Effects of plot size, plot shape, and number of replications on the efficiency of bean yield trials

Abstract does not appear. First page follows.

Introduction

Two types of yield trials are conducted on beans in California, one at the experiment station, the other in counties under the direction of extension service personnel. In the experiment station, care must be taken to maintain genetic purity since the plots serve as the source of seed for the following year. In yield trials the guard rows are not harvested. Most of the work of harvesting the plots at the experiment station is done by hand. The ideal therefore would be plots as small as possible. In the county tests, the introduction of mixtures in harvesting is not so important because the beans are discarded when weighed. Since the county officers are limited in the time they can put into plot tests, the ideal plots would be those which could be planted, cultivated, irrigated, and harvested with commercial equipment. Under these conditions the number of plots rather than their size is the most important factor. Under conditions at both sites, however, the experimental plans should he adequate to obtain least significant differences of 200 pounds per acre. Bean farmers think that a difference of this magnitude has practical significance. It is therefore a worthwhile objective to obtain experimental results precise enough for statistical significance at this level.

Review of Literature

Numerous experiments have been performed to get information on size and shape of experimental plots. (Love (1943))3 has pointed out that plot size depends to a large extent on the nature of the crop; hence the experimental samples must include a number of plants. With wheat, (Day (1920)) found a decrease in plot variability by increasing plot size by adding to either width or length. (Loessell (1936)) studied a field of pea beans in which he harvested 1,890 single-row plots 10 feet long. Then he compared the variability of