A fundamental principle of Roman provincial government was the differential treatment of defeated or submitted enemies, and a specific context of differentiation was land. The settlement of the land after conquest was often a protracted process, in which the army, specialist surveyors known as the agrimensores (literally ‘land measurers’) and the various branches of imperial government (provincial governors, juridical legates and procurators) all played their part. Although Britain was one of the last provinces to be added to the empire, the treatment of land there can be assumed to have followed in broad terms the established practice of centuries of Roman expansionism. The following quote from one of the agrimensores, Siculus Flaccus, summarizes the process.
Certain peoples, with pertinacity, have waged war against the Romans, others having once experienced Roman military valour, have kept the peace, others who had encountered Rome’s good faith and justice, declared their submission to the Romans and frequently took up arms against their enemies. This is why each people has received a legal settlement according to merit: it would not have been just if those who had so frequently broken the peace and had committed perjury and taken the initiative in making war, were seen to be offered the same guarantees as loyal peoples.
As a starting point, the lands of defeated or submitted peoples (dediticii) were considered the property of the people of Rome (imperial property was in some ways a sub-set of this broad category), to be disposed of as she chose. Some lands were habitually seized and re-assigned in specified allotments to Roman colonists (typically time-expired legionary veterans). A proportion of conquered territory commonly remained as state lands (ager publicus and later frequently in the form of imperial estates), which would be leased out and yield revenue. Lands were also reassigned to subject peoples in proportion to their perceived merit (ager assignati), often after a suitable interval of military control and uncertainty. Lands were attached to a civitas or other major town in two ways, directly and indirectly. Direct allocations often included a grant of land to a town, which could be rented out to provide a source of income to support the local administration conducted there. The administrative and financial territory of a town also covered lands indirectly assigned to its control. These generally included substantial areas in private ownership or leasehold, whether created by Rome through personal gift, direct sales or long-term leases. All of the lands that were not state owned or allocated to colonists appear to have carried a vectigal or tribute payment; again, the level of this may have been scaled to reflect the relative favour in which particular groups were held. Direct land sales within the broad territory previously controlled by a particular group could open up ownership to outsiders, whether retired soldiers, Romans from outside Britain (notably from Gaul, but extending up through society to include senators) or incomers from other regions of Britain. External proprietors might sub-let to native Britons or supplant them.
Looked at from another perspective, in extreme cases native Britons might find their lands definitively alienated (after an initial period where they had been permitted to remain on their traditional farms in return for paying tribute) or that they had been converted into tenants of a private estate of an absentee landlord. On the other hand, preferential treatment for certain favoured Britons is to be expected. As Siculus Flaccus noted: ‘among the defeated not all saw their lands seized. In practice, the dignity of certain individuals, the gratitude or friendship of Rome impelled the victor to concede to them their own traditional lands.’
The initial conquest of Britain thus brought about major changes in land proprietorship and, to a lesser extent, land occupancy. The establishment of colonies involved assignation of land, the army carved out legionary territories, the Roman state designated large tracts as ager publicus or imperial property (especially in areas with natural resources). It is apparent that even after a first land deduction, an indigenous community might find a second block of territory taken away for a growing colony in the vicinity – there are hints of just such problems at Colchester before the Boudican revolt.
Many rural districts were placed under the supervision of urban centres, whether coloniae, municipia or civitas centres. The regional juridical power of the magistrates of the coloniae was the greatest, followed by those of municipia. The civitas centres may have had far less extensive regional responsibilities and, while it is certainly possible that they may have supervised some of the small towns (vici) and rural districts (pagi) in their vicinity, this should not be considered a foregone conclusion. Although we cannot reconstruct the fine detail from the surviving evidence, we should imagine a mosaic pattern of landholding across Britain, with the state in one form or another controlling a large amount of territory and a mixture of other block holdings assigned to communities (urban, native peoples, sub-tribal groups) or to individuals or to religious interests. The individual legal settlement between Rome and urban communities will have differed markedly in terms of the amount of land assigned to the town, sold or given to private individuals in its vicinity and the extent to which the local community was expected to supervise financial and administrative affairs across a wider area.
The conquest of Britain and its long-term occupation was no act of altruism; the exploitation of its land and resources was fundamental to the success of the province. While it has been fashionable in recent years to emphasize the participation of the British elite in this and to minimize the potential role of colonists, it is important not to forget that military and other settlers potentially played an important role in many parts of the country – perhaps disproportionate to their numbers. In addition to the legionary veteran colonies at Colchester, Lincoln and Gloucester, settlement of discharged soldiers was probably widespread. For instance, twelve auxiliary veterans are epigraphically attested in southern Britain (two in Norfolk, one in Northamptonshire, one at Lincoln, one in Essex, three in Cheshire/Shropshire, one in London and one just south of the Thames in Kent, two in Gloucestershire). Since the visibility of such veterans is affected by a number of factors, this must represent a minuscule proportion of the total of military personnel who settled back into civil life within the province. The native British contribution to the rural landscapes of the province cannot be denied, of course, but it needs emphasizing that these were equally imperial landscapes, constructed by and for a colonial society in which native Britons participated alongside others, subject to legal and financial limitations. Parallels of a sort exist in the settlement of Dacia (modern Romania) after the bloody annexation at the beginning of the second century. Roman sources have for long been interpreted as implying that Dacia was repopulated with large numbers of settlers following the ‘extermination’ or enslavement of much of the local population. However, archaeological research suggests a more mixed population in the countryside, incomers and indigenous peoples living alongside each other.
Britain under Roman rule comprised a series of predominantly rural societies whose history was largely unwritten and in consequence has been poorly integrated into conventional accounts of the province. A common stereotype of the Roman countryside is that it was structured around elite estates with distinctive Roman-style farms (villae) at their centre. The presence in the countryside of substantial numbers of sites of less characteristically Roman type has rarely been given proper recognition.
