In the summer of 1997, an array of underwater microphones strung across the equatorial Pacific picked up something that should not have been possible. It was a sound of enormous power, ultra-low in frequency, rising and falling over the course of about a minute, and it registered on sensors more than five thousand kilometres apart, which meant it had crossed a substantial fraction of an ocean and still arrived loud. When the researchers at the National Oceanic and Atmospheric Administration sped the recording up sixteen times so that human ears could hear it at all, it came out as a fat, wet, rising note, and somebody called it the Bloop. The name stuck, the spectrogram circulated, and the world was handed a genuine scientific mystery with an irresistible shape: the loudest thing anyone had ever heard underwater, and no known animal on Earth capable of making it.
What followed was two decades of speculation that ran from the reasonable to the gloriously unhinged, taking in surviving megalodons, undiscovered colossal cephalopods, secret submarines, and, inevitably, the observation that the source region sat within a few thousand kilometres of the fictional coordinates H. P. Lovecraft assigned to the sunken city where Cthulhu sleeps. The actual answer, when it arrived, was ice. But the interesting part is not the answer; it is the shape of the mistake, because the Bloop was never unlike anything in nature. It was unlike anything on a very short list, and the list was short because nobody had been listening to the deep ocean for very long. Anomalies of this kind are not statements about the world at all. They are statements about the completeness of a catalogue, which puts the whole genre closer to the beliefs a culture assembles when data is scarce than to zoology, and firmly inside the atlas of blank spaces waiting to be filled in.
The Sound They Called the Bloop
The instruments that caught the Bloop have a history worth knowing, because it explains why anyone was listening at all. During the Cold War the United States Navy built an enormous network of seafloor hydrophones, the Sound Surveillance System, designed to track Soviet submarines by their acoustic signatures across entire ocean basins. It worked, expensively and secretly, for decades. When the Cold War wound down, portions of that capability were made available to civilian science, and oceanographers suddenly found themselves in possession of the most sensitive listening apparatus ever built, pointed at an environment nobody had ever systematically monitored. NOAA deployed its own autonomous arrays in the equatorial Pacific in the early 1990s, primarily to listen for undersea volcanoes and earthquakes along the mid-ocean ridges.
It is worth appreciating how recent all of this is. Humans have been going to sea for many thousands of years and have been able to listen to the deep ocean in any systematic, continuous, wide-area way for roughly three decades, which in the history of any observational science is the first morning. Astronomy had catalogues of stars before it had telescopes. Ornithology had centuries of field notes before anyone recorded a birdsong. Deep-ocean acoustics went from essentially nothing to a global sensor network almost overnight, skipping the long accumulation of ordinary observations that other disciplines built their baselines on, and it did so in an environment nobody could visit to check. That sequence guarantees a particular kind of early result, which is a large pile of signals that nothing in the record resembles.
So the Bloop was detected by an instrument built to find submarines, repurposed to find volcanoes, and pointed at an ocean whose ordinary acoustic behavior had essentially never been characterized. That last clause is the one everybody skipped. A network of that sensitivity, switched on over a previously unlistened environment, was guaranteed to produce a stream of signals that matched nothing in the files, and it did: researchers logged and nicknamed a whole family of them across the 1990s. The Bloop was simply the loudest. Instruments developed for national security and later turned to science have a long track record of revealing more than their designers intended, a pattern running throughout the history of surveillance and detection technology and through the archives of Cold War systems whose real capabilities stayed hidden for decades. The ocean did not start making strange noises in 1997. We started listening.
Why It Sounded Alive
Now the detail that quietly did more work than any other in creating the legend: you have almost certainly never heard the Bloop. What circulates, and what everybody reacts to, is the recording played at sixteen times its actual speed, because at its true frequency the sound sits at the very bottom edge of human hearing or below it, and is close to inaudible without processing. Slowed back to reality, the Bloop is not a bloop at all. It is a minute-long, subsonic, rolling rumble, more like distant thunder rolling through a mountain range than any noise a creature makes.
