The emergence of villa landscapes

Introductory Note

Indications of between 50,000 and 70,000 rural sites of late Iron Age and Roman date are known in England (Mattingly 2004, 14), and there must be many thousands more in Scotland and Wales. These indications mostly consist of surface scatters of finds, and the results of geophysical survey and aerial prospection. The actual number of settlements that have been excavated is relatively small.

The most recent estimate of the number of villas in Britain has put the likely total at over 1,500 (Scott 1993, vi-vii). Previous estimates have been much lower, typically in the region of 500 to 600. The higher figure can be accepted because it is based on a survey of all the site evidence throughout Britain, drawing on the County Sites and Monuments Records. As is the case with the rural sites, only a small percentage of these sites has been excavated, and many of those excavations were carried out at an early date and produced little information about the chronological development of the buildings.

Two separate areas have been selected for more detailed study in this section. The first area for study (Western England) centres on the Cotswolds, a plateau of Jurassic rocks, mostly oolitic limestones, which dips eastwards from a crest which overlooks the estuary of the River Severn. The area also includes the Severn estuary and the parts of the upper Thames valley, a zone with a gravel subsoil. In the immediately pre-Roman period much of the area formed the territory of the Dobunni, a tribe which issued coinage and had access to imported goods from Gaul and the Mediterranean. The map of the area shows only villas which can be confidently identified, usually as a result of partial or full excavation. Records of Roman masonry buildings where nothing is known of their overall plan have been excluded; it is very likely that many of these records represent villas, and thus the distribution shown on the map is a very conservative picture of the density of villas.

The second area (North-East England) comprises much of County Durham and parts of North Yorkshire, West Yorkshire and Humberside. To the west the area is bounded by the Pennines and there are high hills and moorland to the east in the area of the North Yorkshire Moors and the Cleveland Hills. Settlement in the later Iron Age and the Roman period was concentrated in the lower-lying areas. The Parisi occupied much of what is now Humberside, with the remainder of the zone belonging to the Brigantes. The latter were probably a confederation of smaller groups, some of which, in the southern part of their territory, had access to the same range of imports as the Dobunni.

Indication of Pre-Roman People

Rural settlement in the pre-Roman Iron Age across most of southern Britain was characterised by small groups of round-houses associated with field-systems. It is likely that these groups of houses represent kin-groups or extended families, possibly with servile dependents, rather than hamlets or small villages consisting of a number of independent households. In many areas the round-houses were enclosed by a ditch and sometimes also by banks, although open settlements were common. There is some differentiation in the size of the houses. On many sites a large round-house set in its own enclosure was surrounded by smaller round-houses. The larger house might have been occupied by a chieftain and the smaller houses by junior members of the kin-group. Alternatively, the larger house might have been a communal space which was also used for ritual or ceremonial purposes. The houses were usually constructed of timber and clay, with wattle and daub walls and thatched roofs. In upland areas the walls were often of rough stone. Construction in the Roman manner with mortared stone and ceramic building materials is completely unknown in Iron Age Britain.

The economic basis of these settlements would have been agriculture. Craft activities such as pottery-making and metal-working, even if they contributed to extensive trade or exchange networks, would have been subordinate to farming at little more than the level of basic subsistence.

Larger settlements are rarely encountered. The oppida found in southern and eastern England have been described elsewhere (see The Emergence of Civitas Capitals), and some hill-forts were occupied in the later Iron Age. Settlements extensive enough to be classified as villages are almost unknown.

Within the broad similarities of rural settlement in the late Iron Age there are significant regional variations. In Western England there are many so-called curvilinear farmsteads, both in the Thames valley and in the Cotswolds. They are characterised by enclosures with curving sides, some of which were fields and paddocks; others were reserved for buildings (Reece 1984, 182-3). There are some differences in this zone. For example, unenclosed settlements are commoner in the Thames valley, while enclosed settlements are much more frequently seen on the higher ground. It is possible that in the Thames valley there was large-scale farming of stock animals, while the enclosed settlements represent small farmsteads involved in mixed farming (Henig and Booth 2000, 13-15).

In North-East England as in the south the great revolution of the last thirty years has been the discovery, based on pollen analysis, but also on the regularity with which cultivated landscapes are found beneath Roman period excavations, that the landscape was extensively cleared and being farmed in the pre-Roman Iron Age. In East Yorkshire the Romans found the landscape already enclosed by dykes, defining areas containing two or more nucleated arable farms from which drove roads led through the fields and ‘ladder-shaped’ settlements to areas of pasture beyond. Here cattle or sheep ranching was clearly important. Within this landscape there are also rectilinear enclosed settlements (Ramm 1978, 1980; Dent 1983). In one recently surveyed area of east Yorkshire around Holme-on-Spalding, the waterlogged nature of terrain prevented dense settlement of the landscape. Here it has been suggested that the later-Iron Age settlement pattern consisted of enclosed settlements with parcels of arable land in clearings amongst managed woodland, with the smelting of locally available iron ore being an important economic activity (Halkon and Millett 1999).

On the North Yorkshire moors and the area just north, the southern Tees basin, groups of isolated roundhouses and roundhouse enclosures were the commonest settlement types (Inman in Price et al. 1988). Until recently it was thought that rectilinear enclosures were the standard settlement type of the pre-Roman Iron Age north of the Tees, but recently more unenclosed roundhouse settlements have been found (e.g. South Shields, Melsonby, Stanwick, East and West Brunton, Pegswood) and it is now apparent that open settlement occurred widely in the middle to later Iron Age and that the supposed predominance of enclosures has been due to their potential for discovery by air photography. East and West Brunton, on the coastal plain just north of the Wall, show that even when enclosed later Iron Age sites could be complex and form multiple groupings, unlike the small rectangular enclosures once thought typical.

Apart from the oppidum at Stanwick, and the prominent hillfort at Eston Nab, it is difficult (as in the south) to discern any hierarchy of settlement types in this region. The lack of hillforts in the area of the Parisi has been taken to imply a stable polity (Dent 1983) but this is true also of the Tees basin and the Durham/Northumberland coastal plain, where the economic basis of life was a sophisticated regime of mixed farming rather than the primarily pastoral one argued for half a century ago.

