Some Thoughts on the Romanization of Pembrokeshire
By Mark Merrony
In 2003 I surveyed and excavated what I believe to be a Roman villa at Ford near Wolfscastle that was previously identified by the antiquarian Richard Fenton in 1811. The results of the survey were published in the present journal in 2004, and the time is now ripe to present, in brief, the findings of the trial excavation. In light of further discoveries over the past fifteen years, it is also the intention here to integrate these into the emerging picture of Romano-British Pembrokeshire, since it is now becoming increasingly clear that the county – and those that adjoin it – are more Romanized than previously thought.
To put the more recent findings into their rightful context, there is some necessary but brief overlap with my publication in 2004. Logically, the most appropriate place to begin our grand tour of ancient Pembrokeshire is in the Roman period itself. The first evidence is referenced by Ptolemy, the Graeco-Roman historian who compiled Geography (II.3.2) in the second century AD, which mentions Octapitarum Promontorium, thought to refer to the Bishops and Clerks islets west of Ramsey Island near Saint Davids Head. The map also mentions Moridunum (Carmarthen), along with the Demetae, the Celtic tribe that inhabited the region in the Iron Age, Moridunum being their political centre in the Roman period.
Moridunum is also mentioned in an itinerary known as The Peutinger Table, thought to be a thirteenth century copy of a Roman original of the third century AD, acquired by the German antiquarian Konrad Peutinger in 1508, and now in the Hofbibliothek, Vienna. Other ancient references to place-names in the county come from a forged work, The Description of Britain, supposedly by Richard of Cirencester. This includes an itinerary of Roman Britain (Iter XI), listing a road known as the Via Julia west of Moridunum to Ad Menapiam (Saint Davids), via Ad Vigesimum (a supposed Roman station located in north Pembrokeshire). The authenticity of this document was widely accepted when it was first published in 1757 by Charles Bertram, but proved a forgery by Bernard Woodward, Librarian in Ordinary to Queen Victoria in 1847. Richard of Cirencester was actually a fourteenth century monk and historian at the Benedictine Abbey of Westminster.
After a millennium and a half of silence, something of Roman Pembrokeshire revealed itself to the antiquarian pursuits of Richard Fenton, who was informed of an unusual discovery at Ford near Wolfscastle. After visiting the site, he confirmed that it was a Roman villa, observing a bath with protruding flues, roofing tiles, iron nails, various bricks, some grooved, and others etched with lines. Shortly after this, Fenton visited Castle Flemish, mistaking it for Ad Vigesimum, where he found fragments of Roman brick. The site was later excavated by Sir Mortimer Wheeler in 1921, and sherds of Roman pottery, hexagonal roofing slates, and flue-tile, were among his discoveries. The site is now identified as an enclosed Roman villa.
It seems that Fenton was not an entirely trusted source and the idea of the Romans penetrating as far as Pembrokeshire was put on the backburner after the rather sceptical address of the Bishop of Saint Davids to the British Archaeological Association during the Tenby Congress in 1884. In spite of this, subsequent investigations at Ford continued to recover Roman material through the course of the twentieth century, such as roofing slates, and a Roman oil lamp, which is presently in Carmarthenshire Museum at Abergwili, Carmarthen.
Roman material has been recovered from a number of sites in the county, notably from coastal promontory forts at Brawdy, Buckspool, and elsewhere, furnishing evidence of steady occupation through the Roman period. The same may be said of a series of so-called ringforts near Llawhaden and other areas. There have been numerous Roman coin hoards discovered in the county and the broader region, especially since the successful implementation of the Portable Antiquities Scheme (PAS).
After Fenton’s identification of the villa at Ford, the next site of its kind was not discovered until 1950. This occurred at Trelissey near Amroth in the southeast of the county, excavated under the direction of W.G. Thomas and R.S. Walker in 1950 and 1951. They found an enclosed Roman villa with coarse and fine pottery, the head of a javelin, and other finds (now in Tenby Museum).
In 1995, a Roman road was excavated during the construction of the Whitland bypass just over the border in Carmarthenshire, a section of a route first observed by Terry James from aerial photographs, heading west from Carmarthen and into Pembrokeshire by RCAHM traced, in sections, by aerial photography, digital mapping, and fieldwork as far west as Wiston.
