The Repository Abyss Report: Into the Dark

NOAA Filmed a 5 mm Fish in 2019. It Still Has No Name.

September 11, 2026 · Runtime 26:32 · Watch on YouTube

On 1 September 2019, at 1,353 metres in the Northeast Channel off Nova Scotia, the ROV Deep Discoverer filmed a fish a few millimetres long. Seven years later it is still unidentified -- and the record made about it since has drifted further from the record made at sea.

Chapters

Transcript

A fish the size of this sentence 0:00

One thousand three hundred and fifty-three metres down, off the coast of Nova Scotia, a robot's lights are pointed at the seafloor. Something crosses the frame. It is small. If you are watching this on a phone, it is roughly the size of the letters in this sentence. It moves like a fish, because it is one. That is the entire encounter. Nobody caught it. Nobody has named it. And seven years later, nobody can.

One line of text, and what happened to it 0:30

Everything anyone knows about this animal is one line of text. It was typed at sea, on the day, by a person watching this screen. It ends in a question mark. Here is the part nobody expects. In the seven years since, that line has become less accurate, not more. This episode follows it one step at a time. Every step is something you can check yourself in about five minutes.

EX1905L2, dive four, Northeast Channel 0:57

The dive is EX nineteen-oh-five, leg two, dive four. First of September, two thousand nineteen. The ship is NOAA Ship Okeanos Explorer. The vehicle is Deep Discoverer, with the camera platform Seirios above it, lighting it from over its shoulder. The site is the Northeast Channel. That is the deep gap between Georges Bank and Browns Bank, at the mouth of the Gulf of Maine. Forty-one point nine eight degrees north, sixty-five point three eight west. Hold on to that coordinate. It matters later.

This is Canadian water. It was the expedition's last dive on the Canadian side, and the team was there for a specific reason. Just north of the dive track is the Northeast Channel Coral Conservation Area. It has been closed to bottom-contact fishing since two thousand and two, to protect four hundred and twenty-four square kilometres of coral. One more thing to hold on to. The expedition has a name, and the name is going to cause trouble later. Deep Connections twenty nineteen. Exploring Atlantic Canyons and Seamounts of the United States and Canada.

Two vehicles, and that detail matters for what follows. Deep Discoverer does the looking. Seirios hangs on the tether above it, carrying lights and a second camera pointed down at the first. So almost everything on this dive exists twice, from two angles, recorded in parallel. That is why a nine-hour dive produces seventeen hours of video. It is also why a moment that lasts seconds can still be checked from more than one position.

Down to 1,496 metres 2:42

The descent takes about an hour. Between roughly one thousand and one thousand three hundred metres, the lights pick up siphonophores and comb jellies drifting past. None of this is the subject of the episode. It is what the water looks like on the way down. The vehicle reaches bottom at just under one thousand five hundred metres. Soft sediment, cobbles, and a cod-like fish called Antimora hanging in the lights. From there the dive does one thing for eight and a half hours. It climbs.

One thousand four hundred and ninety-six metres at the bottom, one thousand three hundred and twenty-six at the end. A hundred and seventy metres of slope, walked slowly, with two cameras recording the whole time. Two hundred and twenty-six segments of video. About seventeen and a half hours of footage from one working day. And it climbs through the thing it came to see. This slope carries one of the densest cold-water coral and sponge grounds on the Atlantic margin. That is why there is a line on a chart around part of it.

Bubblegum coral, Paragorgia, in heads taller than the vehicle's manipulator. Bamboo coral, Acanella, in thin white thickets. Anthomastus, a soft coral that looks like a mushroom made of jelly, counted in twenty separate segments of this one dive. Then the sponges. Asconema foliata, a glass sponge that grows like a torn paper cup. Hyalonema, which anchors itself in soft mud on a twisted rope of silica fibres. Both are structures other animals live in, which is the entire argument for protecting the ground.

And they do live in them. Brittle stars wrapped around coral branches in the hundreds. Scale worms tucked into sponge folds. Squat lobsters. Sea urchins. Xenophyophores, single-celled animals that build sediment shells the size of a fist. Just after four in the afternoon, at one thousand three hundred and seventy-two metres, the dive finds an octopus. Graneledone, sitting over a clutch of eggs and guarding them. It gets four minutes of dedicated camera and its own highlight clip.

And at one thousand four hundred and fifteen metres, one of the subject lines on this dive is simply the word trash. That is in the public index too. The archive does not tidy that up either. While that is happening, somebody is writing. Not a report afterwards. A running index, typed into the metadata during the dive, describing what is on the screen as it appears.

