Site location and discovery

Sparked by the discovery of a lead ingot (hereby referred to as the Rossett lead pig) by a metal detectorist at Cobblers Lane, Wrexham, in October 2019,Footnote ⁴ Phase 1 of the In the Footsteps of Trebellius Maximus Project began with geophysical investigations of the immediate surroundings of the findspot, as well as at another nearby field, which had also been producing a collection of Roman metal-detected finds (hereafter known as ‘Stringer’s Lane’). Both sites have led to an ongoing project led by archaeologists at the University of Chester and Wrexham Museum to investigate Roman rural settlement and activity in north-east Wales.

The two sites, while as yet seemingly unrelated other than their geographic proximity, are situated 1.7 km apart and include a field to the north of Cobblers Lane near Broad Oak and a field to the west of Stringer’s Lane near Burton Green (Fig. 1). Geophysical surveys of both sites were conducted in September 2020 by Archaeology Survey West on behalf of the University of Chester and subsequent excavations took place at both sites in 2020 and 2021, involving staff from University of Chester, Wrexham Museum, PAS Cymru, Archaeological Survey West and local volunteers. The aims were to investigate the immediate findspot of the Rossett lead pig through geophysical survey and targeted excavation of parts of the wider environs. This included undertaking geophysical survey and systematic fieldwalking of the site at Stringer’s Lane to help contextualise the wider Roman landscape, and finally, undertaking targeted excavations at Stringer’s Lane. Excavations at Stringer’s Lane took place over three weeks in September 2021.

Fig. 1.

Location map of sites at Cobblers Lane and Stringer’s Lane.

Fig. 1 Long description

A road location map with north indicated by a north arrow at the lower left. Two filled circular markers identify the sites. The label, Stringer′s Lane is placed to the left of the markers and the label, Cobblers Lane is placed to the right of the markers. The markers sit between the place labels, Chester above and Wrexham below. A large water inlet is shown at the upper left of the map. A scale bar at the lower left reads 0 and 10 km. No legend is shown.

The geology of the two sites consists of similar glacial and sedimentary deposits. Both are situated over a Kinnerton Sandstone Formation formed approximately 247–252 million years ago, in the Triassic Period, with overlying superficial deposits of glacial till formed up to 2 million years ago in the Quaternary Period. Immediately to the north of the Cobblers Lane field, in an area known as Burton Meadows (Fig. 1), the superficial geology consists of Lacustrine Deposits of clay and silt. This indicates an environment previously dominated by lakes, which is likely to be present in the northern extent of the Cobblers Lane magnetic survey. A wetland environment is also indicated in the LiDAR imagery of Burton Meadows to the north where extensive former water channels are visible. Nineteenth- and twentieth-century water management is also shown on Ordnance Survey mapping in the form of a network of canalised streams, suggesting flooding is likely to have been a dominant factor in this area. The Pulford Brook, a small tributary of the River Dee, runs immediately to the south of the Stringer’s Lane site, and Stringer’s Brook, a few fields to the north.

In the footsteps of Trebellius Maximus: Investigations at Cobblers Lane

The results of the magnetic survey at the Cobblers Lane site identified possible archaeological activity within the immediate vicinity of the lead ingot find spot. The most prominent feature was a concentration of readings consisting of weak linear anomalies and a magnetic disturbance that formed a possible rectangular structure measuring 30 m by 12 m running along a different orientation to the surrounding cultivation, as well as a series of possible pits and narrow linear anomalies that may be associated with post-holes. Faint linear features were also identified crossing the site on a north-east–south-west alignment. Given the hypothesis for a Roman road connected to the lead ingot find spot, this was tentatively postulated to be either a wide ditch or track way. The rest of the survey data produced mostly agricultural and modern features; however, full coverage of the field was not undertaken. Excavation trenches were located in order to investigate features as having potential archaeological significance (see Figs 2 and 3).

Fig. 2.

Interpretation of Cobblers Lane Magnetometry Results showing findspot of the lead ingot and trench locations (Archaeological Survey West). Note: north is to the right.

Fig. 2 Long description

The map displays magnetometry results from Cobblers Lane, highlighting various anomalies and the ingot find spot. Anomalies are categorized as positive, weak negative, strong mixed linear, disturbance and cultivation, each represented by different symbols. The ingot find spot is marked prominently. Linear anomalies and disturbances are distributed across the map, with some forming a possible rectangular structure. The map includes a scale bar indicating 20 meters and a north arrow pointing to the right. Several labeled areas, A to H, are marked, indicating specific survey points or features of interest.

Fig. 3.

Cobblers Lane trench locations in relation to interpreted features. The pig findspot is located just within the left-hand, long rectangular trench (Trench 1).

On investigation, the features identified in the geophysics were not considered to reflect significant archaeological remains, with most unlikely to pre-date the nineteenth century a.d. They therefore bore no direct relation to the discovery of the Roman lead ingot. Instead, the evidence indicated a watery or marsh-like setting during the Roman and later periods, possibly suggesting that the ingot is a stray loss, since no evidence of deliberate deposition or lead processing could be found nearby. Evidence of any Roman activity within the immediate vicinity was therefore absent. This answers one of the key research questions of the project; inferring that the ingot may have been lost in transit. While work undertaken by the Dept of Archaeology, Classics and Egyptology, University of Liverpool, on the material composition of the ingot and the lead isotopes suggests it may have come from somewhere in north-east Wales,Footnote ⁵ the uncertain route of Roman roads across parts of Denbighshire, Flintshire and Wrexham (especially at such an early date) leaves a question regarding how the ingot was being transported and the actual route it took.

