# Milestones and road alignment: feasibility and admission rules

Assessment date: 17 October 2026. These tiers are an editorial assessment framework for this atlas, not an ancient classification, a probability scale, or a substitute for reading the object record. The numbered sequence controls display and review priority. Each assignment needs an explicit publication locator and a short explanation. An empty high tier is preferable to an unsupported admission.

## Feasibility and the intended inference

The proposal is feasible and useful. It gives priority to monuments that can constrain where a road passed and makes that reason visible. A dedicated **rupestral milestone** facet should identify inscriptions cut into rock, while the alignment tier should assess the evidence connecting that inscription to a local road. These are separate questions.

Rock-cut lettering on an undisturbed outcrop can fix the inscription's ancient position much more securely than a transported column. It still does not, on its own, establish that the text is Roman, that it functioned as a milestone, which side or phase of a road it accompanied, or the exact centreline of a pavement. A detached block, movable boulder, modern re-erection, or inscription face relocated for conservation does not acquire an ancient position from its present coordinates. A well-excavated installed column can supply better local evidence than an inadequately documented rock text. T1 and T2 are therefore two strong contextual classes with different safeguards, rather than universally comparable degrees of certainty.

The term “milestone” also needs scrutiny. Alarcão reports that Hübner interpreted the rock inscription CIL II 6211 beside the Paiva as a milestone, but considers it more likely honorific. It should not become a primary road-alignment anchor simply through the label “rupestral.” This is a modern scholarly assessment, not a new autopsy of the CIL entry. [Alarcão 2005, *Conimbriga* 44, p.160](https://ap1.sib.uc.pt/bitstream/10316.2/37810/1/o_territorio_dos_paesvri_e_as_suas_principais_povoacoes.pdf#page=15).

## Cascading categories

| Tier | Display label | Minimum evidence | What it can support |
|---|---|---|---|
| T1 | Fixed rupestral milestone with a documented road relation | Accepted Roman milestone identification; publication documents inscription in original bedrock and its direct local road relation; no unresolved support displacement | Strong local spatial constraint at the ancient inscription position; exact pavement geometry and date still require their own evidence |
| T2a | Milestone in a documented primary road setting | Accepted Roman milestone; publication documents original installation, socket, standing base, or equivalent primary context beside independently identified Roman road remains | Strong local association between marker station and road; not every road segment or roadwork phase |
| T2b | Milestone found with documented Roman road remains | Publication directly ties a local discovery to physically identified Roman road remains; no demonstrated transport or incorporation as displaced spolia before the assessed discovery; later or on-site reuse remains separately flagged; original installation unproved | Strong local co-occurrence; station, exact placement, and contemporaneity remain qualified |
| T3 | Published road association; physical context unresolved | Specific published object-to-road attribution or conditional original-position argument, with source and limits; physical co-occurrence not adequately established | A reasoned road/corridor hypothesis, without promotion to an excavated alignment anchor |
| T4 | Specific discovery context; original road setting unproved | A specific published discovery or recovery place, including reuse contexts; no sufficiently supported original road relation | Discovery/reuse geography; proximity may be explored but cannot recover original roadside placement |
| T5 | Broad locality or corridor only | Published town, region, estate, or named corridor; no admitted specific original find position | Regional distribution at the source's precision |
| T6 | Collection, custody, or proxy location only | Only repository, historical collection, inherited catalogue coordinate, or reporting locality is locatable | Object access, reporting history, or catalogue navigation |
| U | Unreviewed or insufficient evidence | Relevant provenance or support claims not yet inspected | A review queue; absence of review is not evidence of a low-quality ancient context |
| X | Excluded from milestone alignment inference | Serious unresolved authenticity/function problem, or accepted non-milestone classification | A visible comparison or disputed-record dossier, excluded from milestone-anchor claims |

T2a and T2b share the second-tier family but must be visibly distinguished. “Found with a Roman road” can describe an original marker, a fallen nearby stone, or a reused road component. Those circumstances do not have the same implications. At Rantis, the excavator reports milestones and pedestals incorporated into the road, including a milestone used as a curbstone. That is good evidence of reuse in road fabric; it does not recover the original mile station. The explicit displaced-spolia gate overrides a generic “found with road” description. Later or on-site fence-post reuse without demonstrated transport does not, by itself, erase a publication-documented same-site Roman-road association. Such a borderline admission remains T2b, with reuse flagged and original installation unproved; it never establishes an unchanged primary socket. CIL II 4644 at Casa de Campo is a borderline application of this distinction: publication-reported road association and fence-post reuse need to be shown together, while excavation records or fuller discovery documentation remain needed to resolve the precise original station. [Paredes Martín 2023, p.41 / ](https://revistas.uned.es/index.php/ETFII/article/download/36712/28064#page=11). The contrasting Rantis reuse context is documented by [Yossi Elisha, “Rantis (Northwest),” *Hadashot Arkheologiyot* 123 (2011), road description, Table 1 and figs.5–7](https://hadashot.iaa.org.il/Report_Detail_Eng.aspx?id=1826).

If a stone was securely documented in a primary roadside setting before a modern transfer, its historical T2a relation may remain. Its present museum point cannot stand in for that position. Likewise, T1 refers to a documented ancient support position, not to every later display position of a cut-out rock inscription. An admissible original-location geometry must carry its own source.

