ABSTRACT
The importance and value of setting out the experimental material into complete replicates has been discussed in Section 4.1 in relation to the resolvability of incomplete block designs. These comments apply also to row-column designs. Suppose the ks plots in each replicate are arranged in k rows and s columns. Then a resolvable row-column design will be needed if it is required to allow for possible differences between both rows and columns within replicates. For example, Williams and Matheson (1994) consider species trials in forestry where plots are usually square and contain a 5 x 5 or larger grid of trees. In this type of situation a two-dimensional blocking structure of rows and columns within replicates is appropriate, and likely to be much more effective than the one-dimensional blocking structure discussed in Chapter 4.