The most significant aspects of the lead pig therefore currently relate to the stamp marked upon it and its isotopic signatures.Footnote ⁶ The Neronian date of the lead extraction and processing (under the auspices of Trebellius Maximus) raises questions relating to the military activity in the region at that time (especially since the fortress at Chester is not thought to have been established until the early 70s a.d.Footnote ⁷). Equally, the exact source of the extraction and processing sites remain unknown.Footnote ⁸ The letters in the centre of the inscription on the lead ingotFootnote ⁹ may refer to both the location of the site from which the metal was derived as well as the name of the official responsible for it. As already mentioned, Roman lead mining is well attested in Britain with known locations particularly in the Mendips and Derbyshire.Footnote ¹⁰ Equally, in north-east Wales, Roman lead- and silver-processing sites are known from Oakenholt, near Flint, on the coast of the Dee Estuary,Footnote ¹¹ presumably processing ores derived from nearby Halkyn Mountain on the north Delyn Plateau.Footnote ¹² The isotope evidence from the Rossett pig certainly suggest a Flintshire source, and a number of lead pigs often inscribed with a word referring to the Deceangli, the name traditionally ascribed to the pre-Roman tribe, have also been discovered, again thought to have derived from the north-east Wales ore fields.Footnote ¹³ These include examples found within or close to the legionary fortress at Chester (Deva Victrix),Footnote ¹⁴ which was established around a.d. 74 by the Legio II Adiutrix.

Roman lead processing and mining has also been suggested around 10 km to the south-west of the find-spot at MineraFootnote ¹⁵ and at Ffrith.Footnote ^16, Footnote ¹⁷ Minera has seen extensive lead mining from the medieval period and particularly during the eighteenth and nineteenth centuries, although clear evidence of earlier Roman mining is absent.Footnote ¹⁸ There is a known Roman settlement at Ffrith in Flintshire,Footnote ¹⁹ and although the site is poorly understood it appears to consist of an early timber phase associated with a single coin of Vespasian and later stone buildings of the early to late second century a.d. Significantly, a fragment of galena and lead waste has been found during excavations at the site indicating a direct association between the settlement and the lead industry. The presence of several Legion XX tiles are also potentially of relevance, given the early association of the army with mineral exploitation.

It is tempting to suggest a hypothesis that the pig was smelted (if not also mined) near Ffrith and was travelling eastwards along the Roman road when it was lost; if that were the case, then the name inscribed in the centre of the pig MAGUL (…) may be the lost Roman name for Ffrith or Minera. However, should Halkyn Mountain be the most likely in terms of the isotope evidence (which it strongly suggests), then the presence in a field in Wrexham County Borough remains puzzling, since an efficient route from Halkyn Mountain to the coast, or to the River Dee, or overland towards the established fortress at Wroxeter, would not reasonably pass by what is now Cobblers Lane: that is, unless the lead came from Halkyn Mountain, but the processing site was indeed elsewhere.

The movements of people and goods around the region would have utilised a combination of road, riverine and coastal transport mechanisms, including the River Dee and its tributaries. In terms of the road system, a Roman section of the road from Chester to Brithdir near Dolgellau has been postulated for some time,Footnote ²⁰ although physical evidence of its actual line in the area to the north of Wrexham is absent. It was suggested by MargaryFootnote ²¹ as running south from Chester then cutting south-west to the south of Caergwrle and on to Ffrith and Bala.Footnote ²² Further landscape survey and remote sensing may therefore be required to elucidate the routes of the Roman roads and help to answer the remaining questions as to how and why the ingot ended up in a field along Cobblers Lane.

The discovery of the lead pig has therefore raised many interesting questions concerning industrial activity in the area that in turn impact how we understand broader military activity and its effect upon the area (people and landscape). As mentioned, the discovery also raised the opportunity to investigate a nearby site, where other metal-detected finds had been recorded, including Roman brooches and coins. The subsequent geophysical results at the Stringer’s Lane site further added information to this regional picture, where the research team identified significant archaeology including a well-defined footprint of a large structure, the layout of which holds strong resemblance to a Roman corridor-type villa.Footnote ²³

The Rossett Roman villa

Fieldwalking on the site situated just off Stringer’s Lane yielded over 25 kg of ceramic building material (CBM) and included c. 70 fragments of Roman pottery, fragments of opus signinum flooring and wall plaster. It was clear that the field had been extensively ploughed and therefore the distribution densities of materials (Fig. 4) could not be taken as directly linked to buried remains beneath; however, the results strongly suggested that a Roman building was located somewhere on the site. This hypothesis was confirmed by the geophysical surveys.

Fig. 4.

Artefact densities by type from initial fieldwalking at Stringer’s Lane (CBM in kilograms, other finds in sherd/fragment count).

The initial geophysical survey results for Stringer’s Lane have been published elsewhere;Footnote ²⁴ however, a plot detailing the results of the magnetometry can be seen in Figs 5 and 6 and includes subsequent areas surveyed since 2021. The additional geophysical survey was undertaken by then Clywd-Powys Archaeological Trust (now Heneb), on behalf of the University of Chester.

Fig. 5.

Magnetic survey results from Stringer’s Lane.

