Some design-problems in crop experimentation .2. multiple blocking systems

This paper describes the construction, usefulness and randomization of several designs for field experiments in which there is more than one set of blocks, namely: (a) row-and-column designs, in which there are two crossing sets of blocks, treatments being applied to the plots formed by their intersections; (b) row-and-column designs in which factors are applied to complete rows or complete columns, that is, criss-cross (or strip-plot) designs; and (c) split-plot designs, in which the plots in a study of one factor are used as blocks in the study of another. All are examples of a wider class of designs, with many ramifications, said to have 'simple block structure'. It is suggested here that some of the assumptions underlying row-and-column designs are questionable. Some alternative approaches are indicated.