Edited by Neil Holbrook from contributions by Peter Davenport, David Evans, Neil Holbrook, Tony King, Peter Leach and Chris Webster
The start of Roman Britain in the South West is traditionally dated to the campaigns of the Roman army under the future emperor, Vespasian in the years following the invasion of AD 43 under the imperial command of Claudius. This is based on the statement of Suetonius that Vespasian’s forces fought 30 battles, took 20 oppida and the Isle of Wight. The reference to the Isle of Wight is the only geographical location given and has led to the assumption that this campaign was located in the south and west. The traditional picture of the invasion is being challenged even to its landing place and the discovery of a legionary tombstone in Oxfordshire has led some to question the location of the campaigns of Vespasian (Sauer 2005). There is also increasing evidence that Roman “interference” in Britain began in the reign of Gaius, or perhaps even earlier (Creighton 2001), and the presence of pre-conquest Roman imports at sites such as Hengistbury Head is well-known (Cunliffe 1987).
Archaeologically the earliest indication of a Roman presence in the region is a dated lead panel (AD 49, RIB 2404.1); a date that has been questioned, but that is now supported by Claudian pottery from the fort at Charterhouse-on-Mendip (Todd 1995; 2007). Other early material comes from Dorset, for example, Lake Farm and Hamworthy but the construction of the legionary fortresses at Exeter and Gloucester appears to date to the 50s AD. Away from the military sites and centrally planned civitas capitals there is little evidence of change, and earlier settlement forms persisted. It appears that more “Romanised” forms of settlement develop later, such as roadside settlements (perhaps from the 2nd century AD) and villas (primarily a 3rdand 4th-century AD phenomenon). These changes within the Roman period are clearly in need of reconsideration as there has been a tendency in the past to examine each in isolation (the military, the towns, the villas. . . ) with little consideration of their relationships through time and space.
While the origins of Roman Britain are not as clear cut as they might seem at first, defining the end is even more fraught with problems. For example, what is the relationship between the later Roman prosperity of the eastern half of the region and such factors as the drainage of the Somerset and Severn Levels, salt production, the Dean and Mendip mining and other production industries, or the Poole Harbour industries? There is also the question of the region’s presumed inclusion within the late province of Britannia Prima. In the 5th century AD two basic schools of thought exist: a short chronology comprising a total systems collapse, (see for example, Esmonde Cleary 1989) and a long chronology which sees the continuance of Roman ideas for several centuries, albeit without new material culture which would ease its identification (such as proposed by KR Dark 2000). Many of these questions are discussed later but they have their origins in the Roman period itself. The question of what exactly we mean by Roman has been partly addressed by Faulkner (2002; 2004) and will not be re-addressed here, but it is crucial to our understanding of the eschatology of Britain’s role in the Roman world.
There have been many excavations of rural sites within the region but historically most have concentrated on high status settlements, principally villas, although this is changing as development-led excavation has started to redress this imbalance. In addition, many of the villa excavations were focused on the recovery of the ground plan of the main residential buildings, and the reconstruction of their structural history, rather than exploring their social and economic context. The ground plans do allow comparison of size, perhaps a reflection of wealth, while the publication of the mosaic corpus volume for our region has further emphasised the exceptional representation and quality of these sites (Neal and Cosh 2006). In contrast, a change of emphasis towards a better understanding of agricultural, social and economic life has so far had only limited application and will only be advanced through defining a wider set of research aims.
Full modern publication of work upon these high status sites is a rarity, notable exceptions being Gatcombe (Branigan 1977), Halstock (Lucas 1993), Great Witcombe (Leach 1998), Roughground Farm, Lechlade (Allen et al. 1993), Kingscote (although its precise status is still uncertain, Timby 1998), Frocester (Price 2000) and Great Bedwyn (Hostetter and Howe 1997). New discoveries continue to add to the extensive corpus of villa sites across the region, which has the highest density of identified sites in Roman Britain, although their marked absence in Devon (west of the Exe) and Cornwall persists. The only recent discovery in this area is at Crediton (Griffith 1988). Major new sites in Somerset include Dinnington (excavated by Time Team in 2002 and 2005, and now the focus of a research project by Winchester University), Hinton St George and Lopen near Ilminster, Yarford north of Taunton and Iford to the south of Bath. Recent work and new discoveries in Wiltshire are usefully summarised by Walters (2001), to which new sites near Malmesbury (Hart et al. 2005), Groundwell Ridge and Bradford-on-Avon can be added. The last two are now the subject of on-going research projects (Corney 2003 and English Heritage website). New sites in Dorset include Shillingstone and Minchinton, and in south Gloucestershire Badminton, Horton and Hawkesbury. New sites continue to turn up in the Gloucestershire Cotswolds (Holbrook 2006), most notably Turkdean (Holbrook 2004).
There has also been a number of significant investigations of non-villa rural sites, notably on the higher ground of the Cotswolds, Cranborne Chase, Salisbury Plain, Mendip and the Quantock Hills, of both nucleated and more isolated single settlements. Despite these, detailed investigations on the scale applied at Catsgore (Leech 1982a; Ellis 1984) – still a type-site for nucleated Romano-British agricultural settlements – are rare. This is doubtless a reflection of the nature of the great majority of modern site investigations, determined by development threats and a tailored response. Adequate publication of work on many such sites still lags far behind the range and scale of their field investigation, an extensive source of data vital to any understanding of the history and dynamics of rural settlement and land-use in the region.
