THE HEBRIDEAN IRON AGE :

TWENTY YEARS’ RESEARCH

by D. W. Harding

The Hebridean Iron Age: Twenty Years’ Research

This paper reviews progress in Atlantic Scottish Iron Age studies over the past twenty years, with particular reference to a long-term programme of fieldwork in west Lewis undertaken by the University of Edinburgh. It deprecates the survival and revival of older conventional models for defining and dating the major field monuments of the period and region in the face of accumulating evidence for the origins of Atlantic roundhouses in the mid-first millennium BC, and discusses important new evidence for the first-millennium AD sequence of occupation and material culture. The material assemblages of the Hebridean Iron Age are contrasted with the impoverished and relatively aceramic material culture of lowland Scotland and northern England, and the importance of the western seaways in later prehistoric and early historic times as a distinctive cultural region is emphasised.

Scotland U.K and The Outer Hebrides - Location Maps

Key sites in north-west Lewis - Location map

The Legacy in 1980

Atlantic Iron Age studies in 1980 were still heavily dependent upon the governing principals of the 1960s, resulting in classification based primarily upon structural or architectural typology, a compressed chronology for monumental buildings that allowed no antecedents before the first century BC, and a diffusionist framework that inhibited the recognition of local developments and tied innovation to the introduction of exotic imports.

The Atlantic Province had been defined by Piggott as part of his scheme for the Scottish Iron Age (1966), which was essentially an extension into Northern Britain of Hawkes’ 1959/61 scheme for the British (English and Welsh) Iron Age (1959). It shared with Hawkes’ scheme the physiographical determinism which made rivers and estuaries boundaries between regions, rather than arteries of communication, and highland areas watersheds rather than potential entities of coherent community groups. In reality, rivers and estuaries could have functioned in either role dynamically in different periods of prehistory and early history, so that the effect of an inflexible scheme is of variable significance.

The central highlands of Scotland, however, where in later prehistory and into early historic times settlement focused upon the major inland lochs and waterways (Henderson, 1998), manifesting itself conspicuously in the dense distribution of crannogs, was effectively disenfranchised by Piggott’s scheme. As far as Atlantic Scotland was concerned, the Northern Isles and the Hebrides were grouped together as a peripheral zone without regard for distinctive regional differences, and brochs themselves hardly rated consideration. MacKie’s study of 1965, long since superseded by more recent research but seminal at the time, was noted in a postscript but not otherwise taken into account in the text which had been written three years earlier.

The diffusionist model was fundamental to Piggott’s scheme, as it had been to others before him back to Childe. It reached its apogee in Robert Stevenson’s map in the same conference proceedings (1966, 23, Fig. 2), in which arrows bringing immigrants or mobile artefacts are reminiscent of maps produced by the hyper-diffusionists of the early twentieth century. Diffusionism as an explanation of culture change carried with it the corollary of time-lag, which had been compounded by Fox into his Highland/Lowland geographical model in his famous dictum that in the Lowlands new cultures are imposed, in the Highlands they are absorbed (Fox,1932/38, 34), the implication being that the more remote and peripheral the region, the longer it would take for cultural innovation to make its impact. Indeed, the present writer, appointed to a teaching post in the University of Durham in the mid-sixties, was advised by a well-intentioned senior colleague- himself a Roman specialist but an admirer of Fox- that northern England and Scotland were still effectively in the Bronze Age when the Roman armies arrived. Against this background, a compressed chronology for the Iron Age in Northern Britain was a self-fulfilling ordinance.

Local initiative or even progressive development over a longer period of time from local origins was never a strong contender against innovation by external impulse. So for example archaeologists attributed the shift from timber to stone construction of domestic Iron Age roundhouses in the Borders to Roman influence, and dated the transition accordingly, rather than admit the possibility that in regions where stone had been in use at various times and for various purposes since the Neolithic the natives might have been capable of such an innovation unaided if circumstances required.

In Atlantic Scotland, all three factors- diffusion, compressed chronology and classification by structural typology- combined to stultifying effect in the 1960s. Progress in broch studies at this time is inevitably linked to the name of Dr Euan MacKie, though at that time, as he himself acknowledged (1965a, 94) Mackie had not visited the Western Isles, doubtless assuming with the prevailing philosophy that the Outer Hebrides would be no more than a pale reflection of mainland traditions. In a series of now classic papers of the period (1965a; 1965b; 1969; 1971; 1974), MacKie updated the diffusionist approach of Childe and others, first attributing elements in the local Hebridean ‘Clettraval’ style of pottery to migrants from the Iron Age ‘B’ ‘Woodbury culture’ of southern England whose arrival, as late as the first century BC, he saw as a catalyst for broch building, and subsequently identifying the migrants more specifically as refugees displaced by the Belgic invasion of south-eastern England reported by Caesar. Like other examples of the ‘historical catalyst’ explanation (another, equally misplaced, was Wheeler’s introduction of multivallation by Venetic refugees after 56 BC), it required not just a belief in long distance connections, which is perfectly plausible, but demanded that those connections should be of a specific kind (settlement of a refugee population) at a specific time (in the first century BC) and from a specific place (south-east England) (MacKie, 1974, 102). In so far as there is evidence of connections between Atlantic Scotland and Southern Britain, they certainly do not demand such an interpretation. Indeed the ceramic parallels alone point to several different regions of southern England over a span of several hundreds of years, and are certainly not a consistent series of types from the same area and period of time, as might be expected of a genuine population movement. It therefore does not require a isolationist perspective or a committed anti-diffusionist (one thing the present writer has never been accused of) to question the plausibility of attributing the appearance of brochs to refugees from an invasion, the authenticity or circumstances of which many scholars now question, and from a region where there is no tradition of domestic stone building, let alone on the monumental scale of the brochs.

The diffusionist principle was invoked not only to account for the appearance of brochs, but also for the spread of their distribution across Scotland. MacKie favoured a west-to-north expansion, based upon details of architecture; Caulfield pointed to the evidence of quern replacement to suggest that the northern brochs might be the earlier group (1978), as indeed the density of distribution in Caithness and Orkney might have argued. Echoing the contemporary debate about Beaker movements, MacKie even had a ‘reflux’ from north back to west to account for structural characteristics of the Glenelg brochs.

All of this might best have been consigned to the archaeological archives, were it not for the fact that in recent years the sideburns and flares of nineteen-sixties and seventies archaeology have been resurrected and dusted down as if they were what serious students of Scottish Iron Age studies might still be wearing (MacKie, 1995, 661-4). In fact, the same basic hypotheses still largely underpin MacKie’s retention of the late dating of brochs, in combination with the selective use of radiocarbon dates (Topping, 1987, 72-5 and Lane, 1990, 114-15 on Dun Mor, Vaul. Reply to this and all other criticism of his interpretation of the site by MacKie, 1997, 141-80; Harding, 1984, 211 on Rhiroy. Reply by MacKie, 1991, 175-6). The late dating of ‘true brochs’ has been re-asserted without its diffusionist framework by more recent studies based on fieldwork in the southern islands (Parker Pearson and Sharples, 1999).

