Introduction

The Roman-period town of Moridunum lies under modern Carmarthen, situated in West Wales (Figure 1). Although only a handful of excavations have taken place within the town, what has been done has produced sufficient data to allow us to understand the basic layout and its development, as well as affording us a glimpse at the lives of the inhabitants. In 2018, the redevelopment of a section of Priory Street provided the opportunity to excavate an area close to the centre of Moridunum. The data collected from the site suggested that occupation began in the late first to early second centuries AD and continued into the late fourth century. The site was not static, with phases of rapid change followed by periods of both stability and decline. Combining this data with evidence from previous work has allowed us to develop a more detailed picture of life within the town, and the main aim of this paper is to present a new look at the origins and development of Roman Carmarthen throughout its three centuries of occupation.

Figure 1

Map showing South Wales in the early second century AD, showing the locations of Carmarthen and Caerwent, as well as the military bases and road network (derived from Burnham and Davies 2010).

Moridunum was the civitas capital of the Demetae, one of four Iron Age social groupings that Roman writers tell us occupied the area of modern Wales, with the Demetae located in West Wales (Berggren and Jones 2000). The resistance offered to the invading Roman military by the neighbouring tribes of the Silures and Ordovices is well documented by Tacitus (Annals 12.38–39), with the final conquest of Wales not occurring until the Flavian period. The Demetae do not feature in the Roman writings, with little being known about the role they played during the years of conquest and whether they actively resisted the Roman military. After the conquest, the new province was divided into civitates peregrinae, which were units of local government that aided in the administration of the province. As a conquered people, it is unlikely that this status was granted to the Demetae and their neighbours immediately and it has been suggested that the earliest administration of the people living in Wales was centred at Caerleon (Guest 2022). It was not until the early to mid-second century AD that the Civitas Demetae received this status, with the evidence suggesting Moridunum had been invested as a civitas capital by this time (James 2003). The only other civitates peregrina within Wales was that of the Silures, who established their civitas capital at Caerwent, known as Venta Silurum.

The two civitas capitals did not exist in isolation, with other small urban centres developing across both civitates in the aftermath of the conquest. These included a civilian canabae at Caerleon, a nucleated settlement at Cowbridge and roadside settlements at Great Bulmore and Usk (Marvell 1996; Parkhouse and Evans 1996; Yates 2001; Guest and Gardiner 2025). Vici have also been recorded outside a number of the forts in South and West Wales (Burnham and Davies 2021). These sites represent different forms of urbanism, with the civitas capitals and canabae being major towns, while the rest are classed as small towns (Mattingly 2006).

A recently published study of the excavations at the forum-basilica at Caerwent by Guest (2022), has shed new light on the foundation of the town, with an earlier Trajanic date suggested for its foundation as a civitas capital. Together with the new evidence from Carmarthen, which also appears to have its origins in the Trajanic period, this offers new insights into our understanding of the development of urbanism, in particular the major towns, within Roman Wales. This new evidence suggests that these major towns were founded earlier than previously thought. Guest (2022) has suggested that the Roman military may have aided in the construction of the forum-basilica at Caerwent. It is possible that this was replicated at Carmarthen, with the construction of the early public buildings there by the military.

The Discovery of Moridunum

The presence of a town at Carmarthen was established by G.D.B Jones in the 1960s, although parts of a bathhouse had been excavated in the nineteenth century (James 2003). Findspots of Roman coins and pottery were also known from across the town prior to Jones’ excavations at St Peter’s Car Park in the late 1960s. His excavations identified buildings, workshops, a street and the line of the defences along the northern edge of the town. These findings confirmed that Carmarthen was the Muridunum of the Antonine Itinerary (James 1978).

The development of the town was linked to the presence of a small auxiliary fort, located immediately to the west (Figure 2). The fort was excavated in the late 1980s and is believed to have been established between AD 74 and 77 during the campaigns in West and North Wales by Governor Sextus Julius Frontinus (James 2003; Guest 2022). It was connected to forts at Llandeilo, to the east, and Loughor, to the south-east, by the road network (Figure 1). The fort formed one of over 40 fortresses, forts and fortlets constructed across Wales by AD 90 (Guest 2022). The large number of military positions across what is now Wales is a reminder that the people who lived within this varied landscape were not easy to conquer, and it was not until the late AD 70s that they were finally subdued (Guest 2022).

Figure 2

Map showing the location of the fort and town at Carmarthen, as well as the street layout within the town (Thomas 2022).

The fort was reduced in size towards the end of the first century before being abandoned in the early second century (James 2003). It was located on the lowest crossing point along the Afon Tywi (River Towy), which allowed the military to control traffic and shipping on the river. It has been suggested that the fort was located there to police a possible tribal centre, the hillfort of Merlin’s Hill located to the north-east of the fort (James 2003). The only archaeological work on the hillfort consists of the recording of sections through the defences during works on a trackway in the 1980s, with radiocarbon dates indicating a construction date in the Middle Iron Age (James 2003; Murphy et al. 1992; 2012). The interior of the hillfort has not been subject to any excavation, and it is unclear if the site was occupied at the time of the Roman conquest. Several potential Iron Age sites have been identified through aerial survey within the wider landscape of Carmarthen (Ritchie 2018), but none have been excavated. It may have been that there were no settlements within the immediate landscape, and so the fort was situated solely to control the river and its crossing.