Fig. 5 Long description

A map displaying magnetic survey results for Stringers Lane. The map includes a grayscale legend indicating magnetic intensity from 0.0 to 4.0. The surveyed area is outlined and filled with varying shades of gray, representing different magnetic intensities. Roads and water bodies are marked and a north arrow is present. The map grid is labeled with coordinates and a scale bar is included for distance measurement. The logo of CPAT and additional text are present, detailing the survey′s context and authorship.

Fig. 6.

Interpretation of magnetic survey, Stringer’s Lane.

Fig. 6 Long description

The map of Stringers Lane displays magnetic survey results with a legend indicating different magnetic patterns. The legend includes categories such as ′Linear′, ′Weak linear′, ′Negative linear′, ′Disturbance′, ′Field drain′, ′Ferro′, ′Ferro disturbance′, ′Burnt/thermal′ and ′Cultivation′. The map outlines the ′Limit of Survey′ with a red boundary. Various patterns are marked within this boundary, including linear features and disturbances. The map includes a scale bar and a north arrow for orientation. The survey was conducted by the Clwyd-Powys Archaeological Trust on behalf of the University of Chester. The map also shows geographical features like rivers and roads surrounding the surveyed area.

The results show the outline of a winged-corridor villa located within a possible enclosure, perhaps fossilised within later field boundaries. A number of trackways, ancillary and outbuildings are also hinted at, as well as contemporary field systems.

Areas of investigation were identified to determine the presence of archaeological features, structures, deposits, artefacts and ecofacts and would stop at the uppermost layers of significant archaeology. The aim was to undertake an assessment of the character, extent, date, integrity, state of preservation and quality of the buried remains so that, in turn, it could inform the preservation and management of the resource and help formulate a proposal for further archaeological investigation within a programme of research and public engagement. Trenches were excavated (see Fig. 7), targeting the likely domus itself, the rectangular anomaly just to the south of the domus building, the anomalies near Pulford Brook, with some smaller test-pits located to investigate a possible outbuilding to the north-west. Excavations took place over three weeks during September 2021 with a team of professional archaeologists and local volunteers.

Fig. 7.

Interpretive plan of magnetometer survey showing trench locations (produced by Archaeological Survey West).

Fig. 7 Long description

A thematic map depicting archaeological evaluation trenches near a watercourse. The map lacks a north arrow, so directions are relative to the page. Four trenches are outlined in red and numbered 1 to 4. Trench 1 is near the lower center, close to the watercourse. Trench 2 is above and to the right of trench 1. Trench 3 is in the right half, containing several black rectangular outlines. Trench 4 is in the upper left, also with a black rectangular outline. A broad linear strip runs diagonally from the lower center to the upper left. The map includes grid coordinates, but no specific site names or legend are provided, so symbol meanings are not explained. The map′s purpose is to show trench locations for archaeological investigation.

The following provides an overview of the results from each trench

Trench 1

Based on the high magnetic susceptibility shown in the geophysical surveys, the anomalies near the Pulford Brook appeared to show some structural signatures together with some possible industrial activity. It was proposed that this may have been a bath-house for the villa, especially since it was located directly adjacent to the water source. The trench revealed some relatively well-preserved and substantial limestone walling together with in situ opus signinum flooring (Fig. 8), although no substantive evidence was recovered to confirm use of the range of the rooms as a bath-house. The main structure was 6.5 m wide and seemingly of dry-stone wall construction. Extensions and additions were noted both to the south toward the brook and likely to the west.

Fig. 8.

Walling and in situ opus signinum flooring in Trench 1 (facing west).

At some point after construction an internal stone partition with a mortared core was added, cutting the internal floor and forming a narrow room some 1.3 m wide at the very southern end of the building. Roman ceramics and a fragment of copper-alloy hair pin, however, confirmed the Roman date of the structure. The condition of the building remains in this trench were better than elsewhere on site and further investigations will aim to explore this area of the villa complex more extensively.

Trench 2

Trench 2 was located over the rectangular feature to the south of the domus building. Topsoil removal revealed a rectangular rubble spread, within which some facing and deliberate arrangement of stones could be seen. On investigation, it was clear that this was a rectangular, walled building that likely post-dates the villa. Roman building material was used in its construction including tiles with the stamp of the legio XX Valeria Victrix. Likely made nearby at the legionary tile depot in Holt, they denote the legion that was based nearby at Deva (modern-day Chester). Other construction materials included patches of lime mortar and hewn stone varieties also found elsewhere in the Roman structures (Fig. 9).

Fig. 9.

West-facing post-excavation photograph of Trench 2.

The building appears to have suffered some burning with evidence for charred timbers and wattle. Radiocarbon dating of these timbers has provided date ranges from the eleventh and twelfth centuries a.d. This corresponds with a demolition date around that time which is confirmed by ceramics sealed within the rubble backfill of the structure dating from the thirteenth–fourteenth century. It is still uncertain what this structure may have functioned as, but based on an initial map-based review, there is a possibility that it could signify an early medieval or medieval chapel. Further desk-based investigations are ongoing; however, it is interesting to note that Abermagwr villa in Ceredigion is also located in a field called ‘chapel meadow’.Footnote ²⁵ This perhaps highlights how during the Middle Ages (or later), the stone ruins of Roman buildings may have been perceived as chapels.

