Mercury has been shrinking since the day it formed, and the number attached to that shrinking has just moved again: from about 8.3 kilometers of lost radius to about 11.6. The revision did not come from finding more wrinkles on Mercury. It came from working out how many the planet has hidden. If the team behind the work is right, the Moon has hidden more of its own.

The paper, "Underestimation of Planetary Contraction Due To Obscuration by Surface Roughness: The Case of Mercury," ran in Geophysical Research Letters (volume 53, issue 17) on September 10, 2026. Its lead author, Gaku Nishiyama, holds a postdoctoral post at Hokkaido University along with positions at the German Aerospace Center and the University of Tokyo.

Fifty years of counting wrinkles

Rocky planets lose volume as they cool. Mercury lost a lot of it, because it is small, because it has an outsized metal core, and because that core has been solidifying for billions of years. When the interior contracts the crust no longer fits, so it buckles, and the buckling shows up as thrust faults: long curved cliffs called lobate scarps, and lower, broader wrinkle ridges.

Earth erases this kind of record constantly, because plate tectonics recycles the surface. Mercury has one plate and no recycling, so the working assumption has been that every one of those wrinkles is still sitting there. That is what makes it so attractive to geologists. Map every shortening structure, add up how much crust each one swallowed, and you can work backward to how much the radius fell. From that you get constraints on core size, interior temperature, and how much silicon and sulfur got mixed into the iron.

For a planet with a mean radius of 2,439.7 kilometers, 11.6 kilometers is under half a percent of the whole.

The number that refused to sit still

Before MESSENGER, geologists worked from Mariner 10, whose three flybys in 1974 and 1975 brought back pictures of, by NASA's own accounting, not quite half the surface. Photogeological estimates from that era put Mercury's radius loss between 0.8 and 3 kilometers, which never sat well with thermal models. Those wanted more.

In 2014, Paul K. Byrne and colleagues published a global survey in Nature Geoscience using MESSENGER's imaging and reported the opposite problem. Once you counted the belts of ridges and scarps that earlier work had missed, Mercury had contracted by as much as 7 kilometers. The paper's title said it plainly: much greater than earlier estimates.

Seven years later, Thomas R. Watters went back over the same spacecraft data and arrived at 1 to 2 kilometers. His argument in Communications Earth & Environment was about bookkeeping. Earlier surveys, he wrote, had lumped in wrinkle ridges and other relief with no clear evidence of shortening, and had assigned several primary faults to structures that deserved one. Strip that out, and Mercury barely moved.

So the same catalogue, read by careful people, produced answers that differed fivefold. Every one of those arguments was an argument about the contents of the catalogue. Whether the catalogue itself was complete was not the question being asked.

Reading the missing wrinkles out of the rubble

Nishiyama's group started from a mismatch. Global cooling should squeeze a planet much the same way all over, but the mapped shortening structures are not spread evenly at all. That is where the paper opens.

In March 2026 the same group published in The Planetary Science Journal a roughness map covering the whole of Mercury at horizontal wavelengths from 5 to 100 kilometers, which its title presents as the first at kilometric baselines. Roughness is what it sounds like: how bumpy a patch of ground is at a given scale. The new paper treats it as a stand-in for how recently that ground was worked over, on the logic that whatever roughens a planet is also whatever happened to it last.

Building that map took stereo photogrammetry rather than laser ranging. MESSENGER carried a laser altimeter, MLA, good to 30 centimeters, but the spacecraft flew a strongly eccentric orbit with its close approach in the north, so MLA mapped the northern hemisphere and little else. Global topography had to be reconstructed from overlapping images shot at different angles, a technique Nishiyama described to TIME as working the way human eyes build depth from two views.

With the map in hand, the team laid it over the existing catalogues of shortening structures. The correlation was negative and clean. Smooth ground is full of wrinkles. Rough ground is not. The paper's abstract is careful about why: "We find a lack of shortening structures in rough regions, suggesting that roughness-related processes obscure pre-existing structures, hindering their identification, and/or inhibiting their formation."