At all times, the rural population of Britain under Rome will have far outnumbered its urban population (the latter probably never much more than 200,000) or its military community – perhaps in the order of 200,000, including associated civilians. An estimated rural population of c. 1.6 million (80 per cent of the total) seems entirely plausible, given current knowledge of rural settlement. However we wish to play these figures, in common with other provinces Britain was essentially a rural society. If Roman-period population was in fact higher than 2 million, the bulk of the extra numbers will have been in the countryside.
Aerial photography has had a far more dramatic impact in British archaeology than is the case for the Mediterranean countries, but its value is all the greater when combined with field survey. The total number of late Iron Age and Romano-British sites in England alone is now over 100,000. Although the quality of the data is variable, the sheer quantity of information and its geographical coverage creates dramatic new possibilities for analysing settlement at local and provincial levels. It is on this foundation that future studies of the rural geography of Roman Britain will be built.
A fundamental problem to address is the extent to which the academic agenda has prioritized some parts of the landscape and certain settlement types over others. This is best illustrated by reference to the Ordnance Survey Map of Roman Britain. The main categories of site included have remained fairly consistent since the early editions (though nomenclature has changed slightly, the underlying criteria for selection have remained solidly Romano-centric): major towns (coloniae, civitas capitals), defended small towns (settlement, defended), a selection of other nucleated undefended settlements (essentially those along the major roads of the province), spa towns, villas, other substantial rural buildings, temples, legionary fortresses, forts, fortlets, signal stations, supply bases, temporary camps, frontier works, roads, milestones, lighthouses, aqueducts, artificial watercourses, mineral extraction sites, salt-making sites, pottery/tile-production sites, Roman barrows or mausolea, major hoards. Strict site-selection criteria have been applied, but with the specific aim of illustrating the major ‘Roman’ categories of site. What we see are primarily the features of government and domination, and of elite society.
What is not mapped is important: a large number of substantial villages/hamlets (especially those away from the main road network), cemeteries, distinctive regional settlement types – such as fortified major settlements in the north and west of Britain (for instance, the rounds of Cornwall, nucleated defended sites in Wales, brochs and duns in Scotland). This creates a fundamental problem in attempts to map Roman-period Britain. In areas where Roman site ‘types’ are uncommon (notably, but not exclusively, Cornwall, Wales and northern Britain), the maps appear empty apart from Roman military installations standing guard over large capital letters denoting ‘tribal’ names. In reality, these were areas where different rules of government operated and indigenous settlement followed a variety of trajectories. Even the largest native sites are excluded as settlement sites on the Ordnance Survey map, though, to take one notable example, the hillfort of Traprain Law in Scotland is marked (but not named) as the location of a major hoard. Clearly it is impractical to plot on a small-scale map every roundhouse and indigenous site, but even in the use of categories of nucleated settlement, the selection seems to be in favour of the most ‘Roman’ examples. The desire to avoid clutter on a small-scale map is understandable, but what is omitted from such maps is the settlement evidence relating to the vast majority of the population, especially in the north and west, where hardly any non-military sites are actually shown. These academic priorities produce a map of ‘Roman Britian’, rather than of ‘Britain in the Roman empire’, and it is important to remind ourselves that these are not the same thing at all.
The net result of this sort of mapping is to suggest greater homogeneity than was the case and to leave difference under-explored. My alternative vision stresses the heterogeneity of rural settlement and seeks to bring out more detail of the regional variability in settlement morphology and to give greater consideration to the less ‘Roman’ sections of the landscape than is conventionally done. In this diversity lies an important clue to the evolving cultural identity of different groups.
Accounts of Roman rural development tend to generalize, looking for common denominators like villas, road networks and small towns. On the other hand, varied archaeologies have been left by different geographical and social groupings, reflecting divergent regional histories and distinct rural identities. When we examine the regional landscapes as they were remodelled under Roman rule, we can distinguish a number of key features that characterize Britain as a province under imperial rule: the concentrations of military sites, the network of roads and the associated posting stations along them, the financial and security controls at the major harbours and markets, the towns and aqueducts, changes in religious sites and practice. The variability that we see in rural settlement change is also indicative of changing identities under the influence of the imperial regime. These developments in rural settlement must also reflect broad changes in the nature of landholding and ownership. Regional differences in settlement history under Roman rule may also correlate to some extent with variance in the negotiation of power between defeated British groups and the empire.
Maps often portray late Iron Age Britain as carved up into a series of contiguous ‘tribal’ territories, corresponding exactly with the civitates recognized by Rome. The sizes of these implied territoria are vast – most being equivalent to at least two modern counties. This conventional picture of massive civitas units is more problematic than generally admitted; the truth is that we do not know much at all about the extent of the landholdings of either the late Iron Age or Roman-period peoples of Britain. For instance, the widely assumed correspondence between Iron Age coin ‘territories’ and Roman civitates does not stand close scrutiny. To some extent the post-conquest civitates were simply a matter of administrative convenience and we cannot assume exact territorial or social correlation with pre-existent Iron Age groups. Although I shall associate named peoples with particular geographical regions, these are by no means certain ascriptions, and a key argument is that the territories directly allotted to them by Rome were smaller than commonly assumed.
On the other hand, it was certainly the case that Rome found it convenient to delegate a good deal of local administrative responsibility to the towns of the empire. This meant that rural districts (pagi) some way from a defined civitas and its assigned core territory might be placed under the administrative supervision of the civitas. But the extent to which large areas of the countryside may have been differently administered by army or financial officials also merits consideration. Figure 10 (p. 262) offers a hypothetical vision of how this sort of mosaic pattern might have operated, though there is no more hard evidence for this reconstruction than for the more conventional view of enormous civitas-based administrative districts.
When considering the rural landscapes around towns, we need to distinguish between the territory directly controlled by the urban centre and the broader hinterland and hierarchy of settlements that it interacted with. It is also pertinent to consider the different forms of landholding that were practised in Roman provinces. In other parts of the empire, urban territories were generally smaller than those commonly envisaged for Britain, with very large areas of the surrounding landscape given over to private estates, imperial or state lands, religious land, regions under the control of the army, and so on. Our image of the geopolitical map of Britain needs revising accordingly.
