REPORTS MULTIVARIATE correlational analyses of the relation of treatment (five beginning reading instruction systems) to achievement (10 Stanford tests) in the USOE Cooperative Reading Studies Second-Grade Phase data. Multiple partialing was used to control for sex, 10 project sites, and eight readiness tests. Canonical correlations of treatment residuals with achievement residuals showed very slight treatment effects. It is argued that this multivariate re-analysis has a particular elegance and fidelity to the design philosophy of the studies. It is also argued that the lack of differential effectiveness of the treatments should not be allowed to mask the fact that all the instructional systems were effective. The intercorrelations among the achievement residuals from regressions on all predictors were factor analyzed. The single important factor that emerged was of the g type, and it was named general intellectual development. It is argued that this factor provides the best criterion variable for further curriculum research and that it operationalizes the contribution that primary education makes to general intelligence. Growth increments in general intelligence are posited as the fundamental product of primary education.