Since Jones’ excavations only a handful of other sites within the town have been excavated. As the Roman town is overlain by post medieval and modern development, excavation relies on redevelopment, which has occurred infrequently. In the last 50 years, only seven excavations have taken place within Moridunum between 1970 and 2017 (Figure 3). These excavations, along with evidence from watching briefs, have allowed us to piece together the plan of parts of the town, understand the layout of the road system and the development and later redevelopment of the defences (James 2003).

Figure 3

The Roman-period town, showing the locations of previous excavations (Thomas 2022).

Excavations have recovered small quantities of late first and early second century material, concentrated at Church Street and Priory Street (Figure 3). The material from Church Street was believed to have come from an annex of the fort and was not connected to civilian activity (James 2003: 15). However, the material from excavations at Priory Street, to the east of the fort, were thought to have come from a small civilian settlement along the military road, although this was not thought to constitute a vicus (James 2003: 16).

Excavations have recorded at least six metalled streets (Figures 2 and 3) which formed part of the street network laid out across the area of the town (James 2003). The establishment of the civitas capital at Carmarthen is indicated by the laying out of a gridded street network. Woolf (1998) has noted trends within the urbanization of Gaul, one of which was the clear organization of urban space including the imposition of a gridded street network. The layout of Gaulish towns were modelled on cities in northern Italy, although the street grid was flexible and was adapted to suit local topography and existing roads, something that has also been observed in other British towns, including Wroxeter, Exeter and Gloucester (Burnham et al. 2001; Creighton 2006).

At Carmarthen, the first century AD military road formed the Decumanus Maximus, which was the main road running south-west to north-east through the centre of the town. The dating for the layout of the street network indicates that it was constructed between AD 120 and 150. Where Street 1 was excavated within the St Peter’s Car Park, it was found to overlay earlier timber buildings (James 2003), which most likely dated to the late first or early second century AD and may have been demolished to allow the construction of the revised street network. The street network at Carmarthen was not strictly orthogonal, with Streets 1 and 3 not running parallel to the Decumanus Maximus (Figure 2). This design can also be observed at other towns within Britain, particularly those that did not develop on the sites of forts, for example London and St Albans (Burnham et al. 2001; Creighton 2006).

A small number of public buildings are known to have stood within the town, including a temple excavated at 105–111 Priory Street in the 1980s, and thought to have been constructed before the establishment of the street system (James 2003). The temple was short-lived, with evidence of demolition and robbing during the mid-second century AD. The remains of a bathhouse, found within the eastern corner of the town, were discovered during the nineteenth century, with further evidence derived from the excavation at Dyffryn House in 1961 (James 1978). The remains of an amphitheatre survive to the north-east of the town, located along the military road. The amphitheatre was partially excavated in the late 1960s and early 1970s and is thought to have been constructed in the second century AD (Little 1971). James (2003: 19) has also suggested that the partial remains of a building excavated at The Parade may have been a mansio, constructed in the late first to early second century and thought to have been in use into the fourth century AD.

During the mid-to-late second century, the town defences were constructed; this is similar to the dates for the construction of defences around other towns in Roman Britain. A ditch and earthen bank were erected around Caerwent during the mid second century AD with a wall being added in the later second century. At Aldborough, an earthen rampart and wall were constructed during the later second century (Wacher 1995: 384, 403). The first phase of the defences consisted of a bank with a triple ditch system in front of the bank. James (1978; 2003) suggests that the defences were extended and remodelled in the later third century AD, with the earlier bank being spread across the innermost of the three ditches to form a low ramp, which was fronted by a masonry wall across the line of the middle ditch. The outer ditch was then backfilled with a larger ditch being excavated in its place (James 1978; 2003).

Prior to our latest research, little was known of civilian life within the town beyond the nine phases of Roman urban activity, suggested from the excavations at 105–111 Priory Street, dated from the mid-to-late second century to the late third or early fourth century AD. The number of occupied structures recorded during the previous excavation suggested that the high point of occupation occurred during the mid-to-late second century. Several buildings were recorded within these excavations, both domestic and industrial in nature, with evidence of iron smithing and bakeries across a number of periods. James (2003) has suggested that the site had been used for cultivation during the late second to early third centuries AD. Only one high-status domestic building has so far been excavated within the town. This was a large building with tessellated floors and a hypocaust system dating to the fourth century, which was found during the St Peter’s Car Park excavations (James 1978).

From the finds assemblages recovered over these fifty years of excavations, it appeared that life continued within the town up until the final quarter of the fourth century AD. Whether the town was simply abandoned at this point is unclear. Early Christian activity in Carmarthen can be suggested from the site of St Teulyddog just to the east of the walled town, established during the seventh to eighth centuries AD. The Church of St Peter’s within the walled town is thought to be contemporary with St Teulyddog (Ludlow 2004). It is possible that the town was abandoned for a few centuries before being resettled during the Early Medieval period. However, it is also possible that as with many other Roman-period towns, life continued although in a much less archaeologically visible way (see Smith 2007).