A fundamental tool of imperial government was accurate measurement of the extent and quality of landholdings and the Romans were skilled in this area. Iulius Frontinus, governor in the mid-70s, wrote that there were three basic types of land in the provinces: that which was divided and assigned (centuriated), that which was surveyed only in outline (delimited) and that which was not defined by any measurement (ager arcifinus). The last of these was clearly a general term also for all the land declared ager publicus but left undelimited.
In some conquered territories, notably those of the western Mediterranean conquered in the latter centuries BC (Italy, southern Gaul, Africa, Spain, the Balkans), massive schemes of orthogonal land survey were superimposed across huge areas. These rectilinear systems – commonly known as centuriation because of their basis on squares (centuriae) of c.706 m side – have often left distinctive traces in the modern landscapes. The prevalent angles of field boundaries and the incorporation of regular perpendicular grids of roads (limites) within the field systems are the tell-tale markers of ancient centuriation. The lack of certain evidence for centuriation in Britain has often been commented on, though by the time that Rome invaded Britain, such major schemes of landscape transformation were becoming rare in newly conquered territories. On the other hand, Ammaedara, a Flavian colony founded on the site of a decommissioned legionary fortress in North Africa, was linked to a formal gridded land survey of its territorium. It is thus probable that centuriation schemes once existed around the three first-century coloniae in Britain, in relation to the allocation of standard land grants to veteran soldiers. It is much less certain that centuriation was employed on a wider scale, for instance, with regard to lands re-assigned to native civitates.
However, the absence of evidence of widespread centuriation of the British landscape does not equate with a lack of rigour on the part of Rome in delimiting and quantifying existing landholding arrangements within the province. The writings of the agrimensores show that their work continued throughout the Roman period and that in addition to formal centuriation they were frequently employed to carry out more basic surveys of conquered land, measuring area and assessing quality as a basis for taxation and provincial records. Pre-existing field systems were often assessed and adopted as the basis for provincial records. The establishment of the provincial census was undoubtedly a major task of the fledgling Roman bureaucracy in Britain, leading to the codification and registration of British landholding patterns in an unprecedented level of detail.
Centuriated land and delimited land normally had elaborate maps (formae) made to illustrate the overall shape of the scheme and the detail of individual allotments as a form of guarantee to those allocated the land. Although no formae were made of the agri arcifini, it is clear that surveyors were also active in these landscapes. In place of the minutely detailed surveys conducted in centuriated landscapes, the main purpose in agri arcifini was to define in broad terms the extent of lands and the appropriate level of tribute to be levied on the local community. In all types of land, territorial limits of neighbouring blocks were physically established and marked by boundary markers (termini), which could take the form of stone or timber pillars, or upturned amphoras, or of marks on standing trees or natural features such as hills, slopes, water courses, watersheds, or built/dug features such as walls, ditches, roads and so on, in part following the custom of the region. Siculus Flaccus specifically mentions the role of magistri of the rural pagi to extract from the farmers the necessary labour for the establishment and maintenance of the country roads, which connected settlements, but also helped define property boundaries.
A major change under Roman rule was thus that land was quantified, surveyed, subject to title and ownership rights as never before. The Roman penchant for written documentation is sadly not matched by survivals, but the discovery of a writing tablet recording the sale of a small wood in Kent, bordered by other land units in private ownership, is a signal exemplar of just how much documentation has been lost. The purpose of this documentation and land survey was not bureaucracy for bureaucracy’s sake; it was the key to Roman exploitation of Britain. Taxation and tribute assessments depended on written records, based on major land surveys and the formal census. All British lands will have been subject to taxation or tribute (ager stipendiarius) and this might have been either cash-based or (more likely in the first century) in kind. Tribute conditions often also involved the provision of a given number of military recruits for the Roman army, and the fulfilment of labour or pack-animal liturgies (for instance in relation to mines and quarries, the road and transport system). Such terms were generally defined at the communal rather than the individual level.
The apparent lack of evidence for centuriation in Britain is less surprising today than it was a few decades ago, when the extent of pre-Roman land clearance was believed to be far smaller. It is now clear that there had been considerable expansion of Iron Age farming, indicated by the development of field systems in many parts of Britain. So-called ‘Celtic fields’ survive in many upland areas such as Wessex or northern England, but aerial photography has revealed traces of similar systems across extensive parts of lowland Britain. Exploration of the landscapes of Nottinghamshire and southern Yorkshire has been particularly important in this regard, revealing the existence of hitherto unsuspected semi-regular (‘brickwork’) field systems of late Iron Age date extending across a vast area once thought to have been still forested in Roman times.
In the initial post-conquest phase, the situation will have been relatively straightforward, with lands by and large claimed for the Roman state, with the exception of such territory as was parcelled off to client rulers, whose loyalty was recognized in an immediate and exemplary way, although these lands were no doubt expected to be tributary contributors in much the same way as the annexed provincial territory. The emperors will have begun establishing their own land portfolios almost immediately, and these will have grown progressively through confiscations and inheritance. Many British groups were, of course, initially allowed to remain in effective occupation of their traditional lands, subject to such restrictions as the army might impose and to the obligation to make tribute payments. As a condition of good behaviour and depending on the forward movement of the Roman army, the long-term status of these lands would ultimately be addressed as part of the process of transferring groups from military to civil administration. Evidence from other provinces shows that such arrangements were normally negotiated with individual civitas groups; radically different outcomes for neighbouring peoples resulted from Rome’s judgement of their relative loyalty, merit and progress.
The landscapes of Britain were thus transformed under Rome in major ways. The driving through of major roads, the physical demarcation of lands of different uses and for different stake-holders, new types of site and some introduced plants (such as orchard crops) and an expanded architectural lexicon – all these contributed to dynamic change. But much also echoed late Iron Age practice in Britain and to a lesser extent in northern Gaul. The landscapes of Britain under Rome were characterized by types of isolated and nucleated settlements, enclosures, fields, tracks and paddocks. These contrast with the extreme regularity of centuriated field systems elsewhere or the strict application of Mediterranean models of settlement. Britain was in this sense, as in so many others, a peripheral possession of the empire, with distinctive traditions.