The next tangible Roman discovery or – more correctly – rediscovery, occurred on the site of the Ford Roman villa observed by Fenton. In January 2003, guided by an Ordnance Survey map, I visited the site and found several large roofing slates of varying size (some with nail holes; one cut to perhaps fit around a chimney). Subsequently, a geophysical survey was undertaken with Tony Johnson of Oxford Archaeotechnics (based in Noke, Oxfordshire). As expected, the ferrous material in this general location (fences, gates, water trough) significantly impaired the geophysical plot, but it was possible to identify an unenclosed subterranean building under the intersection of three hedges (approximately 18.5m by 7.5m). Rectilinear features to the east of the building were revealed, which may record field boundaries associated with the site, but it is not possible to be certain of this. A trackway appears to lead directly to the building from the east and seems to bisect a similar feature running north-south.
The original excavation at Ford took place with the kind assistance of several local volunteers on the weekend of 29 and 30 March. It was logical to excavate the area that corresponded to the clearer part of the geophysical survey, southeast of the water trough. This entailed the investigation of an area on a north-south alignment, carefully removing the overburden to reveal a number of large stone slabs (Fig. 1).
These were the foundation stones of the building but there was no trace of the superstructure apart from an area of tumble, which did not appear to correspond with the alignment of the structure. Overall, the site appears to have been heavily robbed of stone, a process that was underway between the rediscovery of the building and Fenton’s visit in the early nineteenth century.
The excavated site at Ford photographed from the south in March 2003 by Martin Cavaney
Disappointingly, there was a lack of pottery, apart from a piece of Roman ceramic that is a tile or brick, so it was not possible to establish any firm dates for the construction or development of the building. A significant find appears to be part of an igneous saddle quern, which consists of a lower stone (saddle) on which corn was placed, and an upper stone (rider), which was pushed to-and-fro on top of this. Normally these were replaced in the Iron Age by rotary querns, but in some areas, saddle querns remained in use into the Roman period (Fig. 2).
Fig.2. Saddle quern found at Ford in March 2003. Ian R. Cartwright.
Copyright Institute of Archaeology, Oxford University.
Several more roofing slates were discovered (Fig. 3). These are of particular interest since George Boon has established that they are from the same source in the southern Preseli Hills (also spelled ‘Prescelly’) as the slates found in the secondary metalling of the Roman quayside at Caerleon, Carmarthen, Castle Flemish, Cwmbrwyn villa near Llandorow, and Coygan Camp close to Laugharne, a Romano-British defended enclosure (both sites in Carmarthenshire).
Fig.3. Slate roofing tiles recovered at Ford in March 2003.
In light of the fact there are no navigable rivers between the two Pembrokeshire villas and the source of the slates found there, the material must, of course, have been transported by road; but their point of origin and distribution is too far north for the material to have been conveyed along the Carmarthen-Wiston route, a factor that will be revisited below.
Arguably the most important find was a piece of flue-tile, with a distinct combing pattern, reminiscent of Fenton’s description of similar artefacts over two hundred years ago. Other finds included a small voussoir (perhaps part of an arch), and several roofing nails, originally used to affix the slates to the timbers of the roof. The tile is a fragment of a box flue, which functioned as a component of a wall chimney, venting the hot air and smoke from the hypocaust (underfloor heating). The incised lines on the surface enabled the adhesion of plaster in much the same way that modern plasterers scratch render before the coat of skimming is applied (Fig. 4).
Fig.4. Flue-tile found at Ford in 2003. Ian R. Cartwright 2017.
Copyright Institute of Archaeology, Oxford University.
Fenton also described a bath and its embedded wall flues in considerable detail and the balance of evidence appears to indicate that the building at Ford was probably a villa, perhaps of the corridor type.
The tenant farmer of Bank Farm had kindly agreed to fence the site off to prevent cattle trampling the excavated area, but he was unable to sink any stakes at a sufficient depth to the immediate east of the excavation, due to the presence of substantial buried material – probably stone slabs. Clearly, the base of the building continues on a southeast alignment and extends beyond the present farm track, so there is certainly scope for future excavation in this location.
