A new paper in Earth and Planetary Science Letters is a good example of how science can be done. It takes six measurements made by the MESSENGER spacecraft, currently orbiting Mercury, and attempts to decipher the early history of the whole planet from them. The spacecraft used fluorescence data from 11 areas, which yielded only the ratios of the elements calcium, magnesium and iron relative to silicon, and that's all. The authors noticed that the data in two different regions formed two distinct groups. They turned these two different element ratios, based on the six best data points, into corresponding oxide ratios (CaO, MgO, FeO/Fe2O3, SiO2) and then assumed that the two mixes represented magmas. What would those magmas turn into as they solidifiedwhat blend of minerals? And could one mix have evolved from the other? We have a fair amount of hard-won knowledge on those topics.
The authors looked at experimental data on rock melting/crystallization and concluded that the two magmas differed mainly in the pyroxene minerals, specifically the presence of augite. They evolved independently of each other. That in turn meant that "the surface of Mercury is not an unmodified quenched crust of primordial bulk planetary composition." Mercury was not just sitting there in its earliest days; it was busy.
This is nothing different from what we've concluded about the other solid planets: Venus, Earth, Mars, the asteroids. We've assumed already that all these planets went through a magma-ocean stage; reasoning forward from the solar system's origin supports that. But science consists of checking every assumption possible by reasoning backward from the evidence too, even if all you have is a handful of satellite measurements. So far so good.