Application of a Latin Square Change-Over Design to Dairy Cattle Grazing Experiments

A Latin square change-over design was applied to dairy cattle grazing experiments. Experimental procedure is given. Analysis of the data is presented, including the fitting of constants for nonorthogonal variables. The design gave substantial gains in efficiency over single grouping or complete randomization but required more work to analyze the data and may have led to some confounding of effects of pasture mixtures with effects of other variables.