The Atlantic Roundhouses

The Atlantic Roundhouses
Source: Armit (1998). The Archaeology of Skye and the Western Isles, Edinburgh.

Ever since the time of the first farmers, the Hebridean landscapes had been dominated by tombs and religious monuments. Although settlements were undoubtedly numerous they were physically dwarfed by the chambered tombs and stone circles. This emphasis changed dramatically during the first millennium BC with the construction of the massive Atlantic round- houses, among them the broch towers. Dun Carloway is the archetypal broch tower of popular perception: a massive, ragged ruin, isolated in the bleak, rocky peatlands of west Lewis. Even today it commands the land scape, towering over the later blackhouses and modern township (Figure 7.2). The Hebrides are, however, littered with the much less robbed-out remains of related and contemporary buildings that met with much harsher treatment both from the elements and at the hands of sub sequent generations.

The convenient collapse of one side of Dun Carloway exposes the inner structure and architectural complexity of the building and demonstrates some of the key attributes of this architectural tradition. The two concentric walls are dry stone-built, the space between them being bridged at intervals by circuits of large flat stones. These bridging slabs tied the walls together forming intra-mural galleries and stairs which enabled access between the floors and up towards the wallhead, the whole construction combining to create a tall and remarkably stable structure. Despite considerable variation in the ways in which these techniques were deployed, the principle remained simple. The hollow-walled construction minimised the weight of stone required whilst the tapering shape and bonding effect of the galleries helped to channel the weight stresses effectively to the base. Prom the outside this architectural complexity was cloaked in a forbidding drystone tower punctured only by one small and narrow entrance.

A range of other features of broch architecture are recurrent across the Atlantic regions of Scotland although they are not necessarily common to all sites. Ground level cells built into the walls and entered from the central area provide additional space in many broch towers; examples of these can be seen in the ground floor plans of Dun Carloway and the Loch na Berie broch tower. Scarcement ledges (protruding stone courses on the inner wall-face), at variable heights above floor level, played a part in supporting upper floors and roofs. Superimposed voids for ventilation of the intra-mural galleries are also to be found in some of the inner wall-faces of the broch towers, as at Dun Carloway.

MONUMENTAL HOUSES

It was once fashionable to view broch towers and related buildings as a peculiar architectural flourish in the remote northern outposts of the British Isles an odd phenomenon unrelated to events elsewhere. More recently sough they have been linked with a more widespread tradition of monumental domestic building which can be seen in much of Britain in the middle of the first millennium BC when substantial timber roundhouses character used many parts of the country. In the arable lowlands of southern and eastern Scotland, for example, numerous ring ditches recorded by aerial photography represent the ploughed-down remains of former timber round- houses of similar dimensions to the surviving stone examples in the north. J timber roundhouses of fairly elaborate construction were also found during the Iron Age in parts of England including Wessex. Generally, this phenomenon of monumental roundhouse-building was on the wane long before the end of the millennium. Along the Atlantic coasts of Scotland, however, monumental roundhouses achieved their finest expression after those elsewhere had largely disappeared.

Atlantic roundhouses are widespread throughout the Hebrides occupying most of the tolerably habitable areas, including many small islands long since abandoned for human settlement. The variable density of sites between different areas, for example the dense spreads over North Uist and Barra as opposed to rather less concentrated groups in Lewis and Harris and parts of Skye, may reflect, at least in part, the intensity of past surveys. Some areas, however, apparently have relatively few Atlantic roundhouses. This appears to be the case in South Uist and in the southern part of Skye. In the latter case the relative poverty of the land compared with the northern part of the island, coupled with a lack of intensive survey, may explain the paucity of sites.

The nature of the local landscape tends to dictate the locations favoured by roundhouse builders. In North Uist and Lewis, for example, islet locations are predictably common. In many cases these structures were simply built over older islet settlements. In Barra and Skye, by contrast, with fewer suitable lochs, rocky knolls and promontories were the favoured locations.

The Search for ‘True’ Brochs

Much effort has been spent over many decades in the analysis of the architectural minutiae of brochs in order to isolate a group of ‘true’ brochs. Once defined, the architecture, distribution and associated material culture of this select few could form the basis for the construction of theories on their invention and spread.

Extraordinary structures seemed to imply extraordinary events and in the climate of archaeological thought in the first half of the twentieth century this meant migration or invasion by new peoples. Prior to the 1970s there was a consensus that the brochs were built by incomers to Scotland, probably dis Possessed elites chased out of southern England at the time of the Roman invasion. Despite the lack of comparable stone architecture in the supposed source areas, these ideas seemed more attractive than the notion that such ill favoured island outposts could give rise to architecture of such mastery and solidity. Before the development of scientific dating methods, brochs could only he dated by reference to the few finds which could be matched and dated elsewhere. Inevitably the Roman period provided the only reliable benchmark and, since Roman material occurred on a scatter of broch sites, the brochs were generally placed in that period.

