Often the ultimate measure of performance in a field experiment is the product of a gross weight and a factor relating to quality, as when yield of dry matter is determined from gross yield and percentage dry matter. Since the factor is usually much less variable than the gross yield, yet relatively troublesome and expensive to determine, the practice of making independent estimates of it in every experiment involves diversion of an appreciable part of the total expenditure to a measurement whose inaccuracy will affect only slightly the final experimental error.

Proposals are made for determining the quality factor on only a fraction of all the experiments of one series and applying the averages to the gross yields of others in the same series; formulae and tables relating to the most economic ratio of incompletely to completely recorded experiments are presented.