The estimation of the efficiency of sampling, with special reference to sampling for yield in cereal experiments

The estimation of the experimental yields of cereal crops by sampling methods is considered in the light of results of eighteen experiments which were harvested by these methods at Rothamsted and its associated centres. The harvesting results of the Crop-Weather series of experiments are also included.
A preliminary discussion of the interpretation of the analysis of variance as applied to sampling results is given, and an expression is found for the loss of information arising out of sampling. The results of the discussion are applicable to all types of sampling carried out on replicated experiments.
The sampling and experimental errors of all the experiments have been calculated. The variation between sampling units in the same subplot (usually between l/100th and l/200th acre) was found to be such as would arise if each metre length of row had a standard deviation (about the plot mean) of the order of 25–30 per cent. of the mean yield.