Memory and material culture

For us to interpret objects of the past in the past, we must first demonstrate that time and memory were important elements of prehistoric social structures. Previous studies (e.g. Bradley Reference Bradley2002; Meskell Reference Meskell, Van Dyke and Alcock2003) have highlighted the interlinked nature of memory and social identity (cf. Connerton Reference Connerton1989). Modes of remembering and forgetting are often viewed as integral to the identity of a community (Middleton & Edwards Reference Middleton, Edwards, Middleton and Edwards1990, 10), and certain acts, such as the deposition of objects, may serve simultaneously to commemorate and forget a person, event or practice (Forty Reference Forty, Forty and Küchler1999; Jones Reference Jones2007; Mills Reference Mills, Mills and Walker2008). Forty (Reference Forty, Forty and Küchler1999, 8–12) and Mills (Reference Mills, Mills and Walker2008) outline different ways in which objects might be agents for forgetting and memorializing. The concealing of objects, for instance, may remove the physical presence of the object, but the place in which it was hidden may become a site of commemoration (Mills Reference Mills, Mills and Walker2008). Similarly, multiple authors have emphasized how iconoclasm or the destruction of places and artefacts represents a transformation that ‘forgets’ the original object or site, but creates a new set of social relations between people, objects and places through the event (e.g. Buchli & Lucas Reference Buchli, Lucas, Buchli and Lucas2001, 80; Chapman Reference Chapman2000; Forty Reference Forty, Forty and Küchler1999, 10–12; Mills Reference Mills, Mills and Walker2008). Notably, the enduring nature of monuments and material culture, as well as the extended temporality of landscapes, means people would regularly have been confronted with aspects of their close and distant pasts (Bradley Reference Bradley1998a, 85–100; Gosden & Lock Reference Gosden and Lock1998; Ingold Reference Ingold1993; Jones Reference Jones2007). Further, the sensory and emotional effect of long-lived places and enduring materials would have evoked diverse responses that were temporally and culturally situated, occasionally prompting engagement with elements of the past (Harris Reference Harris, Bille and Sørensen2016; Knight Reference Knight2023a) and the special treatment of objects (Büster Reference Büster2021).

Objects, like monuments, evoke remembrance. Rowlands (Reference Rowlands1993) emphasizes the role of material culture in propagating tradition and memory, suggesting that things can act as stimuli for unconscious memories in a way that language cannot. Even if they are not imbued with a certain memory, people unconsciously (or consciously) make associations influenced by various qualities of objects, including their form and aesthetics, function, condition, depositional context or even the origin(s) of the object and its materials. These aspects are important for understanding the mnemonic potential of objects in the past (Hodder Reference Hodder2012; Jones Reference Jones2007; Lillios Reference Lillios1999; Olivier Reference Olivier and Greenspan2011, 170–4; Woodward Reference Woodward2002), some of which could have taken on an inalienable quality (i.e. so strongly possessing an identity that they cannot be separated from associated persons or cultural ideas; Weiner Reference Weiner1992). Objects were constituted by their relations with people and places through practices, which was enhanced by cultural knowledge of the history of the objects, as well as their aesthetics (Gosden Reference Gosden, DeMarrais, Gosden and Renfrew2004; Jones Reference Jones2007, 56–7). Certain artefacts might embody and communicate ideas, histories, knowledge and even emotional affect (Bell & Spikins Reference Bell and Spikins2018; Büster Reference Büster2021; Reference Büster, Lipkin, Bell and Vare2024). The treatment and deposition of these objects thus became a relational act that connected people, place and objects (Chapman Reference Chapman2000). The durability of artefacts is particularly significant here, as these can transcend the human lifespan and accrue multi-generational age. Of note, much work has focused on prehistoric grave goods showing signs of age, wear and repair (sometimes corroborated through absolute dating) which may have been curated over long periods and functioned as heirlooms passed between generations (Woodward Reference Woodward2002; Woodward & Hunter Reference Woodward and Hunter2015). These became objects that connected people through time, as well as establishing relations between the living and the dead.