Research since 1980, notably in the Northern Isles, but also that promoted by the University of Edinburgh in the Western Isles since 1980 (for location see Fig.1), has contributed to a re-appraisal of the origins and development of monumental Iron Age roundhouses in Atlantic Scotland, demonstrating that broch towers have their origins in a sequence of massive roundhouses, simple or complex, dating from the mid-first millennium BC. To extract the developed broch towers from that context and to dismiss any structure which can be shown to pre-date the first century BC on the grounds that it is not a ‘true broch’ is to create artificial definitions and distinctions which cannot contribute to a balanced view of later prehistoric settlement. Furthermore, it divorces ‘brochs’ from other groups of stone-built roundhouses whose similarities have too frequently been overlooked or suppressed in an attempt to create distinctive structural typologies. As Parker Pearson and Sharples rightly observed, ‘the diversity of stone-built dwellings almost defeats the classification of brochs as a meaningful group’ (1999, 364). At opposite ends of the spectrum, a simple, ungalleried roundhouse and a broch tower may justifiably be seen as poles apart; but they are just that, extremes in a spectrum in which the contrasts of opposites is not the whole or only truth. In practical terms the rigid typological definition of a ‘true broch’ was of little utility to the fieldworker, who was unlikely to be able to tell from a collapsed heap of stonework whether any individual structure displayed the required repertory of architectural features anyway, so that only a handful of the hundreds of known field monuments was ever likely to qualify.

Fig. 2 Atlantic roundhouses, simple and complex: Calf of Eday, Orkney (after RCAHMS, 1946), Howe, Orkney, phase 5 (after Smith, 1994), Rahoy, Argyll (after Childe and Thorneycroft, 1938), Dun Troddan, Glenelg, Lochalsh (after Curle, 1921)

The Structural Sequence; before the brochs and after.

Progress towards a new assessment of brochs and related structures followed the excavations in Orkney of the adjacent sites of Bu and Howe (Figs 2, 3), where radiocarbon dates indicated the existence of ‘proto-brochs’, or at least monumental roundhouses which could legitimately be regarded as antecedents of developed brochs, from the middle of the first millennium BC. To avoid the sterile circular arguments about what constituted a ‘true broch’, Armit (1990; 1992) devised the rather unwieldy phrase ‘complex Atlantic roundhouse’ to cover the range of galleried structures, including developed broch towers and less typologically-correct galleried duns which formed the core of his class of monumental, stone-built structures of circular or sub-circular plan, which characterised the Atlantic north and west. In the Northern Isles these buildings could have developed from the ungalleried, or ‘simple’ (though still substantially-built) roundhouse tradition exemplified by Quanterness or Calf of Eday (Fig. 2), but earlier than the first half of the first millennium BC there appeared and still appears to be a genuine second millennium Bronze Age hiatus. There is another parallel variant to the simple roundhouse in the Northern Isles, best exemplified in the Jarlshof and Clickhimin sequences, in which cellular, or courtyard houses of sub-circular plan, at Jarlshof (Fig. 3) associated with numerous fragments of moulds for diagnostic late Bronze Age artefacts, apparently precede the appearance of stone forts and complex Atlantic roundhouses. The projecting piers which divide peripheral cells from central court seem almost to anticipate the division of space in the later wheelhouses of Shetland and the Western Isles (Fig. 3), even though one might not go so far as to claim them as late Bronze Age prototypes of wheelhouses in terms of structural typology.

Fig. 3 Structures with radial division of internal space: Jarlshof, Shetland ( after Hamilton, 1956), Clettraval, North Uist ( after Scott, 1948), Kilphedir, South Uist ( after Lethbridge, 1952) Bu, Orkney ( after Hedges, 1987), Howe, Orkney, Early Phase 7 ( after Smith, 1994) (Drawings D.W.Harding)

In the Western Isles no equivalent to the late Bronze Age courtyard houses of Orkney and Shetland has yet been identified, and though roundhouses of various kinds dating to the later Bronze Age or earlier Iron Age have been found in the southern islands (Parker Pearson and Sharples, 1999, 364; Marshall et al, 1999) there is still no confirmation that such a thing as a ‘simple’ roundhouse of monumental proportions like those of the Northern Isles existed in the Western Isles.

Dun Bharabhat at Cnip was believed to be such, on the basis of the Inventory description of 1914/28 (RCAHMS, 1928), when selected for excavation, but on the first afternoon of excavation removal of superficial debris on the site had exposed intra-mural chambers (Fig. 4).

Fig. 4 Dun Bharabhat, Cnip, Isle of Lewis; site plan

Dun Bharabhat proved to have several of the structural features of a traditional broch, including galleries, scarcement, intra-mural staircase to an upper floor, and door-rebate, bar-holes, sill-stone and pivot-stone in its entrance passage, but its overall dimensions at ten by eleven metres was smaller than average and its relatively modest proportions would probably have disqualified it from acceptance as a ‘true broch’.

Its island location, with entrance and annexe or forecourt diametrically opposed to the causeway which gave access to the near shore, was replicated elsewhere in the Western Isles, most obviously at the nearby Loch na Beirgh, Riof (Fig. 5), where the original structure would certainly be accepted conventionally as a broch tower. The end of the occupation of the complex Atlantic roundhouse at Bharabhat was brought about by the collapse through subsidence of the north-east sector of its walls. Its shell was then re-used for a secondary roundhouse in which access to the causeway was gained through the widened entrance to gallery 2 and over the dismantled outer wall.

The dating of the complex Atlantic roundhouse at Bharabhat to the second half of the first millennium BC is admittedly based upon just three radiocarbon dates (though they do come from unequivocally significant contexts), one primary and two from probable roof timbers of its secondary phase, placing its main occupation firmly in the second half of the first millennium BC. The associated material assemblage would be consistent with such an early horizon, based upon the emerging evidence of the sequence of material culture from the Bhaltos peninsula and comparanda elsewhere. The terminus post quem for the occupation of the complex Atlantic roundhouse (GU 2436: 2550+/-50bp) coincides with a phase of landscape clearance recognised independently by Kevin Edwards from coring in the loch. The other two dates resulted from an episode of destruction of the secondary occupation of the main building, in which roof timbers were destroyed by fire; these dates (GU 2434: 2010+/-50bp and GU 2435: 2100+/-50bp) presumably therefore indicate a terminus post quem for that secondary occupation. The debris was subsequently levelled in a late secondary phase when the role of the main building and annexe appears to have been reversed, the former serving as a yard through which access was gained to the causeway, while at least one surviving external cell, which combined edge-set slabs with roughly-constructed horizontal coursing in the post-monumental fashion at Beirgh, was constructed against the collapsed stonework in the annexe. This external cell links up with a superimposed sequence of stone structures now under water (of which the lowest unfortunately could not be investigated for reasons of safety) (Dixon, in Harding and Dixon forthcoming). The Bharabhat occupation can be summarised as a sequence in which the complex Atlantic roundhouse is followed by a simple roundhouse which utilised access to one or more pre-existing cells, followed by a latest phase characterised by cellular construction, in effect the same essential sequence which is demonstrated even more clearly at nearby Beirgh.

Fig. 5 Loch na Beirgh, Riof, Isle of Lewis; site plan

The primary levels of the complex Atlantic roundhouse (see Fig. 6 for simplified sequence) have thus yet to be fully examined; only those at first floor gallery level have been accessible, and they almost certainly have been disturbed by later occupation. The waterlogged state of the whole of the ground floor of the structure leaves open a very real prospect that internal furnishings and fittings could survive intact, if the state of organic preservation of the earlier post-Atlantic roundhouse levels is any indication.