Some areas of Britain stand out from their surrounding districts in terms of their archaeology, and this can occasionally be linked to the existence of natural resources that were of importance to Rome. Rome had a number of options with regard to the exploitation of such resources: direct extraction; indirect extraction through contracts issued to individuals or companies; indirect extraction by tapping into preexisting activity through tribute assessed in kind; re-assigning the lands to native communities for them to develop the economic value of the resource; maintaining the lands on which the resources lay as state/imperial property but deferring full exploitation. Comparative studies from other provinces, in the absence of explicit evidence from Britain, would suggest that Rome carefully evaluated the resources available from conquered territory and that deposits of minerals, decorative stone and salt were quite routinely taken into and maintained under state control. Where not exploited directly, there was some preference for issuing contracts to private contractors with the appropriate experience to exploit the resource, rather than placing the onus on local initiative and capital. Where the resource was difficult to exploit efficiently on a large scale, as in the case of coastal salt-working, it is probable that Rome obtained a share of production in lieu of rent or taxation due from local communities. In my vision of a mosaic of different land types, these resources were jealously guarded by the state and were not generally returned to the full control of British civitates.
Our picture of the pre-Roman, Roman and post-Roman landscapes is increasingly detailed. Pollen evidence, derived from cores taken from lakes and peat bogs, provides dated sequences of landscape change. Macroscopic analysis of plant remains (typically carbonized seeds) illustrates the range of both cultivated plants and weed species. Rare survivals of structural timbers and charcoal residues can indicate species present and whether woodland management was practised. Analysis of animal bones likewise provides species lists, and some measure of their relative abundance, but also vital additional data relating to comparative size, physical characteristics and age at death (often indicative of whether the animal was raised primarily for meat or for secondary products such as dairy or wool). Finally, the material culture of farming leaves many traces: ploughshares, scythes, rotary grain mills, wool combs are easily recognizable for what they are. Built farmyard structures can be equally diagnostic whether they be byres, granaries or corn-drying ovens.
From a consideration of the environmental data, it is now clear that land clearance and farming were much more advanced in prehistoric Britain than archaeologists once believed. These new data reveal the large scale and broad range of Iron Age farming and its relative sophistication. Iron Age landscapes were once believed to have been characterized by small areas of human settlement and farming, separated by tracts of dense woodland and marshes, providing a convenient explanation of gaps in the settlement data. Improved environmental knowledge and increased information on rural settlement patterns have demonstrated this image to be false. The major phase of woodland clearance, according to the pollen record, was in the late Bronze Age, over 3,000 years ago. This process continued in many parts of Britain throughout the Iron Age and the instances of Roman date where tree/shrub pollen counts first fall below 50 per cent of recorded pollen are by far outnumbered by those of the late Iron Age.
Regional differences in vegetation history can increasingly be traced. The south-east of Britain saw progressive woodland clearance throughout much of the Iron Age, building on substantial clearance of Bronze Age date. There is ample evidence for agricultural activity, extending to substantial grain storage facilities at many sites, extensive field systems, and so on. East Anglia saw continuing woodland clearance in the Fens and some appearance of cereals before a major phase of marine incursion reduced human activity there. A pollen core from Breckland in Norfolk hints at a more heavily wooded landscape that was giving way to heathland in the late Iron Age. South-west England has produced a varied record, with the Somerset Levels for instance remaining more heavily wooded for longer into the Iron Age. The great moorland massifs of the far south-west such as Dartmoor, Bodmin Moor and Exmoor had all been substantially cleared of tree cover in the Bronze Age, and there are indications that the late Iron Age landscape was a comparatively open one, with numerous settlements practising both stock-raising and agriculture. Pollen samples from the Midlands generally reflect substantial woodland reduction through the Iron Age, though some locally important stands evidently remained. Both north-west and north-east England have produced evidence of major woodland clearances of later Bronze and early to mid Iron Age date. A large group of pollen cores from the area of Hadrian’s Wall suggests that the landscape clearance accelerated in the late Iron Age. There are botanical data that also demonstrate a considerable extension of agriculture at that date, especially in the north-east at sites such as Thorpe Thewles. Both Scotland and Wales show evidence of substantial woodland clearance in the Bronze Age, with an accelerated or revived rate of clearance in the late Iron Age and/or Roman period. Cereal pollen appears in many samples from all the upland areas of Britain.
The Iron Age saw dramatic advances in and expansion of agriculture and stock-raising across Britain. Cereal cultivation, once believed to have been very much a speciality of southern Britain, in fact was well established in the north and west by the latter centuries BC. People had divided up large areas of the landscape with field systems and other boundary features, centred on numerous settlements, typically of roundhouses. Emmer wheat (Triticum dicoccum), the traditional grain of Neolithic and Bronze Age Britain, was progressively less important in comparison to spelt wheat (Triticum spelta) and, to a lesser extent, bread wheat (Triticum aestivum). Hulled six-row barley (Hordeum vulgare) was the most widespread of the less high-quality grains. Both oats (Avena) and rye (Secale cereale) were also cultivated in pre-Roman times. The expansion of the range of grains indicates the spread of agriculture beyond the dry and light soils favoured by earlier farmers, with spelt and barley in particular being suited to a wide range of conditions. Experiments at a reconstructed Iron Age farm at Butser Hill, Hampshire, have shown that many of these grain types could have produced high yields, opening the prospect of an era of surplus production. Weed assemblages also illustrate the extension of agriculture to wetter or heavier soils and this is backed up by widespread evidence of cereal pollen in upland locations of northern and western Britain. Other field crops included peas (Pisum sativum), beans (Vicia faba var. minor) and flax (Linum usitatissimum). Livestock included cattle, sheep and pigs as the three predominant species, though with some significant regional variations in their relative importance (cattle were less common on downland areas with poor water sources, sheep were rarer in wetland environments). Other common domestic animals included horses, dogs and domestic fowl, the last a late Iron Age introduction.
