Current models of Mars predict that it has a thick layer of permafrost - the so-called "cryosphere" - starting at most a few hundred meters below its surface; but that after you get below about 2.5 km deep at the equator or 6 km deep under the poles, Mars' remaining trapped geothermal warmth is high enough to melt it into liquid water trapped in the rock pores, which could conceivably serve as a last redoubt for any surviving Martian microbes.

The big question is whether there are still places where Mars retains enough volcanic activity that deposits of liquid water may be located much closer to the surface, making it much easier for space probes to drill down deep enough to reach them.

The Naiades landers, each about the size of Beagle 2, would be simultaneously released from their bus just before it flew by Mars, and would land in a square pattern only about 16 km from each other in the Dao Valley - an intriguing-looking gash on the flanks of the old shield volcano Hadriaca Patera, which looks very much as if it was produced when the volcano's heat during its early days gradually melted a reservoir of underground ice that then trickled away downslope, finally causing the ground above to cave in.

Hadriaca has been extinct since Mars' early days, but it's one of the most promising places on Mars to have retained enough underground heat that there may still be pockets of liquid water relatively close to the surface.

Europe's Mars Express orbiter will use a low-frequency radar sounder to survey the entire planet for underground water in 2003; the 2005 U.S. orbiter will probably carry another radar sounder to remap the planet at shallower depths but with more accuracy in its depth measurements, and the four little "Netlander" hard landers that France plans to scatter over the planet in 2007 each carry their own small radar to sound the local landing site.

But such radar sounders have limitations. It's hoped that they can penetrate as much as 5 km below the surface, but - since Mars' surface seems to be rich in iron minerals - they may be limited to only about 1.5 km.

And in any case it's quite hard to interpret such radar graphs; different rock layers will reflect back the radar waves in a way that may look very much like layers of either ice or liquid water. The feeling is that these soundings will take a lot of analysis before scientists can estimate how likely it is that some of them may truly have found subsurface water layers.

So the Naiades landers would use a different technique: "magnetotellurics". If far lower-frequency radio waves are beamed into the ground - from a few thousand cycles down to only one cycle per second, as opposed to several million cycles for the radar sounders - they are not reflected back from underground layers of different substances, but instead have their energy absorbed by any layers of electrically conductive stuff they encounter.

And when this happens - since such any single electromagnetic wave consists of a side-to-side oscillation of the local magnetic field - the energy of the wave produces an oscillating electric current in the conductive layer that flows at right angles to the magnetic fluctuation.

Such very low-frequency radio waves are routinely produced in large amounts by natural sources, such as ionospheric fluctuations, that are bound to occur on Mars - and it may even conceivably have radio bursts produced by occasional lightning discharges in its great dust storms.

Each Naiades lander will carry a magnetometer to detect the side-to-side magnetic fluctuations marking such waves - and, after landing, it will also unreel four antennas several meters long from its sides which can measure faint electric currents flowing on the surface of the ground, since a tiny fraction of the oscillating current flow from those buried conductive layers will diffuse upwards all the way to the surface.

Thus it will measure both magnetic oscillations at different frequencies, and look for simultaneous ground electric currents oscillating at the same frequency but at right angles to the magnetic field - a dead giveaway for subsurface conductive layers.

Many ground minerals do conduct electric currents, but liquid water is far better at it - especially if it's briny, as may well be the case on Mars.

And by comparing the relative strengths of the magnetic and electric oscillations for waves of different frequencies, we can get a graph which can be analyzed in various complex ways to get a good estimate of both the depth at which such a conductive water table is buried, and also its thickness. Ground ice can also be detected this way, though with far less sensitivity.

This technique can't be used from orbit, but it is far more sensitive to buried water than radar sounding and can thus locate it at much greater depths - down to tens of kilometers, in fact.

It can also locate water layers with far less ambiguity than radar-sounding graphs, and can measure their thickness (while radar sounding will be able to detect only the top surface of any water layer).

Moreover, the Naiades landers will also sometimes engage in active sounding - transmitting brief pulses of such very low-frequency radio downwards themselves, and precisely timing the tiny period before buried conductive layers absorb their energy and retransmit an electric fluctuation back up to the surface.

This variation on ground radar sounding uses transmitters with far less power than the natural radio sources on Mars, and so would be limited to probing only about a kilometer down - but it's also far more accurate in its depth measurements of any pockets of liquid water that close to the surface, and by listening to each other's transmissions the four landers can also build up a better horizontal map of any such aquifers.

And it's just such possible near-surface water layers - perhaps maintained both by local geothermal heat and by the fact that dissolved salts can greatly lower the melting point of water - that may be responsible for the surprising gullies, apparently carved recently from some fluid oozing from rock strata only 100 meters or so down, that have turned up in MGS photos of a few Martian regions (including slopes in the Dao Valley).

The Naiades landers would probably operate only a few weeks to get adequate sounding data of their landing area, and the only other instruments they'd carry would be rather low-quality surface cameras - but if they could confirm that pockets of liquid water still exist moderately near the surface in this region, it would immediately become one of the most promising sites for future landers engaged in the search for ancient fossils.