Cladh Hallan - Roundhouses and the dead in the Hebridean Bronze Age and Iron Age: Part I: Stratigraghy, Spatial Organisation and Chronology

By Mike Parker Pearson, Jacqui Mulville, Helen Smith and Peter Marshall

This first of two volumes presents the archaeological evidence of a long sequence of settlement and funerary activity from the Beaker period (Early Bronze Age c. 2000 BC) to the Early Iron Age (c. 500 BC) at the unusually long-occupied site of Cladh Hallan on South Uist in the Western Isles of Scotland. Particular highlights of its sequence are a cremation burial ground and pyre site of the 18th–16th centuries BC and a row of three Late Bronze Age sunken-floored roundhouses constructed in the 10th century BC. Beneath these roundhouses, four inhumation graves contained skeletons, two of which were remains of composite collections of body parts with evidence for post-mortem soft tissue preservation prior to burial. They have proved to be the first evidence for mummification in Bronze Age Britain.

Cladh Hallan’s remarkable stratigraphic sequence, preserved in the machair sand of South Uist, includes a unique 500-year sequence of roundhouse life in Late Bronze Age and Iron Age Britain. One of the most important results of the excavation has come from intensive environmental and micro-debris sampling of house floors and outdoor areas to recover patterns of discard and to interpret the spatial use of 15 domestic interiors from the Late Bronze Age to the Early Iron Age. From Cladh Hallan’s roundhouse floors we gain intimate insights into how daily life was organized within the house – where people cooked, ate, worked and slept. Such evidence rarely survives from prehistoric houses in Britain or Europe, and the results make a profound contribution to long-running debates about the sunwise organisation of roundhouse activities. Activity at Cladh Hallan ended with the construction and abandonment of two unusual double-roundhouses in the Early Iron Age. One appears to have been a smokery and steam room, and the other was used for metalworking.

• Language: English
• Publisher: Oxbow Books
• Release date: Oct 31, 2021
• ISBN: 9781789256949

Mike Parker Pearson

Mike Parker Pearson is Professor of British Later Prehistory at University College London. A distinguished prehistorian he has been involved with many major projects, including leading the recent Stonehenge Riverside Project. His many publications include Stonehenge: Exploring the Greatest Stone Age Mystery (2012) and From Machair to Mountains: archaeological survey and excavation in Uist (2012).

Book preview

1 The Cladh Hallan excavations and their context

M. Parker Pearson, P. Marshall, J. Mulville and H. Smith

The machair plains of Scotland’s Western Isles provide some of the most outstanding archaeological landscapes in Europe. Hundreds of settlements and farmsteads have been preserved as mounds or mini-tells, dating variously from the Early Bronze Age to the early modern period. Their deep stratigraphic sequences include the well-preserved remains of house floors and walls as well as myriad layers of middens and other deposits which rarely survive within Europe’s heavily cultivated ploughlands or on the thin soils of Britain’s uplands. This high level of survival is enhanced by the machair’s highly calcareous shell sand, providing exceptional preservation of animal and human bone.

The Western Isles, or Outer Hebrides, are a chain of islands running north–south 60–80km off the northwest coast of Scotland. South Uist is one of the larger southern islands, 30km north–south and 12km east–west. Its geology is Lewisian gneiss, underlying glacial deposits and peat which are overlain by calcareous windblown sand (with pH values of 6.5–8.0) forming machair plains along South Uist’s western seaboard. This western zone of machair soils and vegetation forms a fertile strip along the exposed Atlantic coast. To its east, the machair meets a zone of shallow, acidic peat soils known as ‘blackland’, speckled with numerous freshwater lochs and lochans. The eastern third of the island is rocky and mountainous, descending on the east coast to three sea lochs.

The climate of the Western Isles is cool, cloudy, windy and wet, with mild winters and an unusually narrow annual and diurnal temperature range (8.8°C), thanks to the influence of the Gulf Stream. Wind speeds, however, are among the highest in the world. With most of South Uist’s woodland gone by 2500 BC (Brayshay and Edwards 1996), the island’s vegetation is largely restricted to marram on the coastal dunes, fescue on the west coast pasture and heath-grass and mat-grass on the blackland (Hudson 1991). Before the modern era, the shortage of local timber resulted in reliance for nearly four millennia on driftwood from as far away as eastern Canada, supplied by the Gulf Stream.

Sea-level rise during the Holocene has not only inundated extensive plains along South Uist’s west coast but also caused large quantities of shell sand to be swept landwards and even inland to form an extensive dune system along the west coast. Offshore peats buried beneath this sand indicate that its formation commenced before 3700 BC (Ritchie 1979). Radiocarbon dates from Cladh Hallan and other Early Bronze Age settlements indicate that settlements and their fields were established on top of these machair sands by or around 2000 BC (Sharples 2009).

Although now stabilised under grassland, the coastal dune systems have been subject to periods of extensive wind erosion during which large, active dunes have buried large areas of pasture and fields. Archaeological sites have also suffered through the creation of erosion scars and deflation of stratified layers. Coastal erosion is a continuous and growing threat, as sites such as the Norse farmstead at Cille Pheadair¹ are taken by the sea (Parker Pearson et al. 2018). Yet one of the greatest threats to the machair’s archaeological sites is that caused by rabbits (Parker Pearson et al. 2011). Digging their burrows to depths of well over a metre, they cause considerable damage to the stratigraphic integrity of the settlement mounds. It is only those few locations, such as the main Cladh Hallan site (Area A in this report), covered by a deep mantle of windblown sand that have been protected from the rabbits’ damaging impact.

The settlement evidence from South Uist spans prehistory from the Neolithic to the Iron Age (Parker Pearson et al. 2004). Although not yet represented by findspots on the island, the Mesolithic presence in the Outer Hebrides (c. 8000–4000 BC), suspected from palynological research (Brayshay and Edwards 1996; Edwards 1996), has been confirmed further north in Harris by the discovery of midden deposits at Northton, dated to 7060–6650 cal BC (Gregory et al. 2005).

Neolithic sites (c. 4000–2450 BC) in South Uist include seven chambered tombs in various parts of the island (Cummings et al. 2012) and settlement sites at Loch a’ Choire (Henley 2012) and An Doirlinn (Parker Pearson 2012b: fig. 2.9; Garrow and Sturt 2017). Much of South Uist’s settlement evidence from this period has probably been destroyed by the sea, or lies submerged off the west coast, or is buried inland deep beneath machair sand, or survives undiscovered on crannogs (artificial islets) within freshwater lochs as found elsewhere in the Western Isles (Armit 2003; Lenfert 2012; Garrow and Sturt 2019).