Fig. 6 Loch na Beirgh, Riof, Isle of Lewis; simplified representative sequence of structural phases

Table 1: Loch na Beirgh, Riof, Isle of Lewis. Radiocarbon sample details - errors are expressed at 1 sigma

Table 2: Loch na Beirgh, Riof, Isle of Lewis. Calibrated radiocarbon dates

The next major phase of occupation at Beirgh was indeed the Roundhouse which occupied almost the entire area enclosed by the inner wall of the complex Atlantic roundhouse, much as it did at Dun Mor, Vaul, on Tiree. Some, at any rate, of the intra-mural galleries were re-used in this and possibly successive phases, until the level of the occupation surface rose to a height where access would have been too difficult through the trans-mural passages. It has been argued by Gilmour (2000) that the upper walls and floors of the Atlantic roundhouse were deliberately dismantled, not just for reasons of safety as has been suggested elsewhere, but as a conscious and symbolic reduction of the monumental scale of the building.

In practice a Roundhouse occupying the former interior of the Atlantic roundhouse, but retaining access to its galleries and cells at ground floor level, would have had a distribution of space, central and peripheral, not unlike that of a wheelhouse, and the introduction of radial piers would only have been necessary to retain that formula once access to the intra-mural galleries became impossible. This is exactly what appears to have happened from the latest structural alterations to the Roundhouse.

The anomaly in wheelhouse distribution, most dense in the Western Isles, present in Shetland but apparently absent altogether from Orkney, has often been remarked. Yet the dilemma becomes less acute if sites are compared in terms of social or functional use of space, rather than in terms of architectural typology. Some Atlantic roundhouses or related structures, especially in Orkney where wheelhouses are absent, have internal radial divisions of space, like Bu (Hedges, 1987), in which the hearth significantly forms the focus of the central area, and in at least one phase of the Howe Atlantic roundhouse (Smith, 1994) (Fig. 3). Whether this represents a difference between communal and private space or whether the peripheral compartments are functionally-specific is a matter for speculation, but the circular ground-plan essentially lends itself to this division, in the timber roundhouses of Southern Britain no less than in their stone-built counterparts in the north and west. The origins of the central/peripheral space pattern is equally unclear, but it could certainly be detected in the courtyard arrangement of the late Bronze Age phases at Jarlshof (Fig. 3) and Clickhimin. Though late Bronze Age origins have sometimes been claimed for wheelhouses, it is not necessary for the specific architectural form to have developed in order to sustain the social use of space which it represents. It could be argued (Gilmour, pers. comm.) that the ‘broch villages’ of Orkney represent an analogous form of centre-periphery division of space, but on a much grander scale.

This interpretation of the use of internal space, of course, stands in contrast to the more recent interpretation of the use of internal space in terms of rotational cosmology (Parker Pearson and Sharples, 1999 with earlier references), in which the orientation of roundhouse entrances between south and east is seen to be specially significant. A survey of complex Atlantic roundhouses had shown that the dominant orientation was either east or west (Parker Pearson et al, 1996). For the record, the orientation of the original entrance at Bharabhat is east-south-east; at Beirgh it faces east, in each case diametrically opposed to the direction of access from the shore. The location of the entrance at 180 degrees from the causeway is a recurring pattern of island-based Atlantic roundhouses in Lewis (though not obviously so in the southern islands), as a result of which the orientation of the entrance will be primarily determined by local topography, in effect, by the site’s relation to the nearest shore. Exceptions like Loch an Duna at Bragar, where the south-east facing entrance is oriented at c. 120 degrees from the causeway, could therefore represent a deliberate and significant departure from the norm. If, however, as at Bharabhat, causeway access to the original entrance has subsequently been blocked by collapsed masonry, then a secondary entrance may have to be aligned directly facing the causeway, making accurate reading of derelict surface remains more difficult, and by implication indicating that entrance alignment was no longer a prime consideration.

The catalyst for the development of the wheelhouse as a type is usually regarded as an increasing shortage, in the Western Isles especially, of timber in sufficient quantity and of sufficient length to span a roundhouse roof, a problem that could be alleviated by the introduction of radial piers, truncated so as not to eliminate altogether the central space, from which the rafters of the roof cone could spring. The solution has attendant problems of its own, not least the introduction of open-sided corbelled cells and the effective integration of corbelled peripheral cells with the timber and thatched cone of the central roof, so that there are grounds for believing that conscious segregation of space was as important a consideration as availability of building materials.

One distinction between complex Atlantic roundhouses on the one hand and wheelhouses or other forms of post-monumental structures on the other is that the former are free-standing, whereas many of the latter, including those in the Beirgh Pictish-period sequence, have only a single-faced wall revetted into rubble debris from previous occupation, or, in the case of wheelhouses, typically into machair sand.

Cnip wheelhouse: surviving corbelled cell

The stages in building a wheelhouse were examined and reconstructed by Armit at Cnip, and it is clear that the architecture of the building, and some of its characteristic features like the progressive widening of stonework of the single-faced external wall, were carefully designed to suit its foundation in a subterranean, cylindrical pit, which presumably also facilitated the addition at ground level of backing material for the corbelled cell roofs.

Until recently, only two examples had been excavated of free-standing wheelhouses, at Clettraval, North Uist (Fig. 3) (apparently abutting on its east side the edge of an earlier chambered cairn) and Tigh Talhamanta, Barra; ironically these two anomalies are the only examples illustrated in the standard student text on the British Iron Age (Cunliffe, 1991, Fig. 13, 46). Apart from their radial piers, these are really simple Atlantic roundhouses, further reinforcing the point that we are dealing not with clearly-defined categories of building but with a spectrum of variants from the simplest to the most complex architecturally, in which the presumption that the simplest are early and the most complex are late need not be invariably true. More recent excavations include the aisled roundhouse at Allt Chrisal, Tangaval, Barra (Foster, 1995; Sheffield, 1999).

At Beirgh a significant change in the social order appears to be indicated by the shift to Cellular settlement around the third or fourth centuries AD (Fig.6).

Once again, regrettably few radiocarbon dates are available, though the series such as it is conforms remarkably well with the stratigraphic sequence (Tables 1,2), indicating a span for the Cellular occupation at two sigma calibration between the early third and early seventh centuries AD. Recognising that the maximum size of the buildings is in most cases determined by the practical constraints of their corbelled roofs, there appears to have been a process of fragmentation into smaller units, as if the peripheral cells of previous patterns have broken free and regrouped within the enclosure. It is not clear whether this indicates a change in function or social unit which occupied the site.

The Cellular Phase also displays evidence for metalworking, but this in itself need not indicate a radical change. Not all the buildings of the Cellular Phase are the same. Some evidently were built largely in timber; others may have used turf as part of their construction. But the majority was certainly built in stone, with a characteristic combination of vertical slabbing, revetted against rubble, and horizontal drystone coursing. Some roofs were certainly corbelled, but partial corbelling, involving the progressive inward overlapping of several courses before the roof is braced and completed with a timber and thatch canopy, is also indicated by surviving sections of coursing. Cellular construction in itself, or its characteristic building techniques of edge-set slabs and horizontal coursing, with corbelled or partially-corbelled roof, cannot be regarded as diagnostic of any specific period, since shielings from Medieval to early modern times in the Western Isles used just these techniques. Conversely, of course, there is the very real possibility that genuinely ancient sites have been overlooked on the presumption that they were early modern shielings. In fact, more recent work by Gilmour, Church, Burgess and others around Aird Uig has suggested that this basic form of structure may have an ancestry extending back through several millennia in the same locality.

