Water flow from the bulk soil to a plant root has been viewed (Philip, 1957 ; Gardner, 1960; Cowan, 1965; Molz, Remson, Fundaroli, and Drake, 1968) as flow from a hollow cylinder of soil of infinite length with the root radius forming the inside boundary and the half distance between adjacent roots the outside boundary. The geometry of flow at the rhizosphere (Newman, 1969) is complicated and the physics of the process very poorly understood ; therefore it has been necessary to restrict analysis to this very simple model. Despite the grossly simplified assumption of uniform absorption by parallel roots of uniform radius, these models have a quantita tive value provided the input data are consistent with experimental observations. The models predict the water potential at the root surface (rr) from experimental estimates of the water extraction rate per unit volume of soil (Q), the root radius (r), the length of root per unit volume of soil (= root density) (L), the soil water diffusi vity or unsaturated hydraulic conductivity, and the bulk soil water potential (rb). There is general agreement (Cowan and Milthorpe, 1968 ; Newman, 1969 ; Lawlor, 1972 ; Gardner, 1973 ; Williams, 1974a) that the original predictions (Gardner, 1960 ; Cowan, 1965) of large potential differences away from the plant root resulted because root densities and water extraction rates used in the models were approximately an order of magnitude different from experimental observations. Recent theoretical work (Williams, 1974a) shows that these simple models predict small differences in water potential between root and soil for most root densities, water extraction rates, and soil moisture conditions encountered in the soil plant system. In these calcula tions which were based on a Yolo light clay as originally used by Cowan (1965) a root radius of 5 x 10~2 cm, representative of coarse roots, was used. This note considers the influence of a wide range of root radii and root densities on the predicted water potential differences between root and bulk soil using the models of Gardner (1960) and Cowan (1965). Computational procedures are given by Williams (1974a). Table 1 sets out the relevant values for the input terms to the models.