That sixteen-fold speed-up is not a neutral act of translation, and it is the single most important artifact in the entire case. Compressing a signal in time shifts everything about how a human nervous system categorizes it, packing a slow geological groan into the duration and pitch range where our auditory system expects animal vocalizations, and our auditory system is exquisitely, aggressively tuned to detect animal vocalizations, because for most of our evolutionary history the cost of missing one was severe. Feed it a compressed rumble and it will hand back a creature, confidently, with no sense that it has done anything but hear. Even the NOAA scientists noted early on that the audio profile resembled that of a living thing while insisting the required power output was far beyond any animal, which is a very precise way of saying the pattern-matching had fired on a signal that did not fit the physics. This is the same construction problem that runs through all sensory processing, where the nervous system does not deliver raw data but a best interpretation, the terrain mapped by the science of how nervous systems build perception. The monster was in the playback speed.
The Ocean Is Opaque
Step back and ask why the sea, specifically, generates so many anomalies, and the answer is a physical fact with enormous downstream consequences: the ocean is opaque to almost everything we use to perceive the world. Light attenuates within tens of metres, so the overwhelming majority of the ocean’s volume sits in permanent darkness. Radio waves are absorbed almost immediately, which is why submarines cannot simply use radio and why satellites, which have mapped the surface of Mars in fine detail, cannot see through a hundred metres of seawater. What does propagate, magnificently, is sound.
The practical consequence is a degree of ignorance that people consistently underestimate. More of the surface of the moon has been mapped at fine resolution than the floor of our own ocean, and until quite recently the great majority of the seabed had never been directly measured at all, its depth merely inferred from satellite gravity readings at kilometre-scale resolution. Roughly seventy percent of the planet is covered by water averaging around four kilometres deep, and essentially none of that volume is under continuous observation. Any claim about what is or is not normal down there is being made from a position of near-total sampling poverty, and the honest description of the deep sea is not that it is mysterious but that it is unobserved, which is a different and more tractable condition.
Sound in the sea travels enormous distances, and in a particular layer of the deep ocean it does something remarkable: variations in temperature and pressure create a channel that refracts sound back toward its centre rather than letting it escape upward or downward, so low-frequency signals entering that channel can travel for thousands of kilometres with very little loss. This is why a cracking iceberg near Antarctica can be recorded near the equator, and why the entire Cold War submarine-detection strategy was built around it. The consequence for anomalies is direct: the ocean is an environment where our only good sensory channel delivers signals from sources potentially thousands of kilometres away, in the dark, with no possibility of going to look. The sea remains genuinely under-observed in ways that shape everything from the strategic contest over water and maritime access to the engineering of the cables and pipelines documented in the history of great infrastructure. We cannot see the ocean. We can only listen, from very far away.
Unidentified Means Not in the Library
Here is the mechanism at the centre of every ocean anomaly, and it generalizes far past the sea. When a signal is labelled unidentified, that label describes a failed lookup, not a property of the signal. Some sound arrived, an analyst compared it against a reference collection of known sources, no match was found, and the signal went into a bin marked unknown. Everything depends, therefore, on the size and quality of the reference collection, and almost nothing depends on how strange the signal actually is.
The asymmetry between the two readings has real consequences for how a story travels. A weak claim about an incomplete file is boring and generates no coverage; a strong claim about an impossible sound generates decades of documentaries. So the label itself gets upgraded in transmission, from a technical bin marked pending to a public assertion that science cannot explain this, and nobody has to lie for that to happen. The scientists who logged the Bloop never claimed it was inexplicable in principle; they said it did not match their references and proposed candidate explanations almost immediately. The gap between what was said and what was heard is where the legend lived.
This is worth sitting with, because it inverts the usual intuition. We tend to read unidentified as a strong claim, meaning the thing is unlike anything natural. It is in fact a weak claim, meaning the thing is unlike anything currently on file, and the two are wildly different when the file is thin. In 1997 the file for deep-ocean sound was close to empty in any systematic sense. Nobody had spent decades characterizing what a large iceberg fracturing at the Antarctic margin sounds like from five thousand kilometres away, because until very recently there was no instrument to hear it and no reason to look. The bin marked unknown was therefore enormous by construction, and being in it carried almost no information. Classification against an incomplete reference is a chronic hazard in every detection problem, from training a biological detector to flag a genuine signal to the counterintelligence work of distinguishing an ordinary transmission from a compromised one, as in the long-concealed case of a cipher company that was never what it seemed. Unidentified is a fact about the library. It is not a fact about the ocean.