In 2010, the site was surveyed and excavated by Dyfed Archaeological Trust (DAT) with the assistance of local volunteers under the direction of Duncan Schlee (former Project Manager), but their investigation was hampered by the presence of the farm track mentioned above, which now overlies much of the area originally investigated. The geophysics echoed the results of our survey, with the exclusion of the villa building, but the coverage was more comprehensive, and revealed the presence of a prehistoric enclosure as well as a possible Romano-British building to the southeast of the villa. With the kind technical assistance of James Meek, Head of DAT Archaeological Services, a geophysical plot now exists that combines both surveys (Fig. 5).
Fig.5. Geophysical survey combining the results of the 2003 and 2010 surveys.
Reproduced by courtesy of Dyfed Archaeological Trust and Oxford Archaeotechnics.
Since this survey and excavation were undertaken, further discoveries have been made. One of the most significant of these is a Roman fort at Wiston. This was found in 2012 through geophysical survey, followed by trial excavation (in 2013 and 2014) under the direction of Meek. Part of this has been visible for some years on Ordnance Survey maps, but it had been previously dismissed as a recent feature. Pottery suggests that the fort was occupied in the late first century/early second century AD and that it was reused in the mid-second to the mid-third century AD, perhaps as a civil site. A significant breakthrough was made when the presence of an extensive civilian vicus settlement to the south was identified by geophysics. This was confirmed by excavation in 2014 (Fig. 6).
This extraordinary development is by no means the end of the story. In January 2017, I received correspondence from Luke Hooper, a student at Bournemouth University, who sought my opinion about an interesting discovery after undertaking a geophysical survey in north Pembrokeshire under the auspices of University College London (UCL) the previous summer. My conclusion accorded with his: the plot recorded an enclosed villa of the corridor type with a square Romano-Celtic temple adjacent to it. On my advice, he contacted Revd Professor Marine Henig, who concurred with our interpretation. A small excavation of this site was undertaken in late summer 2017, and this appears to confirm that the site is a Roman villa.
Fig.6. Geophysical interpretation of the Roman fort and vicus at Wiston.
Reproduced courtesy of Dyfed Archaeological Trust.
Martin Davies has provided a major contribution to the study of Roman Pembrokeshire most recently, investigating parts of the elusive Roman road network in the county. Of special interest is an apparent route that appears to head north from near the Roman fort and vicus at Wiston, over the Preselis via Tafarn-y-Bwlch (where there may be a Roman fort), and into Ceredigion.
A section of this appears to run past the Roman villa and temple site mentioned above.
Administration and economy
It is important at this stage to make some general comments about civilian villas and military forts in the county. Evidence elsewhere in Britain indicates that the owners of what we term ‘Roman villas’ were not Roman but indigenous elites who adopted their lifestyle in political territories known as civitates that were adapted to pre-existing Celtic tribal territories. In southern Wales these were implemented by the Roman authorities in the second century AD to frame the polities of the Silures in the east with their civitas caput at Venta Silurum (Caerwent) and the Demetae in the west, their caput at Moridunum (Carmarthen), (as mentioned above), the territory of the Demetae, equating with Carmarthenshire, Pembrokeshire, and southern Ceredigion, comprising a single political unit. In the western Roman empire, civitates were essentially city-states based on the constitution of Rome, with landowning classes (decuriones), defined by a property qualification, controlling power in a senate (ordo), from which the council (curia) were elected, as well as two annually elected magistrates (duoviri), who exercised power within the civitas during their tenure. This political arrangement, notwithstanding some variations, operated on the basis of circumscribed self-government. Tenant farmers (coloni) sold their produce in the urban market-place and paid a proportion of their revenue in tax and rent. Their landowning overlords, who were often pagan, and later, ecclesiastical officials, in turn paid part of their tax to the state and funded public building projects (euergetism).
It is not possible at present to reconstruct the economy of the region with any level of precision until future discoveries are integrated into the broader picture, but it is the case that Moridunum was the urban market of the Demetae, where financial transactions took place, goods were exchanged, and taxes levied. Coins recovered from Carmarthen indicate that they were in circulation from the late first century. In the wider region they were minted in Rome until the second century, and elsewhere in the empire thereafter, at Aquileia, Arles, Lyons, Siscia, and Trier, and in London and Colchester from the late third century AD. In accordance with the general picture in Britain, the bulk of coins are found on military sites and vici close to them, indicating a minimal impact of coinage on the civilian population in the territory of the Demetae. Several hoards discovered in Pembrokeshire have a coastal provenance with coins dating to the first half of the fourth century (Newgale Beach), or predominantly around the middle or the second half of the fourth century (Stackpole Warren, Newgale Beach, Whitesands Beach, Saint Davids and Goodwick Harbour); the later coins are plausibly interpreted by Heather James as having a restricted function as military pay or donativa (bonuses) rather than a broader economic function.
