Introduction
The deliberate destruction and deposition of Bronze Age metalwork is a well-recognized phenomenon, often interpreted according to ritual or utilitarian theories. Yet, to date, no comprehensive methodology has been developed for assessing and interpreting the cause of this damage, with researchers generally relying on obvious features, such as excessive bending or extreme fragmentation. But when can a bent sword be considered ‘deliberately bent’? How can we distinguish an intentionally fragmented axehead from one broken through use? The proliferation of newly-discovered broken and damaged metalwork in excavations and metal-detecting activities across Europe makes it increasingly necessary to determine accurately whether damage on metalwork was the result of deliberate prehistoric agency if we are to interpret this material with confidence.
By considering the material properties of prehistoric copper alloys in combination with recent experimental research, I propose in this article a methodology for a systematic approach to interpreting damage on Bronze Age metalwork, focusing predominantly on British copper alloy objects. Key indicators of intent observed on metalwork are categorized, followed by a set of criteria from which one might consider damage to have been caused by human action. This leads to a simple, coded index, a Damage Ranking System (hereafter DRS) underpinned by an informed rationale. Since studying damage is inherently subjective, it is impossible to achieve total objectivity. However, I conclude this paper by presenting two case studies that exemplify how the DRS can be applied in practice and what potential it offers.
Here, the terms ‘destruction’ and ‘decommission’ (and their variations) refer to rendering an artefact unusable, based on its presumed function, e.g. a sword as a thrusting or slashing weapon or an axehead as a chopping implement. Damage may be intentional, unintentional, or a combination of the two, which may in turn change its function (e.g. a broken sword may become raw material for recycling). However, the purpose of the framework presented here is not to interpret intentional damage as ‘ritual’ or ‘utilitarian’. Intentional damage can be both; what is important is to first assess whether what is being interpreted shows signs of intent.
Classifying Damage
In her study of Bronze Age metalwork from the River Thames, York (Reference York2002: 80) considered deliberate destruction ‘probable’ when an artefact had been:
• ‘chopped across at right angles to its length once or more’;
• ‘struck and crushed in a manner inconsistent with its primary use’;
• ‘bent to breaking point (always a sword)’;
• ‘burnt and maybe twisted, distorted and fused to other objects’.
This broad method of identification characterizes the approach often taken, focusing on ‘obvious’ features of damage. Nebelsick (Reference Nebelsick and Pare2000), for instance, presents numerous case studies of Late Bronze Age hoards from Central Europe selected for the exaggerated nature of the damage. Likewise, objects considered intentionally damaged include crushed and/or plugged socketed axeheads in hoards from Britain (Turner, Reference Turner2010; Dietrich & Mörtz, Reference Dietrich, Mörtz and Brandherm2019), as well as numerous bent, burnt, broken, and/or notched swords from different contexts across Britain, Ireland, and France (Bridgford, Reference Bridgford2000; Quilliec, Reference Quilliec, Hamon and Quilliec2008). In his recent analysis of the Late Bronze Age hoard from Oltárc Márki Hill in Hungary, Tarbay (Reference Tarbay2017) goes further in separating ancient from modern damage and accidental/use-related from deliberate damage to identify what represented a true prehistoric manipulation and what was caused by other processes. A thorough set of criteria is, however, not presented in any of the examples cited above.