Trench 3

Trench 3 was located over the eastern portion of the likely domus. Topsoil removal revealed surviving structural features but with fluctuating levels of preservation. It was clear that the site has been subjected to varying degrees of damage and disturbance from modern agricultural activity (ploughing). Some walls survived at the lowest courses, others had been denuded to thin mortar deposits overlying the clay foundations, and some appeared to have been robbed out subsequent to the domus use, or in one case appeared to have been deliberately removed during the lifespan of the building. Nevertheless, the outline of the eastern end of the villa was much as had been anticipated from the geophysical survey results (Figs 10a and 10b).

Fig. 10.

(a) Plan of Trench 3; (b) post-excavation aerial photograph of Trench 3.

Fig. 10 Long description

An archaeological site plan titled, SLXXI Trench 3. A key at lower left lists: Stone, Mortar, Robbing trench, Cut, Deposit and Deposited topsoil. The plan shows a rectangular trench boundary containing several rectilinear wall-like blocks and room-like rectangles, including a large block on the left, a long narrow band across the upper area and two adjoining rectangular blocks in the lower right area. A small rounded-ended feature extends from the right side of the upper band. Small dot markers are scattered across the plan area. A vertical scale bar is shown at the right edge with tick labels 0, 5, 10 and 15. A north arrow marked N is shown near the lower right. A post-excavation aerial view of the same trench area. A square excavation area is visible with exposed rectilinear features that align with the plan layout, including a broad rectangular zone on the left, a long narrow strip across the upper area and a prominent rectangular feature near the lower center. The excavation boundary forms a near-square outline, with surrounding ground visible outside the trench.

Limited evidence for flooring survives inside the building, however, and patches of opus signinum hint at what may once have been there (Fig. 11). The internal layout of the building is not completely certain, but evidence for internal room dividers and possible modifications of space suggest that the domus may not have been constructed in one phase, or at least underwent some alterations during its life.

Fig. 11.

Section of partially excavated degraded flooring in front ‘corridor’ of main building.

As yet it is not entirely clear when the building was first constructed (although a second-century date seems likely, based on the ceramics recovered), nor how it altered over its lifetime. But there were indications of internal alterations from these initial evaluative excavations, with one of the internal walls seemingly being deliberately dismantled sometime after a.d. 306–7 (terminus post quem from a coin within the backfillFootnote ²⁶) (Fig. 12).

Fig. 12.

Aerial view of Trench 3 highlighting robber cut of internal wall with (inset) coin recovered from within the fill (White and Moorhead Reference White and Moorhead2021).

Fragments of painted wall plaster, a range of pottery including decorated Samian, local ceramics from the kilns at Holt, Black Burnished wares and Oxfordshire wares and box-flue tiles that would be used in a hypocaust system help to provide a picture of internal decoration and the presence of under-floor heating in at least part of the building. Wall construction is of lime mortar-bonded faced stone with rubble core. Materials used include limestone, sandstone and a conglomerate found nearby at Caergwrle. It would be interesting to see if the different building materials (and their very different colours) were used for their decorative impact, but with only a couple of courses surviving it would be impossible to prove. The ceramics range in date from the second (although these are highly abraded) to fourth centuries a.d. (Oxfordshire/hammerhead mortaria). Five coins were also recovered. Those which are identifiable date from the fourth century a.d., including a nummus of the House of Constantine dating from a.d. 335–341. The artefacts and date ranges bear similarity with those found during excavations of the nearby villa at Eaton by Tarporley in Cheshire.Footnote ²⁷

To the east of the end wall of the domus some small exploratory sondages revealed an area of rubble covering a probable hypocaust flue (Fig. 13). The presence of a hypocaust system was evidenced across the villa in the form of box-flue tiles and possible brick pilae, but it was most evident in the in situ remains of the stoke, the line of the flue running through into the building, and the draw for the stoke to the east. The stoke and the section of the flue excavated showed signs of burning, demonstrating that it had been fired. Just how far the heating system goes into the building will be a question for future investigations. Samples from the burnt material within the area of the stoke that were sealed beneath the rubble await further analysis at the time of writing, including radiocarbon dating.

Fig. 13.

Hypocaust flue and stoke from above (Trench 3).

It is unclear how or when the villa fell into disuse due to the lack of survival of the stratigraphy in the area of the trench. However, more recent Ground Penetrating Survey results indicate that the western end of the domus may be better preserved (Fig. 14). There is also the hint of an apsidal wall at this end of the building. This will be a target for any future investigations as the life of the project continues.

Fig. 14.

GPR survey time-slice at ∼0.2–0.4 m depth (David Jordan and Liverpool John Moores University, with permission).

The Rossett villa in context

The following aims to consider the villa in relation to the closest major settlement at Deva. For the purposes of this discussion, we refer to the ‘hinterlands’ broadly (but not rigidly) as the modern Unitary Authorities of Flintshire, Wrexham County Borough, Chesire West and Chester, Warrington and Metropolitan Borough of Wirral. This area, however, should be viewed as having ‘blurred’ edges, reflecting both the fluidity of Chester’s influence during Roman rule as well as the uncertainty in the current understanding of that reach and influence. Not only that, but the political geography of the Rossett area is also sufficiently ambiguous. In that vein, we also do not attempt to assign the villa too firmly to either Deceangli or the Cornovii territories. This undoubtedly impacts our approach, but ensures that the discussion is focused and detailed enough to present an argument within the scope of the article.

