Character of the Middle and Late Bronze Age in Wiltshire

Wiltshire is characterized by upland areas of chalk downs cut by river valleys. From the Middle Bronze Age, extensive field systems, often covering hundreds of hectares, are formalized on the downs (McOmish et al., Reference McOmish, Field and Brown2002). Outside these field systems, extensive areas of open grassland also existed (Gingell, Reference Gingell1992; Bradley et al., Reference Bradley, Entwistle and Raymond1994; Allen & Entwistle, Reference Allen, Entwistle, Fulford, Powell, Entwistle and Raymond2006; Birbeck, Reference Birbeck2006) providing large stretches of less regulated pasture.

Middle Bronze Age settlements are traditionally thought to have had a standard form, characterized by a small rectilinear enclosure or partial enclosure set within the field systems (Ellison, Reference Ellison, Hodder, Isaac and Hammond1981). They contained of one or two roundhouses and a small number of ancillary structures such as pits or multi-post constructions. Typical examples are Thorny Down (Figure 2) (Stone, Reference Stone1937; Ellison, Reference Ellison1987), Fargo Wood (Richards, Reference Richards1990), and Preshute Down (Piggott, Reference Piggott1942). We now know, however, that some Middle Bronze Age settlements such as Old Sarum Spur (Figure 3) (Powell et al., Reference Powell, Allen, Chapman, Every, Gale, Harding, Knight, Mckinley and Stevens2005) and Blenches Mill (Wessex Archaeology, 2004) were unenclosed, while others such as Rockley Down had both enclosed and open phases (McOmish, Reference McOmish, Brown, Field and McOmish2005). The field systems would have been used for both arable and pastoral purposes. Being bounded and measurable, these fields doubtless formed part of well-defined systems of tenure and rights that may have shifted through the agricultural cycle and with exploitation strategies. The large areas of open grassland outside these field systems probably had more fluid rights of access and grazing. The settlements appear to have had a mixed agricultural regime where households maintained some degree of independence in terms of subsistence (Ellison, Reference Ellison, Hodder, Isaac and Hammond1981), although access to the open pastures hints at supra-household levels of interaction, negotiation, and co-operation. Similarly, households were tied into wider society through systems of gift exchange in which bronze circulated (Brück, Reference Brück2000: 290).

Figure 2. Thorny Down: typical Middle Bronze Age enclosure (drawn by the author, adapted from Stone, Reference Stone1937).

Figure 3. Open Middle Bronze Age settlement on Old Sarum Spur (drawn by the author, adapted from Powell et al., Reference Powell, Allen, Chapman, Every, Gale, Harding, Knight, Mckinley and Stevens2005).

Late Bronze Age settlements are typically unenclosed, hence less visible, and several sites possess more structures than their Middle Bronze Age counterparts. Excavations at Dunch Hill revealed four roundhouses along an 8 m wide, 300 m long transect (Figure 4) (Andrews, Reference Andrews2006), while to the east of Quebec Farm activity was identified over a 100 m long stretch of track built for modern army tanks (Ellis & Powell, Reference Ellis and Powell2008: 174). Similarly, excavations to the south of Ford Road in Salisbury (Figure 5) uncovered three roundhouses over a 40 m stretch of a pipeline trench (Powell et al., Reference Powell, Allen, Chapman, Every, Gale, Harding, Knight, Mckinley and Stevens2005). The narrow, linear nature of all these excavations indicates that only a fraction of features was encountered and it is possible that these sites resemble the site of Shorncote Quarry on the Gloucestershire-Wiltshire border, where at least forty-two roundhouses have been identified (Hearne & Heaton, Reference Hearne and Heaton1994; Hearne & Adam, Reference Hearne and Adam1999). These sites imply an increase in size and longevity of settlement compared to those of the Middle Bronze Age. At present it is unclear whether these sites are exceptions or whether they are representative of a more normal and widespread form of settlement.

Figure 4. Late Bronze Age features uncovered over an 8 m wide, 300 m long transect at Dunch Hill (drawn by the author, adapted from Andrews, Reference Andrews2006).