The 2018 Priory Street Excavations

In 2018, the redevelopment of a plot on Priory Street provided a unique opportunity to add to our knowledge of the Roman settlement. The plot was located on the northern side of Priory Street, almost at the heart of Moridunum (Figure 3). A total of six phases of activity were identified during the excavation. Phase 1 was of pre-conquest date and Phases 2 to 6 dated to the last quarter of the first century through until the last quarter of the fourth century (Figure 4). During this time, 14 buildings were constructed with evidence indicating that at least eight others had been present. Other structures excavated across the site included a bread oven, two drying kilns and a street surface. The evidence suggested that few of these buildings had a domestic function but supported commercial or industrial activities.

Figure 4

Phased plan showing all of the Roman period features excavated at 113–117 Priory Street (Thomas 2022).

Phase 2

The earliest evidence of activity dated to the late first to early second century, beginning in the immediate aftermath of the conquest. During this time, three buildings were constructed along the line of the road, which was located just beyond the south-eastern edge of the site (Figure 5). These buildings appear to have been of mixed use, with both domestic and commercial structures present.

Figure 5

Plan of buildings belonging to Phase 2 (Source: Archaeology Wales 2018).

The domestic structure, Building 1, was a rectilinear timber building with at least four rooms. Very little of the building survived, with no intact floor surfaces. The dating of the building is suggested from Samian Ware in one of the beam slots as AD 90–110.

Building 2 was a timber structure that was divided into two large rooms. The remains of what is believed to be a large bread oven in the rear room suggest that the structure may have been of a commercial nature. The structure was demolished within a couple of decades and replaced by the larger Building 3. Although this later building was twice the size of Building 2, evidence of only two rooms survived, both with what are thought to have been large bread ovens in the floors, suggesting it retained its commercial function. The later building had stone foundations, which may have supported a two-storey structure, with a living area on the upper floor.

Phase 3

Activity intensified during this phase, which began in the early second century and lasted until the middle of the same century. Despite being only a few decades in length, this phase appears to have been the peak period of activity within the site. There was evidence for the construction of five buildings (Figure 6), with at least one of these thought to be a shop.

Figure 6

Plan of buildings belonging to Phase 3 (Source: Archaeology Wales 2018).

Building 4 was a timber structure that had been built over the remains of Building 1, following a similar layout, with three rooms evident. There was some evidence for subdivision of two of these rooms and it is possible that this structure began as a domestic building. However, its use quickly altered, with a working hollow and a large square pit surrounded by a metalled surface recorded in a corner of the largest of the three rooms. Slag and hammerscale was found in the pit fills but the small quantities of both suggest metalworking was not taking place within the building but in the vicinity of it.

Building 3 was replaced by Building 5, which underwent partial reconstruction four times, with the south-western external wall being replaced each time. The final period of reconstruction, which took place after fire had damaged the external wall, and the rear room, the wall was constructed on stone foundations rather than the previous timber beam style construction. The building interior had been divided into two rooms, with a possible bread oven in the rear room and the front room appeared to have functioned as a shop. A square timber structure was recorded in the middle of the floor and is thought to have been a free-standing shelving unit, which itself was repaired at least three times. Before this building was demolished at the end of the phase, it underwent a change in function as the oven was backfilled and an opus signinum floor was laid over the top of it. The shelving unit was also demolished, and it may be that it was converted into a domestic structure. At the same time a street was laid out along the south-western edge of this building. The street would have connected to the main street to the south-east and may have joined Street 1 to the north-west.

Building 8 belonged to this phase and was larger than any of the other buildings excavated on the site. Only the north-west corner of the building survived, with an L-shaped foundation trench recorded. It had a substantial foundation trench, consisting of tightly packed rubble. This supported a masonry wall, although only a few stones from the lower course of the wall remained in situ. The substantial foundations are thought to be indicative of a two-storey building. Unlike the other buildings of this phase, which were demolished prior to the mid-second century, Building 8 continued in use into the early third century (Phase 5). It was unclear what the building was used for as there were no internal features bar a single posthole. Given the difference in the size of this building in comparison to the other structures on site, across all phases, a public building or one supporting an official function might be suggested.

Phase 4

This phase began in the mid-second century and ended abruptly around AD 170, when a fire broke out. Although a large number of features were recorded for this phase, there appears to have been very limited activity, with most features relating to two small wooden structures that are indicated by 179 stakeholes. What these stakehole structures were used for is not clear. The lack of activity on the site at this point likely reflects the fact this area saw very little activity during this time, rather than solely due to the shortness of this phase.

The activity across the site appears to have been of a commercial and manufacturing nature, with no evidence of domestic activity although it is possible that living quarters were located in rooms above or potentially behind the shops. Building 10, in the northern part of the site, did have evidence for at least two clay floors but stakeholes cut through these indicate small wooden structures within the floor, which were thought to be related to manufacturing rather than domestic activity (Figure 7).

Figure 7

Plan of buildings belonging to Phase 4 (Source: Archaeology Wales 2018).

The street which had been laid out at the end of Phase 3 was not in use during this phase, with Building 9 having been cut through it, less than a decade after the street was constructed. This structure was similar in construction to Building 5 and had been divided into two rooms (Figure 7). In approximately AD 170, Building 9 burnt down, with the fire creating conditions for excellent preservation of some of the structure and its contents (Figure 8). The oak beams and planks from the exterior walls were excavated along with nearly 3000 sherds of Samian, Black Burnished and Severn Valley wares, which were located in the front room of the building. The vessels were stacked inside each other, indicating that the front room had functioned as a pottery shop.