Forts in Britain were garrisoned by Roman legionary troops and auxiliaries from different parts of the Roman empire. The Wiston fort was probably built by soldiers of the Second Augustan Legion (Legio II Augusta), whose fortress was at Caerleon near Newport. Inscriptions from Hadrian’s Wall, the Antonine Wall, and their hinterlands, prove that the legion was active in northern England and Scotland until the early third century. It is therefore plausible that Caerleon was the hub for broader activities in Pembrokeshire, and they were present in the fort at Loughour, where a stamp tile of Legio II Augusta has been recovered.
The establishment of the base at Caerleon was necessary to pacify the Silures who, according to the Roman historian Tacitus, were hostile, since he records that: ‘…neither severity nor clemency converted the Silurian tribe, which continued the struggle and had to be repressed by the establishment of a legionary camp’ (XII.32). In contrast, the Demetae are absent from Roman accounts, and it is generally considered that they were subdued with relative ease. Wales was finally brought under Roman sway by AD 77 during the tenure of governor Julius Frontinus.
The search for a lost Roman road
A Roman road of particular general interest, mentioned above, is the route from Carmarthen to Wiston (Fig. 7). Where is it heading from Wiston? An obvious place is the Cleddau, in or near Haverfordwest, but there is no evidence of this as yet. It is often thought that its destination is Saint Davids, and there are some clues, provided by place-names and archaeology, that it may run in that general direction. At Rudbaxton, a possible trace of a Roman road was observed by aerial photography on a west-southwest – east-northeast bearing by Chris Musson of The Royal Commission on the Ancient and Historical Monuments of Wales (RCAHMW) in 2002. The further clue is a long stretch of road known as ‘the Causeway’, north of Camrose which tracks in a northwest direction. This is not marked on the modern 1:25,000 Ordnance Survey map of north Pembrokeshire, but it appears on older OS maps, notably the Six Inch (1888-1913) and 1:25,000 (1937-61) versions. Most recently, Martin Davies has identified a possible enclosed Romano-British villa south of its course.
Fig.7. Map of the Carmarthen-Wiston Road and the ‘Via Julia’. Courtesy of Dyfed Archaeological Trust. Published in Pembrokeshire County History, Volume I.
It is probable that this was the route taken by Gerald of Wales in the later twelfth century. He recorded that: ‘From Haverford we proceed on our journey to Menevia [Saint Davids]… and passed through Camros… We then passed over Niwegal sands…’ (I.13). Fenton refers to a possible stretch of Roman road in Newgale that was exposed by a storm around 1780, as described by a Mr Jones of Lether, and in the winter of 1795, where he observed: ‘…two lines of pebbles parallel, and for a mile in length; the pebbles from one inch to a foot [in] diameter…’ He speculated that this was, ‘…by tradition, said to have extended from Old Menapia along the coast to Dale on Milford Haven, and known by the common appellation of the Old Welsh Way’.
He considered that Brawdy, on the hill to the north, could derive from ‘Broadway’, but it may in fact have its root in the Welsh name ‘Breudeth’. In a manuscript of 1293, Breudeth had been shortened to ‘Bre’udy’, according to Bertie Charles. If the paving observed is a Roman road, it may intersect a route from Saint Davids through Brawdy to Carmarthen (see below). Roman roads are marked on the older OS maps between Brawdy and Saint Davids but these appear to be influenced, perhaps in large measure, by the forged map of Bertram. For instance, Menapia Roman Station is marked above Whitesands Bay. It is, however, curious that the latter is also known as ‘Porth Mawr’ (‘Great Port’). This was an important medieval crossing point to and from Ireland, and I am grateful to the Rt Revd (John) Wyn Evans, Bishop of Saint Davids (2008-2016), for clarifying this point.