Continuation of native settlement types

One result of the Roman conquest was the abandonment of the oppida which were not taken over and developed as Roman towns. The immediate effects of the Roman conquest on rural settlements are much less evident. The dating of their development can only be set within a very broad framework, to the extent that at most settlements no clear division can be made the latest Iron Age and earliest Roman phases. Nevertheless, it is clear that even when they were close to major centres of Roman activity, change came very slowly.

Some native-style settlements appear to have been established after the conquest. One example from Western England is Hucclecote, only 5km from the legionary fortress and later colonia at Gloucester, where at least three round houses were established in a series of ditched enclosures. There was no definite evidence of Iron Age occupation, and it is possible that the farmstead developed immediately after the conquest ‘as part of the process of agricultural intensification concerned with military supply, be it officially driven or locally inspired’ (Holbrook in Thomas et al. 2003, 63). Lechlade was another such settlement, which was eventually succeeded by a villa.

In the area of the Thames valley, in the eastern part of the study area in Western England, many of the early settlement sites, whether established in the Iron Age or at the beginning of the Roman period, seem to have come to an end in the early second century. New sites were established in the course of the second century: some replaced abandoned sites nearby, but others had no obvious predecessors in their vicinity (Henig and Booth 2000, 106-7). This disruption seems to have taken place over a fairly short period, which suggests the operation of factors common to all these sites. Whether the disruption was a result of official policy or economic change is uncertain.

South of the Tees there are many examples of these varied settlement forms continuing unchanged for long after the Roman conquest, especially in East Yorkshire and the Wolds (Ramm 1978, 1980; Dent 1983). The marshy Holme-on-Spalding area was slow to be integrated into the emerging Roman economy. Settlement there remained in the native tradition, although there was a growing tendency towards multiple enclosures (which may have begun in the pre-Roman period). Little imported material reached these remote sites in our period (Halkon and Millett 1999). There are a number of enclosed sites in the native tradition in the Tees basin which have pottery, including imports, running down to the late-Roman period. Potto, in the South Tees basin, went from unenclosed roundhouse settlement to enclosed settlement over in the first two centuries, finally being replaced by an open field system. The pottery at Potto is mostly locally made but there is some Roman material (Inman in Price et al. 1988). Similar enclosed sites with predominantly local pottery, probably dating to the first and second centuries, are known at Hutton Rudby and Stokesley. An extensive Iron Age settlement and trading centre at Catcote, near the mouth of the Tees, evidently continued in occupation throughout the Roman period, when the site did acquire rectangular buildings (Long 1988; Vyner and Daniels 1989). But very little is known of its plan. With most of the of indigenous settlements recognised in the lowlands north of the Tees unexcavated and unsurveyed, it is more difficult to know how far they continued in use into the Roman period. Thorpe Thewles (Co Durham) continued in occupation, receiving some Roman pottery – but only until the early-second century or so, when it was suddenly abandoned and succeeded by fields, as had also happened at Potto (Heslop 1987). At East and West Brunton, just north of the Wall, there was little or no Roman material, suggesting they were abandoned before or early in the Roman period.

The slowness of the Roman occupation to have an impact on native settlement is graphically illustrated by two sites. At Lingcroft Farm, only 5km from the legionary fortress at York, excavations have shown that Roman pottery only began to arrive at the site half a century after the foundation of the fortress (Jones in Price et al. 1988). An Iron Age enclosure only 900m south of the important military centre at Catterick still possessed a roundhouse in the mid-second century, as associated Roman pottery shows (Wilson 1988).

Continuation in native tradition (under Roman rule)

Perhaps the most remarkable survival, and indeed development, of the native tradition of house construction is the occurrence of round houses at a large number of villa sites. Roman masonry-building techniques were used rather than the timber construction typical of the Iron Age, but the plans, orientation and interior arrangements of the Roman examples are very similar to those of their pre-Roman antecedents. The walls were usually of unmortared stone blocks, and the floors of earth, clay, stone paving slabs or brick-mortar (opus signinum). The roofs of some of the larger examples were supported at their centres by four posts arranged in a square, as at Shakenoak. None has produced any clear signs of industrial activity, but hearths and areas of burning are sometimes encountered.

Not all circular buildings on villa sites are Roman versions of the Iron Age round house. The shrine at Lullingstone took the form of a circular building, but its function was obvious because of its tessellated floor and a possible podium for a cult statue. However, most of the evidence from circular buildings indicates that they were used for domestic purposes.

The place of these buildings in the sequence of occupation at villa sites is revealing. Some can be shown to have been contemporary with conventional villa buildings. At Shakenoak the circular building was erected after the demolition of the villa building, and at Holme House, Manfield, the circular building overlay rectangular buildings, although their relationships to the main villa buildings is uncertain. Only three examples are definitely known to have preceded the construction of a main villa building. There are thus no grounds for regarding these buildings as a transitional stage between round houses of Iron Age type or construction and rectangular Romanised buildings of masonry construction. Rather, it would seem, they co-existed with Romanised buildings. In cases where circular buildings were contemporary with the main villa buildings, they probably housed dependants of the villa owner, although some were perhaps used as byres or workshops. If an explanation is to be sought for the appearance of this Iron Age building type on Romanised sites, it may lie in unbroken traditions of communal living to which housing of Roman style, with separate rooms for work, cooking and sleeping, was alien.

The notion that the round houses accommodated groups with a different way of life to that of the villa owners finds some support from the extraordinary occurrence of very large numbers of these houses, arranged in neat rows separated by streets, in the Roman fort of Vindolanda at the very beginning of the early third century. It has been suggested that these houses were to accommodate forced levies from the civitates, brought up to Hadrian’s Wall when it was being renovated at about the time of Septimius Severus’s expedition to Britain (Bidwell 1985, 28-31). In effect, the curial classes were required to supply labour for government construction and were obliged to send their agricultural work-force to the northern frontier zone. More recently, it has been suggested that the occupants of the round houses on villa sites and at Vindolanda might have been slaves (Webster 2005, 170-77).

Further development to a Roman villa

Table 1, showing the development of villas in the Western England study area, including those built after the early third century but with earlier phases of settlement on their sites.

Almost all the villas in Britain are concentrated in an area south and east of a line between the River Severn and the Wash, with a dense scatter extending up the east side of Britain into Yorkshire. A much thinner scatter extends further northwards but ends at Old Durham, 28 km south of Hadrian’s Wall. Villas are entirely absent in central and northern Wales, most of north-west England and west and north Devon, although there is a single example in Cornwall, almost at the toe of south-west England.