Patterns of damage on similar objects have often been used as a way of identifying intent, including the presence of different parts of weapons and tools in hoards across Europe (Bradley, Reference Bradley2005: 151–55; Čivilytė, Reference Čivilytė2009) or the careful fragmentation of Late Bronze Age gold bracelets in Britain and Iberia (Gwilt et al., Reference Gwilt, Kucharski, Silvester and Davis2005; Perea, Reference Perea, Hamon and Quilliec2008). Alternatively, patterns of damage have been used to identify changes in practice over time. In his study of Late Bronze Age hoards in the Carpathian Basin, Rezi (Reference Rezi, Berecki, Németh and Rezi2011) recorded fragmentation on different object types. Although this study relied on fragmentation rather than other forms of damage, it allowed Rezi to identify an increase in the extent of damage sustained on objects in hoards from ‘Bronzezeit D’ (c. 1325–1200 bc) to Hallstatt A1 (c. 1200–1125 bc). Rezi used this data-led approach to assess whether sacred or profane interpretations for fragmentation were applicable on a wide scale to hoards in his study region; he concluded that, while some breakage may have been linked to a pre-monetary function, not all hoards nor even every object within a hoard indicated that this was the sole reason for the fragmentation and deposition of objects. Thus, we can begin to see the advantages of developing a methodology for identifying deliberate damage.
While obvious indicators of damage are no doubt important, what of evidence that is initially inconclusive? It is reasonable to consider bent and broken weapons in a hoard to be the result of intentional action, but this is less clear when an isolated fragment of an axehead is found. Moreover, studies focused on the deliberate damage of metal objects rarely draw upon wear-analysis that could indicate how different types of damage were sustained (though see Horn & Karck, Reference Horn and Karck2019). A more cohesive approach is, therefore, desirable, as well as a system of identification and classification building on past studies.
Destruction Indicators
To develop a methodology for identifying intentional damage, we must first assess what might be reasonably considered an indicator of deliberate damage or destruction, hereafter a Destruction Indicator (DI) (Table 1). Seven DIs are proposed, encompassing the most common forms of deliberate damage observed on metalwork. Qualifying criteria are discussed and suggested, but should be considered indicative rather than absolute, and based on our present understandings as informed by material science and wear-analysis. Consequently, aspects of material science, use-related damage and post-depositional effects are inherent to the discussion but they have yet to be widely considered for deliberately damaged metalwork. Importantly, the criteria are starting points that may be subject to revision in the future. Although each listed DI may have been the result of intent, the potential for several indicators having been caused by use or accident must also be considered. Research into these aspects is presented below to determine the criteria for quantifying damage. An awareness of the contexts in which the objects were found, as well as associated objects and damage that can strengthen or weaken the determination of intent, is clearly important for all DIs. Nevertheless, the criteria laid out are designed to be used even when no further information is available. Linked to this is the initial assessment of whether damage is ancient or not. A simple method for identifying ancient damage consists of assessing the consistency between the corrosion of the object and the corrosion of the damage sustained (York, Reference York2002: 79; Roberts & Ottaway, Reference Roberts and Ottaway2003; Horn & von Holstein, Reference Horn and von Holstein2017). There are caveats to this method, as corrosion can form quickly post-recovery and obscure differences in the age of damage. Nonetheless, assessing the consistency of corrosion remains one of the best ways of ascertaining the antiquity of damage, especially when combined with knowledge of post-depositional processes and post-recovery history.
Table 1. Destruction Indicators observed on copper alloy objects.
DI 1: Bending and folding
Bending refers to transverse plastic deformation of an object from the expected trajectory (e.g. a sword that is no longer straight). Folding, an extreme form of bending, refers to plastic deformation resulting in an object being bent to about 180°; it is considered a deliberate act. Objects bent less than this require greater consideration to determine whether the observed damage was intentional. Four aspects should be considered in this assessment: composition, object thickness, angle of bend, and position of the bend. Here, I mainly concentrate on swords and spearheads as the artefacts most commonly found bent but also consider factors relevant to all object types.
Composition
British Bronze Age copper alloys typically consist of copper alloyed with up to ten to twelve per cent tin, and by the later Bronze Age lead is also added (Brown & Blin-Stoyle, Reference Brown and Blin-Stoyle1959). The quantity of tin and lead alloyed with copper will affect the plasticity of the resulting object. For instance, the tensile strength of a tin-copper alloy (i.e. the maximum stress an object can withstand before breaking) will increase until around thirteen per cent, at which point the higher levels of tin will result in a more brittle material, indicating why certain alloy percentages were favoured (Scott, Reference Scott2012: fig. 9.4; Knight, Reference Knight2019: 252). The inclusion of lead (particularly over 5 per cent) will increase the plasticity and ductility of the copper alloy, allowing greater plastic deformation. Quantifying the effects of composition is difficult as other factors play a part, but one should be aware and use this information where possible.
