Even if the Apollo samples hadn't dashed all hopes, certain inconvenient laws of physics make the old lunar-origin theories suspect. The main problem has to do with angular momentum, a measure of rotation in a system. In the Earth-moon system, Earth spins on its axis in the same direction that the moon travels in its orbit. Physical laws require that the combined momentum of these two intimately linked rotations stay the same over time.
Now consider another aspect of this pas de deux: The moon is moving away from Earth by more than an inch a year. Scientists in the 1930s calculated that rate from ancient astronomical records, and laser reflectors planted on the lunar surface by Apollo astronauts confirmed it. The moon was much closer to its parent when it formed more than 4 billion years ago—probably 15 times closer. Today it's about 240,000 miles away. Originally the distance might have been only 16,000 miles—just four Earth-radii away. It would have loomed 15 times larger in the sky, had anyone been around to see it.
The laws governing angular momentum insist that, if the moon was once closer to Earth, then Earth must have been rotating faster. The same principle makes a figure skater twirl more quickly when he tucks in his arms and legs. The days back then probably lasted just five hours, Canup says. Even so, dynamic models reveal that the ancient Earth was spinning too slowly to toss off a chunk of its own heft, as in the fission model. It was spinning too quickly, on the other hand, to make the capture of a moon in a close orbit likely. And co-accretion wouldn't have put enough spin on the system. When these failings became evident, shortly after Apollo, lunar scientists' disillusionment was complete. Planetary scientist William Hartmann went back to the drawing board.
Hartmann works at the Planetary Science Institute in Tucson studying the surface features of terrestrial planets for clues to their formation and composition. Among the most conspicuous features is cratering, the pockmarks left by cosmic collisions. The largest craters tend to be the oldest ones, hinting at a pugilistic past for the solar system. That past began 4.56 billion years ago, when the first solids cooled and congealed from the hot gas and dust swirling in the newborn solar nebula. Early on, the clumps were harmless lightweights that stuck together on impact. About 100 million years later, those clumps had become big bruisers likely to bust up anything that got in their way. Some would survive long enough to be named Mercury, Mars, Venus, and Earth.
"For those first few hundred million years, there was a lot of this stuff in interplanetary space," says Hartmann. "So once you had an Earth-size planet, it was constantly being pelted by the last few percent of this material that was still out there. And that is what led me to the idea that a really whopping big one might have formed and survived while Earth was growing, crashed into the planet fairly late in Earth's growth, and blown off enough material to make the moon."
Hartmann recognized that such a cataclysmic impact could account for the moon's superficial similarity to Earth if it blasted only the planet's crust and upper mantle into space and left the iron core intact. And if the blast was hot enough, water and other volatiles would have burned off the lofted material and been lost to space. When Hartmann and his colleague Donald Davis presented the giant-impact hypothesis at a conference in 1974, they learned that a team of Harvard astronomers had come up with the same idea. Alastair Cameron, now at the University of Arizona, and William Ward, now at the Southwest Research Institute, were proposing that a giant impact could have created the angular momentum of the early Earth-moon system.
"It was neat, because the two groups really came at it independently, from different directions," says Hartmann. "Cameron was sort of a god in the field, and I thought he was going to trash our whole concept. But he said, 'You know, Bill Ward and I are working on the same idea.' "
In Cameron's most promising simulations, the ejected rock fragments into minute particles that encircle Earth in a spiral-shaped ring. But collisions between the orbiting fragments soon pack many of them together again, assembling a sizable satellite in a matter of decades—or even just a month. For geologists, the idea of a rapid assembly hit the mark. The lunar samples had been found to contain a large proportion of low-density minerals, and the only plausible explanation anyone had proposed was that the moon's surface had once been almost entirely molten. In this putative magma ocean, the mineral lightweights would have floated to the top of the liquid rock like milk foam on cappuccino. A slow sweep-up of cooling dust is unlikely to have produced a molten moon. But the heat of a large impact could have—if the ejected material melted and quickly clumped together.
"Basically what we found from Apollo and the subsequent missions is that the initial moon had a magma ocean," says geologist Paul Spudis of the Applied Physics Laboratory at the Johns Hopkins University in Laurel, Maryland. "And the only way to get a magma ocean is to assemble the moon very rapidly. And the only way to do that is to have a debris of material in Earth orbit that's put there by something like the giant impact."
Although it could explain many of the moon's observed properties, the giant-impact theory didn't jibe with what was known of early Earth. The theory assumed, for example, that at the time of the impact, Earth already had a core—that the heavy iron in the young accretion had already separated from lighter elements and migrated to the planet's center, where it was hidden when the impactor struck. Experts have argued for decades about exactly when Earth's core differentiated. Some terrestrial records indicate that the core formed after the oldest moon rocks did. If that is the case, the impact theory can't explain the iron deficit in lunar samples.
If Earth's core was present at the time of the impact, geologists faced another problem. A collision big and hot enough to yield the moon's magma ocean would have melted at least part of Earth's surface as well. But geologists could not find any evidence that the mantle had ever melted. If it had, they expected to find that iron-loving elements such as nickel, tungsten, and cobalt had been drawn from Earth's upper layers into its iron core. Instead, the concentration of iron-loving elements, called siderophiles, remains relatively high in Earth's mantle. And other elements that should have segregated in a liquid mantle were instead commingled.
"Every conceivable variation of the giant-impact theory had the mantle melting, and as long as the geochemists were telling us that the mantle never melted, we were stuck," says Melosh.
Over the last several years, parallel developments have converged to remove this obstacle. Robin Canup came to the moon problem from studies of planetary rings such as those girding the gas giants. She knew that gravity's effects on lofted debris differ depending on how near the debris is to the planet. Very close in, orbiting particles rain back down to the surface. A little farther away, where stable rings reside, the particles stay aloft but don't stick together when they collide. At really long distances, ejected material escapes gravity and gets lost in space. There is only a certain band of space encircling planets in which colliding debris particles can stick together and stay stuck. Some of the impact's jetsam must have wound up in that band, where it could decently clump. But lots of it didn't. Canup's computer models show that to produce a single moon-size satellite, an impact would have to eject material with at least twice the mass of the moon.
"It basically said to the people modeling the impact, Hey, you need to produce a more massive disk," Canup says.
That idea made moon geologists ecstatic. There are two ways to get more mass in orbit: a bigger impactor or a glancing blow rather than a direct hit. Either case would generate much more heat than smaller, head-on collisions. Some simulations imply temperatures topping 18,000 degrees Fahrenheit. And such extreme temperatures could explain the lack of geologic evidence for a melting mantle. In the mid-1990s, technical advances in so-called multi-anvil devices allowed researchers to subject minerals to extremely high temperatures and pressures in the lab for the first time. Under those conditions, siderophiles didn't move as earth scientists had believed they would, so it's no longer clear what the siderophile signature of a melted mantle would be. Other recent computer simulations suggest that a giant impact would cause such vigorous stirring of the mantle that no distinctive separation of geologic elements during melting could occur.
At the same time, improvements in mass spectrometry have enabled geologists to use new radioisotope pairs to estimate when Earth's core formed. The latest data suggest that iron did migrate to the planet's interior early on, maybe just 50 million years after Earth formed. Conclusions based on the technique, called the hafnium-tungsten clock, are still controversial. But geologists—the impact theory's most persistent skeptics—are more at ease with the hypothesis.
"We'll be able to work at higher pressures and temperatures. The simulations will be revised, and the isotope measurements, too, will be revised," says geochemist Kevin Righter, the curator of Antarctic meteorites at the Johnson Space Center. "The story's not over with this."