Optimal Treatment Dispersions in Rectangular Areas.

The classic method for managing spatial heterogeniety in agricultural field trials is the randomized complete block design. When treatments are properly assigned to blocks, the effects of spatial heterogeneity can be estimated using standard analysis of variance. Typically, treatments are randomly dispersed within blocks, but this may result in a systematic bias in error if treatments are not uniformly dispersed. We discuss different methods of restricted randomizations that provide optimal treatment dispersions and compare methods of recovering spatial information. The choice of methods used for spatial analysis may dictate the requirements for optimal dispersion.