Thickness
The thicker an object is, the more force is required to bend or fold it and hence it becomes less likely that plastic deformation might occur accidentally. An experiment trying to bend and break a tin-bronze dagger (containing 10 per cent tin) confirmed that a 3 mm-thick tip could be bent when struck with a granite hammerstone, whereas, at 9 mm thick, the hilt only suffered surface damage and limited plastic deformation (Knight, Reference Knight2018: Appendix C). Furthermore, a study of bent metalwork with associated damage indicating human action (e.g. hammer blows) found that most objects were less than 5 mm thick (Knight, Reference Knight2018). Similarly, Moyler's (Reference Moyler2007: 144–49) attempts to break a replica Early Bronze Age flat axehead with steel tools caused bending and associated cracking, but no breakage. Moyler does not give the thickness of this replica, but it can be compared to bent archaeological examples (see below). Of course, archaeological examples of thick, bent objects are known, for example the large Middle Bronze Age dirk from East Rudham (Norfolk), which must have been twisted and bent while heated, probably by a skilled metalworker with appropriate tools and material knowledge (Wilkin, Reference Wilkin2016). Attempts to deliberately bend replica swords without breaking them came to similar conclusions for the Italian Late Bronze Age (Bietti Sestieri et al., Reference Bietti Sestieri, Salzani, Giardino and Verly2013: 167–9) and the British Late Bronze Age (Knight, Reference Knight2019). Therefore, we can suggest that extreme force and/or specific skills would have been needed to bend thicker objects; by extension it is less likely they would have bent by accident.
Conversely, thinner objects are more prone to plastic deformation, which is supported by recent use experiments. The thin tip of a replica Middle Bronze Age dirk bent when struck against synthetic skeletal tissues (Faulkner-Jones, Reference Faulkner-Jones2016: 3–4), and combat experiments involving replica swords (Crellin et al., Reference Crellin, Dolfini, Uckelmann, Hermann, Dolfini, Crellin, Horn and Uckelmann2018: 295; Gentile & van Gijn, Reference Gentile and van Gijn2019) have repeatedly caused some bending of the weapons, although not to the extent of destruction experiments. Such damage need not decommission an object as the blade can be quickly and easily straightened (Knight, Reference Knight2019: 259). Slighter objects, including tools and ornaments, might similarly bend through accident. Bent awls, knives, pins, and razors are often encountered; given their thinness and the inherent use exerting pressure on the tips, it is hard to determine when damage was deliberate rather than the result of accident or post-depositional processes.
Here, I suggest that bent objects showing no associated signs of damage (e.g. tool marks) but which are thicker than 7.5 mm at the point of the bend could be the result of intentional human action. This figure offers a starting point for categorizing damage and for eliminating the possibility that bending occurred accidentally; further research and experimentation are required to refine this figure. One must also be aware of modern processes, such as dredging or ploughing, both of which could strike and bend bronze objects.
Angle of bend
The angle to which an object is bent is another important indicator of intent. The combat experiments cited earlier indicate that swords might bend up to 20° through use (Gentile & van Gijn, Reference Gentile and van Gijn2019; Knight, Reference Knight2019: fig. 8), offering a starting point for quantifying bending on archaeological blades. In my own experiments, swords did not bend further than approximately 10°, and an analysis of swords from south-western England found that those displaying bending greater than c. 15° were also broken (Knight, Reference Knight2018: 160, Reference Knight2019: table 2). Similarly, Anderson (Reference Anderson2012: 104–5) notes that deliberately decommissioned weapons from northern Britain were bent more than 50°, while those bent through post-depositional activity were rarely bent more than 30°. The angle of bending caused through use is consequently unlikely to be extreme; it is proposed here that any sword with a transverse bend of 30° or more could be considered intentional damage. This does not exclude bladed objects bent to less than 30° from being intentionally bent, but additional damage is necessary to conclusively surmise intent.
