# Expanded miliaria dossier and atlas audit 17 October 2026 · Second research edition The dossier expands from twelve to **38 physical-object records**: ten additional British objects, eight Hispania cases, and eight Italian/Alpine cases. Cesiomaggiore is an explicitly contested road-commemoration comparator; inclusion records the publications and disagreement, not an authentication. The source-preserving transport and bulk epigraphy layers retain their original scope. ## What is now available directly on the atlas Click a typed source pin, choose an object or location in **Checked site locator**, or open a checked result from the atlas search. The map inspector exposes **Object**, **Timeline**, **Locations**, and **Sources** without leaving the atlas. Location cards can pinpoint their own source assertion. Shared-coordinate markers expose every object, including the two Llanfairfechan stones and several museum records; shared coordinates never establish co-discovery. The source panel includes nine rendered pages of the public-domain CIL II (1869) and CIL X/2 (1883) editions. They open in an in-app dialog with full-size viewing. Original scan hashes, PDF/printed page mappings, image hashes, and rendering parameters are in [cil-page-manifest.json](cil-page-manifest.json). Modern publications are cited and linked, not redistributed as page images or PDFs. The expanded data model contains **46 inscription phases**, **74 object-history events**, **four alternative dating assignments**, **112 location assertions**, and **426 claim/source ledger entries**. Sixty-six location assertions have source coordinates; this includes 36 optional catalogue-locality anchors. Coordinates can coincide between assertions. A source-provided coordinate is not necessarily a surveyed archaeological position, and a repository point can identify only an institution. All uncertainty notes remain attached. ## CIL and publication consultation - **Hispania:** the original CIL II scan was read at printed pp.621,632–633,635,640. New object records include 4644,4736,4737=4739,4740,4756,4799,4802,4803. Mantas 2015/2022 and Paredes Martín 2023 provide subsequent readings, chronology, movement history, and, for 4644, an explicit observation coordinate. The old 4735 pilot record was revised using Mantas 2015. Exact locators and consultation limits are in [the Hispania research audit](expansion-hispania.md). - **Sardinia and Cesiomaggiore:** original CIL X/2 printed pp.832–834,836 / were visually checked for seven stones. Institutional records and later scholarship supplement them. Cesiomaggiore's CIL V concordance remains inherited from the university/institutional sources; Basso's original 2019 chapter was directly checked at printed p.417 / local for the unproven authenticity hypothesis. See [the regional audit](expansion-italy.md). - **Britain:** fourteen RIB text editions support ten new physical-object dossiers. CIL VII concordances remain explicitly inherited from RIB; the original CIL VII volume was not directly consulted. The three texts RIB 2252/2253/2256 and three texts RIB 2290/2291/2292 are grouped by physical support. Object and institution map coordinates remain separate. See [the British research audit](expansion-britain.md). ## Material chronology and location findings **CIL II 4735:** the CIL 1869 assignment 134–135 and Mantas 2015's 133–134 series interpretation are alternative datings of one inscription. The text/ordinal problem remains visible. The move to the parish cemetery around 1894, further cross-base cutting, and the 2015 published location report are separate events. Neither the publication date nor the review date becomes a transfer or observation date. **CIL II 4644:** Paredes Martín 2023 prints 38°59′15.58″N,6°19′15.96″W for the stone observed20 November 2020 at Casa de Campo. Decimal conversion is documented. Source datum and positional error are not supplied; the point is a published observation/display position. The proposed original setting remains a separate interpretation without an invented coordinate. The old Claudian repair text is manuscript-dependent and contested in the corpus; the 2020 observation does not date that inscription. **Four catalogue anchors withheld:** the Mérida city anchor for 4644 is superseded by the specific publication point; the Alvarelhos and Vilela anchors conflict with the documented locality contexts; Orri's east-coast Lido di Orrì coordinate conflicts with CIL's southern Nora–Karales context. [The quarantine ledger](location-quarantine.json) preserves the source coordinates, case links, reasons and action. These four points are withheld from map display and road-distance comparison, while raw catalogue records remain searchable and downloadable. **CIL II 4737=4739:** duplicate historical numbering is one support, with a broad Hadrianic assignment and a separate conditional modern135proposal. They are not separate stones or two successive inscriptions. **Roadwork and actors:** explicit restoration, construction, supervision, road designation, and imperial address are distinguished. CIL II 4756's roads-and-bridges restoration and Decius's administrative role do not establish paving material, individual bridge locations, or labour organization. Lost/restored verbs in other records remain qualified; adjacent texts are never merged to invent a work claim. **British contexts:** RIB 2271's corrected location supersedes Hübner's Castle Hill attribution; RIB 2265 retains its titulature anomaly; RIB 2290's undated erased phase remains undated. Wall reuse, river recovery, repository institutions, and catalogue proxies do not offer the primary road-position comparison action. **Cesiomaggiore:** Basso 2019 proposes an early modern imitation of Rabland and explicitly says the hypothesis is difficult to demonstrate. The atlas records a directly supported scholarly challenge, with qualified transmitted chronology, not a verdict of forgery. Villa Tauro's coordinate identifies the repository; the church reuse and modern replica remain distinct assertions. ## Verification The expanded build passed **45 data/source admission checks** and **36 browser checks**, with no browser runtime exceptions. Checks cover source-byte identity, source-preserving geometry, duplicate grouping, ordered date bounds, phase gaps, alternative datings, assertion/source references, context eligibility, coordinate quarantine, CIL image hashes, source-dialog interaction, actual pin and shared-pin clicks, direct atlas inspection, modern-event filtering, repository exclusion, all four basemaps, data exports, keyboard tab navigation, mobile overflow, and unavailable external tiles. Desktop and mobile renders were inspected; the narrow-screen layer layout was corrected after visual review. Detailed machine-readable reports are in `qa/data-audit.json` and `qa/browser-results.json` beside the project. Metadata and browser checks establish a usable evidence instrument, not archaeological truth or exhaustive bibliographic coverage. ## Remaining limits Many CIL find contexts are historical or secondary. Specific original roadside coordinates and current custody remain unknown for numerous objects. Source-overlap between each milestone and each Itiner-e segment is not systematically resolved. Some later publications were available only through indexed text or an institutional abstract; each source's actual consultation status is recorded. RIB institution points establish the institution's location, not current gallery or storage presence. Corpus consultation remains selective; no comprehensive CIL XVII collation or independent scholarly certification is claimed. The original pilot files are retained. Transformations normalize whitespace for display, add typed assertions and cross-links, and quarantine identified coordinate problems; no road geometry or raw source metadata is moved to improve apparent agreement.