The vocabulary is taxonomic, and it is uneven on purpose. Sometimes it is a phylum. Sometimes it is a genus. Sometimes it is a common name in capital letters. What it always is, is contemporaneous. It is what somebody thought while looking, before anyone had time to look it up. That makes it the most honest layer in the whole archive, and also the most fragile. It records a first impression as a first impression. Every later reader has to decide what that deserves.

The other question mark on this dive 6:02

This dive has more than one uncertain entry. At one thousand four hundred and eighty-five metres, a shark comes past the manipulator camera. The subject line for it reads: possible dog fish, question mark. That one got resolved. The dive report calls it a dogshark and moves on. A metre-long shark is a solvable problem. And that is worth sitting with, because this dive is full of fish that did get named. Not vaguely. Precisely.

The fish that did get named 6:34

Antimora, a deep-sea cod, at the bottom of the dive. Grenadiers, over and over, all the way up the slope. Halosaurs. Lantern sharks. And at just under one thousand three hundred and fifty metres, a genus: Apristurus, the ghost catsharks. Same dive. Same two cameras. Same person on watch, working the same keyboard, on the same afternoon. So whatever stopped the tiny fish from getting a name, it was not the equipment, and it was not the observer. Both of those are held constant here, by accident, in a single working day.

18:24:59 UTC, 1,353 metres 7:15

Six twenty-four and fifty-nine seconds in the evening, ship time. Depth: one thousand three hundred and fifty-three metres. The videographer driving the camera sees something almost too small to hold focus on, and holds focus on it. What is in frame is a fish with a large head for its body, sitting on the sediment rather than swimming above it. That posture is the whole of the visual evidence. It is also, as it happens, the posture of about six unrelated families of fish.

Where the word toadfish came from 7:58

Now. Where does the word toadfish come from? Not from the ship. This is a telepresence programme. The dive is streamed live, and scientists on shore watch it in a chat room and type into it. About two minutes after the fish appears, somebody at a console reads that chat out loud. Quote. In the chat room is speculating a teeny toadfish. End quote. That is the origin. A guess, from a viewer, relayed into the room by somebody reading a screen.

Should we get lasers on it? 8:40

Then they measure it, and this part I did not expect to find. Someone asks: should we get lasers on it? The vehicle is repositioned, the camera pulls wide, and a voice confirms the scale. Quote. Those lasers are ten centimetres apart. End quote.

Five millimetres to one centimetre 9:00

And with the lasers on the animal, someone says this. Quote. I would say that even one centimetre, I would put it at five millimetres or something like that. Extremely small. End quote. Then, deciding what to send back to the people watching: quote, five millimetres to one centimetre is our best guess. End quote.

Hold on to that, because of where it goes next. Which is nowhere. That number is never written into the metadata. It is not in the dive summary. It is not in the video caption. It exists in one place only, for seven years: the audio track of a public video that nobody had transcribed.

The line the watch typed 9:57

Here is the line the watch typed for that segment. Semicolon, fish, semicolon, tiny toadfish, question mark. That is it. That is the primary record of this animal, and it is four words long. Now the second record. NOAA publishes a summary of every dive, the same evening. Here is that page for dive four, quoted exactly. A tiny juvenile fish, species undetermined, was spotted by the keen eyes of our videographer.

And a third. NOAA cut the encounter into a short video and put it online. The caption there says the same thing again. Species undetermined. There is one more thing in this catalogue, and it is my favourite detail of the whole episode. Every dive gets a short highlight reel, and somebody has to name the file. On this expedition they named them after whatever the room said out loud.

Dive one is filed as: life in the Gully. Dive two: one of my favorites. Dive three: squid cafe. Dive six is named after a colleague who had died, and it reads: remembering Doctor Brian Kinlan. Dive four, sixty-one seconds long, Northeast Channel, first of September. The file name in the public catalogue, in capital letters, is: it's so cute.

I am not going to tell you that clip is the fish. It could as easily be the octopus, and I have not established which. What it does establish is the temperature of that room. Nobody in there was being cautious for the sake of it. They were charmed, and they said so, in the file name, permanently. And they still typed the question mark. Those two things are not in tension. Together they are what a good record looks like.

Three records, three question marks 11:53

Three official records, produced by three different parts of the same agency, at three different levels of formality. Not one of them names a family without a question mark attached. The archive is behaving correctly. So why can nobody finish the job? Start with the obvious answer, because it is wrong. Knowing why it is wrong is most of this episode.

