13
Settlement and the settlement pattern in the west

Beyond the Jurassic Ridge the landscape of Britain becomes more varied. Considerable expanses of old hard rock have created barren mountainous areas like the core of Wales and the Highlands of Scotland while the predominantly westerly air currents ensure that the Atlantic coasts have a higher precipitation rate than much of the rest of the country. Since height above sea-level (which affects temperature) and mean annual rainfall are crucial factors in controlling crop growth, considerable areas of the north and west lie at or beyond the limit of viable agriculture. Thus, slight fluctuations in climate can cause major dislocations in settlement pattern (pp. 33–4). This factor, and the harsh nature of some of the environments, has tended to create an intense regionalism in settlement form and a high degree of specialization in economy.

In this chapter we focus on the western, Atlantic-facing, regions of southern Britain and each discrete zone is treated separately.

The south-west peninsula

The south-west peninsula has a distinctive character unlike any other area of Britain. Bounded on three sides by the sea and separated from Wessex by the marshlands of Somerset, its communities have developed along individual lines influenced more by the structure and food-producing potential of the land than by external stimulus. The peninsula is dominated by six areas of moorland: the sandstone uplands of Exmoor in the north-east and the granite masses of Dartmoor, Bodmin, Hensbarrow, Carnmenelis and Penwith – all providing some areas of light, if not particularly fertile, soil more suitable for pasture than for arable farming (Figure 13.1). Between these upland masses lie the Culm measures of the middle Devonian and the Permian rocks yielding an intractable clay soil far less congenial to prehistoric settlement.

The literature concerning the Iron Age settlement pattern in the south-west has grown considerably in recent years. For the environmental background the reader is referred to three important papers summarizing the evidence from Cornwall (Caseldine 1980) and Devon (Caseldine and Maguire 1981; Caseldine and Hatton 1996). Cornish defended settlements are discussed in detail by Nicholas Johnson and Peter Rose (1982) and the background to the Cornish Iron Age is assessed by Henrietta Quinnell (1986). The first-millennium BC settlements of Devon are fully treated by Bob Silvester (1979). Two major multiperiod surveys have also been published, on Bodmin Moor (Johnson and Rose 1994) and Exmoor (Riley and Wilson-North 2001).

Figure 13.1 Distribution of enclosed settlements, including rounds and cliff castles, in Cornwall (source: Johnson and Rose 1982).

Moorland settlements

In chapter 3 the principal types of settlement centring upon the Dartmoor massif and dating to the late second and early first millennia were discussed. Briefly summarized, they included large enclosed villages, sometimes with attached stockpens, sited around the south and west limits of the moor; open villages with huts linked by stone walls found in much the same area; and small farmsteads of one or two huts associated with a few embanked fields. This last type is concentrated on the more protected and correspondingly drier eastern edges of the moor. The dating of these Middle–Late Bronze Age settlements is notoriously difficult, for while there can be little doubt that many of them began to be occupied in what is conventionally the Bronze Age, the upper limit of dating is difficult to define. In some cases they may have continued in use as late as the latter half of the first millennium. Three Dartmoor settlements have produced artefacts suggesting an Iron Age date – Kestor, near Chagford, Foale’s Arrishes and Gold Park (Figure 13.2). All three settlements comprise a group of huts scattered among small rectangular fields with a single larger hut protected by an enclosing wall. Strictly, this type of site is little more than an agglomeration of small homesteads but their close spacing and the existence of a single more impressive house might suggest that we are dealing with hamlets or even villages, in which some form of class structure prevailed. A tighter nucleation can be recognized at other, undated, sites such as Broadall Lake in the Upper Yealm valley, Devon, where ten circular huts were grouped along the side of two fields.

The excavation at Kestor has provided a valuable insight into a type of settlement which may tentatively be dated to the fifth or fourth century BC. One of the isolated huts, lying at the junction of three field walls, was totally excavated. It was a simple structure, some 8.2 m across internally, enclosed by an outer wall to take the lower ends of the rafters, which were further supported by a circular setting of posts and a central post to hold up the crown of the roof. Internal fittings were restricted to a hearth and an area of cobbling close to the entrance. The larger hut, built towards the centre of the enclosure known as the Round Pound, was 11.3 m across internally and of a more complex structure, but here too internal posts acted as roof supports – the difference being that a central opening seems to have been provided with a drip- pit beneath to collect and drain away rain-water (Figure 13.4). The reason for the opening is thought to be that the hut was used for iron-smelting and fumes therefore had to be removed. Inside were found a small bowl furnace filled with iron slag and nearby a forging pit, presumably for reheating the bloom prior to the hammering necessary to remove impurities. However, some doubt has been expressed as to the date of the ironworking, one possibility being that it was a medieval intrusion (Silvester 1979, 178–9). The hut was enclosed within an oval stonewalled pound 30–33 m across, provided with a single narrow doorway opening on to a terraced drove road which ran between the surrounding fields. A second, much smaller hut lay close to the pound wall but was completely undated and may indeed belong to the medieval period. While it would be wrong to argue from the evidence of one site alone, and one where there is some doubt about the stratigraphy, the fact that the largest hut may have belonged to the iron- smith could be taken to be an indication of the high status in which the community held such a man.

The close relationship between the pound, the trackway, the other huts and the fields leaves little doubt that all functioned together. Indeed, it was possible for the excavator to show that the ancient ploughsoil stopped 1.8 m clear of one of the huts and that a slight negative lynchet had been formed, demonstrating clearly that the field had been ploughed after the hut had been constructed. Nearby, the ploughsoil was found to overlie a layer of peat, which was shown by pollen analysis to belong to the sub-Atlantic period – a time when the climatic conditions were becoming much wetter. The Kestor settlement and its fields therefore belonged to a community which colonized an area of virgin moorland after the middle of the first millennium, at a time when climatic deterioration had already set in. In all probability it was one of the last inroads to be made on the moor before wetter weather drove the long-established population from the uplands.

