The variability to be expected in replicate plankton hauls, using several methods of hauling and a number of different nets, has been examined by Winsor & Clarke (1940). Some of the variability is due to true sampling variations, inevitable when discrete entities in suspension are sampled, and some is due to inadequacies of sampling technique. Winsor & Walford (1936) found that the variability of their vertical net hauls could be explained on the basis of a random distribution of the population, the variations being ascribed to variations in the volume of water filtered in successive hauls. They fully realized that the alternative, namely a non-random distribution of the population, was not disproved; but they considered that since widely different organisms showed little difference in variability the evidence was strongly in favour of a random distribution. They did note, however, that agreement between these different organisms was less marked when the numbers caught were large.