Houses at Göbekli Tepe: What the First Northern Dig Does and Doesn't Settle

TL;DR: On 13 August 2026, digging began for the first time in the northern section of Göbekli Tepe. Just under the surface came rectangular buildings dated to the site's second phase. The excavation director, Necmi Karul, chose his words carefully: «Our first impression is that they were used as residences.» Eight days later that sentence had travelled through the English-language press and come back as «proved». Here is what it actually settles. The cult-centre reading rested on two legs, and Karul lists both of them himself: no houses, and no water source nearby. The northern dig weakens the first. The second is untouched. By the excavation team's own measurement, the cisterns hold 153.12 cubic metres in total, and the nearest springs are about five kilometres away. There is also a detail almost nobody reported: the German Archaeological Institute revised the «World's First Temples» framing in its own journal back in 2020. Below I trace who said what, where each number came from, and how a hedged sentence lost its hedge in six hops.

What was actually announced on 13 August?

Two people announced it, on the same day, separately. That distinction matters for everything that follows, so let me get it straight at the start.

In the morning, Culture and Tourism Minister Mehmet Nuri Ersoy posted on social media: «Under our Heritage for the Future Project, we have begun excavating for the first time in an area of roughly 80 by 70 metres in the northern section of the mound.» The same post mentioned geomagnetic surveys, and noted that work in the GT1 and GT2 areas and conservation on Structures A, C and D were running in parallel.

Later the same day, the Anadolu Agency published an interview with Prof. Dr. Necmi Karul, coordinator of the Taş Tepeler Project and director of the Göbekli Tepe excavation. What Karul described was more concrete. Digging had started this season in the new area north of the shelter, and rectangular buildings belonging to the site's second phase had already appeared just below the surface.

The olive trees are part of this story too. Trees standing on the plot were moved elsewhere, and only then did the ground become workable. Karul makes a point of it: the trees had not caused serious damage, but had they stayed, the area would have remained impossible to excavate.

Is it 80 by 70, or 1,500 square metres?

Both figures are accurate. The striking part is that both came from Karul.

On 13 August he told the Anadolu Agency that work had begun «on a new area of roughly 1,500 square metres». On 23 August, in a piece he wrote himself for Milliyet, he described «working on a new area measuring 80x70 metres on the northern slope of the mound». Eighty times seventy is about 5,600 square metres. The gap is more than threefold.

Correcting a number is itself a claim, and it carries the same burden of proof. So I am not going to pick one. No source states which figure measures what. The most plausible reading is that 80 by 70 describes the whole area opened to research, while 1,500 square metres is the portion actually made diggable this season. But that is my reading, not the source's statement, and I cannot present it as fact.

I care about this distinction because I walked into the same trap recently. Checking a claim that the Arctic route had «halved» the Suez transit, I found the number was correct and still measuring the wrong thing. Who established the baseline usually matters more than the figure sitting on top of it.

For the record: several outlets presented 5,600 and 1,500 as two conflicting reports. They are not conflicting. They are two numbers from the same person, ten days apart, in two different contexts.

What came out from just below the surface

Erosion helped. Karul's description: the upper courses of the walls wore away long ago, which left the building plans sitting very close to the surface. Some floors were reached, in his phrase, «very quickly».

His Milliyet piece goes further. The buildings were constructed adjoining one another. Floor level has been reached. Some floors are coated with terrazzo, a mixture produced by burning lime. And then the sentence that matters most: there are food-preparation finds inside the buildings.

That last detail is stronger evidence than the «internal arrangement of the spaces» phrasing in the agency report. Internal arrangement invites interpretation. Finding food-preparation equipment inside a room invites rather less.

There is a negative finding as well. The pillar tradition has weakened in these buildings; they hold fewer standing stones. Buildings of the same type had already turned up in an area labelled GT1 in earlier seasons.

What the geomagnetic survey showed, and why two official accounts differ

Before digging started, the area was covered by surface scans and geomagnetic measurement. That is where the first picture of what lies underground comes from. Two official statements describe the same survey differently.

Ersoy's 13 August post says the measurements point to «large public buildings, along with rectangular buildings and huts». Karul's 23 August article says this: «From geomagnetic measurements taken in earlier years we know that here too there are similar, round-plan large buildings.»

Rectangular versus round. At Göbekli Tepe that is not a technical footnote, it is a phase marker. Round-plan monumental buildings belong to the earlier phase, rectangular ones to the later. Which of the two is waiting under the soil changes what the dig will find.

I am leaving both statements as they stand. Karul adds that the team believes it has reached the upper levels of the large round-plan buildings picked up by the survey. Either way, both descriptions get tested in the seasons ahead.