The cumulative picture of the Iron Age landscapes is one of mixed woodland and open land by the late Iron Age. Areas of denser forest were comparatively rare and in a few areas woodland had been reduced to a very low threshold. There is some evidence from waterlogged sites to suggest that woodland management took place in the form of coppicing. Perhaps the greatest change in the rural landscapes of Britain involved the extension of farming on to more varied soils, indicating a whole range of agricultural improvements. Far from being a country of primeval forest and limited agricultural development, it is now clear that farming was evolving rapidly and dynamically in many regions of Britain during the mid to late Iron Age. Yet the Roman period was one of significant change, and just as our evaluation of pre-Roman conditions has had to be revised, so our expectations of the rural history of Britain under Rome also require review.
The natural and agrarian landscapes of Roman Britain were in many respects similar to those of Iron Age Britain in ecological and environmental terms. While this was not a phase of revolutionary and wholesale change in farming, key distinctions appear to be an upward shift in the scale of rural production and greater diversification. For instance, there were important new plant varieties introduced during this period: the grape vine and orchard crops such as apples, medlars, cherries and plums; many garden plants such as asparagus, beet, cabbage, carrot, celery. The look of the countryside itself changed; travellers could have been under no illusion about the nature of the imperial landscapes they were crossing. Far evolved beyond the traditional routes of the Iron Age, the presence of a substantial network of engineered roads, with bridges at many river crossings, presented a new framework for movement. The roads connected forts, towns, markets and religious centres – all of which were to a greater or lesser extent marked out as different from what had gone before. Approaching towns, our travellers would pass cemetery areas, industrial suburbs and, occasionally, aqueducts that again represented dramatic and visible change to the landscape. Even in more conservative rural districts, change would be visible in the form of occasional more Roman-style buildings and probably an enhanced tendency to demarcate property boundaries.
Notions about progressive change inhibit analysis of the Roman-period landscape history. It is sometimes assumed that there was unilinear economic growth and raised standards of living under Roman rule and that this was a self-evident goal that everyone will have aspired to. Put another way, there has been a tendency to see every rural site as a potential villa, that required only improvement of farming practices, new markets and the slow accumulation of capital resources by an eager peasantry to be achieved. There are several objections to this. As already noted, we can now demonstrate that agriculture was highly evolved in late Iron Age times in many parts of Britain. In parts of northern Britain, compared to precocious Iron Age expansion, the level of Roman-period exploitation appears to have been at best static or perhaps even curtailed. Studies of farming technology and analysis of the range and yield of cultivated plants and animals have shown that the degree of change between Iron Age and early Roman farming was relatively slight, with more significant waves of innovation occurring in the latter centuries BC and in the later Roman period. Why was this so?
A clear trend in many areas of Britain was the innate conservatism of vernacular architecture. If ‘Roman Britain’ was the land of the villa, then ‘Britain in the Roman empire’ was characterized overall by the roundhouse. Although increasingly supplanted in some regions by rectangular house forms, it was equally enduring in others. One explanation, of course, is that the issue of whether to embrace new styles of rural building was more pertinent to the upper echelon of late Iron Age society than to the bulk of the populace. A more fundamental objection is that even among the British elite groups we should not assume an unerring desire to adopt Roman architecture, manners and graces. Had the Roman model been so overwhelmingly potent, it is indeed strange to note the slow uptake of villa architecture in many parts of the country. Even elite settlements show a considerable conservatism of approach through the early centuries of Roman rule, despite the construction of some early villas that could have served as models.
Current syntheses suffer from an over-emphasis on upper-echelon sites such as villas. They make general assumptions about the progressive impact of Roman conquest, and the improving effect on the economy of native Britons. They use the dichotomy of a villa-dominated civil zone and a landscape of roundhouses in the military zone. The selection of site plans and reconstruction drawings in many books demonstrates the way that subconscious assumptions lead us to filter out native features. Roundhouses are normally omitted from views of villas, though there is increasing evidence that the two structural types existed side-by-side. There is no denying the importance of villas in our assessment of rural change, nor of the socio-economic significance of the people who were able to fund specialized construction work of this sort. But it is both more problematic and more interesting to look beyond these villas at what else the countryside has to tell us about regional histories and trajectories.
The garrison history has already made clear the existence of long-term regional differences. At the most general level, we can conceive a tripartite division of British landscapes between areas under various forms of civil governance, areas under military administration and extra-provincial areas. Within each of these broad categories, we can hypothesize further levels of difference.
We can only guess at total numbers of rural settlements, and indeed at the overall population of Britain in the Roman period. Estimates of the maximum population have varied wildly over the years, from 400,000 to 500,000 suggested in the earlier twentieth-century accounts, to a 1980s range of 4 to 6 million. Many commentators feel the truth may lie between these two extremes, and a very conservative estimate of 2 million for the mid-second century is adopted here. One of the key factors in the rise in estimates of population has been awareness of a much greater density of Romano-British rural settlement as a result of aerial photography and field survey. A classic study of the making of the English landscape in the 1950s was dismissive of the impact of the Romans – based on the c.100 towns and small towns and 2,500 rural sites of all types then known, the conclusion was that much of Britain was still engulfed by forest or was uncultivated wasteland.
There has been an explosion of new data revealing a density in many parts of the country approaching one site per sq km. This has significant implications for considerations of the potential cultivated area and population, although not all sites were occupied contemporaneously, nor can we assume that we are close to recovering the total ancient site universe. In fact, discoveries continue at a high rate, representing a 40 to 100-fold increase in knowledge of ancient settlements in the second half of the twentieth century.
On the other hand, the total number of known villa sites has grown much more slowly than other categories of rural site, largely because this class of site has a more distinct archaeological signature, with the result that a larger percentage of these elite residences are already known (for example, in the 1950s, villas accounted for c. 20 per cent of rural sites, today they represent c. 2 per cent). The ancient reality was probably even starker, since the numbers of unidentified lesser rural sites will far outweigh the numbers of ‘lost’ villas. The archetypal rural site in Britain was thus not a villa and the villas were likely outnumbered 25:1 or more by other forms of settlement, most with a distinctively Iron Age pedigree in Britain, others of hybrid forms reflecting the impact of new ideas concerning construction and the use of space.