Determining an angle of bend for objects other than swords is problematic as there have been fewer experiments and no data recorded. Such information would be most useful for spearheads, which often present bent tips and, for earlier forms, bent tangs (see examples in Davis, Reference Davis2012, Reference Davis2015). Christian Horn (Reference Horn2013: 13; Horn & Karck, Reference Horn and Karck2019) suggests that spearheads, like swords, might bend from force exerted on the tip. This may occur if the spear was thrust or used for stabbing, but the action failed or was deflected (Bridgford, Reference Bridgford2000: 145). However, Anderson (Reference Anderson2011: 604) found that replica spearhead tips did not bend when thrown or thrust. It is, thus, unclear whether a spearhead might bend through use.
Deliberate hammering of a spearhead tip caused bending, as well as crushing of the socket, and eventually material failure (Knight, Reference Knight2019: 262), indicating that intentional bending may be associated with other forms of damage. In lieu of any quantifiable tests, it is proposed here that the deliberate bending of a socketed spearhead is indicated by an angle of 30° or more. For thinner, tanged spearheads, 45° is taken as an indicator of intent. Where the tip is bent, a 45° angle should be present without the tip showing additional signs of use-wear (e.g. a blunted tip). Parameters are not set here for the angle of bend on other objects.
Position
The position of the bend can be a crucial indicator of intent, especially when combined with other factors. Faulkner-Jones (Reference Faulkner-Jones2016: 3–4) noted a 90° angle of bend to the tip of her replica dirk, indicating that thinner, more fragile parts of artefacts can be susceptible to more extreme bending. In addition to the spearhead tips and tangs mentioned above, the tips, hilts, and tangs of blades and tools are further vulnerable locations. Bending occurring in one or more of these locations may lessen the certainty that such plastic deformation can be considered deliberate.
DI 2: Twisting
Twisting is the lateral plastic deformation of an object along its longitudinal axis (i.e. a rotation), often observed on hafts, hilts, and blades. Twisting may indicate intentional or accidental damage. Minimal twisting could probably have been easily rectified by a competent metalworker.
Significant twisting of blades (e.g. swords, rapiers, or daggers) puts these objects out of action. In several cases, multiple twists are present that could have only occurred intentionally and probably entailed heating the object as part of the process (e.g. Colquhoun & Burgess, Reference Colquhoun and Burgess1988: no. 192). In his study of halberds, Horn (Reference Horn, Uckelmann and Mödlinger2011: 53) suggests that twisted hafting plates and blades indicate wrenching in the process of removing the handle. Twisting is, thus, likely to be associated with bending and torn rivet holes. Horn later suggested that accidental twisting could be caused ‘if the weapon became stuck somewhere, for example, between bones, and it was removed by force in a twisting motion’ (Reference Horn2013: 13). It is difficult, however, to judge the likelihood of such a scenario. I am not aware of any archaeological examples, and experiments involving replicas do not describe any twisting. Other uncertainties, not use-related, concern the extent to which a blade might suffer torsion because of heat-warping (e.g. on a pyre) or through post-depositional processes, emphasizing the need to appreciate the context of the find.
While experiments are needed to quantify a set of criteria, for now, any hilt/haft and/or blade that is twisted beyond 45° along its longitudinal axis will be considered the result of intent. The position of the twist may also indicate the likelihood of intent or accident, and one might expect the presence of other DIs. Other qualifying factors would have to be present for partial twists (less than 45°) to be classed as deliberate.
