Where the Moon Came From

Any theory of where the Moon came from has to explain an odd object. It carries about 1.2 percent of Earth's mass, a far larger share than any planet's satellites in John Wood's 1986 table, with the probable exception of Pluto's Charon. Earth is roughly a third iron by mass; the Moon, by the figures in a NASA Goddard review, is about 8 to 10 percent iron, with a small core some 325 kilometres in radius from GRAIL gravity work. The Moon is short of volatile elements such as potassium, sodium and zinc. Its oldest crust says it was once wrapped in a magma ocean hundreds of kilometres deep. The Earth-Moon pair also holds more angular momentum than the other planets would lead you to expect. Any account of the Moon's birth has to deliver all of that at once, plus the property that turned out to be hardest of all: in isotopes, the Moon's rocks are almost indistinguishable from Earth's mantle.

For most of the twentieth century there were three main candidate stories. George Darwin's 1879 treatise had the Moon flung off a rapidly rotating Earth. Others imagined a Moon formed elsewhere and captured as it passed. A third, co-accretion, had Earth and Moon grow side by side out of the same swarm of debris; Russian accounts credit its development to the Soviet scientists Otto Schmidt, Evgenia Ruskol and Viktor Safronov, and Wood later described it as the model most lunar scientists had accepted, at least at a subconscious level. There was a widespread expectation that Apollo would settle the matter. It did not. Writing after the samples had been studied, Wood said the Moon had preserved "a priceless record of the earliest igneous activity" in a small planet, "but not of how the body formed"; the one clear verdict on origin was that Harold Urey's cold, primitive Moon had been wrong.

The idea that replaced them arrived from two directions in the mid-1970s. In April 1975 William Hartmann and Donald Davis argued in Icarus that the planetesimals building Earth included much larger objects than Safronov allowed, so a late hit by something around a tenth of Earth's mass was to be expected, and that such a collision could throw out more than enough debris for a Moon, most of it iron-poor mantle. The next year A. G. W. Cameron and W. R. Ward worked from the angular momentum instead. The early Earth's spin, they wrote, was presumably imparted by a collision with a major secondary body; if it struck a glancing blow at about 11 kilometres per second, its mass was comparable to that of Mars. At that speed most silicate would vaporise, gas pressure would push debris into orbit, and the resulting Moon should be short of metallic iron and volatile elements. Neither paper proved anything. When lunar scientists met at Kona, Hawaii, on 14 to 16 October 1984, confidence swung toward the collision model, and Wood's summary is candid about why: co-accretion had been shown unable to account for the angular momentum, and the impact was the most plausible idea left standing. It began its career as the survivor.

By 2001 the collision had a standard form. Robin Canup and Erik Asphaug reported in Nature a class of impacts that yields an iron-poor Moon along with the present masses and angular momentum of the Earth-Moon system, using a smaller and therefore more likely impactor than earlier work had needed, late in Earth's growth. The same year a Science paper on lunar oxygen used the name that stuck for the impactor: Theia, after the Titan who in Greek myth was the mother of Selene, goddess of the Moon. The catch was already visible. In the standard simulations, most of the orbiting debris is impactor material. Wood had flagged the worry in 1986: if the impactor formed far from Earth's orbit, he wrote, we should worry about the close match in oxygen isotopes between Earth and Moon. Junjun Zhang and colleagues put a number on the tension in 2012, citing models in which more than 40 percent of the Moon-forming disk came from Theia.

Why that matters comes down to fingerprints. Bodies that formed in different parts of the young solar system carry different mixtures of isotopes, which is why meteorites from different parent bodies can be told apart, and why Mars, as the 2026 SwRI team notes, is compositionally distinct from Earth and Moon. The Moon keeps failing to look different. In 2001 Wiechert and colleagues measured 31 Apollo samples from five landing sites and found them all within ±0.016 per mil of a single line identical, within uncertainties, to Earth's. In 2016 Young's group put the Earth-Moon difference at -1 ± 5 parts per million. In December 2024 a Göttingen team measured 14 Apollo rocks on a newly built instrument and reported 0.2 ± 1.6 ppm, an isotopic match at the sub-ppm level. Titanium, a highly refractory element, tells the same story: the Moon's 50Ti/47Ti ratio matches Earth's within about 4 ppm, 1/150 of the spread seen in meteorites. In March 2023 geochemists at the Vernadsky Institute of the Russian Academy of Sciences listed nine elements, from oxygen and iron to tungsten and chromium, whose isotopes are close in Earth and Moon. For a Moon built mostly from another planet, that is a remarkable coincidence.

