Our planet may once have had dozens of small moonlets, which welded together over the millennia into the object that’s visible today. PHOTOGRAPH BY THE PRINT COLLECTOR / GETTY

Unbeknownst to most earthlings, the moon is experiencing a crisis. Geophysicists will tell you that it’s a “compositional” crisis—a crisis regarding the stuff of which the moon is composed. But it’s also an identity crisis, as much for the scientists as for the object they study.

Our moon formed about 4.5 billion years ago, between twenty million and a hundred million years after Earth took shape. How exactly it got there is a matter of debate. Did it accrete in tandem with Earth, from the same proto-planetary stuff, or did it spin off afterward? Is it our fraternal twin or an identical one? Or was it adopted, drawn into our gravitational sway as it passed by? Since the nineteen-eighties, the consensus has centered on the single-impact hypothesis, sometimes known as the Big Splash or the Big Splat, which supposes that the moon formed when a planet-size object, often called Theia, crashed into Earth and sent a huge mass of debris into orbit. But today, in a paper in Nature Geoscience, a team of Israeli researchers is proposing an equally compelling origin story: the moon, they submit, is the product not of one impact but of at least a dozen—and it isn’t just one moon but an amalgam of the many moons that came before it.

The single-impact scenario has prevailed for so long because it gracefully accounts for several critical facts. For one, the moon is severely lacking in iron compared with Earth, and its core is much smaller relative to its mass than Earth’s is. Computer simulations indicate that such a moon could arise if an object the size of Mars (which has one-tenth Earth’s mass) struck our planet at an oblique angle; the resulting moon would be composed mostly of the impactor, including its small core, plus a scraping of Earth’s iron-poor mantle. An off-center impact would also have set our planet spinning at a rate of once every four or five hours, which most calculations indicate was the case with the early Earth. In the past three decades, as simulations have grown more powerful, scientists have been able to test out all kinds of factors—impactors of various sizes, approaching at various speeds and angles, hitting an Earth spinning at various rates. They all show that it’s not so hard to make a moon, if not necessarily our moon.

Which brings us back around to the current crisis. Studies of the lunar soil and rocks that came back with the Apollo missions, between 1969 and 1972, have shown that the moon’s composition doesn’t quite square with the Big Splat. Elements come in a variety of flavors, known as isotopes; for instance, there’s regular old oxygen, or 16O, which has eight neutrons and eight protons, but also the rare 17O, which has an extra neutron. The moon is deficient in certain Earth elements, as the single-impact hypothesis predicts, but the elements it does have—not only oxygen but also titanium, tungsten, and aluminum, among others—possess the same isotopic signatures as on our planet. In other words, a spoonful of dust from the Sea of Tranquility might have an identical ratio of 16O and 17O to one from, say, the Sahara. Unless whatever struck Earth back then was made of precisely the same stuff as Earth, which isn’t so likely, it’s hard to imagine their collision producing two objects as homogeneously similar as Earth and the moon. “That’s at the core of the issue we’ve all been grappling with,” Robin Canup, an astrophysicist at the Southwest Research Institute and a longtime proponent of the single-impact hypothesis, told me over the weekend.

The new study, led by Raluca Rufu, a Ph.D. student in planetary sciences at the Weizmann Institute of Science, and her adviser, Oded Aharonson, effectively started over. The impetus came from Hagai Perets, a planetary physicist at the Technion-Israel Institute of Technology and a coauthor of the study, who noted that, at the time the moon formed, collisions were commonplace. Given all that activity, it seemed unlikely to Perets that a very large impact—large enough to set Earth and the moon on their current courses—would occur but not be followed by another. Also, all of those impacts must have produced a lot of moons over time. “It’s kind of common to form a moonlet,” Rufu told me. “So where are they?” Maybe, the team thought, all those moonlets are still there, welded into the body of the single object that’s visible today; from many collisions, one moon. To test that idea, Rufu ran more than eight hundred and fifty simulations, involving a wide range of impactors, to see what kinds of moons would result. “I am impacting the Earth, so I’m building the Earth and I’m destroying the Earth,” she said.

Rufu found that it was possible to create a moon from a series of impacts much smaller than those posited in the Big Splat. “If we have about twenty, we can build a moon-size moon,” she said. Each impact would have ejected a disk of molten earth that slowly coalesced into a moonlet. This object would then have migrated far enough from Earth to be safe from the next impact, a few million years later. One by one, Rufu said, the moonlets would have merged to form a single, large, well-mixed lunar mass.

Rufu readily admits that her team has more work to do. Their model assumes that the individual moonlets survived and persisted, whereas, realistically, they may have been just as likely to be reabsorbed or lost. (The next step is to examine the merger process itself.) But for the moment it appears that our moon is statistically as likely to have been produced by multiple impacts as by a single one. “I applaud the group,” Canup said. “They’ve convinced me that maybe it’s now worth considering. Suddenly, the multiple-impact scenario looks equally probable—or improbable, depending on your perspective.” Rufu and Canup wouldn’t mind having a few more moon rocks to study. Although the current analysis is consistent, the rocks come from only a few regions on the moon, and they demonstrate how even a handful of samples can drastically alter what scientists think they know. “If we didn’t have the rocks, we would have convinced ourselves thirty years ago that we’d solved the origin,” Canup said.

What’s clear beyond question is that the moon we see—singular, majestic, the iconic master of time and tides—is only the most recent in a long line of them. Like the human species, maybe, it’s a sole survivor. “Irrespective of whether the moon formed from one impact or many, the last stages of Earth’s growth involved many, many impacts, so we should have had many, many moons before the current one,” Perets told me. “We should have had a ton of them. Probably not more than two, at most three, at the same time.” Rufu said, “That would be fun to see, right?”

Our current moon is also undoubtedly our last, or at least the last one that we’ll see. Its orbit is slowly expanding, sending it a couple of inches or so farther from Earth every year, and one day it will escape our grip entirely. And there will be no new moon to replace it, since there’s nothing big enough left in our neighborhood to help create one—asteroids, maybe, but “nobody really expects a planet-size thing to come and impact Earth,” Perets told me. Then he caught himself. “Actually, Mercury isn’t really stable in its orbit in the long run—we’re talking maybe ten billion years,” he said. “But I don’t think we need to worry about that right now.”