DI 3: Crushing
Crushing is the plastic deformation of an object through compression and applies primarily to socketed implements or ornaments. An object is unlikely to be crushed through use; a socketed axehead, for example, cannot be crushed if it has a wooden haft providing support to the object. Crushing an object is also related to the composition of the material. For example, leaded-bronze can have a higher malleability than tin-bronze, thus it might be more prone to compression.
Post-depositional processes (e.g. ploughing, dredging) may also cause crushing. This can often be deduced from a break in the patina and/or knowledge of the context from which the find derived. Where crushing is demonstrably prehistoric, it should be considered intentional.
DI 4: Notching
Notching refers to the material displacement and plastic deformation of metal and is commonly associated with use, particularly on edged weapons (Horn & von Holstein, Reference Horn and von Holstein2017; Gentile & van Gijn, Reference Gentile and van Gijn2019). However, it can represent intentional damage designed to decommission. Initial observation of this indicator should be concerned with verifying the antiquity of a notch, often indicated by the consistency of the patina, and that it has not been affected by corrosion (Roberts & Ottaway, Reference Roberts and Ottaway2003: 121; Horn & von Holstein, Reference Horn and von Holstein2017: 94ff.). The position, distribution, and depth of notches can all indicate intentionality.
Position
Notches on a blade edge might result from combat against another weapon (e.g. Bridgford, Reference Bridgford2000; Molloy, Reference Molloy, Uckelmann and Mödlinger2011, Reference Molloy2017: 288–90; O'Flaherty et al., Reference O'Flaherty, Gilchrist, Cowie, Uckelmann and Mödlinger2011). Notching in areas that might have been covered (e.g. by a hilt) is less likely to have been combat-related and instead may be the result of deliberate action—a conclusion drawn for notching on a sword from Werkhoven in the Netherlands (Fontijn et al., Reference Fontijn, Theunissen, van Os, Amkreutz, Bakels and Kamermans2012: 207). Anderson also highlights this, stating that even when the hilt of a replica sword was struck, chipping the wood and exposing the shoulder, ‘at no point was the metal component … damaged’ (Reference Anderson2012: 95); this suggests notching the metal part of a hilt requires some intent. In the case of objects that are not prone to edge damage (e.g. small pointed tools, vessels, ornaments), the very presence of notching could indicate intent.
Distribution
Regularly spaced notching along blade edges creating a ‘serrated’ effect has been noted on swords, halberds, and spearheads (O'Flaherty et al., Reference O'Flaherty, Gilchrist, Cowie, Uckelmann and Mödlinger2011: 45). Regular notching is unlikely to have resulted from use and can be taken as a potential indicator of intent. Irregularly spaced notching is harder to judge; even when multiple notches are present, it is difficult to conclude that the action was deliberately inflicted without other DIs or contributing factors to support this conclusion.
Depth
Combat experiments using swords caused notches up to 4 mm deep, which are comparable with those found on archaeological artefacts (Gentile & van Gijn, Reference Gentile and van Gijn2019: 137; Knight, Reference Knight2019: 258; Gentile, pers. comm.). Likewise, experiments investigating edge damage on halberds and swords suggested that notches observed on archaeological specimens were not the result of intentional damage (O'Flaherty et al., Reference O'Flaherty, Gilchrist, Cowie, Uckelmann and Mödlinger2011). Although the dimensions of these notches are not given, O'Flaherty et al. (Reference O'Flaherty, Gilchrist, Cowie, Uckelmann and Mödlinger2011: 43) state that the notch-depth on archaeological halberds falls between 1 and 6 mm, which suggests that experimental notches did not exceed these depths. Deeper notches from combat experiments have yet to be recorded. Therefore, as a starting point, it is suggested here that any notch deeper than 7 mm could be intentional. One must be especially wary here of the impact of corrosion when measuring the true depth of a notch.
DI 5: Breakage
Breakage refers to any material separation or loss that causes an object to fracture into two or more pieces. The parameters presented below are set to help identify an intentionally broken object rather than one broken by accident, use, or post-depositional factors, though these aspects should be considered.
