The proposition that continental crust consolidated during late Archaean-early Proterozoic times as spatially distinct groups of cratons is tested using palaeomagnetic results assigned to the interval 2900-2200 Ma. Geological data indicate the existence of two nuclei following widespread cratonisation at ∼2700 Ma comprising ‘Ur’ (stabilisation at ∼3000 Ma and shallow water cover at 3000-2800 Ma) including central-southern Africa, western Australia and India, and ‘Arctica’ which included Laurentia, Fennoscandia and Siberia (stabilisation by ∼2500 Ma with supracrustal cover beginning at ∼2400 Ma). A third nucleus ‘Atlantica’, comprising much of western Africa and eastern South America, consolidated later (basement stabilisation at 2200-2000 Ma and ∼2000 Ma fluvial-deltaic cover). Using this geometrical premise, palaeomagnetic poles assigned to the interval ∼2900-2200 Ma from the two older nuclei are observed to plot along two limbs of an APW swathe. Over 80% of the poles plot on a single segment running through the west of Africa and terminating in a tight group over Australia and India with ‘Ur’ resided in high latitudes and ‘Arctica’ in low latitudes. The 2900-2200 Ma mean poles from Africa, Australia, Fennoscandia and Laurentia also accord closely with one another and show that only small relative movements occurred between the cratons during these times. This polar grouping does not result from datasets of similar temporal concentration and is due to low rates of continental movement (∼1 cm/year) that are probably the signature of small-scale tectonic regimes. The primeval continental crust consolidated as two linear belts with a configuration approximating to the geoid form and suggesting an aggregation by whole mantle convection. The relative positions of ‘Arctica’ and ‘Ur’ at 2900-2200 Ma were only marginally different from their positions within the putative later supercontinent of Palaeopangaea (∼1100-550 Ma) and imply that the Proterozoic eon was characterised by relatively small intracratonic motions close to limits of palaeomagnetic detection.