Identifying out-of-time Bronze Age metalwork

Although older objects are regularly noted associated with burials, the phenomenon of finding out-of-time objects in other contexts, such as hoards or areas of settlement activity, has received less attention. The most famous multi-period hoard case study is undoubtedly the Salisbury hoard, which contained at least 535 objects spanning the Early Bronze Age to Middle Iron Age and was probably deposited around the second century bce (Garrow et al. Reference Garrow, Gosden, Hills and Bronk Ramsey2009, 83, 88; Stead Reference Stead1998). As an Iron Age example, it is not considered in detail within the present research but the number and diversity of older objects in this hoard suggests that people may have discovered and accumulated multiple earlier deposits and hoards ahead of deposition (Hingley Reference Hingley2009, 154; Stead Reference Stead1998, 123). However, if we consider out-of-time Bronze Age metalwork from later Bronze Age contexts such objects are noted repeatedly throughout the literature, but often ignored in broader discussions, relegated to catalogue appendices, or ignored as unlikely to be associated with later contexts (e.g. Davis Reference Davis2012, 52; Rowlands Reference Rowlands1976, 70, 213–15). This is perhaps due to the period-specific focus of artefactual research, compounded by the problem of defining what exactly constitutes an ‘out-of-time’ object.

By definition, an out-of-time object must be earlier than the context in which it is recovered and can generally be identified through dating relative to its association(s). For Bronze Age metalwork, this relies on placing an object within typo-chronological sequences (Table 1). These are, however, subject to refinement and thus, to ensure that we are dealing with genuinely out-of-time objects, here any object identified in a context that is two or more metalworking assemblages later than the expected relative typological sequence (e.g. a Penard-assemblage palstave alongside Ewart Park-assemblage material) is considered an out-of-time object. Objects from abutting metalworking assemblages in a single context (e.g. Wilburton- and Ewart Park-assemblage material together) may instead be interpreted as deriving from a transitional point in production traditions, hence their exclusion here. This parameter guarantees a minimum of a century or more between the earliest and latest objects, raising the likelihood of a truly anachronistic object at the point of deposition.

Table 1. Bronze Age typo-chronologies for copper-alloy objects in southern Britain. Only object types relevant to the present study have been noted. (Chronology drawn from Needham et al. Reference Needham, Bronk Ramsey, Coombs, Cartwright and Pettitt1997; Reference Needham, Lawson and Woodward2010)

We must also consider how earlier objects end up in later contexts and the complex trajectories they may take for a number of reasons. Knight et al. (Reference Knight, Boughton, Wilkinson, Knight, Boughton and Wilkinson2019, 3) discuss two main trajectories:

1. 1. Objects continued in sustained circulation. This may take the form of continued use beyond the expected typo-chronology, or objects may be curated or retained (for whatever purpose) until a much later date of deposition;

2. 2. Objects represent instances of an older deposit having been (re)discovered by past communities (e.g. through farming, forestry or construction of a later settlement or monument).Footnote ³

Figure 1 highlights that the path any one object might take is not necessarily straightforward and it may go through several iterations of use and adaptation over time—this is equally true for curated objects and rediscovered objects and a key reason why these are referred to as out-of-time objects here, rather than (for instance) relics or curated objects. Different aspects of out-of-time objects and the surrounding landscape can be analysed to aid understanding of their presence in later contexts. One might expect a metal object retained and used over a long period to show signs of extensive wear, repair or curation (e.g. through wrapping),Footnote ⁴ while rediscovered and redeposited objects may present different corrosion from the items with which they were ultimately deposited. An area of long-lived settlement or repeated depositional activity may also increase the likelihood that older objects are (re)discovered.

Figure 1. Hypothetical trajectories of objects in and out of the archaeological record over time.

Importantly, the above trajectories do not offer definite answers for the value and meaning of out-of-time objects, nor their deposition. This necessitates a case-by-case analysis of the object condition and context rather than the application of an overarching interpretation.

Case studies and source critique

Thirty-four instances of certain, probable or possible contexts where out-of-time metal objects were present have been identified in Bronze Age contexts across southern Britain (see supplementary material). This includes two cases from Wales identified by a previous study (Knight Reference Knight, Knight, Boughton and Wilkinson2019). Twenty-eight are hoards containing one or more out-of-time objects. Six case studies represent out-of-time objects from non-hoard contexts; these are primarily settlement-related activity areas (e.g. ditches) with a long temporal span of hundreds of years. Most of the out-of-time objects were deposited during the Late Bronze Age (LBA), with only one potentially deposited during the Middle Bronze Age (MBA). All are listed and numbered in the Supplementary material; when first mentioned in the text, the relevant catalogue number is shown in parentheses (for example, Cadbury Castle (1)).