It is sometimes assumed that the villas represent economic development in Britain and that the less-developed (that is non-villa) landscapes of the north and west are indicative of poor socio-economic evolution. That view is probably too simplistic and we also need to look at other possible explanations. Another way of categorizing territory might be as landscapes of opportunity and landscapes of resistance, with many gradations in between the extremes. This has the advantage over the traditional model in that it introduces another element beyond economic potential to the equation and does not assume that all Britons wanted to live as rich Romans did, or that Rome wished all parts of Britain to see the same sort of development. We have already noted the process of negotiation between the Roman governing power and the indigenous inhabitants about the question of land allocation. This was an unequal negotiation between victor and subject people and did not follow a uniform pattern.
The ‘most successful’ regions thus did not equate neatly with the best agricultural areas and in certain cases we can speculate on the reasons why. Similarly, the relative poverty and underdevelopment of the west and north can only partly be explained by reference to land quality or social make-up of the people. Some areas of Britain were viewed by the Roman state as more suitable or deserving of opportunity and advantage and some were viewed as more resistant and undeserving. The possibility that practical considerations also played their part in determining the nature of land settlement cannot be excluded. For instance, the exigencies of military supply could have affected Rome’s arrangements in the west and north to the long-term disadvantage of the local population. The attitude of the native Britons is another important factor, especially in so far as they felt that the empire presented them with opportunities. Alternatively, the extent to which they found themselves disadvantaged by the resources granted them may have promoted resistance, however subtle or passive.
The villa is one of the best-established categories of rural settlement in Roman provincial studies, although in reality the definition applied varies wildly between different areas and from one scholar to the next. The ancient sources defined the villa as a rural house, but it is clear that they were thinking in terms of a house owned by someone of social standing, not peasant hovels, and they discriminated between the more luxurious type (villa urbana) and productive farms (villa rustica), though in reality the two elements were commonly represented at a single site. In the Mediterranean provinces, however, archaeologists have tended to reserve the term ‘villa’ for the major sites with signs of luxurious living or large-scale productive facilities. Lesser structures, even if stone-built and with tile roofs, are generally described as farms or farmsteads. For Britain, a different approach has mostly been taken: a basic definition of a villa is ‘a rural building of Roman aspect’ (key diagnostic features are the use of stone or brick/tile, rectilinear plan, tessellated pavements or mosaics, bath facilities). One problem is that classification focuses on the ‘house’ element, whereas in most cases there were ancillary structures present, though these have frequently been less well-explored archaeologically. This sort of catch-all description also faces the objection that it must cover wide variation. Compare, for instance, huge country houses such as Woodchester, covering c. 14,000 sq m and comprising over fifty separate rooms and corridors, with smaller buildings such as the 200 sq m, five-roomed building with timber porch at Lockleys. In recent years there has been increasing scepticism of the utility of so broad a definition of the ‘villa’, though it is not easy to see where to draw the boundary between the two extremes just described. On the other hand, it will be apparent that the choice of a more Roman style of building over a more traditional form was in itself a significant one and that the discrimination has some validity. The Lockleys site, for instance, despite its early appearance as a simple row house, went on later to develop further architectural characteristics that favour its eventual classification as a villa.
Just how many villas were there? The minimal position is represented by the Ordnance Survey map, listing only 278 villas and 285 other substantial buildings, for a total of 563. A published gazetteer lists c. 2,500 possible sites and represents a maximal position, since some of these are questionably villas. For the sake of my argument here, let us say there were 2,000, representing only about 2 per cent of the 100,000 rural sites of late Iron Age and Roman date now known in England. Palatial villas were exceptional – perhaps only twenty to thirty across all Britain – with the vast majority of other villas being small, medium or large farms, some decorated with tessellated pavements or mosaics and representing widespread displays of modest wealth. Many of these sites were late developments of the third or fourth centuries.
A series of distinctive types of villa has been identified on architectural grounds. This includes (from simplest to most complex): rectangular houses; corridor houses (rectangular houses fronted by a corridor or portico); winged corridor houses (same as previous but with the addition of projecting rooms or ranges at the end of the corridor facçade); aisled houses (perhaps a development from late Iron Age prototypes); villas with courtyards (buildings arranged around at least three sides of an enclosed yard); elaborate villas of large scale and architectural elaboration (often with courtyards). As noted already, there is huge variation in scale and pretension between these different structures and it is important to consider other factors than simply the architectural type. Room counts, calculated floor area, the richness of interior decor, presence of bath-suites (and scale of these facilities), can all give an impression of the relative grandeur. As a rule of thumb, we might consider as ‘small villas’ sites with fewer than ten separate rooms, as ‘medium villas’ those sites with ten to twenty-nine separate rooms and as ‘large villas’ those with over thirty rooms.
The distribution of villas closely echoes the area of developed urban centres, predominantly in the south and east and with comparatively few in the north and west. However, this often-remarked north–west/south–east separation disguises a great deal of additional variation in the density, typology and dating of villa settlement within Britain. There were comparatively few early Roman villas, and the majority of villas, and virtually all the grandiose houses, were of relatively late construction, especially in the fourth century. This has important implications, rarely fully considered, for our understanding of the impact of Rome on the countryside in the initial centuries after the conquest.
Villas do not occur everywhere or in equal numbers in the civilian zone of Britain and this has often been interpreted as reflecting degrees of economic or social success. It could also indicate differing patterns of land tenure. The general absence of villas in the Fenland, on Salisbury Plain and in parts of the Weald has sometimes been seen as evidence that these areas were maintained as imperial estates. However, there are no good reasons for thinking that imperial estates would have been devoid of villas, since such properties were often contracted out to private individuals to run for profit. In North Africa, for instance, large imperial estates at the heart of the agricultural zone were studded with farms, villages and villas. Nevertheless, the absence of villas in parts of the British landscape may indicate different forms of land-holding, with less emphasis on private ownership of ‘estates’ and perhaps more on communal access to resources and land. The more probable candidates for imperial estates may in fact be found in areas of higher villa development, where rural culture and identity also stand in contrast to the average development. However, villas were not the exclusive preserve of any one group in society and different possibilities of ownership and occupation should be considered. They could relate to any or all of: representatives of the Roman emperors and the state, absentee landowners, the army, members of the British civitates, settler groups (notably discharged soldiers with capital to invest), private individuals (both British and from overseas), religious or entrepreneurial bodies.