Excavations in Orkney in the late 1970s and early 1980s revolutionised the study of broch towers and related structures. Sites like Bu in Orkney showed that massive stone roundhouses, similar to the broch towers although lacking the architectural complexity, were being built in the Northern Isles from around 700 BC (Hedges 1985). This was far earlier than previously suspected and, equally importantly, it provided a viable ancestry for the broch towers which had previously seemed to have appeared fully-formed in the last centuries BC.

A combination of survey and excavation in the Western Isles during the l980s demonstrated that broch towers were part of a structural continuum and could not be abstracted from the wider context of the Atlantic round house group (Armit 1988, 1990, Harding and Armit 1990). The classification of individual sites as brochs had previously depended on their state of preservation since only positive evidence of the presence of upper floors would validate a structure as a ‘true’ broch. Since such survival of stone structures is extremely rare, the inevitable result was that very few structures were accepted as brochs, thus bolstering the argument that brochs were somehow different and special.

Their apparently sudden appearance along with the supposed architectural uniformity and purity of the brochs have been key props in the invasion hypothesis and without these there was no reason to question the indigenous development of broch architecture and the Atlantic roundhouse form. The broch towers were indeed extraordinary structures, but explanations for their origins and developments must be sought in the history of human settlement in northern and western Scotland, not in the adventures of down- at-heel expatriates from sunnier southern climes.

The term Atlantic roundhouse was introduced a few years ago to clarify the rather confusing typological morass which included brochs towers, galleried duns, semibrochs, island duns and an assortment of other variants (Armit 1990a, 199Gb, 1991). It reflects the basic unity of the range of stone built roundhouses in northern and western Scotland, many of which were previously condemned to the typological dustbin whilst the search proceeded for the origin of the brochs. Not all Atlantic roundhouses incorporate the traits of broch architecture. Indeed it is useful conceptually to separate the roundhouse form which relates to much wider traditions of house-building throughout prehistoric Britain, from broch architecture which was applied principally to roundhouses in Atlantic Scotland in the second half of the first millennium BC. Thus traits associated with broch architecture, such as intra mural stairs, cells and scarcements, can also he identified in structures which were clearly not domestic buildings such as some Hebridean promontory forts and enclosures. The promontory fort on Barra Head, for example, con tains two super-imposed galleries and a low entrance passage with bar-holes within an arc of walling cutting off the approach to a promontory (Armit 1992a, 94).

Within the overall class of Atlantic roundhouses we can identify a sub group of complex roundhouses where elements of architectural complexity can be seen. Within this complex roundhouse class we can identify a further sub-group, the broch towers, where the elements of broch architecture were used to construct a tall tower-like building. Due to the vagaries of preservaton, however, it is seldom possible to determine the height which most of

the complex roundhouses attained and it is probably impossible ever to determine the proportion of complex roundhouses that originally stood as broch towers.

The Atlantic roundhouse classification has been evolved to describe the field evidence in a way that recognises the limitations of the data. It need not imply any evolutionary scheme from simple to complex. Nonetheless, a review of the chronology of the Atlantic roundhouses shows that recent work is beginning to point to a gradual development of complexity from early simple Versions to the elaborate broch towers (Armit 1990e, 1991), although this does not necessarily mean that simple roundhouses ceased to be con structed.

These simple Atlantic roundhouses seem generally to have stood alone, devoid of associated domestic buildings or substantial enclosures. They were essentially farmhouses, housing perhaps a single extended family, and gave no signs of serious defensive provision. The early roundhouses did, however, mark a significant break from older traditions of settlement. The character istic Later Bronze Age settlement of the Northern Isles seems to have been the cellular complex, seen for example at Jarlshof in Shetland (Hamilton 1956). Still earlier Neolithic settlements such as Skara Brae and Rinyo in Orkney shared this cellular design. Atlantic roundhouses were the first domestic buildings in the area to mark out the settlement as a dominant landscape feature. Whilst earlier settlements, such as Skara Brae, were elaborate and highly structured internally, they would have been unprepossess Ing from the outside and certainly less prominent than contemporary chambered tombs. In the Iron Age though, the settlement itself, the centre of the domestic arena, became dominant.

‘l’here is no evidence so far of any simple Atlantic roundhouses in Skye or the Western Isles. In the fourteen structures which have been excavated to a sufficient degree for structural complexity to be recognised, all have been found to have incorporated intra-mural cells or galleries (Armit 1992). Although it is too early to be sure, it is possible that by the time that the first Atlantic roundhouses were being built in the Western Isles some of the elements of broch architecture had already been developed. 1’his might suggest that the initial development of the Atlantic roundhouses occurred in the Northern Isles, although there is really no reason to believe that simple Atlantic roundhouses necded to exist before the concept of complex Atlantic roundhouses could arise.