In south-east England villas appeared at an early date. The Neronian predecessor to the Flavian palace at Fishbourne can be compared to a series of other villas on or near the south-east coast of England – Angmering, Southwick and Pulborough - which appear to be of equally early date (Cunliffe 1973, 76-9). Southwick was a courtyard villa of Mediterranean type, as was Fishbourne, and together they represent marked exceptions to the general trends of villa development in Britain.

In recent decades ‘proto-villas’ have been found on a number of sites. There are two early examples in the vicinity of Verulamium. At Gorhambury an aisled barn or hall was set within a ditched enclosure together with a rectangular house. A date in the early first century (that is, in the late Iron Age) has been proposed for these timber buildings, but this has been disputed on the grounds that rectangular buildings of a substantial size are not known in the Iron Age (Neal 1990, 23-34; cf. Smith 1997, 226). Further timber buildings were constructed in the later first century and were replaced by a masonry villa in c. AD 100. Slight remains of a timber house and more extensive remains of a detached masonry baths-block were found at Gadebridge Park. They dated to the early Flavian period. It was not until the Antonine period that the timber house was replaced by a winged-corridor villa, which in the late second or early third century was enlarged with additional wings to create a courtyard in front of the main villa (Neal 1974). In Western England a particularly clear example of a ‘proto-villa’ has been excavated at Barton Court Farm near Abingdon (Miles 1986). An Iron Age enclosure was overlain by another enclosure of the early Roman period which was divided into two parts. One part of the enclosure contained a large timber building with plastered walls and two structures, each represented by six posts, which might have been granaries.

The majority of villa sites have not been extensively investigated and their origins are obscure. In the study area in Western England, in addition too Barton Court Farm, three other villas, Frocester Court, Great Bedwyn and Barnsley Park, are known to have been preceded by timber buildings. More extensive excavations might reveal earlier timber buildings at other villa sites, but the pattern of development seen at Barton Court Farm would not have occurred at all villas. Marshfield is known to have been built on the site of an earlier Roman shrine. At Great Witcombe there was probably Iron Age occupation on the site but it was then abandoned until the villa was built in the early third century. At some villas the earliest masonry buildings that have been excavated were the baths; they might well have been erected at the same time as timber buildings, the remains of which were concealed by later masonry phases of the villas.

Whatever their origins, the range of dates at which masonry villas were built in Western England is similar to that in the rest of the main villa zone of southern Britain. The mid-first century villas near the south coast mentioned above are exceptional. More generally villas begin to appear in the late first and early second centuries, as at Shakenoak, Ditchley and North Leigh. The majority of villas date from the second and early third centuries. In their earliest phases they were quite modest establishments; the largest examples were at Chedworth and North Leigh where there a series of detached buildings. Eventually some of the villas in Western England were transformed into enormous residences with sumptuous decoration, but this was very much a phenomenon of the fourth century.

Table 2, development of villa sites in north-east England (with listing of stone coffins). For sites not in the bibliography, see Gazetteer by Scott (1993).

In North-East England it has been argued that after the Roman conquest in East Yorkshire a series of small farms within rectilinear ditched enclosures (Potter Brompton Wold, Cat Babbleton, Carnsdale, Langton, Langton Whin Fields, Settrington, Crossgates, Rudston) emerged, associated in one area with land division into strips. These sites have been associated with settlement by retired Roman soldiers (Ramm 1978). But there is insufficient evidence for this idea – some (Langton, Rudston) developed into villas, but others (Potter Brompton Wold) have produced mainly late-Roman finds. There is little evidence for the highly Romanised finds one would expect from the Roman military (a single sword from Crossgates, a lead cremation urn from Langton), and at a site like Settrington, where there is a rectangular stone building, the enclosure also contains round houses. At Rillington, Driffield, Garton Slack, Blealands Nook the enclosures produced only pottery in the local Iron Age tradition. It is uncertain whether the many small rectilinear enclosures of second century date north of Hadrian’s Wall represent some form of deliberate settlement of the Roman period or whether they are of pre-Roman Iron Age origin. Such sites are notoriously poor in Roman objects. Somewhat reminiscent of these are some sites of Roman date south-east of Malton, whose form is in the native Iron Age tradition, and there are also post-conquest sites of traditional form (roundhouse in enclosure) with very little Roman material culture (no coins at all) at Garton Slack and Burton Agnes. At Staxton, just below the northern brow of the Wolds, an enclosed roundhouse settlement attracted samian pottery, brooches, etc. in the later first and earlier second centuries (references to all these sites in Ramm 1978).

Types of Roman villa

Most of the villas in Britain seem to have developed from one of two simple types of building plan:

The hall-type, as defined by Smith (1997, 26), which is widespread in the north-western provinces of the Roman Empire. The building provided a large unpartitioned space, and excavation has shown that the interior was divided up into areas for different activities, as in round houses. Variations within the hall-type consists of the broad hall, with a width often in excess of 9m and a plan which is rectangular or almost square, and the narrow hall which is much more elongated (Smith 1997, 23-45). Broad halls often had a nave and two aisles – the so-called aisled house. While the general type is widespread in Germany, the Netherlands and northern France, a division of the interior into separate areas, one certainly for domestic occupation and the other apparently for the shelter of livestock, is a practice scarcely known beyond Britain.

Three examples in the Western England study area can be cited:

Shakenoak (Brodribb et al. fig. 2.4)

Clear Cupboard, Farmington (RCHM Gloucestershire Cotswolds, 56)

Wraxall (Sykes and Brown 1960-1): the plan of this villa seems to have consisted of a long hall (originally identified as a paved courtyard) with baths and other ranges of rooms on three of its sides (Smith 1997, 33, fig. 30). A later third-century date was suggested for its construction, but an earlier date is possible.

In addition, there is a well-published example from Bancroft, Bucks. (Williams and Zeepvat 1994).

The row-type has been defined as a building consisting of ‘a row of two to five rooms … which are interspersed with smaller squarish or corridor-like rooms’; typically, one room in the range is always much larger than the others and might have served as the hall or main living-room (Smith 1997, 46).

Examples in Western England are:

North Leigh: dated to the early second century (Ellis 1999; Wilson 2004).

Shakenoak (Brodribb et al. 2005)

Both types were frequently altered or enlarged. A porticus and pavilions were frequently built onto the fronts of the row-type, producing a winged-corridor villa. The elaborate H-shaped plan of the villa at Great Witcombe has been seen as a development of this type of plan (Ellis 1998, 124). The hall-type usually came to be divided into a series of rooms, sometimes including a baths-suite, by walling across the aisles and at the ends of the nave.

