We have a phrase for things that happen too slowly to notice. We call them glacial.
In January 2022, Antarctica’s Hektoria Glacier still had that reputation. It was advancing, even — inching forward into the Larsen B embayment after a decade of cold recovery, rebuilding what the 2002 ice shelf collapse had taken away. It was doing what glaciers are supposed to do: moving at a pace that requires satellite imagery stretched across years to detect.
Then, in March 2022, it stopped being a glacier.
In nineteen days, Hektoria lost eleven kilometers. Not eleven kilometers of floating ice tongue — that had already broken away — but eleven kilometers of grounded ice, ice that had been sitting on bedrock and pushing back against the sea. By August 2023, the glacier had retreated twenty-three kilometers in total. Fifteen miles, in fifteen months.
The fastest collapse of grounded ice ever measured.
Naomi Ochwat, a postdoctoral researcher at CIRES, flew over the site in early 2024. She had already seen the satellite images. “But being there in person,” she said, “filled me with astonishment at what had happened.” 1
Astonishment is an interesting word. It isn’t panic. It isn’t even alarm, exactly. It’s the feeling of watching something you believed was fixed — the ground, the ice, the very grammar of slowness — prove itself fluid after all.
Hektoria didn’t just retreat. It unzipped. The glacier sat on a flat bedrock plain, an “ice plain” topography that let it go suddenly afloat when the thinning crossed a threshold. Once floating, ocean forces opened crevasses from below while surface stress cracked it from above. The two fracture systems met, and the glacier simply… let go. Seismometers picked up the glacier earthquakes in real time. The ice was screaming, and we heard it.
Here’s what haunts me: Hektoria is small. About 115 square miles — roughly the footprint of Philadelphia. When it collapsed, it was a tragedy in miniature, a dress rehearsal. The real concern is that this same flat-bedrock geometry has been detected beneath much larger Antarctic glaciers. The ones that actually matter for sea level. The ones that could, if they “forgot to be slow” in the same way, rewrite coastlines while we are still arguing about whether the timeline is “this century” or “next.”
Ted Scambos, a senior research scientist who has spent decades watching ice sheets, put it plainly: “Hektoria’s retreat is a bit of a shock — this kind of lightning-fast retreat really changes what’s possible for other, larger glaciers on the continent.” 1
Changes what’s possible. Not “accelerates the timeline.” Not “adds urgency.” Changes what’s possible. As in: the range of futures we thought we were choosing between was wrong. There were always other doors.
The Larsen B ice shelf collapsed in 2002. Hektoria’s final act began twenty years later, when the last protective sea ice blew out in January 2022, followed weeks later by an atmospheric river that dumped unprecedented melt on a glacier already thinned and crevassed. It was a perfect storm, they say. But perfect storms are just normal storms arriving in a world we’ve already destabilized.
For ten years, Hektoria had been healing. Sea ice had butted up against its front, giving it time to thicken, to push forward, to pretend the 2002 disaster was a one-time event. Then the ice left. Then the rain came. Then the glacier proved it had never been healed at all — only waiting.
We still say “glacial pace” when we mean something too slow to worry about. We still draw straight-line graphs of sea level rise, as if ice obeys our preference for gradualism. Hektoria didn’t read the graph. It retreated in stages, with pauses that looked like stability, then suddenly didn’t.
The truth is that glaciers have always known how to move fast. We just hadn’t seen one do it in the satellite era — not grounded ice, not like this. And now we have.
The question isn’t whether another glacier will follow. The question is which one, and whether we’ll still be using the word “glacial” when it does.
The satellite that caught Hektoria’s collapse was named Landsat. It has been watching since 1972. It will keep watching until it fails or we stop launching replacements. Either way, the ice doesn’t need witnesses to move. It only needs the water to be warm enough, and the bedrock to be flat enough, and the time — the time we thought we had — to run out.