Table 1.1 The chronological sequence of activities in Cladh Hallan’s Areas A–D

The Neolithic in Britain ends with the appearance of Beaker pottery, initially within a short-lived Chalcolithic or Copper Age (c. 2450–2200 BC) and continuing within the first half of the Early Bronze Age (c. 2200–1600 BC) to c. 1850 BC. Beaker pottery has been found in South Uist within settlement mounds at Cladh Hallan, on Cill Donnain machair in the middle of the west coast, and on Iochdar machair in the northwest part of the island (Parker Pearson 2012b: fig. 2.10). However, none of the investigated Beaker-period settlement sites have produced material dating to before the beginning of the Early Bronze Age (Sharples 2009).

Middle–Late Bronze Age (c. 1600–700 BC) and Early Iron Age sites (c. 700–200 BC) of similar date to Cladh Hallan are known at numerous locations along the west coast, with a single site in the east (Parker Pearson 2012b: fig. 2.11). The numbers and locations of sites of the Late Bronze Age and Early Iron Age are outlined in Section 1.5.

From the Middle Iron Age (c. 200 BC–AD 300) through the Late Iron Age (c. AD 300–800) to the Norse period (c. AD 800–1266), settlement appears to have been dense along the west-coast machair, with settlement mounds spaced regularly and densely within this strip (Parker Pearson 2012b: figs 2.12–2.15). From the later medieval period onwards, settlements were only rarely located on the machair. The territorial organisation of townships as west–east strips across the island, as recorded at the beginning of the nineteenth century, may have its origins in the Middle Iron Age (Parker Pearson 2012b).

The Cladh Hallan excavations are being published as two monographs. This first volume describes the stratigraphic sequence from Beaker-period cultivation through Early and Middle Bronze Age burials to Late Bronze Age and Early Iron Age houses (summarised in Table 1.1). It concludes with chapters on soil micromorphology, radiocarbon and optically stimulated luminescence dating (OSL), and the pottery. The second volume includes the analysis of artefacts, animal bones, carbonised plant remains and other environmental and subsistence evidence, analysis of the human remains, and a concluding summary.

1.1 The site of Cladh Hallan and its environs

Cladh Hallan is the graveyard for the modern community of Dalabrog (Daliburgh) and its neighbouring townships in the southern part of the island of South Uist (Figure 1.1).² This walled cemetery is located, characteristically for the Western Isles, at a distance from the living, out on the exposed machair dunes between the peatlands and the sea. It lies just 500m from the Atlantic coastline (Figure 1.2), perched upon a sand-covered hill which has been used as a place of Christian burial for at least a thousand years. An ancient cross-marked grave slab in the lower north side of the graveyard dates to the Early Christian period (Fisher 2001), whilst a low mound nearby (Site 49)³ and another mound (Site 50) further south on top of the hill (which, like all ‘hills’ on the machair, is in fact a sand dune) are probably ancient burial sites of the first millennium AD (Figure 1.3; Parker Pearson 2012b: 47–8). The place-name of Hallan is also given to the freshwater loch on the eastern, landward side of the cemetery; it may derive from the name of a Father Allan. The modern graveyard wall was built in the early twentieth century and extended on its south and east sides in the 1990s.

Figure 1.1. The machair dunes, looking east from Cladh Hallan Site 55 towards the modern graveyard of Cladh Hallan

The archaeological remains reported in this volume lie in an undulating landscape of small dunes and prehistoric settlement mounds to the west of the modern graveyard, either side of a sand track which leads westwards from a radio mast to the beach. They share the name ‘Cladh Hallan’ though they are located over 200m from the modern cemetery.

The numerous archaeological sites in the Cladh Hallan area first came to scientific attention in the early twentieth century with a report in the 1912 issue of the Transactions of the Inverness Scientific Society and Field Club by a Mr J. Wedderspoon. Although he did not carry out excavations, he observed remains in the course of destruction by the elements and by the builders of the cemetery walls (see below for an account of his findings).

In 1950 Tom Lethbridge, a Cambridge-based archaeologist, noted prehistoric remains in the dunes below Cladh Hallan cemetery (summarised below). Invited by the charismatic Werner Kissling, a German émigré, to excavate a wheelhouse on South Uist’s machair, in 1951 Lethbridge excavated an unusually well-preserved Iron Age wheelhouse⁴ on the machair at Kilpheder (Cille Pheadair), the next township to the south.

The Cladh Hallan area was later investigated by the SEARCH project (Sheffield Environmental and Archaeological Research Campaign in the Hebrides) of the Department of Archaeology & Prehistory of the University of Sheffield. In 1989, Eddie Moth and University of Sheffield students carried out a limited excavation of prehistoric layers under threat within a sand quarry at Cladh Hallan (Site 54; Area C in this report). Further excavations of this site in Area C were carried out in 1994–1996 and 1998 by Mike Parker Pearson, Jacqui Mulville and Helen Smith. Sand quarrying by local builders of the northern end of a much larger settlement mound (Site 55; Area A; 90m north–south by 80m east–west) revealed archaeological deposits which were excavated in 1996–2003 by the same three, joined by Pete Marshall.

The earlier excavations of Site 54 at Cladh Hallan were carried out at the start of the SEARCH project’s excavation programme, coinciding with the excavation of Cill Donnain Iron Age wheelhouse (Parker Pearson and Zvelebil 2014) and before the excavation of the Iron Age broch of Dun Vulan (Parker Pearson and Sharples 1999).

In these earlier years, the other arms of the SEARCH project were carrying out surveys and excavations on the neighbouring island of Barra (Branigan and Foster 1995; 2000; 2002; Branigan 2005), in addition to environmental studies of these southern isles (Gilbertson et al. 1996) and exploration of South Uist’s interior (Fleming 2012) and east coast (Moreland 2012).

The subsequent excavations of the larger mound (Site 55) ran alongside other major excavations in South Uist, of the Norse farmstead at Cille Pheadair (Parker Pearson et al. 2018) and of the large Norse and Iron Age settlement at Bornais (excavated by Niall Sharples of Cardiff University; Sharples 2005; 2012b; 2019; 2020). Drawing on staff and students from several universities, the Cladh Hallan project was just one element of a major research synergy which pooled substantial expertise and resources, including financial support from Historic Scotland (later to become Historic Environment Scotland).

