Preliminary experiments carried out in the summer of 1928 had indicated that sampling methods might be used to obtain a satisfactorily precise estimate of the yield of small cereal plots(1). It was thought desirable to test the method more thoroughly, and to compare the estimates with figures obtained by large-scale methods. Accordingly samples were taken from all the plots of the three Rothamsted cereal experiments of 1928, the plots later being harvested with a binder in the ordinary manner. Samples were also taken from the 16 plots of a small experiment on barley at Wellingore, Lines., but no direct measurement of total yield was made for these plots. There were 210 plots in all, but of these 50 were sampled in four sections, corresponding with minor differences in manurial treatment; so that the total number of plots dealt with amounted to 360. The large-scale method, however, treated each of the 50 as a single plot, and thus provided only 210 yield figures.