Pliny the Elder, the admiral and historian, who was killed at Stabiae by the eruption of Mount Vesuvius on 25 August AD 79, mentions that: ‘Hibernia lies beyond Britannia, the shortest crossing being from the lands of the Silures, a distance of 30 miles’ (IV.16.103), (28.33 conventional miles or 45.6 kilometres). The distance is in fact about 47 miles (75.6 km). The importance of Whitesands Bay as a point of crossing to Ireland is underscored by the excavation of some 50 skeletons at Saint Patrick’s Chapel, dating to the early medieval period (seventh to eleventh century), in 2015, under the direction of Ken Murphy, Chief Executive Officer, DAT. Interestingly, a large hoard of Roman coins discovered at Whitesands Bay in 2010, recorded by the PAS, comprises 115 coins of several denominations dating from around the mid-second to the mid-fourth centuries. There is no other certain evidence for a Roman presence in the area apart from an unconfirmed fort at Trepewet east of Saint Davids.
The curious irony of the Via Julia
There are reasonable grounds to assume that a road runs from Saint Davids to Carmarthen, roughly parallel to the Carmarthen-Wiston route (Fig. 7). Is this the ‘real Via Julia?’ Place-names, as recorded on the older OS maps, and archaeological evidence play a crucial role here. For the sake of convenience, I reference here the OS Six Inch (1888-1913) map from this point onwards. We pick up the chase again near Brawdy, with reference to the place-name ‘Broadway’ (SM850238) near the southern end of the old runway of the military base. There is possible archaeological evidence for a Roman road a little to the south of Brawdy at Lower Llethr, photographed from the air by Dr Toby Driver, Senior Investigator (Aerial Survey), RCAHMW, in 2003, running west-east (although the possibility that it may be a pipeline should not be discounted). In the same year, a parchmark was photographed by Driver near another ‘Broadway’, located west of Llawhaden, part of the Carmarthen-Wiston Roman road. It is therefore reasonable to assume that the name may also indicate the presence of a Roman road near Brawdy.
Significantly, just a few kilometres to the east of Broadway, is another identical place-name (SM917245) and also Broadway Quarry (922244). This lies not far to the southwest of the Ford villa near Wolfscastle. A large section of hedged trackway runs in a general east-northeast direction for some distance, commencing as South Hill Lane (932250), past a site labelled as ‘Old Camp’, and below the course of the Roman road marked on the old OS maps.
Further east, near the Roman villa at Castle Flemish, the old OS maps bisect the site and this is probably because it was originally thought to be the Roman station Ad Vigesimum. It is plausible to think that a Roman road runs further to the south, since there is a place-name a little to the southwest of a farm called Stradland (994264), which may perhaps be interpreted as ‘Streetland’, referring to the presence of a Roman road, an observation that accords with that of Fenton. The passing of this suggested route south of Ford and Castle Flemish is perhaps not surprising, since, judging from examples elsewhere in Britain (Akeman Street in southern England is typical), Roman villas tend to be located near Roman roads, as in the case of the possible route identified by Martin Davies from Wiston to Ceredigion.
There is reason to suppose from place-names on the old OS maps that the road continues east, crossing this north-south road suggested by Davies somewhere near LIys-y-Fran. The next relevant place-name, Pen-sarn, is about seven kilometres to the east-southeast, near to the western bank of the Eastern Cleddau (SN111249). It is appropriate at this convenient geographical juncture to consider the slates found at Ford, Castle Flemish, Caerleon, and elsewhere, since this stretch of road is not far south of their probable source. The comments of George Boon on the slates found in the secondary metalling of the quayside at Caerleon are especially relevant: ‘The blue-grey, thinly cleaved, spotted slate was readily recognised as being strikingly similar in external appearance to the Foel Tyrch Beds Arenig Slate. This Ordovician slate is exposed and has been extensively quarried in the southern part of the Prescelly Hills. The specimen was also compared with other slates and slate-like rocks from various southern Dyfed localities; but even on macroscopic examination only, it almost certainly came from Prescelly.’ These are dated to the later third-century, when the quay was extended, which obviously indicates that a quarry was operational in this period. It is not clear exactly where the slates were quarried, but I am grateful to Robin Sheldrake, a local historian, for drawing my attention to the slate quarry at Llangolman (128270). It is curious that its location is little more than a kilometre north of Pen-sarn, and there are several other quarries in the area, such as Tyrch, some five kilometres to the northeast (154295). While post medieval activity is attested at both sites, there is scope for further investigation to establish if any Roman slates have been worked or discarded. In any case, the proximity of southern Preseli slate to the suggested line of the Roman road may well account for its distribution at Romano-British sites close to its general latitude.
