It has been proposed that the cosmological constant might be measured from geometric eeects on large-scale structure. A positive vacuum density leads to correlation-function contours which are squashed in the radial direction when calculated assuming a matter-dominated model. We show that this eeect will be somewhat harder to detect than previous calculations have suggested: the squashing factor is likely to be < 1:3, given realistic constraints on the matter contribution to. Moreover, the geometrical distortion risks being confused with the redshift-space distortions caused by the peculiar velocities associated with the growth of galaxy clustering. These depend on the density and bias parameters via the combination 0:6 =b, and we show that the main practical eeect of a geometrical attening factor F is to simulate gravitational instability with ee ' 0:5(F ? 1). Nevertheless, with datasets of suucient size it is possible to distinguish the two eeects. We discuss in detail how this should be done, and give a maximum-likelihood method for extracting and from anisotropic power-spectrum data. New-generation redshift surveys of galaxies and quasars are potentially capable of detecting a non-zero vacuum density, if it exists at a cosmologically interesting level.