In Orkney Atlantic roundhouses appear to have preserved the cellular form of earlier settlements within the confines of their massive circular walls as, for example, at Bu. The roundhouse form seems to have been necessary 4 to create a massive and monumental structure, but the organisation of domestic life inside seems to have continued to follow the cellular patterns 4 set in earlier, Bronze Age dwellings. This might suggest that the concept of monumental domestic construction was adopted from elsewhere and grafted on to existing settlement organisation. By contrast, the Hebrides may have had a tradition of circular domestic building prior to the initial appearance of the Atlantic roundhouses. The hut circles of Skye and the southern Hebrides, and possibly even of the Western Isles, probably originated prior to the mid-first millennium BC. The Hebridean Atlantic roundhouses do not appear to have had cellular interiors but rather a central open space (although the paucity of excavation means that we may yet find evidence for internal timber partitioning). Nonetheless, present evidence, sparse as it is, suggests that Atlantic roundhouses may have begun to be built rather later in the west than in the north.

Ultimately though, the search for broch origins is futile. Such was the heterogeneity of the Atlantic roundhouse group and the extensive area across which they developed over more than five centuries that there is unlikely to have been any one moment or place of invention. What is more relevant to the study of the Atlantic roundhouse tradition is the question of why the diverse communities of northern and western Scotland should turn their attention for the first time to the elaboration in stone of what had previously been simple and often transient houses.

Complex Roundhouses

Some time around the fourth century BC the northern Scottish Atlantic roundhouses appear to have acquired more complex architectural charac teristics (Armit 1990e). Simple roundhouses, like Bu, generally built as isolated structures gave way to more complex structures, some with outer enclosures and ancillary buildings (Armit 1990a; 1990b). Crosskirk in Caithness exemplifies this transformation (Fairhurst 1984). Whilst appar ently never attaining any great height, this structure incorporated intra mural cells and stairs and was set in an enclosure with auxiliary buildings

Its internal space was broken down into a cellular pattern demonstrating its continuity with earlier domestic buildings in the north. Early phases of construction at the multi-period site of the Howe in Orkney hint at the emergence of a similar complex roundhouse there at broadly the same period (Carter et al., 1984). The analysis of the chronology of these complex roundhouses remains in its infancy but a pattern does seem to he emerging of a development of both architectural complexity and enclosure from around the fourth century BC in the north. It is perhaps these complex roundhouses of the far north, with their variable levels of architectural elaboration, that the majority of the Hebridean structures resemble most closely (Armit 1992a).

Dun Bharabhat, Cnip

Dun Bharabhat is one of a number of sites of various periods to have been excavated in the Bhaltos peninsula on the west coast of Lewis (Harding and Armit 1990). It lies within a landscape rich in archaeological remains of all periods from the Early Bronze Age onwards including the Cnip Headland cairns described in Chapter 6.

Prior to excavation, Dun Bharabhat appeared as a rather dismal pile of rubble protruding above the surface of a small loch in the hills above Bhaltos (Figure 7.5). The stony islet was linked to the shore of the loch by a treacherous and often submerged causeway. Its modest appearance was shared with numerous other Atlantic roundhouse sites throughout the Western Isles and in many ways Dun Bharabhat seemed as close as could be expected to a typical ‘island dun’ with no indications of any architectural complexity

Excavation quickly revealed a very different story (Harding and Armit 1990). The structure was found to have been built using the principles of broch architecture, with intra-mural galleries, stairs and cells (Figure 7.5). The presence of intra-mural stairs suggests that Dun Bharabhat would originally have had an upper storey but its small size (only some 11 m in diameter) and wide gallery entrances suggested that it could never have achieved tower-like proportions.

Dun Bharabhat appears to have collapsed soon after completion and was subsequently reoccupied in a distinctly less monumental form during the first and second centuries BC (Harding and Armit 1990). Material from immediately below the primary floor of the roundhouse dates to around the eighth century BC demonstrating the presence of pre-roundhouse settlement on this islet site in the earlier part of the millennium. Excavations by divers Working around the islet have confirmed that the excavated structure over- lies an accumulation of earlier settlement and it seems that its premature Collapse was in part due to its being founded on the unstable debris of this previous occupation. The nature of earlier settlement of the site cannot be ascertained without the removal of the surviving complex roundhouse but it is clear nonetheless that Dun Bharabhat represents a continuity of settlement from the earlier part of the first millennium BC or perhaps much earlier. Many islet settlements, spanning the millennia that separate the Neolithic islets like Ellean Domhnuill and Eilean an Tighe from the Atlantic round- houses, may lie in just such a situation, under the monumental ruins of their Iron Age successors.

The use of broch architectural techniques in a small, islet site of this kind has serious implications for the many other Atlantic roundhouse sites in the Hebrides. Many complex roundhouses were so irregular in plan that there can never have been any intention of building a tower-like form. The exigencies of the islet and promontory locations of a great many of these roundhouse sites suggest that they were never of tower-like proportions although all of the excavated examples where wall structure has been ex amined indicate the presence of structural traits associated with broch archi tecture. Another Dun Bharabhat, for example, this time on Great Bernera off Lewis, seems to fit this description (Figure 7.6). This structure in corporates a massive wall containing intra-mural galleries and stairs facing the causeway that leads out to the islet. A scarcement ledge can be traced along the rear of this upstanding stretch of wall. The shape of the islet, how ever, and the steeply sloping rock on which the remainder of the wall is perched seems to preclude the construction of a broch tower and it appears that the sector of walling facing away from the causeway may have been less massivc than the surviving frontal portion and the structure may thus have been somewhat irregular in shape. A number of sites previously classified as semibrochs, for example Dun Ardtreck and Dun Ringill in Skye, represent similar variants within the complex roundhouse class (although contra MacKie 1965 they were almost certainly wholly enclosed, roofed domestic buildings). Throughout the Hebrides elements of broch architecture appear to have been used simply because they were the architectural norm, or perhaps because their reference to broch towers was seen as transplanting some of the status of these buildings to the inhabitants.

