Best-supported assessment
Supported interpretation: Byzantine-period refuge use is the strongest broad historical attribution in the sources reviewed here. Unknown: no securely contextualized determination of the earliest excavation, or securely dated pre-Christian use deposit at Derinkuyu, was recovered in this bounded review. This is not a claim that such evidence cannot exist. UNESCO associates Derinkuyu with refuge during Arab invasions, but its fourth-century statement concerns the beginning of regional monastic activity, not the foundation of this underground complex. [S206 · UNESCO World Heritage Centre. Göreme National Park and the Rock Sites of Cappadocia, Ou…]
Main limitation: we have institutional estimates and comparative arguments, but not an audited Derinkuyu sample-to-feature sequence. The Ministry estimates construction in the ninth–tenth centuries without supplying the dating evidence on that page. Keep this as its attributed estimate, not an exact foundation date or a scientific consensus interval. [S207 · Türkiye Ministry of Culture and Tourism. Derinkuyu Underground City.]
What the competing claims actually rest on
| Proposal | Evidence recovered | Assessment and missing test |
|---|---|---|
| Hittite or other pre-Christian beginnings | Kahya Yıldırım (2025), p.88, reports Hittite artifacts and separately Roman eagle objects, citing Ayyıldız (1990), p.96. [S208 · Kahya Yıldırım, E. (2025)] | Disputed proposal. The chain reaches a named source, but its page, accession numbers, object descriptions and find contexts have not been inspected. We cannot authenticate a particular “Hittite object.” |
| Phrygian beginnings, eighth–seventh centuries BCE | The Geopark presents this as a possibility attributed to the Ministry. [S213 · Cappadocia Geopark. Derinkuyu Yeraltı Şehri, English overview.] | Disputed proposal. Its summary supplies no construction-linked sample or find catalogue. Repetition is not independent evidence. |
| Regional medieval shelter development | Bixio and Yamaç favour a mid-seventh to early-eleventh-century setting and trace Urban’s earlier-origin hypothesis to Vorland 7 (1973), pp.180–181. [S209 · Bixio, R. and Yamaç, A. (2023)] | Supported interpretation at regional scale. Their comparative pottery examples are other settlements; do not transfer those finds to Derinkuyu. Urban’s original has not been read in this pass. |
The surveyors explicitly retain uncertainty about earlier origins. Nearby Neo-Hittite inscriptions do not establish a direct relationship to underground excavation. Their Derinkuyu phasing diagram is a proposed construction sequence, not an independently dated sequence. [S209 · Bixio, R. and Yamaç, A. (2023)]
Even an authentically early portable object would need a documented relationship to first cutting or use. It could have arrived later, survived as residual material, or come from disturbed deposits. Conversely, later occupation material cannot exclude an older core. Our inference: what matters is the object–deposit–feature relationship, not merely the oldest-looking object associated with the town.
Artifact provenance follow-up — 5 November 2026
Observed evidence: the catalogue lists a 1990 Turkish Eşsiz Kapadokya at AKMB Aydın Sayılı Kütüphanesi, call AKM/K/29649, book accession 0037474. Access limitation: p.96 remains unread; the holding does not verify the second-edition statement or its contents. English editions cannot substitute without checking pagination. [S215 · AKMB Aydın Sayılı Kütüphanesi, catalogue record: Uğur Ayyıldız, Eşsiz Kapadokya (NET, 1990)]
Reported evidence: Okuyucu’s thesis p.49 n.241 places the Roman marble eagle in the garden of Ankara’s Museum of Anatolian Civilizations. Fig.5 (p.162) credits Ö. Demir. This was read in a public third-party transcription; the photograph and institutional original remain unverified. No museum accession, precise find locus or deposition sequence was recovered. The author’s institutional CV dates the thesis 2008, while the mirrored title and subsequent citations use 2007. [S216 · Demet Okuyucu (2007)]
The 2007 conference paper p.515 reports the eagle but its footnote3 concerns the preceding dating-uncertainty statement, not an object catalogue. Üstüner’s 2024 p.79 repeats the claim citing Okuyucu. These are a citation trail, not independent archaeological confirmations. [S217 · Demet Okuyucu, Derinkuyu Yeraltı Şehri, Tarihi Eserlerin Güçlendirilmesi ve Geleceğe Güvenle Devredilmesi Sempozyumu-1 (2007)] [S218 · M. Üstüner (2024)]
Unknown: current custody, museum inventory identity, dating basis and relationship to first excavation. The official Ankara museum directory provides a records enquiry destination, not confirmation of this object. [S219 · Ministry of Culture and Tourism, Ankara Anadolu Medeniyetleri Müze Müdürlüğü, instituti…] The unspecified Hittite artifacts in S208 remain a separate unverified claim; do not identify them with the Roman eagle.