The presence of fields and the discovery of a saddle quern are sufficient to show that arable farming was practised, but the field systems tend to be far more limited in extent than those of the south-east, implying – but by no means proving – that corn production was of subsidiary significance. Of the pastoral aspects of the economy little can be said: animal bones are destroyed by the acid moorland soils, and relevant artefacts, apart from a spindle whorl from Kestor, are exceedingly rare. Flocks and herds must, however, have been a dominant feature of the early first-millennium economy, to judge by the large pounds of Middle–Late Bronze Age date, and in all probability pastoral activities continued to be of first-rate importance throughout the latter part of the period – as, indeed, the multiple-enclosure forts to be described below imply.

Figure 13.2 The settlements at Kestor and Gold Park, Devon (sources: Kestor, A. Fox 1955; Gold Park, Gibson 1992).

The second moorland site to receive detailed archaeological attention was Gold Park on the north side of Dartmoor only one kilometre or so north of Grimspound. The settlement consisted of a number of rectangular fields and droveways, with huts scattered between them and one hut in a circular enclosure. The hut chosen for excavation was an isolated structure. Two very distinct phases of construction were recognized. In the first the house was built of timber and was represented by an external drip gully showing that the house would have been 8 m in diameter with the weight of its roof supported on a circular setting of timber uprights (Figure 13.2). In the second constructional phase the structure was rebuilt in stone, with an internal diameter of 4–4.5 m. Radiocarbon dates for the timber house suggest a Middle Iron Age date in the fourth to second centuries, with the possibility that the stone building belongs to the first century BC. The few potsherds recovered would support this later date. Little evidence was found to reflect upon the economy of the settlement but its moorland fringe location was ideal for a family exploiting the arable and woodlands of the lower slope and the grazing of the higher moor.

Scattered huts among small rectangular fields is a pattern of settlement reflected over most of the south-west peninsula from Dartmoor to Land’s End. But while many such sites are known, few have been excavated outside Dartmoor itself. One site, Bodrifty on Mulfra Hill near Penzance, Cornwall (Figure 13.3), has however provided some details in response to limited excavation. Here more than twenty simple circular huts were spread over a tract of land 0.4 km across, but at least nine are clustered together and were later enclosed by a pound wall built probably in the second or first century BC at about the time when some of the huts show signs of rebuilding. In the original pre-pound settlement, beginning perhaps as early as the sixth or fifth century, the individual huts were joined by lengths of walling rather like the technique employed on some of the Dartmoor villages such as Stanton Down. The huts themselves (Figure 13.4) are closely similar in structure to the Dartmoor examples, with wide stone walls, central hearths and sometimes internal settings of posts to help support the rafters in the case of the larger buildings. The excavation produced few finds, but spindle whorls and saddle querns give some hint of the agricultural and pastoral activities of the community. The significance of Bodrifty lies in the relatively large quantities of pottery recovered, ranging from types which would not be out of place in sixth- to fifth-century contexts in Wessex to jars of the second to first centuries decorated in the South-Western Decorated style. The ceramic evidence allows the possibility that occupation continued well into the first century BC. Other Cornish sites show that once established many of them continued in use into, and often throughout, the Roman period.

Thus the Dartmoor and Cornish settlements are very close in details of planning, structure and economy, but on Dartmoor while there is evidence to suggest limited occupation continuing to the first century BC no sites are known to have continued into the Roman period. The contrast with Cornwall is striking, but can be explained if it is assumed that the onset of sub-Atlantic conditions forced the inhabitants of the high moors to abandon their traditional farmlands, while those living around the fringes of the less elevated and more hospitable Cornish moors were able to continue and even expand their territories. If this is correct, it could mean a gradual shift of population away from Dartmoor and Bodmin Moor during the latter part of the Iron Age. It is necessary, therefore, to consider where these dislocated pastoralists might have settled.

Figure 13.3 The settlement at Bodrifty, Cornwall. Letters refer to the identifications in the excavator’s report (source: Dudley 1957).

Multiple-enclosure forts

The lowland areas of Devon and eastern Cornwall are densely scattered with settlements commonly referred to as multiple-enclosure forts (Figures 13.5–13.7), evidently designed for pastoral communities and (where evidence is available) not built until the fourth or third century BC. It is tempting to see these structures as the successors of the moorland settlements, colonizing the richer low-lying soils of the Devon and Cornish hills, but until the dating of a sufficient sample has been established it would be unwise to be too dogmatic as to their origins and relationships.

Characteristically, the multiple-enclosure settlements are often sited, more for pastoral convenience than for strength, on hill-slopes or the ends of ridges overlooking springs or river valleys. Normally the inner enclosure, between 0.2 and 1.6 ha in extent, is defended by a fairly massive bank and ditch laid out in a circular or sub-rectangular plan with a single entrance in one side. The additional enclosures were created in several ways (Figure 13.5). At Milber Down Camp, Devon, two outer banks and ditches were arranged concentrically with the inner earthwork, both enclosing a considerable additional hectareage. An even more impressive example of this type is Clovelly Dykes, where two enclosures are concentric with the main camp and two additional areas are attached to one side of the outermost, thus providing four separate enclosed areas in addition to the central element. All were interlinked with entrances, the main entrance being so sited as to give easy access to nearby springs. A variation on this type can be seen at Castle Dore, Cornwall, at Denbury, south Devon, and at several other sites where the outer enclosure is pendent upon the inner, the two ramparts being close-spaced for part of their length but diverging widely towards the entrance to form an outer enclosure. A third type occurs in which the main camp is provided with a separate annex attached to the side with the entrance. Strictly, the little fort at Blackbury Castle, Devon, belongs to this class, but here the passage between the two entrances was flanked by banks and ditches, giving rise to the so-called barbican approach. More distant cross-banks were also extensively used, usually to cut off the neck of the promontory or spur upon which the main enclosure was situated. This type of arrangement frequently enclosed a substantial hectareage, which almost invariably included springs or streams.