The context for the large-scale excavations at Cladh Hallan, Bornais, Cille Pheadair and Cill Donnain was provided by a series of surveys and further smaller excavations elsewhere in South Uist. These have produced information on a multi-period background of settlement histories from the Neolithic to the modern period (Parker Pearson 2012a). Whilst surveys on the blacklands and moorlands in the eastern part of the island identified numerous prehistoric sites, the machair survey along the west coast identified an unusual density of Bronze Age–Iron Age sites in the form of settlement mounds of varying sizes (the site numbers in this volume derive from this machair survey; Parker Pearson 2012b). Twenty such sites were identified within 1km of Sites 55 and 54 at Cladh Hallan, most of them discovered in 1995 (Parker Pearson 2012b: fig. 2.4).

This high number of Bronze Age–Iron Age sites in the vicinity of Cladh Hallan reveals the significance of this area in later prehistory. Linking Wedderspoon’s and Lethbridge’s accounts to these sites recorded by the machair survey is difficult, though there are possible correspondences for some of them (see below).

Figure 1.2. Late Bronze Age/Early Iron Age occupation in South Uist and the southern isles of the Outer Hebrides

Figure 1.3. Prehistoric sites in the vicinity of Cladh Hallan; the excavations took place at Areas A–D on Sites 55, 54 and 53

1.2 The Bronze Age to Early Iron Age settlement at Cladh Hallan

The SEARCH investigations at Cladh Hallan consisted of four excavations in discrete locations, labelled as Areas A–D (Figure 1.3). Areas A–C were rescue excavations on sites already damaged by sand quarrying and wind erosion whereas Area D was a small trench to investigate the southern end of the large settlement mound (Site 55) into which the largest excavation trench (Area A) was dug. The excavations in Areas A and D cover about a tenth of the full extent of this large settlement mound (Site 55).

Area A

Area A is the northern third of a quarry-damaged settlement mound (Site 55). The most complete structures discovered are three Late Bronze Age sunken-floored roundhouses dug into the machair sand. The houses have shared or ‘party’ sand walls, faced with stone, attaching one to the next to create a terraced row. The houses are identified by context numbers: House 1370 is the northernmost; House 401 is the largest, in the middle, with a forecourt building; House 801 is the southern house (see Figure 1.21). Most of the settlement mound south of Area A is still covered by a deep sand dune; we estimate that we excavated the northern part of what may have been a row of six to eight Late Bronze Age roundhouses.

Area A’s upper layers of inhabitation are Early Iron Age; the base of the stratigraphic sequence dates to the Early Bronze Age. All the houses in Area A were preceded by a cremation cemetery and pyres and, before that, by a Beaker-period cultivation horizon. Pre-dating the terraced row of three roundhouses were three other Late Bronze Age house structures (Houses 2835, 2477 and 3260). Another two date to within the row’s sequence (Late Bronze Age House 2190 and Early Iron Age House 1500), and an Early Iron Age double roundhouse (House 640) was the final building at the end of the occupation.

Each of the three roundhouses in the terraced row contained one or more human burials beneath its primary floor. Detailed analysis of the skeletons has revealed that two of these bodies were composites of multiple individuals and that they were originally mummified.

Area B

Area B (Site 53) was discovered within a crater of eroding sand in which large stones, cultural deposits and artefacts were observed (see Chapter 2). It is not a settlement mound as such and no certain structures were revealed on excavation (see Figure 2.28). It produced unstratified finds from the Bronze Age and Iron Age (see Figure 2.29).

Area C

Area C (Site 54) is a settlement mound with an Early Iron Age double roundhouse (House 150) in its uppermost layers (see Figures 1.11–1.13 and Chapter 10). Due to modern quarrying, it survives as a knoll several metres above the surrounding quarry floor. The base of the mound consists of deposits containing Early Bronze Age Cordoned Urn pottery, above which are midden layers with Late Bronze Age ceramics. A Middle Bronze Age skeleton was disturbed by quarrying from a nearby but unidentified context.

Area D

Area D (south end of Site 55) is a small trench dug in 2001 into the southern, undamaged edge of the large settlement mound. Its stratigraphic sequence included the sand walls of a roundhouse considered on ceramic typological grounds to be of the same phase as the row of three Late Bronze Age sunken-floored roundhouses (Houses 801, 401 and 1370) excavated in Area A.

1.3 Previous discoveries

Mr Wedderspoon visits

In 1912 Mr J. Wedderspoon reported to the Inverness Scientific Society and Field Club his discovery of archaeological remains at Cladh Hallan (1912: 327–30). His observations are worth quoting in full:

‘The next discovery was made on the machar [machair] near Daliburgh in South Uist. The old graveyard of Hallan situated on the machar between Loch-Hallan and the shore, was until a few years ago unfenced, and in a very neglected condition. Eventually the matter was taken in hand, and arrangements were made to erect a wall round it. The stones required in the work were to be carted from the beach, and to that circumstance the discovery of the middens was due, for shortly after the work was started a succession of gales stopped the operations, but the wind found a weak part in the surface of the machar, which had been broken by the carts carrying the stones, and excavated a number of pot-shaped holes in the underlying sand, and also demolished a large grass-covered dune, which contained a bee-hive shaped structure of dry stonework.

The Contractor, whose antiquarian sense was badly developed, looked on the find as a godsend, and quickly demolished it to the last stone, and the only relic left of the house is a large boulder with a basin-shaped cavity somewhat similar to the one at Baleshare.

The little information I could gather about the structure was not of a conclusive nature. It was described as shaped like a bowl, and nine or ten feet in height. It was full of sand inside, which was not removed, and no door or exterior passage was observed. About 100 cart loads of stones were taken from it; these might average about five cwt. a load, or 25 ton, in all.

When I visited the place the site of the house was covered with sand, and nothing remained to tell of the structure but the basin-shaped stone, which has now been removed to Askernish.’

This demolished structure could just possibly have been the double-roomed House 640, interpreted as a smokehouse/ steam room and dating to the Early Iron Age (see Chapter 12). Excavated in 1998, it lies within the northeast corner of Area A, immediately south of the track between the cemetery and the beach, where the northern edge of the settlement mound (Site 55) is cut into by the track (see Figure 1.16). Without any holes for roof-bearing posts and with interior spaces no wider than 4.20m, House 640 is best interpreted as having been a stone-corbelled beehive-shaped structure. This would fit well with Wedderspoon’s description of a structure standing 3m high, much larger than a corbelled burial cist of the type found on Rosinish machair (cf Crawford 1978: figs 2–3) and more like the exceptionally well-preserved Middle Iron Age wheelhouse at Cille Pheadair excavated by Lethbridge and Kissling (Lethbridge 1952) just a mile to the south (Site 64 in Parker Pearson 2012b: fig. 2.4). Although no doorway was noted by the cemetery-wall builders, this could have been because of the door’s small size (like that found in the other double-roomed building, House 150, in Area C).

