Ancient Britain — Source Register Purpose: Single consolidated register of every source named across the Ancient Britain research sessions, for use alongside ancient-britain-knowledge-base.md (NotebookLM ingestion and wiki generation). Standard: Only sources actually named in the sessions are listed. Nothing has been added from outside them. Where a citation is incomplete (no author, journal or URL was recorded), it is marked as such rather than filled in. Scope: Mesolithic → Roman Britain (with post-Roman naming for context), matching knowledge base v4. Status: Living document — v1, compiled 2026-10-05 from sessions of 2026-09-09 to 2026-09-12. Matches knowledge base v4. How to read this file Provenance — how the source entered the research: GK — Claude's general knowledge in the session; no search run for that turn. Not independently checked. WS — surfaced by a web search in the session. Search results were not retained, so no URL is on record. Citation completeness: Full — enough detail to locate it (author, work, date). Partial — named, with year and/or journal or institution, but not enough to cite formally. Name only — described, but no author, title or venue recorded. Confidence stamps (as used in the Core Sample build): CONFIRMED · CONTESTED · INFERRED · UNRECOVERABLE A. Documentary sources (written, ancient) All are outsider accounts. None is a native British text. The Iron Age had no British writing system. Source Date Used for Tier Provenance Completeness Julius Caesar, Commentarii de Bello Gallico 50s BC Druidic religion: training, oak/mistletoe rites, reincarnation, human-sacrifice claims, claim that druidic learning originated in Britain. Describes Gaul mainly; Britain treated as an offshoot. Political motive (justifying conquest), so treat sacrifice claims sceptically. Tier 1 GK Full Diodorus Siculus — Partially independent corroboration of Caesar, via Posidonius Tier 1 GK Name only Strabo — As Diodorus Tier 1 GK Name only Posidonius — Lost; survives only through Diodorus and Strabo Tier 1 (indirect) GK Name only Pliny — Druids: sacrifice, omen-reading, mistletoe cutting, Anglesey as stronghold Tier 1 GK Name only Pytheas of Massalia c. 320 BC First record of Pretannikē ("the Pretannic Isles"), root of "Britain" Tier 1 GK Partial Gap: the source for Boudica invoking Andraste (60/61 AD) was never named in the sessions. B. Medieval literary sources (late, non-British) Source Date written Used for Tier Provenance Completeness Ulster Cycle (Irish) from 8th c. AD Celtic mythology, Otherworld Tier 3 GK Name only Mythological Cycle (Irish) from 8th c. AD As above Tier 3 GK Name only Mabinogion (Welsh) medieval As above Tier 3 GK Name only Written by Christian monks in Ireland and Wales, centuries after paganism ended there. Used only as evidence for a broader Celtic mindset, not specifically pre-Roman British belief. Post-Roman context only (outside project scope): Beowulf and Sutton Hoo (Anglo-Saxon belief), Augustine's mission (597 AD), Æthelstan (927 AD). All GK, name only. C. Archaeological sites and finds Site / find Location Period Evidence and use Stamp Provenance Completeness Star Carr Yorkshire Mesolithic, from c. 9300 BC Waterlogged organic preservation; antler headdresses; Britain's oldest known house structure (c. 8500 BC); 21 red-deer antler frontlets; excavated from the 1940s; University of York CONFIRMED (site); INFERRED (ritual) GK + WS Partial Howick Northumberland Mesolithic Large circular dwelling, radiocarbon dated c. 7600 BC CONFIRMED GK + WS Partial Caldey Island south Wales Mesolithic Isotope analysis of remains: heavy marine-protein diet; variation read as seasonal coast–inland movement INFERRED WS Partial Cheddar Man (Gough's Cave) Somerset c. 10,000 BC Genome sequenced 2018: dark-to-black skin, blue eyes CONFIRMED GK Partial Colinda harpoon North Sea Mesolithic Barbed antler point trawled up in 1931; first concrete evidence of a land connection to the continent CONFIRMED WS Partial Doggerland (submerged