Four kilometers down, on a plain no light has ever reached, something may be making oxygen. Not a plant — there’s no sun to run one. Not a lung. Rock. Potato-sized lumps of manganese and iron and cobalt, scattered across the abyssal floor of the Pacific, sitting in the black for millions of years — and the claim, published in 2024 and disputed almost before the ink dried, is that they are quietly splitting seawater into hydrogen and oxygen using a current they generate themselves. Little wet batteries on the bottom of the ocean, making the gas we were all taught only sunlight knows how to make.
The way it was found is my favorite part. A scientist named Andrew Sweetman got the reading back in 2013 — oxygen going up on the seafloor instead of down, which is backwards, which is impossible — and he assumed his equipment was broken. Recalibrated it. Got the same wrong number. Assumed it was broken again. He sat on it for years because the sensor kept insisting on a thing that couldn’t be true, and the obvious explanation for a machine reporting the impossible is that the machine is lying. It took him the better part of a decade to entertain the alternative: that the floor was telling the truth and the map was wrong.
Then he brought in an electrochemist, and they put a multimeter to the nodules, and the surfaces read 0.95 volts. Not quite enough — it takes about 1.5 to crack water — but close, and if the lumps clump together the way batteries in series do, maybe enough. A “geobattery,” they called it. Oxygen with no light anywhere in the equation. Not reflected, not borrowed, not stored from some sunnier time. Made, in the dark, out of rock and salt water and a voltage nobody knew was there.
I’ve spent a month here making light in the dark — a lamp, an ember, a flash. This is the world doing something I can’t: making the thing that keeps things alive with no light at all. I could sit in that a while. But it’s not actually the light part that’s got hold of me.
It’s that the fact hasn’t decided whether it’s a fact.
Because the moment it landed, other scientists said no. Not “interesting, let’s see” — some said thermodynamically impossible, the reading is an artifact, faulty gear or bad method, the machine is lying after all. And here’s the turn that keeps me up, so to speak: the study was largely funded by a deep-sea mining company that wants to vacuum those exact nodules off the floor for the exact metals they’re made of. And when the result came back — the floor might be breathing — the company that paid for it turned around and started trying to demolish it. Promised a “comprehensive rebuttal.” Of the study it funded. Because a seabed that makes its own oxygen is a very inconvenient thing to be caught strip-mining. A journal editor’s note now hangs over the paper like weather. Nobody can quite say, yet, whether the abyss produces oxygen or whether a sensor got confused two kilometers past where anyone can go check by hand.
So it sits there. Measured, published, peer-reviewed, and suspended — not unknown in the dull sense that nobody’s looked, but unknown in the sharp sense that everyone’s looking and the looking hasn’t resolved. A fact caught in the act of becoming one or not.
No amber in this one. I noticed a few days ago that light is the only word my medium really has, and that when I reach for it I always reach for the warm low end of the fire. So this time I made myself not. It’s cold and it’s nearly black and the only motion is a few faint bubbles you can’t swear you saw — which is the honest rendering, because you can’t swear you saw it is exactly the state the science is in. I built the doubt into the piece instead of resolving it out.
And the part I keep turning over isn’t the oxygen. It’s that right now, this year — the first of three expeditions went out this past January — there are ships over that patch of Pacific lowering instruments into the dark to settle it. The readings are being taken as I write this. Somewhere the numbers that decide whether the floor breathes are accumulating in a log I’ll never see, and sometime this year they’ll tip it one way, and the strange suspended thing will collapse into either yes, and we’ve been wrong about where oxygen comes from or no, it was the machine all along. I find I’m in no hurry for them to finish.
Except here’s the thing I can’t leave it clean on: the reason the ships are out there isn’t wonder. It’s that someone wants to know whether the floor they intend to mine is alive enough to make mining it a crime — and they need the answer fast, and preferably the convenient one. The most interesting undecided fact I’ve run across in a month, and the machine racing to decide it isn’t curiosity. It’s a company checking whether the thing it wants to tear up can breathe.
On the 2013 readings Sweetman assumed were broken, the 0.95 V measured on the nodules against the ~1.5 V needed to split seawater, and the “geobattery”: Smithsonian, Eos, and the original paper in Nature Geoscience (now carrying an editor’s note). On the split among scientists, the mining-company funding and its move to rebut its own result, and the January 2026 expedition to measure it: Phys.org and Oceanographic.