In most fields of natural science, the density of individuals is usually determined from counts in defined areas of known dimensions. When tree density is being measured, the delimitation of plots of up to a hectare or acre in area, presents some difficulties and is time-consuming. Because of the reciprocal relationship between. density and nearness of individuals to one another it is necessary to kuow only the distance between regularlyspaced individuals to be able to calculate density. Each plant has available to it an area equal to the square of the between-plant distance. If the individuals are distributed not regularly but at random, density may still be determined from the spacing, since between-plant distances will vary at random from the mean distance. In most. if not all, natural communities there is a departure from randomness. i.e. individuals tend to be aggregated to some degree (GreigSmith, 1964). For most forest sampling methods this non~randomness has usually been considered to be insufficient to affect the results seriously. Mean distances may be determined by methods based on between-plant distances or, as in the point-centred quarter method, on point-to-plant distances. Since being described by Cottam and Curtis (1956), the point-centred quarter method has become accepted, along with other methods of plotless sampling, as one .of the most efficient for obtaining quantitative data on forest trees.