Figure 5. The Late Bronze Age settlement to the south of Ford Road, Salisbury (drawn by the author, adapted from Powell et al., Reference Powell, Allen, Chapman, Every, Gale, Harding, Knight, Mckinley and Stevens2005).

A shift in the mode of agricultural production accompanies this change in the character of inhabitation during the Late Bronze Age. Linear earthworks are constructed across the downs, the longest examples being Old Nursery Ditch and Old Ditch West stretching up to 11 km and 16 km respectively (McOmish et al., Reference McOmish, Field and Brown2002: 61). Evidence from Quarley Hill (Hawkes, Reference Hawkes1939) and Windy Dido (Cunliffe, Reference Cunliffe2000) shows that many were constructed in a piecemeal fashion. Although unlikely to have originally been conceived as a grand system, over time elements were added and removed to fit a much larger project. The chronology of most of these earthworks is relative but, where excavations could ascertain their construction dates, they mostly appear to have been established in the later stages of the Late Bronze Age (Bradley et al., Reference Bradley, Entwistle and Raymond1994). On Snail Down and Tidworth Down, linear earthworks cut obliquely across earlier field systems, suggesting that fields had gone out, or were put out, of arable use. This has led to assumptions that they represent a shift towards pastoralism (Bowen, Reference Bowen, Limbrey and Evans1978; Brück, Reference Brück, Haselgrove and Pope2007), but the relationship is not a simple one. On Cherhill Down, Overton Down, and along the Old Ditch West, the earthworks separate field systems from areas of open pasture (Fowler, Reference Fowler2000: 224; McOmish et al., Reference McOmish, Field and Brown2002: 64; Kirkham, Reference Kirkham, Brown, Field and McOmish2005). At Dunch Hill, asymmetric lynchets either side of a linear ditch imply that arable production was abandoned on one side but continued on the other (McOmish et al., Reference McOmish, Field and Brown2002: 63). Therefore, while linear earthworks are indicative of an increase in pastoral exploitation of the land, they do not represent the complete abandonment of arable production on the downs.

Social Reproduction in the Middle to Late Bronze Age

Fundamental to the discussion of Middle Bronze Age society is the model where communities compete for economic and political power through the local and interregional exchange of bronze (Rowlands, Reference Rowlands, Barrett and Bradley1980: 33). This is underpinned by an evolutionary model of society that postulates that societies are constituted through their productive relationships and are reproduced through the social structures that dominate the productive processes (Friedman & Rowlands, Reference Friedman, Rowlands, Friedman and Rowlands1977: 203). Productive success, in the form of agricultural surpluses and the ability to throw communal feasts, is seen as closely linked with supernatural forces, in particular those of the ancestors. Groups whose ancestors are able to provide supernatural assistance in terms of superior production are of demonstrably superior lineage. The production of economic surpluses thus indicates the ‘supernatural proximity’ and ‘genealogical superiority’ of one's ancestors (Friedman & Rowlands, Reference Friedman, Rowlands, Friedman and Rowlands1977: 207). This genealogical superiority is reflected in the higher social value of (specifically female) marriage partners in matrimonial exchanges. In return for the chance of buying into a superior lineage a bride price is agreed and this forms a channel for the circulation of wealth (Friedman & Rowlands, Reference Friedman, Rowlands, Friedman and Rowlands1977: 208). This system of exchange acts as a mechanism through which society is tied together through a mesh of matrimonial alliances. These allow groups to control trade and manipulate bronze in systems of competitive gift exchange to accumulate wealth and political influence, although the gift exchange system is ultimately based on agricultural production and supernatural support (Barrett, Reference Barrett, Barrett and Bradley1980a; Rowlands, Reference Rowlands, Barrett and Bradley1980). Bronze thus fulfilled a key role within the reproduction of the relations of production through which society was organized. It was a socio-political currency rather than just a commoditized store of wealth or pseudo-currency (Needham, Reference Needham, Haselgrove and Pope2007) and, although along with other valuables or prestige items it would not in itself be the foundation of rank, it symbolized status (Friedman & Rowlands, Reference Friedman, Rowlands, Friedman and Rowlands1977: 208). While small local groups were largely self-sufficient entities during the Middle Bronze Age (Brück, Reference Brück2000: 290), they were tied into a larger social arena as their social reproduction relied on production (of bronze and matrimonial exchanges) based outside their territory.