To continue with Wedderspoon’s account:

‘Retracing my steps towards the burial ground which occupies the landward side of a grass-covered sand-hill, about 50 feet in height, I came on a pot-shaped hole about 30 yards in diameter, neatly excavated by the wind, which had piled up the blown sand into a new dune on the landward side of the hole. The sides of the pit were nearly perpendicular, unless towards the north-east, which sloped upwards to the new dune. The hole was about nine feet deep, the bottom fairly level with a low ridge running across it from north to south, and on its landward side part of a huge midden was exposed. The midden had evidently been undisturbed since the covering sand was removed, probably about two months before I saw it.

The first find was a hammerstone with several indented marks, which resemble the marks on a bone comb found at one of the mounds at Bernera in Harris.

Near where the hammerstone lay a small conical heap of dark coloured matter was observed, about six inches wide, and three or four in height, enclosed by a small ring of a brown earthy substance which contained a few fragments of unglazed pottery, some of them marked with the so-called herring-bone pattern, which determined the nature of the find. The clay of the urn was soft and gritty, and must have crumbled away on being exposed to the air.

The little heap of dark matter consisted of charred bones, and a soft humid substance like wood ashes. Resting on the top of the heap of bones was a piece of clay slate shaped like the letter S, which must have been fashioned by hand.

At the north end of the midden, where it disappears in the sandbank, a few fragments of another urn were found on a little heap of charred bones similar in every respect to the other find.

A feature of the midden was the numerous fragments of deer horn, some of a very large size, but all soft and badly decayed. Ox teeth were very numerous, also tusks, and teeth of other animals, but horse teeth were rare⁵; four only were found after a close search, and two human teeth.’

Again, locating the position of this sand blowout with its midden and cremation burials is difficult. Wedderspoon may have been describing one of the settlement mounds on either side of the track – that to the north (Area C) or that to the south (Area A). The ‘herring-bone pattern’ on the pottery probably describes Early Bronze Age urns (cf Crawford 1978: fig. 4) and ties in with the later discovery of Early Bronze Age levels in Area A and Area C.

Returning to Wedderspoon:

‘Continuing the exploration among the sandhills along the cart track another wind-swept hollow was found at the bottom of the hill occupied by the burial ground. The excavated surface is about a quarter of an acre in extent, in which part of a very large midden was exposed which extends in the direction of the burial ground till hid by the sand about 30 yards west from the enclosure. One part of the midden was composed almost entirely of cockle shells; the rest of the surface was covered with shells and bones indiscriminately.

A considerable number of burnt stones in groups were found at several places, which probably indicates the sites of the fire hearths.

On a smaller mound a little detached from the large midden, a considerable number of flints were found, all of which showed distinct marks of flaking. Many of these flints are in no way different from those found on the east coast, while the others resemble more the nodule flints of Antrim.

This is the only place in the Hebrides where I have found wrought flints, though I have often searched for them. A well-marked hammerstone was found with the flints, one end damaged by fire, and a bone needle.

Several hammerstones were got at the large midden, two bearing marks of fire, also a bronze knob, several human teeth, a bone bodkin, broken bone needles, fragments of coarse pottery, and a number of wrought bones of unknown use. The latter were found near the centre of the hollow on a smaller midden which was entirely covered with splinters of horn and bone.

On a subsequent visit to this midden, another small heap of charred bones was found, with a few fragments of the urn marked in the herring-bone pattern. These were got on a part of the midden which was covered with sand on my previous visit, and near a ring of fire-marked stones. To all appearance a small part only of one side of the deposit is exposed. The highest part where it disappears in the sandhill is about 8 feet above the old land surface, and has an upward slope into the sandhill towards the burial ground.

These middens are situated on the east or landward side of the hollow, and rest on the old land surface, which consists of a light sandy loam. Thirty yards from the west side the loam ends, and the surface, which falls away towards the sea, is covered with gravel to the edge of the sand face, which at that point is 10 feet in height. The distance from there to the present shore line is about 250 yards.’

Once again, this site is unlocated but it has to lie between the track from the beach and the graveyard. The only possible location is east of the track and within the hollowed-out dunes closest to the graveyard’s western wall. This is most probably Site 215 (NF 7324 2198), a large Earlier Bronze Age mound, with more than 2.30m of stratigraphic deposits, the surface of which was partially exposed in the mid-1990s within the two northernmost hollows in the dunes east of the track where it bends southwards around the settlement mound of Area A. This site has produced two flint flakes and six sherds, one of which has a fine fabric and could come from a Beaker vessel (Parker Pearson 2012b: 66–7).

From Wedderspoon again:

‘Proceeding southward along the machar towards Kilphedar, the grass-covered dunes give place to low-lying sand flats, which on the land side enclose an inland basin about two acres in extent, and known locally as the Old Loch. The bottom is covered with a dark slimy mud, intersected by shallow pools of water. Near the north side of this basin the remains of another midden were found by seeing the broken shells and bones showing white on the dark coloured mud. The midden was traced to the shore line of the old loch, and extended 20 yards northward from that point till lost in a sand-hill. Where it disappeared in the sand-hill, two hammerstones were found, several horn pins, and broken bone needles, a shaped piece of horn of the toothpick pattern, several human teeth and charred bones, a few fragments of pottery with herring-bone markings, a number of flint flakes, and a chipped stone, which may have been used to break marrow bones, or for settling personal disputes in the fashion of the times.

The bones and shells of the midden are very much decayed on account of their long exposure to the elements. The shells are mostly in fragments, and the bones are brown and brittle with age.’

This site has not been located, although seven settlements of different periods have been found south of Cladh Hallan and north of the Cille Pheadair township boundary (Parker Pearson 2012b: fig. 2.4). From Wedderspoon’s description, it appears to be closer to the old loch than any of the sites found during the machair survey in the 1990s.

Tom Lethbridge at Cladh Hallan

In 1950 Lethbridge noted an Early Bronze Age site somewhere in this area: ‘one of these [sites], just beneath the modern Daliburgh graveyard, belonged to the earliest Bronze Age. Here barbed-and-tanged arrowheads of flint, minute thumb scrapers and microlithic barbs lay about in profusion. Most of the middens, however, from Daliburgh to Kilpheder, belonged to the wheel-house people. There may have been a dozen of them at one time.’ (Lethbridge 1954: 180–1). The site cannot now be located but may not have been far from the area of Beaker-period cultivation that lies beneath Site 55 (Area A). It could well be the same small mound covered with a considerable quantity of ‘wrought flints’ that Wedderspoon spotted in the hollow near the bottom of the cemetery hill and is most likely Site 215.