landscape) North Sea Early Holocene Sonar, seismic, cores; see section D CONTESTED (mechanism of loss) WS Partial Stonehenge Wiltshire Neolithic Solar/lunar alignments; implies organised labour INFERRED (meaning) GK Name only Avebury Wiltshire Neolithic As above INFERRED GK Name only Long barrows, chambered tombs, henges various Neolithic Ancestor veneration inferred from burial practice INFERRED GK Name only Llyn Cerrig Bach Anglesey Iron Age Swords, chariot gear, cauldrons deliberately deposited in a lake; corroborates classical claims of water veneration CONFIRMED (deposit); INFERRED (meaning) GK Name only Bath, Sulis Minerva Somerset Roman Curse tablets (people petitioning the goddess against thieves); Romano-British inscriptions naming Sulis CONFIRMED GK Name only Lydney, Nodens Gloucestershire Roman Inscriptions naming Nodens, used to work back to Iron Age religious geography INFERRED GK Name only London Mithraeum London Roman Mithraic temple CONFIRMED GK Name only Deities inferred from inscriptions and Iron Age/Romano-British syncretism: Epona, Sulis, Andraste, Cernunnos, Nodens. All GK; no inscription corpus was cited. D. Scientific studies and research programmes D1. Genetics and population Study What it showed Stamp Provenance Completeness 2018, Nature Ecology & Evolution First genomic data from British Mesolithic individuals (6 people), compared with 67 Neolithic individuals. Mesolithic Britons stayed genetically close to Western European hunter-gatherers (WHG / "Villabruna" ancestry) until separation from the continent. Neolithic transition absorbed essentially none of them, except for at least two Early Neolithic individuals in western Scotland with local hunter-gatherer ancestry. CONFIRMED WS Partial (no authors) 2022 Palaeolithic genetics study First Palaeolithic human genetic material from Britain: two distinct populations roughly 600 years apart, both arriving after the last Ice Age CONFIRMED WS Name only (no journal or authors) 2022, Cambridge Archaeological Journal Challenges the "clean replacement" narrative; minority position CONTESTED WS Partial (no authors) Cheddar Man genome, 2018 See section C CONFIRMED GK Partial Genetic sequence in the knowledge base: Mesolithic hunter-gatherers → largely replaced by Neolithic farmers (c. 4000 BC) → largely replaced by Bronze Age Beaker migrants (c. 2500 BC). D2. Mesolithic behaviour and method Source Point Stamp Provenance Completeness Grahame Clark, original Star Carr work "Seasonal round" model: winter lowland aggregation, summer upland dispersal following red deer. Built partly on ethnographic analogy, itself contested. CONTESTED WS Partial (no title or date) Star Carr and Howick structures Complicate the purely nomadic picture; suggest seasonal or partly permanent occupation CONTESTED WS Partial 2024 synthesis paper Most British Mesolithic evidence is undated lithic scatters; core methodological problem CONFIRMED (as a limitation) WS Name only Conneller, 2021 Typological scheme aimed at fixing the dating problem — WS Partial D3. Doggerland Source Point Provenance Completeness Vincent Gaffney (University of Bradford), "Europe's Lost Frontiers" / "Doggerland and the Lost Frontiers Project", 2015–2020 Seismic mapping, coring and sedaDNA to model the landscape from the start of the Holocene (~12,000 years ago) to submergence WS Partial Repurposed North Sea oil-industry seismic data Mapped ancient river valleys; 3D seismic surveys cost tens of millions of dollars, beyond normal archaeology budgets WS Name only Vibrocoring study 78 cores from 60 locations along one ancient river system WS Name only Cribdon, 2021 PhD thesis, University of Warwick sedaDNA from North Sea cores to reconstruct plant and animal profiles WS Partial Gaffney et al. (tsunami study) Topography surveys and underwater rock cores: the Storegga tsunami (c. 6200 BC) hit only small residual islands; gradual climate-driven sea-level rise was the real cause WS Partial Bensharada et al. (University of