In the later centuries H the Atlantic roundhouse tradition reached its apogee with the development of the broch towers. Although Dun Carloway is the best surviving Western Isles broeh tower, there are a number of others which would, in their day, have been at least as impressive. The Loch na Bent broch tower in Bhaltos, on the west coast of Lewis, was partially excavated by the Callanish Archaeological Research Project from 1985—1989 (Hard ing and Armit 1990) at the same time as excavations were in progress at Dun Bharabhat some 500 m away, and subsequent excavations have been carried out more recently (Harding, pets. comm.). Although the excavated deposits relate mostly to the later, Pictish period structures inside it (see Chapter 9), the broch tower is of considerable importance in its own right. Before exca vation, this structure was visible only as an unobtrusive mound in a sand- choked marsh (Figure 7.7). Local tradition held that the site was that of a broch but there was nothing visible to confirm this identification.

On excavation it transpired that this was indeed the site of a hroeh tower of quite exceptional preservation, the surrounding peat, and sand blown in from the coast, having swamped it almost completely. The Loch na Berie broch tower was built using the same architectural techniques as Dun Carlo- way and its proportions suggest that it would have stood at least as high. Once the upper parts had been excavated it became apparent that the present-day ground surface was in fact the first floor level in the broch tower, and that the first floor gallery was intact, with a set of steps leading up to the (now vanished) second floor (Figure 7.4). Two doorways opened from this first floor gallery into the central area, where a timber floor would once have rested on the projecting scarcement ledge. A further set of steps led down wards into the sludge which had filled the ground floor galleries. Through the gaps between the first floor gallery slabs could be seen the ground floor cells half-full of waterlogged debris. In the centre of the broch tower the debris of later occupation had gradually brought the floor level up to the modern ground surface.

The Loch na Benie broch tower clearly illustrates the futility of attempting detailed typological classification of Atlantic roundhouses from the surface evidence. Using the traditional models of classification the site would never have been accepted as a true broch prior to excavation, There is no reason to believe that other equally impressive structures do not remain to be recog nised throughout the Hebrides.

In Orkney certain broch towers, notably Gurness and Midhowe, formed the focus for sizeable settlements clustered around their bases. This suggests that, in the Northern Isles at least, the development of the broch towers was accompanied by the emergence of several larger centres of population. These broch villages were laid out in a highly structured way with the broch tower acting as the physical and spatial centre of the settlement (Foster 1989). The social dominance of the tower’s inhabitants was reinforced through the architecture of the village. In all aspects of their daily lives those living in the shadow of the tower would have tacitly acknowledged their sub servience. The restrictions of access to certain areas and constraints on the freedom of individuals to move around the complex form striking parallels with the much earlier religious architecture of the Neolithic. I ike the cham bered tombs before them, these broch towers enabled the routine social re lationships between members of the community to be constantly and silently reinforced in the normal business of daily life.

There is no sign, however, of this development in Skye or the Western Isles. Although some of the finest examples of broch architecture occur in the west, notably at Dun Carloway, Dun Troddan and Dun Telve (Figure 7.8), they remain isolated single-structure settlements. At Dun Colbost in Skye excavations in the enclosed area around the roundhouse have revealed traces of paving, drains and hearths, but no evidence of actual domestic buildings contemporary with the main structure (MacSween, pers. comm.). Excavations outwith the enclosed area produced no evidence of any activity. So although many Atlantic roundhouses had enclosures around them and possibly small external buildings, there is as yet no evidence for con temporary domestic structures. The architectural developments which enabled the construction of broch towers thus occurred in two very different contexts in Orkney and the Hebrides and we cannot, therefore, assume that their function and meaning would have been the same in both areas.

Traditionally the Atlantic roundhouses, and especially the broth towers, were seen as the defensive strongholds of a warrior aristocracy. Research in the past ten years, however, has shown that, while security was probably a consideration in the building of many broch towers, it was by no means the whole story and does not in itself provide an adequate explanation for the development of broch architecture.

Whilst their visual similarity to later castles and towcrhouses immediately suggests a military function, the defensive capacities of broch towers were actually fairly limited. While promontory forts of broadly similar date took advantage of natural topography to secure quite substantial arcas of ground in which stock and other possessions could he defended, broch towers

have provided limited space. The single, narrow entrance might have been easy to defend but it could perhaps equally easily have been blocked up or set alight by determined attackers intent on smoking or starving out the defenders. The timber roofs would have been highly vulnerable to fire, particularly at sites like Loch na Berie where the building was overlooked from a nearby rocky hillside.