Prior to his work on South Uist, Lethbridge had excavated widely on sites in the Cambridge area and, coincidentally, had used evidence from a round barrow that he investigated at Snailwell to argue that Bronze Age Britons had practised mummification (1950). One of the Snailwell burials was missing its feet, suggesting post mortem loss sometime after death, and another had ‘scorched’ bones which Lethbridge took as evidence for over-enthusiastic drying-out of the corpse over a fire. Sadly the Snailwell skeletons cannot be located in either the Museum of Archaeology and Anthropology or the Duckworth Laboratory collections in Cambridge so there is no way of re-evaluating Lethbridge’s claims, although his arguments do not appear particularly convincing today. In retrospect, he may have been right for the wrong reasons. It is deeply ironic that he probably walked over the very site which, some 50 years later, would produce evidence of Bronze Age mummification.

1.4 Survey, test excavations and trial-trenching 1988–1996

In June 1988 an archaeological team surveying the island’s dunes and machair for the SEARCH project discovered the site now known as Area C (Site 54; Parker Pearson 2012b: 48, fig. 2.4). Located about 200m inland from the beach and about 30m north of the track from the radio mast to the beach, this 4m-high dune was being cut in half by sand quarrying. Two midden layers could be seen in the upper face of the dune and large quantities of animal bones, shells and sherds were seen eroding out of the lower of the two midden layers. There was also a possible hearth in this layer but, by June 1989, further sand extraction and natural erosion had cut back the central part of the dune, destroyed the hearth and uncovered a third, lower midden layer.

The 1989 excavation

Figure 1.4. Excavations in 1989 at Cladh Hallan sand quarry (no grid co-ordinates have survived)

In June 1989 Eddie Moth directed a small excavation in Area C as part of the SEARCH project (Figure 1.4; Moth 1989a; 1989b). The site was given a geographically incorrect side code (KIL89); all the archaeological sites around Cladh Hallan are in Dalabrog township, but in 1989 the area was misidentified as being part of Kilpheder township (Cille Pheadair), which lies to the south. At some point since 1989, the surveying records for KIL89 were mislaid, preventing precise mapping of the trench or its sections and plans.

Figure 1.5. The 1989 north–south section at its south end

Figure 1.6. The 1989 north–south section towards its middle

Figure 1.7. The 1989 north–south section towards its north end

Figure 1.8. Plans of the lower layers of the midden, within the stepped section of 1989; layer 16 was a midden layer and the rest were sand layers

Figure 1.9. Pottery from layer 009 (equivalent to phase 9 in the later excavations) and from unstratified surface exposures, collected in 1989

Note that Moth’s excavation was located east of the double roundhouse House 150. The section locations in Figure 1.4 are the approximate positions of Moth’s sections (Figures 1.5–1.7) of his trench, on the eastern side of the dune (in fact a settlement mound on top of a dune) as it then survived. The area of dune or settlement mound examined by Moth has completely disappeared: the double roundhouse (House 150; see below) is now pedestalled on a remnant of the mound, to the west of the location of all Moth’s investigations. In 1989 the active quarry was also being used as a rubbish tip and Eddie’s team spent their tea-breaks in an open-air living room equipped with old sofas and broken chairs and tables, where they provided visitors with refreshments and soft furnishings.

Where the sand quarry had cut into the east side of the mound, Moth’s team exposed a north–south section, cutting it back to a straight, vertical face and excavating it in steps descending from the top of the mound to the west to the base of the quarry to the east. This allowed the 4m depth of stratified deposits within the mound to be recorded in section (Figures 1.5–1.7) and plans to be drawn of layers exposed by the steps (Figure 1.8). An excavation trench, c. 1m deep, was also dug on the western, seaward side of the long section, to the top of a midden layer (007) and was cut down in one location to a lower layer (009).

Some sections and small plans survive from the 1989 excavations, but without co-ordinates, so their location in relation to the 1994–1996 excavations cannot be established with any precision. From Eddie Moth’s recollections, his 1989 north–south section must have been about 5m or so east of a deep section subsequently drawn in 1998, when the layers at the bottom of the stratified deposits were exposed in a small 2m-deep trench (see Chapter 2).

Figure 1.10. Flint scrapers and other artefacts from the 1989 excavations

Post-excavation work established that the plain, coarse pottery (Figure 1.9) from the middle layers of the Area C midden dates to the Late Bronze Age, on the basis of a radiocarbon date of 1390–1010 cal BC at 95% probability (OxA-3352; 2960±75 BP) on disarticulated cattle bone (see Table 14.1).⁶ Worked pumice and struck flakes of quartz and flint were also recovered, as well as a whale vertebra (Figure 1.10).

Figure 1.11. Area C in 1992 after extension of the sand quarry westwards, thereby obliterating the 1989 trench, photographed from the east

The quarry between 1990 and 1993

Sand quarrying continued and, after a visit in 1991, Eddie Moth reported that the site had been eaten into like an apple, with large chunks taken out of the mound on its east and west sides (Figures 1.11–1.12). In that same year, Uist Builders Construction Ltd reported the discovery of a human skeleton, disturbed while they were quarrying the mound, and Angie Foster (a former Sheffield student living on Barra) visited the find-spot, about 5m southeast of the midden in Area C, and collected the remains for analysis.

This skeleton, of a 9–12 year-old child, was incomplete (SK 4906; the osteology report appears in Volume 2). A femur from SK 4906 is dated to 1610–1385 cal BC at 94% probability (AA-48585; 3205±50 BP), within the Middle Bronze Age. In 1996 a test pit at the spot indicated by Angie Foster located a missing humerus and limb bones belonging to this skeleton but there was no sign of the context from which it came. It has been said locally that the bones of a second individual found in this area were reburied in the graveyard in 1992. There are other unverified local rumours of further skeletons being found during sand quarrying in Area C before 1994.

In 1992 local resident Don MacPhee took MPP to the site to show him more archaeological remains that had been exposed by quarrying and wind. Area C was as reported by Eddie Moth the year before but there were now quantities of pottery, animal bones and shells falling out of midden layers on the east side of the quarried mound and a well-preserved stone building exposed in the near-vertical face of the mound’s west side (Figure 1.12).

The 1994–1996 excavations

In June 1994 a small rescue excavation was carried out on Area C, exposing the remains of the stone building and the midden into which it was set (Figure 1.13). The structure was an Early Iron Age double roundhouse, at that time a unique find in the Western Isles; sadly, its east and west ends had been lost through quarrying. The excavation of the double roundhouse (House 150), its dating and associated finds are reported in Chapter 10, and the dating of the lower layers of the midden appears in Chapter 2. The house had filled with windblown sand which contained an assemblage of smashed Early Iron Age plain ware pots.

