Some 30 years ago the planetary science community was surprised when the Mariner 10 spacecraft flew by the planet Mercury and detected an internal magnetic field (1). Earth's internal field is produced by a magnetic dynamo sustained by convective motions in the planet's molten, iron-rich outer core. Although Mercury's high bulk density indicates that its dominantly iron central core is the largest by fractional mass among the planets (2), the detection of its magnetic field was surprising because Venus has no field and Mars and the Moon show evidence only for ancient global fields. With a mass about 5% that of Earth, Mercury had been expected to have cooled internally to the point where either the core had solidified or core convection no longer occurs. A necessary condition for Mercury's magnetic field to arise from an active Earth-like dynamo is that at least the outer shell of its core be molten. On page 710 of this issue, Margot et al. report new observations of variations in Mercury's spin rate made with Earth-based radar, providing strong evidence that this condition is met (3).
The radar measurements constitute a triumph of two theoretical ideas developed decades ago. Shortly after the Mariner 10 discovery, Peale (4) outlined a procedure to determine whether the planet has a fluid outer core. His method was based on the observation that Mercury is in an orbital state in which the planet completes three rotations about its spin axis for every two revolutions around the Sun. The procedure requires the measurement of the small oscillation in the planet's spin rate (libration)--a few hundred meters in amplitude--forced by solar torques as Mercury follows its 88-day eccentric orbit. Additional parameters that must be known include the tilt of the spin axis and the components of the planet's gravity field describing the degree to which the field is flattened at the poles and out of round along the equator. The last two quantities have been estimated, albeit with low precision, from Mariner 10 tracking observations made during the probe's three encounters with Mercury during 1974-75, but the libration amplitude and a sufficiently accurate pole position were not known before now.
The second theoretical development, by Green (5) and Holin (6), stems from the recognition that irregularities, or speckles, in the radar signal returned from a planetary target rotate in space as the planet spins. Under suitable geometric constraints, analysis of radar signals recorded at two stations on Earth can detect this rotation as the speckle pattern sweeps coherently across Earth's surface. By combining many such paired measurements at different times and observing geometries, the position of the target planet's spin axis and periodic variations in the spin rate may be ascertained.
Margot and his team (3) applied these two theories with spectacular results. From radar signals bounced off Mercury and recorded at pairs of radio antennas in California, West Virginia, and Puerto Rico during more than 20 observation periods from 2002 through 2006, the group determined the position of Mercury's spin axis with a precision two orders of magnitude superior to the previous best estimate. Equally important, they detected Mercury's forced libration and determined its amplitude for the first time. The amplitude is sufficiently large that Mercury's solid mantle and crust must be decoupled from the planet's core on an 88-day time scale. This result indicates that Mercury has a molten outer core at 95% confidence, a level limited at present by uncertainty in the knowledge of Mercury's gravity field.