Figure 13.4 House plans from south-west Britain (sources: Bodrifty, Dudley 1957; Kestor, A. Fox 1955).

Figure 13.5 Multiple-enclosure forts in Devon and Cornwall (sources: A. Fox 1953, 1961b).

Figure 13.6 Distribution of multiple-enclosure forts in south-west Britain (source: A. Fox 1961b).

The intention behind the siting and planning of these settlements is clear enough: the inner enclosures, some of which were comparable in size to enclosures of Little Woodbury type though others were smaller, presumably formed the inhabited area where the owner and family lived while the outer enclosures were designed to protect the homestead pastures and their watering-places.

Figure 13.7 The multiple-enclosure fort of Castle-an-Dinas, Cornwall (photograph: Dr J.K.S. St Joseph. Crown Copyright reserved).

The most extensively examined example of a multiple-enclosure fort is the double-ditched enclosure at Killibury, Cornwall. Excavation within the inner enclosure has produced ample evidence of long-lived occupation in the form of a dense scatter of post-holes associated with a depth of stratified occupation levels. This contrasts markedly with the space between the two defensive enclosures where a single trial trench exposed only a series of shallow gullies, showing that activity in the outer enclosure had been slight. Much the same picture is provided by Castle Dore, Cornwall. In a reassessment of the 1936–7 excavation, it has been argued (Quinnell and Harris 1985) that the densely occupied interior was in use from the fourth to first centuries BC, during which time there were a series of circular timber buildings sited mainly against the rampart in the southern half of the enclosure. The outer enclosure was not extensively examined but in the late phase a single hut was identified close to the outer gate.

Apart from pottery, artefacts are not particularly prolific from the few multiple-ditched enclosures that have been excavated but signs of moderate wealth have come from Milber Down Camp, in the form of an iron dagger handle, and from Castle Dore where armlets of bronze, shale and glass have been found as well as glass beads (Fitzpatrick 1985b; Henderson 1985). Items of this kind, together with the effort required to construct the impressive earthworks of these sites, suggest that multiple-enclosure settlements were the homesteads of an élite. It is hardly surprising therefore that Castle Dore should have produced sherds of imported wine amphorae of first-century BC and first-century AD date.

The form of the enclosures and their general orientation towards rivers or streams has long suggested that the economy depended heavily on livestock. The nature of the husbandry is difficult to assess, since animal bones seldom survive in the acid soils of the area. The consistent occurrence of spindle whorls does, however, point to the significance of sheep and one can hardly doubt that cattle too were important. Field systems were notably absent from the vicinity of the enclosures, but that some cereal cultivation was carried out seems certain in the light of the evidence from Killibury, where emmer wheat, spelt wheat and oats were identified. The relative balance between husbandry and cultivation necessarily remains unclear but the overall impression must be that the economic wealth of the élite was founded on their control of flocks and herds.

Rounds (Figure 13.8)

Another type of habitation site occurs widely in Cornwall and north-west Devon, usually in hilly country 60–120 m above sea-level. These settlements, called rounds in Cornwall, consist of a simple banked and ditched enclosure, seldom exceeding a hectare in extent and sited invariably on good arable land. Inside, a few huts were usually built close against the bank. In size and form, however, rounds display a considerable variety (Johnson and Rose 1982). Several have been excavated: the round at Trevisker, Cornwall, produced evidence of two successive Iron Age phases, the later comprising a 1.2 ha enclosure. The settlement, which dated to the second or early first century BC, superseded an earlier unenclosed settlement dating back to the second millennium. At Castle Gotha, Cornwall, a 0.6 ha enclosure and its two huts, built late in the second century BC, continued in occupation for about 500 years.

Detailed fieldwork in Cornwall (Thomas 1966b; Johnson and Rose 1982) has shown that rounds were densely distributed, varying from one per 2.1 square km to one per 4.5 square km (Figure 13.1), but since only a few have been excavated, it is impossible to say that all were in use at the same time; many of them may not have been built until the Roman period. Nevertheless, as a settlement type rounds were probably widespread before the first century BC. The nature of the economy is difficult to assess but rotary querns and a possible iron sickle from Tre- visker indicate that cereal growing played a significant part.

Courtyard houses (Figures 13.9–13.11)

A different type of settlement, the so-called courtyard house, is found on the high ground in west Cornwall. This is essentially a central paved ‘courtyard’ surrounded by rooms and byres, the whole complex being enclosed with a massive stone wall, usually with a single entrance leading into the yard. This conventional view – that the ‘courtyards’ were open – has, however, been challenged by the suggestion that the entire structure was covered by a single roof. If so then the ‘courtyard’ should be interpreted as a communal living space with rooms off (Wood 2001). While the suggestion is possible, one might question whether the substantial timbers needed to create the roof structure would have been readily available. When courtyard houses occur in clusters, as they do at Chysauster (Figures 13.9 and 13.10), the appearance is not at all unlike a deserted medieval village.

Figure 13.8 Plans of Cornish rounds (source: A.C. Thomas 1966b).