The site in numbers: 9 hectares, 1,500 years, at least three phases

Popular coverage tends to reduce this place to a handful of stone circles. The scale is worth setting out, because the argument only makes sense against it.

Fifteen centuries is a long time. Count back that far from today and you land around the completion of the present Hagia Sophia. This is the first reason the question «what was Göbekli Tepe» has no single answer. A place is not obliged to serve one function for fifteen hundred years.

Karul summarises the architectural drift in his own words. The earliest inhabitants build small round huts; over time those become more permanent dwellings; and towards the end of the period, people discover the corner and start building on a rectangular plan. In the history of architecture, discovering the corner changes how a building can grow.

What «second phase» actually means

Turkish coverage says «the 8000s BC». English coverage says «the early 9th millennium BCE». These are the same range under two naming conventions, since 8000 to 8999 BC is the ninth millennium. The difference is vocabulary, not data.

The buildings in the north belong to Layer II. That places them in the more modest family of structures that either overlapped with the monumental enclosures or followed them.

The radiocarbon side has hard numbers too. In the sequence published by the excavation team, a sample from the wall plaster of Structure D returned 9,984 ± 42 radiocarbon years, while charcoal collected close to bedrock clusters between 9664 and 9311 cal BC at 95.4% confidence.

One piece of background belongs with those dates. Göbekli Tepe went up as the world was emerging from the last Ice Age. Periods when climate moves on that scale have always been thresholds for our species, and we now watch the weakening of ocean circulation through the same lens. The difference is that nobody back then could measure it.

Whose temple thesis was it? Schmidt's own sentence

One person deserves to be described accurately here, because he has taken the worst of the last week's coverage without being around to answer.

Klaus Schmidt began the excavations in 1995 and read the site as a cult centre. The idea attributed to him, that the temple came before the city, inverted the standard account of the agricultural revolution. People did not settle down, take up farming and then build temples. Large ritual centres may have pulled people together first, and settlement and farming followed.

The abstract of Schmidt's 2010 paper in Documenta Praehistorica puts it plainly: «It was not used for habitation; it consists of several sanctuaries in the form of round megalithic enclosures.»

That was a firm claim, and it had firm support. Decades of excavation had produced little evidence of domesticated plants or animals, and no pottery at all. Schmidt died in 2014. He never saw most of the data that followed.

The objection arrived in 2011

E. B. Banning published «So Fair a House» in Current Anthropology in 2011. The argument had two layers.

The first was methodological. Archaeologists tend to project a modern Western split between sacred and profane onto prehistory, while anthropology keeps showing those two domains tangled together in most societies. The second layer was direct: the buildings at Göbekli Tepe may in fact be houses, rich in symbolic content but domestic all the same.

Banning pointed to in-house burials, caches and wall paintings as evidence that the sacred was leaking into daily life rather than being sealed off in a separate building type.

The excavation team answered in 2017

The German Archaeological Institute's Göbekli Tepe team replied at length on their research blog. The title Jens Notroff gave the post tells you the register: «A Sanctuary … or so fair a House?»

Notroff accepts that Banning's objection to the sacred-profane split is valid. Then he sets a counterweight against it. The T-shaped pillars are unmistakably anthropomorphic, carved with arms, hands and a garment, and they are distinctly larger than life. Life-sized sculpture was well within reach at the time; the so-called Urfa Man is considered the oldest known life-sized human statue. So the abstraction and the scale were choices, not technical limits.

One regional observation in that post reframes the whole debate. Research over the previous twenty years had shown that almost every extensively excavated settlement of the tenth and ninth millennia BC separates living quarters from workshop areas, and also produces special buildings or open spaces for communal or ritual activity. Çayönü's Skull, Terrazzo and Flagstone Buildings. Nevalı Çori's Terrazzo Building. The communal structures at Jerf el Ahmar and Mureybet in northern Syria.

So «dwelling or special building» is not a dilemma unique to Göbekli Tepe. Across the region, the answer is usually both.

The real revision happened quietly between 2019 and 2020

This is the part almost none of last week's coverage mentions.

In 2019, Laura Dietrich and colleagues published a paper in PLOS ONE. More than seven thousand grinding slabs, handstones, pestles and mortars had been analysed, a density unusual for contemporary sites in the region. Phytolith analysis confirmed that cereals were present at the site in quantity. And the finding that matters most here: there is a clear link between plant-food preparation and the rectangular buildings.

What turned up in the north in August 2026 had already been seen in the main excavation area in 2019. The rectangular buildings were doing kitchen work.