The end of this system is equated with the later stages of the Late Bronze Age, when large quantities of bronze were deposited into the ground as hoards. The quantity of material in hoards of the Ewart Park metalwork tradition is more than five times that of earlier phases, and most of this late material is dated to the tradition's latest phases, i.e. 900–800 bc (Needham, Reference Needham, Haselgrove and Pope2007: 53). While some consider that this period of hoarding represents a rapidly accelerating system of votive deposition as a means of negotiating and maintaining power (e.g. Bradley, Reference Bradley1990), votive deposition (as distinct from other forms of hoarding) cannot be identified in the quantities required for it to form a mechanism for the continuation of political power (Barrett, Reference Barrett, Jones, Pollard, Allen and Gardiner2012: 14).

Needham (Reference Needham, Haselgrove and Pope2007) argues that by the time we reach the Llyn Fawr metalwork phase, bronze's supra-functional quality as indicator of status seems to have been superseded. The preceding period of hoard deposition represents the removal of material supporting this supra-functional role from circulation and, from this point on, bronze is almost solely used to produce functional goods such as tools, cauldrons, and weapons. Needham presents a strong argument, proposing that the increased period of hoarding results from a crisis of confidence in the ability of bronze to continue fulfilling its social or supra-functional roles. Although the system's breakdown possibly happened over several generations, the period of greatest bronze deposition represents the point at which bronze fails as a primary driver of social reproduction. With bronze no longer acting as an impetus for the creation of alliances and facilitating the flow of wealth around society, we would expect a change within the relations of production and the reproduction of the structures that govern these. To understand this transformation in reproductive institutions, it is essential to know what comes to replace bronze.

The reorganization of the landscape in the latter stages of the Late Bronze Age reflects a change in the means of production and a more mobile way of life dominated by animal husbandry. It supported the pasturing of animals, opened up the landscape, and promoted movement over longer distances (Tullett, Reference Tullett, Sterry, Tullett and Ray2010a). The dating of these changes to the latter stages of the Late Bronze Age (900–800 bc) makes them concurrent with the major period of bronze deposition (Needham, Reference Needham, Haselgrove and Pope2007). This suggests that the change in the means of agricultural production was contemporary with the fall from grace of bronze, and it is tempting to see the two factors as connected. It shows an emphasis on animals, with an ultimate evolution being reached in the Earliest Iron Age.

Earliest Iron Age Evolutions

At the onset of the Earliest Iron Age (800–600 bc) there are elements of continuity in landscape exploitation, together with further evolution in settlement patterns and dramatic changes in material culture. A small number of downland settlements are identifiable at Overton Down (Fowler, Reference Fowler2000), Cow Down, Longbridge Deverill (Chadwick-Hawkes, Reference Chadwick-Hawkes1961), Battlesbury (Ellis & Powell, Reference Ellis and Powell2008), and Cockey Down (Trott, Reference Trott1991; Lovell, Reference Lovell1999); however, these downland sites appear to be exceptions, with activity concentrated around the Vale of Pewsey where a series of black earth sites representing large midden deposits have been identified. Those that have been excavated, like All Cannings Cross, Potterne, East Chisenbury, and Stanton St Bernard, have produced evidence for deposits of stabling waste (Macphail, Reference Macphail and Lawson2000, Reference Macphail, McOmish, Field and Brown2010) up to 2 m thick yielding large assemblages of pottery, animal bone, and artefacts. The bone assemblages imply substantial permanent populations with significant seasonal augmentation (Serjeantson, Reference Serjeantson, Haselgrove and Pope2007; Serjeantson et al., Reference Serjeantson, Bagust, Jenkins, McOmish, Field and Brown2010). It illustrates that the inhabitants of this landscape consisted of both fixed sedentary core populations and fluid mobile elements. This reorientation of settlement from downs to valley has been associated with dairy production (Serjeantson, Reference Serjeantson, Haselgrove and Pope2007; Tullett & Harrison, Reference Tullett, Harrison, Baker, Allen, Middle and Poole2008) as well as the exploitation of rich sources of iron ore (Barrett & McOmish, Reference Barrett, McOmish, Roulière-Lambert, Daubigny, Milcent, Talon and Vital2009).