Courtyard houses are often associated with field systems. One of the best preserved of these is an extensive system found to extend along the slope on which Chysauster was built. Survey and trial excavation suggest that the system began to be constructed in the Bronze Age, but the fields around the settlement were later modified to form a more irregular pattern, strongly lynchetted (Smith 1996), perhaps in an attempt to conserve soil as the village of courtyard houses developed.

At Carn Euny, where excavation of a cluster of courtyard houses has been thorough, the complex growth of the settlement can be appreciated by observing the relationship of the various walls. Here the courtyard houses represented the last phase in a settlement which began probably as early as the sixth or fifth century with circular timber buildings. It was early in this sequence that a fogou (an underground chamber) was built.

Whether single or in groups, courtyard houses are usually unenclosed but several sites are known where houses are contained within a round (Figure 13.11). At Porthmeor at least three houses and their gardens were enclosed in this way at a date subsequent to their erection, while at Goldherring the round contained a single house with its cultivation plots outside. Where evidence of date is available, it is generally found that courtyard houses could not have begun before the first century BC and the majority of them continued in use throughout the Roman era. In a thorough review of the dating evidence Henrietta Quinnell has argued that courtyard houses are best regarded as a Roman phenomenon (1986, 120).

Figure 13.9 Courtyard houses at Chysauster and Carn Euny, Cornwall (sources: Chysauster, H.O’N. Hencken 1933; Carn Euny, Christie 1978).

Figure 13.10 Chysauster, Cornwall (photograph: English Heritage).

Cliff castles (Figure 13.12)

Cliff castles, situated as their popular name suggests on rugged headlands jutting into the sea and protected from landward approach by complex lines of banks and ditches, feature prominently in the archaeological literature. Their similarity to comparable structures in Brittany has been used to argue for an immigration of Breton Veneti into the south-west in the first century BC. Such a view must now be discarded since several of the cliff castles have produced pottery of considerably earlier date and it is simplest to see the Cornish and Breton sites as the obvious response, of communities requiring defence, to closely similar landscape potentials. This is not, however, to deny contact between the two peninsulas over many centuries.

The choice of dramatic location and the investment of energy in creating the, often substantial, defences imply a desire to impress, but the purpose of these structures is still quite uncertain. Those which have been examined by excavation have usually produced domestic debris and traces of small circular houses, but the remoteness of many of the sites, their exposure to extremes of weather and the comparatively limited area within suitable for habitation together make them far from ideal settlement locations. While they could have been places of refuge or even the homes of the élite it is not unreasonable to suggest that they may have served as sacred locations symbolically sited in the liminal space between land and sea. They are all dramatic locations and some incorporate natural phenomena like the great rock-stack of Trevyn Dinas (Sharp 1992; Herring 1994). A thorough excavation of one of the cliff castles may throw some light on the question.

Figure 13.11 Courtyard houses built into rounds (sources: Goldherring, Guthrie 1969; Porthmeor, Hirst 1936).

Summary

The landscape and settlement pattern of the south-west peninsula is diverse but the majority of the settlements can be divided into two broad categories: minor enclosures, i.e. the rounds; and major enclosures, the multiple-enclosure forts and, perhaps, the cliff castles. In addition there is a scatter of unenclosed farmsteads about which very little is known. What are noticeably absent are the large hillforts of Wessex type which, we suggest, represent the power centres of large territorial groupings. In other words this level of social organization, so evident in the centre south, does not appear to be present in the south-west. Instead the most appropriate model would seem to be a simple three-tier structure with the major enclosures representing the residences of the élite while the minor enclosures were the homesteads of their vassals. Below these would have come the unfree living in the unenclosed settlements. In such a system one might expect the élite to be more concerned with wealth in the form of livestock, leaving the vassals to till the land and to provide them with grain on a tithe basis in return for protection and patronage. The archaeological evidence would not contradict such a model but is hardly yet sufficient to support it or to suggest an alternative.

Figure 13.12 Cornish cliff castles (sources: The Rumps, Brooks 1964; Gurnard’s Head, A.S.R. Gordon 1941).

It is tempting to suggest that the pastoral-dominated economy of the south-west mitigated against the growth of political centralization to the extent to which it appears to have developed in the south-east. Instead, there arose a society composed of individual lordships broadly of equal status across much of the region. It was a system that was to persist long after the Roman invasion and into the Dark Ages.

Wales and the Borderlands

That part of the British Isles that lies west of the Severn–Dee line is, for the purposes of this book, regarded as part of the western zone of Britain although strictly it is divided into two quite separate regions by a central mountainous spine – a coastal ocean-facing region and an inland border region permeated by the tributaries of the rivers Severn and Dee (Figure 13.13). The coastal region itself divides into several sub-regions but for convenience we will make a somewhat arbitrary divide between the south and west and the north.

Figure 13.13 Defended settlements in Wales (source: Davies and Lynch 2000).

Figure 13.14 Walesland Rath, Dyfed (sources: Wainwright 1971a; G. Williams 1988).

South and west Wales

The southern coastal area of Wales, stretching from the Usk valley to Pembrokeshire, was in many ways similar in its settlement pattern to the south-west peninsula of England, a similarity which can be explained, in part at least, by the geomorphological likeness of the two areas. The Carboniferous and Triassic rocks of Glamorganshire and Monmouthshire and the Ordovician rocks of the western counties give rise to soils comparable to those of Devon and eastern Cornwall, while the craggy coastline west of Swansea Bay could easily be mistaken for Cornwall. Similarly, the predominantly north–south flowing rivers tend to cut the territory, lying between the mountains and the sea, into strips and blocks, making communication by land difficult.

