Published online by Cambridge University Press: 27 March 2009
Three sampling experiments were conducted to examine the effect of sample size and sampling intensity on the precision of dry-matter content and botanical composition estimates of perennial rye-grass-white clover herbage. One of these experiments examined the between-sample variability of these attributes and of dry-matter yield in relation to other sources of experimental error in a small-plot sward trial. The sample sizes examined were 800 g, 400 g, 200 g, 100 g, 50 g and 25 g green weight. In general the accuracy of dry-matter content and botanical composition estimates decreased with decreasing sample size. The between-sample variabilities of 25 g and 50g samples were high in relation to their between-plot variabilities. Single 100 g samples provided reasonably good estimates of these attributes and of dry-matter yield, but single 200 g samples provided a more satisfactory margin for error. Samples larger than 200 g appeared to be unnecessary. When weight-for-weight comparisons of single and duplicate samples were made there appeared to be little advantage in duplicate sampling. A theoretical examination of measurement inaccuracies inherent in the techniques used in small-plot sward trialssuggested that variation in plot length measurements in particular may make an undesirable contribution to the variability of such trials. A procedure for the conduct of small-plot trials is recommended. It is concluded that, where plot size and replication are limited, further improvement in the precision of such trials will not be readily attainable.