Is Mercury's Polar Ice Symmetric? A Pre-Registered Quantitative Framework for Testing the North–South Composition Assumption with BepiColombo's MGNS Instrument

<jats:p>MESSENGER's Neutron Spectrometer provided strong evidence for polar water ice at Mercury's north pole, but its eccentric orbit never sampled the south pole. Every existing estimate of Mercury's polar water-ice inventory therefore assumes, untested, that the south pole shares the north's composition. We present a pre-registered, quantitative framework testing that assumption with BepiColombo's Mercury Gamma-ray and Neutron Spectrometer (MGNS). We synthesize independent, MESSENGER-era ice-thickness estimates into a pooled Monte Carlo ensemble (median 43.1 m, 95% CI [0.9, 92.3] m) and propagate this through an MGNS response model calibrated from the instrument team's own published sensitivity figures. Our north-pole prediction (median suppression 3.40%, 95% CI [1.38%, 5.74%]) is consistent with MESSENGER's directly measured 2.4% signal , and our south-pole minimum-detectable threshold (median 0.145 wt% water-equivalent hydrogen) is of the same order of magnitude as the MGNS team's stated 0.1 wt% sensitivity; both are insensitive to how the thickness ensemble is constructed. A sensitivity analysis shows that the dominant source of predictive uncertainty (96.9% of total variance) is not ice composition or thickness, both of which saturate the response model, but the instrument's spatial-response geometry, not yet confirmed for MGNS. We formulate three falsifiable delivery-mechanism hypotheses for the south pole's ice origin: a suppression signal below ~9.25% favors ballistic delivery over the combined alternative, while symmetric accretion and atmosphere-mediated delivery prove statistically indistinguishable at every level tested. All evidence, code, and a reproducible analysis notebook are archived publicly.</jats:p>