Two very useful surveys of settlement in the region have been published: the Royal Commission has surveyed all the Iron Age settlements in Glamorganshire (RCHM(W) Glamorganshire 1, 1976) while the settlements of Carmarthenshire have been considered in an environmental and historical framework by George Williams (1978, 1979, 1988).

Unenclosed hut groups, generally undated, occur in Carmarthenshire and Pembrokeshire but they are ill-known compared with the Dartmoor sites. One group, partly excavated in 1899, lay on the top of Moel Trigarn in the Prescelly Mountains. Aerial photography has shown that some of the huts pre-date the multiple walls of the later hillfort and must therefore belong to an open settlement of some size.

Multiple-enclosure settlements of the types defined in the south-west occur in some number in all four of the south Welsh counties. Site location again indicates a preference for valley-sides dominating springs or streams, suggesting that here, as in the south-west, the enclosures were constructed by predominantly pastoral communities more concerned with watering their flocks and herds than with purely military considerations.

The West Country round also has its counterpart in the rath of south-west Wales. It is estimated that of the 580 hillforts in Wales about 230 enclose less than 0.4 ha, and of these three-quarters lie in the south-west (Figure 13.13). While it is true that many were occupied in the Roman period, and some may even have been constructed then, a high percentage probably date back to the Iron Age. Total excavation of Walesland Rath near Haverfordwest, Pembs., provides a rare insight into the structure and development of this type of establishment (Figure 13.14). Here an oval-shaped area, 64 by 49 m internally, was enclosed by a bank and ditch pierced by two entrances. The south-east gate was massively constructed, with three pairs of timbers which must have once supported a tower, while the western entrance was flanked by dry-stone walling, with its gate set back within the line of the ramparts. Internally the enclosure was packed with timber structures, including at least six circular timber huts, several of which show signs of rebuilding on a number of occasions. A more unusual type of building was constructed from three pairs of large timbers; this resembles the structures found in the Wessex and Welsh Borderland hillforts. The provision of an eaves-drip trench around it shows that it was roofed and may possibly have been a house but does not necessarily preclude the possibility of its serving as a granary in this context. A maze of post-holes arranged in alignments around the periphery of the enclosure close to the rampart are best interpreted as a zone of four- and six- post storage buildings, reconstructed on numerous occasions. Dating evidence at Walesland Rath was sparse, but radiocarbon determinations allow the first phase to be assigned to the third century BC or a little earlier, while the second phase, defined by the rebuilding of the rampart on a more massive scale and the blocking of the west gate, began in the early part of the first century BC. Occupation continued into the third century AD.

A second excavated site in the same category is Coygan Camp, sited on a promontory of Carboniferous limestone overlooking Carmarthen Bay. An initial occupation, tentatively assigned to the eighth to second centuries, was followed by the construction of the enclosure bank and ditch. Two entrances were provided: a north-west gate, the approach to which was further protected by a length of additional rampart built across the neck of the promontory, and a south-west entrance leading down to the marshy pasture at the base of the hill. Contemporary internal structures were not found, nor was the material culture particularly rich, apart from a pair of bronze early La Tène bracelets, but a large quantity of animal bones was well preserved in the alkaline soil. The collection from the enclosure phase shows that numerically cattle predominated, amounting to 64 per cent of the total, with sheep/goats a mere 16 per cent, closely followed by pigs at 15 per cent. Allowing for the fact that a single cow produced about seven times the meat yield of a sheep, it will be evident that the basic diet was beef, the intake of mutton and pork being negligible by comparison. The relative importance of meat to grain cannot be assessed, but quernstones were very rare and storage facilities unrecorded. In all probability, the inhabitants were pastoralists, using the fertile pastures of the neighbouring Devonian soils and the marshland fringes at the foot of the hill for rearing and maintaining herds of cattle. The proximity of the sea provided an additional food source: fish bones have not been preserved but shellfish were collected in quantity from the estuary of the Taf and from the rocky pools at the base of the Pendine Cliffs. Like Walesland Rath, Coygan Camp continued to be occupied well into the Roman period. The neighbouring sites of Trelissey and Cwmbrwyn, both earthwork enclosures, also contained substantial Roman-style masonry buildings. It is not unreasonable to suggest that here too we are dealing with Iron Age settlements which remained in use for some centuries after the Roman Conquest.

Several other settlements which could be classed either as raths or as multiple enclosures have recently been examined. At Parc Cynnog, not far from Coygan Camp in Carmarthenshire, the earthworks of a partially multivallate enclosure, about 0.4 ha in area, were sectioned, but little evidence of internal occupation was discovered. However, the excavation of Castell Coyan, Carm., produced clear indications of circular buildings of timber within a small enclosure, but dating evidence was lacking. At Pen y Coed, Llangynog, Carm., the almost total excavation of a small embanked and ditched enclosure yielded a complete plan of an Iron Age farmstead (Figure 13.15) with one large roundhouse, a yard divided by fences and a four-post granary set against the bank. The entrance to the enclosure was approached by a road flanked by ditches similar to the banjo enclosures of Wessex. A similar arrangement is evident at Woodside Camp, one of a pair of presumably contemporary enclosures barely 50 m apart (Figure 13.15). Excavations at Woodbarn Rath, Caer Cadwgan and Pembry Mountain, though on a less extensive scale, have all shown occupation to have been quite intensive, the last two yielding significant samples of cereals (see below p. 434).

