This paper presents a study of the relationship between compensation and conservation in 80 children, ages 5-6. Findings are compared to those of an earlier experiment which was designed to determine whether or not prior knowledge of the compensation rule increases kindergarten children's susceptibility to conservation training experiences. It is suggested that the key finding of the first experiment (a correlation between pretraining performance on a compensation test and posttraining conservation performance) could have been an artifact of a spurious correlation between these two variables and an uncontrolled third variable (pretraining conservation performance). The previously reported experiment was replicated with the third variable controlled. Results of the present experiment provide no support for the hypothesis that prior knowledge of compensation does not predict relative trainability. (CS)