The Repository Deep Time Archives

The Fish Was Still Breathing When the Impact Reached It

August 26, 2026 · Runtime 9:58 · Watch on YouTube

In a fossil site in North Dakota, there is a fish with a bead of impact glass lodged in its gills, not swallowed after death, but caught in the exact structure it used to breathe. The site, called Tanis, is one of the most extraordinary, and most contested, claims in modern paleontology. This episode covers both.

Chapters

  1. 0:00 A fish breathing when it arrived
  2. 0:23 Not exactly what peer review says
  3. 0:46 66.043 million years ago
  4. 1:07 Chicxulub, Mexico, and Tanis, North Dakota
  5. 1:30 1980: a thin line of iridium
  6. 2:04 One bed, laid down at once
  7. 2:27 A metre and a half of jumbled debris
  8. 2:46 Glass raining from the sky
  9. 3:06 A bead trapped mid-fall
  10. 3:32 The seiche mechanism
  11. 3:55 Spherules in the gill rakers
  12. 4:19 Under an hour, minute by minute
  13. 4:43 An extraordinary claim
  14. 5:02 The magazine ran first
  15. 5:25 What outside scientists said
  16. 5:46 Private land, one gatekeeper
  17. 6:10 A second team, a narrower question
  18. 6:25 Growth rings and isotopes
  19. 6:49 Accusations of fabrication
  20. 7:10 The Manchester investigation's verdict
  21. 7:37 Same season, from rivals
  22. 7:54 Why the season matters
  23. 8:29 What almost nobody disputes
  24. 9:05 What verification looks like
  25. 9:39 Next: the fern that cooled a planet

Transcript

In a fossil quarry in North Dakota, there is a fish with a small bead of glass lodged where its gills should be filtering water, not debris. It did not swallow that bead after death. It was caught in the one structure built to catch exactly that kind of particle, while the animal was still using it. Something arrived while this fish was still alive, and breathing.

You may have heard that this site, called Tanis, caught the actual day the dinosaurs died: feathers, skin, eggs, all preserved in the same afternoon. That is not exactly what made it through peer review. What actually got published is narrower, stranger, and it came bundled with one of the ugliest disputes in modern paleontology.

This episode sits at a single, precisely dated line: sixty-six million and forty-three thousand years ago, the boundary between the Cretaceous and the Paleogene periods. Radiometric dating pins it to within about eleven thousand years, either way. It is the last afternoon of the age of dinosaurs.

Two places matter here, roughly three thousand kilometres apart. Chicxulub, on Mexico's Yucatan Peninsula, where an asteroid about ten kilometres wide hit a shallow sea and blew out a crater more than one hundred and fifty kilometres across. And Tanis, a streambank in the Hell Creek Formation of southwestern North Dakota, nowhere near an ocean.

None of this started with Tanis. In 1980, a physicist and his geologist son found a thin dark clay band, about a centimetre thick, sitting at the same boundary in Italy, Denmark, and New Zealand. It was unusually rich in iridium, an element rare on Earth's surface but common in asteroids. That thin, ordinary line, found on three separate continents, is what convinced most of the field an impact had happened at all. Decades before anyone claimed to have found the actual afternoon.

Most of the Hell Creek Formation preserves fossils the ordinary way: bones scattered by rivers, buried gradually, mixing different years and even different decades in the same rock. Tanis is not that. It is one thin, graded bed, laid down all at once, sitting directly beneath the thin clay layer that marks the boundary itself.

That bed runs up to a metre and a half thick in places: sand and coarser debris, packed with fish, tree trunks, and burrow casts, all jumbled together the way a flood leaves things. Not the way a river slowly buries them, one season at a time, for years.

Mixed into that same bed: small glass beads, up to five millimetres across. Molten rock, blasted out of the crater by the impact, that cooled into droplets as it arced back down through the atmosphere. The same glassy fallout turns up in impact debris fields across the world.

Some of that falling glass never touched the ground. It hit trees instead, and got sealed inside hardening resin, which is exactly how amber forms: a bead of impact glass, trapped in tree sap, mid-fall. The moment frozen the instant the resin closed over it. Amber like that is part of why researchers can be this specific about what was raining down, and roughly when.

Tanis sat nowhere near an ocean when the asteroid hit. It was on the edge of the Western Interior Seaway, a shallow inland sea running the length of the continent. The proposed mechanism is a seiche: the same sloshing you get in a bathtub, but triggered by earthquake-scale shaking of an entire inland sea, sending a surge of water and sediment up the rivers that fed it.

Here is the detail that ties the fish to the moment itself, not to some later flood. Under a scanning electron microscope, roughly half of the fish examined had glass spherules lodged in their gill rakers, the one structure that only catches particles from water a living, breathing animal is actively filtering. The fish were alive, breathing, when it happened.

Run the physics backward, and the whole sequence compresses into well under an hour. Seismic waves from the impact would have reached this stretch of the continent in nine or ten minutes, triggering the seiche. Ten to twenty minutes after that, the glass began falling from the sky. A second surge, minutes later, buried what the first one had already killed or stranded.

A claim like that, an actual afternoon held inside a single bed of rock, is about as extraordinary as paleontology gets. Extraordinary claims are supposed to get the hardest scrutiny science has. This one got scrutiny, but not in the order anyone would have chosen.

