Within each stratum we have to determine the total area (or length if sampling along rows) to be measured, and how this is divided up. For example, if the unit is divided into suitable strata and one stratum has an area of 500 m 2 we could decide to measure the crop on 50 m 2. This could be one 50 m 2 square quadrat or ten 5 m 2 quadrats. The total sample area required depends on the precision required and on the variation in crop yield across the stratum. However, if the study involves replicated plots and farms, as is often the case, it is only possible to improve the precision of the estimate of crop yield from a single unit to a limited extent by increasing the sampled area within it.