In 1995 the house floors and parts of the surrounding midden in Area C were excavated (see Section 1.6 for the excavation methodology applied to this and all other South Uist sites excavated by MPP et al. and by Sharples).

Although quarrying of this settlement mound north of the machair track had now ceased, and the surviving fragment of the mound was being planted with grass, a new and bigger sand quarry (Area A, also known as Site 55) was opened up on the south side of the track, about 150m to the southeast (Figure 1.14). Some of the sandy soil removed by machine from Area A was dumped to the west of the new quarry; since this sand contained archaeological finds, this caused some confusion as to whether yet another site lay here (recorded as Site 56 in the machair survey). By 1996 the clean sand on top of what was now incontrovertibly identifiable as a settlement, and some of the uppermost archaeological layers, had been quarried away in Area A, leaving a flat surface, about 20m east–west by 30m north–south. This was an exposed area of bone-rich midden, in which no pottery was visible (Figures 1.15–1.16). Arrangements were made with Uist Builders Construction Ltd and with the Dalabrog Grazing Committee for trial excavations to be carried out in 1996.

Figure 1.12. Area C in 1993 with the Early Iron Age house at the top of the sequence cut into Late Bronze Age midden layers, photographed from the west

The 1996 fieldwork included geophysical survey, final excavations of the double roundhouse (House 150) in Area C and trial excavations in Areas A and B. In Area C, sections were cut against the east and west faces of the upper midden layers (i.e. cut as vertical sections at the top of an eroding slope to record the midden layers’ stratigraphic sequence). All these midden layers (see Chapters 5 and 6) were earlier than the double roundhouse (see Chapter 10). The interior of the surviving portion of the house and the surface of the midden were turfed over by the project team to preserve House 150 for public interest.

An intermittent spread of bones, sherds and shells extended 600m north and north-northwest from Area C, most of it exposed in rabbit burrows skirting the eastern edge of a very large dune. Two concentrations of finds were identified within this spread (Sites 52 and 53; see Figure 1.3). Immediately south of Site 53 and about 20m to the east of the very large dune, a small sand blowout had exposed stonework and midden deposits; a small excavation trench was dug here in Area B. The results of this excavation are reported in Chapter 2.

On the quarry floor in Area A, Dr Mike Hamilton of Cardiff University surveyed up to seven 10m × 10m grids with a transverse interval of 1m and a sampling interval of 0.50m (800 readings per grid). The equipment used was a Geoscan FM36 fluxgate gradiometer, a Bartington MS2B magnetic susceptibility probe, and a RM4 resistivity meter with a DL10 data-logger. The data were processed on a Triumph-Adler Workstation 386 SX using Geoscan Geoplot 1.1 and 2 programs and printed on a Hewlett Packard Deskjet 340.

Although the geophysical results were good, it was only in retrospect, after several seasons of excavation, that they could be correctly interpreted (Figure 1.17). Three roundhouses (Houses 1370, 401 and 801) in a north–south line, and the forecourt of the central roundhouse, are visible as areas of low resistance. The central and northern roundhouses (House 401 and House 1370) are visible as areas of high magnetic susceptibility, but only House 401 can be detected in the magnetometer plot, as an area with low magnetic responses surrounding a highly magnetic anomaly (the roundhouse’s sequence of central hearths). Overall, the earth resistance plot provided the clearest indication of the outlines of the three sunken-floored roundhouses.

In Area A, the large expanse of midden exposed by the quarrying of sand from the centre of the mound was trowelled and planned in 1996. In the same season, three 1m-wide trial trenches (A1, A2 and A5) were excavated into this surface to establish what remains were preserved in the upper layers of the midden. In addition, two test pits (A3 and A4) were dug into the lower layers of the midden in the southern end of the quarried area.

Figure 1.13. Area C, showing the 1994–1996 excavation trench and the approximate position of the 1989 trench

Figure 1.14. The Area A sand quarry before archaeological excavation, viewed from the north

The only surface clues in 1996 were a few protruding stones, some of which turned out to be the final phase of House 401’s circular wall. At the centre of the quarried area, one of the trenches (A1) confirmed the presence of this roundhouse but no other structures were found. The other test pits demonstrated that the stratigraphic sequence was well over a metre deep. In retrospect, the smaller and deeper trenches did more harm than good. In particular, Trench A4 (2m × 1m) cut through the north side of the floor of House 801 without providing any indication at the time that this was the interior of a roundhouse.

Figure 1.15. The quarry (Area A) at the start of archaeological excavation in 1995 with Area C in the centre background (beyond and immediately left of the JCB), viewed from the southeast

Figure 1.16. The settlement mound of Site 55, showing the location of the sand quarry on its north side

Figure 1.17. Plots of geophysical surveys within the quarry before the start of archaeological excavation: magnetometer (top), magnetic susceptibility (middle), and earth resistance (bottom)

1.5 The evolving research design

The project results published in this pair of monographs in many areas surpass our original aims and objectives. The level of detail about the lives of the roundhouses’ inhabitants that the project specialists have extracted from the assemblages is more impressive than we could have hoped. Advances in scientific techniques in the first decades of the twenty-first century have given us information on such things as cattle aDNA, stable isotopes and the occurrence of mummification in British prehistory, information that was simply unforeseeable when the research design was rewritten for each field season between 1996 and 2003.

The reader should therefore bear in mind that this section lays out our thinking at the time. We have updated some of the references to comparative material, and for clarity need to mention some of our findings and conclusions, since these affected the development of the research design over the years of the project, but otherwise we present the research design as it was (i.e. we have not attempted to pretend that we could foresee the future, and no research design predicted finding the sub-floor burials!)

Aims and objectives 1996–1998

Prior to 1996 the rationale for investigation was a rescuedriven response to continuing damage to a settlement that appeared to be Late Bronze Age or Early Iron Age, a period poorly understood. That year the research aims were formalised around five main themes:

Period rarity

Bronze Age and Early Iron Age settlements, dating to the period prior to brochs and wheelhouses, are extremely rare in the Western Isles and the only comparison at that time was the poorly preserved horizon excavated at the Udal (Crawford 1986; Selkirk 1996; Armit 1996: 103–5). For the British Isles generally, there are few sites, most of them in Orkney, where there are substantial standing remains of houses from that period. Cladh Hallan also appeared to have remains of the Early Bronze Age, a period well-represented in South Uist by known sites although these were largely unexcavated (Figure 1.18).

