Abstract: | This paper proposes a model for outliers and fertility jumps in agricultural field trials when the fertility is correlated in two directions. Uncertainty concerning the occurrence of these anomalies can be substantial, especially when outliers occur near a fertility jump. The proposed estimation technique can take this uncertainty into account rather than condition on a particular set of anomalies. Diagnostics for the detection of these anomalies are also discussed. The procedures are illustrated with a wheat trial. |