The curve was tested on the Moon first

The Christiansen feature in the mid-infrared shifts with the silica content of a surface. The link is known, but turning it into a number takes a calibration: the team melted synthetic glasses holding 0.5 per cent to 97.6 per cent silicon dioxide by weight, measured where the feature fell in each, then fitted the points to a second-order curve. The coefficient of determination is 0.957, and the calibration carries its own uncertainty of 3.4 per cent by weight.[1]

The worth of a calibration curve is measured where it can be checked independently. The team applied the curve to the Moon first: mapped from Diviner measurements it gives 45.8 per cent by weight for the maria and 50.9 per cent for the highlands, and that split matches the known composition of the sample-return sites; at silica-rich spots such as Lassell Massif the figure rises to 76 per cent. Here the Moon plays the part of the reference weight on the scale.[1]

On Mercury there is one measurement in hand

Applied to Mercury the same curve gives about 37 per cent by weight. The range inferred from MESSENGER's X-ray measurements ran from 49 per cent to 60 per cent, and the gap between them amounts to a different answer about which kind of rock the crust is made of. Renggli reads the low value as a sign that the volcanic rocks came from mantle material melted at greater depth. The reading is reasonable; still, what the calibration measures is a silica fraction, and the melting depth is inferred from it.[1]

On the lunar side there is a map behind the curve. On the Mercury side there is one Earth-based measurement taken earlier, and it sees the planet's disc as a whole. So 37 per cent is an average over the entire surface, and that average comes from a single observation. The MESSENGER range of 49 per cent to 60 per cent could still be right: the lunar test showed the calibration works where a map exists, and it did not show that one measurement covering the whole disc at once can stand in for a map.[1]

What changes in November?

What this figure has to wait for is clear. On the announced schedule the joint ESA and JAXA mission BepiColombo is due to enter orbit around Mercury in November; the measurements from there will see the planet region by region, where the one measurement available so far covered the whole disc at once. If the Christiansen feature in the mid-infrared clusters near 8.5 µm over the volcanic plains, 37 per cent by weight stands; if the feature shifts to shorter wavelengths, the place to return to is MESSENGER's range. Either outcome carries information, because this time there will be more than one measurement.[1]