Later prehistoric change from a nucleated to dispersed settlement pattern

With the completion of the South Uist machair survey (Parker Pearson 1996a), we were better able to assess the context of the settlement at Cladh Hallan. Whereas Middle Iron Age and later settlements are located at regular intervals along the machair of South Uist’s west coast, the Late Bronze Age/Early Iron Age settlements are found only sporadically. Early Iron Age deposits under the Dun Vulan broch (Parker Pearson and Sharples 1999: 50–5), at an islet settlement in Upper Loch Bornish (Marshall and Parker Pearson 2012) and at Kirkidale (Moreland 2012) as well as settlement mounds at Hornish Point (Barber 2003), Ormacleit (Site 10), Staoinebrig (Site 29), Smercleit (Site 74) and Machair Mheadhanach (Sites 136, 152, 154) are the only sites of this period (Figure 1.19). Site 136 is as large as Cladh Hallan (Site 55), about 80m in diameter and 3m high.

The distribution of these sites suggests three broad groups:

•a southern group at Hallan;

•a central group in the Dun Vulan area;

•a northern group in the Iochdar area.

This is a partial picture since not only does the pottery of this period weather very quickly on the surface, but LBA/ EIA layers may be buried deep within long-term occupied settlement mounds of ostensibly later date. Our working hypothesis in setting out a research design was, however, that the LBA/EIA settlement pattern was more irregular and nucleated than in later periods.

Figure 1.18. Early Bronze Age sites in South Uist

Domestic use of space: from double roundhouses to brochs and wheelhouses

A third avenue of research was the architectural context of this period, not only in terms of house construction and tradition but also the use of space in and around them. The double roundhouses are a unique and distinct style from a period before the division of Atlantic Scottish architecture into wheelhouses and brochs. We initially speculated that the forecourts of South Uist wheelhouses (Lethbridge 1952; Young and Richardson 1960; Fairhurst 1971) might even have derived from transformation of the eastern of such roundhouses’ two rooms. Our excavations revealed an absence of internal piers in any of the Cladh Hallan houses, indicating a significant transformation from open interiors in the Late Bronze Age and Earliest Iron Age into the cellular partitioning within wheelhouses of the Early and Middle Iron Age (Armit 2006).

Figure 1.19. Late Bronze Age and Early Iron Age sites in South Uist

Culinary practices, economy and social status

Whilst the excavations in Area C prior to 1996 had produced valuable evidence for construction, use and abandonment, including a ceramic assemblage smashed in the abandoned double roundhouse, very little research had been done on the food remains. The copious midden refuse from Area C included shells, animal bones and carbonised plant remains, and there were evidently deep and extensive archaeological deposits in Area A. Excavating these would offer an opportunity to fully characterise economy and culinary practices during a period for which such evidence is scarce. Furthermore, these practices could be compared to those of the Middle Iron Age (with the emerging culinary distinctions between brochs and wheelhouses; Parker Pearson et al. 1996; Parker Pearson and Sharples 1999: 353–5) and of the Early Bronze Age at sites such as Northton and Rosinish (Simpson 1971; 1976; Shepherd 1976; Shepherd and Tuckwell 1977).

Sunwise ordering of activities within roundhouses

In 1994 Andrew Fitzpatrick produced a sunwise interpretation of the use of space in British Iron Age roundhouses, based on the distribution of artefacts from the postholes of a roundhouse excavated at Dunston Park at Thatcham in Berkshire. His evidence can be criticised on the grounds that the Dunston Park house had no surviving floor, and that there is a depositional problem – did debris from interior activities end up in postholes? For Dunston Park, there is also the possibility that the house was actually two consecutive roundhouses, a later house built upon an earlier house. Fitzpatrick’s model has occasionally been denigrated for its strongly symbolic interpretation; some favour a purely functionalist explanation, arguing, for example, that doorway directions were dictated solely by practical considerations of shelter from prevailing westerly winds and maximising morning sunlight within the house interior (Pope 2007).

In spite of the caveats, the Fitzpatrick model has proved to be provocative and powerful: it potentially explains why roundhouses were round, why so many have their entrances facing east or southeast (Oswald 1997), and how the interior space might have been divided into a south side for daily activities and a north side for sleeping. Quite simply, the activities within the house were, according to Fitzpatrick, conducted in relation to the sun’s movement around the sky. After sunrise, food was cooked in the southeast. As the sun moved round to the south, daytime activities and crafts were carried out on the south side. Finally, as the sun set in the west and passed below the horizon to the north, the northern part of the house was where people slept.

During 1997 and 1998, we found preserved house floors in House 401, the central roundhouse in Area A. This thus offered the perfect circumstances to explore the Fitzpatrick model in better preserved contexts than the truncated site at Dunston Park. Here was an opportunity to find out whether the model applied to somewhere as distant as the Western Isles and, if so, whether it was regionally variant and whether or how its use changed through time.

In 1996 MPP and Niall Sharples commissioned Adrian Chadwick to illustrate a fourfold model of the Fitzpatrick scheme, its modification to fit the excavated evidence from Middle Iron Age wheelhouses for the diurnal cycle, a homologous model of the house as metaphor of the life cycle from birth to death, and a model of seniority and sacredness within the roundhouse (Figure 1.20; Parker Pearson and Sharples 1999: fig 1.10).

Refinement of the research design: 1999–2001

From 1999 onwards – for the final three years of excavation – a more comprehensive research design was employed, once the full size and complexity of Area A had been realised. By the end of the penultimate excavation season in 2001, we had recovered a sequence of occupation at Cladh Hallan of around a millennium.

The project’s aims and objectives related to seven themes, set out below. As mentioned above, there was no indication of the presence of burials below the houses until the last floor of each was excavated in 2001, and therefore the research design contained no aims concerning human remains.

Figure 1.20. The Fitzpatrick model and modifications, drawn in 1996. The ‘toss and drop’ zones in d) [after Binford 1978] were added during post-excavation. In a) and b) the descriptions around the edges of the house are unchanged from before excavation whilst the horizontal labels have been added as a result of the excavations

Chronology

The Late Bronze Age to Early Iron Age transition is a significant period in British prehistory. The Cladh Hallan settlement has a remarkable stratigraphic sequence from the Beaker period through to the bronze–iron transition and Early Iron Age, with corresponding changes in material culture, particularly in its large ceramic assemblage. Successive ceramic styles, spanning most of the second and first millennia BC, were found in Cladh Hallan’s exceptional sequence.