Various factors were considered when interpreting these cases. Crucially we must be aware of differences between the hoard and non-hoard contexts under study here. The prolonged use and re-use of settled landscapes, with multiple phases of occupation and construction, provide a more active environment of interactions increasing the possibility of the discovery of earlier material, compared with the relative static action of depositing a hoard. Much material might be rediscovered and redeposited over several hundred years (e.g. Büster Reference Büster2021). A hoard, by contrast, provides a snapshot at a specific moment in time, even if the accumulation of the hoard may have been more dynamic. However, much metalwork recovered from settlement areas is fragmentary and undiagnostic (Needham in Longley Reference Longley1980, 24ff.) so may not be recognized as being of age, by either Bronze Age communities or modern archaeologists. Additionally, we are faced with modern biases because of excavation strategies and the date of discovery. It is rare to excavate a prehistoric settlement or landscape in its entirety—though greater coverage is achieved through developer-funded excavations—and thus it is highly probable that excavations that sub-sample a site would miss any isolated deposits. It is striking that the case studies identified comprise large-scale excavations and landscape surveys. By contrast, the modern discovery of a hoard typically means all objects are recovered and the immediate context is sometimes investigated. Historic discoveries, however, often lack detailed find circumstances and objects might be dispersed, meaning we are left with a partial record.

Because of these differences, hoards and non-hoard contexts are initially presented separately, though discussed together. Additionally, following Hingley (Reference Hingley2009), a scale of certainty was required to categorize all case studies as ‘certain’, ‘probable’ or ‘possible’. This relied on:

• • the security of the context and association, influenced by the accuracy of the records kept at the time of discovery;

• • the known object history post-recovery; and

• • the likelihood that incomplete objects definitely represent an older artefact.Footnote ⁵

The certainty of any association is qualified in the Supplementary material.

Hoards

Distribution and depositional location

Multi-period hoards are concentrated in central southern and eastern England and along the Thames Valley (Fig. 2), biased partly by the comprehensive cataloguing and research investigations into hoards and sites in this region (Davies Reference Davies2018; Turner Reference Turner2010; Weller Reference Weller2014); these are also areas where a significant volume of LBA metalwork was deposited. Even considering other areas that have seen dedicated research attention such as southwest England (Knight Reference Knight2022a; Knight et al. Reference Knight, Ormrod and Pearce2015; Pearce Reference Pearce1983), the geographic concentration of out-of-time objects (and the hoards with which they are associated) in southern and eastern England appears to represent a genuine pattern, albeit one probably related to intensive hoarding practices during the late Ewart Park assemblage. Few multi-period hoards have been recorded in western or northern Britain (Knight Reference Knight, Knight, Boughton and Wilkinson2019).

Figure 2. Distribution of sites where out-of-time objects are present. Numbers relate to supplementary material. (Contains Ordnance Survey data licensed under the Open Government Licence v3.0.)

Most deposition sites are concentrated in elevated locations, coastal areas and river valleys, reflecting common landscape locales for metalwork deposition (Dunkin et al. Reference Dunkin, Yates and Bradley2020; Yates & Bradley Reference Yates and Bradley2010). Four multi-period hoards were positioned on hilltops (Breage I (8); Cleeve Hill (28); Yattendon (21)) or high on hillslopes (Crooksbury (9)). Others were situated on or near coastlines and estuaries (Gorleston-on-Sea (11); Shoebury I (19); Southall (20); Southchurch I (34)), or within river valleys (Mawr Community (14)).

Object types, conditions and associations

The types of out-of-time objects identified are varied, but typically only represented by single instances (Fig. 3). The key exceptions are spearheads and earlier flat or flanged axeheads, which occur both in hoards and non-hoard contexts; palstaves overwhelmingly dominate. This is an unsurprising trend as palstaves are the most characteristic and numerous metal object of the MBA in southern Britain and were deposited in large numbers (Rowlands Reference Rowlands1976); the same is true of earlier axeheads. Of all earlier objects, different forms of axeheads are the most likely to be rediscovered or retained.

Figure 3. The frequency and completeness of types of out-of-time objects occurring across hoard and non-hoard contexts. ‘Fragment’ refers to any object where 25 per cent or less survives.

