If hopes of colonizing the Moon are ever to become reality, water needs to be found there. Sizeable amounts of ice could provide Moon-dwellers with drinking water and, if they split it into its atomic components, hydrogen for fuel and oxygen to breathe. If lunar water is scarce, colonizers would have to carry water and fuel with them from Earth, or perhaps extract hydrogen from the rocks, either of which could be hugely expensive.

But a paper published in Nature this week (see page 835) strengthens the argument that radar data previously interpreted as showing ice could reveal nothing more than rough ground.

The Shackleton crater at the Moon's south pole is a good place to look for deposits of water ice, because parts of it are in permanent shadow from the Sun and so perpetually frozen. Areas of the crater's rim are in permanent sunlight; together these factors make it a good candidate for a lunar base.

But scientists have argued for a decade about whether there is really ice in the crater. The evidence for Shackleton containing ice deposits is based on the fact that radar signals reflected off the crater are polarized the same way as the incoming pulse (called a high circular polarization ratio, or CPR) — a property of ice. Rough surfaces have the same effect, but the Moon has few rough areas, because apart from young features such as recent impact craters, the surface has been sandblasted to a fine dust.

But astronomer Donald Campbell, of Cornell University in Ithaca, New York, has never bought this argument. Now he has gathered the highest-resolution radar images made of the Moon so far, and claims that they dash hopes of finding large lumps of ice at Shackleton.

Campbell and his colleagues sent radar signals from the Arecibo radio telescope in Puerto Rico to the Moon, and captured the reflections using the Robert C. Byrd Green Bank Telescope in West Virginia. Campbell says that Shackleton shows the same pattern of high CPR as the similarly shaped Schomberger crater — which is dried out by sunlight. Campbell also found similar patterns in the walls and floors of other young craters. So, he argues, the signal from Shackleton is probably caused by rough rocks.

On the rocks: astronomers might have mistaken rough craters for lunar ice deposits. Credit: NASA

Campbell did see some evidence of ice at Shackleton, but as dispersed particles in loose rock, not as lumps. He urges those planning Moon missions to take his findings into account. “Planning for future bases on the Moon should be predicated on there being low concentrations of water ice,” he says.

Predictably for such a long-running debate, not everyone agrees. “Don sees one thing and says 'no ice'. I see the same thing and say 'maybe ice',” says planetary geologist Paul Spudis of the Lunar and Planetary Institute in Houston, Texas. Spudis has believed there is ice on the Moon since the 1994 Clementine mission found high CPR in dark polar regions of the Moon. In 1998, spectrometers on NASA's Lunar Prospector found high levels of hydrogen in the same areas, again suggesting the presence of ice.

Spudis disagrees that the radar signals from Shackleton and Schomberger craters are the same: “Those patterns look different to me.” The CPR signal from Schomberger weakens as the crater deepens, consistent with roughness at its rim, Spudis says, whereas the signal at Shackleton starts halfway down the crater and continues deeper in, as might be expected from ice deposits (see Figure 2, page 837).

Spudis is involved with NASA's Lunar Reconnaissance Orbiter (LRO), set to launch in October 2008, which aims to provide a more detailed map of the Moon. In April, NASA announced that the mission would also include a probe that, by crashing into the Moon, might clear up the ice debate. The probe will split in two, with cameras and a spectrometer on the upper stage recording and analysing the plume of material produced by the lower stage's impact with the Moon. The rest of the probe will crash into the Moon 15 minutes later.

A strong signal of water vapour from the impact could confirm the presence of ice. But an absence of water vapour is unlikely to end the debate, as Spudis argues that the probe could simply hit a dry patch.

The only way to find out for sure will be to land on the Moon and take a sample, says Spudis. Bernard Foing, project scientist on the European Space Agency's Smart-1 lunar orbiter — which ended its three-year mission by crashing into the Moon on 3 September — agrees, and is lobbying hard for the agency to develop a lunar lander that could be launched by 2012.