Field trials when observations are correlated with those in neighbouring or nearby plots in one dimension are examined. The relationships between maximum likelihood solutions and the well-known Papadakis estimator are clarified and it is shown that the Papadakis estimator is appropriate under one covariance assumption but not others.