In groups of experimentally infested sheep, worm distributions are markedly skew. In controlled anthelmintic tests, the worm burdens of treated and control sheep have different distributions and this invalidates the use of transformations. Five experiments are described, of which the first three describe the evolutionary steps taken to find a suitable method for interpreting the data. A non-parametric method was evolved and the last two experiments demonstrate the use of this test to interpret the results. The entire method is explained and tables are included which simplify its use for biologists with no statistical training.