The ruler is bigger than the subject 12:17

The obvious answer is size, and we just watched them run into it. The ruler bolted to that vehicle is ten centimetres wide. The animal is about half a centimetre. They were measuring with an instrument ten to twenty times longer than the subject, which is why the answer came out as a range and not a number. That sounds like a camera problem. It is not, and this channel can prove it, because we already made the episode that proves it.

The species named from video alone 12:46

Episode one of this channel was about Duobrachium sparksae. A comb jelly, filmed by this same programme, by this same class of vehicle, in twenty fifteen. In twenty twenty it became a new genus and a new species. It was described entirely from video. No specimen was ever collected. The footage itself was deposited as the type material. It sits in the Smithsonian's National Museum of Natural History, where anybody can go and look at it.

That was not a small thing to do. Naming a species from pictures alone had been argued about for years, on a reasonable objection. A specimen can be re-examined by the next generation. A video, historically, could not be. This one drew very little of that argument. Two reasons. The footage was good enough to show the characters that mattered. And it was deposited in a museum collection, catalogued, permanent, and open, exactly as a jar of preserved animal would have been.

Allen Collins, one of the authors, on what made that possible. Quote. The cameras on the Deep Discoverer robot are able to get high-resolution images, and measure structures less than a millimetre. End quote. Structures less than a millimetre. So the camera is not the limit, and it was not the limit in twenty nineteen either. The limit is somewhere else, and it is a much more interesting place.

Where the deciding character sits 14:19

A comb jelly wears the answer on the outside. Two tentacle arms. A body that reads rectangular from one angle, and round from ninety degrees away. Those are the characters that separate it from every other comb jelly, and they are all pointed at the lens. A fish does not work like that. What separates one fish species from another is mostly counting. Rays in each fin. Vertebrae. Spines. Bones in the gill arch. These are called meristic characters, and most of them are counted on a dead animal, under a microscope.

And this one is a juvenile, which makes it worse in a specific way. Young fish frequently look nothing like the adults of their own species. Different proportions, different pigment, sometimes spines and trailing fin filaments the adult simply does not have. So here is the finding the fish literature is quite blunt about. When you identify a larval or juvenile fish by looking at it, family is usually as far as you get. Genus and species normally need the specimen, or its DNA.

Read the line from the ship again with that in mind. A family. With a question mark. That is not a sloppy guess. That is somebody working at the exact ceiling of what looking at an animal can deliver, and then flagging that they were at it. And while we are here, the guess itself deserves defending. Toadfish is not a silly thing to have typed.

A blunt head that looks too big for the body. No obvious swim bladder lift. Sitting on the sediment on its fins instead of hovering over it. That is the toadfish silhouette, and it is a real visual category, learned by looking at a lot of screens. It is also the silhouette of sea toads, batfishes, sculpins, and a handful of other groups that solved the same problem the same way. Sitting still on the bottom with a big mouth is a job, not a family. Several unrelated lineages have taken it.

One last thing about that typed line, because it is easy to miss what it is for. Without it, this animal is four seconds inside one of two hundred and twenty-six unlabelled video files. The index is the only reason anybody can find the moment again, including me, making this. The guess that turned out to be shaky is the same thing that makes the record retrievable. You cannot have the second without somebody being willing to risk the first.

Nobody collected it 16:57

The last piece is collection, and it is undramatic. The dive did collect. Three sponges and a bubblegum coral went into the bio-box that day. Nobody tried for the fish. A few millimetres of animal, loose on sediment, is not something those arms can take. That is where the record stood, correct and incomplete, until the first of September, two thousand twenty-six. Seven years to the day after the dive, the fish was in the news.

Seven years later, to the day 17:28

The coverage was affectionate and it was widely shared, and this is not an episode about a bad journalist. It is an episode about what happens to a hedge as it travels. Here is what the published version said. That the fish measured five millimetres to one centimetre. That scientists thought it might be a juvenile toadfish, family Batrachoididae, also called the sea toad. And that it was seen in the Northeast Canyons and Seamounts Marine National Monument, off New England.

The part that is right 18:00

Start with the one that is right, because it is the most surprising thing in this episode. The size is correct. Five millimetres to one centimetre is not a number anybody invented later. It is, word for word, what the control room said out loud with the lasers on the animal, seven years earlier.

So the figure that reached the public is the team's own estimate, carried accurately across seven years. And it is the one thing NOAA's written records dropped. It never made it into the metadata, the dive page, or the caption. The most durable number in this story survived in an audio track, not in a database.

Toadfishes stop 275 metres up 18:42

Take the family first. Toadfishes, Batrachoididae, are shallow-water animals. They sit in sand and mud, in estuaries and on coastal bottoms. The depth range usually given for the family runs to about two hundred and seventy-five metres. This fish was at one thousand three hundred and fifty-three metres. That is not near the edge of the family's range. It is roughly five times past it.