Many out-of-time objects were complete when recovered, the exceptions being spearheads and rapiers, which were more often deposited incomplete,Footnote ⁶ though remained identifiable (Fig. 3); similar patterns were identified for out-of-time objects in hoards elsewhere in Britain (Knight Reference Knight, Knight, Boughton and Wilkinson2019, 24). The completeness and wear of objects are crucial to understanding how objects may have been treated and/or valued. Worn and abraded fragments may have been residual pieces in long-term circulation (and their fragmentary nature may in fact hinder their identification as out-of-time objects), while complete or repaired objects indicate rediscovered deposits or actively curated artefacts.

At several sites (e.g. Cumberlow Green (10); Rayne (17)), complete palstaves were deposited alongside complete and fragmentary LBA material. The Shoebury I hoard comprising largely complete and well-used objects also contained a complete, unused MBA palstave. Turner (Reference Turner2010, 84) suggests the palstave may have been recovered from an earlier context, which seems likely considering the palstave would have been centuries old when (re-)deposited, as well as contrasting with the used condition of the other objects. Additionally, unfinished palstaves are characteristic of MBA hoards in Essex (Turner Reference Turner2010, 70). That earlier palstaves were not also fragmented in numerous instances may indicate certain values attributed to them as objects of the past. As rediscovered objects, it may have been important to respect and maintain the condition in which they were found.

The Yattendon hoard is particularly interesting as it included an Early Bronze Age (EBA) flat axehead, three MBA basal-looped spearheads and two palstaves, alongside 52 LBA pieces (Coghlan Reference Coghlan1970). It is the most diverse multi-period hoard under consideration here. These objects were deposited in ‘wet drab clay’ (Colquhoun & Burgess Reference Colquhoun and Burgess1988, 76), perhaps indicating an originally damp context. This accumulation has typically been considered scrap (Britton Reference Britton1963, 270), but while some objects show signs of deliberate fragmentation, others appear to have been reworked and kept sharp prior to deposition (Coghlan Reference Coghlan1970). This includes the flat axehead, one of the palstaves and a LBA faceted socketed axehead: these objects had not exhausted their practical usefulness. Furthermore, the flat axehead and palstaves contained very little lead in their composition (<1%), confirming an earlier production date (Coghlan Reference Coghlan1970, table 1). Likewise, at Stoke Ferry (25) a complete copper halberd may have been associated with broken LBA copper-alloy weapons (Hawkes Reference Hawkes1954, GB.8). Similarly, earlier unleaded bronzes at Nettleham (15) and Southall found alongside later leaded bronze objects suggest these objects derive from previous manufacturing events (Britton Reference Britton1960, GB.51; Brown & Blin-Stoyle Reference Brown and Blin-Stoyle1959, 204). It is striking that these objects were not melted down nor recycled to make contemporary objects. In the case of Southall, earlier palstaves form the bulk of the hoard alongside a thick annular ring and a bronze mould for casting later socketed axeheads.

Most palstave associations also include socketed axeheads. At four sites, small deposits were recovered consisting of only palstaves and socketed axeheads.Footnote ⁷ The three palstaves at Paston (24) significantly predate the axehead and were broken and heavily corroded with a different patina to the axehead (Langmaid Reference Langmaid1966, 29), suggesting the rediscovery of an older deposit. Similarly, at Owslebury (16), the palstave was worn and abraded, while the socketed axehead was freshly cast (Dale Reference Dale1906). By contrast, at Hankley Common (12) the two axeheads and the palstaves all show damage and wear to the blade edge (Lowther Reference Lowther, Oakley, Rankine and Lowther1939, 163), with the socketed axeheads being poorly cast and probably not fit for use (Phillips Reference Phillips1967, 11). The Crooksbury axeheads were all complete and seemingly in good condition when recovered during ditch digging in the nineteenth century, despite the fact they chronologically span several centuries (Lasham Reference Lasham1895, 152).

It is possible some out-of-time objects were kept and used beyond the periods suggested by their typologies—this might be the case for the two earlier palstaves found alongside a Penard–Wilburton socketed axehead and an incomplete spearhead at Rotherwick, Hampshire (18) (Fig. 4), or the possible association of an Arreton flanged axehead with Taunton-assemblage material at Brabourne, Kent (27) (Fig. 5) (see also Knight Reference Knight, Knight, Boughton and Wilkinson2019, 29–30). Some other older objects might have been relatively common rediscoveries. Palstaves are also one of the more frequent out-of-time metal objects in Iron Age contexts (Hingley Reference Hingley2009). The survival of so many complete palstaves in later contexts when they could easily have been recycled may highlight the social significance of the tool, either as a link to the past or for the general usefulness and endurance associated with its function.