The earlier villas commonly consisted of a series of detached buildings, as for example at North Leigh where the main residence is flanked on one side by a block of baths and on the other by a hall, possibly a work-place rather than additional living space. At Chedworth there were four detached buildings; they consisted of a baths-block and three row-type buildings, one of which would have been the main villa residence. Beyond Western England, in Buckinghamshire,

the second-century site at Stanton Low consisted of at least six large buildings including barns and possible warehouses; the site has been described as an ‘extended villa’ or ‘some unrecognised type of wealthy settlement’ (Woodfield and Johnson 1989, 260-1). The villa at Beddington in Surrey, built in the late second century, consisted of seven separate buildings, including two large barns of timber construction (Howell 2005).

In North-East England, simple rectilinear stone buildings or ‘cottage houses’ occur in enclosures at Holme House, at Settrington, near Malton, and in the second-century development at Langton. It is clear that a good many simple rectilinear houses of second-century date could await discovery or have gone undetected on sites with later villa buildings. Where plans of more developed villa buildings are known in this region they are of the simple development from the row-type known as the wing-corridor villa, almost a standard plan. At both Ingleby Barwick and Dalton-on-Tees we see the ubiquitous pairing together of a wing-corridor villa and aisled hall-type building, the latter in its typical position in front and to one side of the main house. Neither site is at present closely dated, and the other known examples of this arrangement in the north-east date to after AD 200.

Endowment of villa

Evidence for the decoration of villas up to the early third century, although copious, is very fragmentary. Most villas were rebuilt and often greatly enlarged in the later third and fourth centuries, and the mosaics which survive intact and the reconstructable schemes of decoration are of that later period. There is no doubt that many villas had mosaics and wall-paintings by the early third century, although on a much more modest scale than is typical of the late villas of Western England which bear comparison with the most luxurious examples in the Western Empire.

Although the villas of North-East England were equipped with the usual amenities in the third and fourth centuries, there is at present much less evidence of this for the period up to AD 200. Holme House certainly had a bath suite and mosaics in the second century. The complex of buildings at Well perhaps had baths and tessellated pavements as early as the second century, although it is not certain that this is a villa. The second-century house at Langton was, however, extremely basic, with no evidence for hypocausts or mosaics. At Ingleby Barwick the villa is accompanied by a separate, small hypocausted building, presumably intended for some kind of baths, but of a most basic and unorthodox type, as if a Roman bath was being interpreted by builders unfamiliar with the type. The main houses with mosaics and hypocausts at Langton, Rudston and Dalton Parlours were all built after the end of our period.

Diffusion of Roman villas

The zone selected for study in Western England has the densest distribution of villas in Britain. The influenced of geology on the pattern of settlement has already been considered, but there were other factors:

Relationship to major towns. There is a ‘generally high density of settlement’ in the neighbourhood of the colonia at Gloucester, with many farmsteads that originated in the Iron Age or the early Roman period and continued in occupation down to the fourth century AD (Hurst 1999, 127-30). Some of these settlements had developed into villas by the early third century, as at Frocester and Hucclecote, but many continued as simple farms. The villas were of modest status, and many of them were presumably owned by citizens of the colonia. In the vicinity of Cirencester there is again a high density of settlement with many villas, although only in a few instances, such as at Barnsley Park, Chedworth and Lechlade, is much known about their early development.

Relationship to smaller towns. Small towns are common in this area. Their main function was probably to serve as market centres for the surrounding villas, so their frequency is a reflection of the density of villa settlement. Smaller settlements are equally common, and might have actually formed part of villa estates (Reece 1984, 185).

The northern and western edges of the study area show much a thinner distribution of villas. To the west lay the Forest of Dean and beyond it South Wales, a zone which contains some villas and two civitas capitals, at Caerwent and Carmarthen, but which also had a strong military presence at the legionary fortress of Caerleon. The northern part of the area represents the limit of the main villa zone in southern Britain.

It has been suggested that some parts of southern Britain were imperial estates, but their existence, although likely, has not yet been proved. An inscription from Combe Down, near Bath, refers to the restoration of a principia by an imperial freedman who was a procuratorial assistant (RIB 179). The inscription dates to the early third century, and the building it mentions was probably a procuratorial office rather than a military headquarters building. But whether this office was concerned with the administration of an imperial estate or with control of mining in the Mendips to the west is uncertain.

Until recently there seemed to be large areas of North-East England which showed no trace of the pattern of towns and villas that would be expected of a developed province of the Roman empire. Villa settlement seemed confined to certain areas of Yorkshire and largely absent north of Catterick and the Cleveland Hills. Holme House seemed a possible exception, related to the nearby native centre at Stanwick. The one candidate for a villa north of the Tees, Old Durham, was so fragmentarily known that it was considered an uncertain villa.

But now, thanks to an increased incidence of developer-funded archaeology, and some enterprising research by local societies, more Roman villas have actually started to appear in the northern part of this region. The most notable examples are at Quarry Farm, Ingleby Barwick, and Dalton-on-Tees (Chapel House Farm: Brown 1999). There is also a Romanised rural settlement, which may have developed into a villa, at Faverdale near Darlington. The importance of these sites, still low in absolute numbers, is that they restore the reputation of the ‘villa’ at Old Durham, and in combination with that site suggest that there are many more to be discovered and that the distribution could well run far north of the Tees towards Hadrian’s Wall. Small towns or marketing centres to accompany these villas must also await discovery (the recently found site at Sedgefield is a probable example of such: Carne and Mason 2006). The Humberside and East Yorkshire region makes an interesting comparison, for here where more villas have long been known, urban centres, such as Shiptonthorpe, have also only been recognised in recent times.