Sea toads are a different family 19:10

Now the nickname, and this is the strange one. Sea toad is a real name for a real family, and that family is not Batrachoididae. It is Chaunacidae, the coffinfishes. Different order, different animal. And Chaunacidae live exactly where this fish was found. They are continental slope animals, recorded down to at least two thousand four hundred and sixty metres. The nickname landed in deeper water than the family name it was attached to.

Be careful with what that does and does not mean. It does not make this fish a coffinfish. This channel is not naming it either. What it means is narrower and stranger. The sentence disagrees with itself, and one half of it is in the right ocean. Third, the place. The Northeast Canyons and Seamounts Marine National Monument is real, and it is spectacular. It is also about a hundred and thirty miles east-southeast of Cape Cod. Thirty-nine point seven north, sixty-seven point one seven west.

The wrong country 20:16

The dive was at forty-one point nine eight north, sixty-five point three eight west. That is a couple of hundred kilometres away, in another country's water, under another country's protection scheme. The likeliest source of that one is not carelessness. It is the name of the expedition. Deep Connections twenty nineteen. Exploring Atlantic Canyons and Seamounts of the United States and Canada. Read that quickly and you get canyons, seamounts, and a monument.

The four versions, in order 20:47

Line the four versions up in order and the shape is obvious. A question mark, on the ship. Species undetermined, in the dive report. Species undetermined, in the video caption. A family, a nickname and a monument, seven years later. The uncertainty survives every step that has access to the dive, and dies at the first step that does not. That is not a coincidence, and it is not dishonesty. It is a property of the chain.

Why the hedge is paid last 21:19

Because a question mark is very hard to publish. Species undetermined is not a headline. Somewhere along the line the hedge has to be paid for, and it gets paid at the step with the least to lose and the most people watching. Notice which way the two curves run. Going down the chain, the audience gets larger at every step. The metadata entry is read by specialists. The dive page, by enthusiasts. The news story reached more people in a day than the other three have in seven years.

And going down that same chain, direct access to the dive gets smaller at every step. Maximum readership arrives exactly where contact with the primary record is thinnest. That is the whole mechanism, and it is not anybody's fault. The uncomfortable part is that the fix is not being more sceptical. Every version of this story links back to NOAA. The sourcing was never the problem. The problem is that the primary record and the popular record are two different documents.

Checking it yourself 22:25

Which is worth ending on, because the primary record here is not hidden. It is a public catalogue entry, with an identifier, a timestamp, a depth to one decimal place, and a coordinate. It took minutes to pull for this episode. If you ever want to do this yourself, it is three steps, and none of them need an account or a subscription.

One. Find the expedition code and the dive number. They are usually in the caption or the credit line of whatever image you are looking at. Here they were EX nineteen-oh-five leg two, and dive four. Two. Open the dive summary that the agency published on the day. That is where the coordinate and the depth range live, and it is written before anyone knows the story will travel.

Three. Compare. Look for a name, a place or a number in the popular version that is not in the day-of record. That is not automatically wrong. But it came from somewhere else, and it is worth knowing where. And one more step, which I only learned while making this. Play the audio. These videos carry the control room talking, in the public domain, untranscribed. The best number in this entire story sat in an audio track for seven years.

Undetermined is a finding 23:47

What would actually close this one is not complicated either. Another sighting on that slope, with a suction sampler running and a jar at the end of it. And that is not fantasy. This is a monitored, protected piece of seafloor with a management case attached to it. Vehicles go back. The animal is presumably still there. So the rule this channel keeps arriving at, from a different direction each time. When a record says undetermined, that is not a gap in the record. That is the finding. Somebody looked hard, reached the edge of the method, and wrote down where the edge was.

The complete honest description 24:27

Here is the honest description of this animal, complete. A juvenile fish, a few millimetres long, on soft sediment. One thousand three hundred and fifty-three metres down, in the Northeast Channel. First of September, two thousand nineteen. Species undetermined. And think about the other fish for a second. The dogshark, same dive, an hour and a half earlier, at one thousand four hundred and eighty-five metres. It also arrived with a question mark on it.

Its question mark came off, because it was a metre long and the characters were showing. Same water, same lights, same afternoon, same person typing. The only variable that changed was how much of the answer the animal was carrying on its outside. That is not a failure. It is a fish that got seen once, by accident, by somebody with good eyes. And the seeing was written down properly enough that you can still audit it seven years on.