How the Bloop Was Identified
Watch how the Bloop actually got solved, because the process is the argument. Nobody found a creature. Nobody built a better theory in an office. What happened is that NOAA kept deploying hydrophones, and kept deploying them further south, into the Scotia Sea and toward the Antarctic margin, in order to study seafloor volcanoes and earthquakes at higher latitudes. Once instruments were sitting near the ice, they recorded enormous numbers of icequakes: the acoustic signature of large icebergs cracking, fracturing, and grinding against the seafloor. And the spectrograms looked like the Bloop.
The identification was thorough. NOAA’s account of the case notes that the broad-spectrum sounds from 1997 are consistent with icequakes from large icebergs cracking and fracturing, that hydrophones in the Scotia Sea recorded numerous icequakes with spectrograms very similar to the Bloop, and that based on the arrival direction the responsible icebergs were most likely between the Bransfield Strait and the Ross Sea, or perhaps at Cape Adare, a known source of cryogenic signals, as documented in the NOAA acoustics program’s page on icequakes. The detail that makes the whole thing land is what came next: the icequake signature became a tool, and NOAA used it to acoustically track the iceberg A53a as it disintegrated near South Georgia in early 2008. The sound that had been a monster became an instrument for watching ice break up. By 2012 a NOAA seismologist could state that the hydrophone network picks up tens of thousands of Bloop-like sounds every year, which is the whole lesson compressed into one number. It took sustained instrumentation of a hostile environment, the unglamorous grinding work behind most real discoveries in the catalogue of great technical undertakings, and it turned on understanding the physical behavior of materials under stress, the same geological literacy that underpins the science of where the earth’s useful materials come from. The signal never changed. The library did.
The Rest of the Chorus
The Bloop was not alone, and the fate of its siblings makes the pattern unmistakable. Across the 1990s, NOAA researchers logged and nicknamed a whole family of unexplained sounds: Upsweep, first noted in 1991 and recurring seasonally; Train, in March 1997; Slow Down, in May 1997; Whistle, in July 1997; and Julia, in March 1999, which lasted nearly three minutes and was loud enough to register across the entire array, and which got its name because it sounded unnervingly like a muffled human voice. Each one was genuinely strange, each generated its own small mythology, and each was, for a while, unidentified in the strict sense of not matching anything in the files.
The naming convention deserves a moment, because it quietly shaped the reception of all of them. Working scientists give informal nicknames to unclassified signals for the entirely practical reason that a catchy label is easier to refer to in a meeting than a timestamp and a bearing. But a nickname is also a small act of characterization, and calling a signal Julia because it resembles a human voice, or the Bloop because it sounds like a cartoon bubble, imports a suggestion of agency into what is really a filing decision. Nobody at NOAA intended the names as claims. They were shorthand, and shorthand is exactly the kind of thing that survives the journey out of a laboratory while the caveats do not.
They have been falling to the same explanation, one after another. Slow Down, Train, and Julia are now generally attributed to icebergs, specifically to the enormous grinding and fracturing that occurs when a berg runs aground on the seafloor or scrapes across it, and their arrival directions point consistently toward the same stretch of the Antarctic margin. Upsweep, which behaves seasonally and has been detected for decades, is associated with volcanic activity at a seafloor site in the South Pacific. What is striking is the uniformity of the resolutions: every named ocean mystery that has been solved has resolved toward ice, geology, or biology, and not one has resolved toward anything exotic. That is precisely what you expect if the anomalies were catalogue gaps rather than genuine novelties, and it mirrors how apparent mysteries in animal behavior keep resolving into ordinary and beautiful mechanisms, as with the magnetic sense that guides migrating birds and the accumulating record of what animals actually know and transmit. Six mysteries, one direction of travel.
The Biotwang and the Machine
The mechanism is not a historical curiosity; it ran again recently, in public, and the case is close to a laboratory demonstration. In October 2014, an autonomous underwater glider operated by Oregon State University, drifting near the Mariana Archipelago, recorded a call nobody could place: a low sonorous groan followed by a squeaky, metallic, almost synthetic flourish. Researchers named it the Western Pacific Biotwang. It was clearly biological in character and presumably came from a baleen whale, but it matched no known whale call in the reference collections, and without a visual sighting to pair with a recording, the species could not be assigned. It went into the unidentified bin and stayed there for a decade, accumulating the usual speculation.