Figure 8

Fire damaged pottery from the remains of Building 9 (Source: Archaeology Wales 2018).

Large quantities of burnt grain were recovered from along the line of the internal wall, with the amount suggesting that at least one sack of grain had been stacked against the wall at the time of the fire. Structures excavated outside the building are believed to have been a bread oven, with bread-like material being recovered from the structure, and two drying ovens located just beyond this. The drying ovens were not contemporary; rather, one was a replacement for the other. It is likely that a bakery operated out of the back room, with the grain being dried, stored and then baked into bread on site. It was not clear if the grain was milled on site, as no evidence of millstones or querns was found. If it wasn’t milled on site, then it is likely that a mill existed within close proximity.

The fire that brought this phase to an end was contained within this building and it is likely that the fire started in the bakery. The positioning of the burnt timbers and the contained spread of material from the building suggests that it was deliberately demolished to contain and extinguish the fire (Figure 9).

Figure 9

Burnt timbers from one of the walls of Building 9 (Source: Archaeology Wales 2018).

Phase 5

This phase began in the aftermath of the fire and continued until the end of the third century. Although this was the longest of the Roman phases of occupation, the levels of activity were very low. Buildings 8 (Phase 3) and 10 (Phase 4) continued to be occupied, with three further buildings, Buildings 11, 12 and 13, constructed during the first half of the third century (Figure 10). All of these buildings had been heavily truncated, with very little of each structure surviving.

Figure 10

Plan of buildings belonging to Phase 5 (Source: Archaeology Wales 2018).

At the beginning of this phase, a street, which followed the line of the street from Phase 3, was constructed across the levelled remains of Building 9 (Figure 11). The street surface was repaired at least once. A fourth-century rubble deposit overlay part of the street, and it is likely that the street remained in use throughout Phases 5 and 6.

Figure 11

Part of the street that was constructed across the remains of Phase 4 Building 9 (Source: Archaeology Wales 2018).

Building 8 was demolished in the early third century, being replaced by Building 11. Only a single foundation trench remained of this building, which contained three large postholes, each approximately 1m in diameter, with two having a large iron spike in the base. The spikes were likely used to help support the large posts that were set within these postholes. The size of the postholes suggests they supported a substantial building. A postpipe was recorded within one of the postholes, indicating that the building was left to decay and not deliberately demolished. It may be that this building continued in use into the fourth century.

Phase 6

The final phase of Roman occupation spanned the fourth century, with extremely low levels of activity across this phase. It is thought that Phase 5 Buildings 11 and 12, continued in use during this phase, along with the street.

A single building was constructed within this phase, Building 14, which was located along the main street (Figure 12). Little remained of the building, but the remains of a clay floor cut by a number of postholes and stakeholes indicate it likely had a commercial or manufacturing function. The head of a copper alloy figurine was recovered from one of the backfilled beamslots that made up the building. The figurine is thought to be Apollo, although it was very badly corroded and may be his sister Diana or potentially Venus. A statue of Diana depicted with a quiver and a stag is also known from Cynwyl Elfed, approximately 9km to the north-west of Priory Street (Durham 2012).

Figure 12

Plan of buildings belonging to Phase 6 (Source: Archaeology Wales 2018).

The end of Phase 6 was signalled by a number of layers which appear deliberately laid, likely associated with the demolition of standing structures and the levelling of the site. It is possible that the levelling deposits were associated with a wider abandonment of the town during the mid-to-late fourth century. They marked the end of occupation within the Priory Street site, which was not reoccupied until the twelfth century.

Rethinking Moridunum

In her analysis of the excavations in Carmarthen between 1978 and 1993, James (2003) discussed the idea that Moridunum had evolved into the civitas capital of the Demetae from a vicus, located along the military road to the east of the fort. The presence of vici has been demonstrated at several forts in Roman Wales (Burnham and Davies 2021). Vici typically comprised simple buildings that included workshops and shops, as well as residential buildings. James (2003: 16) believed that the lack of evidence for a ‘ribbon’ like settlement along the military road meant that the settlement outside the fort was not a vicus; rather she refers to it as a focus of settlement in the area around the temple.

However, the evidence from our excavations suggests that this early occupation was more widespread than previously thought. The 2018 excavations have shown that buildings were constructed along the main road out of the fort very quickly after the fort was established. The ceramics from the buildings’ demolition deposits suggest that they had been constructed in the final decades of the first century AD, similar to those found in the previous excavations along Priory Street. In the 1960s, Jones found the remains of timber structures underneath Street 1, to the north-west of Priory Street and there is late first century activity from Church Street. James (2003) suggested the latter belonged to an annex of the fort, related to a small number of stock enclosures and hearths. It is unclear why these were thought to belong to the fort as they could equally have been civilian features. It is thought here that the latter is more likely based on the growing evidence of wide-spread civilian settlement at this early date. The evidence suggests that this early settlement may well have been a vicus, which developed into a civitas capital during the reign of Trajan.

Recent research in south-west England (Holbrook 2015), has shown that the known major towns, with the exception of Dorchester, evolved from forts during the late Flavian and Trajanic periods, including Exeter, which was founded within the remains of the legionary fortress (Holbrook 2015). Roadside settlements are also situated in close proximity to abandoned forts, at Pomeroy Wood, Woodbury Farm and Shortlands Lane in Devon, that originated in the second century (Thomas 2021). Recent work at Cirencester has revised previous interpretations that features were part of a fort annex and vicus and these are now considered to belong to an earlier phase of the town, with public buildings being erected in the late first century AD (Holbrook 2015).

