This page is generated from site-status.json — the single source
of truth. Counts are typed so source records, live search rows, curated editions, and
AI-indexed items are never conflated. machine-readable JSON ↗
Count types
- live queryable live index (DB/API rows)
- snapshot frozen point-in-time export
- source raw source records before dedup
- curated hand-edited scholarly editions
- draft not yet reviewed
- ai-indexed items in an AI/bridge index (NOT inscriptions)
- reference reference-work nodes (LCL/BNP/CIL/PIR)
- teaching course/teaching units
- generated derived/generated catalogue
| Count | What it is | Type | Source · provenance |
|---|---|---|---|
| 11 families |
Top-level content families · 顶级内容门类
SINGLE SOURCE OF TRUTH for the sitewide family count. Was drifting 8/8/9/10 across home.html ("eight kingdoms"), builds.html ("eight families") and families/index.html ("nine kingdoms", ten cards). stamp_counts.py syncs every surface; check_family_counts.py gates it. Per-family page counts live in doc["families"]. |
generated | schema/family-*.json (families/index.html card taxonomy) |
| 511,973 rows |
Unified inscription list · 统一铭文列表
Search rows BEFORE cross-system dedup — NOT distinct inscriptions. |
live | list.sqlite (EDCS∪EDH∪TM metadata) |
| 81,883 rows |
EDH index · EDH 索引 | snapshot | edh.sqlite (Heidelberg, archival) |
| 66,079 rows |
Papyri index · 纸草索引 | snapshot | papyri.sqlite (TM-keyed) |
| 5,467 records |
KCL Matrix-Hub inscriptions snapshot · KCL 铭文快照 | snapshot | KCL IAph/IRT/IRCyr static snapshot |
| 2,248 works |
Literature works (CTS) · 文学作品 | snapshot | canonical greekLit/latinLit TEI |
| 6,617 normalised decree shards |
Formula-dossier decree shards · 套语法令切片 | generated | formula-dossier mining |
| 54 formulae |
Formulae with corpus counts (dossier bank) · 有语料计数之套语（正典库）
SINGLE SOURCE OF TRUTH for the sitewide "N formulae" claim — derived from the served dossier bank, all with corpus counts. schema/formula-network.json lists 58 uids: the closed-vocabulary superset incl. 4 Latin tier-2 formulae without corpus counts; only discovery.html / formula-browser.html (which display that artifact) cite 58. stamp_counts.py syncs every surface — was drifting 54/58/50. |
curated | matrix-hub/formula-dossier/dossier.html "F.x" bank (corpus-counted subset of schema/formula-network.json, 58 uids) |
| 21 case studies |
Formula-dossier case studies · 套语案例研究
Standalone interpretive deep-dives (the dossier's most reader-facing artefact). stamp_counts.py syncs every surface — was drifting 20/21/22 (22 = a glob that wrongly counted the case_studies/index.html landing). |
curated | matrix-hub/formula-dossier/case_studies/*.html minus index.html + TEMPLATE.html |
| 4 periods |
Formula-dossier periods · 套语涵盖时期
4 periods across 5 centuries. A "5 periods" surface was the drift — 5 is centuries. |
curated | Classical Greek · Hellenistic Greek · Republican Latin · Imperial Latin |
| 669,000 indexed items |
Symbolon cross-evidence bridge · Symbolon 跨证据桥
AI cross-evidence INDEX items across 5 streams — NOT inscriptions, NOT one vector space (identity/concept bridge; see Symbolon §0.5). |
ai-indexed | bridge_*.jsonl (grc/lat/pap/lit/ed, concept+identity) |
| 188 artifacts |
Identity-spine artifacts · 身份脊柱条目
By class: {'literary-passage': 94, 'edition-collection': 1, 'inscription': 92, 'section-landing': 1}. |
curated | registry.json (register.py harvest of #magalia-id blocks) |
| 186 records |
Catalogue records · 编目记录 | curated | catalogue.json |
| 42 curated editions |
Governance curated editions · 治理类精编版本
Canonical curated count of the interlinked Greco-Roman public-documents core (17 BC–AD 247). A DIFFERENT categorisation from governance_stamped (all stamped pages sitewide); the Matrix Hub counts-explained note defines both subsets. |
curated | governance-corpus.json |
| 81 stamped editions |
Governance stamped editions · 治理类盖印评注本
SINGLE SOURCE OF TRUTH for the sitewide "N stamped editions" claim. stamp_counts.py syncs every surface to this; do not hand-edit those numbers. |
generated | matrix-hub/governance/*.html (pages with #magalia-id) |
| 41 photogrammetry captures |
3D inscription scans · 三维铭文扫描
8 fully spine-joined (EM⇄IG⇄PHI⇄edition); 17 carry the EM↔IG pair. The homepage '32 inscription + 9 sculpture' splits this same set — reconcile to 41. |
curated | models3d-identity.json + models3d-join.json |
