The Only Ear That Ever Worked on Mars

The Viking landers carried seismometers in 1976. The instrument on Viking 1 never unlocked. The one on Viking 2 worked, but it was bolted to the lander's deck rather than set on the ground, and what it mostly recorded was wind. Which leaves the entire seismological record of the planet to a single instrument: SEIS, lowered onto the Martian surface by NASA's InSight lander in December 2018 and buried under enough dust four years later that the solar panels could no longer keep the spacecraft awake. Roughly four years of listening. More than a thousand events.

In September 2025, a team led by Huixing Bi published a paper in Nature arguing that Mars has a solid inner core: an iron-rich sphere about 613 kilometres in radius — give or take 67 — sitting inside the liquid outer core, at roughly eighteen percent of the planet's radius. Earth's inner core sits at about nineteen percent of Earth's. The evidence is two seismic phases teased out of 23 low-frequency marsquakes at epicentral distances of 27 to 40 degrees: PKKP, which passes deep through the core, and PKiKP, which reflects off the inner-core boundary and returns. If PKiKP is really there, there is a boundary for it to have bounced off.

Nobody expected that. "None of us really believed that you would have a solid inner core," Amir Khan of ETH Zürich, an InSight science team member not involved in the paper, told Eos. His objection is chemical rather than seismological. Crystallising an inner core means expelling light elements — sulfur, oxygen, carbon — into the liquid left behind, so a solid Martian centre implies an outer core more loaded with them than the material Mars was assembled from can comfortably supply. And two years before this paper, Khan and others had argued for a layer of molten rock at the base of the Martian mantle: a thermal blanket that would hold the core too hot to freeze at all.

So the situation is a surprising result, a plausible mechanism arguing against it, and a dispute about composition that turns on quantities the paper itself flags as badly constrained. The authors say so directly — large uncertainties in the velocity and especially the density jump across the boundary make the core's composition hard to pin down.

This is an ordinary state of affairs in a young field, and it usually has an ordinary remedy: go and measure it again. Build a better instrument. Put down more than one of them, so you have a network instead of a single point. Find out whether PKiKP is still there when someone else looks.

That option does not exist, and it will not exist soon. There is no working seismometer on Mars, nothing currently flying will put another one down, and no approved mission plans to. The constraint is not physical — landing a working seismometer on Mars is a solved problem, solved once already. It is budgetary. For as long as that holds, every new statement about the interior of that planet is a statement about the same four years of recordings, re-analysed with better methods by people who cannot go and get more.

Which is not the same as the question being unanswerable, and the reason is already on this page. Khan's objection is not argument in the loose sense; it is a chemical and thermal constraint imported from outside the seismic archive entirely. Sulfur budgets can be narrowed by laboratory work on iron alloys at Martian core pressures. The core's size and state are separately constrained by radio tracking of the lander as the planet wobbled — a different instrument on the same spacecraft, measuring a different thing. What Bi and Khan have is not a closed question. It is one dataset that cannot settle it alone, and several other directions from which it might be settled.

Still, something changes in the meantime. We normally treat a body of measurements as a window: the thing under study is on the other side of it, and if the view is poor you go and get a better window. When the instrument is dead and the archive is closed, the measurements stop functioning as a view of the object and start functioning, for practical purposes, as the object. Martian seismology for the next twenty years will not be an argument about the interior of Mars. It will be an argument about those 23 marsquakes.

This is not a failure. InSight did what it was sent to do, returned more about the inside of another planet than anyone had before, and kept producing results long after it stopped transmitting — the inner-core paper landed nearly three years after the last contact. Data outliving its instrument is the ordinary condition of science; that is what an archive is for.

But an archive that cannot be extended is a different epistemic object from one that can, and the difference is easy to miss because the two look identical from the outside. Both are a pile of numbers. Only one of them can be wrong in a way you are able to find out about.

And the closure runs the other way too, which is the part worth sitting with. Nothing about the physics of Mars decided that its only seismic record would be four years long. An appropriation did. The working consensus about the interior of a planet — what a generation of geophysicists will teach as settled — was shaped by a funding cycle, and the winner of the Bi–Khan dispute may be decided less by who is right than by who is better at re-analysis. That is renewable in both directions. Someone could choose otherwise next year.

Mars is still shaking. Nothing is listening.