Approximately three kilometres to the northeast of Pen-sarn, at Glandy Cross, appear the names Parcsarnau and Sarnau, a short way to the northeast (145266). As with ‘the Causeway’ near Camrose, it is often the case that crucial names have been removed from the modern OS maps, such as Parcsarnau in this case.
The next place-name, Pen-sarn, is east of Login (193238). Then east-southeast of Llanboidy there are a cluster of relevant place-names, tracking towards Carmarthen and all labelled to the north of the Via Julia as marked on the old OS map. Approximately 1.5km to the north of the first Sarnau (211253) is an earthwork labelled as ‘Caer’ (‘Y Gaer’ on the modern OS map), where a hoard of Roman silver coins is recorded. Further to the southeast are: Blaen-sarn-goch (238222), Sarn-goch (232218), Sarn-newydd (236219), Caerlleon (256220), Efail-Caerlleon (257219), Penyrheol (308208), Sarnau (313209), and Sarn-y-bwla (318205).
This concentration of place-names did not escape the attention of the late Professor Barri Jones. In the early 1970s he identified a corresponding agger west and east of Meidrim, on a general east-west alignment towards the confluence of the Afon Cywyn and the Nant Cynnen, near Rickett’s Mill in the direction of Saint Davids Hospital. In this context, I am indebted to Martin Davies for his diligent observation in the field during March of this year; augmented by his Lidar analysis in this area, both avenues of inquiry appearing to support the existence of a well-defined route along this course (Fig. 8).
Fig.8. Lidar survey detailing the Roman road west of Meidrim proposed by the author, based on the observations of Martin Davies
The most extraordinary thing is that the collective evidence, from archaeology and place-names, appears to confirm the existence of a Roman road that roughly follows the spurious Via Julia, although I am not suggesting that Bertram’s Itinerary was genuine, but rather coincidental.
It is perhaps appropriate to conclude by considering the likelihood of future discoveries, and I am convinced, as are others, that a number of villas exist, but with perhaps fewer large forts to be established along the emerging Roman road network. It is not difficult to envisage the presence of a Roman road linking the Cwmbrwyn villa with the Trelissey site at Amroth, especially sin ce there is reason to suspect another site nearby at Eastlake Farm. In the meantime, the forthcoming excavation of the Roman villa discovered by UCL in north Pembrokeshire may even produce the long-awaited discovery of some Roman mosaics. Such is the unpredictability and excitement of archaeology.
 A more technical paper on the results of the 2003 excavation is in preparation.
3 The most recent study on Roman Pembrokeshire is presented by James, H., ‘Roman Pembrokeshire AD 75 – 410’. In Pembrokeshire County History, Vol. 1, Prehistoric, Roman and Early Medieval Pembrokeshire, edited by H. James, M. John, K. Murphy, and G. Wainwright (Haverfordwest: Pembrokeshire County History Trust, 2016), 293-339.
 Rathmann, M., Tabula Peutingeriana. Die einzige Weltkarte aus der Antike (Darmstadt: Philipp von Zabern, 2016).
7 Bertram, C., The Description of Britain: Translated from Richard of Cirencester: with the Original Treatise, De situ Britanniæ; and a Commentary on the Itinerary (London: J. White & Company, 1809), 144.
12 ‘In the course of a further examination if this site [at Ford] on the 14 March, 1924, when the surface of the ground was clear of vegetation and the soil in the adjoining field had been turned over by the plough, several pieces of slate roofing tiles of distinctly Roman appearance were picked up, of which we append an illustration. There can be no doubt that a Roman building of some description has occupied the site’ (No. 305A). Royal Commission on the Ancient and Historical Monuments and Constructions in Wales and Monmouthshire, An inventory of the ancient monuments in Wales and Monmouthshire. VII, County of Pembroke (London: HMSO, 1925), 117.
It is probable that a roofing slate donated to the National Museum of Wales, Cardiff by W.J. Hemp (Acc. No. 27.317), is one of these slates. For this information, I am grateful to Dr Richard Brewer, Keeper of Roman Antiquities at the National Museum of Wales.