Formerly the villas in this region were seen as relating to some known major urban centres: Brough, Malton, the area between Aldborough and Catterick, a lost urban site at Bridlington (probably Gabrantovicum Sinus, destroyed by coastal erosion). There is a curiously low number of known villas around York. In view of the more recent discoveries, these groupings may represent merely the most conspicuous and prosperous examples, and a much wider distribution of villas that have not left such substantial archaeological traces must be expected, along with the marketing centres that will have accompanied them. There will be exceptions: throughout Britain villas favour the most attractive land for mixed farming, and the north-east was no different. For example, the marshy Holme-on-Spalding area failed to adopt the model; its sites have few imports and in contrast to the nearby areas around Brough, Hayton and Shiptonthorpe, all along a Roman road, there are few known Roman coin finds and no villas, despite intensive recent fieldwork. A pottery industry developed here, but only after our period. In general, however, although settlements in the native tradition clearly persisted in certain areas, it now appears unlikely that there was any fertile lowland part of the north-east region south of Hadrian’s Wall where the idea of the villa house and estate on the Roman provincial model was not penetrating by AD 200.

Given the extent of the villa distribution now emerging in the north-east, the complete absence of villas west of the Pennines, in Cumbria and Lancashire, becomes all the more striking. There is still not a single example known on the western side of northern Britain, despite the fact that Carlisle had become a flourishing town at the centre of the civitas Carvetiorum by the end of our period. It is unclear whether this utter absence is due to unevenness in the opportunities for archaeological discovery – after all, this was the situation that until recently prevailed for the Tees basin – or whether it signifies some profound difference in the pattern of land exploitation in the Roman period.

In the north-east there is, of course, a stark difference between the areas south and north of Hadrian’s Wall. All Roman period sites to the north of the frontier remain in the native tradition. This does not mean that the same communities continued to live there uninterruptedly. In some areas north of the Wall settlements may have been newly founded in the Roman period and others abandoned; at present very few of these sites are closely datable. But it is clear that there was never the development of villas that occurred within the province.

Economical basis of the villa

Agriculture

Surveys and excavations across Britain have begun to reveal how villas realted to their agricultural landscapes. A notable example is at Stanwick (Northants.)

In Western England much of the agricultural regime of the Iron Age survived into the Roman period, when it became the main support of the villa economies. However, there were a number of important changes which resulted in an intensification of agriculture. Improved, larger breeds, particularly of cattle, were introduced; new crops were cultivated, and types of wheat, for example bread wheat, which were scarcely known in the Iron Age became much commoner. In the Upper Thames Valley, there was ‘a new emphasis on the development of hay meadows’ (Henig and Booth 154-9).

New technologies were also introduced. The Roman period saw the replacement of the ard by the plough, with an iron coulter and a mould-board. The speed and efficiency of reaping was much improved by the use of scythes with long blades. Corn-driers (ie raised floors which were heated from underneath by flues) could be used to save harvests affected by wet weather that might otherwise have been discarded, to prepared grain for storage, and to roast germinated grains for malt (van der Veen 1989). A few examples are known in the earlier Roman period (but none from the Iron Age), but they are particularly common in the third and fourth centuries. Examples at Foxholes Farm (Herts.), a rural settlement occupied mainly during the second to fourth centuries, including a particularly elaborate version. Another great improvement in the efficiency of crop processing was the introduction of the watermill, using undershot vertical wheels (ie with paddles which are turned as the water pushes against the bottom of the wheel). The only earlier Roman example which has been excavated in Britain is at Haltwhistle Burn, near the fort of Greatchesters on Hadrian’s Wall, but finds of mill-channels and large milling-stones suggest that they were widespread in the civilian zone (Spain 1984).

The size of villa estates is an important factor in understanding their economies. In the absence of surviving boundary-markers, the only way of calculating their extent is by looking at the distribution of villas in relationship to Anglo-Saxon estates of the ninth century and later. In some parts of England, particularly in Western England, it seems likely that the limits of villa estates are reflected by later land allotments. Reece (1984, p. 185) estimates villa estates around Cirencester of c. 2,000ha.

Industry

Although agriculture was the staple of the rural economy, in some areas there were other activities, such as pottery manufacture, quarrying and mining, which might have contributed significantly to the income of villa estates. An example from Western England is iron-working. On the western side of the Bristol Channel, in the area of the Forest of Dean, there was exploitation of the local iron ore deposits from the earlier second century onwards. The Rivers Severn and Wye provided good connections with the western sea-ways, and it is likely the iron industry was serving the large military markets in Wales and the northern frontier zone. The villas, small industrial sites and larger settlements all seem to have been involved in this industry. Most of the evidence so far recovered comes from the later Roman period, but it is likely that iron working played a part in the economy of villas such as Boughspring, Woolaston and Clearwell from their very beginning (Fulford and Allen 1992, 205-6).

The economy of villa landscapes in North-East England

There is some archaeological evidence in the region for the kinds of changes in landholding and farming practice that may have led to the development of villas. Enclosed sites in the native tradition at Potto and Thorpe Thewles were abandoned and overlain by field systems in the course of the second century. The newly discovered small town of Sedgefield emerges as a Roman settlement in the same period as Thorpe Thewles, only 7km away, is abandoned, or rather overlain by fields. Naturally this could be coincidence, but it signals the possibility that in some areas drastic new patterns of landholding could have displaced or replaced earlier settlement patterns, as a market economy geared to the Roman roads and urban centres developed. Some archaeologists have theorised an increase in the practice of arable agriculture, at the expense of pastoralism, with the development of the Roman provincial economy. Certainly the form of settlement in the native tradition, especially in East Yorkshire, suggests that cattle rearing was very important in the pre-Roman Iron Age, but few would now doubt that cereal cultivation was also important in the pre-Roman period (van der Veen 1992). But the production of grain for the market rather than for subsistence may lie behind the development of some villa estates and villa buildings. The evolution of the site at Langton suggests this kind of economic development. Here we see the amalgamation in the second century of a group of Iron Age enclosed settlements into a more extensive, unified field system. This implies a shift from dispersed subsistence to a unit of production organised for marketing produce, and the development of a villa at the site demonstrates that the reorganised agricultural system was capable of generating and selling a surplus.

A further basis for villa development may have lain in the economic prosperity of the individual urban centres around which villas grouped: the military-fuelled markets of Catterick, Malton, and Piercebridge, and the civitas capitals at Aldborough and Brough. The lost town near Bridlington mentioned above, which was probably a coastal trading centre now destroyed by erosion, must have provided some of the wealth behind the group of villas – Grindale, Rudston, Harpham, Tuft Hill, and no doubt others, that clustered close by.