Even that result is argued over at the edges. The same Göttingen laboratory, in a 2014 Science paper by Herwartz and colleagues, had reported a 12 ± 3 ppm difference and presented it as Theia's signature finally found in lunar rock. Its 2024 data do not reproduce that, and the authors write that the exact cause of the discrepancy could not be resolved. In 2020 Erick Cano, Zachary Sharp and Charles Shearer reported that Earth and Moon in fact have distinct oxygen compositions, that lunar values vary with rock type, and that samples from the deep lunar mantle, heavier than Earth, may best represent Theia. So the field does not yet agree whether the difference is zero or merely tiny.

There are three logical ways out, as the NASA Goddard review sets them out: the Moon is mostly Earth mantle after all; Earth and impactor material mixed until their fingerprints matched; or Theia was made of the same stuff as Earth. Each has champions. In 2012 two papers in Science pushed the collision harder. Ćuk and Stewart showed that a typical late impact onto a fast-spinning proto-Earth could make a disk derived mainly from Earth's mantle, and Canup showed that much larger impactors, comparable in mass to the proto-Earth they struck, could give the disk the composition of the planet's mantle. Both leave the system spinning too fast and need the excess drained later through an orbital resonance with the Sun. In 2018 Simon Lock, Sarah Stewart and colleagues went further: a high-energy impact can leave a structure too hot and fast-spinning to be a planet in the usual sense, a doughnut of rock vapour they named a synestia, inside which the Moon condenses while bathed in vapour of Earth's composition. Potassium offered this family some support. Wang and Jacobsen found lunar rocks about 0.4 per mil heavier in potassium isotopes than Earth and chondrites, a pattern they read as condensation from vapour at more than 10 bar, inconsistent with the low-energy disk equilibration model and supportive of a high-energy impact.

Other groups changed the impact itself. Raluca Rufu, Oded Aharonson and Hagai Perets proposed in 2017 that no single blow was needed: a succession of smaller collisions could each leave a debris disk and a moonlet, and if the moonlets drifted outward and merged efficiently they could build the Moon. Natsuki Hosono of JAMSTEC and colleagues struck a proto-Earth still covered by a magma ocean. Because molten rock heats far more under shock than solid rock, their simulations drew much of the Moon-forming debris from the magma ocean even in a glancing collision; the Japanese release from JAMSTEC, Kobe University and RIKEN described this as a possible resolution of the contradiction between the Apollo isotope measurements and conventional giant-impact simulations. In 2022 Jacob Kegerreis and colleagues at NASA Ames and Durham University found that at high enough resolution a giant impact can place a Moon-like satellite directly into orbit, in NASA's words "in a matter of hours", its outer layers around 60 percent proto-Earth material. And on 1 September 2026 Adeene Denton of the Southwest Research Institute and colleagues added something earlier models had left out, the strength of rock. With otherwise identical parameters, a hot but solid Theia produced an intact Moon while a colder, stronger Theia made a classic debris disk, which ties the style of the Moon's birth to the timing of the impact.

Where Theia came from is its own open question. In November 2025 Timo Hopp, Nicolas Dauphas and colleagues measured iron isotopes in six Apollo samples, 15 terrestrial rocks and a range of meteorites and, combining them with other elements in a mass balance, concluded that all of Theia came from the inner solar system, possibly closer to the Sun than Earth. That would ease the isotope problem, since neighbours share fingerprints. Their institute's release keeps a caveat worth repeating: the Moon may be mostly Theia, mostly early Earth mantle, or an inseparable mix of both. Around the same time Duarte Branco, Pedro Machado and Sean Raymond published dynamical simulations in Icarus that, in a scenario built to match cosmochemical estimates that Earth took in some carbonaceous material, gave roughly even odds that Theia was a carbonaceous object, either an embryo from the outer solar system or an inner one that had absorbed such an embryo, and Cano's 2020 oxygen work had pointed to a Theia formed farther from the Sun than Earth. There is even a proposal that pieces of Theia are still here: Yuan and colleagues argued in Nature in 2023 that two continent-sized, seismically slow regions at the base of Earth's mantle may be buried relics of Theia's mantle.