What do villas signify? The question is not easy to answer, especially in view of the heterogeneity of sites in Britain. Some villas evoke comparison with the ‘stately homes’ of more recent British aristocracy. This is especially true of the large, multi-roomed sites, with lavish use of mosaics, painted wall-plaster, imported marble, statuary and so on. There is an interesting historiographical link here, in that the first villa excavations in the eighteenth and nineteenth centuries often took place on the estates of landowners who were keen to establish a connection with Roman antecedents.
There is ample evidence from other parts of the empire to show that there was a strong tradition of elite construction of luxurious rural houses, often contained within productive farm complexes. The elites in the Mediterranean provinces were strongly linked in to the practice of government and display at urban centres, so they generally maintained households in both town and country, moving between the two areas, and making use of bailiffs and tenants to conduct the economic exploitation of their lands. It is uncertain whether the British elite behaved exactly as those in Italy or Africa; the presumption should perhaps be that there was a degree of difference. There is some evidence to suggest that by the fourth century elite wealth was being spent more lavishly on rural building projects than in the towns, though it also remains to be proved that those responsible for the most extravagant villas were also members of the urban governing class, as opposed to imperial officials or wealthy immigrants or largely absentee owners from other provinces. The eighteenth-century British aristocracy was more country-based, though many maintained houses in London or major resorts, such as Bath, and we should be cautious about assuming too close a degree of similarity between them and the owners of major villas. There is no doubt, however, that the large villas as a group were elite residences of considerable scale – in the most impressive examples comparable with almost any others north of the Alps.
Another idea is that villas were elements of Roman-style estates, where landownership and tenancy were controlled by formal deeds and legal documents of types that were indisputably Roman. To this extent the villa was symbolic of economic relationships as well as social ones. But the long time-lag in the widespread adoption of villas (and the uneven spread even in south-east Britain) suggests that the construction of a villa was an optional extra and not a necessary condition of a Roman-style estate. Many farms that lacked villa-type buildings will nonetheless have been fully documented landholdings in the Roman manner. On the other hand, in the north and west of Britain, the relative absence of villas suggests the possibility that most of the land there was accorded different status in law.
Villas have most commonly been viewed as Roman-style farms. That is to say, farms that took on an architectural form that echoed urban standards of construction in sharp contrast to the norms of both the late Iron Age and of the Romano-British countryside in general. The choice of rectangular building plans, often using mortared-stone walls or footings and making use in some cases of heavy roofing materials, such as tiles or slates, will have contrasted with predominantly curvilinear building types, such as the roundhouse, and a general preference for timber, wattle and daub, and thatch. Even a small ‘villa’, little more than a three-roomed cottage, could still represent an architectural statement about identity and aspirations.
The development of villas mirrors the slow evolution of towns in Britain. There were few villas in the first century and some at least of these appear to have been exceptional commissions, perhaps to indulge imperial officials, prominent immigrants or uniquely important native Britons. Some of the earliest villa development took place in the southeast, essentially in the area within which the coinage of the eastern and southern kingdoms had previously circulated. The best-known example is the palatial villa at Fishbourne, just outside Chichester, generally identified as a residence built for Togidubnus. This building is outstanding for first-century Britain in terms of its scale, its architecture, its use of marble and mosaics.
There are a few other well-appointed medium villas of early date and again these may have represented special construction projects. A good example is The Ditches villa, established within an Iron Age enclosure closely adjacent to the oppidum at Bagendon. The rarity of early stone villas of any degree of elaboration is shown by the fact that fewer than ten villas with second-century mosaics are known for the whole of England. More commonly, early villa development is represented by small rectangular timber structures replacing Iron Age roundhouses at a number of sites.
There was undoubtedly a major phase of development in the third and fourth centuries, representing a gradual growth in numbers rather than a sudden revolution. Many sites reveal a long sequence of gradual accretion of signs of wealth and luxury (additional rooms, baths, hypocausts). Numerous long-established rural centres only developed stone-built ‘villa’ buildings in the late third or early fourth century. In the fourth century, a number of structures were built or expanded on an exceptional scale and lavishness of ornamentation (Bignor, Chedworth, Turkdean, Woodchester are the obvious examples). It needs to be stressed that these were exceptions not the norm.
The reasons why there was enhanced villa development in the fourth century have been much debated, with explanations often focusing on an assumed in-migration of wealthy Gallic families, fleeing barbarian invasion in northern France and buying up British estates. The increased fragmentation of the Roman imperial bureaucracy in Britain following the Diocletianic reforms has also been highlighted as a possible explanation of some of the most exceptional villas. Others have argued that the fourth century marks the apogee of Roman economic development in Britain, with some local families achieving previously unheralded levels of wealth. There are objections to each of these theories, though the truth may encompass all of them.
There has sometimes been a tendency to describe Roman rural development as a natural, unilinear and ever-upward trend. However, the process of villa construction was retarded in many regions and numerous villas did not survive in use through to the end of the provinces’ existence; indeed many had already failed by the mid-fourth century. Moreover, the level of investment in construction and maintenance was sometimes beyond the pocket of the owners. For instance, some bath-houses at villas appear never to have been completed and heated up, illustrating what an expensive and lengthy business the construction of such facilities was.
Traditional British roundhouses have often been dismissively referred to as ‘circular huts’, subconsciously emphasizing their difference from elite villa houses. Some excavations have revealed a transition from roundhouse to villa on the same rural site and this sequence has been identified as a key element of the ‘Roman’ countryside. However, there has been an increasing recognition that the roundhouse was a much more enduring feature of the Romano-British landscape and that the social understanding of this building form was variable and complex like the ‘villa’. Even at villa sites, circular buildings were constructed through much of the Roman occupation. Moreover, roundhouses should not automatically be viewed as low-status or poor dwellings. Although these were commonly constructed of non-durable materials (timber, daub, thatch), the dimensions of the larger roundhouses are impressive, with internal diameters commonly in excess of 5 m and examples as large as 16 m recorded. These were substantial constructions and provided extensive internal spaces, which shaped many social rituals and interactions in traditional British societies. The larger and more elaborate roundhouses were thus impressive markers of a sophisticated society.