Bradford) Carbon dating of North Sea cores: sea level began rising roughly 10,000 years ago; multi-proxy method (thermal measurements, organic biomarkers, radiocarbon, chemostratigraphy) WS Partial Wessex Archaeology for Vattenfall Sediment cores from Doggerland channels, recovered during pre-construction surveys for Norfolk offshore wind WS Partial Assessment on record: the tsunami-as-final-blow theory is being displaced by a gradual-drowning model, but this is still an open question. Stamp: CONTESTED. D4. Sea level and Meltwater Pulse 1B Source Point Provenance Completeness Global sea-level rise, ~120 m from the Last Glacial Maximum (~20,000 years ago) to ~7,000 years ago Not in dispute; supported by coral cores, marine sediment cores, isostatic rebound data and the Doggerland record WS Name only Fairbanks, Barbados coral cores (1990s) Proposed MWP-1B: ~28 m in ~500 years around 11,300 years ago WS Partial Bard et al., Tahiti coral cores At most an acceleration in a steady rise; Barbados figure possibly a tectonic-uplift artefact WS Partial 2025 Great Barrier Reef core study Supports the Tahiti reading; no evidence of an abrupt >11 m jump WS Name only Svalbard ostracod cores, 2020 Evidence of a real but smaller rapid event in the same window WS Name only MWP-1A (~14,600 years ago) Comparatively well established; sits before the Mesolithic WS Name only Stamp: sea-level rise CONFIRMED; the shape of the curve at ~11,300 years ago CONTESTED. E. Institutions named University of York (Star Carr lead) · Historic England · British Museum · University of Bradford (Gaffney, Bensharada) · University of Warwick (Cribdon) · Wessex Archaeology · Vattenfall (commissioning body) F. Section-to-source map (knowledge base v4) KB section Principal sources 2.1 Doggerland Colinda harpoon; Gaffney; Cribdon; Bensharada; Wessex/Vattenfall; MWP-1B note (Fairbanks, Bard, GBR 2025, Svalbard 2020) 2.2 Who lived there Cheddar Man genome (2018) 2.3 Key sites Star Carr; Howick 2.4 Genetics 2018 Nat Ecol Evol; 2022 Palaeolithic study; 2022 Camb Archaeol J dissent 2.5 Diet Caldey Island isotopes 2.6 Settlement Clark; Star Carr and Howick structures 2.7 Ritual Star Carr antler frontlets 2.8 Methodological ceiling 2024 synthesis; Conneller 2021 2.9 Institutional grounding York, Historic England, British Museum 3 Neolithic Stonehenge; Avebury; barrows (all GK) 4.2 / 4.2.1 Iron Age and druids Caesar; Diodorus; Strabo; Posidonius; Pliny; Pytheas; Llyn Cerrig Bach; Irish and Welsh literature 5 Roman Britain Bath; Lydney; London Mithraeum (all GK) G. Open issues in the source base No URLs retained. Every WS item came from search results that were not saved. Re-run the searches to capture links before treating these as citable. Many citations are partial. The 2018, 2022 and 2024 studies need authors, titles and DOIs. Roughly half the list is general knowledge only (all of sections A and B, and most of C). No source text was checked. This matters most for the Tier 1 classical sources, since the project standard is primary-source grounded. Chronology tension to resolve. The knowledge base starts the British Mesolithic at c. 11,000 BC, but MWP-1B is dated ~11,300 years ago (c. 9350 BC), roughly 1,650 years after that start. The session described it as "right at the start", which only holds under the alternative convention (c. 8800 BC). Decide on one Mesolithic start date for the schema. Undated or unnamed items: the source for Boudica and Andraste; any inscription corpus for Sulis, Nodens and Epona; the Pliny passage. Candidates mentioned but not yet used as sources: Tacitus (suggested as a documentary source for pre-conquest practices in the schema discussion). Not carried over: NotebookLM supported-formats lookup (a tooling note, not a historical source). Changelog 2026-10-05 — v1 created. Consolidated from the 2026-09-09/12 knowledge base sessions (v1–v4 content, MWP-1B discussion) and the Ancient Britain schema discussion. No sources added beyond those named in the sessions.