Figure 4. The Rotherwick hoard, Hampshire (HAMP-0C6D32). (© Winchester Museums Service. CC BY-SA 4.0, with minor adaptations by M. Knight.)

Figure 5. The Brabourne hoard, Kent (KENT-775A23). (© Kent County Council. CC BY 2.0.)

Accumulations in fragment-dominant hoards

We should be wary of applying significance to all material, however. Several worn and fragmented out-of-time objects were recovered from fragment-dominant LBA hoards and may have been scarcely recognizable as archaic objects, their fragmentation and wear resulting from long-term circulation in an active accumulation of fragmented material where pieces were regularly fragmented, added and removed before the material ended up in the ground (cf. Needham Reference Needham1990). The age of some of these pieces may have held limited value, such as a probable rapier fragment found alongside later axehead fragments and casting jets at Mawr Community, or a possible MBA sickle fragment alongside 62 complete and broken LBA weapons, tools and ingots at Southchurch I. The Gorleston-on-Sea hoard, buried during the Ewart Park assemblage (c. 920–800 bce), contained c. 122 objects typologically spanning 300 years, including MBA fragments of a palstave and two rapiers and residual material from the preceding Wilburton assemblage; it may have been deposited in an organic bag (Clough & Green Reference Clough and Green1978). It is hard to argue these older fragments held any particular social significance within larger hoards except where deliberately structuring or care is evident for the fragments. One example may be a MBA torc fragment inserted into the socket of an axehead at Breage I (Fig. 6); Dietrich (Reference Dietrich2014; Dietrich & Mörtz Reference Dietrich, Mörtz and Brandherm2019) has referred to such plugged sockets as miniature hoards, occasionally situated within large hoards, and the collection and inclusion of fragmented out-of-time objects may have been part of adding significance to these smaller groups.

Figure 6. The Breage I hoard, Cornwall. The arrow indicates the MBA torc fragment. (Photograph: author, courtesy of the Royal Institution of Cornwall.)

The hoards from Gorleston-on-Sea and Southchurch I also contained objects stylistic of the Carp's-Tongue/Boughton-Vénat tradition (875–800 bce), suggesting they were deposited during the latest stages of the Bronze Age (Brandherm & Moskal-del Hoyo Reference Brandherm and Moskal-del Hoyo2014; Turner Reference Turner2010). By extension, it also means this is the stage when earlier objects would be at their oldest; at least eight Carp's-Tongue hoards containing out-of-time objects have been identified (Table 2; Turner Reference Turner2010, 37).Footnote ⁸

Table 2. Carp's-Tongue hoards (c. 875–800 bce) containing out-of-time objects

The character of these hoards suggests that material may have been accumulated over time with pieces added and removed either before or after deposition, perhaps in stockpiles. Older objects may represent part of the original accumulation but then never entered the archaeological record due to later removal (for whatever purpose). The out-of-time objects we observe in hoards may be the remnants of a larger number of older objects that stayed in longer circulation. In turn, the out-of-time objects that remained in hoards potentially reveal the length of time over which hoards were accumulated and their ‘active’ nature, with hoarded deposits re-opened and reiterated through the addition and removal of pieces over time (Knight Reference Knight, Bertrand, Durham, Hall, Keily and Knight2022b; Needham Reference Needham2001; Reference Needham, Burgess, Topping and Lynch2007, 280). Evidence for containers, such as the possible bag at Gorleston-on-Sea, highlights the possibility that metal hoards were not static and may have been gathered and moved around.

Non-hoard contexts

Six instances of the deposition of out-of-time metal objects in secure non-hoard contexts were identified. Four of these are distributed in southeastern and eastern England, mirroring the distribution of the multi-period hoards; the two remaining sites are from the Cadbury Castle hillfort landscape in Somerset (Fig. 2).