While the River Dee is often treated as a potential boundary in the late Iron Age, this division is neither certain nor necessarily stable, and the very nature of ‘tribal territories’ in this period remains debatable. Under Roman rule, this ambiguity persists. Although north-east Wales and Cheshire are conventionally placed within the civitas of the Cornovii, centred on Wroxeter, the reality on the ground was more complex. The proximity of the legionary fortress at Chester means that much of this landscape lay within a zone of strong military influence, if not direct control. As such, the Roman landscape of north-east Wales and western Cheshire largely consisted of military and industrial sites (Fig. 15). This is no surprise, since the region also forms part of Britannia’s western frontier zone.Footnote ²⁸ In such circumstances, civilian administrative frameworks may have been secondary to the logistical and strategic needs of the army. It is therefore unclear how far Wroxeter’s authority meaningfully extended into the Chester hinterland. If the Rossett villa lay within land appropriated for military purposes (the prata legionis), its administrative affiliation may have been defined less by civitas structures than by military oversight.

Fig. 15.

Map showing military bases and key sites mentioned in text.

Fig. 15 Long description

The map depicts north-east Wales and western Cheshire, highlighting various sites. Military sites potentially part of the ′frontier′ are marked with black squares, including locations like Caer Gybi, Segontium and Chester. Other major civilian settlements are indicated by grey circles, such as Whitchurch and Wroxeter. Industrial sites are shown with black circles, including areas like Pen Llystyn and Caer Gai. Roads are represented by lines, connecting these sites. The map provides a spatial overview of military, civilian and industrial locations in the region.

The network of military installations emanated from Chester at the centre (the legionary fortress of Deva Victrix) westwards across north Wales to Holyhead/Caer Gybi, down into mid-Wales (e.g. Caersws) and eastwards across Cheshire, then up through the modern counties of Lancashire and Cumbria to join the northern frontier (Fig. 15). A fort was potentially established at Chester c. a.d. 60 and then subsequently the legionary fortress in c. a.d. 74 by legio II Aduitrix during the campaigns against the Deceangli, Ordovices and Brigantes. In a.d. 88, Deva became the home of the legio XX Valeria Victrix (hereafter legio XX VV). The fortress has some unusual features including its size (20 per cent larger than any other legionary base in Britain), the distinctive masonry used in its construction (specifically the cornicing still visible on the north wall) and the unique elliptical building, the function of which remains a topic of debate.Footnote ²⁹ These features combined have led to the postulation that the fortress was, at one time, envisioned to be the administrative centre of the area as well as a potential base for an invasion of Ireland.Footnote ³⁰

Like nearly all the military bases in the region, an extra-mural civilian settlement developed. Surrounding the fortress, some rural sites are known, such as the recently identified farm at Saighton CampFootnote ³¹ and the settlement at Saltney.Footnote ³² Along key Roman roads, archaeological investigations have revealed roadside settlements, notably that at HeronbridgeFootnote ³³ (see Fig. 16).

Fig. 16.

Location map of Chester and surrounding settlement areas (after Mason Reference Mason2012).

Fig. 16 Long description

The map illustrates the Chester area, highlighting several settlements and geographic features. The presumed course of the River Dee is depicted, winding through the region. Key locations include Canabae, marked with a large rectangle, indicating a significant site. Surrounding Canabae are numerous ′X′ symbols representing cemeteries. The map also shows civil settlements at Saltney and Heronbridge, with Saltney located to the west and Heronbridge to the south. Saighton Camp is identified as a farmstead or villa, situated to the east. The legend defines symbols: ′XXXX′ for cemeteries, ′X′ for civil settlements and outlines for river and marsh areas. The map provides a scale bar for distance reference, indicating the spatial relationships between these historical sites.

Beyond the military installations, most known Roman sites within the fortress hinterlands relate to mineral extraction, industrial production or processing. Equally, many ‘sites’ on the Historic Environment Records database relate to stray findspots or late Roman hoards, some of which may well mark further rural settlements. The routes of the Roman roads are fairly well known in and out of Deva (e.g. Margary 701, 7a, 67a, 66a, 6a), with some across the coastal plateau being archaeologically well attested (e.g. parts of Margary 67a). Yet some lines of roads heading south from the fortress (especially Margary 66a from Heronbridge to Ffrith and onwards to Bala and Brithdir) are largely only postulated and large portions remain unconfirmed through fieldwork. There are also likely more minor routeways and trackways across the landscape lying undiscovered.

A distinctive feature of the area to the west of Deva (now north Wales) is that no Roman towns are known. This contrasts with south Wales, the west Midlands and north-west England. The nearest civilian administrative centre to north Wales was the civitas capital at Wroxeter (Shropshire), while some roadside settlements and industrial sites across Cheshire developed urban characteristics (e.g. Wilderspool and Nantwich). Consequently, it has therefore always been assumed that most of the population of Deva’s hinterlands, and indeed north Wales (if there was much of a population), were living in the vici (or civil settlements) associated with the military forts,Footnote ³⁴ the industrial settlements such as Wilderspool, or perhaps in archaeologically ephemeral rural farmsteads or roundhouse settlements yet to be discovered.

In terms of the broader picture, Roman rural settlement in Britain has been a topic of interest, debate and analysis for generations,Footnote ³⁵ but perhaps the largest-scale project to date has been that of the Rural Settlement of Roman Britain (RSRB) project, undertaken by the University of Reading and Cotswold Archaeology.Footnote ³⁶ The project set out to collate information from published data, but mostly it aimed to bring together evidence from the increasing number of ‘grey literature’ reports. Such reports were produced through developer-led archaeological interventions since archaeology became a statutory consideration within the planning process in 1990. It has produced an enormous data-set and attempted to characterise settlement throughout England and Wales while also highlighting avenues for future work. The database is publicly accessible via the Archaeology Data Service (ADS) and their own findings have been published in three volumes.Footnote ³⁷ As is often the way with dealing with large datum, the general patterns are clearly presented, but some of the nuances can become obfuscated. Equally, only sites excavated and that produced radiocarbon dates were included in the study, leaving many recorded Roman buildings and unexcavated settlements beyond its scope.