The construction of three public buildings at Carmarthen: the temple, the bathhouse and the possible mansio (Figure 3), is believed to date to this earlier settlement, pre-dating the establishment of the civitas capital (James 2003), and may have been of military construction (see below). The gridded street network was laid out around these buildings, and the excavations at 105–111 Priory Street suggest that the street frontages were demarcated by fence lines, with evidence of property boundaries separating plots across part of the excavation (James 2003). The construction of a street grid and the demarcation of property boundaries suggests that a central authority, likely an Ordo, controlled the implementation of the town’s layout (Thomas 2022). It is possible that the Ordo comprised members of the local population and potentially retired Roman military veterans from the auxiliary fort. Woolf (1998: 119) suggests that the implementation of a street grid across a town ‘implies a single moment of foundation…’ which, in conjunction with the construction of public monuments, may have served to legitimize the social position of a new elite within society (Woolf 1998: 124–125). The demarcation of the street grid and property boundaries within Moridunum is tangible evidence of this moment of foundation.

The presence of the fort would have acted as a stimulus for the development of the town, with the early inhabitants trading with the soldiers. The fort would have opened up the area to traders from other parts of Britain and the Roman Empire, who took advantage of the fort to trade with local populations. Several coins are known from Carmarthen, although the assemblage is small, with only 130 reported in total. This is in comparison to 1625 coins from Exeter, 1915 from Dorchester and 1123 from Caerwent (Holbrook 2021). Most of the coins from Carmarthen are early, with coin usage dropping off in the later second century. The number of Flavian and Trajanic coins suggests that a market centre developed soon after the establishment of the fort. Using coinage to date the beginnings of the town is complicated by the fact that some may also have continued to circulate for decades after minting, which is known to be the case for Flavian coins (Guest 2008).

The monetization of Wales in the Roman period has been demonstrated to be linked to the military, with Wales becoming monetized quickly after the conquest (Guest 2008). The study by Guest (2008) established that small bronze denominations were more commonly found as single finds along the coast and inland along the river valleys in West Wales. Guest (2008: 51) suggested that this was indicative of bronze coins being used in commercial exchanges. It is then reasonable to suggest that the earlier coins in Carmarthen were introduced by the military, with the population using them to purchase goods.

Piecing together the evidence of Moridunum and its inhabitants shows that the town underwent periods of rapid change within the first two hundred years of its occupation. The 2018 excavations demonstrated that buildings were frequently altered and repaired before being replaced. Building 2 was constructed within Phase 2, however, it only stood for a few years before being pulled down and replaced by a larger structure. Both buildings had what are believed to have been bread ovens and may have functioned as bakeries. The demolition of Building 2 does not necessarily mean that the ownership of the plot changed in between Buildings 2 and 3 being constructed, and it may have been that the owner had wanted to expand the bakery, which led to the construction of Building 3.

Both Building 5 (Phase 3) and Building 9 (Phase 4) were also believed to be partly or wholly used as bakeries. These were both constructed within the same plot as Buildings 2 and 3, and may have been owned by the same family, showing longevity of occupation within the town. This may also be the case for Buildings 1 and 4 (Phases 2 and 3), which had very similar floor plans with Building 4 directly overlying Building 1. A similar picture is evident from the 1980s excavations at Priory Street, where plot boundaries appeared to survive throughout the life of the town, although periods of near desertion were evident, if short lived. The boundary lines between plots were refreshed a number of times across the centuries, with new ditches and fence lines evident (James 2003).

The fact that most of the buildings within both Priory Street excavations were of a more commercial and manufacturing nature indicates that the town may have been zoned, which has been suggested for other towns such as Wroxeter (Gaffney et al. 2013). Commercial structures within the centre of Moridunum are to be expected, and although the forum basilica of Moridunum has yet to be found, the 2018 excavation area must lie in very close proximity, where building owners might take advantage of this almost prime position. Other sections of the town have produced evidence for zoning, Jones found what he believed to be workshops within his excavations, but these appeared to have been confined to the northern side of Street 1, with residential housing on the southern side (James 1978).

Evidence from the current excavations suggests that there was a steady decrease in coin loss towards the end of the second century and that a market economy was never truly established in this part of Britannia. The small quantities of later coinage may mean that the population had returned to Iron Age traditions to acquire goods. It may also mean that we are seeing a biased picture of coin usage and loss, with coin circulation potentially having been much higher. Only further investigations will confirm whether coin use was truly limited in the later period. The drop in coin loss does link to a downturn in activity levels recorded by both the 1980s and 2018 excavations at Priory Street. This, along with the cultivation soil recorded during the 1980s excavations, paints a picture of a town that was sparsely inhabited by the third century AD. The cultivation soil is not unique to Carmarthen, with evidence of a similar deposit from Exeter (Bidwell 1980; Holbrook 2015). The cultivation terraces at Carmarthen were not long lived and had been reused for industrial activity within a few decades. Yet the fact that the street surfaces in Carmarthen were being refreshed into the late fourth century, with a number of repairs over the centuries (James 2003), indicates that there was a small but thriving population within the town, which was still governed by an Ordo, who would have coordinated the road repairs amongst their other duties (Thomas 2022). This is further supported by the presence of a Crossbow brooch, dating to the middle to end of the fourth century, found during the previous work at Priory Street. These brooches are recognized as markers of official rank that were worn by both civilian and military officials (Swift 2000).