A range of less easily categorized sites are also coming to light on the downs. They usually consist of just a few features: one or two shallow pits or postholes and the occasional hearth. Finds are rare, with occasional pieces of pottery and animal bone. To the west of Breach Hill, three hearths, eight shallow pits, and four stake holes were found spread along a 550 m long stretch of the Old Ditch linear earthwork (Birbeck, Reference Birbeck2006). At Odstock Road, Britford (Figure 6), two shallow pits tentatively dated to the Earliest Iron Age were found next to a trackway and another linear ditch (Wessex Archaeology, 1997). At Boreham Farm (Figure 7) two pits and postholes next to a seasonal palaeochannel produced six fragments of Late Bronze Age ceramics and two sherds of Earliest Iron Age pottery (Ellis & Powell, Reference Ellis and Powell2008: 141–44). Other similar sites dated to this period exist at Dunch Hill (Bradley et al., Reference Bradley, Entwistle and Raymond1994) and underlying later occupation at Coombe Down North (Fulford et al., Reference Fulford, Powell, Entwistle and Raymond2006) and Oliver's Camp (Cunnington, Reference Cunnington1908). Given the ephemeral nature of these sites and lack of datable finds, it is likely that many more are yet to be discovered or have been overlooked as undated features. It seems highly probable that similar types of sites outside Wiltshire have also received scant attention for these reasons. None of these examples can be taken to represent permanent settlement and the remains have usually been ignored or merely cursorily recorded. Although the features do not form obvious structures, the sites have a direct relationship with linear features such as linear earthworks, trackways, and palaeostreams, all of which are associated with the movement and management of animals. Located away from the known settlements, they appear to represent the ephemeral remains of temporary camps associated with animal husbandry systems (Tullett, Reference Tullett, Sterry, Tullett and Ray2010a).

Figure 6. Trackway uncovered next to a linear ditch on land off Odstock Road, Britford (drawn by the author, adapted from Wessex Archaeology, 1997).

Figure 7. Late Bronze Age/Earliest Iron Age features identified next to a seasonal palaeochannel at Boreham Farm (drawn by the author, adapted from Ellis & Powell, Reference Ellis and Powell2008).

The first early hilltop enclosures, the precursors of the hillforts, are established during this transitional period. Martinsell on the northern margin of the Vale of Pewsey uses a major linear earthwork as its northern boundary (Cunliffe, Reference Cunliffe2004) and is built with a relatively slight perimeter bank and ditch. A magnetometer survey of the interior revealed no visible features, suggesting that it was never intensively occupied (Payne et al. Reference Payne, Corney and Cunliffe2006, 118–23). It represents the best example of an early hilltop enclosure in the region, primarily because it never evolved into a later hillfort and was thus not extensively altered. Liddington Castle, where excavations recovered Earliest Iron Age ceramics (Hirst & Rahtz, Reference Hirst and Rahtz1996), is a second example. The excavators also revealed a putative palisade trench inside the main bank (Payne et al., Reference Payne, Corney and Cunliffe2006: 118); it is possible that this represents a first phase before the site was remodelled, into a hillfort with ramparts. Like Martinsell, it lies on the edge of a scarp and is associated with linear earthworks. Sidbury is a further possible example, occupying a nodal point in a group of linear ditches that predate its construction. A small trench to assess damage to the rampart recovered ‘Iron Age A/B sherds’ (Megaw, Reference Megaw1967) but Early All Cannings Cross wares were found in the area around the hilltop, raising the possibility of an earlier origin (Bradley et al., Reference Bradley, Entwistle and Raymond1994).