Parallels from other colonial societies alert us to the potential complexities. The memoirs of Nelson Mandela contain an illuminating description of his guardian’s farm in the Transkei. Jongintaba was a local chieftain recognized by the British colonial authorities, and his home consisted of two whitewashed rectangular buildings of European style, complete with tin roofs, and a cluster of ‘rondavels’ (superior roundhouses). The European-style buildings were a visible symbol of his colonial powers and responsibilities, and Nelson had seen nothing like them in an African village before he arrived there following his father’s death. Social behaviour and use of space at Jongintaba’s ‘villa’ was complicated and transitional between two traditions. The rectangular buildings were used for receiving and entertaining people, but the chief’s extended family continued to sleep and carry out many other daily activities in the six traditional roundhouses to the rear. This is not to imply that villas in Britain should be read in this way, but it raises interesting questions about social choices that are taken in colonial societies. Can one assume, as some commentators have in the past, that Britons if they had the choice would automatically have adopted the comforts of the villa over the roundhouse? The comparatively slow and uneven take-up of villa building may suggest otherwise for at least some part of the British elite. The other main possibility to consider here is that the economic and social encouragement offered by Rome was highly focused on a small group of the British elite class.
Roundhouses were thus present even on villa sites, but were actually far more ubiquitous generally within the rural landscapes, though each region tended to have its own distinctive character. There were also Iron Age traditions of rectilinear building, so the increasingly common appearance of rectangular structures from the first century could be due either to trends already present in Iron Age society or to Roman influence. Until the florescence of villa building in the third to fourth centuries, much of Britain remained dominated by native forms of rural settlement, with a thin veneer of new types of buildings and structural forms. The persistence of pre-Roman building forms for domestic use was even more common and consistent in the north and west of Britain.
However, it would be wrong to imply that building forms were unchanging. The construction of stone buildings marked a major departure in many parts of the countryside, even if not all examples were rectangular in form. There are examples of substantial stone-footed round buildings of late second-to fourth-century date and some at least were evidently for domestic use. Other possible functions of circular structures include shrines or industrial or agricultural buildings. A major change during the Roman period was the widespread addition of rectangular buildings to the traditional curvilinear forms. Many of the new structures utilized timber and daub just like the roundhouses, but some adopted more complex carpentry techniques (indicated by sleeper trenches or footings for sill beams rather than earth-fast posts) and some buildings incorporated tile or slate roofs. The latter are significant markers of change as the technology and tolerances of supporting a heavy solid roof are very different from those for a thatch construction.
One of the most popular forms of rectangular building in Roman Britain was the aisled hall. Sometimes described as ‘aisled barns’, these appear to have been multifunctional structures, with excavated examples serving both as domestic accommodation and a range of agricultural functions. Some aisled halls can be dated to the first century and it is possible that the form was already present in south-east Britain in the pre-conquest phase. At any rate the aisled hall can be seen as a peculiarity of Britain, there being no strong correlation with the vernacular architecture of the nearest provinces in Gaul and Germany. There appears to have been a major expansion in numbers of aisled buildings from the second century onwards.
The older view of aisled buildings was that they represented accommodation for bailiffs and workers and/or served as ancillary farm buildings on villa estates, but increasing numbers are now known where the aisled house, often with its main walls constructed on stone footings, was clearly the main dwelling on an estate, with mosaics and bath units incorporated into a subdivided aisled plan. The collapsed gable end of an example at Meonstoke provides a graphic illustration of the architectural refinement and impressive vertical elevation of these structures. The aisled hall became a standard architectural feature in many parts of the civil zone and it is to some extent a hybrid form, common to both villa and non-villa sites. It combined the large internal space of the roundhouse, with the same flexibility of use, but it could also reflect adoption of new structural forms and an increased scale of architectural display. Some scholars have speculated on the suitability of such large communal spaces for a society still based on extended clan groups. The precise details of these social relations are generally elusive, given the quality of archaeological evidence (though an attempt has been made to define functional and gender divisions within an aisled villa at North Wanborough). It is increasingly clear that the social architecture of many rural sites in Britain differed significantly from Continental parallels. Even in the adoption of Roman outward forms of architecture and display, British societies followed their own local rhythms.
Many late Iron Age rural settlements were erected within enclosures, to be characterized as ‘compounds’. A typical compound site might have two to three roundhouses in contemporary occupation, suggesting that the social unit present was an extended family or clan. Although there are examples of isolated homesteads, these compound sites indicate that British society commonly revolved around such extended units. Attempts have been made to identify the same tendency at some villa sites, where within an enclosed ‘compound’ the domestic accommodation appears to be duplicated (among the clearest examples are villas such as Halstock, Rockbourne, Gayton Thorpe, with two houses of relatively equal size and status). Such ‘unit villas’ as they are known may indeed be evidence of a British tradition translated into Roman architectural terms, but the arguments have sometimes been pushed too far in this direction, with almost every villa plan being re-evaluated for its possible divisions between extended family groups. An intriguing example is a double villa at Bradford-on-Avon, where despite near identical ground-plans the two buildings seem to have had very different functions. If we see unit villas everywhere, we are in danger of replacing one over-generalizing model with another. Nonetheless, the archaeological evidence suggests that they were a potentially important feature of indigenous rural society.
Another area of great interest is the nucleated rural settlement – essentially, villages or hamlets. This group in part overlaps with the class of sites often described as small towns, especially when they occur alongside major roads (though the broad class of secondary agglomerations are often not tied into the main road network). There is an increasing recognition of the existence of villages alongside villas, and some of these were no doubt integrally linked to one another. On the other hand, some villages evidently had pre-villa origins and other examples are situated some distance away from villas. They represent another pre-Roman social tendency in parts of Britain for people to live in larger social groupings and the fact that we should find some element of continuity of such settlements under Rome should not surprise us.