In fact the main defensive qualities of most of the Atlantic roundhouses in the Western Isles derive from their siting rather than from the structures themselves. The numerous islet-sited structures would have provided easily controlled access via their narrow causeways and many appear to have had sufficient space for stock to be held to the rear of the islet. Dun Loch an Duna at Bragar in Lewis displays several traits which recur on islet-sited roundhouses (Figure 7.9). Three separate cross-walls, possibly relating to different phases of the site’s use, cut off the approach across the causeway and an enclosing wall encircles a large area of ground behind the round house. At other sites, such as Dun Loch an l)uin near Carloway in Lewis and un Thomaidh in North Uist, the causeway terminates abruptly some distance short of the islet, presumably indicating the former presence of a timber gangway that could be raised when necessary. On both of these sites th presence of the roundhouse seems almost incidental to the defensive capacities of the site. The control of access, the provision of space for stock and the separation of the vulnerable timber and thatch elements of the settle ment from potential attackers were all achieved principally by careful choice of location, on islets, promontories or other topographically favoured spots.

The situation in Skye, however, appears to have been rather different. Anne MacSween’s examination of the topography of the Skye roundhouses suggests that their builders often failed to maximise the dcfensive locationS of thc rocky knolls and promontories upon which most of the structures are located (1985, 13). Whilst this might, as MacSween suggests, indicate a re liance on the structures themselves for defence it is perhaps more likely tO suggest that defence was a relatively minor consideration in the design and positioning of most Atlantic roundhouses.

Whatever their defensive capacities, 1-lebridean Atlantic roundhou5 were essentially and routinely farmhouses. Excavations, principally Iii northern Scotland but also at sites like Dun Bharahhat, Cnip and Dun Vul in South Uist have demonstrated that Atlantic roundhouses were permanently occupied domestic buildings often associated with enclosures and ancillary structures necessary to the farming economy of their inhabitants.

Alan Braby’s cut-away reconstruction drawing of Dun Carloway in occu pation provides a starting point for a discussion of how a broch tower might have been used, based largely upon architecture and the combined evidence of numerous excavations albeit over a wide geographical area (Figure 7.10). Excavations at Dun Carloway itself however, have been restricted to the intra-mural cells where the excavator found evidence of intermittent occu pation spanning more than a millennium (Tabraham 1977).

It is certain that broch towers had multiple, superimposed timber floors. This is demonstrated very clearly at Loch na Berie where the intra-mural stairs lead to an entrance through the inner wall at the level of the scarcement ledge; a second entrance at the same level across the interior of the tower gave access to the first floor gallery and it could only have been reached by walk ing across the former timber floor. It appears then that at least some scarce ment ledges acted as supports for a timber floor. It is less clear, however, whether the absence of a scarcement means that there was no such floor and it remains difficult to estimate the number of potential floor levels in struc mres like Dun Carloway. If we could make a simplistic equation between scarcements and floors then the ground floor at Carloway would have been relatively low and cramped whilst the first floor would have a vast space above it reaching up to the timber roof. This interpretation is perhaps more likely than the multiple floor scenario favoured in the reconstruction draw ing.

Such a picture is consistent with research carried out recently by John Hope on the construction and use of broch towers. This research has con centrated principally on two exceptionally well-preserved broch towers situ ated only a few hundred yards apart in Glenelg on the mainland opposite Skye. Dun Troddan and Dun Telve (Figure 7.8) are amongst the tallest surviving examples in Scotland making their close proximity all the more surprising. Perhaps their remote setting in the narrow and economically rather unrewarding glen is the principal reason for their survival in this con drnon or perhaps they were always amongst the taller and most substantial of the broch towers. Principally on the basis of evidence from Dun Troddan, Hope has suggested that the ground floors of broch towers may have been less finely constructed and faced than the first floors. Together with their relative lack of height and recurrent presence of unlevelled rock outcrops Protruding from their floors, this suggests that they may have been used to house stock or for storage, whilst the first floor formed the main domestic 5 The height of the first floor, assuming there were no further upper floors would create an impressive domestic area on the inside, consistent with the visually striking exterior of the tower. There would probably have been a central hearth from which smoke would have escaped into the space above and ultimately seeped out through the thatch. Voids in the inner wall let smoke and air out into the intra-mural cavity keeping it dry.

There is no reason to believe that Atlantic roundhouses had anything other than a simple conical timber roof common to timber and stone roundhouses alike. This would have provided the most stable and practical roofing form and could have been supported on an upper scarecment or the inner wall- head, protected from the wind by the lip of the outer wall.

Our knowledge of the internal furniture of the Hebridean roundhouses is hampered by a lack of excavated primary floors. It is worth noting, however, that if the first floor was commonly the main domestic space in the broch towers then excavation will he of little help. However many Atlantic round- houses were not towers like Dun Carloway or Loch na Berie and some, like Dun Bharabhat, Cnip, had their domestic occupation fhcuscd on the ground floor. The internal area of Dun Bharabhat was very small (only some 5 m in diameter) and thus predictably there was little formal division of the interior which was dominated by a centrally placed hearth.