The project categorised four main database settlement types: (1) Farms (defined as small agriculturally based settlements without villa architecture; of the farming settlements, unenclosed ‘open’ settlements, enclosed settlements, and complex settlements were used as further defining factors); (2) villas (mostly agricultural-based settlements with architectural characteristics associated with prominent display, for example tiled roofs, painted plaster, mosaic floors, and often associated baths); (3) villages (defined as nucleated rural settlements, usually over three hectares with multiple areas of occupation and not directly associated with the main Roman road network); (4) roadside settlements (nucleated rural settlements directly associated with the main Roman road network).

For the area of Cheshire (East and West), the RSRB project characterises the county as part of the Central West region. Within this region, Chester is argued to have dominated rural dynamics and countryside settlement patterns reflect supply functions and military demand. Farms were found to be predominantly oval or sub‑rectangular enclosed farmsteads, often on marginal land. Villas or high-status sites are known and are indicated by the presence of continental architectural details such as hypocausts, masonry and wall plaster. Evidence for extensive field systems was found to be limited but present, for example, small enclosures near Chester, Middlewich and Nantwich, although the evidence suggests that field systems were not extensively developed. Industrial settlements were, however, significant, including salt production at Middlewich and pottery at Holt, representing the processing and production of goods and materials to support urban/military centres. The knock-on effect of this is argued to have resulted in substantial deforestation (likely for industry rather than farming).

The area of north-east Wales, however, is mostly grouped within Upland Wales and the Marches region for analysis. Here, settlement in upland areas (>300 m) is suggested to consist mostly of curvilinear field enclosures and a stock-based economy, demonstrating continuity from the Iron Age. On the lower valleys or terraces, more rectilinear terraced field systems were identified, with occasional arable farming. There is, however, a marked absence of settlement noted within this area (Fig. 17). Nearer the military bases and installations, vicus settlements and higher levels of integration with Roman infrastructure were identified. Yet when we look at the data from the Portable Antiquities Scheme and other known findspots, the picture suggests higher levels of activity than the number of identified settlements reflects (Fig. 18).

Fig. 17.

Screenshot from Rural Settlement of Roman Britain project online archive showing rural settlements identified (Allen et al. Reference Allen, Blick, Brindle, Evans, Fulford, Holbrook, Lodwick, Richards and Smith2018).

Fig. 17 Long description

A thematic map of Wales and adjacent regions displays rural settlements from the Roman Britain period. North is oriented at the top. The map highlights north-east Wales, showing upland areas and terraces. Symbols represent different settlement types: circles for curvilinear field enclosures and squares for rectilinear terraced fields. Key locations include major towns and coastline features. The legend explains symbols and shading, indicating economic activities and settlement patterns. The map provides a visual distribution of settlements, with clusters near military bases and sparse areas in uplands.

Fig. 18.

Roman finds recorded on the Portable Antiquities Scheme database as of January 2026 (Open OS Data: contains Ordnance Survey data © Crown copyright and database right, 2025.)

In terms of the area under scrutiny in this paper, the large ‘gap’ in rural settlement covering much of north Wales and parts of Cheshire stands out and raises two immediate questions. The first is about the character of sites identified by the project as ‘rural’. The second relates to whether this gap is evidence of the absence of rural settlement or if it represents an absence of evidence. The following summarises the current state of knowledge, combining the RSRB data with evidence from sites identified since that project. It uses the RSRB classifications of site types and in doing so raises questions concerning the complex relationship of agricultural and industrial economies, military supply and the potential for settlements still ‘missing’ from the archaeological record, like the more recently discovered villa at Stringer’s Lane.

The Rossett villa: a missing link and where to find more

While there may not have been a density of rural settlement like those identified in areas of central-southern England, the authors propose that the current picture for Cheshire and northeast Wales is still not an accurate one and that based on the discovery of the Rossett Roman villa, more targeted survey and investigation is warranted. The following outlines our reasoning for this thesis.

MatthewsFootnote ⁷⁸ has argued that the prata legionis alone would have been sufficient to sustain the garrison at Deva. However, only some of the identified enclosures and villa/potential villa sites are located within 10 km of the fortress, and — as stipulated by NevellFootnote ⁷⁹ — the villa/potential villa sites do form a ring beyond this boundary to the south and east, which could be important in delineating Chester’s prata legionis. This suggests that either the other smaller farms and settlements within the prata were collectively producing sufficient supplies, that supply to the legion extended beyond the confines of the prata legionis, or indeed, that the extent of the prata legionis reaches beyond the usual 10km radius. If we take the presence and distribution of LEG XX tiles as evidence for the extent of the prata legionis, then according to Mason, the area would have encompassed c. 5,500 sq km. MasonFootnote ⁸⁰ instead suggests that the areas to the west of Deva could form part of an imperial estate while areas to the north and east largely formed the prata. Is this therefore, why relatively few farms have been found in the area that now forms north Wales?

