USE OF MOVING MEANS IN WHEAT YIELD TRIALS

Abstract

The efficiency of adjustments employing repeated controls was compared with the efficiency of moving mean adjustments in yield of wheat. Results reported show the moving mean of adjacent hybrid plots to give superior control of the experimental error. The best number of adjacent controls was tested and found to vary widely from test to test. Plant breeders may have to run several analyses to obtain the most accurate adjustment. An analysis of covariance was made on some trials to see if this technique would avoid overadjustment and give lower error variance. With few exceptions neither the use of covariance nor control plots gave as great a reduction in error variance as the optimum adjustment obtained by using the moving mean of adjacent plots.