Most of the moon's water is locked deep below the surface, new analysis suggests

For roughly four billion years, the consensus held that the moon was dry. Then, in 2009, a team at Brown University reported traces of water trapped inside tiny volcanic glass beads returned by the Apollo 15 and Apollo 17 missions, and the orthodoxy began to crack. Sixteen years later, a synthesis of those same samples, plus a wider archive of lunar rocks, points to a more striking conclusion: most of the moon's water is not on the surface at all. It is chemically bound inside the deep interior, locked into mantle minerals by the same heat and pressure that shaped the rest of the rocky body.

The picture matters for two reasons that do not usually appear in the same sentence. It reframes the moon as a wet, differentiated world, with implications for how the Earth-Moon system formed; and it complicates the practical case for lunar water ice, the resource that has done the most to underwrite plans for crewed bases at the south pole.

What changed in the samples

The early Apollo-era analyses returned a verdict that shaped textbooks for decades: the moon was effectively anhydrous. Basalts and highland rocks sampled between 1969 and 1972 carried vanishingly small amounts of hydroxyl, the chemical signature of water bound into mineral structures. That reading was always slightly strange, because some of the same rocks showed other volatile species, and because the giant-impact hypothesis for the moon's formation, popularised in the 1970s and 1980s, predicted that any water present at the impact stage should have been boiled off into space.

The 2009 finding, by Alberto Saal and colleagues working on the Apollo pyroclastic glass, was small in absolute terms but enormous in implication. The beads had erupted from the deep mantle as fire-fountain droplets, quenching so quickly that volatiles were trapped before they could escape. Inside the glass, ion-microprobe measurements found tens to hundreds of parts per million of water. That number was an order of magnitude below Earth's upper mantle, but it was no longer zero.

Since then, reanalysis of older samples has multiplied the data points. Plagioclase and olivine-hosted melt inclusions, apatite grains in mare basalts, and a long list of highland rocks have all been re-examined with modern secondary-ion mass spectrometers and nanoscale techniques. The pattern that emerges is consistent: wherever lunar volcanics have been measured carefully, water shows up, in concentrations that vary with the depth and temperature of the source region.

Why deep, not surface

The new synthesis draws on the geological logic of partial melting. A wet interior does not behave like a wet interior that has been melted through and degassed. The lunar mantle has melted extensively, producing the mare basalt plains that fill the near-side basins, but the water signature in the source regions of those basalts correlates with how deep, and how hot, the melting occurred. The deeper, hotter the source, the wetter the parent melt. That is the opposite of what would be expected if water had been delivered to the surface by impacts after the magma ocean had frozen.

The most economical reading is that the water is primordial. It survived the giant impact, or was reacquired quickly afterwards, and has been sequestered in the deep mantle ever since, only occasionally rising to the surface as dissolved load in mantle-derived melts. Surface water ice, where it exists in permanently shadowed craters near the poles, is a separate inventory, deposited by comets and by interaction of the solar wind with oxygen in the regolith, and is not the main reservoir.

That distinction matters because the two reservoirs have very different implications. A wet interior explains the geological record. It constrains formation models that have, for decades, had to assume a dry moon in order to match the giant-impact scenario. A wet interior also changes how planetary scientists think about the magma ocean, the global melting event thought to have produced the anorthosite crust. If water was present during that stage, the phase diagrams that describe lunar crust formation need to be redrawn.

The resource question

For lunar planners, the headline is less comfortable. Most of the resource-rich literature of the past decade has assumed that water ice in polar craters would be the workhorse supply for any sustained human presence: drinkable water, oxygen for breathing, hydrogen for fuel. If the deeper interior is the dominant water reservoir, that ice is still real, but it is not the bulk inventory. It is a veneer.

This publication notes that the practical implications are not catastrophic. The shadowed craters still hold ice, and the ice is still orders of magnitude easier to access than mantle rock kilometres below the surface. But the strategic framing changes. A moon whose water is mostly kilometres down is a different industrial proposition from a moon whose water sits in convenient cold traps. The economics of in-situ resource utilisation, already a tough sell, get a little tougher.

There is a secondary counterpoint worth raising. Some lunar scientists, particularly those working on the South Pole-Aitken Basin and on more recent remote-sensing datasets, argue that the surface ice inventory may have been systematically underestimated by orbital neutron spectrometers, which struggle to distinguish hydrogen bound in ice from hydrogen bound in regolith minerals. If the surface veneer is thicker than the instruments suggest, the resource picture is rosier than the deep-interior finding implies. The honest position is that the geological and the resource questions are being answered by different instruments, with different systematic errors, and the next few years of sample return will need to reconcile them.

What is still unknown

The synthesis is built almost entirely on Apollo samples, supplemented by a small number of meteorites of lunar origin. None of those samples came from the south pole, from the far side, or from the mantle directly. The Chang'e-5 and Chang'e-6 missions returned younger basalts and the first samples from the far side, and those are still being worked through. NASA's Artemis III crewed landing, repeatedly delayed but still nominally aimed at the south polar region, is meant to return samples from a part of the moon that has never been sampled, and where the water question, in both its geological and resource senses, is most acute.

For now, the dry-moon orthodoxy is dead, the wet-interior picture is gaining ground, and the lunar ice is a real but smaller resource than the more enthusiastic plans have implied. The next decade of sample return will determine whether the mantle reservoir is the main story, or simply the first chapter of a longer one.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• https://en.wikipedia.org/wiki/Giant-impact_hypothesis
• https://en.wikipedia.org/wiki/Lunar_water
• https://en.wikipedia.org/wiki/Apollo_15