The slight perimeters of sites such as Martinsell have been taken to indicate a primary function associated with communal stock management rather than defence (Cunliffe, Reference Cunliffe2004). Their relationship with linear earthwork networks would also support the notion that they formed part of large-scale animal management strategies.

Ceramic styles change dramatically across the transition period with a transformation from plain wares to highly decorated Early All Cannings Cross wares. Late Bronze Age Plain Ware clearly grows out of the Middle Bronze Age Deverel-Rimbury tradition, with little innovation in production technology or materials. It is also a largely qualitative distinction often defined through relative proportions of decorated wares across assemblages (Barrett, Reference Barrett1980b). Early All Cannings Cross wares, however, manifest innovation in their production and fabric (Middleton, Reference Middleton1987), with a notable increase in decoration and a greater range of forms such as bowls and cups (Barrett, Reference Barrett1980b). The disjunction between the styles implies a change in ceramic technology, in concepts about how pottery should be manipulated, and in its social positioning.

Bronze's False Heirs

Several new drivers for social reproduction have been posited as replacements for bronze on the basis of some of the most visual elements of Iron Age archaeology: pottery, grain storage pits, and hillforts. A rise in the number of features associated with the storage of grain such as storage pits and four- or six-post raised granaries is seen as originating at the start of the Iron Age (e.g. Gingell, Reference Gingell1992: 158; Cunliffe, Reference Cunliffe2004: 76; Bradley, Reference Bradley2007: 234). This has been taken to represent an intensification of agricultural production with the manipulation of grain surpluses as a social device.

Barrett (Reference Barrett, Stig Sørensen and Thomas1989) points to the increase in decoration and the introduction of new forms of pottery associated with the preparation and service of food as indicative of consumption taking place in increasingly public arenas. He relates this phenomenon with a shift from the gift exchange of bronze to commensal production and consumption as the chief mechanism of social reproduction. Here it is the obligations created during feasts rather than gift exchange that dominate social relations, drawing on fertility within the arable cycle (Barrett, Reference Barrett, Stig Sørensen and Thomas1989: 309).

Cunliffe (Reference Cunliffe2005) links the large number of grain storage features within hillforts such as Danebury to the centralization of communal grain storage and its exploitation to acquire exotic or non-local resources. Thomas (Reference Thomas1997) states that an intensification of arable production is ‘broadly’ coincidental with a move to enclosed settlement and indicates control over land and arable production, replacing long-distance (bronze) exchange networks as the basis of social power. Williams (Reference Williams2003) hypothesises that, for the later prehistoric period, society manipulated the idea of the unchanging agricultural cycle as a metaphor for the timeless nature of social reproductive systems. He suggests that the metaphor was used to imply the permanence of social practice, and in particular of grain as especially emblematic to Iron Age society.

The dating of the escalation of grain storage facilities is often fudged into an ‘Iron Age’ or ‘first half of the first millennium bc’ time span without clearly identifying when the phenomenon starts. It is thus pertinent to assess prehistoric grain storage facilities on sites in Wiltshire. Not just any pit will do to successfully store grain. For the Iron Age, we believe grain would probably only have been stored within large beehive-, bell-, or barrel-shaped pits rather than in pits with a smaller cylindrical or bowl-like profile (Reynolds, Reference Reynolds1979). Thomas has likewise argued that the shallow, bowl-like nature of most Neolithic pits would preclude them from storing grain (Thomas, Reference Thomas1991: 63). Depth and profile seem key to the interpretation of pits as grain stores.

