An investigation into inter‐plot interactions, in experiments with mildew on barley, using balanced designs

SUMMARY Serial designs, balanced for effects of neighbours, were used in 1975–1977 to investigate interactions between plots of spring barley given different treatments for the control of powdery mildew ( Erysiphe graminis DC. f. sp. hordei Mérat). Differences in amounts of disease, between similarly treated plots, could be related largely to the treatments applied to the neighbouring plots and to wind directions. Amounts of disease in plots were usually increased if the upwind neighbour was untreated but untreated plots themselves were also affected by neighbours, having least disease where the upwind neighbour was sprayed early. Differences in mildew due to treatment were reflected in grain yield. Yields also provided evidence of interactions between plots. Interference, as well as altering the average response to treatments, can also contribute to variability in experiment results and lead to substantial losses in efficiency. The yields in 1975 and 1976 provided strong evidence of profiles of fertility. In the analyses, adjustment by covariates, employed to allow for these profiles, gave substantial net gains in accuracy.