Four species of grassland plant, Plantago lanceolata, Holcus lanatus, Lolium perenne and Rumex acetosa, were grown as monocultures and mixtures in pots of nutrient poor soil in a glasshouse for 8 months. There were four plants per pot and these were arranged in two competition modes: either root and shoot interactions were permitted, or only roots allowed to interact by using above-ground partitions. Time of introduction of seedlings was varied to give a range of plant size ratios at the start of the experiment. The factorial design catered for all combinations of species, competition modes and planting times, replicated in four blocks. The shoots were clipped at a fixed height at each of five harvests. Rumex grew badly and was mostly omitted from analysis of the data.
By (i) following the change in the relationship of clip dry weights against planting time with successive harvests, (ii) plotting the change in the logarithm of the ratio of cumulative clip dry weights with time and (iii) the use of de Wit logarithmic ratio plots it was demonstrated that each monoculture and mixture combination's ratios of plant weights converged towards stable equilibrium values. Three hypotheses are put forward to explain why in monocultures a smaller plant was at a competitive advantage relative to a larger neighbour and was not suppressed in its growth by the latter. In mixtures this plant size effect was superimposed to different extents on the relative aggressiveness of the species considered. It was concluded that in a nutrient poor soil, when competition for light was low, root interactions can promote the co-existence of neighbouring plant species.