In Wiltshire, pits are rare on Late Bronze Age and Earliest Iron Age sites. Excavations (admittedly sometimes covering only limited areas) of Late Bronze Age sites at Dean Bottom, Rockley Down, Burderop Down, and to the east of Fargo Wood (Richards, Reference Richards1990; Gingell, Reference Gingell1992) failed to reveal any contemporary pits as did those of Earliest Iron Age sites at Erlestoke, Blackpatch, Knoll Plantation, and Longbridge Deverill Cow Down (Foster & Roddham, Reference Foster and Roddham2002; Chadwick-Hawkes, Reference Chadwick-Hawkes, Hawkes and Brown2012). Where pits were found on Late Bronze Age sites, they turned out to be slight: three pits excavated to the north-east of Kill barrow (out of ten revealed during stripping of the topsoil) had depths ranging from 0.14 to 0.17 m (Wessex Archaeology, 1995). At Dunch Hill, five out of nine pits were shallow scoops, the other four being less than 0.4 m deep (Andrews, Reference Andrews2006). An evaluation in advance of construction of the Westbury bypass revealed two relatively insubstantial Late Bronze/Earliest Iron Age pits, 0.3 m and 0.44 m deep (Wessex Archaeology, 2004). Of the five pits excavated at the Late Bronze/Earliest Iron Age Triangle Site, South Marston, four were shallow (depths: 0.24 m, 0.32 m, 0.32 m, and 0.64 m) and one was larger: 2.55 m in diameter, 0.97 m deep, with a bowl-shaped profile (Wessex Archaeology, 2011). Similarly, the pits identified in the deposit at Potterne were noted as shallow and unsuitable for the storage of grain (Lawson, Reference Lawson2000). Exceptions do occur and at Combe Down North excavations revealed a 0.8 m deep Late Bronze/Earliest Iron Age pit, next to a 2 m deep Earliest/Early Iron Age pit probably relating to a pre-enclosure phase of activity (Fulford et al., Reference Fulford, Powell, Entwistle and Raymond2006: 30–31).

Larger multi-period sites have produced more evidence for pits. At Boscombe Down West numerous Iron Age pits were found but only one, Q15, could be placed with any certainty in the Earliest Iron Age compared to forty-five+ for the Early Iron Age (Richardson, Reference Richardson1951). While sixty pits were dated to a broad Late Bronze–early Middle Iron Age period at Battlesbury Bowl, only seven (further) pits could be definitively dated to a Late Bronze/Early Iron Age phase. Furthermore, these fell within a depth range of 0.45 to 0.7 m (mean 0.5 m) (Ellis & Powell, Reference Ellis and Powell2008: 31–32). Similarly, at Scotland Lodge the eight Early Iron Age pits only averaged a depth of 0.42 m (Wessex Archaeology, 2002; Leivers & Moore, Reference Leivers and Moore2008). At All Cannings Cross seventy-five pits were excavated but it is impossible to separate them into Earliest and Early Iron Age phases (Cunnington, Reference Cunnington1923). A similar problem of attribution to the two phases was encountered at Latton Lands in the Upper Thames Valley where at least twelve pits were associated with the combined period (Powell et al., Reference Powell, Laws and Brown2009). On both sites most pits probably belonged to a post-600 bc phase.

What these multi-period sites indicate is that, although there are exceptions, most pits fall into the post-600 bc date range; large pits become a pervasive feature only in the Early Iron Age. At Codford Circle, where thirty-four pits have been identified, the only excavated example, 3.6 × 2.4 m across and 2.5 m deep, is dated to the Early Iron Age (Allen & Gardiner, Reference Allen and Gardiner2006). Highfield, Salisbury, contained sixty pits with depths varying between 1.5 and 3 m (Stevens, Reference Stevens1934: 582) and the 107 pits at Fifield Bavant had a depth of 1.04 to 2.46 m, with seventy over 1.52 m deep (Clay, Reference Clay1924: 487–96). Other contemporary sites have yielded large numbers of pits: Swallowcliffe Down (ninety-three; Clay, Reference Clay1927), Groundwell West (c. 200; Walker et al., Reference Walker, Langton and Oakey2001), Bodenham Hill Plantation (c. 150; Borthwick & Canham, Reference Borthwick, Canham, Catherall, Barnett and McClean1984), the Headlands enclosure (c. 200; Linford, Reference Linford2004), and Little Woodbury (c. 360; Bersu, Reference Bersu1940: 64).