Wider, more integrated, surveys such as those undertaken on Salisbury Plain (Fulford et al. 2006), in South Somerset and North Dorset (Leech 1982b), on the Somerset and Severn Estuary Levels (Rippon 1997a; 2000) and recently completed on the Quantocks and their environs, by English Heritage (Riley 2006) and Winchester University, have the potential to set both high status and other rural settlements in their wider agrarian, social and economic contexts. Methodologically in-depth, closely defined surveys such as that at Shapwick (Aston and Gerrard 1999; Gerrard and Aston forthcoming) and the South Cadbury Environs project (Tabor 2002; 2004b) in Somerset have demonstrated the potential to resolve key questions at the micro level but within a multiscalar context. Several of the above have seen the use of large-scale geophysical surveys in tandem with excavation which has produced dramatic results. Other work, in Somerset (Leech et al. 1993), Gloucestershire (Holbrook 2003), Wiltshire and Cornwall (AM Jones 2000–1) has shown the high potential for this use of large area geophysical survey.
Because of the concentration on the plans and structures, the origins of villas in the South West require much more investigation. The few sites extensively excavated in modern times have shown the potential for adding to both our understanding of the Iron Age to Roman transition and of the early post-Roman period. Outside Devon and Cornwall, the concentration of villas around towns such as Cirencester, Ilchester and Dorchester, with their floruit so markedly in the 4th century, is especially notable and these need to be tied into broader, social, enquires into urbanisation (and deurbanisation), wealth accumulation and population movement.
The chronologies, transitions and context of nonvilla settlements are equally problematic. The potential for further research and excavation of sites known predominantly through field survey on Cranborne Chase, Salisbury Plain, the upper Thames valley (Leech 1977) and on the uplands and marginal lands of Devon and Cornwall (Griffith and Quinnell 1999) is underlined through the extensive excavation of Trethurgy in mid-Cornwall (Quinnell 2004a). This is a high priority for, with a few notable exceptions – Bradley Hill (Leech 1981b; Gerrard 2004a) and Catsgore (Leech 1982a; Ellis 1984), sites on Purbeck such as Ower (Sunter and Woodward 1987), Bestwall (forthcoming, see http://http://www. bestwall.co.uk) or Worth Matravers (Graham et al. 2002), Butcombe (still largely unpublished), Chisenbury Warren (Fulford et al. 2006), Birdlip Quarry on the Cotswolds (Mudd et al. 1999), sites in the Gloucestershire Severn valley around Tewkesbury and Gloucester (A Thomas et al. 2003; Walker et al. 2004) and Claydon Pike in the upper Thames valley (Miles et al. 2007) – so little is known of the lesser status rural settlements, villages, farmsteads, hamlets and other forms of settlement in which the majority of the population of the region must have lived. For Devon we still lack the complete plan of any rural settlement, and for the south-west peninsula as a whole, the data from Trethurgy has emphasised the potential for investigation of settlements with a high degree of survival of internal structural detail as stone has been used as the primary building material.
Rural settlement dynamics need to be set within a wider geographical, economic, social, demographic and temporal context (for Cornwall, see Herring forthcoming). There is a particular need for more work on the Roman/Post-Roman transition focusing on settlement forms, ownerships, communications and land-use/field-systems which may have continued functioning at different levels within this period.
The region contains one colonia (Gloucester: Glevum) and four, or five, civitas capitals: Exeter (Isca Dumnoniorum), Dorchester (Durnovaria), Cirencester (Corinium Dobunnorum), Ilchester (Lindinis) and just possibly Bath (Aquae Sulis). Other sites which might fall under the catch-all term of small town/market centre include Bath (Aquae Sulis), Shepton Mallet, ?Keynsham/Bitton(Traiectus?), Sea Mills (Abonae) Bourton-on-the-Water, Wycomb, Dymock, Coln St Aldwyns/Quenington, Wanborough (Durocornovium) and Mildenhall (Cunetio). Very small or uncertain towns include Tewkesbury, Dorn, Sandy Lane (Verlucio?), Camerton (Wedlake 1958), Easton Grey, Nettleton Shrub, Old Sarum (Sorviodunum), Badbury Rings (Vindocladia), Pomeroy Wood, Honiton or Woodbury, Axminster (Moridunum?) and perhaps Catsgore (Leech 1982a; Ellis 1984). The recently discovered site at Hall End, Wickwar appears to be another roadside settlement (Young 2003).
The known distribution, as outlined above, is biased towards the northern and eastern end of the region, with very few towns (in the broadest sense) west of Ilchester and Dorchester and with the main concentration in north Somerset and Gloucestershire; a pattern which does not seem to relate to physical landscape types. While, to some extent, this reflects the traditional (and more recent) concentration of archaeological work in these areas, no new sites have been discovered during extensive and intensive survey in several areas and it seems likely that the known distribution reflects long-accepted views on the different nature of societies in the east and west (or north and south?) parts of the region. To this extent understanding towns will have significant role in understanding the subtleties of societies that in one case welcomed towns and, in the other, rejected them.