## Keep the other dimensions independent

1. **Support form:** original bedrock; detached rock face; movable boulder; freestanding column; fragment; unknown. “Rupestral” describes the carved support, while physical continuity with bedrock is an independently supported claim.
2. **Function and authenticity:** accepted milestone; probable/disputed milestone; other road inscription; honorific/boundary/non-road text; disputed authenticity. A fixed road-building inscription can be important in a separate road-inscription category without being renamed a milestone.
3. **Location role and precision:** original station, findspot, secondary reuse, recovery, historical custody, repository, or catalogue proxy. Preserve all events. Unknown accuracy remains unknown; decimal places, zoom level, and a nearby road do not supply metres of precision.
4. **Temporal evidence:** inscription phase, original road installation, works claim, erasure, reuse, discovery, and relocation are separate events. A spatially fixed but undated Roman text can constrain location without validating a dated road phase. A broad phase envelope is not evidence of continuous road operation.
5. **Source independence:** the milestone and the Itiner-e segment may descend from the same discovery report or edition. Agreement then shows consistency with the reconstruction's input, not an independent test. Independence is assessed per segment and claim, not once for an entire object.
6. **Inference scope:** point; local road association; corridor; segment; whole route. No single marker automatically validates the complete connecting polyline, caput, road surface, imperial workforce, or route history.

RIB 2283 illustrates why narrative provenance outranks a generic metadata label: its record includes an “in situ” institution field, but describes re-erection about 180 m from discovery and later movement. The atlas must preserve that movement chain. [RIB 2283, Support: Find context, Other events, Modern location; Commentary](https://romaninscriptionsofbritain.org/inscriptions/2283).

The independence check is necessary because Itiner-e's authors explicitly report using milestones, earlier reconstructions, archaeological evidence, and other sources in road identification and digitisation. The methods also describe inaccurate milestone positions and missing chronology. A nearest-line result does not resolve that source genealogy. [Itiner-e dataset paper, Methods: road digitisation, steps one to three](https://www.nature.com/articles/s41597-025-06140-z).

## Chronology, roadwork, and actors

The tier measures a location relationship. It must not silently upgrade the inscription's date, an emperor's practical role, or the date of the nearby road fabric. Maintain the Latin action and grammatical subject: an imperial name or honorific dedication is not automatically a construction event; even an explicit works formula requires an evidence-backed local interpretation. Kolb distinguishes construction, repair, improvement, and erection of milestones, while treating road archaeology as necessary for an itinerary's exact course. [Anne Kolb, “Via ducta – Roman Road Building,” in *Roman Roads: New Evidence – New Perspectives* (2019), pp.11–13](https://doi.org/10.1515/9783110638332). These pages were checked in the source edition, .

Road fabric and milestone chronology can diverge even in an extensively investigated place. The Via Augusta excavation at Ianus Augustus identified successive road deposits, but the authors could not securely date each sequence and cautioned that the geographical scope of epigraphically reported repairs remains unknown. Such correlations should remain relative or conditional when direct stratigraphic dating is absent. [Gutiérrez-Rodríguez et al. 2022, “Microstratigraphic analysis … Ianus Augustus,” Discussion, subsection on reconstructing the Via Augusta and its repairs](https://link.springer.com/article/10.1007/s12520-022-01602-6).

## Atlas presentation and bounded audit

Each marker should show its tier label, subtype, assessment date, source locator, reasoning, and unresolved gates beside its timeline and location history. Provide independent filters for rupestral form, tier, secondary reuse, and coordinate fitness. At T2b show “Original installation unproved” prominently. Allow top categories with no admitted point to remain searchable in the dossier; lack of a usable coordinate must not erase the scholarly category.

For comparison with Itiner-e, expose the original-position geometry source, its precision, the selected segment identifier and version, that segment's certainty and bibliography, temporal compatibility, and source-independence status. If these are unknown, show them as unknown. Any measured separation remains exploratory; a discrepancy may reflect network geometry, movement, mistaken attribution, or imprecise coordinates.

The audit should reject automatic promotion from material words, database object type, museum location, road proximity, or an isolated “in situ” field. Each high-tier assignment needs a direct source locator. Raw catalogue records remain U until reviewed, rather than acquiring a false low or high ranking. Tiers can be revised in either direction when new context, relocation evidence, or a reclassification is published.

## Consultation record

- Newly inspected online: Alarcão 2005 p.160; Elisha 2011 final report; RIB 2283 provenance fields; Itiner-e Methods; Gutiérrez-Rodríguez et al. 2022 Discussion. These are direct scholarly publications or editions, not search snippets used as evidence.
- Newly inspected local scholarship: Kolb 2019 pp.11–13/ and Rathmann, “Miliaria in der Provinz Lusitania,” pp.303–306/. Rathmann stresses integrating corpus editions with archaeological and infrastructure evidence; his statements about the state of editions and counts are historical research status, not current corpus totals.
- Existing project controls inspected: `research/milestone-research.md`, `local-inventory.md`, `local-inventory.json`, `local-cases.json`, and `milestone-cases.json`. These supplied context and paths; new high-tier object assignments remain the responsibility of the separate case audit.
- A local CIL II Supplement scan was searched for no.6211 but yielded no reliable entry transcription during this bounded methods task. CIL II 6211 is therefore cited here through Alarcão's explicit discussion. No new CIL-autopsy claim is made.

Machine-readable rules: `research/hierarchy-criteria.json`. Object assessments must additionally retain the evidence source and scope, rather than reporting a tier code alone.