Early four-post structures are equally rare, with two identified at the Late Bronze Age settlement of Dunch Hill (Andrews, Reference Andrews2006) and a single example from the Earliest Iron Age Triangle Site, South Marston (Wessex Archaeology, 2011). Elsewhere, there is a general problem of dating four-post structures on multi-phase sites where they could easily belong to any period of occupation from the Earliest to the Middle Iron Age, something that applies to the twenty-three four-post structures at Latton Lands (Powell et al., Reference Powell, Laws and Brown2009), the two at All Cannings Cross (Cunnington, Reference Cunnington1923), the three at Groundwell West (Walker et al., Reference Walker, Langton and Oakey2001), the ten at Battlesbury (Ellis & Powell, Reference Ellis and Powell2008), and the seven at Groundwell Farm (Gingell, Reference Gingell1982).

Hillforts reflect this pattern, and those that may have early phases such as Martinsell or Oliver's Camp have failed to produce evidence for pits during a programme of geophysical survey (Payne et al., Reference Payne, Corney and Cunliffe2006); as for the two pits excavated at Liddington Castle, they were of Early Iron Age date (Hirst & Rahtz, Reference Hirst and Rahtz1996). Other hillforts where geophysical survey has suggested extensive areas of pitting, such as Oldbury or Barbury, are also undoubtedly of later Early or Middle Iron Age date (Payne et al., Reference Payne, Corney and Cunliffe2006).

From this it is clear that before 600 bc most pits dug on settlements in Wiltshire were cylindrical and tended to have an average depth of less than 0.5 m, much smaller than the classic Wessex grain storage pits. It does not mean that larger isolated examples do not occasionally occur, nor does it necessarily preclude the use of the smaller pits for the storage of grain, but it does suggest that smaller volumes were stored than would be the case in later periods. This is true not only for Wiltshire's ‘upland’ chalk regions but also the ‘lowland’ gravels of the Upper Thames valley.

Hillforts constitute a further mechanism that has traditionally been considered a driver of social reproduction. Sharples (Reference Sharples, Haselgrove and Pope2007) points to the conspicuous consumption of resources associated with their creation and renovation as forming the new medium of social competition and social reproduction. The gathering of communities for such events would enable the exchange of information and knowledge between groups and allow the strengthening of social bonds. Sharples focuses on the Early Iron Age and later, largely because his model is settlement-based and his study area (Dorset) had few transitional settlements for him to work with (Sharples, Reference Sharples1991). J.D. Hill on his part views hillforts as a device for the congregation of scattered households and the enactment of social rituals that facilitate social reproduction (Hill, Reference Hill, Hill and Cumberpatch1995, Reference Hill, Champion and Collis1996). In both models, hillforts are key to social reproduction.

In Wiltshire hillforts such as Figsbury (Cunnington, Reference Cunnington1927), Yarnbury (Cunnington, Reference Cunnington1933), Vespasian's Camp (Hunter-Mann, Reference Hunter-Mann1999), Old Sarum (Montgomerie, Reference Montgomerie1947), Ogbury Camp (Grinsell, Reference Grinsell and Pugh1957), and Scratchbury (Annable, Reference Annable1958) all had an initial phase of construction in the Early Iron Age (600–350 bc). Martinsell, Liddington, and Oliver's Camp possibly had earlier phases but as rather slight hilltop enclosures. Other undated examples are also likely to have been established in the Early Iron Age or during a later phase of hillfort development in the Middle-Late Iron Ages; consequently the main period of hillfort construction in Wiltshire lies after the transitional period.

With an end to the role of bronze between 900 and 800 bc and a rise of hillforts and grain storage facilities after 600 bc we are faced with a lacuna during which a further form of dominant project must have existed among the currently identified social devices.

Social Reproduction across the Transition

At one level, Middle Bronze Age households appear to have had a high degree of control over their means of production including land and labour denoting that they were largely self-sufficient in terms of subsistence and craft production (Brück, Reference Brück1999). However, participation in the bronze exchange system would clearly have meant some dependence on external groups, possibly beyond that of the immediate communal group. Maintaining access to seasonal pastures or raw materials would probably have needed the power of a communal group, and the demands of the agricultural calendar would have meant that households were tied into a reciprocal labour system (Tullett, Reference Tullett, Sterry, Tullett and Ray2010b). Settlements may superficially seem largely self-sufficient, with production based on the household, but the reality appears to be a more complex, tiered system of reciprocal obligations.