This raises questions of definition, and more work needs to be undertaken to clarify what we mean by “town” (or more realistically agree the broad parameters within which various understandings of “town” can be discussed). This is particularly pressing for the so-called “small towns”. For example, why is Camerton usually described as a town and Catsgore as a village? Both lie on a road, have industry and evidence of social differentiation. Similar questions can be raised over other “villages”, on Salisbury Plain and elsewhere. Differentiation between a village and a small town will only come through an understanding of which activities actually went on in a given place, rather than a concentration on morphology. Some detailed study, including excavation, has been undertaken in an ad hoc way (for example Corney 2001; Leach and Ellis forthcoming; Timby 1998) but there is a need for more and, of course, full publication of work already done.
The definition of larger towns, for our purposes, civitas capitals and coloniae, is a simpler proposition, but there are still uncertainties. Ilchester (Leach 1982; 1994) is now largely accepted as a late sub-civitas capital, but how it acquired this status is unknown; did this have as much to do with its pre-Roman status as any economic, administrative or patronage issues? Did its late elevation require the construction of the usual structural elements of an earlier civitas capital, principally a forum and basilica? We equally have little idea of the relationship of economic to administrative status or, on this fringe area of urbanisation, the relationship between wealth (expressed in later Roman Somerset and the Cotswolds in massive building schemes in the countryside) and the size and sophistication of towns. Published and unpublished work in Cirencester (Holbrook 1998; Darvill and Gerrard 1994), Bath (Cunliffe 2000b), Ilchester (Leach 1982; 1994) and Shepton Mallet (Leach and Evans 2001) may go some way to answer this. In Gloucester (see Hurst 1988; 1999b;c) there are many important excavations dating back to the 1960s which still await publication and in particular we still know little about the later history of the colonia, especially in the 4th century.
In other parts of the region, the character and success or otherwise of the civitas capitals is unclear. Exeter, for instance “does not have the usual trappings of an urban centre; few substantial buildings have been located, the material culture is relatively ‘poor’, there is very limited epigraphic evidence and there is no evidence for villas/supporting agricultural economy in the hinterland. Major excavations are underway in the town but watching briefs in the supposed suburban area have been uninformative” (Peter Weddell, pers. comm.). This leaves questions such as was Exeter really atypical, did it reflect the character of the region, or was it a failure as an urban foundation?
The same, with less justification, could be said of Dorchester (see PJ Woodward et al. 1993 for an overview). While the topography of the town is slowly coming to light (such as the forum gravelling over a large length of recent service trenches in High East Street), little enough of its character is yet known. Its hinterland is better known and there is the clear possibility that the relationships of town to country over time can be studied in the future.
We are getting to grips with Bath (see Cunliffe 2000b); its origins and economy in the earlier Roman period are becoming clear, and there appears to be a change in the later Roman period. Dating of this and the later sub- and post-Roman changes are less clear. Bath, of course, is atypical and its status is unknown. Work at Shepton Mallet (Leach and Evans 2001) is very helpful and points up the importance of Somerset (real or is it chance?) in providing opportunities to study the late- to post-Roman transition. Ilchester (Leach 1982; 1994) is perhaps one of the most important potential sites for a range of these studies.
Several of the small towns have been examined in recent years, with Shepton Mallet one of the most extensively excavated in the country (Leach and Evans 2001; Leach and Ellis forthcoming). Others that have seen modern publication include Kingscote and Wycomb (Timby 1998), Mildenhall and other sites in Wiltshire (Corney 2001) and Woodbury, Axminster (Silvester and Bidwell 1984; Weddell et al. 1993; Cole and Linford 1993).
Towns were presumably foci for trade, and finds and environmental issues are fundamental to an understanding of this topic. The identification of small ports on the Severn coast, and rivers leading to it, in the medieval period suggests we should be able to identify some for the earlier ones: Sea Mills (Ellis 1987), Crandon Bridge (currently being written-up by Stephen Rippon at Exeter University) and Combwich might repay closer attention. Ports might also be expected on the Camel estuary around Padstow and perhaps in Barnstaple Bay (although no trace of Roman activity has so far come to light beneath the medieval towns of Barnstaple or Bideford (Holbrook 2001). On the Channel coast, Hamworthy, Seaton, Topsham, Mount Batten, St Michael’s Mount (Herring 2000) and probably somewhere on the Fal estuary are all likely ports.
It is probable that galena deposits on Mendip were exploited in the pre-Roman period, but the main phase of the lead industry there begins just after the conquest (Todd 1996; 2007). Fort construction at Charterhouse-on-Mendip dates from the late 40s AD, linked to surface extraction and digging of pits into mineral-rich veins from AD 50/75 onwards (Todd 1996; 2007; Whittick 1982). The highly characteristic series of ingots confirms the growth of the industry during the 1st and 2nd centuries AD, and their distribution indicates export mainly to the east and south via a road running from Mendip to Salisbury, Winchester and Bitterne, whence to Gaul and beyond (RIB II 2404.4–13; Britannia 27, 1996, 446–8, nos 11–13; Elkington 1976; Todd 1996; 2007). Official involvement is thought to have declined by the 3rd century AD, and exploitation passed to private interests. As a consequence, later Roman evidence in the Charterhouse area is less clear, but is still definite (Todd 1996; 2007). More conclusive is the large quantity of Late Roman lead products, including coffins, water pipes and pewter (Elkington 1976, 195–7). It is possible that the lead and silver deposits on Exmoor and at Combe Martin might have been exploited although evidence is currently lacking.