The Final Pictish-period occupation at Beirgh is also cellular in essence, but on a scale which once again occupies the whole interior of the former Atlantic roundhouse, comprising a principal room and two subsidiary cells in what is fundamentally a figure-of-eight configuration.

Fig. 7 Later Pictish-period figure-of-eight structures; Beirgh, Riof, Isle of Lewis; and Bostadh, Great Bernera, Isle of Lewis (after Neighbour and Burgess, 1996)

Beirgh: Figure of eight house partly excavated, showing vertical slabbing and horizontal coursing

It is not absolutely clear whether either of these smaller cells could have been corbelled in their entirety, thus affording yet another permutation and re-ordering of the wheelhouse theme of principal room with timber and thatch roof and subordinate cells with corbelled roofs. It may equally be that structures of this general class were covered by a timber and thatch roof of hipped construction, as has been inferred for the reconstruction on Great Bernera, based upon the excavations at Bostadh by the University of Edinburgh Department of Archaeology and its Centre for Field Archaeology (Neighbour and Burgess, 1996). The evidence for partial corbelling at Beirgh and the presence of a third, guard cell, however, argues against this form of overarching roof in this instance. The chronological concurrence of the Final Pictish phase at Beirgh and the Bostadh figure-of-eight buildings (Fig. 7) certainly supports Gilmour’s contention (2000) that this was a late expression of the cellular mode of construction. No radiocarbon dates are available for the later Pictish-period occupation at Beirgh, but distinctive material types, like penannular brooches, composite bone combs and pipe-bowl crucibles, indicate a span within the seventh to ninth centuries AD. Unlike Bostadh or some sites in the Northern Isles, there was no evidence of Norse occupation at Beirgh, though Norse burials have been recorded in the immediate vicinity.

Dun Bharabhat, the Cnip wheelhouse and the Beirgh sequence together provide a convincing sequence from early and developed monumental complex Atlantic roundhouses, through an immediate post-monumental phase represented by a less substantial roundhouse or wheelhouse to a period of structural fragmentation when small, but clustered cellular buildings are the norm, and finally to the more formal figure-of-eight buildings which characterised the later Pictish occupation of the Western and Northern Isles alike. Whether this sequence will prove to be applicable to other regions of the Atlantic west remains to be seen, but recent research has certainly taken Hebridean Iron Age studies well beyond the brochs, chronologically in both directions. There are certainly points of correlation between the Bhaltos sequence and that more recently excavated at Dun Vulan in South Uist, notably in the introduction of a secondary cellular occupation and in the broad progression of pottery styles. A point of digression is the first century BC date for the construction of Dun Vulan, claimed by the excavators on the basis of two radiocarbon dates from contexts which in the final report, if not previously, are recognised as less than definitive, an issue that might have been placed beyond doubt if the primary occupation within the building had been sampled. From the report it is fairly clear that there could be several intermediate phases of occupation between the cellular structure exposed in the interior of Dun Vulan and the primary occupation below, the actual date of which remains to be confirmed (for review, see Armit, 2000). In the event, Dun Vulan may have been constructed in the first century BC, but even if this proved to be the case it hardly seems a basis for insisting that all complex Atlantic roundhouses belong to such a late horizon, which no longer has any compelling significance once the old diffusionist myth of refugees from Belgic invaders catalysing the development of ‘true brochs’ has been dispelled.

One form of construction which is apparently not represented in the Beirgh sequence is the rectangular plan, which elsewhere in Atlantic Scotland, as for example on the Kintyre stack site of Dun Fhinn, makes its appearance in the early historic period. At Dun Vulan rectangular buildings apparently date from the second century AD, but are not accorded a function as dwellings.

Bostadh: Recnstruction of later period Pictish period house

Taking the Bostadh model, it is possible that a shift towards rectangularity in superstructure may not always mirror itself in ground-plans, but with that qualification, the introduction of rectangular building should reflect a more fundamental shift in the use of internal space than such a progressive transition might imply.

By way of a postscript to this survey of the post-Atlantic roundhouse structural sequence in west Lewis, we might consider what form of building should be anticipated as evidence of the early Christian settlement of the Western Isles (Harding, forthcoming,a). Notwithstanding the reasonable presumption that the islands came under the influence of Christianity in the century or so following the Columban mission to Iona of AD 563, very few ecclesiastical sites have been identified which can be dated earlier than the fourteenth century or thereabouts. The probability is that the earliest churches shifted their locations, if only by a few hundred yards, as the Christian communities expanded, so that the earliest foundations may survive only as very residual outlines. But some of the earliest were doubtless more remote cells, whose structures would be little different from the small, cellular constructions of their secular neighbours, and whose religious identity, if not suggested by a monastic layout within an enclosure, would only be proclaimed by the presence of a cross inscription or the like. It remains tempting in the absence of tangible evidence to think that a structure like the small, square cell with its pair of aumbries, built in the debris of the collapsed dun wall at Bharabhat, and overlooking the island still known as Pabbay Mor, might have had such associations.

Material Culture of the Atlantic Iron Age

Apart from its monumental architecture, the most striking aspect of the Atlantic Iron Age is the relative wealth of its material culture. By contrast with lowland Scotland, where later prehistoric material culture is not only virtually aceramic but also impoverished in terms of other surviving material types, or indeed with Britain from the Trent northwards, where pottery is not widely represented before the late Iron Age, Atlantic Scotland has a distinctive ceramic repertory and is reasonably well-represented in terms of other domestic artefacts. The fact that the ceramic tradition in the Northern Isles is not the same as in the West testifies to regional autonomy of culture, but they stand together in contrast to the rest of mainland Scotland.

What then accounts for the virtual absence of pottery and the paucity of material artefacts through much of northern England and Scotland, and why has it excited so little comment from archaeologists? Forty years ago it would readily have been explained as a concomitant of a pastoral economy. Piggott’s ‘Celtic cow-boys and shepherds, footloose and unpredictable, moving with their animals over rough pasture and moorland’ (Piggott, 1958, 25) would have had no place for pottery, vulnerable to breakage among semi-mobile communities, who doubtless would have been expected to use leather or wood instead. The problems with this view were several, and have been recognised for many years. First, considerable tracts of northern England and Scotland are not just rough pasture and moorland, but agricultural land of good quality in prehistoric times as today. Second, in these areas, and indeed in the Border uplands, there is ample evidence of Iron Age settlement with permanent large roundhouses on no less a scale than those of Wessex, not infrequently displaying episodes of rebuilding which suggests stability and longevity of occupation. Piggott’s contrasting economies have long been regarded as an oversimplification to the point of distortion, and whilst seasonal transhumance was doubtless a component of Iron Age society in northern Britain (as indeed it was on a local scale in other parts of Britain), there can be no doubt that communities in most regions of Scotland, particularly in the Atlantic North and West, were basically sedentary. As far as material culture is concerned, we know from underwater excavation in central Scotland and in the Hebrides that other materials such as wood, leather or wickerwork were used, both in ceramic and aceramic communities, so that the presence or absence of these does not require or explain the absence of pottery.

