The use of statistical methods in agricultural experiments

Abstract

There is nothing new in the idea of agricultural experiments. Modern agriculture is largely the result of successful experiment, and the student of agricultural history finds that the facts of one generation are the fruits of the experiments of previous ones. Our own industrious Department of Agriculture has for over 30 years carried on a country-wide series of experiments on all types of crops and stock and every phase of agricultural effort. It is not generally appreciated what an enormous amount of agricultural experience has thereby been accumulated; few countries can boast that their agriculture is so well or so widely charted. It is not, however, with such schemes of multiple experiments that this paper is intended to deal; it may be remarked, however, in passing that from the broad point of arriving at a result from sheer weight of evidence the simple system of trials generally practised in these experiments is perfectly satisfactory. A comparison of different treatments in side-by-side plots in one hundred different centres is bound to provide at least interesting evidence of their relative merits under differing conditions.