| 1,674 reference artifacts |
Reference-work nodes (LCL/BNP/CIL/PIR) · 参考文献节点
LCL 28 · BNP 1504 · BNPSII 128 · CIL-vol 6 · PIR-vol 7. |
reference | registry-references.json (build_reference_registry.py) |
| 183 databases/projects |
DH database/project directory · DH 数据库目录
Homepage prose must read this number (was drifting 146 vs 183). |
curated | dh-tracker |
| 11 bundles |
Live project bundles · 在线项目集
Header chip + prose + meta must all read this. The homepage prose enumerates 11 live bundles: §IX Greek Historians and §X Attic Orators are SEPARATE sections (separate historiae/ and orators/ dirs). Reconciled 10→11 on 2026-06-16 — the source list previously omitted Attic Orators. To revert, fold Orators into Greek Historians. |
teaching | site sections (Happy Latin · Cicero · Pottery · Aesop · Periplus · Matrix Hub · Latin Exam Prep · DH Tracker · ML/AI · Greek Historians · Attic Orators) |
| 100 gold items (abstention sample) |
Symbolon/assistant honesty eval · Symbolon 诚实性评测
From the own-ground-truth eval (eval/RESULTS.md). The validation page is SHIPPED and live at /symbolon-validation.html (HTTP 200, verified 2026-06-16) — was mislabelled "(planned)". |
ai-indexed | eval/ harness (P2) — Qwen3.7-plus: abstain 1.00, 0 hallucinations |
| 6 systems with verified deep-link templates |
Outbound ID systems (record-linkable) · 可直链外部标识系统
Of 39 ID systems in registry.json equivalents; the remaining 33 are print corpora expanded via sigla-index (no fabricated URL). |
generated | schema/id-systems.json (build_id_systems.py) |
| 144 partner corpora a TM-bearing text resolves across |
Cross-corpus federation (TM-partner) · 跨语料联邦（TM 合作语料）
5 carry a live-verified per-id deep-link; the rest resolve through the Trismegistos relations hub (no fabricated URL). |
reference | schema/id-systems.json federation (FAIR-epigraphy datasources) |
| 122 sigla → full citation |
Sigla with full attribution · 可展开著录的著录略语
89 from EpiBib Heidelberg + 36 hand-curated (3 flagged uncertain). |
reference | schema/sigla-index.json (EpiBib + supplement) |
| 537,262 Latin inscriptions scanned |
EDCS inscriptions audited for restoration · 已审计修复率的 EDCS 铭文
Finding: 11.5% of legible letters are editorial restoration; 27.5% of inscriptions carry conjecture; 14,241 flagged. 'History from Square Brackets'. |
generated | edcs-restoration-auditor → matrix-hub/restoration-atlas.json |
| 132,909 geotemporal Latin inscriptions |
Dated + located Latin inscriptions (LIRE) · 有纪年且定位之拉丁铭文
97.6% of LIRE's 136,190 features are dated+located. Epigraphic-habit peak 150–175 CE. A geotemporal VIEW of the merged record — overlaps the EDH/EDCS snapshots, not additive. |
generated | LIRE v1-2 (EDCS∪EDH merged) → matrix-hub/epigraphic-habit.json |
| 15,061 PIR² persons |
Imperial prosopography persons (PIR²) · 帝国群体传记人物（PIR²）
3,988 carry a cursus honorum; top attested office Consul (cos.) (n=1631). Citation+offices authority (names abbreviated in open export); DPRR Republican layer not yet ingested. |
reference | PIR-public export → schema/person-authority.json |
| 477 years |
Consular fasti years (T1, DPRR) · 执政官名表年数（T1，DPRR）
-509..-20; 867 consul + 69 suffect attestations; 55 eponymy-exception years. T1 Republic data-driven; T2/T3 not yet built. |
reference | DPRR data.sql → build_fasti_republican.py → fasti/consular.json |
| 8 figures |
Chronology figures linked to the fasti (DPRR) · 已连接执政官名表之编年人物（DPRR）
curated figures resolved to their exact DPRR person (homonym-safe); consular years backfilled into figures-authority.json; H7 round-trip clean. |
reference | build_fasti_join.py → fasti/figures-backfill.json + figures-authority.json |
| 42,121 ancient places profiled |
Ancient places (Pleiades gazetteer) · 古代地名（Pleiades 地名辞典）
34,563 located (82%); the spatial backbone beneath the site — 81 of these places carry Magalia editions (enriched with Pleiades type/precision). Spatial profile only. |
reference | Pleiades v4.1 → matrix-hub/pleiades-atlas.json |
| 23 centers scored on the enclave rubric |
Roman production centers scored · 已评分之罗马生产中心
1,399 attested OXREP mines mapped + dated; 23 centers scored on the 7-axis enclave rubric, 12 qualify as bounded fiscal-legal production enclaves. Citations P5-verified against EDCS. |
finding | roman-economy-centers dossier (OXREP Mines v3.0 + CEIPAC + LIST/EDCS) -> matrix-hub/production-centers.html |