Where the population of the town originated from is unknown, but it is very likely that most of the inhabitants were local people from the surrounding region. It is possible that military veterans formed part of the population of the early town, and it may be that traders from elsewhere in the wider region, province, or from even further afield lived within the town, taking advantage of trading opportunities. The cemeteries that would have been outside the town boundaries, which might shed light on these points, have never been found, although a cremation was excavated to the east, between the town defences and the amphitheatre. The cremated remains, which dated to the late first or early second century, were contained within a greyware vessel that included a fragment from a glass vessel (James 2003). Although some distance away from the fort, it is possible these were the mortuary arrangements of a soldier. A second cremation was excavated approximately 2km to the west of the town. The cremated remains had been buried with two oil lamps, one of which carried the stamp Fortis which is dated to AD 90–150. The lack of proximity to the fort suggests that this second individual may have been local to the area. Unfortunately, no isotope analysis has been conducted on either cremation, which would help to answer such questions.

Wherever the inhabitants of the town came from, we do have an idea of how some of them ate, what they ate and how they looked. The ceramic assemblage indicates that the inhabitants of the town were well connected to a range of trade routes, with both Spanish and Italian amphorae, Samian Ware, Black Burnished Wares and Severn Valley Wares well represented. Forms and products from the Nene Valley and the Oxford kilns are also represented, as are Mortaria from Verulamium, the New Forest, Wroxeter, North Gaul and Lower Germany. Fabrics produced in South Wales, such as Caerleon Ware, were also well represented but Black Burnished Wares represent the bulk of the ceramic assemblages from across all the excavations.

The most prevalent ceramic forms from Moridunum are jars and bowls, though beakers, dishes, flagons, mortaria and amphorae are also well represented. These forms indicate a shift away from the traditional way of eating and drinking evident in the Iron Age, when communal eating from large vessels appears to have been the norm (Cool 2006). The idea of serving and eating individual portions appears to have arrived in Britain through contact with the Roman world, which is evident in the appearance of small vessels such as dishes and beakers (Pitts 2005; 2010). This new tableware and drinking vessel forms appear in the record at Carmarthen during the earliest phase of occupation, and so, some inhabitants at least, began to adopt new social practices very quickly after the conquest. Alterations to drinking practices were also evident in the glass assemblage, with a number of cup fragments recovered. These changes to long-held social practices are significant and hint at the upheaval caused by the Roman invasion of South-west Wales.

The presence of Spanish olive oil and Italian wine amphorae shows changes to not only the way people were eating, but also what they were eating. The mortaria suggest that cooking practices altered with new food preparation methods being introduced. The charred plant remains included large numbers of peas and beans, which are not common finds from Roman sites in Wales with larger quantities known only from Caerleon. Beet was also recovered from the remains of Building 9, which was introduced into Britain during the Roman period, and it appears this is the first time beet has been found on a Roman site within Wales. This is also the case for the fennel, which was recovered from a third-century gully within the site and was unknown in Roman Wales until this occurrence (Carruthers 2022).

It is clear that eating habits altered in the aftermath of the conquest, but did these changes offer limited information about how people presented themselves to the world? We have little understanding of what the Demetae looked like, with few finds linked to clothing, personal adornment and grooming found within the region. Fragments of two brooches were found in situ with a crouched inhumation at Plas Gogerddan, which dates to the Late Iron Age, along with the fragmentary remains of a possible dress fastener from another grave on the site (Murphy et al. 1992). Such items are better represented after the conquest, with brooches, bracelets, hair pins, toilet implements and beads all recovered from Carmarthen. The presence of these items suggest that individuals did manipulate their appearance in new and different ways after the conquest through the use of jewellery and toilet implements. Overall numbers of these objects are fairly low, which is partly due to the low levels of excavations but also suggest that these changes were not wholesale and for many of the inhabitants of the town, traditional Iron Age practices of dress and bodily manipulation may well have continued to be the norm throughout the lifespan of the town. Newer ways of styling the body may have been something that only wealthier members of society practiced.

Mattingly (2006: 292) states that towns were places where new identities and social hierarchy were negotiated. The change in social practices, evident in the way food and drink were consumed, how food was prepared and the way in which certain individuals styled themselves, suggest that this was occurring within the population of Moridunum. These urban centres were a complete shift away from earlier settlement forms and the conquest would have disrupted the older socio-political networks, which enabled these new identities and a new elite to form within the town, and likely within the wider region, although this is beyond the scope of the current paper.