It is a fact of insular Iron Age pottery that generally goes unremarked that it is very limited in the range of types, and rather restricted in its range of vessel size. Until the introduction of wheel-thrown wares classification seldom demands division of form beyond ‘jar’ and ‘bowl’, and by comparison with other cultures or periods the size-range hardly extends beyond small or medium. These are presumably factors of production, the result with some exceptions of local and domestic, rather than professional or industrial manufacture, so that the observation need occasion no great excitement. Only with the introduction into Britain of wheel-thrown pottery at the end of the southern British Iron Age do we see vessels designed for a specific purpose, such as platters or cups, for example, and even then the range is limited by comparison with Mediterranean or Middle Eastern assemblages. The limitations of size, quality and design of Iron Age pottery, however, must have limited the utility of the medium. Large scale storage would simply not have been practicable in these pottery vessels, and given the quality of fabric, storage of liquids must have been less than satisfactory. In fact, pottery would probably not have been nearly the all-purpose utility artefact that we sometimes imagine it to be, hence the use of alternatives which, because of the lower survival rate of examples, we too frequently relegate to minor supplementary roles.

Fig. 8 Pottery from Dun Bharabhat, Cnip, Isle of Lewis

The limitations of Iron Age pottery cannot simply be attributed to ignorance or lack of skill on the part of the potter (though we might be tempted to explain the wilful crudity of Dunagoil ware that way), because alongside rather poor quality vessels in the Western Isles are much better-quality wares with walls often almost eggshell-thin and frequently decorated with very fine incised designs (Fig.8). Twisting the surviving fragments under a spotlight in the study to catch the ornament, one wonders what the purpose of such decoration could have been within a windowless Atlantic roundhouse, but perhaps the point was that it was there rather than that it was visible. It could have been enhanced by infilling with a contrasting paste, as deep-scored haematite-coated wares in Wessex commonly were, though no such infilling seems to survive. Unlike Wessex, however, a suitable paste is not readily derived from the local geology of the Western Isles, and crushed shell would still be rather coarse. An obvious locally-available substance would be seabird guano, which would surely wash away after the pottery was deposited in the ground. Apart from the possibility of infilling of incised ornament, simple painting of pottery should not be discounted. Once again a surface wash applied after firing might not be expected to survive under normal circumstances, but odd examples of what appear to be crudely-painted stripes on some vessels from the underwater excavations at Dun Bharabhat, perhaps baked hard and thus more durable through use by the hearth, could well be indicative of a more widespread fashion.

Fig. 9 Pottery from Beirgh, Riof, Isle of Lewis; post-complex Atlantic roundhouse, pre-Cellular phases

Archaeological interest in Hebridean pottery has hitherto focused less upon its social or economic function than upon its potential as a means of dating an assemblage or identifying its cultural context. Establishing a usable sequence for the later prehistoric and early historic periods has been an objective since Sir Lindsay Scott’s pioneer work (1948), which was later developed by Young (1966) and MacKie (1974). Scott’s analysis unsurprisingly for its time was predicated upon a diffusionist model, but it did hint at a sequence in which applied ornament outlived incised ornament on Iron Age pottery, with plain wares coming at the end of the sequence. Young too recognised that everted-rim vessels, decorated with cordons and incised decoration and sometimes with internal fluting of the rim gave way to plain wares in the later Iron Age, though her identification of the diagnostic types of the earliest Iron Age was less certain. MacKie, on the basis of the stratigraphic sequence at Dun Mor Vaul, identified ‘Dunagoil ware’, ‘Vaul ware’ together with Balevullin vases and internally-decorated bases as potentially early. Everted-rim ware, with or without cordons, and ‘Clettraval ware’, with cordons and arcaded ornament, were recognised as essentially middle Iron Age, with ‘degenerate Clettraval ware’ and ‘Dun Cuier ware’ being assigned to the period of wheelhouses in the third and fourth centuries AD. MacKie’s analysis came in for severe criticism from Lane (1990, 111ff), both on methodological grounds and for its dependence upon the site’s controversial chronology, but more recent work may nonetheless vindicate the basic sequence which it proposed. The interpretation of that sequence and elements within it was, however, still determined by the model of one-way diffusionism which at times strained credibility and placed undue constraints on chronology.

Topping’s study of Hebridean Iron Age pottery using neutron activation analysis (1986) demonstrated fairly conclusively, and not altogether surprisingly, that it was consistently of local manufacture, a conclusion that in general would have endorsed conventional analyses based upon vessel form and decoration. A recurrent problem, however, was the lack of a sequence of stratified material, covering a sufficient span from early to late Iron Age, which might have permitted changes in form, fabric or decoration to be convincingly demonstrated. Once again the compressed chronology for Atlantic roundhouses, centring on the first century BC and early centuries AD meant that any perceived differences were still likely to appear to be broadly contemporary.

It was central to the original research strategy for a campaign of excavation in the Bhaltos peninsula that, within such a relatively small and discrete area, in sampling the three principal structural types recognised by conventional classification, that is, (island) broch, (island) dun and wheelhouse, any significant difference in chronology as indicated by radiocarbon determinations would surely be reflected in the material assemblages, or at least conversely any notable distinction in material culture would have to be explained as a factor of chronology or possibly of social differences rather than as the product simply of regional variation. In retrospect one might have been less confident that change would register in the archaeological record during the currency of these monumental structures in a region where cultural conservatism is endemic, but no-one had anticipated uncovering a site in which at least a thousand years’ occupation was contained within one extended stratigraphic sequence.

Within the Bharabhat sequence it is hard to detect any clearly defined changes between the earlier and secondary occupations in terms of their associated pottery (Fig 8). Incurving-rim jars, a variant perhaps of MacKie’s Vaul ware jars, are well represented throughout, as are coarse wares with applied cabled ornament. Bases with internal dimple decoration, also characteristic of the pre-broch assemblage from Dun Mor Vaul, were found only in the earlier contexts, but their occurrence at Beirgh in a wider range of phases, though perhaps residually, raises a doubt about this as a diagnostically early trait. Everted-rim jars occur at Bharabhat, predominantly in secondary contexts, but never with ‘Clettraval’-style arcaded ornament.

Because of its extended stratigraphy, the Beirgh excavation has afforded an opportunity to amplify the ceramic sequence. From the outset it was recognised that plain pottery only, generally of slack body-profile and with tall, very slightly flaring necks to the rim, was recovered from the later Pictish occupation levels in the interior of the site, whereas in the intra-mural galleries, even at first floor level, decorated wares residual from the much earlier Atlantic roundhouse occupation or its immediate successors were commonly present. In essence this reflected what Scott and Young had implied, that the later Iron Age saw a decline in decoration of pottery and a tendency towards slack body-profiles. In due course it became clear that this represented an oversimplification. Coarse wares, some displaying applied cable ornament around shoulder or neck, together with finer wares which might also bear applied cables, occur throughout the post-Atlantic roundhouse and Cellular phases, in the Cellular phases combined distinctively with circular, looped, or horse-shoe designs (Fig. 10). Incised linear or geometric curvilinear designs were common into the post-Atlantic roundhouse phases, notably in the exterior sequence (Fig. 9). Forms typical of Dun Bharabhat, including incurving-rim jars and jars or bowls with short, everted rims, are well represented in the post-Atlantic roundhouse, pre-Cellular phases (Fig. 9). Several of these incurving-rim vessels came from the first floor gallery of the complex Atlantic roundhouse, and could just be residual from that phase depending upon circumstances of deposition; they are here assigned to the Roundhouse phase 10, however, on the grounds that it is more probable that these contexts resulted from secondary activity. Everted rims with internal fluting appear at this stage, but the shallow-tooled arcading of the ‘Clettraval’ style at Beirgh is represented most clearly in the ensuing Cellular phase. Finally, the later Pictish-period occupation has slack-profiled plain wares broadly comparable to those described by Lane (1990,117 and Ill.7.3). Now that a working ceramic sequence is beginning to emerge, particularly through the work of Melanie Johnson, it should in due course be possible to order associated non-ceramic types into a similarly workable sequence. It has to be acknowledged, however, that the earlier Iron Age assemblage appears to be much more limited in the range and quantity of pottery and material types than that which characterises the middle and later Atlantic Iron Age.