Use of space

The analysis of the many categories of artefactual and ecofactual material recovered from the Cladh Hallan house floors in large quantities using our intensive sampling techniques would provide a firmer empirical footing to the study of the use of space within roundhouses (see Foster 1989a; 1989b; Fitzpatrick 1994; 1997; Parker Pearson and Richards 1994; Giles and Parker Pearson 1999; Parker Pearson 1999; Parker Pearson and Sharples 1999: 16–23; Giles 2012: 85–90; 3; Sharples 2010: 174–237). Given the absence in the archaeological record of well-preserved roundhouse floors, rare for this period of British prehistory, the lack of evidence concerning the use of domestic space equally weakens the critiques of the sunwise model (e.g. Pope 2007; Webley 2007). With 17 house floor plans recovered from Cladh Hallan, this has substantially augmented the number of complete and well-preserved late second to first millennium BC house floors excavated in Britain. Other examples from Scotland include Dun Bharabhat (Harding and Armit 1990), Lairg (McCullagh and Tipping 1998), Carn Dubh (Rideout 1995), Cul a’Bhaile (Stevenson 1984) and Tofts Ness (Dockrill 2007).

Settlement layout

The Cladh Hallan settlement was a community of perhaps as many as six, seven or even eight, roundhouses forming a north–south terrace constructed and first occupied in the eleventh century BC (constructed in 1080–1020 cal BC [95% probability]; see Chapter 14).

Similar linear arrangements are known from a few sites in Britain such as Blackford, Perth & Kinross (O’Connell and Anderson 2021), Kintore, Aberdeenshire (Cook and Dunbar 2008: 86–97), Lairg, Sutherland (McCullagh and Tipping 1998), Black Patch, East Sussex (Drewett 1982), Itford Hill, East Sussex (Burstow and Holleyman 1957; Ellison 1978; Sharples 2010: 224–5), Scarcewater (Jones 2015) and Trethellan Farm, Cornwall (Nowakowski 1991), all in the Middle Bronze Age. Blackford and Kintore also have groups of round houses con tinuing into the Late Bronze Age and Early Iron Age while those at Douglasmuir, Angus, date to the Early Iron Age (Kendrick et al. 1995). In all cases, the exact degree of contemporaneity between houses is often difficult to establish with certainty without close stratigraphic control. An exception, although not in a linear arrangement, is the group of roundhouses in the waterlogged and burnt-down Late Bronze Age settlement of Must Farm, Cambridgeshire (Knight et al. 2019). Cladh Hallan with its party walls and deep stratigraphy has important implications for our knowledge and understanding of a settlement’s formation and organisation over many centuries during the Bronze Age and Iron Age.

Structured deposition

The Cladh Hallan houses provide some remarkable instances of structured deposition. Sheep, dogs, antlers, cattle scapula shovels and pots all received careful treatment in burial and placing. There are also distinct differences in deposition between households. Structured deposition was an important aspect of Bronze Age and Iron Age social life which requires high-quality evidence and careful interpretation of its potential complexities (Wait 1985; Hill 1989; 1995; Bradley 2003; Brück 1999; 2006; Brudenell and Cooper 2008; Garrow 2012).

Origins of east-facing doorways

The easterly orientations of the Cladh Hallan roundhouse doorways continued throughout the settlement’s life. Elsewhere in Britain, east-facing roundhouse entrances were considered (at the time of writing the research designs in 1999–2001) to have been preceded in the Middle Bronze Age by south-southeast facing doorways (Parker Pearson 1996b; 1999; Oswald 1997). Although we now know that regular use of east-facing doorways began in the Early Bronze Age (Hannah Bullmore pers. comm.; in prep.), we had hoped that the beginning of the Cladh Hallan sequence might shed light on such origins.

Milking

The controversy about whether milking was important in Hebridean prehistory was unresolved when excavations began (Parker Pearson et al. 1996; 1999; Gilmour and Cook 1998) and we hoped that this could be resolved using data from Cladh Hallan, with studies of lipid and protein residues on the ceramics (Craig et al. 2000; 2005) being integrated with studies of the animal bone assemblage.

Agriculture and stock-raising

Cladh Hallan provided important information on agriculture and stock management which, thanks to other excavations on South Uist, can now be viewed within the long-term perspective of the last four thousand years (Smith and Mulville 2004). The large proportion of deer bones at Cladh Hallan overturned ideas about the absence of deer on the Uists in prehistory (Mulville 1999: 273).

1.6 The 1997–2003 excavations

Uist Builders Construction Ltd had been granted planning permission to quarry an area to the east of the site which was not archaeologically sensitive. However, sand was still extracted from Area A after 1995 by other people until excavations recommenced in June 1997. After that quarrying stopped, except for a brief revival in 1999, an episode in which a large lorry became stuck in the quarry and leaked diesel oil into the sand above the southeast quadrant of House 1370.

From 1998 onwards, each excavation season at Cladh Hallan was expected to be the last. Yet every year the site produced new phases of activity and new structures; every year we thus had to persuade Historic Scotland⁷ to contribute to another season’s fieldwork. Thanks to the support of successive regional inspectors, who recognised both the complexity of the site and its importance, the project continued until the last major field season in 2002.

Methodology

House 401 was excavated in four quadrants, divided by north–south and east–west sections bisecting the house. Each opposing quadrant was excavated in stages, so the depth of each section increased over each field season (see Figure 1.23 for the quadrants under excavation). The section lines were maintained and re-established each year to ensure stratigraphic recording of two crosssections through the entire depth of the house. This was difficult and time-consuming but struck a balance between recording (and sampling) in plan and recording in section. By contrast, the floors of Houses 801, 1370, 640, 150 and other minor structures with shallower stratigraphy were excavated in plan.

Archaeological deposits were excavated by hand and sieved on site through a 10mm mesh to standardise recovery. At Dun Vulan we had developed an integrated environmental and scientific approach that allowed us to investigate these kinds of house floors in considerable detail, gridding them at half-metre intervals for geochemical and geophysical sampling and for intensive flotation of the entire floor (Parker Pearson and Sharples 1999; Smith et al. 2001). Each floor and outside ‘yard’ surface at Cladh Hallan was thus sampled in its entirety in 0.50m × 0.50m squares (with samples for magnetic susceptibility and phosphorus taken at 0.50m intervals across each floor). One exception was floor 595 (phase 12) in House 401, where only the southeast and northwest quadrants were sampled on this half-metre grid because of concerns in that field season about lack of time and resources. Also, in the early days of the excavation of Area A, the first two floor layers encountered within House 401 (floors 452 [phase 15] and 462 [phase 14]) were excavated by quadrant to maintain stratigraphic control, but not sampled in half-metre squares – there are therefore no distribution plots for these floors, other than those finds that were spatially recorded by co-ordinates.

In addition to samples from the house floors, samples for flotation (normally c. 30 litres or more), soil micromorphology, OSL dating, magnetic susceptibility and soil phosphate analysis were taken from other deposits. As mentioned above, all deposits were dry-sieved on site through a 10mm mesh, other than entirely sterile windblown sand layers and those contexts that were removed entirely as samples for environmental and other scientific