The early date for lead exploitation contrasts with the mainly late Roman evidence for tin extraction in Cornwall, probably due to the easy availability of tin from Iberia in the 1st and 2nd centuries AD (Todd 1987, 231; Penhallurick 1986). Direct evidence for Roman tin streaming is lacking, except at Boscarne, near the early Roman fort of Nanstallon, which may imply an early official interest in the mineral, never followed up (Penhallurick 1986, 210–11; Fox and Ravenhill 1972, 90–1). Recent work on Dartmoor is also now trying to prove the Roman date of working widely assumed to have taken place (Thorndycraft et al. 2004). There are, however, Roman finds near the tin grounds, such as at Reskadinnick and, of course, the small villa at Magor (Penhallurick 1986, 158). Better evidence comes from artefacts, notably the tin vessel (96% pure) from near Caerhays Castle, containing a coin hoard dating to the 270s (Penhallurick 1986, 187). The artefactual evidence, including the small assemblage of ingots, suggests a greater level of activity from the 2nd century, and especially in the later Roman period (Penhallurick 1986, 200–13), the Trethurgy Round tin ingot being a good example (Quinnell 2004a, 72–3, fig 47; Penhallurick 1986, 228–9, fig 128).
Primary iron-ore working (as opposed to secondary working) is focused mainly on the Forest of Dean, but recent fieldwork has cast new light on the regionally important iron industry on Exmoor (Riley and WilsonNorth 2001, 78–81; Wilson-North pers. comm.). Iron working also occurred widely on the Blackdown Hills from an early date, production sites stretching as far west as Gingerlands, Cullompton (Griffith and Weddell 1996; Reed 1997; Bill Horner pers. comm.).
In the Forest of Dean, the clearest evidence comes from the two adits into ore-bearing strata at Lydney (Wheeler and Wheeler 1932, 18–22), from Newent, which appears to be an early Roman industrial settlement of c.47ha, with evidence for iron working, similar to that at Weston-under-Penyard (Herefordshire) on the north side of the Forest (Walters 1992, 82–4), and from villas such as the Chesters, Woolaston, where late Roman furnaces have been excavated (Fulford and Allen 1992). At the Chesters, ore was brought in from nearby, whilst the furnaces were fuelled by charcoal from managed woodlands. The villa-based iron working seems to have replaced the larger industrial settlements in the Late Roman period, probably at a smaller scale and as part of a mixed economy with farming (Fulford and Allen 1992, 200–5; Walters 1992, 91, 95). A fourth centre of iron making lay in South Gloucestershire around Mangotsfield where iron-making settlements have been partially examined (Holbrook 2006).
Coal was present at the Chesters villa, of local origin and probably used as a domestic fuel, not for ironsmelting (Fulford and Allen 1992, 191). The distribution of Roman coal finds in the South West is confined largely to Somerset, Gloucestershire and western Wiltshire, and the coal is derived from the local coalfields in the Forest of Dean and the Bristol and North Somerset area, as might be expected (Dearne and Branigan 1995; AHV Smith 1996; 1997). Use of coal seems to have been locally significant and becomes more widespread in the Late Roman period, but did not replace wood or wood-charcoal throughout most of the region.
Stone for building was much more readily exploited, and was clearly more significant economically. The higher quality stone sources in the region, notably Purbeck, Ham Hill, Chilmark stone, Lias limestones and oolitic limestone from the Cotswolds and the Bath area, were all fully exploited from early in the Roman period, as building material and architectural and other details on both public and domestic buildings clearly testify (Blagg 2002; Palmer 1996; Williams 1971). Purbeck was exported out of the region for fine-grain uses such as inscriptions (e.g. RIB I 91, Chichester, 1st century AD), as was oolitic limestone for monuments and sculpture (for example, Classicianus’s tombstone, London, 1st century AD, Grasby and Tomlin 2002, 47–9), but generally usage was local to the quarries for most building stones, due to transport costs. Similar considerations applied to ceramic building materials such as roof and floor tiles, as the well-documented inscribed tiles of the Cirencester and Gloucester region demonstrate very well (McWhirr and Viner 1978).
Other stone products included querns and millstones from sites in the Forest of Dean (Shaffrey 2006), Mendip and also at Pen Pits on the Somerset/Wiltshire border.
Shale suitable for working into artefacts is present in Britain only in Purbeck, and consequently this became a widely distributed specialist product of the South West. Iron Age production is well attested, utilising specialised flints as tools and making use of simple lathe technology. The use of the lathe increased during the Roman period, and led to the large-scale production of armlets, tableware and other artefacts (Sunter and Woodward 1987, 6, 145). Larger items such as table legs were hand-carved. Excavation of production sites in the Purbeck area (Sunter and Woodward 1987) shows that production was often combined with stone working, pottery and salt production, so that Purbeck and the Poole Harbour area can be regarded as a region that was probably more dependent on these industries than on agriculture. This combination of desirable products led to a thriving export trade, mainly by sea, as far as northern Britain and Gaul.
Poole Harbour was a major centre of salt production, as sites such as Ower (Sunter and Woodward 1987, 6–8) testify but other coastal zones in the South West region also have evidence for salt production, notably the coastal parts of the central Somerset Levels where over 100 sites are recorded in the HER (Leech and Leach 1982, 70–1; Rippon 1997b, 65– 77). Development work has also produced more evidence in the North Somerset Levels at sites such as St Georges, Weston-super-Mare (Vince Russett pers. comm.). There is also evidence for saltmaking on Lundy (Quinnell 2004b), a location which would appear unlikely for large-scale production due to lack of fuel and transport difficulties.
