Twenty thousand years ago, at the height of the last Ice Age, so much of the world’s water was locked up in ice sheets that sea levels stood roughly 120 meters lower than they do today. Britain was joined to Europe. Australia was joined to New Guinea. A broad plain stretched across what is now the Persian Gulf, and the islands of Indonesia were the hills of a single landmass.

Then the ice melted, and the sea came back. Over the next 13,000 years or so, it rose by more than a hundred meters, sometimes slowly, sometimes in sudden surges. During one episode about 14,600 years ago, known to geologists as Meltwater Pulse 1A, global sea level rose by roughly 15 to 20 meters in a few centuries.

For anyone interested in the deep human past, this is one of the most important facts in all of prehistory, and one of the least discussed. People have always favored coasts, river mouths, and estuaries, where food is plentiful and travel is easy. The coastlines of the Ice Age world were exactly where we would expect to find many of its people. And almost all of them are now underwater.

The Problem of the Missing Coast

As we noted on our Lost Civilizations page, this creates a genuine blind spot in the archaeological record. Archaeologists excavate on land. The land that was coastline 15,000 years ago is now tens of meters beneath the waves, buried in sediment, and expensive and difficult to search.

The absence of evidence from those drowned plains is therefore not evidence of absence. It tells us mainly that we have barely looked. What happens when we do look is the subject of some remarkable recent work.

A Lost Land Under the North Sea

The best studied drowned landscape in the world is Doggerland, the land that once connected Britain to mainland Europe. Fishing trawlers have been dragging up mammoth bones, antlers, and occasional stone tools from the North Sea for well over a century.

In 2026, a team led by Prof. Robin Allaby of the University of Warwick, working with Prof. Vincent Gaffney of the University of Bradford, published a study in the Proceedings of the National Academy of Sciences that reconstructed this lost land in unprecedented detail. Instead of looking for artifacts, they analyzed ancient DNA preserved in sediment: 252 samples from 41 seabed cores taken along the course of a prehistoric river.

The results were striking. Oak, elm, and hazel were already growing in southern Doggerland more than 16,000 years ago, earlier than pollen records from Britain had suggested. Lime trees appeared about 2,000 years earlier than previously recorded on the British mainland. The team even found DNA from Pterocarya, a walnut relative thought to have vanished from northwestern Europe some 400,000 years ago.

In other words, Doggerland was not a frozen wasteland but a wooded, habitable country. Allaby called it “the best evidence that Doggerland’s wooded environment could have supported early Mesolithic communities.” Parts of it survived the great Storegga tsunami of about 8,150 years ago, and some areas may have stayed above water until roughly 7,000 years ago. The study also suggests that the scarcity of early Mesolithic sites on mainland Britain may simply reflect the fact that many people were living on land that no longer exists.

A Time Capsule in Danish Waters

In August 2025, underwater archaeologists in Denmark reported a Stone Age settlement about 8 meters (25 feet) beneath the surface of the Bay of Aarhus, dating to around 8,500 years ago. Divers working with Moesgaard Museum excavated about 40 square meters of seabed and recovered animal bones, stone tools, arrowheads, a seal tooth, and pieces of worked wood.

Wood rarely survives on land sites of this age. Underwater, sealed in oxygen-free sediment, it can last for millennia. As the museum’s Peter Moe Astrup put it, “It’s like a time capsule.” The team hopes future seasons will turn up harpoons, fishhooks, or the remains of fish traps, the kind of everyday coastal technology that is almost entirely missing from the dry-land record.

Walls Beneath the Adriatic

Off the coast of Croatia, a series of finds is showing that Neolithic communities built substantial structures on coastlines that have since disappeared.

In late 2025, a team led by Mate Parica of the University of Zadar reported a submerged site off the island of Pašman, about 5 meters underwater. It includes a dry stone wall and rows of densely set wooden posts, part of a structure running more than 300 meters along the seabed. Radiocarbon dating places its construction at about 4700 BC, when the Adriatic stood more than five meters lower than it does today. The researchers believe it was a settlement on a landform that was separate from the coast even then, and suggest these communities were oriented toward the sea and possibly involved in early maritime trade. Why they chose that spot, they admit, “remains a mystery.”

Nearby, near the islet of Ričul, a Bronze Age harbor settlement about 3,500 years old lies just two to three meters down, close enough to the shore that visitors can swim out to it. Excavations there have produced thousands of artifacts, along with a stone embankment about 300 meters long and an olive stone believed to be the oldest yet found on the eastern Adriatic coast.

These Croatian sites are younger than the end of the Ice Age, and they were drowned by the slower sea level rise of later millennia. But they make an important point: even modest changes in sea level, a few meters, have erased significant settlements from view.

The Oldest Drowned Village

Off the Carmel coast of Israel lies Atlit-Yam, a Neolithic village about 9,000 years old, now 8 to 12 meters underwater. It is the best preserved of more than twenty submerged prehistoric sites known along that stretch of coast. Its finds include stone houses, water wells, burials, and a semicircle of large standing stones. Atlit-Yam shows that complete, organized villages from the early farming era can survive on the seabed, if anyone goes looking.

What About Dwarka and Yonaguni?

Any discussion of drowned worlds raises the more famous claims: the underwater formations near Yonaguni in Japan, and submerged structures in the Gulf of Khambhat and off Dwarka in India. These sites have devoted advocates, and they deserve investigation. But unlike Doggerland, Aarhus, Pašman, and Atlit-Yam, none has yet produced the kind of securely dated, excavated, peer-reviewed evidence of human construction that would settle the question. Many geologists regard Yonaguni’s terraces as natural rock formations. We keep these sites on our list of open questions, not established facts. For a closer look at the Indian sites, see India’s Drowned Cities.

Why This Matters

The mainstream picture of prehistory is built almost entirely from sites that survived on dry land. Those surviving sites were often inland locations, uplands, and caves. The world’s Ice Age coasts, which were likely among the most densely populated places on Earth, have contributed almost nothing to that picture, because they were drowned before anyone thought to look.

The drowning of those coasts is not proof of a lost advanced civilization, and we would caution against leaping there. What it does mean is that our map of the human past has a large blank region at its edges, and that the blank is a matter of where we have looked, not what was there. Every time underwater archaeology gets a serious chance, as at Doggerland or Aarhus, it finds people. The honest expectation is that there is a great deal more to find.

Sources