Sanctuaries at villas

In Western England, Marshfield provides an example of a villa superseding a shrine which was established in the Iron Age and continued in use until c. 250. No other example of a transition from religious to domestic use is known on a villa-site in Britain. In the later Roman period there is much evidence for the religious use of some parts of villas or for separate religious buildings. In the earlier period such remains are scarcer. Examples are as follows:

Cellars

Cellars occur at some town-houses and at a number of villas in Britain. Some were certainly cult rooms, mainly associated with ‘an imported fertility cult, influenced by both oriental and western elements, but in which Isaic ideas and practices may have dominated’ (Perring 1989, 294). There are cellars in the early villas at Shakenoak and The Ditches, Woodmancote, in Western England, but there are no clear signs that they were used for ritual purposes. The best early example is at Lullingstone in Kent, a villa constructed in the later first century AD which included a cellar, the so-called Deep Room, with entrances both from the interior of the villa and from outside (Meates 1979). It seems to have been built originally for the storage of agricultural produce. Towards the end of the second century the cellar was converted into a nymphaeum: a small square well was sunk near the middle of the room and the walls were decorated with paintings of mythological subjects, including water nymphs.

Pools

At a number of villas large pools were constructed near the main building. The best known examples are at Gadebridge (Herts.) and Bancroft (Bucks.) but are of late Roman date. An earlier example, probably dating to the later second century, is at Well in North-Eastern England. As at Well, these pools have sometimes been regarded as open-air baths, but they seem too large to be associated with the modest baths-buildings found at villas. Some form of ritual purpose, possibly associated with water deities, seems likely. From the filling of the pool at Well there came a panel with S-shaped motifs, an architectural element which was used to crest the pediments of temples.

Shrines

At Lullingstone in Kent a circular shrine with an internal diameter of 4.72m was erected a short distance north of the villa in c. AD 100 (Meates 1979). Part of its floor was tessellated and opposite the entrance there seems to have been a rectangular podium built of wattle and daub which was presumably for a cult statue. Its sides were plastered and painted in white and red. In Western England, shrines are only known from the later phases of villas, as at Chedworth, but it is possible that they replaced earlier, smaller shrines.

Temple-mausoleum

At Bancroft (Bucks.), about 350m north of the villa site, a temple-mausoleum was erected in the second half of the second century (Wiiliams and Zeepvat 1994). A burial vault 4.0m square, which had probably contained one or more stone sarcophagi, formed the basement of a square cella which seemed to have served as a shrine. It was surrounded on all four sides by an enclosed corridor, and in plan it resembles a Romano-Celtic temple. It was built on the site of irrregularly-shaped enclosure of Iron Age origin, just outside of which there had been a small cremation cemetery. The temple-mausoleum has a later Roman parallel at Lullingstone (Kent), a building of similar plan, also with a burial vault, which was built near the circular shrine mentioned above.

Indications of villa owners

Inscriptions are hardly ever found at villa-sites, and the few that are known, particularly religious dedications from Western England, seem to date from the third or early fourth centuries. The names of villa owners can sometimes be deduced from place names. Villa Faustini occurs in the Antonine Itinerary; it was somewhere in East Anglia and obviously refers to an estate owned by Faustinus. In the Ravenna Cosmography there is an example which seems to have been in Western England – Anicetis, ‘probably the designation of an estate, based on the Latin personal name Anicetus’. This might have been the Q. Pompeius Anicetus (RIB 148), or a relative of his, who dedicated an altar to Sulis at Bath. The name suggests that he was a Greek or eastern freedman.

The generally accepted model is that villas were the country residences of members of the curial class. They maintained town-houses so that they could take part in the affairs of their civitas or colonia, but also had villas from where they could supervise the main source of their wealth, which was agriculture. Aggrandizement of their town and country properties went hand in hand as their wealth increased. Whether this general picture can be applied to all villas in all the parts of Britain where they occur is uncertain. Writing in 1965 about the most northerly suspected villa in the Roman empire, Old Durham, Salway admirably distilled the problem: ‘The most interesting question of all remains unanswered, for the evidence is insufficient to prove whether the thoroughly Roman aspect of the site was the result of Romanisation of the local native population or whether the owners of this establishment were, in origin at least, Romans from elsewhere in Britain or abroad’ (Salway 1967, 155). Forty years later, even though we have growing evidence that Old Durham does not stand alone, there remains a basic uncertainty about whether these hyperborean villas were the homes of a native elite, or wholly or in part the product of an immigrant population. The question of the origin of the owners also remains open in the case of the much more substantial and prosperous villa settlement of Yorkshire.

We saw earlier that there is a clear relationship between several villas and a pre-Roman Iron Age settlement of agricultural system on the same site. However, in no case is it possible to demonstrate that the descendents of the pre-Roman community were responsible for developing the site into a villa, or even to show that the development of the villa was a gradual adaptation of a pre-Roman site by the same community. Langton, with the amalgamation of early enclosures and the eventual development of the site into a villa, perhaps best fits the model of indigenous farmers becoming villa owners. Even if this could be demonstrated, it would then be difficult to prove continuity of ownership within the Roman period. The idea that certain rectilinear enclosures in East Yorkshire, such as those that preceded the villa at Langton, were the settlements of military veterans, shortly after the conquest of the region, has already been mentioned, and it has been noted that there is little real evidence for this. Publication of the pottery and finds from a recently excavated villa site like Ingleby Barwick, will, it is hoped, allow comparison with assemblages from military sites in the area, so that some assessment can be made as to whether the villa drew on the same sources of supply, something that might be expected if it had really been established by settlers with close ties to the army.

At three places in rich farming land in Ryedale, north-west of Malton, stone sarcophagi have been discovered. Although no Roman buildings are as yet known, these places have been suggested as the sites of villas, on the basis of the coffins are more likely to indicate the presence of a villa than any other kind of site. Two are inscribed. That from Hood Grange (where nearby crop-marks, possibly of a villa, have been observed) was inscribed by Aurelius Serenus to his wife. That from East Ness says that ‘Valerius Vindicianus had this coffin made for his wife, Titia Pinta aged 38, and his sons, Valerius Adiutor, aged 20, and Varialus, aged 15’. These inscriptions probably date at the earliest to the end of our period. The use of highly Latinate stone epigraphy, generally unusual at villas throughout the province and normally most commonly associated with major urban and military centres, has been taken to suggest that these villas were built by settlers, in particular retired soldiers, but of course there can be no proof of this. Nevertheless the inscriptions are of rare importance in giving us, in all probability, the names of villa owning families. These unusual finds may indicate a limited group of settlers, not necessarily typical for a region where the enduring pattern of native settlement, and the slow and late development of villas, can equally be taken to suggest a social evolution from the pre-Roman Iron Age.