It was also the Poole Harbour area that was the location the most successful pottery production in the region. South-East Dorset Black Burnished Ware (BB1) was made from the conquest period, deriving from pre-Roman traditions, until the late 4th century AD (Tyers 2004). At its greatest extent in the 2nd and 3rd centuries AD, the pottery was distributed throughout southern Britain and had a strong presence on Hadrian’s Wall and the Antonine Wall, probably due to military contracts (Allen and Fulford 1996; Holbrook and Bidwell 1991, 88–114). Remarkably, the pottery was hand-made and relatively primitively fired, but was successful due to its thermal qualities as cooking ware (Farrar 1973). Production sites have been recently excavated at Ower (Sunter and Woodward 1987, 44–123), Worgret (Hearne and Smith 1992) and Bestwall (Ladle 2004). Black Burnished wares were also made during the late 1st to mid3rd centuries AD elsewhere, perhaps the Ilchester area, (South-Western BB1), but only achieved a more localised distribution (Tyers 2004; Swan 1984, fiche 4: 582–95; Williams 1977, 192–3). The suggestion that some of this material was being made on salt mounds in the Somerset Levels has not been borne out by recent excavations (Richard Brunning, pers. comm.). The burnished ware tradition also influenced several forms produced in South Devon, mainly in the late Roman period (Tyers 2004; Holbrook and Bidwell 1991, 177–81).
The other significant pottery production area was in North Wiltshire, where Savernake-type grey wares were produced at several sites in the vicinity of Mildenhall (Cunetio) in the early Roman period and achieved a fairly prolific but regional distribution in Gloucestershire, the Bristol area and Wiltshire (Tyers 2004; Timby 2001; Hodder 1974). There was also an extensive grey-ware industry in North Somerset (Rahtz et al. 1992, 148; Watts and Leach 1996, 98– 99) but the only kilns located have been at Congresbury (Usher and Lilly 1963/4; Scammell 1969: there is 280kg of pottery, yet to be published, in North Somerset Museum), although others clearly exist (Jane Timby, pers. comm.). It is possible that Roman production was set up on sites where Iron Age pottery had been produced, but by immigrant potters from the east who were following the Roman army westwards in order to supply the military market (Swan 1975). Military supply was also the motivation for the manufacture of highly distinctive pottery types at Kingsholm and Gloucester during the 1st century AD (Timby 1991).
Severn Valley Wares, previously considered also to have been set up to supply the military, are now seen to be pre-Roman in origin and separate from the Kingsholm/Gloucester types (Timby 1990). The manufacture of Severn Valley wares ranged from Somerset to Shropshire on several kiln sites of 1st- to 4th-century AD date; the distribution in the South West region was largely confined to Gloucestershire and the Bristol area (Tyers 2004). Red-slipped mortaria and other forms are known from Shepton Mallet (Scarth 1866) and there are other industries, such as South Devon Ware and South West White Slipped Ware that are poorly understood as yet.
Understanding of the ways in which communities identified themselves in Roman Britain, and their differing attitudes to the acceptance or resistance (passive or otherwise) of aspects of the Roman “package” are much debated at the moment (see Mattingly 2002; 2004 for useful summaries). The South West has much to contribute to this topic. The recent report on Trethurgy makes much play of the distinctive culture of the Roman period in Cornwall, and it is be wondered to what degree the area west of Bodmin Moor was integrated into the Civitas Dumnoniorum (Quinnell 2004a). By virtue of a common administration there has been a tendency in the past to view the civitates as uniform entities, whereas it is now clear that there are region patterns both within and across the them (see for example Hurst 2005). The actual political boundaries of the civitates are therefore probably irrelevant to a consideration of regional and subregional identity. For instance regional patterns in the distribution of villas are well known, and presumably reflect choice just as much, if not more, than the ability to generate local economic wealth. Differing burial practices, such as a continuance of local Late Iron Age traditions of inhumation into the Early Roman period, also help to identify societies.
Artefact studies also have a role to play in recognising the ways in which groups identified themselves. For instance study of the production and distribution of metalwork allows patterning to be identified, although on a broader geographical scale than the fine grain often detectable in coarse pottery. There are types of brooch which clearly show a south-western distribution. Made of leaded bronze, their main area of production may lie on Mendip (moulds have been found at Compton Dando; Butcher in Quinnell 2004a, 71). Ellen Swift (2000, 129, fig. 174) has also recognised a type of 4th-century strip bracelet which shows a clear distribution focused on the Severn estuary. The identification of a class of late 4th-century bronze belt buckles decorated with outward-facing horse-heads (Hawkes and Dunning 1961, type IB) has long been known and recently Mark Corney and Nicholas Griffiths have been working on a new study of the distribution of this material. The results of this important research are not as yet fully published, although Swift (2000, 2, 185) has summarised some of their key conclusions. Corney and Griffiths have considerably enhanced the original distribution map and show that over 70% of all finds are from the Cotswolds and Wessex, a distribution seemingly centred upon Cirencester. They suggest that production was limited to the late province of Britannia Prima and that there was military or official regulation of production and distribution. Whether such accoutrements were solely the preserve of soldiers is open to debate, but nevertheless the fresh study of these seemingly familiar artefacts is providing valuable new insights on the nature of state control in the latter half of the 4th century in our area.
