When Mars orbiters decades ago spotted valleys and other fluvial landforms on the surface of the Red Planet, a tantalizing idea came to the fore. Perhaps, planetary scientists ventured, ancient Mars was blanketed by a thick atmosphere that kept the planet much warmer and wetter than it is now, with flowing water, lakes and maybe even an ocean covering at least part of its surface. And if Mars had water some three billion to four billion years ago, they wondered, why not life?

But recent mineralogical data have put a twist on that idea, according to a new study. For although Mars indeed seems to have passed through a warm, wet phase, the bulk of the action looks to have occurred in the crust, beneath a surface that remained mostly dry and frigid. In this scenario, Mars's valleys and lakebeds would have been shaped by ephemeral, episodic flows, rather than by long-lived bodies of water. But belowground the environment may have been very different—warm, watery and possibly even hospitable to life.

That is the thrust of a review article published in the November 3 issue of Nature. (Scientific American is part of Nature Publishing Group.) The study's authors examined how clay minerals, which form in the presence of liquid water and have been detected at sites all over Mars, came to exist on the surface. The evidence, the researchers argue, suggests that the clays formed belowground rather than in rocks near the surface exposed to liquid water, as they often do on Earth. The minerals would have been churned up to the surface by meteor impacts in the billions of years since their formation.

"The more information we saw, the more it looked like clays weren't forming in the way that they did on Earth," says lead study author Bethany Ehlmann, a planetary scientist at the California Institute of Technology. "I think it's interesting that the warmer, wetter early Mars may be beneath the surface." In the first billion or so years of Mars's history, ample heat was available in the crust, due to extensive volcanism and bombardment by large impactors.

One line of evidence for subsurface clay formation comes from the chemical composition of those minerals, which closely resembles that of the volcanic rocks that reacted with water to form the clays. That similarity points to an enclosed system without much inflow or outflow of chemical elements due to flowing water or atmospheric exchange. "There hasn't been a lot of water moving in and out of the system, and if you're on the surface, it's hard to imagine how that would happen," Ehlmann notes. "The clays and the basalts are of similar chemical composition."

Another important point is that some of the clay minerals identified on Mars form in high-temperature environments of up to 400 degrees Celsius, indicating pockets of hydrothermal activity rather than open-air bodies of water as the source of the clays.

If the geochemical evidence for water on Mars indeed traces a hydrologic history dominated by subsurface activity, that would mean that the flows that carved valleys and floodplains on the Red Planet are mere outliers in the planet's watery past. "There's sort of a disconnect between the morphological evidence and the mineralogical evidence for water," Ehlmann says. The morphological features may have been carved by brief, episodic floods of melting ice during anomalously warm periods in Mars's unstable climate. "The evidence is consistent with the fact that those were very short in time," Ehlmann says.

Although Martian landforms suggestive of water have been recognized since the 1970s, that morphological evidence does not tell the whole story, says Raymond Arvidson, a planetary scientist at Washington University in Saint Louis who did not contribute to the new study. "What we're finding when you combine that with the mineralogical evidence is: 'Whoa. Take off your Earth-colored glasses,'" he says, "because this is a planet where there has been extensive water activity, but nowhere near as extensive as on Earth."

A cold, dry Martian exterior agrees well with some climate models for the planet, which struggle to reproduce a temperate paleoclimate. "We no longer, based on our modeling, believe in a warm, wet scenario for Mars," Robin Wordsworth, a postdoctoral researcher at the Laboratory of Dynamic Meteorology (LMD) in Paris, said at a planetary science conference in France last month.

If Mars was once blanketed by a thicker atmosphere than it has today, the models indicate that it would indeed have been somewhat warmer. "In spite of that, we found that whatever the thickness of the atmosphere, the climate would have been too cold to sustain a long-term, Earth-like, warm and wet climate," says Wordsworth's LMD colleague Francois Forget. "Overall, our climate simulations point toward a cold early Mars where ice and snow would have played a major role," he says, with localized or episodic melts providing liquid water.

"The real Mars, I think, is beginning to stand up," Arvidson says. "Mars was occasionally warm on the surface and occasionally wet on the surface, but far more in the subsurface." Understanding where and when liquid water existed for long periods on Mars is the key to designing targeted searches for signs of past life on the Red Planet. "It will eventually lead us to the right place to see if Mars had the right stuff to sustain a habitable environment," Arvidson says.