers by means of three conservation pretests (discontinuous quantity, correspondence, and continuous quantity) were randomly assigned to one of five treatment groups (discontinuous quantity training, correspondence training, continuous quantity training, control, and control language groups). The four hypotheses were as follows: it is possible to train retardates to conserve quantity; one form of conservation training transfers to other forms of conservation; performance on conservation tests is related to mental age; and discontinuous quantity conservation appears before continuous quantity conservation. In general, the results of the posttests were consistent with the four hypothesis. Testing and training procedures are appended. (Author)