Days before the peer-reviewed paper came out, in 2019, a long magazine profile described dinosaur skin, feathers, even eggs, all supposedly recovered from the same layer. None of that is in the paper. It mentions exactly one dinosaur bone fragment, and only in a supplementary document, not the main text.

Paleontologists outside the project noticed the gap immediately. Some called the geochemical evidence thin. Others pointed out that a single bed cannot, on its own, prove timing down to the minute the way the coverage implied. Reasonable doubt is not the same as being wrong, but it is not nothing either.

And there was a structural problem, separate from the data itself. The site sits on private land, under the control of the one researcher who excavates it. For years, that meant almost nobody outside his own team could examine the rock in person, or check the claims against the original material: the kind of independent look that usually catches a mistake before it goes to print.

A separate line of evidence came from a different team entirely, chasing a narrower and more testable question. Not 'did this happen in an hour', but something a fish skeleton can actually answer: what season it was.

Sturgeon and paddlefish, like trees, lay down annual growth rings in their bones: a dense band through the fast-feeding warm months, a thin band through winter. Oxygen and carbon isotopes locked into each band track the fish's migration cycle, year by year. Read the very last band before death, and you can read the season the fish died in.

The two teams did not publish quietly, or together. Melanie During's paper on the fish bones was still under review at Nature when the site's original excavator published his own paper, reaching a similar conclusion, from a separate set of isotope data. During accused him publicly of fabricating results to beat her into print.

A formal investigation followed, run by the University of Manchester. It found no fabrication, and evidence the isotope work had genuinely begun before During ever visited the site. But it also found real research misconduct: sloppy handling and presentation of the isotope data. Serious enough that the university said so in public, even while leaving the paper's actual conclusion standing.

That is the part worth sitting with. Two teams that ended up publicly accusing each other of scientific misconduct, working from different fish, different labs, different methods, and they landed on the same season: boreal spring, or at latest early summer.

Season is not a trivia detail. Spring in the north is breeding season for many species, and autumn in the far south, the same moment, on the other side of the planet. An impact that lands mid-breeding-cycle for one hemisphere, and mid-dormancy for the other, does not kill everything the same way. That asymmetry may be part of why extinction across the K-Pg boundary was so selective. During and her colleagues proposed exactly that: that the timing itself may have shaped who made it across the boundary, and who did not.

Strip away the fight, and here is what almost nobody in the field actually disputes: an asteroid did hit Chicxulub at the end of the Cretaceous. The Tanis bed really was deposited by a seismically triggered surge, close enough in time to the impact to matter. Fish really did have impact glass lodged in their gills. What stayed contested was never the basic physics. It was the precision: whether an afternoon, a season, a single hour, can really be pinned down that tightly, and whether everyone involved was honest about how they got there.

None of this is the clean story the magazine article promised. It is messier: a real site, a real fight, a formal misconduct finding, and a conclusion that still held up anyway, because it was checked twice, by people who actively distrusted each other. That is closer to what verification actually looks like than any tidy version of it. The lesson is not to distrust every extraordinary finding. It is to ask, every time: has this actually been checked, by someone who had reason to want it to be wrong?

The next time this channel follows carbon, it will not be a single afternoon. It will be a fern: a freshwater fern that turned an entire ocean fresh, floating on the Arctic for more than a million years, right as Earth's climate tipped from greenhouse toward the icehouse world we still live in.

Description and sources

Tanis is a single graded bed in North Dakota's Hell Creek Formation, tied to the Cretaceous-Paleogene boundary at 66.043 million years ago. Its key evidence: impact glass lodged in the gill rakers of roughly half the fossil fish examined, proof they were breathing when the debris fell. But a magazine profile ran days before the 2019 peer-reviewed paper, describing dinosaur skin and feathers that never made it into the actual study, and outside scientists flagged thin evidence and a site controlled by one excavator on private land. A second question, what season it happened in, was answered by a rival team, and the two researchers ended up publicly accusing each other of misconduct. A University of Manchester investigation found no fabrication, but did find real misconduct in the isotope work, without overturning the conclusion. Both independent methods still converge on the same season: boreal spring. That contested, checked-twice result is closer to real scientific verification than the clean magazine version.

PRINT-READY, FROM THIS CHANNEL

The Deep Time Record, Vol. 01 — the cores, the curves and the clocks

https://therepository.gumroad.com/l/deep-time-record

PRIMARY SOURCES

ARCHIVE PLATES — public-domain documents, with author, date, archive and licence.

... and 3 more plates, each credited on screen in the video and listed in this project's manifest.

The 16 generated plates were made for this channel with FLUX.1-schnell. No person is depicted in any of them.

Deep Time Archives rebuilds the pre-human Earth from the physical record — outcrops, drill cores, golden spikes, catalogued specimens. Every episode starts from an object you can go and check, and the sources are listed above.

About the images. The living paddlefish and the Hell Creek Formation outcrop are archive photographs, credited on screen. The impact scene and the reconstructed disputes are AI-generated, because nobody photographed the minutes after Chicxulub. This video is labelled as containing altered or synthetic content. The diagrams are drawn in code from the published numbers, not generated.

#DeepTime #EarthHistory #Paleontology #Chicxulub #ScienceDocumentary