The moon's origin has been the subject of scientific speculation since Galileo in 1609 showed that the moon is a rocky body like our earth. Just before the Apollo landings began in 1969 there were three different theories of its origin: The fission theory, proposed by G. H. Darwin, Charles Darwin's son, supposed that the moon was spun out of the earth's mantle during an early era of rapid rotation of the ancient earth. The capture theory supposed that the moon formed somewhere else in the solar system and was later captured in orbit about the earth. The co-accretion or "double planet" theory supposed that the earth and moon simply grew together out of a primordial swarm of small "planetesimals". When confronted with the evidence of the lunar rocks, none of these three theories could be confirmed--all made predictions at variance with the observations that the moon has no substantial metallic iron core, that its rocks are grossly similar in composition to the earth's mantle (its oxygen isotopic ratios are identical to the earth's), but that the lunar rocks are slightly enriched in refractory elements and are strongly depleted in volatiles. This disagreement was resolved in 1984 when a new theory of the moon's origin began to gain attention. The new theory stemmed from the recognition that the early solar system 4,500 million years ago was a more violent place than had been previously assumed. Rather than being filled with swarms of 10 km diameter planetesimals accreting directly into the four inner planets, it was realized that accreting matter would form embryonic planets with a large range of sizes in closely spaced orbits. The final stages of planetary formation would involve the coalescence of often rather large bodies, punctuating this era with giant impacts in which bodies of comparable size crashed into one another at high speed. The chaos of this era explains the wide variations in orbital inclinations, eccentricities, rotational periods and spin axis directions observed among the inner planets at present.

A giant impact provides just the right circumstances for a body with the moon's peculiar chemical composition to arise. The vapor squirted from the contact point between the proto-earth and the impacting smaller protoplanet would consist predominantly of material from the mantles of the two objects and should exclude core metal. Condensing in space, the high-speed cloud of rock vapor would preferentially incorporate refractory elements, while volatile elements would be slow to condense and hence may be greatly depleted. The large amount of angular momentum brought in by the projectile would mostly go into the orbiting debris, although the proto-earth would also be spun up. From the angular momentum of the present earth-moon system the projectile must have had a mass comparable to that of the planet Mars.

The computer-generated images below illustrate the first 30 minutes after a Mars-size protoplanet collides with the protoearth with a velocity upon contact of 8 km/sec. This contact velocity corresponds to a relative velocity of nearly zero at the initial large separation. These images are the result of a 3-dimensional hydrocode computation using the CTH code at Sandia National Laboratories. Each of the four images is separated by about 400 seconds. The metallic cores of the projectile and target are shown in red and pink and their dunite mantles are shown in brown and green, respectively. The vapor plume of mixed projectile and target mantle material is well developed in the second and third frames. This plume eventually condenses into dust, some of which remains in orbit to later accrete into a proto-moon with an initial orbital radius of about 10 earth radii.

These computations were performed by M. E. Kipp of Sandia National Laboratories and H. J. Melosh of the University of Arizona. This computation using the 3-D hydrocode CTH required a Cray I/XMP computer with a solid state disk approximately 100 hours of CPU time to complete. The computer graphics were prepared by the staff at Sandia National Laboratories and all picture credits should go to them.