Fig. 10 Pottery from Beirgh, Riof, Isle of Lewis; Cellular and later Pictish phases

If the evidence from Dun Bharabhat and Beirgh proves to be typical, it really does appear that for the Early Iron Age in the Western Isles the non-ceramic material culture is fundamentally utilitarian and not in fact represented by a wide range of types; hammer-stones and pounders are fairly prolific (and may in due course be amenable to closer classification), while spindle-whorls and strike-a-lights are likewise not uncommon, but hardly diagnostically distinctive. Indeed it has been argued elsewhere that the lack of diagnostic types may be one reason why the earlier phases of multi-period sites have not always been adequately recognised in the past (Harding, 1995). The only exotic type at Bharabhat was glass beads, mostly not closely dateable, but one spiral-inlaid form belonging to the secondary occupation of the dun around the turn of the millennium. The post-monumental phases, on the other hand, see an increase in the number and range of material types.

Beirgh: moulds for doorknob spearbutts and pins

The Cellular period at Beirgh has produced evidence of metalworking and the production of handpins or proto-handpins and doorknob ‘spearbutts’ in the form of mould fragments. These types were evidently being manufactured in the Western Isles, but the distribution of moulds for doorknob spearbutts in Scotland in particular, and their absence from Ireland in spite of the much greater numbers there of the end product, is bound to prompt a question over the origin of the type.

Beirgh: penannular brooch from later Pictish-period occupation

A new programme of XRF analysis has now been initiated by the major museums involved in Britain and Ireland (A. Heald, pers. comm.), which may well cast light on this vexed issue. Meantime the Beirgh finds provide one of the few reliable contexts with associations for doorknob spearbutts and the handpin series, which on present evidence places the currency of the former somewhat later than might have been anticipated, and that of the latter rather earlier.

Heald’s work on the non-ceramic artefacts of Atlantic Scotland at this period will equally demand that new criteria for the recovery of metalworking materials be built into the operational strategy of any future excavation from the outset, and not just responsively as too often has been the case in the past. The discovery of hammerscale at Beirgh has shown that the potential by-products of every stage in the metalworking process need to be identified and anticipated, so that the generalised conclusion that metalworking was taking place can be refined to define exactly which activities were taking place on any particular site. Metalworking debris occurs so regularly on Atlantic Iron Age sites that it would be easy to conclude that small-scale metalworking was undertaken domestically, which is at the same level of understanding which might conclude that they ate sheep on most sites. The possibility that neighbouring sites operated in concert on different stages in production, or assigned the production of specific types to particular sites, still needs to be explored. The fact that Beirgh, for example, has only produced moulds for pins (ring-headed and hand-pins) and doorknob spearbutts seems singular in itself, a problem which is not helped by the improbability that spearbutts ever adorned the butts of spears. Artefact studies have to change gear too, addressing not just the time-honoured and still vital questions of date and association, but also the social and economic circumstances of their production.

Economy and Environment

It is axiomatic in Hebridean archaeology that palaeo-environmental studies are integral to archaeological research. Even at the most basic level of plotting distributions of human settlement, it is necessary first to determine the land available for habitation, which requires an assessment of factors such as post-glacial rise in sea-level and consequential changes in coastal geomorphology or the spread of peat across the inland parts of the islands which progressively constrained settlement within limited coastal zones. Whether in the fields of palaeobotany, palaeozoology or in the study of soils and sediments, it is no longer sufficient for the archaeologist to enlist non-archaeological specialists to provide a commentary on excavated material; the palaeo-environmentalist must be an archaeologist and must be party to the determination of research strategy. That has been the policy adopted at Calanais and in the research undertaken in its environs, and rather belatedly in the field research programme in west Lewis likewise.

When Finlay began her doctoral research in 1981 on the faunal evidence for prehistoric economy in the Outer Hebrides to AD 400 the greatest handicap to progress was the lack of any major faunal assemblages (with the exception of that from the long-term excavations at the Udal to which she had partial access), still less of any which had been excavated to modern standards. In consequence her conclusion, that ‘the fauna show little change in type or character of species, with stock introduced after domestication and with wild species similar to those available today’ (Finlay, 1984) hardly occasioned surprise. Based upon fifteen years of excavation in the Bhaltos peninsula Cook (forthcoming) was able to reach more startling conclusions. First, the high proportion of deer bones and the nature of bones represented from the later Pictish assemblage at Beirgh strongly argued for a degree of stock management which would not normally have been anticipated in the exploitation of wild herds. Second, the selective representation of bones from the assemblage, even taking into account possible shortcomings of the excavation in recovery of smaller bones, suggested the possibility of communal culling and butchery at an off-site location, followed by distribution of the proceeds, perhaps even based upon a hierarchical allocation of portions. The special role of deer has subsequently been noted from the nearby wheelhouse assemblage at Cnip, and to a lesser degree in the assemblage from Bostadh just across the water on Great Bernera. It remains to be seen whether there is any longer-term shift visible in the thousand-year span of the Beirgh sequence, but the potential for clarifying economic practice in Atlantic Iron Age society is plainly substantial, provided that the fundamental requirements of recovery and sampling are adequately built into the excavation strategy from the outset.

This constraint is even more essential in terms of the recovery of palaeobotanical and related environmental data. From the several excavated sites in west Lewis the survival of plant macrofossils was in general extremely good, and significant advances are being made in the understanding of plant management and procurement, including the management of timber and heathland resources, as well as of the nature of Iron Age agriculture itself (Church, forthcoming). Increasingly attention is focussing on site taphonomy and the interpretation of observed deposits in the stratigraphic sequence. At Bharabhat and Beirgh floors which comprise stone paving are hardly in dispute, but ‘floor levels’ or ‘occupation levels’ represented by mixed soils containing potsherds, animal bones and other debris are less certainly identified. Where considerable quantities of very small potsherds are present we might suspect a trodden surface, and at Bharabhat such sherds were found exceptionally pressed into slabs of hard clay as if they had been trodden in to a floor surface. At Beirgh the major horizon between the Roundhouse occupation and the succeeding phase of Cellular structures was marked by the introduction of a thick and hard peat surface, the purpose of which was undoubtedly in part to raise and level the occupation surface in readiness for the next episode of construction, but which might equally have been a symbolic act, given the effort of importing peat involved, designed to underline the totality of the change. The interpretation of lenses of clean sand, whether wind-blown or artificially introduced, is equally controversial. Wind-blown deposits might correspond to periods of abandonment, or at least to periods in which structures were not roofed, but in the Western Isles a considerable accumulation may be the product of a relatively short-lived episode. How clean occupation surfaces would have been kept whilst buildings were in use has been a topic of debate, highlighting the difficulty of recognising levels which correspond to actual occupation as opposed to those which reflect use in dereliction before total collapse of a building. There really are no a priori grounds for believing that Iron Age communities, highly skilled in structural and artefactual technology and living within a complex economic and social order, should blunder about in their houses amid the squalor of rotting domestic debris and animal excrement, as has sometimes been claimed on the basis of excavated evidence. We should remember that by definition what we excavate is what was left behind or accumulated when the site was abandoned, which, except in catastrophic circumstances like sudden inundation by volcanic action or shipwreck in deep water, is unlikely to leave for the archaeologist an authentic impression of its use in occupation. At Bharabhat the problems of interpreting stratigraphic episodes are further complicated by the fact that a metre of deposits underwater could be the equivalent of what on land is compressed within a tenth of that depth. In sum, what survives within an occupational sequence, as Armit has recently argued (forthcoming), may represent the abnormal rather than the normal use of a building .