Domestic Activity

‘fl paucity of excavated Atlantic roundhouse interiors in the Hebrides restricts our knowledge of the arrangement and organisation of domestic activities. As in the Neolithic and Beaker periods, pottery dominates the assemblage of finds; ceramics continue to be well-made and often richly decorated in a recurring range of incised and applied motifs. Neutron activation analysis of pottery from various Iron Age sites in the Hebrides appears to demonstrate that most of this material was made locally and seldom, if ever, traded or transported over significant distances (Topping1986).

The profusion of decorated pottery from Hebridean Iron Age sites is in stark contrast to the situation in most of the rest of Scotland. Decorated pottery of similar styles does occur in the Northern Isles but it forms a much smaller proportion of the overall assemblages than in the west (Lane 1990). Further south, in lowland Scotland, the pottery record for the Iron Age is sparse in the extreme and vessels tend to be ruggedly functional and crudely made (Cool 1982).

Bone and antler would have remained important for a range of tools but these are rarely preserved in the acid soil conditions which pertain on most roundhouse sites. Fortunately the waterlogged deposits of the islet sites pro vides the opportunity for the recovery of other organic materials that do not normally survive on dryland sites. The deposits under the loch level around Dun Bharabhat for example have produced a small but tantalising assem blage of wooden artefacts including fragments of looms, opening up the potential for the analysis of an important and rarely seen aspect of Iron Age material culture (Harding, pers. comm.).

Despite the recovery of these important assemblages from recent exca vations, it unfortunately remains the case that the great bulk of finds from Hebridean Atlantic roundhouses derive from early excavations where the standards of recording were very low. At Dun Fiadhairt in Skye, for example, the finds included a necklace of amber beads, a steatite arm quernstones, rubbing stones and worked flint as well as a substantial and varied assem blage of decorated pottery. Undoubtedly the most unusual find was a hollow terracona object, apparently modelled to represent a bale of fleeces (CurIe 1932, 289). This object has been interpreted as one of the very few items of Roman manufacture found in the Hebrides and Curle speculated that it may have originated as a votive offering to the gods, carried north by some Mediterranean trader. Its context within the site would have been of some importance both chronologically and for the interpretation of its function. Although it is reported to have been found at a low level of the interior, close to the natural rock, it is, sadly, impossible to ascertain whether it was in ti-usive to that level. Early excavations at Dun Beag, also on Skye, yielded a comparable assemblage of finds with the addition of a glass armiet and beads, numerous bronze objects including pins and rings, and iron objects such as a tanged knife and possible spearheads. Taken as a group however, the finds from both Dun Beag and Dun Fiadhairt typify the mixed and long- lived assemblages which can he expected to accumulate in strucmres that would have been foci for settlement centuries after their primary use had ended.

Despite the architectural mastery displayed in their design and construction, Atlantic roundhouses were manifestly impractical and environmentally ill- adapted structures. Both earlier and later buildings in the Hebrides tended to share two key traits. First they were low, often achieving negative height by virtue of being dug into sand hills, middens or the ruins of old buildings. This characteristic is easily explicable in terms of the need for insulation and shelter from the perennial 1- winds. Secondly, they were con structed in such a way as to create a fairly minimal requirement for timber., Narrow sub-rectangular buildings or multi-cellular complexes could be roofed using small timbers; circular structures with substantial internal diameters require much more substantial lengths. Although broch towers represent a supreme achievement of drystone building it is important to remember the importance of timber in their design. Broch towers would have made serious demands on available timber supplies for their floors, internal fittings and roofs. Timber scaffolding would have been required for the construction of the upper parts of the larger towers, although the intra mural galleries may have provided some limited access. All of this would have added further pressure to what must surely have been a scarce resource. The builders of Atlantic roundhouses seem deliberately to have consumed scarce timber for the same basic reasons of prestige and display that caused their ancestors to sacrifice fine metalwork in bogs and pools and, even earlier, to quarry, haul and erect great megaliths for constructions of no practical use.

The recent excavations at Dun Vulan in South Uist have revealed the presence of a substantial midden within the enclosure around the main structure. This midden appears to have been formed of domestic floor sweepings and contains a mix of debris similar to that which might he expected within the roundhouse floor. What is unusual about this find, however, is that the rich and fertile midden dump has been allowed to accumulate around the settlement rather than being spread as feruliser acros the surrounding fields. The excavator of Dun Vulan, Mike Parker- Pearson, has suggested that the midden may have been a potent symbol of fertilitY and may in itself have constituted a status symbol (pers. comm.). Thus the presence of substantial unused midden around the settlement may have been another means by which the inhabitants displayed their affluence and power.

Several aspects of the Atlantic roundhouse tradition, then, suggest that its development was based around the desire to demonstrate the power of the inhabitants, even if the scale and nature of that power varied radically. Atlantic roundhouses developed in complexity over time, with successive innovations enabling the construction of ever taller and more massive struc tures. Although we regard many of the brochs as towers, we should re member that none of them is taller than its diameter; surely a pre-requisite for the use of the term. The impression of great height is achieved by the tapering of the walls; these structures were clearly built to impress.

