Lost Worlds · Near Adilcevaz, Lake Van, eastern Turkey

Lake Van Underwater 'Castle'

A viral 'lost castle' beneath Turkey's greatest lake — and a lesson in how headlines outrun evidence

Mainstream: Probably largely medieval fortification, possibly reusing Urartian (c. 9th-6th century BC) blocksAlternative: Reported worldwide in 2017 as a 3,000-year-old Urartian castle — an 'Atlantis of Lake Van'38.79°, 42.73°

Lost Worlds

At a glance

In November 2017, news outlets worldwide announced that a 3,000-year-old castle of the Urartian kingdom had been discovered beneath Lake Van, Turkey's vast alkaline inland sea. The find was real: a dive team led by underwater photographer Tahsin Ceylan, working with Van Yuzuncu Yil University, had filmed impressive stretches of cut-stone walls outside the harbour of Adilcevaz on the lake's north shore. But the dating dispute began almost immediately. Archaeologists noted that much of the masonry looks medieval, that surveys in the 1950s had already described a castle at Adilcevaz reusing Urartian blocks, and that Lake Van's dramatically fluctuating water level has drowned shoreline structures of many eras. The 'underwater Atlantis' framing said more about the media than the masonry.

The mainstream view

What archaeology says

Lake Van sits at the heart of the ancient kingdom of Urartu (roughly 9th to 6th centuries BC), whose capital Tushpa stood on the Van citadel rock, so an Urartian find there is inherently plausible — and the lake genuinely conceals archaeology, because it has no outlet and its level has swung by many metres over the centuries, drowning harbours and shoreline buildings. Ceylan's team, which had previously documented the lake's microbialite towers, a sunken Russian ship and other features, filmed walls running for hundreds of metres at shallow depth near Adilcevaz, with squared blocks and what they described as a castle or fortification.

The caution came from archaeologists who pointed out that the structure appears in earlier records: a 1958 paper in Anatolian Studies by Charles Allen Burney and G. R. J. Lawson describes a medieval castle at Adilcevaz on the north shore of Lake Van, built partly of reused Urartian masonry. Specialists who reviewed the 2017 footage, including commentators quoted by Live Science, judged that much of the visible construction — mortared rubble and roughly coursed stone — is characteristically medieval, perhaps with genuinely Urartian ashlar blocks incorporated, exactly as Burney and Lawson had described on land.

The mainstream conclusion is therefore not that the find is fake, but that it is probably a known medieval fortification, partly recycled from Urartian material, drowned by a later rise in lake level — and still worth proper investigation. No systematic underwater excavation has yet been published, so the balance between Urartian and medieval elements remains formally unresolved.

Key evidence cited

Burney and Lawson's 1958 Anatolian Studies survey describes a medieval castle at Adilcevaz reusing Urartian blocks — likely the same structure later filmed underwater
Specialists reviewing the 2017 footage identified mortared, roughly coursed masonry characteristic of medieval rather than Urartian construction
Lake Van's level is known to fluctuate by many metres over historical time, readily explaining the drowning of shoreline buildings of various eras
Adilcevaz has confirmed Urartian remains on land, including the fortress of Kef Kalesi, making reused Urartian ashlar in later walls unsurprising
The dive team included no archaeologist, and no peer-reviewed excavation report has yet substantiated an Urartian date for the submerged walls
Urartian sites around the lake basin follow known patterns of citadel construction that differ from the layout filmed underwater

The alternative view

What the skeptics propose

The 2017 coverage is a case study in how a modest discovery becomes an Atlantis. Headlines proclaimed a lost 3,000-year-old castle, some outlets speaking of a vanished civilisation beneath the waters; social media added imaginary details, and the story recirculates every year or two with fresh superlatives. The Urartu angle gave it mythic depth: Urartu is linked to the biblical Ararat, and the region already carries flood traditions — Mount Ararat, the supposed resting place of Noah's Ark, overlooks the wider region — so a drowned Urartian city slotted neatly into an ancient-mysteries narrative.

Fringe writers went further, connecting the 'sunken castle' to claims of a lost Armenian or pre-flood civilisation beneath the lake, to the local legend of the Lake Van monster, and to the idea that official archaeology was downplaying a revolutionary find. The fact that the discoverers were divers rather than archaeologists, and that Turkish officials initially promoted the Urartian date enthusiastically, gave the story an appealing outsiders-versus-experts shape.

Yet the episode also shows the value of the alternative impulse when it is disciplined. Ceylan's team were not cranks: they were documenting a genuinely under-explored lake, they found real submerged architecture, and their work pushed institutions to take Lake Van's underwater heritage seriously. The dispute is a fair one between an optimistic reading (Urartian fortress) and a cautious one (medieval castle with Urartian spolia) — and it is exactly the kind of question more diving, not more headlines, can settle.

Key evidence cited

Tahsin Ceylan's team filmed extensive cut-stone walls, reported as running for around a kilometre, at shallow depth outside Adilcevaz harbour in 2016-2017
Some squared blocks in the footage resemble Urartian ashlar, consistent with at least partial Bronze-Iron Age origins
Global media presented the find as a 3,000-year-old Urartian castle, and Turkish coverage embraced the ancient dating
Lake Van was demonstrably lower in the Urartian period, so genuinely Urartian shoreline structures could lie underwater elsewhere in the lake
Fringe accounts link the find to lost-civilisation and flood traditions of the Ararat region, and even to the Lake Van monster legend
The lake's proven underwater surprises — giant microbialite towers, wrecks — encourage expectations of further undiscovered structures

Genuinely open questions

Are the submerged walls at Adilcevaz the medieval castle recorded by Burney and Lawson, an Urartian structure, or a composite of both?
Can stratigraphic underwater excavation or mortar dating resolve the construction sequence?
How much genuinely Urartian shoreline architecture lies beneath Lake Van, given the lower lake levels of the Iron Age?
Why has no full scientific excavation followed the 2017 publicity, and what would a systematic survey of the lake's shallows reveal?

Worth knowing

The same dive team's earlier finds in Lake Van include underwater towers of microbialite — rock built by microbes — up to several metres tall, so the lake really does contain castle-like structures that no human ever built.