TWM

Bibliography

Allen, T. G., Darvill, T. C., Green, L. S., and Jones, M. U. (1993) Excavations at Roughground Farm, Lechlade, Gloucestershire, Thames Valley Landscapes: The Cotswold Water Park, Volume 1: Oxford Archaeological Unit.

Bidwell, P. T. (1985) The Roman Fort of Vindolanda, Historic Buildings and Monuments Commission for England Archaeological Report No. 1.

Blockley, K. (1985) Marshfield, Ironmonger’s Piece, Excavations 1982-3: An Iron Age and Romano-British Settlement in the South Cotswolds, British Archaeological Reports British Series 141.

Branigan, K. (ed.) (1980) Rome and the Brigantes: The Impact of Rome on Northern England, Sheffield.

Branigan, K. (1980) ‘Villas in the North: Change in the rural landscape?’, in Branigan (ed.) 1980, 18-27.

Brodribb, A. C. C., Hands, A. R., and Walker, D. R. (2005) The Roman Villa at Shakenoak Farm, Oxfordshire, British Archaeological Reports 395.

Brown, J. (1999) ’Chapel,House Farm, Dalton-on-Tees’, Teeside Archaeological Society Bulletin, 4, 6-16.

Carne, P., and Mason, D. (2006) ‘The Sedgefield Archaeology Project’, in Archaeology County Durham 2006, County Durham Books.

Collingwood R. G., and Wright, R. P. (1965) The Roman Inscriptions of Britain, Volume 1:Iinscriptions on Stone, Oxford.

Corder, P., and Kirk, J. L. (1932) A RomanVilla at Langton, near Malton, Leeds.

Cunliffe, B. (1973) The Regni, Duckworth: London.

Dent, J. S. (1983) ‘The impact of Roman rule on native society in the territory of the Parisi’, Britannia, 14, 35-44.

Ellis, P. (1999) ‘North Leigh Roman villa, Oxfordshire: a report on excavation and recording in the 1970s’, Britannia, 30, 199-245.

Fincham, G. (2002) Landscapes of Imperialism. Roman and Native in the East Anglian Fenland, British Archaeological Reports British Series 338.

Fulford, M. G., and Allen, J. R.L. (1992) ‘Iron-making at the Chesters villa, Woolaston, Gloucestershire: survey and excavation’, Britannia, 23, 159-215.

Gilyard-Beer, R. (1951) The Romano-British Baths at Well, Yorkshire Roman Antiquities Committee Research Report No. 1.

Goodburn, R., 1979 The Roman Villa, Chedworth, National Trust: London.

Gosden, C., and Lock, G. (2003) ‘Becoming Roman on the Berkshire Downs: the evidence from Alfred’s Castle’, Britannia, 34, 65-80.

Halkon, P., and Millett, M. (eds.) (1999) Rural settlement and industry: studies in the IronAge and Roman archaeology of lowland East Yorkshire, Leeds.

Hartley, B., and Fitts, L. P. (1988) The Brigantes, Duckworth: London

Haselgrove, C. (1984) ‘The later pre-Roman Iron Age between the Humber and the Tyne’, in Wilson, Jones and Evans (eds.) 1984, 9-25.

Haselgrove, C. (2002) ‘The Later Bronze Age and the Iron Age in the Lowlands’, in Brooks, Daniels and Harding (eds) 2002, Past Present and Future: the Archaeology of Northern England (Proceedings of a Conference held in Durham in 1996), Durham, 49-69.

Hayes, R. H. (1988) North-East Yorkshire Studies: Archaeological Papers, Leeds.

Hayes, R. H. (1988) ‘Small square or rectilinear enclosures in North-East Yorkshire’, in Hayes 1988, 51-56.

Henig, M., and Booth, P. (2000) Roman Oxfordshire, Sutton Publishing: Stroud.

Heslop, D. H. (1987) The excavation of an Iron Age settlement at Thorpe Thewles, Cleveland, 1980-82, London.

Hingley, R. (1989) Rural Settlement in Roman Britain, Seaby: London.

Hodder, I., and Millett, M. (1980) ‘Romano-British villas and towns: a systematic analysis’, World Archaeology, 12, 69-76.

Hodgson, N., Stobbs, G. C., and Van der Veen, M. (2001) ‘An Iron–Age Settlement and Remains of earlier prehistoric date beneath South Shields Roman Fort, Tyne and Wear’, The Archaeological Journal, 158, 62-160.

Holbrook, N. (2004) ‘Turkdean Roman villa, Gloucestershire: archaeological investigations 1997-1998’, Britannia, 35, 39-76.

Hostetter, E., and Howe, T. N. (1997) The Romano-British Villa at Castle Copse, Great Bedwyn, Indiana University Press: Bloomington and Indianapolis.

Howell, I. (ed., 2005) Prehistoric Landscape to Roman Villa: Excavations at Beddington, Surrey, 1981-7, Museum of London Archaeology Service Monograph 26.

Hurst, H. (1999) ‘Topography and identity in Glevum colonia’, in (ed.) Hurst, H., The Coloniae of Roman Britain: New Studies and a Review, Journal of Roman Archaeology Supplementary Series 36.

Inman, R. (1988) ‘Romano-British settlement in the south Tees basin’, in Price, Wilson, Briggs and Hardman (eds) 1988, 219-234.

Johnston, D. E. (1994) Roman Villas, 4^th edn, Shire Books: London.

Jones, B., and Mattingly, D., (1990) An Atlas of Roman Britain, Blackwell: Oxford.

Leach, P. (1998) Great Witcombe Roman Villa, Gloucestershire; A Report on the Excavations by Ernest Greenfield, British Archaeological Reports British Series 266.

Long, C. D. (1988) ‘The iron Age and Romano-British settlement at Catcote, Hartlepool’, Durham Archaeological Journal, 4, 13-35.

Mattingly, D. (2004) ‘Being Roman: expressing identity in a provincial setting’, Journal of Roman Archaeology, 17, 5-25.

Meates, G. W. (1979) The Roman Villa at Lullingstone, Kent, Monograph Series of the Kent Archaeological Society, No. 1.

Miles, D. (1984) ‘Romano-British settlement in the Gloucestershire Thames Valley’, in Saville 1984, 191-211.

