A reanalysis of ten isotope systems points to Earth's inner Solar System origins

Two ETH Zurich scientists reanalysed published meteorite data and found that lines fitted through inner-Solar-System meteorites predict Earth's rock in all ten isotope systems they checked, once the predictions are combined. No new samples were measured. The study leaves room for up to about 2% of Earth to have come from farther out, and the paper makes no claim about where the oceans came from. The water line came from the university's release.

8 October 2026 · How we rate evidence

Nobody measured a single new meteorite for this study. Paolo Sossi and Dan Bower of ETH Zurich took published data on ten kinds of isotope fingerprints in meteorites, Mars and Earth, ran them through what Sossi, in ETH Zurich's release, calls statistical calculations "rarely used in geochemistry," and found that Earth lines up with inner-Solar-System material in every one.

The paper came out in Nature Astronomy on 27 March. On 16 September ScienceDaily reposted ETH Zurich's release as "Scientists 'truly astonished' by discovery that challenges Earth's origin story." It says outer material makes up "less than two percent of Earth's total mass, and possibly none at all," and that "most volatile substances, including water, must already have existed in the inner Solar System." An April MSN headline went further: "Study argues Earth formed from inner solar system material, including its water."

What they did. Meteorites come in two families that formed in different parts of the young Solar System: non-carbonaceous ones from the inner region, and carbonaceous, water-rich ones from beyond Jupiter. For years the argument has been how much of Earth came from each. The figures in play are about 6% outer material (if Earth's main ingredient resembled enstatite chondrites) or about 40% (if it resembled a rarer group called ureilites); both, the paper argues, rest on only some of the isotope systems measured, or on extra assumptions.

Sossi and Bower used ten at once: isotope ratios of calcium, titanium, chromium, iron, nickel, zinc, zirconium, molybdenum (two ratios) and ruthenium, taken from a public compilation. The only Earth values they changed were for molybdenum, recalculated from two recent studies after earlier values were found to carry artifacts from the way labs corrected their measurements.

What they found. Earth's mantle is not a blend of known meteorites. In every isotope system, it sits beyond the end of a line running through ordinary chondrites, Mars and enstatite chondrites: an extreme version of inner-Solar-System material that no meteorite in our collections matches. Predicting each of Earth's ten values from the other nine, using that line, and combining the nine predictions for each, gave answers within the measured values' error bars, zirconium's only just. The carbonaceous meteorite most often invoked, a type called CI, never fell on the line. Earth's own values were left out of every fit, so this was a test Earth could have failed. The data compilation, the code and the reviewers' reports are all public.

Then they asked how much CI material could be stirred into Earth's mantle before its isotopes moved outside their error bars. Calcium, an element that stays in rock, allowed at most about 2% of Earth's whole mass, which is up to about 1.6 Moons' worth. Ruthenium and molybdenum, which mostly sank into the core, allowed less than 0.1 to 0.3% of the mantle. Both numbers are calculated ceilings, not measured amounts.

What it doesn't show. It doesn't show that Earth contains no outer material. The paper's own abstract says "exclusively"; its results put the ceiling at "<2% of the bulk Earth," and the authors wrote to reviewers that "a small amount is allowed."

How tight is the fit? Not as tight as the abstract suggests. The abstract says the line "always" meets Earth "to within 1 standard deviation" for "any two" isotopes. The supplement's table of all 90 pairwise predictions shows 15 falling outside that band, and one pairing of calcium and molybdenum missing by almost four times the band. The combined predictions do match. We rate the pattern Promising, close to Early: the call turns on whether that check against Earth's held-out values outweighs the fact that every prediction sits beyond the range of the meteorites the lines were fitted to, with only five to nine bodies per fit. The supplement also labels its error bars as two standard errors, and the authors confirmed "2 se" to a reviewer, while the main text calls the match "1 standard deviation"; if the bars are two standard errors wide, the test is gentler than it sounds. That doesn't loosen the ceilings on outer material: narrower bars would only lower them. One signed reviewer, Gregory Brennecka, wrote in his report that most values were "BARELY within uncertainty ... so this is not a slam dunk, but cool, no doubt."

Tracers left out could disagree. Silicon, the basis of one of the 40% estimates, was excluded because labs disagree on how to correct it. Sulfur and zinc, elements more easily lost to heat, may tell a different story: a 2025 sulfur study attributes about a quarter of Earth's sulfur to outer material, and earlier zinc studies the paper itself cites put about 30% of the mantle's zinc there.

What about the water? The paper's only link to water is a single calculation: adding CI material equal to about 1% of the mantle's mass would supply all the hydrogen in the mantle. Its ruthenium ceiling would cap that share at around a tenth, by our arithmetic, but its calcium ceiling would allow all of it. The paper notes that the hydrogen isotopes of the oceans and air are altered enough to complicate tracing. Its second reviewer, Mario Fischer-Gödde, pressed the authors on exactly this and cited a 2020 estimate that about 4% of the hydrogen in the ocean and air came from CI-like material. The claim that water "must" have come from the inner Solar System comes from the press release, not the paper.

What would have to be true. For "inner only" to hold, the ten fingerprints must keep lining up when another team compiles the data, and the tracers left out must agree. Nothing in it breaks physics: the finding concerns where Earth's ingredients formed, not how planets work.

Why it's still interesting. Using ten isotope systems at once turns a tug-of-war between single-element estimates into one test that Earth could have failed, and passed on the combined predictions. It also says Earth was built from inner material that no meteorite in our collections represents. And it makes a testable prediction: Venus and Mercury should be even more extreme than Earth. With no rocks from either planet, that remains a proposal, but it tells future sample-return missions what to look for.

Open questions.

Does the result survive someone else's spreadsheet? All the inputs come from one compilation, and the grouping of meteorites was chosen after seeing the data. An independent team compiling its own averages, with new measurements for groups with thin data, would show whether the pattern belongs to the meteorites or to the bookkeeping. If it holds, the case against heavy outer delivery gets much stronger.

Do the elements left out tell the same story? Silicon, sulfur and zinc could each be added once labs agree on their corrections. If they also fit the line, "inner only" becomes hard to escape; if volatile elements need more outer material, Earth may have a rocky body from nearby with a light top-up from farther out.

Where did the oceans come from? Hydrogen and nitrogen isotopes in deep-mantle rocks, compared directly with enstatite chondrites, and better estimates of how much hydrogen went into the core, could show whether inner material alone can supply Earth's water. The answer changes how likely water-rich rocky planets are around other stars.

A note on timing. The paper was published on 27 March 2026, with a first wave of coverage in early April. ScienceDaily reposted the release on 16 September, almost six months later. The science didn't change in between; we found no independent reanalysis published since.