At Woodbastwick, Norfolk (6), an incomplete Taunton-assemblage palstave was recovered during metal-detecting with adhering calcined flint and charcoal from a possible burnt mound (National Bronze Implements Index BAI-6139Footnote ⁹; HER 29362Footnote ¹⁰). Charcoal (Maloideae sp.) embedded in the corrosion products on the surface of the palstave was radiocarbon dated producing a M–LBA date, 1210–910 cal. bce (2865 ± 45 bp; OxA-4505; Needham et al. Reference Needham, Bronk Ramsey, Coombs, Cartwright and Pettitt1997, 61, DoB 18).Footnote ¹¹ As Needham et al. (Reference Needham, Bronk Ramsey, Coombs, Cartwright and Pettitt1997, 60) note, this may represent the accidental conflagration of the palstave, but its inclusion in a possible burnt mound may instead point to the deliberate inclusion of an already old object in a later monument. A fragment of a second palstave was subsequently found nearby, also with charcoal adhering, but this was not dated. Further details of the site are unfortunately lacking.

At Heathrow Airport (2), a MBA spiral finger ring was recovered from the fill of a LBA waterhole. This waterhole intersected a MBA well, which was lined with wattle work radiocarbon dated to 1500–1210 cal. bce (3086 ± 51 bp; Wk-10024: Lewis et al. Reference Lewis, Leivers and Brown2010, 141, 152). The recutting of an earlier feature raises the possibility that the ring represents a rediscovered object that was then redeposited. Other apparently rediscovered and reburied objects from the same site include a complete Neolithic stone axehead from the basal fill of a different MBA waterhole located close by, alongside a wooden haft for a socketed axehead and a wooden ard spike (Lewis et al. Reference Lewis, Leivers and Brown2010, 156–8) and a MBA basal-looped spearhead recovered from a M–LBA waterhole (Lewis et al. Reference Lewis, Leivers and Brown2010, 142).

At Iwade (3) and Shrubsoles Hill (5), both Kent, MBA objects were deposited in LBA ditches. The landscapes at both sites built on MBA traditions and, importantly, these sites are approximately five miles apart, separated by the Swale river channel, and may have been intervisible. These out-of-time deposits may thus represent similar social ideas regarding objects of the past and the communities that produced them within a wider landscape of interactions.

At Iwade, an Acton Park–Taunton-assemblage palstave was deposited in a LBA trackway ditch at least a century after its initial production; the depth of the palstave indicates it was deposited late in the ditch silting history (Barber in Bishop & Bagwell Reference Bishop and Bagwell2005, 52). At Shrubsoles, an incomplete MBA side-looped spearhead was found in a LBA enclosure ditch (Fig. 7; Coles et al. Reference Coles, Pine, Preston, Coles, Hammond, Pine, Preston and Taylor2003). Other deposits in and around these ditches included 37 cremation burials and pottery, indicating the importance of this area for formalised deposition. Many of the cremation burials cut the infilled ditch of Enclosure A, suggesting they represented a practice post-dating the deposition of the spearhead (Coles et al. Reference Coles, Pine, Preston, Coles, Hammond, Pine, Preston and Taylor2003, 18). Wood fragments (species unknown) found inside the spearhead socket placed the last hafting of the spearhead around 1010–810 cal. bce (2758 ± 41 bp; KIA11047: Taylor in Coles et al. Reference Coles, Pine, Preston, Coles, Hammond, Pine, Preston and Taylor2003, 42–3; Coles et al. Reference Coles, Pine, Preston, Coles, Hammond, Pine, Preston and Taylor2003, 51). The spearhead may have been used over a long period or else was rediscovered and re-hafted especially for deposition (Coles et al. Reference Coles, Pine, Preston, Coles, Hammond, Pine, Preston and Taylor2003, 53). The shaft of a side-looped spearhead from the Heathrow Airport site also produced a radiocarbon date stretching into the LBA (Lawson Reference Lawson and Scott2010), strengthening the idea that these were sometimes long-lived implements or perhaps that this was a long-lived object form that was occasionally produced beyond our expected typo-chronologies. The presence of numerous cremation burials, both inside and outside the Shrubsoles enclosure, as well as in the enclosure ditch, may infer a connection being made between people and their pasts through the act of interment. The re-hafted spearhead may have been an heirloom linked with this practice.

Figure 7. Enclosure A at Shrubsoles, Kent, showing location of the spearhead within the enclosure ditch. (Adapted from Coles et al. Reference Coles, Pine, Preston, Coles, Hammond, Pine, Preston and Taylor2003, figures 1.6 and 1.17, courtesy of Thames Valley Archaeological Services.)