The same team found no large storage facilities and concluded that food was produced for immediate consumption. Laura Dietrich's 2021 book sharpens the picture: porridges and beers predominate over bread-like food, cooking took place around the monumental buildings, and the sheer quantity of tools points to feasting in addition to daily meals.

Then, in 2020, the institute running the excavation wrote this in its own publication, under Lee Clare's name: the work carried out between 2015 and 2019 is leading to a revision of former interpretations, the ones that «still fuel the popular media image» of the site. The phrase in quotation marks in that text is «World's First Temples».

The cult-centre case stood on two legs. One is gone.

Karul's own sentence is the honest way to frame this section, because he is the one who lists what the case rested on.

From his Milliyet article: on the cult-centre reading this was a place people gathered at certain times for feasts, and «the absence of dwellings and of water sources in the immediate vicinity supported that view».

Two legs. No houses, no water.

The northern dig reaches the first one. The second is exactly where it was, and its numbers come from the excavation team as well.

153.12 cubic metres is roughly 153 tonnes of water. I am not going to work out how many people that supports for how long; the team has published its own assessment and it sits in the table above. Notroff's phrasing in 2017 lands harder anyway: the spot itself is «hostile to settlement». The highest point of the Germuş range, a landmark visible for kilometres, and a hill with no water on it.

Which is why the honest answer is not binary. «Temple or village» is probably a badly built question. The reading that the monumental buildings served a special purpose is still standing. What has changed is the assumption that nobody lived around them.

Does a terrazzo floor prove domestic use? Careful here.

Karul reporting terrazzo floors in the new buildings looks at first like a point for the domestic reading. A carefully finished floor suggests permanence.

Except that the same material shows up on the other side of the argument. Two of the Pre-Pottery Neolithic «special purpose buildings» Notroff lists in 2017 are named after it: Çayönü's Terrazzo Building and Nevalı Çori's Terrazzo Building. Those are classified as communal or ritual structures, not houses.

So terrazzo says neither dwelling nor temple. It demonstrates labour and permanence. It does not demonstrate function.

I am not writing that as a criticism. It is simply how the discipline works, and it is exactly why drawing a conclusion from a single find is difficult.

What will actually settle this? Karul already said.

This is the sentence that got trimmed away in almost every retelling, and it may be the most important thing the excavation director said.

In his agency interview, right after suggesting the rectangular buildings may have served as dwellings, he adds: «But of course it is not enough to say this simply by exposing the walls. In particular the soil samples we are collecting inside these buildings will tell us about the function of the spaces. And that is something we can say once we reach floor level, from the samples taken off the floor.»

So the answer is in the soil, not the walls. What do you look for in floor-level samples? Phytoliths, the microscopic silica casts left behind by plant tissue. Charred seed remains. Hearth traces. Micro-debris distribution, which shows which corner of a room was used for what. That was precisely the method behind the 2019 cereal study.

None of this finishes during a dig season. It goes to a laboratory, gets sorted, and eventually becomes a publication. Karul makes that point too: what separates an archaeological excavation from an earthmoving job is the sorting of data in the lab, the soil analyses and the documentation, and every few years a season is paused so the work can happen indoors.

How a sentence lost its hedge in eight days

Now to the part I find most interesting. Karul's careful sentence went out in Turkish on 13 August. By the time it returned to Turkish on 21 August, it had become «proved».

I traced the chain link by link.

Four distinct drifts, and none of them malicious.

The first is attribution. The person who posted on social media was the minister, not Karul. The second is shape: Karul describes round-plan buildings in the geomagnetic data, while the word «rectangular» from Ersoy's list merged with «public buildings» into a single category. The third is method. You see internal layout by digging, not by scanning. Geophysics gives you the trace of a wall; it does not tell you what is inside the room. In the Archaeology Magazine sentence those two have swapped places.

The fourth drift is modality, and it is the heaviest. The distance between «our first impression is that they were used as residences» and «proved» is the distance that protects a scientist's career. In the same interview Karul had said explicitly that exposing walls is not enough to make the claim. After the first link in the chain, that sentence appears nowhere.

I ran into the same pattern while looking at an AI security incident: the figure 141,006 was widely attributed to the wrong organisation, and once a number slips owners it does not slip back. Six hops in eight days. At the end of it nobody has lied; everyone has summarised the person in front of them.

I went and looked at Karul's own accounts

«Explained on social media» is a checkable claim, so I checked it.

Two social media posts are embedded in the IFLScience article. The first belongs to the Culture and Tourism Minister's verified Instagram account. The second is from an archaeology news account, with the official Taş Tepeler account listed as co-author. That second post refers to Karul but was not written by him: the text is third person, and the content is a summary of the agency interview.

