# Milestone evidence pilot: research and inspection record

Prepared 17 October 2026. This online research component contains **nine selected objects**, not an empire-wide milestone census. The machine-readable records are in `milestone-cases.json`. Six carry coordinates taken from the RIB website; three remain without an admitted WGS84 point. Geographic incompleteness is retained for review.

## What the evidence permits

The proposed comparison is feasible if each map symbol opens an explicit statement of its evidentiary role. A milestone can attest an inscription, an imperial intervention claim, a measured distance, a recovery context, or a later reuse event. Those propositions require separate fields. Their union does not establish the exact alignment, building date, pavement, or uninterrupted existence of a road.

A named emperor normally supplies an honorand or authority. It does not identify the engineer, contractor, work unit, local contributor, or actual workforce. `fecit`, `refecit`, `restituit`, and `munit` should remain visible with their objects and restorations. Imperial titulature without an intervention verb does not support a construction event. Even an explicit roadwork claim is a claim made by an inscription, requiring comparison with excavated fabric and context before assigning a local construction type.

Dates also require separate dimensions: inscription phase, event claimed in its text, discovery, relocation, erasure/reinscription, and reuse. A phase interval is not a continuous period during which every mapped road segment operated. Stones with successive inscriptions need object identity plus multiple phase records, rather than duplicated monuments.

The map must distinguish findspot, inferred original emplacement, reuse findspot, repository, and catalogue locality. A repository point cannot challenge a route reconstruction merely through proximity. Absence of recorded stones is similarly inconclusive without a survey and preservation model.

## Case inventory and exact source locators

| Pilot ID | Main identifier | Inspected publication locator |
|---|---|---|
| rib-2244 | RIB 2244 = CIL VII 1169 | RIB Support; Edition 5–6; Apparatus 5; Commentary 2–3 |
| rib-2224 | RIB 2224 = CIL VII 1150 | RIB primary/secondary editions; Date versus Commentary |
| rib-2272 | RIB 2272 = CIL VII 1175 | RIB Support; Edition/Apparatus 5 |
| rib-2283 | RIB 2283 = EE III 119 | RIB Support; Apparatus; Commentary; Tomlin, *Britannia* 48 (2017), p.490(c), cited by RIB |
| rib-2298 | RIB 2298 = CIL VII 1186 | RIB Support; Edition 7–8; Apparatus |
| rib-2301 | RIB 2301 = EE VII 1110 | RIB Condition; four editions; Commentary; RIB 2299 Support for group context |
| tre-fontane-2009 | Ceraudo 2009 | *Epigraphica* 71, pp.107–117, especially pp.109,111–114, figs.1–3 |
| gerulata-2011 | Kovács and Schmidtová 2011 | *Slovenská archeológia* 59/2, pp.311–318; English summary pp.317–318; figs.1–6 |
| rabland-cil-xvii-4-1 | CIL V 8003 = CIL XVII/4, 1 = IBR 465 | Official CIL XVII/4,1 index ; Grenoble university transcription; Bozen municipal museum notice no.1 (February 2012) |

Each JSON case supplies a direct URL, locator, short Latin passage, newly written English rendering, and explicit source-inspection status. No AE number was assigned to the two newly published cases without verification. Related CIL examples cited in an article are not identifiers of that article’s new object.

## CIL consultation and its limits

The official [CIL volume catalogue](https://cil.bbaw.de/en/homenavigation/the-cil/volumes) was inspected to establish the relevant series. In particular, CIL XVII/1,1 covers Hispania citerior (2015), XVII/2 Narbonensis/Galliae/Germaniae (1986), and XVII/4,1 Raetia/Noricum (2005). The catalogue now lists the Pannonian fascicle XVII/4,3 (2025). Older digital publications therefore cannot establish the current completeness of CIL coverage.

The official [CIL XVII/4,1 index](https://cil.bbaw.de/fileadmin/user_upload/Archiv/Index_XVII4_1.pdf) was downloaded, text-extracted, rendered, and inspected. Its first page explicitly assigns entry 1 to Claudius in 46 CE. This is direct consultation of an official CIL index, **not inspection of the full CIL object entry**. RIB’s CIL VII concordances are publication-reported; the original VII pages were not inspected by this research component. Local-volume checks undertaken elsewhere in the project must be documented separately.

## Audit findings

- A web extraction of the [Ceraudo PDF](https://www.epigraphica.it/wp-content/uploads/2025/01/2009-LXXI-min.pdf) produced an extra X in the Tre Fontane numeral. The downloaded and rendered printed p.111 reads `[X]XVI`, consistent with the 26th-mile caption. The admitted case uses 26. Search snippets must not decide numeric readings.
- The [older Grenoble Rabland page](https://droitromain.univ-grenoble-alpes.fr/Varia/via_claudia_Augusta_CIL1.htm) carries a tentative 47 CE heading; the official CIL index controls this pilot’s dating. The museum notice independently supplies the custody history.
- [RIB 2224](https://romaninscriptionsofbritain.org/inscriptions/2224) has an apparent structured-date inconsistency. Both values are disclosed in the record.
- The RIB HTML `mapSetup` function provides location coordinates. Inspection of its public script showed that its `300` argument must **not** be treated as a 300 m accuracy estimate. The pilot leaves accuracy unknown.
- [RIB 2283](https://romaninscriptionsofbritain.org/inscriptions/2283) narrates displacement despite an institution label reading in situ. Narrative provenance takes precedence in this pilot.

CIL, EDCS, EDH and modern publications may transmit the same edition or discovery report. Agreement among those services is not automatically independent corroboration. A future comparison should carry source genealogy, a reproducible spatial query, route-version identity, temporal compatibility, and an explicit statement of whether the milestone was used in constructing the route under test.

## Retained downloads available to the build coordinator

The following temporary files preserve the inspected source bytes. They are not redistributed inside the HTML by this research component; the coordinator may retain them for local audit with appropriate source attribution.

| Temporary file | SHA-256 |
|---|---|
| `[workstation path omitted]` | `ad87e8c32ba67b9644b8a66c175283d741772bd10b8eb7968ef1b38120fa71bd` |
| `[workstation path omitted]` | `6700bb20f26270a2a3b3d5bbccdb0959d5ec4a1aa737b967f501da7da624a4b5` |
| `[workstation path omitted]` | `567ead9b7cd86e5756ad79344066b90753d4712b27eb13f80dba6fed29d98803` |
| `[workstation path omitted]` | `29521a6a03cc4a5ba9fb67db73bdb41f8848a4f02e80c0abf54e87e89d075de0` |

The [Gerulata article](https://cevnad.sav.sk/aktivita_1_1/slovenska_archeologia_2011_2.pdf#page=111) was inspected through its primary PDF and English summary. The [Bozen museum notice](https://www.gemeinde.bozen.bz.it/ocmultibinary/download/21100/1152538/34/91845209ef6876ae1816343948ac3662.pdf/file/6ed9625322207d7a0b78ba2647714d5a.pdf) required a direct download after the web renderer failed. Failed fetches did not become admitted evidence.