A similar situation can be observed at Cadbury Castle hillfort (1) and the surrounding landscape where activity spanned the Neolithic through to the Romano-British period. During excavations at the hillfort site, 47 pieces of complete and incomplete Bronze Age metalwork were recovered from a variety of contexts within the hillfort enclosure, mostly dating to the LBA (Barrett et al. Reference Barrett, Freeman and Woodward2000; Knight Reference Knight2023b; O'Connor Reference O'Connor1994). Three pieces, however, compositionally and typologically predate this period of activity, including a fragment of a MBA side-looped spearhead within a pit containing LBA pottery. Further, a complete EBA flanged axehead was found alongside LBA and Early Iron Age pottery and metalwork in a pit, lined with clay-daubed stone and a ceramic terminus post quem of c. 400–200 bce (Alcock Reference Alcock1971, 5; O'Connor Reference O'Connor1994). The axehead was thus about 1500 years old when deposited. Given the long period of activity on this site, the objects probably represent rediscovered deposits. That the axehead was deposited in such a carefully constructed context, though, suggests there was some recognition and potentially some reverence of the historic nature of this object, which aligns with Hingley's (Reference Hingley2009) suggestions for the Iron Age.

On the western slope of the hillfort, at Milsoms Corner (4), a bronze shield was found in the M–LBA enclosure ditch (Coles et al. Reference Coles, Leach, Minnitt, Tabor and Wilson1999; Needham et al. Reference Needham, Northover, Uckelmann and Tabor2012). This Yetholm-type shield was carefully crafted and elaborately decorated with hammered ribs and bosses (Fig. 8; Coles et al. Reference Coles, Leach, Minnitt, Tabor and Wilson1999, 40; Knight Reference Knight2022a, 87–90); the impressive aesthetics, as well as the time and labour necessary for production, may have enhanced the mnemonic capacity of this inalienable object. The shield was seemingly well maintained prior to deposition, and metallurgical analysis showed that its composition is consistent with the Penard assemblage (1275–1150 bce), in which Yetholm shields appear (Needham et al. Reference Needham, Northover, Uckelmann and Tabor2012, 482).

Figure 8. The Milsoms Corner (or ‘South Cadbury’) shield and a plan of the Milsoms Corner Bronze Age enclosure ditch, showing the various depositions and phases of ditch construction. Phases 1 and 2 date to the MBA, while Phase 3 dates to the LBA. (Shield image © Somerset County Council and South West Heritage Trust, acc.no. TTNCM 110/1998/1. Plan of Milsoms Corner courtesy of Richard Tabor.)

Upon deposition, the shield was placed face down over a stake-hole at the junction of two ditches forming the southern and eastern boundaries of the enclosure (Coles et al. Reference Coles, Leach, Minnitt, Tabor and Wilson1999, 35–7). The shield was then penetrated three times from behind by a blunt, non-metal object—possibly a wooden stake (Coles et al. Reference Coles, Leach, Minnitt, Tabor and Wilson1999, 38–9, 45–6). The stake-hole over which the shield was placed also contained a red deer or cattle pelvis, in contact with the shield rim, which produced a LBA radiocarbon date, 1210–810 cal. bce (2810 ± 80 bp; BM-3152, 95% probability), probably 1060–830 cal. bce (68% probability), meaning that although the dates overlap with the Penard assemblage, it was probably curated for up to a century or more after production and before deposition (Coles et al. Reference Coles, Leach, Minnitt, Tabor and Wilson1999, 37; Needham et al. Reference Needham, Northover, Uckelmann and Tabor2012, table 2). This is perhaps strengthened by its unusual deposition context compared with other Yetholm-type shields, which are almost always deposited in watery contexts.

About one metre west of the shield was a LBA post-hole within the ditch packed with burnt stone and containing an unburnt human lower leg bone from an EBA burial disturbed during the first phase of ditch construction during the MBA (Fig. 8; Coles et al. Reference Coles, Leach, Minnitt, Tabor and Wilson1999, 35–7). Tabor (Reference Tabor2008, 91) has suggested this packed post-hole and the shield deposition occurring in the upper fill of the ditch were linked. The disturbed leg bone may have been viewed as a relic or a powerful ancestral artefact, perhaps requiring special treatment; its incorporation into practices that involved the deposition of an already old artefact (the shield) and the closing of the ditches was probably significant. Reburying the disturbed leg bone and the deliberate decommissioning of the shield may have served to finalize and sever any inalienable ties while also producing a performative and memorable event (Knight Reference Knight2022a, 90).