The long corridor-like approach to the enclosures of Woodside, Pen y Coed and Dan y Coed is a common characteristic of settlements in the south-west and a number of sites have been recognized from aerial photographs in which the outer ends of the corridor ditches splay out to create an outer enclosure. These concentric antennae enclosures (James 1990) are quite distinctive but what function the outer enclosure had is unclear. While it is possible they provide enclosed pastures or an enclosed cultivation zone it is equally possible that the defined zone had ritual significance as a kind of liminal area between the inside world of the home and the outside world of production. Similar concentric enclosures are found in north Wales (pp. 299–301).

The multiple enclosure banks of sites like Pen y Coed and Woodside Camp and the prominent positions of others like Coygan Camp are likely in some way to reflect on the status of the lineage occupying the enclosure. In this context Castell Henllys, Pembs., is of particular interest. Here a comparatively small enclosure sited at the end of a ridge was provided with unusually massive defences across the line of approach, with a very impressive gate and a chevaux de frise. Although a number of circular timber-built houses were found inside, not all were in use at any one time and the site is never likely to have housed more than one extended family. The implication of such elaborate defences, however, is that the site was one of élite status throughout the Iron Age.

Figure 13.15 Woodside Camp and Dan y Coed enclosure, Dyfed (source: G. Williams 1988).

Few of the rath-like enclosures in Glamorganshire or Monmouthshire have been adequately excavated, with the notable exception of Mynydd Bychan, Glam., overlooking the Ewenny valley, where a small enclosure was found to contain a group of timber-framed circular huts, probably dating to the first century BC. Rebuilding in stone took place in the middle of the first century AD (Figure 13.16). At Whitton, South Glam., a roughly rectangular ditched enclosure established in the early decades of the first century AD was provided with seven timber roundhouses of different dates. Romanization, in the form of rectangular timber buildings, becomes apparent at the end of the first century AD (Figure 13.16). Both sites are best regarded as single family farmsteads.

Many of the promontories of south-west Wales were defended by banks and ditches, turning them into the equivalent of the Cornish cliff castles (Figure 13.17). Several have been partially examined by excavation. On St David’s Head, Pembs., a complex rampart protected a small group of six conjoined stone-walled huts 4.5 to 6.0 m in diameter, excavated in 1900. The material culture was sparse but whetstones, spindle whorls, hammer stones and fragments of iron were found together with a few glass beads and decorated shale pendents. No precise evidence of the date range of the settlement has come to light. An excavation on a closely similar promontory fort at Tower Point, St Brides, Pembs., has demonstrated a two-phase construction for the inner stone-faced rampart and has shown that here, too, huts were built within the protection of the rampart, but little cultural material was found and there was no direct evidence of date. Limited work at Llanstephan Castle, Carm., an inland promontory enclosure beneath the later medieval castle, has however produced a group of radiocarbon assessments indicating a date in the sixth century BC for an early phase in the occupation sequence. Trial excavation at Great Castle Head, Dale, Pembs., showed a similarly extended occupation, with an inner bank demonstrating three phases of construction during the Early and Middle Iron Age and a massive outer bank added in the Late Iron Age.

Further east several enclosed sites have now been partially examined by excavation. At the Knave, a promontory fort near Rhossili in Glamorganshire, two widely spread ramparts protected a cliff-top, the inner area of which was barely 46 m across. Pottery akin to South-Western Decorated Wares suggests a date in the second or first century BC but occupation could well have begun much earlier. The 0.1 ha promontory fort at Bishopston valley, Glam., produced pottery of similar date together with samian ware, showing that occupation continued into the Roman period, and at High Penard, Glam., a 0.8 ha promontory fort with widely spaced ramparts was examined, but yielded only Roman objects. Limited excavation at the Bulwarks, Porthkerry, Glam., a major promontory fort of some 4.1 ha, showed that here too occupation which began in the Iron Age continued into the Roman period.

Of the inland multiple-enclosure forts, Harding’s Down West Fort on the Gower peninsula, Glam., is the most extensively excavated. The main enclosure was small, only 0.6 ha, and strongly defended, but the outer compound, presumably for stock, more than doubled the size of the protected area. The approach to the main enclosure was made along a trackway which flanked the side of the annexe earthworks. The gate itself consisted of two pairs of posts 1.4 m apart which gave direct access to the enclosure, in which three house terraces could be recognized. Excavation of two of these has shown them to have been occupied by substantial timber-built houses. Artefacts were few, but pottery of Late Iron Age burnished type was found, together with an iron bloom. The evidence is therefore sufficient to suggest that the enclosure was probably occupied by a single family unit during the latter part of the Iron Age.

Finally, the excavation at Castle Ditches, Llancarfan, a small univallate hillfort, 5 km inland from the coast of the Vale of Glamorgan, revealed a complex structural history beginning with a small stone-walled enclosure which was later extended with an earth rampart and ditch to enclose about 2 ha. Internal occupation, when sampled against the rampart, proved to be intensive and produced a few sherds of South-Western Decorated Wares. Evidence of ironworking was apparent throughout the sequence.

Figure 13.16 Welsh settlements (sources: Mynydd Bychan, Savory 1954, 1956; Whitton, Jarrett and Wrathmell 1981; Pen y Coed, Murphy 1985).

Figure 13.17 Welsh promontory forts (sources: St David’s Head, Baring-Gould, Burnard and Enys 1899; Caerau Henllan, Ordnance Survey; The Knave, A. Williams 1939b).

From the above descriptions it is clear that the distinction between raths, promontory forts, univallate forts and multiple enclosures is blurred. The small promontory forts like the Knave and High Penard could strictly be classed as multiple-enclosure settlements modified to suit a promontory position, while the areas enclosed by many of them correspond to the areas of raths on inland sites. It is doubtful, therefore, whether the promontory forts should be regarded as economically or socially distinct from the other types of enclosure. It is simpler to suppose that most of them were merely variants of the multiple-enclosure and rath types.

