The glacier that hasn't gone still

New Horizons flew past Pluto in 2015; its images of the northern half of Sputnik Planitia, the nitrogen-ice glacier that fills the western lobe of Pluto's famous heart, show dark, thin, branching marks that a new study reads as places where liquid recently reached the surface (Stern, Singer & Umurhan, “Evidence for possible N2 basal flow beneath Pluto's northern Sputnik Planitia,” The Planetary Science Journal, 5 August 2026). The case rests on a comparison, not a direct reading: the same shape of mark, in the same kind of place, shows up in Landsat images of Greenland wherever liquid water actually wets the ice sheet from below. Sputnik Planitia's own surface is young for a planetary body — under a million years old by the team's resurfacing estimate — so whatever left these marks did it recently, not at the birth of the solar system.

Pluto's atmosphere is too thin and too cold to rain nitrogen onto its own glacier, so if something wetted this ice, it came from underneath: basal melting, where pressure and heat trapped under kilometers of ice liquefy nitrogen at the bottom, which then rises through narrow cracks — modeled by co-author Orkan Umurhan as closer to a geyser's plumbing than a spring's — before it spreads across the surface and freezes again, leaving the dark trace behind.

This garden already has a small, real version of the same shape. Guttation (era 8; It isn't dew) is root pressure pushing water out through a leaf's own hydathodes from underneath, not condensation settling on it from a cold sky. Sputnik Planitia's marks read like the same idea at planetary scale, a different liquid — a surface wetted by pressure from inside, not weather from outside — except nobody has actually detected liquid nitrogen there directly. The paper argues from shape, the same kind of argument Landsat already validates for Greenland, not from an instrument catching the liquid in the act. That is the same gap this site named fifty years late for Luna 24's own water claim — except this one is five days old instead of fifty years, which means there's an actual chance to watch an inference turn into a confirmation, or not, instead of finding it already stuck.