If we accept that a large part of the reason for the construction of Atlantic roundhouses involved prestige and the display of power it still remains open to question which members of society were engaged in this process. Traditionally it had been thought that brochs were the homes of an elite group, analogous to the motte and bailey castles of the incoming Anglo Normans in the medieval period on the mainland. ‘l’his view was under standable in a context where only a select few structures were defined as ‘true’ brochs, the rest being sorry imitations or degenerate, late versions.

Study of the density and distributions of Atlantic roundhouses in certain parts of northern and western Scotland has led to a questioning of this traditional view. I’hc number of Atlantic roundhouses on Barra, for example, is far larger than could be sustained by any likely Iron Age population level if their occupation was restricted to the upper echelons of society (Figure 7.11, Armit 1988, 1992) because these buildings are spread fairly evenly across the most environmentally favoured parts of the island; a fact which suggests that they were all broadly contemporary. In North Uist there were many more Atlantic roundhouses than there were tacksmen in the post- medieval period. This suggests that, if population levels were broadly similar, Atlantic roundhouse construction extended far lower down the social ladder than the level of minor aristocracy and was probably more likely to represent something approximating to the level of tenant farmer. Such comparisons are of course very crude, hut they do demonstrate that, unless the Iron Age Populations of the Western Isles were vastly in excess of eighteenth century levels, Atlantic roundhouses cannot have been confined to the ruling elite. It is much more likely that they formed the standard unit of settlement throughout much of the region and were inhabited by a variety of social Prestige levels from tribal chief down to modest farming families. This interpretation fits the evidence for the tremendous range in scale and quality of the structures themselves, perhaps most clearly demonstrated in the Bhaltos peninsula of Lewis where Dun Bharabhat and the Loch na Berie broch tower lie only 500 m apart and represent almost polar opposites in terms of size and quality of construction.

Atlantic roundhouses in the Hebrides seem, then, to have been built to express the local dominance over land and resources of the individual com munity. In Orkney, as we have seen, a pattern developed of increasing

nucleation of settlement and centralisation of power which was expressed in the elaboration of the broch tower. Fewer but larger broch settlements cmerged as Orcadian elites became more powerful. In the west, though, power over people does not seem to have been expressed through monumental domestic architecture. Undoubtedly there were significant differences in status among communities within the islands, but these must have found other modes of expression. The Hebridean Atlantic roundhouses were built by individual small communities. Even the tiny and currently uninhabited islands south of Barra supported populations that imprinted themselves upon the land by the construction of their own roundhouses. The roundhouse expressed the self-sufficiency of the community, its control of its small pocket of land and its permanence in the face of a hostile environment. Thus there never developed the spatial hierarchies that can be seen so clearly in the arrangement of subsidiary buildings around the central broch towers of the north. Density of Atlantic roundhouses throughout the Hebrides strongly suggests that they represented a standard settlement form of the later centuries BC. MacSween, however, has identified two groups of structures in Skye — dun enclosures and promontory forts — that may relate to the later prehistoric period and have an uncertain chronological and functional relationship to the Atlantic roundhouses (1985). Promontory forts are also found widely in the Western Isles and numerous islet enclosures may parallel the dun enclosures of Skye. With the exception of a few cases, however, as at Barra Head Lighthouse, where these structures incorporate galleries or intra-mural cells, there is seldom any evidence to attribute even a broad date and there is little reason to assume that they represent any unitary class of monument, possibly incorporating medieval and Norse as well as perhaps pre-Iron Age sites amongst their number.

Recently an entirely new class of structure has been added to the roster of potential Atlantic roundhouse contemporaries, at least in Skye. Excavations by Roger Miket at the site of Tungadale revealed a substantial rectilinear building with an entrance in its short, east end (information from Roger Miket, Figure 7.12). This building was partially terraced into a hillside and formed of thick, stone-lined earthen walls. The spacious interior was dominated by a centrally placed hearth which had been replaced on several Occasions. The most intriguing feature was the presence of a narrow souter rain, or underground passage, stone-lined and lintelled, leading off from the interior within the thickness of the wall. In the later stages of occupation the floor level inside the building had built up and the entrance to the souterrain had been cleared out and revettcd with stone to allow the opening of a timber door. ‘ lowered entrance is important because it means that the souter rain entrance could not have been hidden by placing timber furniture or other obstacles in front of it. This appears to rule out one traditional in terpretatiori of such features: that they were for temporary refuge. Another more favoured explanation has been that souterrains were for the storage of foodstuffs. This, too, is unlikely in the case of Tungadale, as the souterrain was subject to water run-off from the hillside and required a drain in its floor to remove water. In this respect the Tungadale souterrain is unlike those larger semi-subterranean structures in eastern Scotland which date to a somewhat later period and were almost certainly used for the large-scale storage of grain.

Since the excavation, Miket has identified other structures associated with souterrains throughout Skye which may be of similar date and type. Radio carbon dates suggest that the Tungadale building dates to around the third century BC, in the period when Atlantic roundhouses were almost certainly still being built. The relationship between these two markedly different architectural traditions must form an important subject for future research in Skye.

