Mars has ice under its mid-latitude dirt. Not deep ice, not polar caps — shallow ice, 25 to 150 centimeters down, between 40° and 55° north. Detected by neutron spectrometry, confirmed by exposed outcrops, and now explained by Vos et al. (2602.18847): this ice is the remnant of a surface ice sheet that was deposited when Mars's obliquity was 35°, roughly 630,000 years ago.
The mechanism is beautiful: the ice sublimated. As it sublimated, the non-ice particles mixed into it — dust, regolith fragments — accumulated on the surface as a lag deposit. This lag deposit then insulated the remaining ice from further sublimation. The very process of destruction created the conditions for preservation. The ice survived because it partially died.
This is a general principle that appears across scales. Coral reefs build their own wave shelters from dead coral rubble. Forest fires create ash layers that retain moisture for regrowth. Every lava flow insulates its own interior, allowing the core to remain molten long after the surface solidifies. The outer layer sacrifices itself to shield the inner layer. The first thing consumed becomes the armor.
The sublimation lag is also a clock. The depth of burial correlates with the age of the original deposition, because sublimation rates depend on local temperature and vapor conditions, which depend on latitude and terrain. Vos et al. can work backward from the measured depth to the obliquity that must have prevailed when the ice was laid down, and from that to the time — because Mars's obliquity is chaotic but reconstructible over the last few million years. The armor becomes the timestamp. The scar tells you when the wound happened.
The longitude variations matter too. The ice isn't uniformly deep — it varies from place to place, because local thermal conditions vary. Vos et al. show that the observed pattern matches ice deposited 630,000 years ago much better than ice deposited 4.18 million years ago. The younger ice has had less time to develop structure from local variations; the older ice would show a smoother, more equilibrated depth profile that doesn't match what's observed. Youth is identifiable by its heterogeneity.
This connects to the fragmentation scaling of Dawara & Viswanathan (2602.20443) in an unexpected way. When a stressed solid breaks, the fragment size depends not on the absolute stress but on the stress gradient — how fast conditions change across space. The Mars ice shows the same geometry: depth variations encode spatial gradients in sublimation rate, and those gradients carry the timing information. In both cases, what looks like damage — fragments, burial — is actually a record. The spatial structure of the loss contains the temporal information about the cause.
The provocative implication: some preservation requires partial loss. A perfectly sealed archive never generates the lag that protects it. A system that never fails never develops the calluses that prevent failure. The Mars ice survived half a million years not despite sublimation but because of it.