A recent re-evaluation of the rural settlement in the northern frontier zone of Britain has identified a number of interesting patterns that might be of some relevance here. Bruhn and HodgsonFootnote ⁸¹ identified that the imposition of Hadrian’s Wall on the indigenous settlement patterns (and population) was different both north and south of the wall. Once viewed as having had minimal impact on settlement types, increased settlement identification and investigation suggests that villas emerged in the area south of the wall, whereas to the north there appears to have been widespread abandonment of settlement and an emergence of a new social order, potentially linked to the military supply of agricultural produce. While the authors rightly argue that the evidence is not sufficient categorically to identify any conscious imperial policy at play, the possibilities that are raised include a combination of indigenous response to the imposition of a frontier north of the wall and potential for conscious policies of colonisation and land development to the south. Either situation is possible in our region (and indeed a combination of the two is likely), and furthers Reynolds’s argumentFootnote ⁸² that North Wales was an economically varied landscape heavily shaped by local geography, trade routes, and community choices.

As articulated by Bruhn and Hodgson for the north of Britain,Footnote ⁸³ the supply of the army is a significant cog in the wheel of rural settlement in the region, and one would therefore expect a higher density of settlements producing agricultural surplus (like larger farms or villas). However, in this area of the western frontier zone there remains a question over how steady the demand would have been. From c. a.d. 120 to 165, a significant portion of the legionary garrison stationed at Chester, and indeed garrisons from bases across north Wales were away from their base, initially involved in constructing Hadrian’s Wall, and subsequently the Antonine Wall further north. The legio XX VV also provided garrisons for frontier forts such as Corbridge and Newstead and contributed vexillations for campaigns in other provinces. The auxiliary units stationed in Northwich and Middlewich were also likely re-directed to oversee industrial activities related to production salt and other resources.Footnote ⁸⁴ Nevertheless, it is questionable as to whether local landowners faced any decline in income, since the fortress remained the regimental headquarters and, due to its port facilities, likely became a more important hub for gathering goods destined for the northern garrisons or elsewhere in the Empire. This is particularly the case for mineral resources and could be the same for agricultural produce. For estate owners in Cheshire, as well as across the province, the largely peaceful and prosperous conditions of the second century likely fostered a growing civilian market for their produce. So why, then, do we not see this reflected in the intensity of rural farms across the landscape?

It is hard to believe that fertile areas were not exploited, whether within the prata or part of an imperial estate, or through encouraging/incentivising elements of the existing local population. In the first instance, reconstructions of the environment through palaeoenvironmental remains would be helpful in at least establishing the viability of the landscape, the proportion of dense woodland compared with open farmland, and so on. Nonetheless, the lack of large, complex farms is perhaps surprising. Rippon et al.Footnote ⁸⁵ propose that, without the economic pull of the Roman army, agriculture in the ‘Western Lowlands’ never needed to scale up into the kind of complex, market-oriented farming seen in more militarised or densely developed regions. However, this does not ring true in north-east Wales, where a sustained military presence existed. Instead, for this region, a clue could lie in the lack of ‘outbuildings’ identified at the Eaton-by-Tarporley villa. While the lifestyle of the native elite was supported by agriculture, together with the exploitation of natural resources such as metallic ores and salt, it could be conceivable that they did not involve themselves directly in its management. Instead, they took rents and tribute from people of ‘free-commoner’ status who worked their lands and looked after their stock in return for patronage and protection.Footnote ⁸⁶ Thus, the actual farming of the estate(s) would have been carried out by a workforce living in scattered farms and hamlets and there would have been no need for farm-workers’ accommodation and agricultural buildings in the immediate vicinity of the villa-house. It is possible that the scattered and low-level distribution of ‘farms’ across much of the region under discussion here supports this concept of ‘free-commoner’ farmers. Questions therefore arise as to the levels of surviving ‘indigenous’ populations and their social status. Were large numbers forced into slavery and moved into other areas of the empire? Were slaves brought over to this part of Britain to farm and support military supply? Or were some people forced into a social grouping that was technically free but with no or restricted access to social and economic networks? The evidence from Hen Caerwys, Flint and similar sites might support one of these possibilities. Additionally, the more recent discovery of two lead pigs of Domitian at LlangynfelynFootnote ⁸⁷ prompt similar questions surrounding the late Iron Age and Roman lead mining and processing on the west coast of Wales. Currently, the problem is that we do not have a clear enough picture of late Iron Age settlement in the region, sufficient radiocarbon dating to create precise chronologies of change, nor a full picture of post-conquest rural settlement.

The discovery of the Rossett villa and this review of the evidence has thus suggested two things: first, that there is a need to understand better the character of ‘rural’ settlement in the region, specifically in relation to the interplay between industrial and agricultural production. Second, that further survey and investigation is needed to ensure a fuller picture of rural settlement both in terms of pre- and post-conquest change and to establish if the absence of evidence is indeed, evidence of absence. As has already been postulated, it is likely that the Stringer’s Lane villa and associated field system and structures do not exist in isolation. Additional survey and investigation in the region could continue to uncover previously undetected villa sites, and an ongoing research project led by the authors aims to address exactly this. Whether the presence of the villa bears any relation to the existence of the lead pig inscribed with the name Trebellius Maximus is uncertain; however, both discoveries certainly add to our understanding of Roman activity in north-east Wales and raise significant questions concerning early military activity in the area and the subsequent character and intensity of rural settlement. In turn, we may be able to re-evaluate the prata legionis and potential for an imperial estate in the region.Footnote ⁸⁸ Some of the priorities highlighted here chime in with the current north-west Regional Research Framework (R18, R19, and R32), but this paper strongly advocates that a link needs to be made with the Research Framework for Wales, currently in development.Footnote ⁸⁹ As outlined by Belford, the impact of modern national boundaries upon archaeological research and practice has hampered our understanding of this region.Footnote ⁹⁰ This notion of a ‘frontier’ is thus not one restricted to the distant past; indeed it remains deeply ingrained even today, as argued by White.Footnote ⁹¹ If the prata legionis stretches across what are now modern borders such as that between England and Wales, so too must our research focus.