Composition
As with plastic deformation, composition is a vital consideration when assessing fragmentation. Again, one should consider the presence or absence of lead in the copper alloy. Fragmentation experiments have demonstrated that, when heated and struck, lead-bronze objects break more easily than tin-bronze objects and with less plastic deformation (Knight, Reference Knight2017, Reference Knight2019). This is because lead is immiscible in a tin-copper mixture; so, at a microstructural level, it creates planes of weaknesses within the alloy (Scott, Reference Scott2012: 242). When heated and struck, the copper alloy breaks easily—a process known as ‘hot-shorting’ (Knight, Reference Knight2019: 252). Conversely, tin is soluble in copper, resulting in a stronger microstructure that is more prone to plastic deformation before fracturing and requires higher temperatures to achieve the same effect. Importantly, while the presence of lead lowers the tensile strength of an object when hot, it increases this same property when cold. Unheated leaded-bronze objects are more prone to plastic deformation than breakage.
Manufacture and casting flaws
Flaws during the casting process, such as impurities in the metal or unintended inclusions (Figure 1), weaken an object. For instance, casting flaws have been highlighted in several broken flat axeheads (Moyler, Reference Moyler2007: 147). Consequently, any break in which macroscopic casting flaws can be observed without additional damage suggesting human agency (e.g. tool marks) cannot be automatically taken to be intentional, especially if such breaks have occurred at weak junctions in the object.
Figure 1. Casting flaws present in the break of a palstave. By permission of the Royal Albert Memorial Museum and Art Gallery, Exeter.
Even where tool marks are visible, working the object and other manufacturing processes may have caused fragmentation; for example, copper alloys work-harden, which can increase the brittleness and lower the tensile strength, increasing the likelihood of fracture. Here, Kuijpers’ (Reference Kuijpers2018) work is informative: his study of Early Bronze Age axeheads highlights the impact of the skill of a craftsperson on the resulting objects, including broken examples. He notes several axeheads likely to have ‘hot-short’ during working when less skilled metalworkers have struck a heated axehead (Kuijpers, Reference Kuijpers2018: 148, 181, 220–21). Such an assessment draws on compositional knowledge, as well as the general evidence of working on an object; importantly, it underlines the need to consider other aspects of the object, such as its manufacture or use, in conjunction with any damage to make a reliable interpretation.
Patterns of breakage and weak points
Here, patterns of breakage refer to the same types of object suffering similar breakages in similar places. Two possible explanations may account for such breakages:
(1) There was an inherent design flaw that predisposed these objects to breaking at the same point during repeated types of activity; or
(2) Objects were repeatedly broken intentionally in the same places.
Palstaves provide a possible example of the former incidence; these objects are frequently found broken at the stop ridge through the hafting plate and flanges (Figure 2). This pattern of breakage occurs at a crucial point of hafting, suggesting that during use the palstave may be prone to breaking there. Published experiments involving palstaves are limited and breakage has not been recorded (yet), so the likelihood of these objects breaking through use is currently unknown. However, the way palstaves are produced—by casting the flanges rather than hammering them—creates an inherently weak internal structure at this point that would increase the likelihood of the object breaking under stress. Therefore, although replica palstaves have not broken in use experiments, breakages should not be automatically considered to be the result of intent. Potential weak points on other objects include side-loops, rivet holes, and blade tips. Of note, the same blow that caused the replica Friarton dirk tip to bend also caused part of the wooden hilt and metal rivet hole to break (Faulkner-Jones, Reference Faulkner-Jones2016: 3–4). At Tormarton (Gloucestershire) and Dorchester-on-Thames (Oxfordshire), broken bronze spearheads were found embedded in skeletons (Knight et al., Reference Knight, Browne and Grinsell1972; Osgood, Reference Osgood, Parker Pearson and Thorpe2005), suggesting that earlier (and possibly later) forms of socketed spearheads may suffer material failure through use across the tip or the blade-socket junction—in this case when used as fatal weapons.