The growth in the size of settlements from the Late Bronze Age onwards suggests that there has been a move from household or small extended family groups to cohabitation in multi-household or broader, extended family groups. This would have facilitated some pooling of labour and allowed land that was perhaps a patchwork of tenurial claims to continue to be exploited from a central location. There is evidence of a greater concentration of craft production at midden sites, which made it possible to share skills, knowledge, and expertise throughout the community (Brück, Reference Brück, Haselgrove and Pope2007). New ceramic forms were integral to this increased scale of productive unit, with food forming a medium to negotiate meetings between groups and manage these multiple relations (Tullett & Harrison, Reference Tullett, Harrison, Baker, Allen, Middle and Poole2008).

Cunliffe posits that, prima facie, there appears to be ‘a massive desertification of the chalklands in the early first millennium’ (Cunliffe, Reference Cunliffe2000: 202), the result of a decline in soil fertility or climatic deterioration (Cunliffe, Reference Cunliffe2004), leading to an emphasis on the chalklands’ exploitation as pasture and a focus on control over fertility as exhibited by the sites of the Vale of Pewsey (Cunliffe, Reference Cunliffe2000: 202). Although midden deposits are known outside the Vale of Pewsey, their scale, with the exception of Runnymede Bridge (Needham & Done, Reference Needham and Done1991), never matches that of the Wiltshire sites, hinting that other factors are at play in this region (Barrett & McOmish, Reference Barrett, McOmish, Roulière-Lambert, Daubigny, Milcent, Talon and Vital2009).

Barrett has tried more recently to move the discussion away from exchange-based models; he suggests a continuity or growth of Bronze Age agricultural systems into the Iron Age that reflects the continuation of social reproduction via stable ‘values embodied in the practices of selfhood and identity’ through agricultural practice (Barrett, Reference Barrett, Jones, Pollard, Allen and Gardiner2012: 14). But, rather than stability, there are significant changes to the mode of agricultural production. The reorganization of the landscape also points to an increase in the scale of social groups that could be mobilized for projects. The substantial blocks of land created by the linear earthworks indicates land tenure beyond the reach of individual households and implies a scale of ownership at a community level which was able to mediate the rights of access and use by its members (Tullett, Reference Tullett, Sterry, Tullett and Ray2010b). They also indicate the mobilization of a communal labour supply that doubtless depended on serving the interests of the participants with a system that involved large herds of animals. This is supported by evidence from Potterne, East Chisenbury, All Cannings Cross, and Stanton St Bernard. These sites all produced huge bone assemblages, which represent considerable numbers of animals beyond the scale of individual household ownership or management, and required a large workforce working together for their common interest. Transhumance presents an image of a mobile, fluid society organized around animal husbandry practices. It clearly shows the importance of animals to society and their value in terms of communal wealth, such that the landscape and society were reorganized to accommodate their management.

The demands of a transhumant system, directing domestic resources to communal endeavours and dictating that household members spent time away from the settlement, would weaken the power of the household. The process separates smaller social units such as individuals, nuclear families, and households from control over their own means of production. Instead they became part of the larger communal means of production, which tied them into a thick network of social obligations and bonds. Communities must have been able to maintain tenure over pasture located away from permanent settlement, mobilize the labour required for building the early hilltop enclosures, and facilitate the management of the large animal herds; these are all factors that see their ultimate expression in the Early Iron Age with the construction of hillforts in downland locations.

The transition, rather than representing a change in metal technology, can therefore be interpreted as a change in the scale of productive units and its implications for social reproduction. It represents a move from a society where households maintained some control over their means of agricultural production, if not their own reproduction, to one where communities became self-sufficient and able to control the means of production on behalf of their members. It denotes a transformation in social relationships and social reproduction of communities in a changing world in which a social value system based on bronze was replaced by one focused on animals (Needham, Reference Needham, Haselgrove and Pope2007).