The success of a breeding program depends upon the ability of the plant breeder to not only develop new lines, but also to determine which lines or varieties are superior. For this purpose, yield trials are an important yardstick for determining the value of the trial in assessing the merits of the different lines. Variation encountered is due to several factors; with true genetic differences, environment, and human error being of major importance. In order for yield trials to give meaningful information on a line or variety, it is essential that variation due to human error and environment be reduced to a minimum. Uniform space planting, careful cultivation, and irrigation can help reduce the human error. However, the effect of environment, mainly soil heterogeneity, is not so easily handled. Randomization, replication, and planting experimental plots of the proper shape with sufficient size to give the desired information are techniques devised to minimize the effect of soil heterogeneity. In general, it has been found that the larger the experimental plots, the smaller the variation (1, 23). Also, plots with a large length to width ratio lying in the direction of greatest soil variability have been found to decrease the variability of the yield trial (2, 8, 32, 35). However, both expense and convenience will modify the size and shape of plot that is desired. The optimum size and shape of a plot would be one that gives a maximum amount of information at a minimum cost and be of convenient handling dimensions. Such information must be determined for each individual crop.