As for Karul's own accounts. His last post on X is dated November 2022, with nothing at all in August 2026. His Instagram account is active, but the only thing he posted himself between 13 and 19 August was a poster for a lecture he was giving at the Kazdağı Museum on the 15th, and it carries no caption.

So the sentence is his. No question about that. He said it to a journalist. What was on social media were two other things, and both belonged to somebody else.

One caveat for the record: Instagram stories vanish after 24 hours and nobody can verify them afterwards. Which is why I am writing «I could not find one» rather than «there was none».

What was hiding under the olive grove?

The northern plot lay under olive trees for years, and that grew its own mythology online.

In the same Milliyet edition, Mesut Alp lists the claims in circulation. Excavation was being deliberately blocked. Trees had been planted on purpose to cover secrets that would rewrite history. Foreign powers had stopped the dig.

The actual reason he gives is duller and far more familiar. Until recently the land consisted of privately owned fields. Planting fruit or timber trees before or during an expropriation process, in order to raise the compensation, is a well-known practice in Turkey. Once ownership was settled and the trees were relocated, the excavation team could move in.

The trees were not hiding a secret. They were inflating a payout.

Alp raises one more point worth repeating. When a season is skipped, the reason is usually laboratory work, and Turkey's Ministry of Culture and Tourism has published its excavation reports openly every year since the 1980s. How much of Anatolia's scientific record is already accessible came up for me before, in the work that pulled 1,700 years of aurora sightings out of the archives. The data is usually sitting there. Nobody opens it.

Where the rest of Taş Tepeler stands

Göbekli Tepe is no longer a site on its own. The Taş Tepeler Project, launched by the Ministry of Culture and Tourism in 2021, runs excavations at several locations; in his article Karul names Karahantepe, Sayburç, Çakmaktepe, Gürcütepe, Harbetsuvan, Söğüt Tarlası and Sefertepe. Part of what makes the northern dig meaningful is what the neighbours have produced.

On 19 August, after a presentation by Karul on Karahantepe and Sayburç, Minister Ersoy shared fresh figures. In its sixth season, Sayburç has exposed roughly 70 dwellings and special buildings across 2,700 square metres. The settlement dates to the middle of the ninth millennium BC and is a multi-layered mound, running from the Neolithic through the Roman period into modern village life.

What matters about Sayburç is this: the excavated buildings show architecture shifting from round to rectangular plans, and they yield evidence about the emergence of the house and the household. Another instance of the same transition now visible north of Göbekli Tepe.

At Karahantepe the finds include the earliest known examples of architectural elements such as the staircase and the window. The ministry plans to turn Sayburç into an open-air museum, and the traditional village buildings on the expropriated land into a village museum. Worth stating plainly that this is a plan, not something already done.

Which visitor number should we trust?

A walkway is planned around the new excavation area so visitors can follow the work closely. That sent me to the tourism figures, where I found more disorder than I expected.

Consistent so far. The trouble starts with earlier years. The Anadolu Agency's January 2026 report gives 850,000 for 2022, 317,253 for 2023 and 709,643 for 2024. The same agency's October 2025 report gives 837,811, 512,164 and 731,794 for those same years.

The gap for 2023 is nearly two hundred thousand people. One series probably counts ticketed visitors and the other total entries. No source says so, and I am not going to publish my guess as data. The table above therefore carries only the figures with a single, clearly named source.

While I was at it, a travel site turned up in the search results claiming 1.4 million visitors for 2025, against 781,000 from the governor and the ministry, sourced to an unnamed «museum director». Putting a number like that into this article would make me the seventh link in the chain I just spent a section taking apart.

Frequently asked questions

Has it been proved that Göbekli Tepe was not a temple?

No. What has been established is that rectangular buildings lie just below the surface in the northern section, with food-preparation finds inside them. The excavation director calls this a first impression and says function will be determined from soil samples. The reading that the monumental buildings served a special purpose stands separately.

Does this finding make Klaus Schmidt wrong?

Partly. Schmidt wrote in 2010 that the site was not used for habitation, and that claim is weakening. But his central argument was that the monumental buildings were communal and ritual in purpose, and that part remains open. He died in 2014 and did not see most of the later data.

How large is the new area?

The ministry and Karul both say 80 by 70 metres, roughly 5,600 square metres. Karul also says work began this season on about 1,500 square metres. Which figure measures what has not been explained.

How could anyone live on a hill with no water?

There is no settled answer. Pits on the western slope may be cisterns with a calculated total capacity of 153.12 cubic metres; by the excavation team's assessment that would support a longer stay but probably not the entire rainless summer. The nearest accessible springs are about five kilometres away. Seasonal or periodic use therefore remains on the table.