Urbanism in Roman Wales

Urban centres developed in lowland Britain during the pre-Roman Iron Age, with a number of centres of political power emerging across this area in the decades before the Claudian conquest, such as Colchester (Camulodunum) and St Albans (Verulamium) (Perring and Pitts 2013). Urban sites did not exist in Wales prior to the Roman invasion, and no obvious centres of power or hierarchy emerged in the Later Iron Age. Instead, settlement in South-West Wales appears to have been based upon hillforts and defended enclosures of which some 787 sites have been recorded (Murphy and Murphy 2010). Very few of these have been excavated and where they have been, the larger enclosed sites appear to have their origins in the Late Bronze Age and Early Iron Age, with the smaller enclosures beginning to appear from the Late Iron Age (Driver 2023). Dating enclosures where excavation has taken place has been complicated by the fact that South-West Wales was largely aceramic, meaning that dating primarily relies on radiocarbon analysis. Defended enclosures remained the dominant form of settlement within South-West Wales until the end of the Romano-British period (Murphy et al. 2012). Given this gap in the data it is difficult to be certain how society in South-West Wales was structured during the Later Iron Age.

There are many similarities to be drawn between the communities within South-West Wales and those of the South-West peninsula in England during the Later Iron Age. There is little evidence of an elite within society in Devon and Cornwall during the Late Iron Age. Society appears to have been structured around defended enclosures, with a large number of promontory forts, hillforts, multiple enclosure forts and rounds, which are small, defended settlements unique to the region. All of these sites were enclosed and so are believed to have been of some status (Quinnell 2004; Thomas 2021). The sheer number of such sites, including over 60 promontory forts from the coastline of Cornwall alone, makes it difficult to determine social stratification within settlement types. However, during this time, high-status metal work, such as decorated mirrors, shields, swords, scabbard mounts and chariot fittings along with brooches and other personal adornment items become visible in the archaeological record (Thomas 2021). Much of the metalwork, the decorated mirrors and Atlantic type brooches in particular, is thought to have been produced within the South-West peninsula (Tyacke 2002–3; Joy 2008; 2010). This emerging elite within the South-West then seemingly disappears around the time of the conquest, with social structures shifting in the face of the new power imbalance created by the arrival of the Roman military. Instead, a handful of new elite sites appear around the time of the conquest, which are marked out by the number of personal adornment items and the presence of imported ceramics within their assemblages, suggesting that these families or small communities had direct contact with the military and those people involved with the administration of the province (Thomas 2021).

Only two items of metalwork of Late Iron Age date are recorded in the Portable Antiquity Scheme within South-West Wales. Other metal work items, in the form of chariot fittings, were found with the remains of a chariot that was excavated in Pembrokeshire in 2018 (Lamb 2022). The chariot and its richly decorated fittings do hint at the presence of an elite within South-West Wales although the limited data available from excavations within the region means that it is difficult to see any other elite members of society within the archaeological record. This lack of evidence of higher status families and communities within the region makes the presence of wealthy members of society at Carmarthen within 50 or so years of the conquest of Wales very significant. As discussed above, some of the founding members of the Ordo would potentially have been military veterans who had received their honesta missio, which would have included a monetary settlement allowing them to buy property and integrate into civilian life. They may also have been drawn from several of the local leaders of the Demetae. It is possible that wealthy families existed during the Late Iron Age but have yet to be recognized in the archaeological record, or it may also be that the arrival of the Roman military and the integration of the region into the Roman empire created an opportunity for families to become wealthy and assume power.

The fort at Carmarthen was suggested to have been constructed partly to keep an eye on the local populations, with the hillfort of Merlin’s Hill suggested to have been a possible centre of occupation. However, as there is no evidence for occupation within the hillfort at this time, we cannot be certain of the significance or scope of this centre as it appears that there is little settlement within this area at the time of the conquest. The fort would have acted as a stimulus, attracting civilians who settled outside the defences. As already discussed, the military presence would have opened the area to civilian traders from other parts of Britain and the Roman Empire, who may have taken advantage of the fort to trade with local populations. It is possible to imagine local families profiting from these trading opportunities and accumulating a store of wealth, with the conquest then being the stimulus for the creation of a new elite tier of society.

The Roman state may also have lent a helping hand with the foundation of the town, as has previously been suggested for the foundation of Exeter. Around the time of the conquest there is little evidence of a local elite within South-West Wales who could have supplied the money necessary for the construction of public buildings. It may have been that Roman officials supplied funds for the construction of the public buildings erected within Carmarthen (Holbrook 2021; Thomas 2021). The recent study of the forum-basilica at Caerwent (Guest 2022) has highlighted the possibility that the Roman military may have contributed to the construction of the building. A well-preserved sestertius of Trajan was recovered from deposits below the forum-basilica, and its condition suggests that construction may have begun during Trajan’s reign. Guest (2022) suggests that, given the short span of time between the conquest and the construction of the forum-basilica, only the military had the ability and resources to carry out this project.

Similarly, the temple built within the area of the town at Carmarthen, which predated the layout of the civitas capital, was likely of military construction. The temple had been demolished by the time the gridded street network was laid out in around AD 120–150 and so most likely was a Trajanic construction (James 2003). It may have been that one or more of the Dematae families had become wealthy enough by this time to pay for the construction, but no building of this sort is known from the Later Iron Age or conquest period within the region, and the idea of worshipping in such a place would have been unfamiliar to the Demetae. Its early construction is therefore intriguing. Several temples and shrines within Britain were supported by the Roman military, with investment being provided for both their construction and restoration. This support began during the reign of Nerva, with several examples of military support noted between AD 96 and the death of Hadrian, including the construction of the Romano-Celtic temple at Vindolanda (Walsh 2020) located outside the fort, within the area of the vicus (Blake 2003).

