Abstract
In the course of the last 20 or 25 years it has been established clearly that agricultural experiments carried out under field conditions are subject to an appreciable error, due chiefly to soil heterogeneity. In order to avoid the effect of this disturbing factor, which greatly affects the reliability of data given by field trials, two questions had to be carefully considered: (1) how to secure greater uniformity in soil conditions; and (2) what is the best method of getting a criterion of the accuracy of the experiment and an adequate interpretation of the results obtained?