From the text...SUMMARY: '1. The most straightforward method of assessing the degree of non-randomness, if any, of a plant population is to collect a sample of distances from random points to the plant individuals nearest them. A knowledge of the density of the individuals, independently determined, is also necessary. 2. As an index of non-randomness a = pDw is suggested, where D is density and w is the mean of the squares of the point-to-plant distances. a is equal to, less than or greater than (n-1)/n according as the population is random, regular or aggregated. The significance of a departure of a from this value is easily found since 2na is distributed like x² with 2n degrees of freedom. Observed values of a from two non-random populations may be compared by a t-test. 3. The advantage of using this index is that it will reveal all the non-randomness present and not merely the smallest scale of non-randomness as an index based on plant-to-plant distances would. Also no distances need be measured from randomly chosen plants; the selection of truly random plants is exceedingly laborious and a biased sample is useless as it is likely to give most misleading results. 4. Owing to the fact that point-to-plant distances may sometimes have the same frequency distribution in random, regular or aggregated populations, the observed distribution of this variate will not necessarily, by itself, reveal non-randomness. The writer is at present investigating the distribution of point-to-plant distances in regular populations.'