The Hebrides and the Western Seaways

The sustained programme of research in the Western Isles (not of course exclusively concerned with the Iron Age), whilst at last consigning to history, we may hope, the older conventional interpretations based upon architectural typology and cultural diffusion, has nonetheless rekindled interest in another older model, that of the role of the western seaways in prehistory and early history (Bowen, 1971; Harding, 1995). The recognition that the archaeology of Atlantic Europe at various times and in varying degrees shows elements in common, such as the practice of burial within megalithic tombs or common fashions in later Bronze Age industries, does not of necessity require the hypothesis of cultural or ethnic uniformity or the transmission of ideas and fashions from any single source. In some instances, nonetheless, most obviously in the case of Norse settlement, attested by historical and place-name evidence as well as archaeological, the western seaways evidently did serve as an artery of cultural diffusion of a kind which archaeologists will not normally concede in the absence of the evidence of parallel disciplines. It is noteworthy that Bowen saw this as a case of ‘the seaways in reverse’, revealing a diffusionist pre-supposition that the normal route for the movement of new fashions or new peoples was from south-east to north-west, exactly as Christopher Hawkes in his lectures at Oxford used to liken the map of Europe to a billiard table in which Britain was the top corner pocket. We need make no such presumption, but may still regard Atlantic Europe as a coherent region in which the western seaways played a crucial role as an artery of communication between communities whose shores faced the Irish Sea and western approaches.

A further presumption, still prevalent in the 1960s and based upon the model advanced by Gordon Childe, that archaeological distributions might equate with ethnically or cultural distinct population groups has long been discredited, and it remains a fact conversely that historically-documented episodes like the Dalriadic settlement of western Scotland are notoriously difficult to equate with any diagnostic archaeological types, either structural or portable artefacts. This need occasion no surprise if we accept that adjacent communities, united rather than divided by water, will have been subject to mutual and reciprocal influences over a long period of time, and doubtless had common ties of kinship as well as language to varying degrees over the centuries or perhaps even millennia before cross-channel links between Ireland and Atlantic Scotland manifested themselves in the historical record. Unravelling the complexities of palaeolinguistics is a potential minefield, especially for archaeologists, but it is clear that there is no simple correlation between any coherent archaeological assemblage, La Tene or non-La Tene, in Atlantic Britain or Ireland, and either of the traditionally-recognised branches of Q- and P-Celtic. Drawing lines down seaways, however, is an inflexible construct which is unlikely to reflect the dynamics of past reality, and it is arguable whether even older place-name elements, for example, could give a reliable indication of the linguistic map of western Europe in the late Bronze Age.

However we address the challenge of reconciling the evidence of Iron Age archaeology and the evidence of Celtic linguistics it is reasonably clear that Atlantic Scotland and the western seaways encompassed a cultural continuum which stands apart from that of central Europe with its sequence of cultures from Urnfield through Hallstatt to La Tene. To regard one as being ‘core’ and the other as ‘periphery’ would be an over-simplification, if not downright misleading. The occupants of Calanais, Beirgh or Bostadh in the Neolithic, Iron Age or Norse periods respectively were all very much in the mainstream of Atlantic Europe.

References

Armit, I., (ed) 1990. Beyond the Brochs; Changing Perspectives on the Atlantic Scottish Iron Age, Edinburgh.

Armit, I. 1992. The Later Prehistory of the Western Isles of Scotland, BAR Brit. Ser., 221, Oxford.

Armit, I. 1997. ‘Cultural landscapes and identities; a case study in the Scottish Iron Age’, in Gwilt, A. and Haselgrove, C. (eds) 1997, 248-53.

Armit, I. 2000. Review of M. Parker Pearson, N. Sharples with J. Mulville, Between land and sea; excavations at Dun Vulan, South Uist,' Sheffield, 1999, Antiquity, 74, No 283, March, 2000, 244-5.

Armit, forthcoming. ‘Floor formation and structured deposition: Cnip wheelhouse, Lewis’, Scottish Archaeological Forum.

Bowen, E. G., 1972. Britain and the Western Seaways, London.

Caulfield, S., 1978. ‘Quern replacement and the origins of brochs’, Proc. Soc. Ant. Scot., 109, 129-39.

Childe, V. G. and Thorneycroft, W., 1938. ‘The vitrified fort at Rahoy, Morvern, Argyll' , Proc. Soc. Ant. Scot., 72, 1937-38, 23-43.

Church, M., forthcoming. ‘The Botany of Brochs- Twenty Years On’, Scottish Archaeological Forum.

Cook, M, forthcoming 2000. The Iron Age Settlement at Beirgh, Riof, Isle of Lewis: Faunal Report on the Later Pictish Occupation, Edinburgh

Cunliffe, B. W., 1991. Iron Age Communities in Britain, 3^rd ed. London.

Curle, A. O., 1921. ‘The Broch of Dun Troddan, Gleann Beag, Glenelg, Inverness-shire,’ Proc. Soc. Ant. Scot., 55, 83-94.

Finlay, J., 1984. Faunal evidence for prehistoric economy and settlement in the Outer Hebrides to c. 400 AD, unpublished thesis, University of Edinburgh; summary in Univ. Edin. Dept. Arch. Ann. Rpt., No 31, 30-31.

Foster, P. 1995. ‘Excavations at Allt Chrisal’, in Branigan, K and Foster, P., eds., Barra: Archaeological Research on Ben Tangaval, Sheffield, 49-160.

Fox, C., 1932/38. The Personality of Britain, Cardiff.

Gilmour, S., 2000. ‘First Millennia Settlement Development in the Atlantic West,’ in J. Henderson (ed), The Prehistory and Early History of Atlantic Europe, BAR, Oxford.

Gwilt, A. and Haselgrove, C. 1997. Reconstructing Iron Age Societies; New Approaches to the British Iron Age, Oxbow Monographs 71, Oxford.

Hamilton, J. R. C., 1956. Excavations at Jarlshof, Shetland, HMSO, Edinburgh.

Harding, D. W., 1984. ‘The function and classification of brochs and duns,’ in R. Miket and C. Burgess (eds), Between and beyond the walls, Essays on the Prehistory and History of North Britain in Honour of George Jobey, Edinburgh, 206-20.