It is possible that the military also constructed the bathhouse and possible mansio within Carmarthen. Neither building has been extensively excavated but are believed to be of similar date to the temple. Again, such buildings would be unfamiliar to the local Demetae, and it may be that the baths were constructed for use by the soldiers, with inhabitants of the early settlement being allowed to use the facilities.

Discussion

It has long been suggested that the civitas capital of Moridunum developed under Emperor Hadrian when urbanization in Britain is believed to have expanded. It has been suggested that Hadrian’s visit to Britannia in AD 121 or 122 stimulated urban growth with new civitas capitals being established across Wales and North England (Wacher 1995; Guest 2022). The excavations across Carmarthen during the later twentieth century have been synthesized by James (2003) with the data largely fitting into this Hadrianic narrative, the later civitas capital being discussed largely as a separate entity to the earlier settlement that existed outside the auxiliary fort from the late first century. Reinterpretation of the town suggests that it evolved organically from a small settlement, established outside the fort in the late first or early second century. Similar work at other towns within the province is showing that this story is echoed in places such as Corinium (Holbrook 2015). The formal street networks and public buildings that mark these sites out as civitas capitals were then inserted later.

The earlier evidence of settlement at Carmarthen consists of an assemblage of late first and early second century pottery, which was recovered from Priory Street, with a further assemblage of the same date from Church Street. These assemblages were not linked in previous discussions but used to suggest a small civilian settlement east of the auxiliary fort, with the material from Church Street coming from what was likely an annexe of the fort. I argue here that this was not the case, and the Church Street features relate to civilian occupation within the earlier settlement. The 2018 excavations demonstrated that this earlier settlement was established within a couple of decades of the fort being constructed, with timber buildings, potentially bakeries, located along the main east road out of the fort. Similarly, early timber buildings are also known from the north-western corner of the later town, which were demolished in order to make way for the gridded street network, laid out between AD 120–150. The construction of the temple and bathhouse during this phase of occupation were likely funded and constructed by the military with soldiers and civilians alike using both structures.

The temple was demolished when the street network was laid out, with no other temple currently being known within the town. James (2003) notes that two stone altars were recorded just off Church Street in 1804, with a pipeclay figurine of Venus also found within the town. Along with the head of Apollo, found in Building 14, these items indicate that individuals within the town continued to worship. These items also show that traditional forms of worship had altered for some of the inhabitants, with Roman gods now being venerated. It is likely that the altars had been moved but their presence indicates that another temple stood within Moridunum, highlighting the fact that there is still much to learn about the town and its inhabitants.

The fact that the early settlement at Carmarthen was far more extensive than previously suggested, with at least two public buildings standing prior to AD 120, suggests that this settlement should be thought of as a small town, with its location outside the fort indicating it started life as a vicus. This would mean that the civitas capital of the Demetae has its origins in the Trajanic period. However, the foundation date of the civitas capital itself was later, with the street grid not being laid out until between AD 120–150. If the interpretation of the foundation date for Carmarthen presented here is correct then the civitas capital at Caerwent predated that at Carmarthen, with construction at the former beginning during the reign of Trajan, rather than under Hadrian as previously thought (Guest 2022).

The relatively rapid development of Carmarthen raises interesting questions about the people of South-West Wales. It has been demonstrated that urbanism was a new concept to the native inhabitants of modern Wales. Previously, the presence of conjectured tribal centres within hillforts has been suggested as a reason for forts being established in certain locations, such as at Carmarthen and Caerwent, but this is really an outdated view with no real evidence to support it (Guest 2022). It may be that future work on Late Iron Age settlements does provide evidence for larger-scale tribal centres within South-West Wales, which could be deemed proto-urban sites; however, at present, it would seem that the towns of South Wales had no predecessors. It would appear that the military presence was the stimulus for the small town to develop at Carmarthen, which later became the civitas capital. The good transport links, over land and over water would have allowed trade connections to be established that went beyond supplying the fort, resulting in an accumulation of wealth for some families. The development of Carmarthen, as well as Caerwent, would have helped integrate the peoples of South Wales into not only the province, but also the empire.

The finds assemblage suggests that some of the local inhabitants of the town did just that, with new eating habits and ways of presenting the body clearly represented in the objects from the town. Both eating practices and how the body was styled are facets of an individual’s identity (Pitts 2004; 2007; Thomas 2021). The artefact assemblage from Carmarthen indicates that the conquest of the region and the foundation of the town impacted how at least some of the local population presented themselves and how they dined. As discussed above, the conquest of the region would have altered socio-political structures and would have allowed a new elite to form (Thomas 2021). The small quantities of dress items and new tableware forms, as well as the new foodstuffs that were imported into the town, indicate that not all the town’s inhabitants embraced new ideas, and it may be that only the elite or wealthier members of society presented themselves in a new manner. These new modes of dress and eating practices would have shown their standing within society. Perhaps for most members of the population the traditional practices of dress and eating continued. The evidence from the town indicates that Moridunum and its inhabitants were well connected to the wider Roman world despite being on its fringes.

Acknowledgements

I would like to thank Rowena Hart and Charley James Martin at Archaeology Wales for their help and support with presenting and publishing the results of our excavations at Carmarthen. I am also very grateful to Dr Jaqueline Cahill Wilson for her encouragement and for reading and providing comments on the drafts of this article.

Competing Interests

The author has no competing interests to declare.