The leading hypothesis for the moon's formation contends that a massive impact billions of years ago knocked a wealth of planetary material off of Earth, which coalesced into our lunar companion.

But that coalescence was not a serene process—the heat of formation left the nascent moon coated with an ocean of magma, researchers studying samples returned by Apollo 11 theorized in 1970. As that ocean cooled, its most buoyant components floated to the top, forming an outer shell over denser layers of rock.

Now, a geological survey of the moon conducted using data collected by the recently retired Kaguya spacecraft supports this magma-ocean hypothesis, finding that the upper layer of the moon's crust is indeed rich in low-density rock of exceptional purity. The results from the Japan Aerospace Exploration Agency's Kaguya, formerly known as the Selenological and Engineering Explorer (SELENE), are published in this week's Nature. (Scientific American is part of the Nature Publishing Group.)

The Kaguya orbiter looked for anorthosite, a type of rock enriched in the relatively low-density mineral plagioclase. The spacecraft's survey found that in fresh craters and basins across the lunar highlands, the terrain that constitutes the majority of the moon's crust, anorthosite was not only prevalent but was almost entirely pure—that is, the rock was nearly 100 percent plagioclase. The study's authors speculate that exceptionally pure anorthosite may make up a global layer of the moon's crust from roughly three kilometers to 30 kilometers below the surface.

The ubiquity of the buoyant crustal rock points to a formation process that worked the same way across the moon, such as a magma ocean. The Kaguya results show that anorthosite formation "is really a global process," says John Longhi, a petrologist at the Lamont-Doherty Earth Observatory of Columbia University and currently a visiting scholar at Duke University, who did not contribute to the study. "Forming it by any other means than a magma ocean seems very doubtful, even though I've proposed one of those models. This research seems to suggest that there's really no other way."

Longhi's comments are significant, given that Paul Warren, a geochemist at the University of California, Los Angeles, calls Longhi "the author of, I think, the leading contender versus the magma ocean." In that model, plagioclase-rich rock rose to the upper crust as a result of reheating after an ocean of magma had already crystallized. Some other theories dispense with the need for a magma ocean entirely.

Warren, who did not participate in the Kaguya study, says that the new anorthosite research "implies that you had an awfully efficient process for purifying that mineral." The rock's prevalence and purity "is hard to reconcile with any kind of serial model where you have things happening in a piecemeal way," Warren says. "All that tends to weigh for a globally consistent process that is most readily explained with a magma ocean model."

Both Longhi and Warren harbor some skepticism about the extraordinary level of plagioclase purity in the anorthosites detected by Kaguya but maintain that the general implications remain relevant. Extremely homogenous composition is nearly impossible to find via remote sensing of the moon, Warren notes, because lunar soil is such a hodgepodge, and it is bound to contaminate any given surface to some extent. "Whatever extreme compositions might occur in the underlying crust, they are bound to be dampened" by deposited soil, he says.

"I find it a little hard to believe in detail," Warren says of the Kaguya paper. "But nonetheless, even if the details might be a little off, the import of what they've got is very great."