Miles, D. (ed., 1986) Archaeology at Barton Court Farm, Abingdon, Oxon: An Investigation of Late Neolithic, Iron Age, Romano-British and Saxon Settlements, Council of British Archaeology Research Reports 50.

Neal, D. S. (1974) The Excavation of the Roman Villa in Gadebridge Park, Hemel Hempstead, 1963-8, Reports of the Research Committee of the Society of Antiquaries of London, No. 31.

Neal, D. S. (1996) Excavations on the Roman Villa at Beadlam, Yorkshire, Leeds.

Neal, D. S., Wardle, A., and Hunn, J. (1990) Excavation of the Iron Age, Roman and Medieval settlement at Gorhambury, St Albans, English Heritage Archaeological Report No. 14.

O’Neil, H. (1952) ‘Whittington Court Roman Villa, Glos.: A report of the excavations undertaken from 1948 to 1951’, Transactions of the Bristol and Gloucestershire Archaeological Society, 71, 13-87.

Partridge, C. (1989) Foxholes Farm, A Multi-Period Gravel Site, Hertfordshire Archaeological Trust.

Percival, J. (1981) The Roman Villa, Batsford: London.

Perring, D. (1989) ‘Cellars and cults in Roman Britain’, Archaeological Journal, 146, 279-301.

Price, J., Wilson, P. R., Briggs, C. S., and Hardman, S. J. (eds) (1988) Recent Research in Roman Yorkshire, British Archaeological Reports British Series 193, Oxford.

RCHM Gloucestershire Cotswolds Iron Age and Romano-British Monuments in the Gloucestershire Cotswolds, Royal Commisssion on Historical Monuments (England), 1976.

Reece, R. (1984) ‘The Cotswolds: an essay on some aspects and problems of rural settlement’, in Saville 1984, 181-90.

Radford, C. A. R. (1936) ‘The Roman villa at Radley, Oxon.’, Oxoniensia, 1, 24-69.

Rahtz, P., Hayfield, C., and Bateman, J. (1986) Two Roman Villas at Wharram Le Street, York University Archaeological Publications 2.

Ramm, H. (1978) The Parisi, London.

Ramm, H. (1980) ‘Native settlements east of the Pennines’, in Branigan 1980, 28-40.

Ramm, H. (1988) ‘Aspects of the Roman countryside in East Yorkshire’, in Price, Wilson, Briggs and Hardman 1988, 81-88.

Richmond, I. A., Romanns, T., and Wright, R. P. (1944) ‘A civilian bath-house of the Roman period at Old Durham’, Archaeologia Aeliana, 4^th series, 22, 1-21.

Rivet, A. L. F. (ed.) (1969) The Roman Villa in Britain, Routledge Kegan Paul: London.

Salway, P. (1965) The Frontier People of Roman Britain, Cambridge University Press, Cambridge.

Saville, A. (ed.) (1984) Archaeology in Gloucestershire from the Earliest Hunters to the Industrial Age, Alan Sutton: Cheltenham.

Scott, E. (1993) A Gazetteer of Roman Villas in Britain, Leicester Archaeology Monographs No 1.

Shaw-Mellor, A., and Goodchild, R. (1942) ‘The Roman villa at Atworth, Wilts.’, Wiltshire Archaeological and Natural History Magazine, 49, 46-95.

Smith, J. T. (1997) Roman Villas: A Study in Social Structure, Routledge: London.

Spain, R. J. (1984) ‘Romano-British watermills’, Archaeologia Cantiana, 100, 101-28.

Spratt, D. A. (ed.) (1982) Prehistoric and Roman Archaeology of North-East Yorkshire, Oxford.

Stead, I. M. (1980) Rudston Roman Villa, Leeds.

Sykes, C. M., and Brown, G. A. (1960-1) ‘The Wraxall villa’, Proceedings of the Somersetshire Archaeological and Natural History Society, 105, 37-51.

Thomas, A., Holbrook, N., and Bateman, C. (2003) Later Prehistoric and Romano-British Burial and Settlement at Hucclecote, Gloucestershire, Bristol and Gloucestershire Archaeological Report 2.

Van der Veen, M. (1989) ‘Charred grain assemblages from Roman-period corn driers in Britain’, Archaeological Journal, 146, 302-319.

Van der Veen, M. (1992) Crop husbandry regimes: an archaeobotanical study of farming in northern England, 1000BC-AD 500, Sheffield.

Vyner, B. E., and Daniels, R. (1989) ‘Further investigation of the Iron Age and Romano-British settlement site at Catcote, Hartlepool, Cleveland, 1987’, Durham Archaeological Journal, 5, 11-34.

Webster, J. (2005) ‘Archaeologies of slavery and servitude: bringing ‘New World’ perspectives to Roman Britain’, Journal of Roman Archaeology, 18, 161-79.

Webster, G. (1981) ‘The excavation of a Romano-British rural establishment at Barnsley Park: Part I’, Transactions of the Bristol and Gloucestershire Archaeological Society, 99, 21-78.

Williams, R. J., and Zeepvat, R. J. (1994) The Late Bronze Age and Iron Age Settlements and Roman Temple-Mausoleum. The Roman Villa, The Buckinghamshire Archaeological Society Monograph Series No. 7 (2 vols).

Wilson, D. R. (2004) ‘The North Leigh Roman villa: its plan reviewed’, Britannia, 35, 77-113.

Wilson, P. R. (1997) ‘Behind the frontier: Romans and natives in the military zone of Northern England. Case Study – North East Yorkshire’, in Groenman-van Waateringe, W. et al. (eds.) Roman Frontier Studies 1995 (Oxbow Monograph 91), Oxford, 1997, 587-93.

Wilson, P. R., Jones, R. F. J., and Evans, D. M. (eds.) (1984) Settlement and Society in the Roman North, Bradford.

Woodfield, C., and Johnson, C. (1989) ‘A Roman site at Stanton Low, on the Great Ouse, Buckinghamshire, excavated by Margaret Jones, 1957-58’, Archaeological Journal, 146, 135-278.

Wrathmell, S., and Nicholson, A. (1990) Dalton Parlours Iron Age Settlement and Roman Villa, West Yorkshire Archaeology Service.

Wright, R. P., and Gillam, J. P. (1951) ‘Second report on Roman buildings at Old Durham’, Archaeologia Aeliana, 4th series, 29, 203-12.