Abstract
Information about experimental units is usually available before they are assigned to treatments in agricultural or biological research. Such information, however, is taken into consideration in few available designs, if any. Based on the objective of homogeneity between groups, a new criterion (X- criterion) is suggested for measuring the goodness of allocations or designs. Under certain conditions, the ^.-optimum design is equal to D-optimum and A-optimum designs proposed by Harville (1974a), for a covariance model