The spherical harmonic and random filling methods, used to account for incomplete sky coverage in dipole calculations, are investigated. A study of their comparative performances is carried out and the former method is shown to be more accurate. Using both these methods, an attempt is made to find the regions responsible for the dipole in the Lahav optical catalogue. We find, in accordance with Lahav et al., that the Virgo–Centaurus region provides most of the dipole power. We also find strong indications that these structures are embedded within a global underlying anisotropy which could account for the dipole even if the Virgo–Centaurus region were absent and the anisotropics in the rest of the sphere, as quantified by the expansion of the galaxy surface density to quadrupole order, were extrapolated to this region. A picture of a global anisotropy in the distribution of galaxies throughout the unit sphere on which the various ‘attractors’ are superimposed is favoured.