Published online by Cambridge University Press: 27 March 2009
(1) The edge rows give significantly higher yields than the inside rows, indicating thereby the inadvisability of using edge rows in yield trials.
(2) The variation between rows is very much greater than within rows. Different parts of the same drill row should therefore not be regarded as subject to independent error. The present investigation emphatically confirms Clapham's conclusions on this point.
(3) A slight advantage may be gained by the subdivision of the area to be sampled, without additional labour.
(4) In order to study the effect of the structure of a sampling unit of given size five types of unit have been examined. Of these, method (1), in which the “sampling unit” consists of four parallel half-metre lengths on adjacent rows, appears to be the most precise, and may be recommended on the basis of this trial. The half-metres within such sampling units appear to be negatively rather than positively correlated, and a significantly lower sampling error is obtained in consequence.