field are seen in the THEMIS data (Fig. ), yet are absent or barely discernible in the MOC wide-angle camera (MOC WA) and Mars Orbiter Laser Altimeter topographic data. One MOC NA image crosses several of these features and reveals that they are exposures of underlying bright terrain where dunes are not present (Fig. ). These features are probably locations of aeolian scour, which has exposed unit E from beneath the pervasive dune field.The ubiquitous exposure of unit E within unit Ph strongly suggests that it underlies all of the haematite-rich plains. Unit E is exposed in craters of all sizes, down to the resolution of the MOC NA images (,2 m), indicating that unit Ph is very thin. In all cases, unit E appears as bright, high thermal inertia outcrops that contrast sharply with unit Ph. Evidence of expansive dune fields and embayment of positive topographic expressions of unit E suggests that the thin, dark, haematite-bearing layer is mobile and may be the primary soil constituent. Unit Ph is probably a lag deposit made of dense, coarse, haematite-rich grains derived from the erosion of previously existing overlying layers, and this is the source of the haematite signature detected by the TES instrument. Conversely, unit E consists of in-place underlying bedrock, as is evident from its high thermal inertia and its layered exposures seen in images.In the Opportunity landing site crater, ,1/2-m-thick, finely layered units are exposed in the rim 3 and embayed by dark, haematite-rich soil 15 . These outcrops are probably the uppermost expressions of unit E and correlate with light-toned layers observed throughout the haematite-bearing plain, an area of 1 ยฃ 10 5 km 2 . Moreover, unit E outcrops along roughly half of the unit Ph margin and is mappable over hundreds of kilometres to the east, north and west. In places, unit E disappears underneath younger mantling materials, masking its full extent. In totality, exposures of this etched material cover an area of 3.3 ยฃ 10 5 km 2 that spans 208 of longitude and 148 of latitude, more than three times the extent of the haematite-bearing plain itself. Opportunity has detected high concentrations of sulphur-bearing minerals and unique textures in the observed bedrock that are indicative of water saturation for long periods of time 3 . Thus, whatever aqueous process altered, and perhaps formed, the layered units at the landing site 3 must have acted over an extensive area and a large volume of material. If the outcrops at the landing site are indeed a result of deposition in a sea, then that body of water had to be comparable in area to Earth's Baltic Sea and must have been deep enough, and persisted long enough, to build up at least a 0.5-km-thick stack of sediments. For this to occur, the ancient global climate of Mars must have been very different from its present climate and have lasted for an extended period.A