Missing-plot techniques.

In experimental work it frequently happens that one or more experimental units is missing from the data or has to be rejected because of conditions outside the control of the experimenter. It should be cautioned that observa? tions should be rejected in the analysis of results only under extreme circumstances, when it is quite obvious that the treatment being studied is not responsible for the ap? parently anomalous results. One of the first papers on the subject of estimating the yield of a missing unit in field experimental work was published by Allan and Wishart (1). They derived form? ulas and illustrated their use for a single missing plot in a randomized block and in a Latin square experiment These methods were extended by Yates (7) to cover several missing units in a given experiment The formula given by Yates for estimating the yield of a single missing unit in a random? ized block experiment is