RECENT attempts to explain the glacial aridity at upper elevations in the Galapagos Islands have failed to take into account either concurrent events in neighbouring areas1–6 or modern causes of climate. Yet, any reconstruction of the glacial climatology of the South Pacific must be consistent with all these data. Recognised glacial features of the region that must be taken into consideration and the empirical evidence which supports them include: increased upwelling along the equatorial divergence (enhanced plankton production7, simultaneous build up of guano8), upper elevation (650 m) aridity relative to the present in the Galapagos Islands (dated palaeomicrofloras1); increased precipitation above 2,000 m along the western slopes of the Peruvian Andes (deposits of oxides9,10, traces of humid cycle stream cutting and scouring of alluvium9–11, remains of sheet flooding at the base of the mountains9,10 a disproportionate lowering of the west facing glacial snowline12, relics of formerly more continuous plant and animal distributions13,14); continuous aridity throughout the Pleistocene along the Peruvian coast (unbroken sequence of arid cycle processes9–11); and exposure of large areas of the western Indo–Australian region following the 100–120 m sea level drop16 (fossil terraces16,17). It can be deduced from the position of this exposed land that the modern pattern of vernal mixing of water from the eastern and western Pacific could not have taken place17 and that the eastern Pacific anticyclone did not, therefore, weaken in the glacial summer17. Thus, the pressure gradients between the two sides of the ocean would have remained fixed throughout the year and the eastern anticyclone and the south-east trade winds would have retained high intensity in both summer and winter.