Our inference: even a confirmed Roman sculpture would date its manufacture, not automatically the underground deposit or cutting. No earliest-construction date changes. The access log and exact records needed are saved in research/data/derinkuyu/artifact-trail.json. No external enquiry has been sent.
Demir photograph and inventory search follow-up
Access limitation: the photograph itself and a matching Ankara accession remain unrecovered. Demir editions now have concrete retrieval routes: a 1990 revised English copy at AKMB Aydın Sayılı Kütüphanesi (AKM/K/45095; library use only), a 1979 Google Books metadata record, and a restricted Internet Archive copy catalogued as the 1997 eighth revised edition. None is established as the undated Turkish edition cited by Okuyucu. [S220 · AKMB Aydın Sayılı Kütüphanesi, Ömer Demir, Cappadocia cradle of history Göreme (1990 revised edition)] [S221 · Google Books bibliographic record, Ömer Demir, Cappadocia: Derinkuyu: Cradle of History (Ajans-Türk, 1979)] [S222 · Internet Archive, Ömer Demir, Cappadocia: cradle of history, item cappadociacradle0000d…]
Unknown: whether Demir took the photograph or reproduced it. The figure credit establishes attribution to his work, not original photographic authorship. No museum inventory identity follows from a library book number or a museum ticket catalogue number. Turkish/English searches did not recover an object record; this is not proof that unpublished records are absent.
The structured artifact log now includes edition-specific access failures and unsent library/museum questions. UP-DK-ORIGINS was reviewed again: no construction-linked date was recovered, so its unplotted status remains appropriate. Reconstruction remains pending.
Yenipınar’s site-improvement report
Observed evidence: the complete report by Nevşehir Museum Director Halis Yenipınar (published 2002; conference 2001), pp.205–210, has now been visually read. Its subject is modern site management and improvements. It reports no eagle, museum accession or dated archaeological deposit. This limited silence does not disprove the artifact claim. [S223 · Halis Yenipınar (2002)]
Disputed proposal: p.205 explicitly leaves cutting dates uncertain and explains Hittite attribution through settlements above or near some underground sites. That is regional association, not a dated Derinkuyu construction context. Its Xenophon reference supplies no site identification; the Armenian-route limitation established above still applies. [S223 · Halis Yenipınar (2002)] [S210 · Xenophon. Anabasis 4.4.1–4 and 4.5.25–34, English translation hosted by the Circle of A…]
Reported evidence: p.205 describes upper-level storage/stabling before the reported 1965 opening; p.206 documents planning and works in 1992–2001. These attest modern reuse and intervention, not ancient foundation. Access limitation: text and captions are legible, but the scanned plan and photographs are too dark for fine-detail identification. The page-by-page scope, file fingerprint and retrieval limitations are in research/data/derinkuyu/yenipinar-report.json. Atlas record UP-DK-ORIGINS remains unplotted and now includes S223. [S223 · Halis Yenipınar (2002)]
Urban’s Phrygian hypothesis: original-access audit
Access limitation: Urban’s original 1973 articles remain unread. The series comprises Vorland no.6, “Der Befund” (pp.149/150–153); no.7, “Geschichtlicher Rahmen und Deutungen” (pp.174–181); and no.8, “Maginotlinie der Frühgeschichte” (pp.205–212). ARGE Höhle und Karst Stuttgart lists all three. MUSE’s holding covers nos.6–7. Catalogue identification is not full-text verification. [S224 · ARGE Höhle und Karst Stuttgart, library holdings: Martin Urban’s three-part Vorland series (1973)] [S225 · MUSE, Periodici cartacei della Biblioteca, Vorland holding.]