What is being done about earthquake risk to the pillars?

In Structure C, the last surface fragments of a re-erected pillar have been put back in place. In Structure D, the walls were reinforced and then the floor connections of the two facing central pillars were strengthened, and the pillars are being secured against tremors with steel cables.

Might monumental buildings turn up in the north as well?

Karul says the team believes it has reached the upper levels of the large round-plan buildings detected by the geomagnetic survey, and the ministry's statement also refers to large public buildings. So the odds look good, but nothing monumental has been exposed yet.

When will this excavation be finished?

Karul's plan for this year is to strip the surface level across the whole area and expose the remains immediately beneath it. Beyond that, no timetable has been announced. Pauses between seasons for laboratory work are a normal part of the process.

Three things I will be watching in September

The soil samples come first. Analysis of the floor-level samples is the only real arbiter of the «were they houses» question. When the phytolith and micro-debris results are published we will see how well the first impression held.

Second, the large buildings visible in the geomagnetic survey. Rectangular or round? Two official statements say two different things, and the soil can settle that dispute this season.

Third, and this one may sound odd to watch: how the same story gets told a year from now. A piece that says «proved» today stays that way online unless a correction is published. Even if the soil results confirm the first impression, that sentence will not have been right because it was verified. It will have been right by accident.

One closing note. Every outlet in that chain had the same source available: the Anadolu Agency report of 13 August, free, open, five minutes to read. Karul's caveat was sitting inside it. It appears in none of the six links that followed.

The detailed piece Karul wrote himself came out on 23 August, after the chain had already run its course. The terrazzo floors, the food-preparation finds and the two legs of the cult-centre case all come from there. The text published after the news cycle ended said more than the entire news cycle did.

Sources

• Anadolu Agency, Rectangular buildings uncovered in new excavations at Göbeklitepe, 13 August 2026 (Karul's interview and Ersoy's post)
• Prof. Dr. Necmi Karul, Göbeklitepe's new route: questions the new excavation area will answer, Milliyet Sunday supplement, 23 August 2026 (80x70, terrazzo, food-preparation finds, the two legs of the cult-centre case)
• TRT Haber, First excavation north of Göbeklitepe, 13 August 2026
• Klaus Schmidt, «Göbekli Tepe – the Stone Age Sanctuaries. New results of ongoing excavations with a special focus on sculptures and high reliefs», Documenta Praehistorica 37 (2010), pp. 239-256
• E. B. Banning, «So Fair a House: Göbekli Tepe and the Identification of Temples in the Pre-Pottery Neolithic of the Near East», Current Anthropology 52(5), 2011
• Jens Notroff, «A Sanctuary … or so fair a House?», Tepe Telegrams (German Archaeological Institute), 24 January 2017 (cistern capacity, distance to springs, list of special purpose buildings)
• Laura Dietrich, Julia Meister, Oliver Dietrich, Jens Notroff, Janika Kiep, Julia Heeb, André Beuger, Brigitta Schütt, «Cereal processing at Early Neolithic Göbekli Tepe, southeastern Turkey», PLOS ONE 14(5): e0215214, 2019
• Laura Dietrich, Plant Food Processing Tools at Early Neolithic Göbekli Tepe, Archaeopress, 2021
• Lee Clare, «Göbekli Tepe, Turkey. A brief summary of research at a new World Heritage Site (2015–2019)», e-Forschungsberichte des DAI, 2020, doi:10.34780/efb.v0i2.1012
• Oliver Dietrich, Çiğdem Köksal-Schmidt, Jens Notroff, Klaus Schmidt, «Establishing a Radiocarbon Sequence for Göbekli Tepe. State of Research and New Data», Neo-Lithics 1/13, pp. 36-47
• IFLScience, New Excavations At Göbekli Tepe Are Challenging Old Theories, 19 August 2026
• Archaeology Magazine, Dwellings Detected Near Monuments at Göbekli Tepe, 20 August 2026
• Euronews Türkçe, New excavations at Göbeklitepe challenge established theories, 21 August 2026
• CNN Türk, Minister Ersoy: we are preparing for a new era in Şanlıurfa, 19 August 2026 (Sayburç, 2,700 m², roughly 70 buildings)
• Sabah, Visitor figures announced for Göbeklitepe and the Şanlıurfa Archaeology Museum, 3 January 2026 (Governor Şıldak: 781,381)
• Anadolu Agency, 4.5 million visitors to Göbeklitepe in its eighth year on the UNESCO list, 1 July 2026
• Verification of embedded posts: the two social media embeds in the IFLScience article and the posting dates on the @necmikarul accounts, browser inspection, 24 August 2026