| 6,617 decree inscriptions |
Greek decree inscriptions (formula-traced) · 希腊政令铭文（套语追踪）
4,370 dated; 22 commonest formulae traced across centuries/regions; peak century 2c_BCE (n=1691). Overwhelmingly Greek, Hellenistic-weighted, Attica-centric. |
generated | formula-dossier → matrix-hub/formula-diffusion.json |
| 91.5 % exact (Aeneas, leakage-free) |
Latin restoration benchmark (best exact) · 拉丁修复基准（最佳精确率）
Specialist Aeneas 91.5% exact, ECE 0.0531 (calibrated); general LLM 5.5% exact at ECE 0.818 (confidently wrong). Surfaced from the existing eval/ harness; no model run by the page. |
ai-indexed | eval/runs/*-metrics.json → matrix-hub/latin-restoration-eval.json |
| 4,784 EpiDoc inscriptions |
Sicilian inscriptions profiled (I.Sicily) · 已剖析之西西里铭文（I.Sicily）
9 languages (Greek/Latin/Sikel/Elymian/Punic/Oscan/Hebrew), 3,813 dated, 19 bilingual — the multilingual contact zone across the centuries. |
reference | I.Sicily → matrix-hub/sicily-languages.json |
| 98 EpiDoc inscriptions |
Praetorian-prefecture inscriptions profiled (PPRET) · 已剖析之禁卫军长官铭文（PPRET）
76 dated (289–450 CE), 8 text-types (top Honorary 63.3%) — the late-antique administrative epigraphy of the empire's top civil officials. |
reference | PPRET → matrix-hub/ppret-prefecture.json |
| 397 EpiDoc inscriptions |
Cretan institutional inscriptions profiled (I.Cret) · 已剖析之克里特制度铭文（I.Cret）
373 dated (700 BCE–212 CE), 271 distinct institutions named 3,149 times (top kosmos 554) — the civic vocabulary of the Cretan poleis. |
reference | I.Cret → matrix-hub/cretan-institutions.json |
| 66,079 papyri & ostraca |
Papyri profiled by century & findspot · 已剖析之纸草（分世纪与出土地）
65,077 dated across 1,974 findspots; peak 2nd c. CE (n=15619). The documentary habit — the everyday twin of the inscribed record (Oxyrhynchus, the Fayum, Thebes). |
snapshot | papyri.sqlite → matrix-hub/papyri-distribution.json |
| 27,953 epitaphs |
Latin epitaphs profiled (Dis Manibus) · 已剖析之拉丁墓志（Dis Manibus）
25,191 dated; the Sacrum variant (D M S) is a western/African signature — Baetica 96% vs the Danube/Rhine ~1%. Filtered to EDH's epitaph class; Sacrum over legible-formula subset. |
snapshot | EDH dis_manibus → matrix-hub/funerary-formula.json |
| 100 leads |
Embedding-outlier discovery leads · 嵌入离群点发现线索
model-only, unreviewed queue: 18,387 of 30,000 sampled Greek texts scored across 222 (region × century) cells; high score = locally isolated in embedding space, NOT a proven error. |
ai-indexed | phase-16 semantic index → matrix-hub/embedding-outliers.json |
| 80 candidate clusters |
Candidate formulae (open-vocabulary discovery) · 候选套语（开放词表发现）
REVIEW-ONLY curator queue over 4,434 register decrees vs background; nothing is auto-promoted into the 58-formula bank; every stored attestation offset re-slices or the build fails. |
generated | I.PHI register mining → schema/formula-discovery.json |
| 12 PROIEL texts |
Treebank texts profiled for word order · 已剖析语序之树库文本
Object-before-verb share: classical Latin 85% (verb-final) vs Greek 40%; the Vulgate 37% calques Greek order. One construction (obj↔verb), no model. |
generated | PROIEL treebank → matrix-hub/word-order.json |
| 3,048 numbered §§ |
Smyth's Greek Grammar sections · Smyth 希腊语法节数
All 3048 canonical sections (§1–§3048) + 213 dialect notes; betacode converted to Unicode polytonic, cross-source-verified vs the 1920 Textkit edition. The Greek companion to Woodcock's Latin syntax reference. |
reference | neelsmith/smyth (Perseus-derived TEI) → greek-grammar/smyth.html |
| 129 comparanda |
Comparative Greek–Latin comparanda · 希腊-拉丁比较项
Across syntax 81 (Whiton's parallel rules), morphology 12 (Sihler), semantics 10 + pragmatics 12 (Pinkster/Oniga), vocabulary 14 (Mason). Anchored on Smyth (Greek) + Woodcock (Latin) §§; the four sampled dimensions are page-cited, not bulk-OCR'd. |
reference | Whiton/Sihler/Pinkster/Mason/Oniga → comparative-grammar/comparative.html |
| 29 works |
Secondary-literature works cited · 所引二手学术著作
Grounds 13 pages; each work linked to a page only where that page genuinely relies on it (MacMullen 1982→epigraphic habit, Badian 1989→restoration, Whiton/Sihler/Pinkster/Mason/Oniga→comparative grammar, LIRE/EDH/PROIEL/I.Sicily/Pleiades/PIR→the data findings). Distinct from registry-references.json (Loeb/Pauly/CIL/PIR series). |
reference | bibliography.json (scholarship + dataset authorities + reference grammars) |