Figure 2. Common breaking points on palstaves. Arrows indicate fracture points. By permission of the Royal Albert Memorial Museum and Art Gallery, Exeter.
Alternatively, objects which have repeatedly broken in the same place could be the result of intentionality. For instance, the separation of a socketed axehead cutting edge from its body is unlikely to be a use-related break, yet it is repeatedly observed in the archaeological record (Figure 3). Associated damage (e.g. hammer marks) could support this interpretation, as seen on socketed axeheads. Such an assessment partly relies on the availability of experimental data alongside a suitable body of archaeological data confirming that such a pattern exists (Knight, Reference Knight2017).
Figure 3. Cutting edge fragments of socketed axeheads (Royal Albert Memorial Museum and Art Gallery, Exeter, acc. no. 455/2007.1, and Torquay Museum, acc. nos. A367, A368). By permission of the Royal Albert Memorial Museum and Art Gallery, Exeter, and Torquay Museum.
Multi-piece breaks
Any object broken into more than two pieces could indicate intent. Some objects may break through use, but one might expect only a single breakage. The plasticity and toughness of bronze means that, although it is possible for an object to break in half, it is less likely that a bronze artefact would fracture into several pieces through use. This has typically been considered for swords but applies to other object types.
Quilliec (Reference Quilliec, Hamon and Quilliec2008: 70) argues that the breaking of a sword into more than two pieces cannot be accidental, especially when some swords were broken into ten pieces. Multi-piece breaks are always noted when some or all the refitting pieces are present. It is, however, often overlooked (or at least not noted) that the discovery of a mid-section of a sword blade or the body of an axehead implicitly suggests that the object was once in three or more pieces.
Some multi-piece breaks were possibly the result of post-depositional processes, which could be verified through analysis of the consistency of the corrosion. There is unfortunately no general methodology that can be applied to the study of objects broken into two or more pieces and use experiments have yet to demonstrate that this may occur through use. However, experiments deliberately breaking replica objects have produced archaeologically comparable multi-piece fragments, suggesting this is how such prehistoric fragmentation may have occurred (Knight, Reference Knight2017, Reference Knight2019). Knowledge of the metallurgical composition contributes to understanding the likelihood that an object may break in this fashion. It is reasonable to conclude that any object indicating a multi-piece break is the result of intentional fragmentation.
Associated marks
Marks associated with breakage are defined here as damage that can indicate the process by which an object was broken (e.g. tool marks or bending). Associated marks can confirm intentional breakage and indicate how objects were broken. Turner (Reference Turner1998: 36–37, 54–55), for instance, suggested that the absence of hammer marks but presence of bending associated with breaks on objects in the Grays Thurrock hoard (Essex) indicates that the pieces were probably snapped by hand. Chisel marks similarly indicate forceful blows designed to decommission an object (Figure 4).
Figure 4. A sword fragment with a chisel mark.
By permission of the Royal Institution of Cornwall.
If deliberate destruction was the work of skilled metalworkers who had a developed understanding of the material and objects (see Nebelsick, Reference Nebelsick and Pare2000; Knight, Reference Knight2019: 267–69), the object may present no associated damage. Heating an object before breaking it, for instance, might only require a single strike of limited force to fracture the object, leaving no marks (Knight, Reference Knight2019: 265). Alternatively, the section that was struck may be absent. Associated marks are therefore undeniably valuable when assessing intentional damage, but broken objects without marks should not be ruled out.
DI 6: Burning
Some Bronze Age metalwork has a charred appearance, even though bronze does not actually burn. Such an appearance is caused by the corrosion process and might vary across an object. The microstructure of an object might reveal whether it had been burnt prior to deposition, and examples where burning is macroscopically visible (e.g. the Late Bronze Age Duddingston Loch assemblage, Scotland) may help determine if other objects had been deliberately burnt (Bridgford, Reference Bridgford2000: 51–52). Cremation experiments have also demonstrated that the dendritic microstructure of bronze pieces is destroyed when subjected to high temperatures (700°C) on a pyre (Marshall, Reference Marshall2011: 32). Further microstructural analysis is an important avenue for future investigation.