In the aftermath of the invasion of AD 43 the Roman authorities created a framework of regional government based upon civitates, each with an urban centre where the administration was based; a system that had been successfully adopted elsewhere in the western Empire. The South West comprised a number of civitates, whose approximate bounds can be reconstructed mostly on the evidence of Ptolomy’s geography (Rivet 1964; Branigan 1976). It should not be assumed that the Roman civitates necessarily reflect the extent of pre-Roman groupings, and indeed in the case of the Dobunni this is almost certainly not the case.
In the far west were the Dumnonii whose eastern boundary is normally assumed to lie along the lines of the rivers Parrett or Axe in Somerset, and the east Devon Axe. Their centre lay at Exeter (Isca Dumnoniorum). To the east lay the Durotriges. The distribution of late pre-Roman Iron Age coins normally ascribed to the Durotriges, and so-called Durotrigian pottery, covers the area of Dorset, south Somerset, south Wiltshire and south-west Hampshire. Dorchester (Durnovaria) must have been the civitas capital, although it lacks its tribal suffix in the document known as the Antonine Itinerary. An inscription found on Hadrian’s Wall records work done by people from the civitas Durotrigum Lendiniensis. Lindinis (Lendiniae) is Ilchester, and suggests that by the 3rd century AD the Durotriges had been split into two parts, one centred on Dorchester, the other on Ilchester. The civitas of the Belgae was centred on Winchester, and Ptolomy records that it contained Bath. If Ptolomy is correct, it creates an odd shape for the civitas, so much so that it is generally regarded as a new creation of the Roman government carved out of parts of the territory of the Atrebates in the east and the Dobunni in the west. The southern boundary of the Belgae perhaps ran along the southern edge of Mendip and then eastwards towards Old Sarum before turning south along the Hampshire Avon. The Avon or Frome, and the Thames perhaps defined the boundary between the Dobunni and the Belgae.
In the later Roman period the province of Britain was divided into two, and then again into four provinces, the latter event probably in the reign of Diocletian or perhaps slightly earlier. By deduction the province of Britannia Prima must lie in the west, and two pieces of evidence have been advanced to champion the theory (often repeated fact) that Cirencester was the capital of Britannia Prima. First that it was the largest walled town within the bounds of the province (which are nowhere attested) and second that it has produced an inscription which records a rector (official) of Britannia Prima. It must be stated, however, that interpretation of this inscription is not straight-forward, and it is not conclusive proof that Cirencester was the capital. Indeed Richard Reece has suggested that Gloucester could just as easily have fulfilled this role (Darvill and Gerrard 1994, 74).
Overall, despite the considerable quantity of pottery manufactured in the South West and the export of BB1 to North Britain, the balance of ceramic trade was probably a negative one. In the early Roman period, fine wares were supplied from Gaul and elsewhere, whilst in the later Roman period, the Oxfordshire and New Forest kiln centres were able virtually to monopolise fine ware and even much of the coarseware distribution into the South West. However, other products discussed above, such as shale, certain building stones, tin and lead, were all exported from the South West, both to the rest of Britain and beyond. This was a significant element in the economy of the South West during the Roman period, and may have approached agriculture in importance in those areas where production was concentrated.
In the early years of Roman Britain, Roman religion acted as a unifying influence which brought and held Roman and Briton together. Later on, particularly in the 4th century AD, its effect was more divisive because of competition between cults (Christianity and Mithraism, Christianity and other pagan cults) and between different Christian heresies (Ferguson 1970; Green 1976; Henig 1984; Webster 1986). From the surviving material evidence the Olympian deities (such as Mercury, Mars, Apollo, Minerva and Hercules) dominated religious life in urban and high status rural communities, although frequently in concert with native deities: Apollo Cunomaglos at Nettleton Shrub (Wedlake 1982) or Sulis Minerva at Bath (Cunliffe and Davenport 1985). Local cults such as the Genii Cucullati (gods of healing, fertility and the other world) seem to have been popular in the Cotswold region, sometimes in association with a Mother Goddess, also from Celtic religion (Henig 1993). More research is required into the extent of these native cults and deities, their associations with Roman deities, and their manifestations as beliefs or observances within all levels of society (A Woodward 1992).
Relative to other regions there are many rural Romano-Celtic temples known in the South West, a high proportion of which have been examined using modern techniques. These include, in Somerset: Brean Down (ApSimon 1965), Pagans Hill (Rahtz and Harris 1957), Lamyatt Beacon (Leech 1986) and Henley Wood (Watts and Leach 1996), while temples are suspected at Brent Knoll, Cadbury Castle, Worlbury and Pedwell among others. To the south, in Dorset, lie Maiden Castle (Wheeler 1943) and Jordan’s Hill whilst in Gloucestershire are Lydney Park (Wheeler and Wheeler 1932; Casey and Hoffman 1999), Uley (A Woodward and Leach 1993) and others suspected at Chedworth, Lower Slaughter, Sapperton and Wycomb.
In Wiltshire, Nettleton Shrub (Wedlake 1982) and Cold Kitchen Hill are known with other suspected sites at Marlborough, Westbury Ironworks, and perhaps at Silbury Hill and Winterslow (Robinson 2001). The predominantly Late Roman floruit demonstrated at so many of these sites, their potential relationship with the Late Roman prosperity of the region and its great landowners, as epitomised by the wealth of the villas (Leech 1980), require further research. In this context recent research on sites such as Frampton (suggestive of a Gnostic cult, Perring 2003), or Littlecote (Bacchus, Walters 1996, although see Ling 1997), has shown the potential for regional re-evaluation of cult structures and practices within high status sites in particular.