In summary, it may be said that the settlement pattern and economy of south and west Wales had much in common with Devon and Cornwall, the reason being chiefly that the climate and geomorphology of both areas encouraged the development of basically pastoral economies, which in turn directly influenced the nature of the social structure. The emphasis appears to have been on individually defended homesteads, sometimes large enough to house not only the owner and family but also a considerable entourage. The absence of large hillforts strongly suggests a lack of centralized government. Cattle rearing played an important part in the economy, the animals no doubt serving as a manifestation of wealth which could be treated as currency. In the more mountainous regions sheep may have been more significant but positive evidence is at present lacking. The relative importance of cereal-growing is uncertain: querns are not uncommon but no extensive traces of Celtic field patterns have been recognized in the area, although cereal grains have been recovered from Caer Cadwgan and Pembry Mountain. At present, all that can be said is that limited corn growing was practised but in many areas may have been subservient to animal husbandry.

North Wales

The study of the settlement pattern of north Wales is made difficult by the almost total absence of dating evidence from the many excavated sites, but the high quality of the fieldwork carried out, particularly in Caernarvonshire, makes it possible to describe in some detail the basic settlement forms belonging to the pre-Roman Iron Age (Smith 2001).

Discussion must begin with the small multivallate enclosure of Castell Odo near Aberdaron, Caerns. (Figure 13.18), where as a result of extensive excavation a development sequence can be recognized, beginning with an open settlement composed of several circular timber houses associated with a small quantity of pottery similar in some forms to southern British assemblages dating to the fifth or fourth century. At some stage, while the huts were in use, work began on the construction of a timber palisade, but it appears never to have been completed – a fact which might be linked to the destruction of one of the houses by fire. In the third phase the settlement was enclosed by a bank of earth surrounding an area approximately 76 m across, in which were built several circular huts of stone. Later still the bank was revetted back and front with dry- stone walling and a new, similarly constructed bank was erected inside it, leaving a space of about 12 m between the two. Finally, possibly as the result of the Roman invasion, the defences were slighted and circular stone houses were built over them.

In terms of size and social structure, Castell Odo compares closely with the raths of southwest Wales. In all probability, it developed as the homestead of the local chieftain’s family, providing perhaps some protected accommodation for dependants within the enclosure. The intervallum space created later could well have served as a safe corral for stock, functioning in much the same way as the multiple-enclosure forts of the south-west. Elsewhere in north Wales this type of small defended settlement is by no means uncommon; several are found in the Aber- daron peninsula and along the coastal strip. Further inland at Dinas Emrys, Caerns., traces of a fenced settlement, possibly of similar type, have been recognized.

A rather different form of settlement is represented by the rectangular earthwork of Bryn Eryr, Anglesey, which has been extensively excavated and thoroughly published and therefore serves as a convenient type-site. Occupation began in the Early Iron Age with a single clay-walled roundhouse set within a timber stockade, but later, in the fourth or third century BC, a second circular building was attached to the first and the settlement was enclosed by a new rectilinear perimeter composed of a bank with an external ditch. Within the enclosure several rectangular ‘granaries’ were constructed and there is evidence in the form of quernstone and grain of a well-developed agricultural economy. The settlement continued in use well into the Roman period. Bryn Eryr is morphologically comparable to a number of other enclosures found on Anglesey and on the adjacent mainland (Longley 1998, figures 1 and 21) and appears to represent a distinct category of homesteads, possibly of élite status. They are closely similar to the enclosures, characterized by Whitton, found in south Wales. It is tempting to suggest that rectangularity was deliberately chosen in both regions as a symbol of the status of the occupying lineage.

Figure 13.18 Settlements from north Wales (sources: Castell Odo, Alcock 1961; Llwyn du Bach, Bersu and Griffith 1949; Bryn y Castell, Crew 1985; Crawcwellt, Crew 1998; Bryn Eryr, Longley 1998).

A third class of enclosed homesteads, generally known as concentric circle sites, has been recognized in Caernarvonshire. These are comparable to the concentric antennae enclosures of the south-west. A typical example, Llwyn du Bach, consists of a circular stone-built hut 9 m across, with two concentric enclosing walls 26 and 60 m in diameter (Figure 13.18). No indication of dating was found, but it has been suggested (Hogg 1966) that concentric circle sites belong to the pre-Roman Iron Age and may have formed the prototype from which some at least of the native Roman Iron Age enclosures developed.

A fourth type of settlement is simple stone-walled enclosures of varying forms, well typified by the large-scale excavation of Bryn y Castell in Snowdonia (Figure 13.18). Here the roughly pear-shaped enclosed area measured only 40 by 25 m and consisted of a cobbled yard with a single circular house in one corner. The economic basis of the community who used this exposed site is unclear but it is evident from a considerable mass of debris and several structures that the smelting and forging of iron was an important activity. Dating evidence would suggest occupation beginning in the first century BC and continuing to the first century AD. A closely comparable upland site was carefully excavated at Crawcwellt West, Merioneth (Figure 13.18). Here the settlement began c. 300 BC with a group of circular houses built in timber with stake- walls. These were later reconstructed with stone foundations, with occupation continuing into the first century AD. In both phases the buildings were associated with iron-smelting during the course of which large slag dumps accumulated downhill of the settlement area. It was estimated that during the life of the settlement about half a tonne of fully refined bar iron would have been produced.