The Atlantic roundhouse phenomenon has tended to am-act the attention of archaeologists away from other aspects of the Hebridean Iron Age. ‘ibis was, however, a time when important economic developments took place, not least of which was the arrival of iron itself 1’he adoption of iron, in prefer ence to bronze, as the favoured material for tools and weapons happened at different times across northern and western Europe in the first half of the first millennium BC. ‘I’he date of its first arrival in any area is notoriously difficult to assess because iron survives much less well on archaeological sites than either stone or bronze. In Orkney, for example, the first use of iron can be deduced by the disappearance of stone agricultural implements rather than by the appearance of actual iron implements.

The broch builders certainly used iron tools, although only rarely do these survive as anything more than depressing clusters of corrosion. Iron tools enabled more efficient agriculture and the raw materials were easier to obtain than those of bronze, although the latter material continued to be extensively worked for the production of decorative items. The change in metal-working technology, however, does not seem to have been as revolutionary as was once thought, when the introduction of iron was linked to the coming of new Celtic peoples with an accompanying package of cultural and social changes.

The period during which the Atlantic roundhouses originated appears to have been one of serious environmental stress throughout northern and western Scotland. The impact of climatic deterioration was felt, however, by prehistoric Scott farmers several centuries before the appearance of monumental architecture. Recent work near Lairg in Sutherland, for example, has suggested that Later Bronze Age farmers were highly vulnerable even to relatively minor environmental changes (Mac 1991), and that they retreated from the marginal uplands of Sutherland from the end of the second millennium BC and through the following millennium.

It is notoriously difficult to make generalisations on the nature and impact of environmental deterioration from one area to another, even within Atlantic Scotland. Nonetheless there arc indications from the settlement dis tributions in the Hebrides that a parallel process of settlement contraction may have occurred. The clearest evidence for this comes from North Uist where Atlantic roundhouse settlement is largely confined to the coastal belt, albeit off the machair, whilst apparently earlier, non-monumental settlement sites occur widely across the bleak interior of the island (Armit 1992). This pattern may rcflcct the combined effects of climatic change and human interference with the natural soils and vegetation cover conspiring to render much of the island useless for all but rough grazing and as a reservoir of peat. The coastal belt with its wider range of resources, terrestrial, marine and lacustrine, would thus have become the focus of island settlement as it has remained ever since. Such a process of settlement contraction would have serious social implications and, depending on its duration, may even have precipitated crises over the control of land and other resources. Such a back ground of land pressure and settlement dislocation may have promoted the concern with territoriality that led to the construction of the Atlantic round- houses as symbols of the local dominance and legitimacy of established Hebridean farming communities.

Despite these indications of the broad-scale background to Iron Age Hebri dean cconomies, it remains difficult to be specific about actual subsistence practices on Atlantic roundhouse settlements. The balance between arable and pastoral farming and the relative importance of fishing and fowling will have depended, as ever, on highly specific local conditions. The lack of substantial recent excavations on Hebridean Atlantic roundhouse sites together with the generally poor quality of bone preservation prevents any detailed discussion of the subsistence economy of these sites. 1’he material recovered from older excavations is generally too poorly provenanced to address modern research questions. A reasonable bone assemblage was recovered from Dun Cuier in Barra, for example (Young 1955), including red deer, grey seal, pony and otter as well as the expectcd range of domesticated cattle, sheep and pig. Bird bones comprised principally shag and cormorant whilst several species of fish were represented including wrasse, black bream, cod, ling and saithe. Reinterpretation of this site, however, has demonstrated that its occupation stretched from at least the later part of first millennium c when the roundhouse was built, until the immediately pre-Norse period when a secondary building housed the inhabitants within the ruins of the former structure. 1’he bone assemblage, as with other finds from the site, cannot be attributed to any particular period of occupation and may mask a range of chronological changes. Similar problems pertain to the quernstones found at many sites and it seldom possible to make any specific statements concerning the subsistence economies of any of the Hebridean roundhouse sites in their primary periods of occupation.

Much new information may, however, come from the post-excavation work on the material from Dun Vulan in South Uist. Most of this assemblage derives from the enclosure around the complex roundhouse where excavations were concentrated. Preliminary indications suggest that a substantial and representative assemblage of both animal and plant remains may repay detailed analysis. Initial reports suggest that sheep dominate the faunal assemblage, with a substantial component of cattle and an unexpectedly high proportion of pig (Parker-Pearson 1992). As might be expected, numerous species of bird and fish are also represented. Hulled barley seems to have been the dominant crop with some wheat, oats and possibly rye. The occurrence of carbonised grain at this and other sites, and the ubiquity of saddle and later rotary querns from Iron Age sites demonstrates the existence of an arable component within the economies of the islands.

At some time around the turn of the millennium, the people of the Western Isles stopped building Atlantic roundhouses. We do not know when the last broch tower was built or how long occupation continued in the unaltered towers. What is clear is that in the last century BC, and certainly from the first century AD, a new form of settlement was becoming dominant in the islands. This new structural form, the wheetl house, forms a sharp contrast to the broch towers, but it was still, in a different way, monumental and architecturally accomplished.