There are also sites where “ritual” objects have been found whose exact nature is as yet unclear; these include Cadbury, Tickenham where an altar to Mars from the hillfort is now in North Somerset Museum and Steep Holm where a carved head of “RomanoCeltic” type is known (Rendell and Rendell 1993, 21– 22). Another head exhibiting high quality carving has been recovered from Frome (Bird 1985).
Romano-Celtic temples are not known from Devon or Cornwall but ritual shafts, into which offerings were placed, have been found within the hillfort of Cadbury Castle, Tiverton (Fox 1952b) and at Bosence, St Erth (Haverfield 1924, 8). In the Isles of Scilly, an unusual structure with a rich artefactual assemblage at Nornour (Butcher 2000/2001) is probably religious and an uninscribed altar is known from Hugh Town (Todd 1987, 230–1). Investigation and research into this aspect of rural life is important both for our understanding of the area and to perceptions about the extent and influence of Roman ideas and religious influence.
A review of temple sites in the region found that while there might be a siting policy for temples which placed them at boundaries in “neutral” zones, sufficient were placed away from postulated boundaries for it not be the sole consideration. Another suggestion that high places were a favoured locations was also examined and again found not to be a consistent feature. A study of intervisibility of temples might also be valuable as preliminary work showed that there was intervisibility from site to site (or very close by) from Lydney to Groundwell Ridge to Lamyatt Beacon to Cadbury Castle to Brean Down and back to Lydney (A Smith 2000).
In contrast is the almost total absence of known religious structures within urban contexts. Bath is here the significant and outstanding exception to the rule (Cunliffe and Davenport 1985; Cunliffe 2000b). The religious and cultural impact of this cosmopolitan centre upon our region deserves more study as does the relationship between pagan Roman and later Christian uses. The marked lack of evidence from the other major urban centres may be rectified by future discoveries. Portable finds or structural evidence from Cirencester, Gloucester and Dorchester hint at the presence of shrines or temples there and probably at some of the smaller nucleated settlements, such as Camerton (Wedlake 1958).
The survival of pagan Roman temple/shrine structures contrasts with the paucity of evidence for late Roman or early post-Roman Christian structures or portable finds, although this is by no means peculiar to the South West. The possibility of eremetic successors has been postulated at temple sites such as Brean Down, and Lamyatt Beacon in Somerset (Leech 1980), and the suggestion of Christian churches or other activity at Uley, Nettleton Shrub and possibly Bath requires reassessment in their wider context (Heighway 2003). Paganism appears to have been strongly entrenched among the landowners of the South West, but the unequivocal Christian symbolism at Hinton St Mary or at Chedworth demonstrates that this was not universal. In urban contexts Christianity makes fleeting appearances, except through burial, where a good case can be made at Dorchester, in the key Poundbury cemetery (Farwell and Molleson 1993). Christian burials can also be inferred in a number of other urban cemeteries and in at least one smaller settlement: Shepton Mallet (Leach and Evans 2001; see also 179).
There are many cemeteries and burials of the Roman period known within the region, some extensively recorded and discussed in the literature. Across the South West it is the cemeteries, and especially cremation burials, of the earlier Roman period which are least well understood (although see Foster 2001 for Wiltshire). Research has shown the potential of these data for better understanding the Iron Age/Roman transition and the impact of the Roman military upon the funerary and social landscape (Whimster 1981). A Late Iron Age tradition of inhumation burial, often crouched, in the Cotswold/Severn region persisted well into the Roman period in some places (A Thomas et al. 2003; Moore 2006a).
Cemeteries of the later and early post-Roman periods have been more extensively examined, the evidence from them, including the skeletal material, representing a considerable resource for further statistical analysis following on from previous research agendas (Philpott 1991; Rahtz 1977), including important questions relating to religious identity and belief.
With the notable exception of Poundbury (Farwell and Molleson 1993), extensive examination and analysis of urban cemeteries has been relatively limited. Smaller samples are published from the Allington Avenue cemetery, Dorchester (Davies et al. 2002), from Ilchester (Leach 1982; 1994) and from Cirencester (McWhirr et al. 1982). Both Gloucester and Bath had extensive cemeteries, and with Cirencester, have important groups of tombstones and memorials, but are otherwise relatively little explored (or in the case of Gloucester, published) while even less is known of Exeter’s cemeteries.
Rural cemeteries and individual burials are widespread and are probably to be found in association with almost every settlement type. Once again, modern publication of substantial groups is rare: examples include Bradley Hill (Leech 1981b), Frocester Court (Price 2000) and Hucclecote (A Thomas et al. 2003). Many settlement sites will have a few burials (see, for instance, the Wiltshire evidence summarised in Foster 2001 to which the newly excavated cemetery at Boscombe Down can be added). One other large and outstanding exception is Cannington, (Rahtz et al. 2000), where burial extended into the 7th century AD at least. Cannington also exemplifies an exceptional wealth of late and early post-Roman cemeteries in this region, notably in Somerset, such as Henley Wood (Watts and Leach 1996), Brean Down (Bell 1990), Lamyatt Beacon (Leech 1980) or Shepton Mallet (Leach and
Evans 2001). These sites and others in Dorset and Gloucestershire are of outstanding importance for exploration of the Roman/post-Roman transition in the South West and the impact of Christianity (Rahtz 1977), though it should be remembered that virtually nothing is known of burial practices in Devon and Cornwall.