A fifth type of settlement has been called the enclosed group (Gresham 1973). These can be defined as consisting of two or more separate huts situated close enough to each other to be considered as part of the same social unit. The individual huts are often linked by walls and are usually associated with a small field system composed of separate cultivation plots edged with boulders cleared from their surfaces. In general appearance they closely resemble the Dartmoor settlements described above (pp. 276–9). A detailed field survey in the Cwm Ystradllyn district of southern Caernarvonshire has brought to light a large number of such settlements which tend to cluster on the gentle mountain slopes around the 300 m contour. The one excavated site, Braich y Cornel, has produced no dating evidence, but it may well pre-date the Roman occupation.

Many other types of enclosed settlement are known in north Wales and have been discussed from time to time (Gresham 1973; Hogg 1966; C.A. Smith 1974, 1978b; G. Smith 2001), but many if not most of these are likely to be of Roman date, at least in their final form.

One of the most dramatic aspects of the north Welsh settlement pattern is the development of large hill-top settlements defended by stone walls. The evidence from the Caernarvonshire forts of Garn Boduan, Tre’r Ceiri and Conway Mountain (Figures 13.19 and 13.20) shows that many of them were occupied by sizeable communities of between 100 and 400 people living in circular, but largely undated, stone-walled huts, there being some twenty to eighty huts in each settlement. Normally the huts were totally enclosed by dry-stone ramparts, but occasionally, as at Garn Fadrun, Caerns., a considerable part of the settlement lay on the surrounding slopes. Further east at Dinorben, Denbigh., a settlement of comparable size was recognized, but in this case the huts were of timber and some at least pre-dated the construction of the hillfort.

Figure 13.19 Plan of Tre’r Ceiri, Gwynedd (source: Hogg 1962).

The Welsh Borderland

The wide strip of countryside west of the Dee–Severn axis, extending into the foothills of the Welsh mountains, has produced abundant evidence of pre-Roman Iron Age occupation, but mostly from the excavation of large hillforts which have tended to attract the attention of archaeologists at the expense of smaller settlements. Settlements are, however, becoming increasingly well known, particularly in the upper reaches of the Severn and the foothills of the Berwyn Mountains where enclosures of less than 1.2 ha abound, as they do between the Wye and Usk in the south (Spurgeon 1972).

In the Upper Severn valley between Montgomery and Oswestry many settlements have been discovered almost entirely as the result of systematic aerial survey (Whimster 1989; Gibson 1999). The enclosures can be broadly classified into curvilinear, rectilinear and hybrid. That the great majority of them have been ploughed out reflects their siting on good agricultural land. One of these sites at Collfryn, 10 km north of Welshpool, has been extensively excavated (Figure 13.21). The settlement began life as an unenclosed settlement, but some time around 300 BC a triple-ditched enclosure was created with widely spaced ditches. At this stage it covered an area of about 2.5 ha, but some time later, probably in the first century BC, the area was reduced by recutting the original inner ditch and cutting a new ditch immediately within it. Throughout most of the period the settlement seems to have comprised three or four timber- built roundhouses and a number of four-post granaries and would probably therefore have supported a single extended family. The economy was based on animal husbandry and crop production, including wheat, barley and oats, and the inhabitants had access to salt from Cheshire. The rich and varied soils of the Severn valley would have been ideal for mixed farming regimes. The status of the lineage occupying Collfryn is difficult to judge, but the multiple enclosure earthwork might be thought to indicate an élite status.

Figure 13.20 Plans of two north Welsh hillforts (sources: Garn Boduan, Hogg 1962; Conway Mountain, Griffiths and Hogg 1957).

Figure 13.21 Collfryn, Powys (source: Britnell 1989).

Of the excavated hillforts, several show signs of rebuilding and extensive occupation over a considerable period of time, and in the southern part of the area at least, many of the basic elements of the south-eastern economy are found.

The west: a summary

Much of the region we have designated as the west of Britain has, in recent years, been subjected to an intensity of archaeological survey quite unparalleled anywhere else in the country. As a result very large numbers of Iron Age settlements have been identified, showing that in many parts of the region the density of occupation must have been very considerable.

In the eastern part of the region – the Welsh Borderland – the presence of large hillforts suggests that the socio-political system may have been similar to other hillfort-dominated parts of central southern Britain, but in the far west, the Atlantic-facing communities inhabiting the peninsulas of Devon and Cornwall, south-west Wales and north-west Wales developed a very different system based on the single homestead without recourse to larger and more complex structures. Within the general category of ‘homestead’ there are a variety of forms, but all seem to have been of single or extended family size and it was the general rule that the settlement should be defined in some way, by a palisade, a bank, a ditch or these boundaries in combination. That some adopted multiple boundaries and some chose more prominent locations or constructed massive banks is most likely to be an expression of the different status of the lineages. There is nothing unreasonable in supposing that in the Iron Age, as we know to have been the case in the Saxon period, physical aspects of the settlement were closely related to grades of status. In such a system the social ranking of a family would have been displayed by readily identifiable features such as multiple lines of enclosure, height of bank, presence of gate tower, etc. Thus in the archaeology of settlement form may lie evidence of social structure.

Another factor which seems to be shared by all areas where good excavated evidence is available is the increase in the number and permanence of settlements in the Middle Iron Age. Once established, occupation usually continued for centuries, lasting in many areas well into the Roman period. This longevity of occupation argues for a degree of social and political stability, in marked contrast to the south-east where quite dramatic dislocations are evident. The persistence of established settlements along the Atlantic-facing zone from around 400–300 BC may, at least in part, be the result of the self-contained nature of each farmstead able, from the variety of resources within its reach, to provide for itself without recourse to external support. In this self-sufficiency may lie the reason why the material culture of the region is far more limited in its variety than that of the contemporary south-east.