In 2005, after years of effort, researchers had all but given up on finding bacterial colonization in the hyperarid core of Chile's Atacama Desert. It was simply too dry, they concluded, to support life.

But then, in an unnamed salar (salt flat) near Yungay, Chile, Jacek Wierzchos discovered cyanobacteria living inside rocks made of halite: sodium chloride, common table salt. That discovery could guide the future search for life on Mars.

In the six years since his discovery, Wierzchos and a number of other researchers have been studying these bacteria, trying to understand more about their unique environment. How is it that, in a region where there is almost no external moisture available, the halite can maintain an internal microenvironment capable of supporting life? How do the bacteria colonize the rocks? What factors determine which rocks are colonized and which are not?

Progress is being made at answering all these questions, but on my third day in the Atacama, much of the focus was on yet another question. These strange, knobby, halite structures, structures not found anywhere on Earth but in the Atacama, structures that are the last refuge of life in this parched environment: how did they get here?

To help answer that question, we took a trip to a location a few miles up the road from Yungay, a site that has been dubbed "Bea Hills." Bea Hills would be a good place to film a movie about exploring Mars. (For a panoramic view, click here.) The first thing you notice when you drive by them are dark grey boulders littering the reddish-brown terrain.

No-one is quite sure how the boulders got there. There's no obvious source for them. Nor is anyone sure how long they've been there. Perhaps for millions of years, being slowly reshaped, not by water but by the sandblasting action of wind.

There are no halite knobs in Bea Hills, and no bacteria. The salars where halite knobs form are not found on hillsides but rather in low-lying areas. But what one does sometimes find on hillsides, like in Bea Hills, are polygonal patterns appearing in the surface soil. Polygonal patterns very similar to those found on Mars.

When you scrape away the overlying soil, you find gypsum, and beneath the gypsum, halite. The crystal growth of the halite creates subsurface polygons, and these induce the patterns visible in the soil at the surface.

"When we see a polygon on the surface, the stuff sitting under the polygon is gypsum. And so our first guess in the past was that those were gypsum polygons," says Alfonso Davila, of NASA Ames Research Center in Mountain View, California.

"But every time we've dug out a polygon and we've gone deep enough, we've hit a halite polygon." On this expedition, Davila and Octavio Artieda, a geologist with the Universidad de Extremadura in Plasencia, Spain, focused much of their attention on polygons.

The understanding researchers have started to piece together is that as the salt crystals at the edges of these polygons grow, the edges begin to buckle and the polygons begin to tilt up at an angle, creating teepee-like shapes. This process occurs in places other than the Atacama. But in the hyperarid core of the Chilean desert, the process goes a step further.

"If you don't have any rain or anything that resets the system, these teepees become the focus of evaporation," Davila says. As small amounts of moisture wick upward, carrying molecules of salt with it, the salt, which cannot evaporate with the moisture, gets deposited along the top edges of the teepees.

And it is along the top edges of these teepees that the halite knobs are thought to grow. "The edges of the polygon continue being affected by dissolution and precipitation and they become rounded into pinnacles," he says. He likens this rounding effect to a very slowly melting ice-cream cone, except that in the case of the salt, unlike that of an ice-cream cone, the pinnacles also continue to grow.

The polygon-teepee-pinnacle connection is far from obvious at first glance. When you walk around a salar filled with halite knobs (as I like to call the pinnacles, because "knobs" sounds rounder to me), the knobs appear to grow willy-nilly, at random. But there are some places where you can see the pattern.

If you spend enough time walking around Yungay, and pay careful attention - or if, like me, you walk around with a scientist who's already figured it out and can show you what to look for - you occasionally stumble across places where the polygonal pattern is noticeable at the surface. And if you spend even more time walking around, and paying even more careful attention, you notice places where the polygons are forming teepees, and where knobs are forming along the peaks of the teepees.

That's if the halite is at the surface. "The halite on the surface here gets reshaped into the pinnacles," Davila says. "But if the halite in the polygons is buried it doesn't develop these kinds of shapes." If one scraped off a subsurface halite polygon "and we [left] it exposed, though time my guess is that it would develop the pinnacle shapes on the sides. But not if it's buried."

The pinnacle-formation process takes place far too slowly for anyone to watch the progression occur. The steps of the process have been pieced together from years of observation in the field and a lot of head scratching. And recently, a little help from a nearby mining company.

Next to the site of the former research station in Yungay there is a water pump that delivers water to a mine a few miles away. Last year the pump burst and flooded a small section of the Yungay salar, the size of a couple of football fields, with water.

The spill effectively "reset" the halite, dissolving all the knobs in the flooded region and washing them clean of sand. When the water evaporated, which in the Atacama didn't take long, it left behind a flat slab of pure white salt.

Over the course of the past year, as the salt grew - salt is a type of crystal, remember - polygons began to form. And then the polygons began to buckle and tilt. At present, teepees are visible throughout the flooded area. No knob-like structures are visible yet, but researchers suspect that when they return in a year or two, small knobs may have begun to grow along the tops of the teepees.

"I think it's interesting in terms of astrobiology," Davila says, because "we can see how a habitat is created from something relatively simple and non-habitable."

A similar process could occur on Mars, he says. "It's probably too dry today to drive any ecosystem in the salt." But, "ten million years ago, [Mars's] obliquity was higher, so the polar caps were closer to the equator," and "at least in polar regions, you had 100 to 1000 times more water in the atmosphere."

Davila thinks halite deposits would be excellent targets for future life-detection missions to Mars. They probably exist. Unfortunately, they're difficult to detect. Chloride has been found on Mars, but sodium chloride, or halite, cannot be specifically detected by any of the instruments flown to date on orbiting spacecraft.

In part this is a technical issue. But before the discovery of colonized halite in 2005, halite on Mars wasn't a target of particular interest. And since then no new orbiting spacecraft have been sent to explore the planet.

And there's another small problem. There's a lot of dust and sand on Mars. Even if an instrument capable of detecting halite from orbit were sent there, any halite present on Mars is probably buried. Unless the halite had formed recently, Davila says - and that's an unlikely prospect - all "you would detect" would be "a layer of dirt."