Harding, D. W., 1995. ‘Atlantic Scotland and the Western Seaways’, paper given to X Internat. Congress Celtic Studs., Edinburgh, 1995, University of Edinburgh, Department of Archaeology.

Harding, D. W. 1997. ‘Forts, Duns, Brochs and Crannogs: Iron Age Settlements in Argyll,’ in G. Ritchie (ed), The Archaeology of Argyll, Edinburgh, 116-40.

Harding, D. W., forthcoming,a. ‘The Outer Hebrides in the Early Christian Period’, in E. Rynne (ed) Essays in Memory of Liam de Paor.

Harding, D. W., forthcoming,b. ‘The Atlantic Scottish Iron Age: External Relations Reviewed’, Tall Stories?, Proceedings of Conference in Lerwick, Shetland, July, 2000.

Harding and Dixon, forthcoming, 2000. Dun Bharabhat, Cnip: An Iron Age Settlement in West Lewis, Vol. 1. Structures and Material Culture, Calanais Research Series, 2, Edinburgh.

Harding and Gilmour, 2000. The Iron Age Settlement at Beirgh, Riof, Isle of Lewis: Excavations 1985-95, Vol.1 The Structures and Stratigraphy, Calanais Research Series, 1, Edinburgh.

Hawkes, C. F. C.,1959. ‘The ABC of the British Iron Age’, Antiquity, XXXIII, 170-82.

Hedges, J. W., 1987. Bu, Gurness and the Brochs of Orkney, BAR British Ser., 163-5, Oxford.

Henderson, J. 1998. ‘Islets through time: the definition, dating and distribution of Scottish crannogs’, Oxford Journ. Arch., 27.2, 227-44.

Lane, A., 1990. ‘Hebridean Pottery: Problems of Definition, Chronology, Presence and Absence’, in I. Armit (ed), Beyond the Brochs, Edinburgh, 108-30.

Lethbridge, T. 1952. ‘Excavations at Kilphedir, South Uist, and the Problem of the Brochs and Wheel-houses’, Proc. Prehist. Soc., 18, 176-93.

MacKie, E. W., 1965a. ‘The origin and development of the broch and wheelhouse building cultures of the Scottish Iron Age,’ Proc. Prehist. Soc., 31, 93-146.

MacKie, E. W., 1965b. ‘Brochs and the Hebridean Iron Age’, Antiquity, 39, 266-78.

MacKie, E. W., 1969. ‘Radiocarbon dates and the Scottish Iron Age,’ Antiquity, 43, 15-26.

MacKie, E. W., 1971. ‘English migrants and Scottish brochs’, Glasgow Arch. Journ., 2, 39-71.

MacKie, E. W., 1974. Dun Mor Vaul, an Iron Age broch on Tiree, Glasgow.

Mackie, E. W., 1991. ‘The Iron Age semi-brochs of Atlantic Scotland: a case study in the problems of deductive reasoning,’ Arch. Journ., 148, 149-81.

MacKie, E. W., 1995. ‘The Early Celts in Scotland,’ in M. Green (ed), The Celtic World, London, 654-70.

Mackie, E. W., 1997. ‘Dun Mor Vaul revisited: Fact and Theory in the Reappraisal of the Scottish Atlantic Iron Age,’, in G. Ritchie (ed), The Archaeology of Argyll, Edinburgh, 141-80.

Marshall, P., et al., 1999. The Late Bronze Age and Early Iron Age Community at Cladh Hallan, South Uist; Excavations in 1999, Sheffield.

Parker Pearson, M., Sharples, N., and Mulville, J., 1996. ‘Brochs and Iron Age Society’, Antiquity, 70, 57-68.

Neighbour, T and Burgess, C., 1996. ‘Traigh Bostadh’, Discovery and Excavation in Scotland, 1996, 113-4.

Parker Pearson, M. and Sharples, N., 1999. Between Land and Sea, Excavations at Dun Vulan, South Uist, Sheffield.

Piggott, S., 1958. ‘Native economies and the Roman occupation of North Britain,’ in I. A. Richmond (ed), Roman and Native in North Britain, Edinburgh, 1-27.

Piggott, S., 1966. ‘A scheme for the Scottish Iron Age’, in A. L. F. Rivet (ed) The Iron Age in Northern Britain, Edinburgh, 1-25.

RCAHMS, 1928. The Outer Hebrides, Skye and the Small Isles, Edinburgh.

RCAHMS, 1946. Orkney, Edinburgh.

Scott, Sir Lindsay, 1948. ‘ Gallo-British Colonies: The Aisled Roundhouse Culture in the North,’ Proc. Prehist. Soc., XIV, 46-125.

Sheffield, 1999. Allt Chrisal, T17: A Preliminary Report on the Excavation of Iron Age and Later Structures, 1996-1999, Sheffield.

Smith, Beverley Ballin (ed), 1994. Howe: four millenia of Orkney prehistory, Society of Antiquaries of Scotland Monograph Series, 9, Edinburgh.

Stevenson, R. B. K., 1966. ‘Metalwork and some other objects in Scotland and their cultural affinities,’ in A. L. F. Rivet (ed) The Iron Age in Northern Britain, Edinburgh, 17-44.

Topping, P. G, 1986. ‘Neutron activation analysis of later prehistoric pottery from the Western Isles of Scotland’, Proc. Prehist. Soc., 52, 105-29.

Topping, P. G., 1987. ‘ Typology and chronology in the later prehistoric pottery assemblages of the Western Isles’, Proc. Soc. Ant. Scot., 117, 67-84.

Young, A., 1966. ‘The Sequence of Hebridean Pottery,’, in A. L. F. Rivet (ed) The Iron Age in Northern Britain, Edinburgh, 45-58.

---

Acknowledgement of Funding

The excavations at Dun Bharabhat and Loch na Beirgh were funded exclusively, apart from private donations, by the University of Edinburgh through its General Council Trust, Development Fund and Moray Fund. Publication was made possible by a grant from the Faculty of Social Sciences Research Fund.

No financial support was received at any stage from any central or local government agency, nor from any external body with one exception. The British Academy in 1996 made an award of stlg 4725 for the preparation of a report on the later Pictish-period faunal assemblage from Beirgh, a full report of which was submitted in July, 1997, within two months of completion of the study. That report will be published in the Calanais Research Series as soon as funding is available for that purpose.

The support of all those bodies and individuals who have contributed to the programme is gratefully acknowledged.

This paper was submitted to Antiquity on 1 November, 1999. It was rejected on 10 December by the Editor, who on 14 December invited the author to re-submit it in amended form. This was done on 16 February, 2000. Having still not had a decision by mid April, the author formally withdrew it on 19 April, 2000.

Readers of and contributors to archaeological periodicals may question why peer reviewers should remain anonymous, since it is thereby unclear whether they genuinely command peer-authority in the field under review, or whether their judgement is unduly influenced by the fact that they represent a contrary point of view. In judging the contributions of younger scholars, peer review might even inhibit the publication of views with which established scholars disagree, but without discussion of which the discipline cannot progress. Archaeology especially, in which specialist fields may involve relatively small groups of academic or professional practitioners, needs to exercise particular care to ensure that the system does not fall into disrepute. Quis custodiet ipsos custodes?

last edited 19/5/2000