Beyond Latin squares

Abstract

Field experiments oft en show heterogeneity and trend in both rows and columns. It is therefore useful to consider blocking in both rows and columns of the field layout of plots. We provide a brief review of row-column designs and demonstrate the particular advantage of resolvable designs, which allow the treatments to be spread out over the experimental field by latinization of rows and/or columns of plots and by evenly distributing treatments between complete replicates, thus largely avoiding a clumped placement of replications of a treatment in a limited area of the experiment. An example from a field trial with silage maize (Zea mays L.) is used to illustrate the analysis of a latinized design using mixed model procedures. Show more