Disputed proposal, mediated through later research: Bixio and Yamaç’s p.156 describes Urban’s eighth–seventh-century BCE buffer-zone hypothesis between Phrygian and Assyrian powers, citing part2 pp.180–181. They explicitly report that decisive archaeological/textual confirmation remains lacking. Nearby Neo-Hittite inscriptions are not demonstrated links to underground construction. The defence interpretation provides a possible historical setting, not a Derinkuyu founding determination. [S209 · Bixio, R. and Yamaç, A. (2023)]
Ayhan’s thesis p.27 separately attributes Urban’s dating to investigations of “millstones.” The recovered university-indexed passage supplies no identifiable object or dated context. Original German wording and figures must resolve whether grinding stones, circular doors or another comparison is meant. We have not verified a Phrygian diagnostic artifact. [S226 · Arda Ayhan (2004)]
Our assessment: retain an older core as possible, but do not promote this regional hypothesis to established Phrygian builders at Derinkuyu. The structured audit at research/data/derinkuyu/urban-argument.json records acquisition routes, competing explanations and tests. UP-DK-ORIGINS was reviewed; no dated marker or ancient reconstruction added.
Millstones and the search for dated early use
Observed source trail: Ayhan’s indexed p.105 bibliography names Urban’s 1973 work but supplies no journal, part or page. His p.27 attribution therefore remains unresolved at the object level: the recovered text does not identify a stone or its context. The complete thesis and Urban originals remain unread. [S226 · Arda Ayhan (2004)]
Reported evidence, original text now checked: Kahya Yıldırım’s p.84 leads to Emge’s 2011 Turkish paper, PDF p.5. Emge reports a Hittite handmill from a deeper Derinkuyu level, but explicitly rejects it as sufficient dating evidence: similar tools remained in use and it could have arrived later. He provides no object-specific source, inventory or stratigraphy. His p.4 separately describes millstone-shaped doors. These are distinct categories; neither identifies what Ayhan meant. Emge postdates Ayhan and cannot be his source. [S208 · Kahya Yıldırım, E. (2025)] [S227 · Emge, A. (2011)]
Our inference: the handmill’s cultural attribution is unverified, and even an authentic early manufacture date would not establish early deposition or use. Emge’s broad statement about centuries of reuse is not a measured occupation interval. We cannot yet equate this tool with Urban’s stones or Ayyıldız’s unspecified artifacts.
Bounded negative result: a case-insensitive all-field search of C-Turkey’s 3,390 radiocarbon rows for Derinkuyu, Malakopi and Malakope returned no matches at commit 81e1e6651c3bd6f73a83fef3f1ac6421f62ee9e9. This does not exclude missing, differently named or unpublished dates. No securely contextualized pre-Christian use determination was recovered from this review; the eagle and other artifact claims retain their provenance gaps. [S228 · Altınışık, N.E. (2026)]
The reproducible search, source locators and access failures are saved in research/data/derinkuyu/millstone-audit.json. Emge’s HTTPS endpoint failed, but the bibliography’s HTTP address worked; relevant pages were visually checked. Atlas/globe record UP-DK-ORIGINS remains unplotted, now citing S227. Reconstruction remains pending.
Handmill provenance follow-up — 6 November 2026
Observed source evidence: the author-hosted English and German versions of Emge’s paper also place the reported handmill in a deeper Derinkuyu level. Both lack an object-specific reference. The German p.5 explicitly questions precise dating, in addition to possible later introduction. The English p.5 note12 names Yörükoğlu and colleagues’ 1989 booklet, but its marker belongs to the following general Kaymaklı/Derinkuyu discussion. It is a retrieval lead, not a demonstrated source for the handmill. These versions are not independent attestations. [S229 · Emge, A. (2011)]
Unknown: finder, discovery date, exact room/layer, inventory number, current custody and the basis of the “Hittite” attribution. No original find report or matching museum record was recovered. The Gülyaz/Yenipınar and Krassmann works in adjacent references remain candidates without an established object-specific link; Krassmann’s direct PDF endpoints failed. No typological comparison can be audited without an identifiable object.
Next records route: official ministry pages identify Nevşehir Museum as the regional enquiry route for Derinkuyu, not as a confirmed holder of this object. Its published contact is nevsehirmuzesi@ktb.gov.tr, +90 384 213 14 47. Ask for cleaning/excavation registers, accession and transfer records, photographs and the original dating rationale. Do not assign the handmill to Ankara on the basis of the separate eagle claim. [S230 · Ministry of Culture and Tourism, Nevşehir Müze Müdürlüğü and associated sites.]
The access log, parallel-text comparison and specific record questions are saved in research/data/derinkuyu/handmill-provenance.json. No enquiries were sent. The atlas/globe assessment was reviewed and remains unchanged: no securely dated early-use event has been established; an older core is not excluded.