Not everyone accepts a giant impact at all. Erik Galimov, director of the Vernadsky Institute of Geochemistry and Analytical Chemistry (GEOKHI) of the Russian Academy of Sciences, told the Presidium of the Russian Academy of Sciences in December 2003 that his group held a different hypothesis, with no catastrophic impact, in which Earth and Moon formed from primary material as a double system. In his model the pair condensed from a single rarefied cloud of gas and dust, and evaporation stripped iron from the hot early embryos; he argued from rubidium and strontium that the Moon's material spent its first 50 million years dispersed. GEOKHI colleagues have kept the question alive. A 2023 institute review, summarised in Kommersant with comments from Mikhail Marov and Sergei Ipatov, weighs co-accretion, multiple impacts, the megaimpact and Galimov's condensation model, and lists weaknesses in each, Galimov's included: other researchers' work has not supported a condensation lasting tens of millions of years, and his model does not say where 95 percent of its iron-poor inner material went. Ipatov's own 2018 variant, which the review judged the best supported today, has the Earth and Moon embryos form together from a condensation spun up when two clouds collided. These ideas sit outside the collision family that NASA says almost all current theories belong to, but they press on the same fact: the isotope match is what every model has to meet.

The samples are where the argument is tested. The Apollo crews brought back 382 kilograms, and NASA's summary builds the core of the modern picture on them and on later orbital mapping: lunar basalts strikingly similar to Earth's, oxygen that matches, white anorthosite crust that floated on a magma ocean, and a Moon poor in volatile elements. The Soviet robots Luna 16, Luna 20 and Luna 24 added about 0.3 kilograms (101, 30 and 170.1 grams by NSSDCA's figures) from three eastern sites no crew visited, including highland samples from the Apollonius region and drill samples from Mare Crisium. Their part in the origin debate has been quieter: the 2001 and 2024 oxygen datasets and the 2025 iron study described here were all measured on Apollo material. China's samples have started to test whether the Apollo picture holds for the whole Moon. Chang'e-5 basalts, dated to 2,030 ± 4 million years, extended known lunar volcanism by some 800 to 900 million years, which bears more on how the Moon cooled than on how it formed. Chang'e-6 returned to Earth on 25 June 2024 with 1,935.3 grams from the South Pole-Aitken basin on the far side. Li Yiheng, Wang Zaicong and colleagues found that its basalts carry the same heavy sulfur signature as the near side, about 2 per mil heavier than Earth's mantle, and attribute it to global volatile loss in the Moon-forming impact. Then in January 2026 Tian Hengci and colleagues reported that Chang'e-6 basalts are heavier in potassium isotopes than Apollo and Chang'e-5 basalts, most likely because of evaporation during the South Pole-Aitken impact itself: a reminder that later giant impacts left isotopic marks too.

Even the date is disputed. NASA's summary, drawing on Apollo ages, puts the Moon's formation around 60 million years after the solar system began. Published ages range from 4.35 to 4.51 billion years depending on whether whole rocks or individual zircon grains are dated. In December 2024 Francis Nimmo, Thorsten Kleine and Alessandro Morbidelli argued in Nature that the many 4.35 billion year ages record a later tidal remelting as the Moon's orbit evolved, which would allow the Moon itself to have formed within a few tens of millions of years of the solar system's start. The Denton result makes timing matter more, since in those simulations the temperature of the colliding bodies, and so the date of the impact, decides whether the Moon forms intact or from a disk.

So the honest answer, as of 23 September 2026, comes in two parts. A giant collision is the family of explanations nearly all current work sits within, and NASA reads the magma ocean and the lost volatiles as signs of an intensely energetic start. Which collision, how large, how fast, how hot, one blow or many, a disk or an intact Moon, and whether the Moon is mostly Theia or mostly Earth: all of that is open, and papers from 2025 and 2026 pull in different directions. The Göttingen oxygen team drew one practical lesson worth keeping: returned samples are worth more for this work than meteorites, which may have interacted with terrestrial water. John Wood's sentence from 1986 still holds for everything beyond the broad outline: how the Earth's Moon was formed is still not known.