Published online by Cambridge University Press: 27 March 2009
1. Four cereal experiments, comprising 210 plots each about l/40th acre in area, were harvested by a sampling method. Three of the experiments were later harvested by large-scale methods, so that a direct comparison could be made.
2. The field technique is described, and an account is given of the small combined thresher and winnower which was constructed for the purpose of dealing rapidly with the numerous small sheaflets.
3. The results are analysed in detail and it is shown that the samplingerrors per plot lie between 5 and 6 per cent. of the mean yield, and that these errors are sufficiently low for there to be little loss of information.