In a previous paper in this series a model was proposed lor the competition between plant populations with different rooting depths. This model predicts that in mixtures of plant populations with different rooting depths the Relative Yield Total will exceed unity. Secondly it predicts that in these mixtures the relative crowding coefficient of the deep rooting population with respect to the shallow rooting one will decrease with increasing plant frequency. Two competition experiments in deep pots are carried out to test these predictions. In the first experiment a comparison is made between a series in which nutrients limited plant growth severely and a series in which nutrients were applied in a quantity that allowed high productivity. In the nutrient poor series the Relative Yield Total exceeded unity by about twenty per cent, while in the nutrient rich series only a small deviation from one was observed. The second experiment was divided into a series of deep pots in which different rooting depths were possible and a series of shallow pots in which the two species were forced to share the same space. In the series of shallow pots the relative crowding coefficient of the two species was found to be independent of plant frequency. In the series of deep pots the relative corowding coefficient of the two species with respect to each other did decrease significantly with increasing frequency.The observed frequency-dependence for the shallow rooting species could be explained by an extension of the theory presented previously. In the spacing series used in the same experiment a decrease of the shoot to root ratio with increasing plant density was observed. Furthermore, in the replacement series the shoot to root ratio in mixture was found to be lower than in monoculture. The consequences of these phenomena for the estimation of the relative crowding coefficient and the Relative Yield Total on the basis of shoot weights are discussed. However, it is argued that the fact that shoot to root ratios change with plant frequency does not provide an alternative explanation for the frequency-dependence observed.