The resolution came from closing exactly the gap that had created the problem. In 2018, NOAA scientists ran a combined visual and acoustic survey in the Mariana Archipelago, and on ten occasions they had eyes on groups of Bryde’s whales; on nine of those occasions, the biotwang was recorded simultaneously. Then a team applied a machine-learning classifier trained on more than two hundred thousand hours of audio to the long-term recordings, allowing them to trace the call’s distribution and seasonality across years of data no human could listen through, and published the result in 2024, reported in the study in Frontiers in Marine Science identifying the source. The biotwang belongs to Bryde’s whales, apparently to a specific western North Pacific population, and its seasonal pattern tracks their migration, with detections spiking when a warm El Nino year drew more animals into the area. A decade of mystery ended not with a discovery of something new but with the pairing of a sound to a sighting, the kind of ground-truth correspondence that underlies all honest claims about what other creatures do and experience, from the difficult science of animal sentience to the effort to read signals directly from nervous systems. The whale was always there. The entry was missing.
The Phantom Bottom
If you want the cleanest historical proof that the catalogue is the whole story, it comes from sonar rather than hydrophones, and it is one of the great episodes in the history of oceanography. During the Second World War, sonar operators hunting submarines began reporting something deeply alarming: a solid reflecting layer, a seafloor return, at a few hundred metres depth in water known to be thousands of metres deep. It was a false bottom that had no business existing. Worse, it moved. The phantom seafloor rose toward the surface each night and sank back down at dawn, which is not behavior anyone expects from geology, and for a while it was a genuine operational problem for a navy trying to distinguish real returns from whatever this was.
Notice the structure of that discovery, because it is identical to the Bloop’s and it happened fifty years earlier. A sensing technology built for a military purpose was pointed at an environment nobody had characterized; it immediately returned a signal that matched nothing in the reference material; the signal was strange enough to look like an instrument fault or an impossibility; and it turned out to be an enormous, entirely natural phenomenon that had been going on continuously and unobserved for as long as the relevant animals had existed. The anomaly was not evidence of something exotic. It was evidence of a brand-new instrument meeting an old, unexamined world.
The explanation, established after the war, is one of the most extraordinary facts about the planet. The reflecting layer is biology: dense aggregations of fish, squid, siphonophores, and zooplankton, many with gas-filled structures that bounce sound beautifully, forming a living stratum across the world ocean. It rises toward the surface at dusk to feed and descends at first light to hide from predators, and this diel vertical migration is the largest coordinated movement of animal biomass on Earth, occurring every single day, entirely unsuspected until an instrument built for war accidentally ran into it. The most alarming sonar anomaly in naval history turned out to be an enormous, ordinary, previously uncatalogued biological phenomenon, and the resolution again came from filling in the library rather than from revising physics, a lesson that recurs wherever biology turns out to be doing something cleverer than expected, as in the natural world’s elaborate strategies of concealment, and one that grows more relevant as autonomous platforms proliferate alongside the drones and robotic systems now surveying everywhere. The false bottom was real. It was just alive.
Unidentified Submerged Objects
Which brings us to the sea’s other anomaly tradition, the unidentified submerged object, generally shortened to USO. The category covers reports of objects moving underwater at implausible speeds, sonar contacts that appear and vanish, and claims of craft crossing between air and water, and it has a small canon of incidents, most famously the 1967 event at Shag Harbour in Nova Scotia, where multiple witnesses reported lights descending into the water and Canadian authorities conducted a genuine search that found nothing. The category has been revived in recent years by the wider interest in unidentified aerial phenomena and by claims that certain reported objects were transmedium, capable of operating in air and sea alike.
The honest analysis has to start with how bad sonar is at answering the question people want it to answer. A sonar display is not a picture; it is a reconstruction assembled from timing and intensity of returns, and it is riddled with well-understood ways of producing convincing objects that are not there. Thermoclines refract and reflect sound, generating false targets and shadowing real ones. Wakes, bubble clouds, whales, and fish schools all return signal. Multipath propagation delivers the same echo twice by different routes, which can render a single object as two, or place it where it is not. Speed estimates depend on correctly associating successive returns with the same target, and misassociation produces apparent velocities that no physical object achieved. None of this means every report has been explained, and it is fair to say plainly that some remain open; but the base rate of instrument artifacts in this environment is extremely high, which is exactly the caution now applied systematically to the modern investigation of unidentified aerial phenomena and which anyone working with active sensing understands, as in the engineering of directed-energy and radar systems. The most famous underwater sonar mystery of all was a false bottom made of fish.