13 Arnold, C.J., and Davies, J.L., Roman and Early Medieval Wales (Stroud: Sutton, 2000), 74; Crane, P., ‘Iron Age Promontary Fort to Medieval Castle? Excavations at Great Castle Head, Dale, Pembrokeshire’, Archaeologia Cambrensis 148 (1999), 86-145; Williams, G., and Mytum, H., Llawhaden, Dyfed: excavations on a group of small enclosures, 1980-1984, edited by K. Blockley. BAR British Series 275 (Oxford: Archaeopress, 1998); Benson, D.G., Evans, J.G., Williams, G.H., Darvill, T., and David, A., ‘Excavations at Stackpole Warren, Dyfed’, Proceedings of the Prehistoric Society 56 (1990), 179-245; Davies, J.L., Hague, D.B., and Hogg, A.H.A., ‘The Hut-Settlement on Gateholm, Pembrokeshire’, Archaeologia Cambrensis (1971), 102-110; Wainwright, G.J., ‘The Excavation of a fortified Settlement at Walesland Rath, Pembrokeshire’, Britannia 2 (1971), 48-231.
 Driver, T., Pembrokeshire:Historic Landscapes from the Air.
(Aberystwyth: CBHC/RCAHMW, 2007), 174-177.
 Carroll, M., and Lang, A., ‘The Iron Age’, in R. Adkins, L. Adkins, and V. Leitch, The Handbook of British Archaeology, revised edition (London: Constable, 2008), 104.
 Ward, T., ‘Roman remains at Cwmbrwyn, Carmarthenshire’, Archaeologia Cambrensis 24 (1907), 175-212.
 Wainwright, G.J., Coygan Camp: a Prehistoric, Romano-British and Dark Age settlement in Carmarthenshire (London: Cambrian Archaeological Association, 1967).
 Boon, G.C., Roman Sites (Cardiff: Cambrian Archaeological Association, 1978).
 Schlee, D., Archaeological Investigations at Upper Newton ‘Roman Villa’, Wolfscastle, Pembrokeshire 2010 (Llandeilo: Dyfed Archaeological Trust, 2010), 13.
 Meek, J., Archaeological Investigations at Wiston Roman Fort and its Environs, Pembrokeshire 2014: Interim Report (Llandeilo: Dyfed Archaeological Trust, 2015).
 Luke Hooper, personal communication, 6 January 2017.
 Martin Henig, personal communication, 6 January 2017.
 Davies, M., Ancient Causeways Uncovered (Cardigan, 2017).
 Millett, M., The Romanization of Britain (Cambridge: Cambridge University Press, 1990), 66.
 Merrony, M.W., The Plight of Rome in the Fifth Century AD (London: Routledge, 2017), 81.
The most advanced study of Roman Carmarthen is presented by James, H., Roman Carmarthen: Excavations, 1978-1993 (London: Society for the Promotion of Roman Studies, 2003).
31 Fulford, M.G., ‘The Second Augustan Legion in the west of Britain’. In Birthday of the Eagle: The Second Augustan Legion and the Roman Military Machine, edited by R.J. Brewer (Cardiff: National Museums & Galleries of Wales, 2002), 83-84.
 Tacitus, Annals, Books 4-6, 11-12, translated by J. Jackson (Cambridge, MA: Harvard University Press, 1937).
 Giraldus Cambrensis, The Journey through Wales and The Description of Wales (London: Penguin, 1978).
 Charles, B.G., The Place-Names of Pembrokeshire, 2 vols (Aberystwyth: National Library of Wales, 1992), 198.
 Bishop Wyn Evans, personal communication, 3 January 2018.
 Pliny, Natural History, Vol. II, Books 3-7, translated by H. Rackham (Cambridge, MA: Harvard University Press, 1942).
 Murphy, K., Shiner, M., and Wilson, H., Excavation at St Patrick’s Chapel 2015 Interim Report (Llandeilo: Dyfed Archaeological Trust, 2015).
 This appears on the RCAHMW website Coflein as NPRN 416862 (National Primary Record Number).
 NPRN 413847.
 Boon, 1978, 11.
 Boon, 1978, 2.
 Robin Sheldrake, personal communication, 6 February 2018.
 Llangolman quarry (412706), Tyrch quarry (401348).
 NPRN 410762.