Silicon dioxide behaves like a clock in a cooling planet, which is why the size of this number matters more than its novelty. As a mantle cools, the minerals that crystallise first take comparatively little silica with them, so the leftover melt grows richer in it, and lava erupted later carries more [4]. Read backwards, a silica-poor surface points to melt drawn from hotter material deeper down, which is the reading the authors take [3].
The arithmetic behind "up to 25% less" is worth doing out loud, because the source states it as a relative shortfall rather than in percentage points. If 37 percent is a quarter below what was accepted, the figure it displaces sat near 49 percent by mass, and the gap is roughly 12 percentage points of crust [13]. That is a large correction to a first-order property. For scale, terrestrial volcanic rocks run up to about 75 percent silica [12], so Mercury's crust now reads at about half the silica of Earth's most evolved lavas [14].
The thing this does not tell you is which of two histories produced the number. The team notes that Mercury may instead have started with more silica in its crust and gradually lost its oxygen [5]. Deep hot melting and oxygen loss both end with a silica-poor infrared spectrum, and a single mass fraction cannot separate them. So the study constrains the family of acceptable models without picking one member of it.
The control deserves attention, because it is the part that would be easiest to skip. Mercury has never been landed on and has yielded no samples, leaving remote infrared as the only route to composition [6]. Calibrating that route against half-millimetre laboratory glasses of known silica content is unremarkable in itself [7]; running the resulting relationship on the Moon first, where NASA's Lunar Reconnaissance Orbiter has been mapping infrared emission since 2009 and where returned rocks provide independent checks, is the step that makes the Mercury number arguable rather than merely asserted [8][9]. Lead author Christian Renggli calls the Moon a touchstone and a stepping stone [15]. It is a good one. It is also not Mercury, and no returned sample exists to catch a bias that the lunar test happens not to expose.
One boundary on the claim: the timing is not what moved. Mercury's volcanism is thought to have stopped as early as about a billion years after the planet formed, leaving a continuous solid crust [11], and that framing predates this measurement. What the 37 percent figure revises is the depth and temperature of the melting that fed those eruptions [3], not the date they ended. The Mercury spectra themselves come from ground-based observation, including the Bok Telescope [10].