It is also still open. The video is public. The metadata is public. If somebody films that animal again on that slope, with a manipulator ready, the question mark comes off. Until then it stays on, and it should.

They look so big on our screen 25:46

And the last word should go to the room. A minute after the estimate, with the vehicle pulling back off the bottom, somebody says this, to nobody in particular. Quote. Sometimes when we zoom in, I forget how small things actually are. They look so big on our screen. End quote.

Next: the fish that holds its breath 26:07

One thing does get settled, though. That nickname. Sea toads are real, they live down there, and they do something no other fish on Earth is known to do. They inflate their whole body with seawater and hold their breath, for up to four minutes, on the bottom. Next episode. The fish that stopped breathing on camera, and the scientists who worked out why.

Description and sources

The person annotating this dive typed four words and a question mark: TINY TOADFISH? NOAA's dive summary that evening called it a tiny juvenile fish, species undetermined, and so did the video caption. Three official records, three times the uncertainty kept. Seven years later to the day, the coverage gave it a family whose depth range stops five times shallower, a nickname belonging to a different family, and a location in another country's water. The size everybody repeats is the one part that holds -- and it is the one thing NOAA never wrote down.

PRINT-READY, FROM THIS CHANNEL

Read This Dive, Vol. 01 — the method, and five dives read with it

https://therepository.gumroad.com/l/read-this-dive

PRIMARY SOURCES

NOAA Ocean Exploration, Dive 04: Northeast Channel (EX1905 Leg 2), 1 Sept 2019: https://oceanexplorer.noaa.gov/ex1905-dailyupdates-sept1/ -- the day-of summary: 41.98381 N, -65.37541, 975-1,494 m, and 'A tiny juvenile fish, species undetermined, was spotted by the keen eyes of our videographer.'
NOAA NCEI geoportal, granule EX1905L2_DIVE04_20190901T182459Z_ROVHD -- the primary annotation: keywords FISH and TINY TOADFISH?, 2019-09-01T18:24:59Z, 1352.9-1353.0 m. Source of every depth and timestamp here.
NOAA Ocean Exploration, Deep Connections 2019: https://oceanexplorer.noaa.gov/expedition/ex1905/
NOAA Ocean Exploration, ROV Deep Discoverer: https://oceanexplorer.noaa.gov/technology/subs-deep-discoverer/ -- the two lasers mounted 10 cm apart.
NOAA Ocean Exploration, 'NOAA Scientists Virtually Discover New Species of Comb Jelly': https://oceanexplorer.noaa.gov/news/ctenophore/ -- source of Allen Collins's remark that the cameras 'measure structures less than a millimeter'.
Ford, M.D., Bezio, N. & Collins, A.G. (2020). Duobrachium sparksae. Plankton and Benthos Research 15(4): 296-305 -- the description whose type material is the video itself.
FAO Species Catalogue, Batrachoididae: https://www.fao.org/4/y4161e/y4161e28.pdf -- the family's shallow depth range.
Chaunacidae (sea toads), continental slope to at least 2,460 m -- U.S. Fish & Wildlife Service: https://www.fws.gov/species/sea-toads-chaunacidae
Ko, H.-L. et al. (2013). Evaluating the accuracy of morphological identification of larval fishes by applying DNA barcoding. PLOS ONE 8(1): e53451 -- why juvenile fish key only to family.
Northeast Canyons and Seamounts Marine National Monument proclamation, Federal Register, 21 Sept 2016: https://www.federalregister.gov/documents/2016/09/21/2016-22921/northeast-canyons-and-seamounts-marine-national-monument
Fisheries and Oceans Canada, Northeast Channel Marine Refuge: https://www.dfo-mpo.gc.ca/oceans/oecm-amcepz/refuges/northeast-nordest-eng.html
Popular Science, 'NOAA scientists spot outrageously tiny fish', 1 Sept 2026: https://www.popsci.com/environment/noaa-scientists-spot-outrageously-tiny-fish/
Long, N.P. & Farina, S.C. (2019). Enormous gill chambers of deep-sea coffinfishes. Journal of Fish Biology 95(2): 502-509. doi:10.1111/jfb.14003 -- the source for the next episode.

Abyss Report reads the deep ocean through the record that resolved it -- instrument, date, position and the primary sources above.

About the material. Every moving frame is NOAA Ocean Exploration video from EX1905L2 dive 04, public domain, credited on screen with dive, depth and position. Three cuts drop the narration and play the original control room audio; each of those windows was re-transcribed in isolation twice and used only where both passes matched. Every diagram is drawn in code from measured values. Nothing here is generated.

#DeepSea #NOAA #OceanExploration #MarineBiology #Taxonomy