In the context of current research concerns there is very great potential for the scientific study of the skeletal populations now in museum storage, for biometric data, genetic data, for strontium isotope analyses and other emerging techniques. Data on contrasts between urban and rural populations, population movements, on migrations into and from Britain, for example into and from Germania, Gaul, or Britanny, could be of the greatest interest. In looking further at the Roman/post-Roman transition any opportunities to examine burial populations within early churches and churchyards should be taken where possible.
The traditional account of the Roman invasion based on Suetonius’s account appears to be borne out by the discovery of early Roman military sites and material along a line through Dorset, South Somerset and Devon (such as Lake Farm, Hamworthy, Whaddon Hill, Maiden Castle, Hod Hill, Cadbury Castle, Ham Hill and Hembury). The legionary fortress at Exeter and some other forts, appear to be slightly later foundations. We still lack a good understanding of the purpose of such forts in lowland Britain. They do not appear to follow the pattern seen in the upland areas, where forts are sited in a network to control a hostile population, and might be seen more as winter quarters (Maxfield 1987). The pattern might reflect less the need to control a population and more a need for land to supply large numbers of troops. Some forts such as those at Charterhouse-on-Mendip and Nanstallon were associated with early control of mineral resources while it is conceivable that the distribution of forts around the fringes of the Blackdown Hills and on routes leading to Exmoor reflects military control of iron extraction in these areas. Pottery types normally associated with Roman military sites in Devon have been recovered in association with iron-smelting debris at Yarcombe (Bill Horner pers. comm.). The lack of forts in the northern part of the region may be a result of complex political origins on which which we can only speculate. The forts (and their foraging troops) might not, for instance, have been imposed on tribes which allied themselves to Rome, with the exception of Cirencester which might have housed a prestige formation to support the rulers at Bagendon. Whilst more forts have been discovered in recent years, for instance, Rainsbury (Riley and Wilson-North 2001, 71) and Topsham (Sage and Allan 2004), these have all been in Somerset and Devon suggesting that the distribution pattern that we currently see is essentially correct. The majority of the forts are situated in the Dartmoor/Exmoor fringes. Two forts on the north Devon coast at Martinhoe and Old Burrow indicate a military interest in the security of the Bristol Channel in the 1st century AD.
The chronological pattern at the legionary bases appears to be fairly clear with the establishment of a small base at Kingsholm, Gloucester in the late 40s, followed by fortresses at Exeter in the early–mid-50s and Gloucester in the mid-60s, although the exact movements and identities of legions involved is still a subject for debate. The smaller forts are less well dated although there is good evidence from Pomeroy Wood (from the 60s to the ?80s, Fitzpatrick et al. 1999), from Bury Barton (late 50s–80, Todd 2002) and from Bolham Farm, Tiverton (c.65–85/90 Maxfield 1991; Todd 2002 believes that the foundation date may be at least 5 years earlier).
The fortresses at Exeter and Gloucester were abandoned by the mid-70s as troops were redeployed to South Wales but the evidence from Cadbury Castle may indicate some continuing strife in the South West up c.AD 70 (Barrett et al. 2000). The fort at Tiverton remained in use until c.85/90 (Maxfield 1991). Although the Roman military remains in the South West are not as extensive as those of the upland areas, they were built during an important period of change in the Roman army whilst their early abandonment means that much of this evidence is undisturbed by later development.
All of the principal towns of the South West were walled, most first in earthwork with later stone replacements. In common with the rest of the province the appearance of defences at this date (the earthen ones appear to be late 2nd century) is a peculiarly British phenomenon and reasons to explain it differ (see Henderson 2001, 66 for a short summary). The colonia at Gloucester was walled early (around the beginning of the 2nd century AD), but of the civitates Exeter seems to have been the earliest masonry defence (perhaps no later than the early 3rd century AD, Henderson 2001, 74). Others followed in the 3rd and 4th centuries AD. Defences were also added to the smaller towns of Bath and Mildenhall (Corney 1997). A small Late Roman defended site at Dorn on the Fosse Way has more in common with socalled “burgi” found in the Midlands and an interpretation as a small official strong point seems likely (Timby 1998). The enigmatic site at Gatcombe was also walled (Branigan 1977); interpretations of this site differ, a private or public estate or perhaps even a temple?
There is currently very little evidence of a military connection with the South West after the end of the 1st century AD. A tile stamp of the Second Augustan legion, which probably dates to the 3rd century AD, has been found at Seaton. Holbrook (1987) sought to associate this with antiquarian references to a possible stone fort in the town, although evidence of the true nature of the site is still awaited. The enigmatic site at Oldaport on the south Devon coast which has been suggested in the past to be a possible Late Roman defended enclosure (for example by Todd 1987, 260– 62) has recently produced a Late Saxon date for its walls (Rainbird 1998; Rainbird and Druce 2004).
We are grateful for additional comments or contributions from Gail Boyle, Toby Catchpole, Peter Herring, Jane Hill, Bill Horner, Charlie Johns, Andrew Jones, Bob Jones, Steven Membery, Rebecca Newman, Andrew Smith, Jane Timby and Peter Weddell.
Part of the Research Frameworks Network