Duplicate detn. on each of 2 field samples from 3 replications of 5 fertilizer treatments on Alyce clover, Alysicarpus vaginalis, provided data for estimating variances due to plots, samples and chem. detn. These estimates were obtained for % of P, K, Ca and Mg in the clover hay. The accuracy of treatment means involving different numbers of plots, samples and detn. was examined. Relative costs for the 3 phases of the procedure were estimated and the cost per unit of information was computed for the various schemes under study. The opt. ratio of plots: samples: detn. was calculated for a constant variance of the mean. In general, the relatively high cost and low variance of the laboratory detn. require that this part of the technique be reduced to a minimum. The opt. ratio of total samples to total detn. per treatment varied from 4 for Mg to 16 for K. Except for the unusually low plot-to-plot variance of Mg, the opt. number of samples per plot ranged from 2 for P to 4 for Ca.