The claim is not simply that scarps got buried. Rough, recently resurfaced ground may also never have recorded them in the first place. Either way the tally comes out short, and the suspected mechanism is unglamorous: impacts. Ejecta from large craters, including relatively new ones more than 150 kilometers across, blankets and churns whatever was underneath.

"Mercury's surface preserves a record of how the planet has cooled and contracted, but we found that this record is incomplete," Nishiyama said in Hokkaido University's announcement.

Correcting for it means taking the density of shortening structures in smooth terrain and extrapolating it onto the rough terrain where they are missing. The abstract puts the resulting bias at several kilometers, leaving Mercury's radial contraction up to 30 percent larger than previously thought. The university's English release gives the pair of numbers: 8.3 kilometers before the correction, 11.6 after.

That difference registers directly in the planet's interior. "More shrinking means Mercury could have a larger metal core, less light elements like silicon mixed into the metal core, or a higher starting temperature," Nishiyama told TIME.

November 21, and no second attempt

BepiColombo, the joint ESA and JAXA mission launched on October 20, 2018, is arriving at Mercury now. Its transfer module separated on September 3. Orbit insertion is set for November 21. ESA's Mercury Planetary Orbiter and JAXA's Mio part company on December 9 and 10, and the science phase opens in April 2027.

It is nearly a year late. Tech Times reports that a power conditioning fault found in the transfer module in April 2024 forced the fourth Mercury flyby to be flown 35 kilometers lower, stealing extra braking from the planet's gravity, which pushed insertion past its original December 2025 date. On the day itself, signals will take about 11 minutes each way, so the burn has to run without anyone on Earth able to correct it as it happens.

Hokkaido University's release names the instrument it expects to sharpen the number: BELA, the laser altimeter riding on the European orbiter. The University of Bern team that built it lists its objectives as figure and topography plus surface characteristics including roughness, local slopes, and albedo, even inside permanently shadowed polar craters. The quantity Nishiyama's correction is built on stops being inferred from stereo images and becomes measured directly. MPO's near-polar orbit also covers the southern hemisphere that MLA could never reach.

Mio is Japan's contribution, and it studies the magnetosphere and plasma environment rather than the ground, so the hardware that will check this work is European. Japan's stake is the mission existing at all, and a lead author whose institutional line runs Sapporo to Tokyo to Berlin. The money splits the same way, between JSPS grants on the Japanese side and the Alexander von Humboldt Foundation and a DFG Emmy Noether grant on the German.

The Moon is where this bites next

The abstract's final sentence points away from Mercury: the same bias, the authors write, could matter just as much for other rocky worlds, the Moon among them.

Nishiyama has also talked about what that would mean for the Moon. Measured lunar contraction, he told Gizmodo, comes to less than a kilometer, which does not sit well with thermal models. Hokkaido University's release adds that on average the Moon is more than 50 percent rougher than Mercury. If rough ground hides wrinkles, the Moon should be hiding a larger share of its own.

The data to check that already exists, and part of it is Japanese. SLDEM2015, the standard lunar terrain model, merges NASA's LOLA altimetry with the Terrain Camera flown on JAXA's Kaguya. Kaguya's own laser altimeter, LALT, laid down more than 30 million ranging points and produced a global topographic map that, unlike Clementine's earlier survey, included both polar regions. A roughness correction of the kind Nishiyama ran on Mercury has everything it needs waiting in archives that have been public for years.

Most of the world met this story as a headline about a planet getting smaller. In Japan it ran as a Hokkaido University result, with the researcher's name attached. Both leave out what will still matter in ten years: a widely used way of measuring a planet had a hole in it, and somebody found the hole by looking at how bumpy the ground was.

Does any of this reach you where you live, and is the November 21 burn something your national broadcaster will even mention?

References