Despite its 400 °C surface temperatures, physicists originally predicted that Mercury's small mass – about 5% that of the Earth – would have allowed its core to cool down enough to solidify long ago. But their predictions became much less certain in the 1970s after NASA's Mariner 10 spacecraft flew by the planet and detected a small internal dipole magnetic field. Although some claimed that the field could have been a fossil of an earlier one "frozen" into the crust, others maintained that this was very unlikely, and that dipole magnetic fields in terrestrial planets are normally a result of the convection of molten iron producing a dynamo.

Now Jean-Luc Margot of Cornell University and physicists from other US and Russian institutions have used two previously-untried techniques to settle the dispute. The first technique required the measurement of the small oscillation in the rate at which Mercury spins on its axis, which on average is three rotations for every two 88-day orbits around the Sun. The second involved tracing how the "speckles" in radar images returned from the planet rotate as the planet spins. By combining data from both of these, they could calculate the periodic variations in Mercury's spin. Then, because the Sun's gravitational field affects the spin of planets differently depending on their composition, these variations would tell whether Mercury is solid throughout or has a detached, molten core.

Margot's team took measurements over five years from three telescopes – the National Science Foundation's Robert C Byrd Green Bank Telescope in West Virginia, the Arecibo Observatory in Puerto Rico and NASA/Jet Propulsion Laboratory antennas in California. These, together with previous estimations of the tilt of the spin axis and components of the gravitational field made by Mariner 10, enabled them to determine the periodic variations in Mercury's spin rate with an accuracy of one part in 100,000. They found that the variations were relatively large – characteristic of a planet with a molten core. This means that a lighter element, such as sulfur, must have alloyed with the iron in the core to lower the melting temperature and hence prevent it from solidifying.

The discovery also means that Mercury's magnetic field is almost certainly due to dynamo action. However, at just 1% the strength of Earth's, the field detected by Mariner 10 is too small to be have been produced by a completely molten core. Therefore the question still remains as to how deep into the core the molten iron goes, which is only likely to be answered when NASA's MESSENGER begins three flybys within 200 km of the planet next January. "MESSENGER carries a very good magnetometer and is magnetically clean," Sean Solomon, principal investigator on the MESSENGER mission, told Physics Web. "A lot of the theoretical models of Mercury's [core composition] depend on the geometry of the field. Not only will we get a very good measure of the dipole strength, we will also measure many of the shorter wavelength components of the field, which will give this geometry."