THE efficacy of the Latin-square design 1 in eliminating two-dimensional variation from treatment comparisons, particularly in field experiments, has long been established. Kendall 2 states that ...