Oil on troubled waters.

either by changing the ice sheet's basal thermal regime or by introducing water into the path of the ice stream through periodic drainage events. If this proves to be the case, it would substantially change our understanding of the effect of subglacial lakes on ice-sheet stability. It indicates that basal hydrology needs to be accounted for in numerical models of ice-sheet motion because it has the potential to rapidly induce fast flow. The recent sudden acceleration of ice streams in Greenland 6,7 , and its resultant effect on the mass balance of the Greenland ice sheet 8 , underscores the importance of understanding the dynamics of ice-sheet response to climatic change. Might the geographical coincidence of the Recovery lakes and the start of fast flow be just that, a coincidence? So far we have not seen any other large subglacial lakes at the heads of ice streams. But Antarctica remains one of the most poorly mapped parts of our planet — we know more about the surface of the dark side of the Moon than we do about the bedrock or even the surface topography of Antarctica. Antarctica's surface is being mapped at ever higher resolution with remote-sensing satellites such as NASA's ICESat or the planned European Space Agency mission CryoSat-2, but radar imaging of the bedrock is not keeping pace. Proposed missions to put ice-sounding radar instruments on board a satellite, with the goal of mapping the bed in Greenland and Antarctica, have yet to be approved. For now, continued airborne and ground-based geo-physical exploration is the only way forward for imaging the ice sheet's base. During the forthcoming International Polar Year, a number of land traverses and aerial campaigns are planned for Antarctica. In 2008, a Norwegian–US traverse will survey the Recovery lakes area. This was last visited by the US–UK traverse from the South Pole to Dron-ning Maud Land in 1965–66. Interestingly, that early expedition, which was equipped with one of the first radars, also crossed over the lakes discovered by Bell and colleagues (Fig. 2). The radar equipment was relatively primitive and used only intermittently, so the lakes were not identified in the field. But a logbook duly noted that Gordon de Q. Robin, then head of the Scott Polar Research Institute in Cambridge, UK, and one of the pioneers of radar glaciol-ogy, had suggested the existence of a " …pos-sible melt-layer at the bottom of …