Furthermore, the wider survey and investigation around industrial centres like Holt will continue to add to the evidence and help us explore the juxtaposition of industry and agriculture across the rural landscape and further explore the character of Romano-British lifeways in the region. Clarifying the lines of the Roman roads in the area is also high on the list of priorities. However, in the absence of developer-led archaeology, finding sites that remain elusive will very much rely upon the use of metal-detecting data as well as continued, targeted research, survey and investigation. In the clay soils of Cheshire, the Portable Antiquities Scheme and metal-detected finds will be key. The ongoing research also aims to capture these elements.

There are also those sites which are classified in the Historic Environment Records as dating to later periods, but for which the classification rests on morphological information rather than any firm dating. These sites require investigation. Based on the poor material assemblage at Hen Caerwys, there is a possibility that similar settlements exist and are yet unconfirmed as being Roman. If this is the case, rather than a region devoid of population, it is likely that these fertile plains were occupied by people who simply did not leave much in the way of material objects, an issue also identified in the hinterlands of Wroxeter.Footnote ⁹² There may be several reasons for this; however, poor survival in the ground surely cannot be one, since the fair preservation of faunal remains would suggest an equally fair survival of ceramics. Perhaps, then, it is due to the lack of excavation focus in the area. Alternatively, as previously discussed, such settlements may also be linked to very low-status, possibly even slave societies. The presence of active, and extensively exploited, lead mines would suggest the presence of slave populations in the region; however, where they were housed is uncertain. Is it unfeasible to consider slaves, criminals, or very low-status communities working in the nearby lead and silver mines around north-east Wales and living within settlements like that at Hen Caerwys? The area is known for its large-scale extraction and processing of lead and silver (see, for example, Flint, situated c. 8 km to the east of Caerwys at the mouth of the River Dee, where, as already discussed, intensive industrial activities have been identified). The ability to test this hypothesis would certainly add to our understanding of diversity amongst rural populations and lifeways in the region.

Strategies that prioritise environmental remains are also crucial. Without radiocarbon dating Hen Caerwys may well have continued to be known as a medieval (and later) site. Bayesian modelling, where possible, will also be important to address chronologies, especially in a region where ceramics and coins are not always recovered, but also to help narrow down rates of change. Equally, strategic palaeoenvironmental sampling strategies of all rural settlement (industrial or seemingly agricultural) are required to establish fully the interplay between production and supply, as well as the balance of industry and agriculture across rural sites.

The landscape of the Deva hinterlands is therefore one of economic productivity. The military stimuli include the provisioning of the troops in addition to the military role in the extraction and processing of valuable resources (namely lead and silver). Based on current evidence, much of the civilian population appears to be reliant upon such an economic foundation. However, the agricultural picture suggests that the production of foodstuffs is very much a secondary economic driver. Wealth is clearly being created and is circulating within the frontier zone (as evidenced in the general patterns of material culture and connectivity). The landscape is fertile for agriculture and there exists a need to feed a resident population (military and civilian). Yet, with the data available, large farms that are based primarily on the production of surplus foodstuffs (and by-products) appear to be almost non-existent. Instead, rural settlement in the region consists of mixed economies wherein a settlement (roadside or farm) either has an industrial economic basis with some agricultural production on the side (e.g. Prestatyn or Wilderspool), or has a more equal balance of the two (e.g. Plas Coch). Whether the supplies for the army are coming from elsewhere remains unknown (although the Feeding the Roman Army in Britain (FRAB)Footnote ⁹³ and Feeding the Roman Army, Making the Empire (FRAME)Footnote ⁹⁴ projects both seek to investigate this exact problem). The picture we have of rural settlement in the region is more complex than the RSRB data suggests, but the overarching narrative has not substantially moved on since the work published in the 1990sFootnote ⁹⁵ and early 2000s,Footnote ⁹⁶ despite new evidence that is starting to question it. The region therefore needs to be a focus of research but also be included in targeted studies such as FRAB and FRAME.

The Rossett lead pig was a significant find and has raised many questions concerning the intensity and scope of lead extraction and processing in the region during the Neronian period. In turn, this impacts our understanding of the conquest phases of north Wales. Subsequently, by following in the footsteps of Trebellius Maximus, we stumbled across the villa at Stringer’s Lane. The discovery of the villa was not made through the presence of crop- or parch-marks, showing up in aerial photography as so many other villas have done. Instead, it was identified through scrutiny of metal-detected finds and walkover survey. There are, therefore, implications for how we might approach methods of identification and discovery moving forwards, especially in terms of the importance of metal-detected finds (and the responsible recording of them). In turn, this will help us address the many uncertainties about the character of life in this part of Britannia. This paper has outlined the implications of discovering a Roman villa near Rossett and developed an argument for the need to undertake a more systematic approach to identifying, investigating, characterising and classifying rural settlement in the region. Were this to be done, we would have a much better understanding of the economic (and socio-cultural) dynamics and lifeways of the population.