The Baltic Sea anomaly of 2011 illustrates the whole dynamic in miniature. A commercial treasure-hunting team released a side-scan sonar image of a large circular structure on the seafloor between Sweden and Finland, and the internet promptly identified it as a crashed craft, complete with an apparent access ramp. Geologists who examined recovered samples identified the material as rock deposited by glaciers, which is exactly what the northern Baltic seafloor is made of, and the striking circular outline turned out to owe a great deal to the way side-scan imagery is processed and rendered. A glacial deposit, imaged at low resolution and presented as a picture, became a spacecraft.
Ocean Anomalies in 2026
The interesting development is that the catalogue is now being filled at machine speed, and this is changing the anomaly landscape faster than anything in the previous century. Autonomous gliders and floats drift for months collecting acoustic and oceanographic data without a ship in sight. Permanent cabled observatories stream continuously from the seafloor. Machine-learning classifiers, of the kind that cracked the biotwang, can process years of continuous audio that no human team could ever audit, and they are being used to build genuine reference libraries of ocean sound. Environmental DNA sampling can establish which species are present in a body of water from a jug of seawater, without seeing or catching anything. And an international effort has been working to map the entire ocean floor at usable resolution by the end of this decade, against a starting point where the great majority of the seabed had never been directly measured at all.
There is a specific reason machine classification matters so much here, and it is not that the algorithms are clever. It is that the bottleneck in ocean acoustics was never analysis but attention: hydrophone networks generate continuous audio for years on end, and a human analyst can meaningfully review a vanishing fraction of it, which means the historical reference library was built from whatever fragments people happened to listen to. Automated classification changes the sampling fraction from a sliver to nearly all of it, which is precisely the intervention that fixes a catalogue-coverage problem. The biotwang did not yield to a new idea. It yielded to somebody finally being able to listen to everything.
The result is that the unidentified bin is draining, steadily and in one direction. What is worth stressing is that this is not a story about the ocean becoming boring, because the opposite is happening: the same instrumentation keeps turning up genuinely new species, unexpected behaviors, previously unknown vent systems, and calls nobody had documented. The unknown in the ocean is vast and real and being actively reduced. It simply is not the kind of unknown the anomaly tradition wants, since every increment of it resolves into biology, geology, or ice, and the governance questions this raises are correspondingly practical rather than exotic, concerning who may map, mine, cable, and fish the deep sea, of a piece with wider arguments about how to govern commons that no one owns. The mystery is shrinking. The discoveries are not.
The Catalogue Was the Monster
Strip the ocean anomalies down to their mechanism and the lesson is one sentence with unusual portability: unidentified is a claim about a reference library, not about the world, and the size of the unidentified bin measures how recently you started paying attention rather than how strange the universe is. The Bloop was never impossible. It was a large iceberg fracturing in the Antarctic, doing a thing that happens tens of thousands of times a year, recorded by a network that had been listening for a handful of years to an ocean that had been making the sound for as long as there has been ice. The gap was not in nature. It was in the files.
The test this yields is simple enough to apply to any anomaly claim, in the ocean or out of it. Before accepting that something is unlike anything in nature, ask how long anyone has been systematically observing the relevant corner of nature, how much of it has actually been sampled, and how large the reference collection was at the moment the label was applied. If the honest answers are a few decades, a tiny fraction, and almost nothing, then unidentified is doing no work at all.
And the direction of resolution is the part worth carrying away, because it has been perfectly consistent. Every named ocean mystery that has been solved resolved toward the ordinary and the magnificent at once: ice cracking off a continent, a volcano venting on a ridge, a population of whales with a call nobody had catalogued, a living layer of animals so vast it reads as a seafloor. The sound that made people think of a monster is now used to track icebergs breaking apart, which is about as complete a reversal as an anomaly can undergo, and it is the reason this whole family of cases earns its place among the resolved entries in the catalogue of Fortean phenomena rather than the open ones. Something enormous really was down there in 1997, loud enough to cross an ocean. It was the Antarctic ice sheet, coming apart, and nobody had written down what that sounds like.