The familiar version goes like this. Early Mars was a warmer world with rivers and lakes and a thick atmosphere. Its internal dynamo died, the planet lost the magnetic field that had shielded it, and the solar wind stripped the air away over billions of years, leaving the cold desert we see now.

The outline is broadly right. The water was real, the dynamo did fail, and the solar wind does strip the atmosphere. But several links in that chain are less settled than the clean telling suggests, and a recent finding shows that not all the missing air went to space.

What the rocks actually show

The strongest part of the story is the water. Dried river valleys, lake beds, deltas and water-formed minerals across the older Martian terrain make it clear that liquid water flowed on the surface, at least intermittently, before roughly 3.5 billion years ago.

That much is not in serious dispute.

Whether the planet was warm is another matter. Keeping early Mars warm is difficult in climate models, because the young Sun was fainter than it is today. One camp argues for a persistently warm and relatively wet climate held up by a dense carbon dioxide atmosphere and extra greenhouse gases. Another argues for a mostly cold and icy planet, frozen for long stretches and warmed only in episodes. The geological and climatological case for a warmer, wetter early Mars has been made in detail, and the cold-and-icy alternative has been modelled just as seriously. The argument has run for decades and is not resolved. “Once warmer” is a reasonable shorthand, but it papers over a real disagreement.

When the dynamo died

Mars no longer has a global magnetic field, but it clearly had one. The crust carries magnetic imprints locked in when ancient rocks cooled, which means a core dynamo was running early in the planet’s history and then stopped. The usual estimate places the shutdown somewhere between about 4.2 and 3.7 billion years ago, around the same era the atmosphere appears to have thinned.

The timing is not pinned down. Palaeomagnetic work led by Sarah Steele at Harvard, published in Science Advances in 2023, argues the dynamo may have lasted longer and behaved in a more complex way than earlier reconstructions assumed.

The death of the field is real. Exactly when, and how cleanly, is still being worked out.

What MAVEN measured

The clearest evidence comes from NASA’s MAVEN mission, which has orbited Mars since September 2014. MAVEN was built to measure how fast gas escapes the upper atmosphere and what drives it. It found that the solar wind does strip ions from Mars into space, and that the rate jumps sharply during solar storms. A strong storm in March 2015 served as a natural experiment, with escape rates spiking while the spacecraft watched.

Measurements of argon isotopes point the same way. Argon is useful because it is not recycled through rocks and volcanoes the way carbon dioxide is, so its isotope ratio records loss to space fairly directly. On the MAVEN team’s reading, the argon ratio implies that most of Mars’ atmospheric loss happened upward, to space. Carbon dioxide is more complicated, because some of it could also be locked into rocks.

Whether the magnetic field was really the shield

This is the part most often stated too confidently. It is tempting to treat the sequence as simple cause and effect: field on, atmosphere safe; field off, atmosphere gone. The physics is not that tidy.

A planetary magnetic field deflects some of the solar wind, but it can also open channels along which charged particles escape, and in some configurations a field may speed up loss rather than prevent it. Venus is the standing complication. It has no internally generated magnetic field and sits closer to the Sun, yet it holds a dense atmosphere. The match in timing between Mars losing its dynamo and losing its air is suggestive, and the protective reading is the leading one, but the causal claim is not established. Some researchers have argued that a longer-lived dynamo might even have helped Mars lose water rather than save it.

The carbon that did not go to space

There is also the question of where the carbon dioxide went. If early Mars had a thick carbon dioxide atmosphere, large amounts of carbonate rock should have formed as the gas reacted with water and stone, and for years orbiters did not find nearly enough of it.

In 2025, a team led by Benjamin Tutolo reported in Science that the Curiosity rover had found abundant siderite, an iron carbonate, in the sulphate-rich layers of Gale crater. Scaled across similar deposits, the team estimated the rocks could hold the equivalent of a few to several tens of millibar of atmospheric carbon dioxide. That is a genuine reservoir, and part of the missing inventory, though it falls well short of the roughly one bar thought necessary to warm the surface. Some of the buried carbon also appears to have been released again, which would make it a partial, imbalanced carbon cycle.

So the air left by more than one route. Some was stripped to space, as MAVEN shows is still happening. Some was locked into the crust as rock.

What remains open is the balance between those routes, the true early climate, and how much the dying magnetic field actually mattered. Better palaeomagnetic dating of the dynamo, and missions that sample more of the sulphate layers, are the things likely to narrow it. The cold desert is the settled part. The path it took to get there is still being filled in.