Yörükoğlu booklet: edition and access audit
Access limitation: the original 1989 booklet remains unread. The Cappadocia Academy bibliography lists Yörükoğlu and colleagues’ Underground Cities in Cappadocia (Ankara, 1989). An indexed English seller listing names Aşık Ofset; a directly accessible Turkish listing describes an 80-page Beğendik edition from Ankara, 1989. These identify acquisition routes, not verified contents. Later 1990 citations must not be silently treated as the same pagination. [S231 · Cappadocia Academy bibliography, Yörükoglu et al.1989 entry; bookseller edition/acquisi…]
Unknown: whether the booklet contains the handmill report or another securely contextualized early Derinkuyu find. No readable original passage, artifact plate or find catalogue was recovered. Emge’s adjacent general citation remains a candidate trail, not proof of dependence for the handmill. [S229 · Emge, A. (2011)]
The access log and edition-specific retrieval checklist are in research/data/derinkuyu/yorukoglu-booklet.json. No purchase or enquiry was made. Atlas/globe coverage was reviewed; no new dated event supports changing UP-DK-ORIGINS or the pending reconstruction.
Two historical records that need careful geographical handling
Observed evidence: Xenophon’s Anabasis 4.5.25 describes underground houses. The preceding route and the villagers’ identification in 4.5.34 place this episode in Armenia. It neither names Derinkuyu nor identifies its passages. It is therefore an analogy for subterranean dwelling, not a fifth/fourth-century BCE date for this site. [S210 · Xenophon. Anabasis 4.4.1–4 and 4.5.25–34, English translation hosted by the Circle of A…]
Observed evidence: Dawkins’s Modern Greek in Asia Minor (1916), p.24, supplies direct testimony at Malakopi, identified with modern Derinkuyu by the British School at Athens. He describes exploring rooms below the house of the singer Khristos, protected by five stone doors and connected by steep steps to a well shaft. His itinerary on p.1 covers visits in 1909, 1910 and 1911. This establishes the existence of that complex by his fieldwork, not when it was excavated. Its exact correspondence to today’s ticketed sector remains unverified. [S211 · Dawkins, R. M. (1916)] [S212 · British School at Athens (2020)]
Dawkins is the earliest firsthand underground exploration account directly inspected in this pass, not a claim to the earliest surviving record. His p.24 footnote points to Arkhelaos (1899), p.124, for water depth; that older source is a next lead. His separate reference to a church founded in AD 970 concerns a church site, not the excavation date of the underground rooms. His p.17 accounts of regional refuges likewise should not be silently reassigned to Derinkuyu. [S211 · Dawkins, R. M. (1916)]
Provisional sequence, not a founding timeline
- Possible early core: Hittite/Phrygian claims remain unverified as construction events.
- Medieval refuge and enlargement: the working historical interpretation; neither first cutting nor every room is securely dated.
- Documented later existence: the Malakopi complex seen during Dawkins’s 1909–1911 journeys; not the beginning of occupation.
These are different kinds of evidence. They must not be joined into one continuous occupation range. The 2016 engineering paper also acknowledges uncertain origins; its technical subject does not make repeated ancient-origin claims into excavation results. [S214 · Nývlt, V., Musílek, J., Čejka, J. and Stopka, O. (2016)]
Why this conclusion, and what would change it?
The early claims currently lack a recoverable context tying them to construction. The strongest alternative is an older core altered during later refuge use. To test it, recover Ayyıldız’s cited page and underlying museum records, then map any objects to their original deposits. Read Urban’s original argument rather than relying on later summaries. Seek sealed cutting debris, initial floors, dated inscriptions or a stratigraphic sequence that connects datable material to a particular chamber. A credible early use date would change that phase’s assessment without automatically dating the whole complex.
The structured ledger at research/data/derinkuyu/investigation.json preserves claim IDs, source locators, alternative explanations and inaccessible-source limits. This pass did not produce an occupancy model, a tunnel-network survey or a complete engineering reconstruction.
Reconstruction and atlas/globe coverage
Illustrative reconstruction: none created. Environment, climate, seasonality, resources and ancient daily activities remain pending in the reconstruction record; modern terrain and later observations are not a source-certified ancient scene.
Atlas/globe coverage checked. UP-DK-ORIGINS records the unresolved chronology with sources. No founding marker, invented coordinate or combined “possible age” interval was added. The firsthand modern attestation is retained in the dossier rather than plotted as an ancient construction event. [S207 · Türkiye Ministry of Culture and Tourism. Derinkuyu Underground City.] [S209 · Bixio, R. and Yamaç, A. (2023)] [S211 · Dawkins, R. M. (1916)] [S212 · British School at Athens (2020)]