If evidence of burning can be identified, several issues become obvious. Objects may have been burnt accidentally, e.g. falling into a fire, or may have been placed on a fire, e.g. during a cremation. Alternatively, heating objects to work and anneal them is often intrinsic to the manufacturing process and affects the metallic microstructure (Kuijpers, Reference Kuijpers2018: 85–86). How might we distinguish between these processes and objects that were deliberately heated for fragmentation? This DI requires particularly careful assessment, and I would argue that isolated evidence of burning should not be interpreted as deliberate destruction. Where present, it requires additional DIs and/or contextual information to confirm intent.
DI 7: Plugged sockets
Socketed objects include axeheads, spearheads, gouges, and knifes. The decommissioning of socketed axeheads by blocking or ‘plugging’ the socket has been noted across Europe (Hansen, Reference Hansen1998; Dietrich, Reference Dietrich2014; Dietrich & Mörtz, Reference Dietrich, Mörtz and Brandherm2019). Sockets can be filled with fragments of other objects, sometimes of diverse materials (e.g. amber or gold), and occasionally display fracturing around the socket mouth, where the objects have been hammered in. This act requires not only putting the socketed implement out of use, but often necessitates the fragmentation of other objects to fit them inside the socket. Plugged sockets should be considered as representing intention when encountered.
Damage Ranking System
The assessment of Destruction Indicators on Bronze Age metalwork and quantitative criteria has been necessarily thorough to accommodate the various ways damage can occur. DIs typically occur at the end of an object's use-life, but this does not always mean that the object goes out of circulation. Accurately identifying and understanding the various DIs on objects is a crucial element for understanding the biography of an object and the interactions Bronze Age communities may have had with it. This is further informed by the depositional process, which is often another intentional action. The DIs thus offer a benchmark from which to approach damage on Bronze Age metalwork. To simplify the analysis process, a ranking system is proposed here. The Damage Ranking System (DRS) is a simple six-category index corresponding to a scale of likelihood of damage being intentional (Table 2). Some damage is object-specific, although the criteria have been presented as a set of considerations applicable to most Bronze Age copper alloy metalwork (Table 3). Significantly, the DRS is designed to rank individual forms of damage, not an object overall, because many objects display multiple unassociated forms of damage (use-related, post-depositional, deliberate, etc.). Therefore, an object may be attributed multiple rankings. Here, two case studies illustrate the practical application of the DRS proposed.
Table 2. Summary of the Damage Ranking System.
Table 3. Summary of criteria and considerations for applying the Damage Ranking System.
The spearhead from Sandy's Farm
Two conjoining pieces of a Middle Bronze Age socketed spearhead (Figure 5) were recovered while metal-detecting at Sandy's Farm in Devon (Knight et al., Reference Knight, Ormrod and Pearce2015: 46, no. 180). This spearhead is broken across the middle of the blade and the tip is missing (Table 4). Both side-loops are broken, and the mid-blade break has a patina consistent with the rest of the object, suggesting the break is ancient. There are no macroscopic casting flaws in the break and the break was not over a socket hollow. The conjoining break shows no apparent blow marks; but, when the pieces are refitted, there is a notch on the blade edge at the point of the break. This notch is V-shaped, 11.2 mm long and 9.2 mm deep. The corrosion of the notch suggests that the damage happened in antiquity and the depth of the notch suggests intent plausibly related to the breakage that might be damage-ranked 2, though there are additional considerations. The broken side-loops and missing tip are both structurally weak points, so should be damage-ranked 1. In isolation, one might come to a mixed interpretation of this object: perhaps the tip and loops broke by accident, followed by a deliberate decommissioning of the spearhead through notching and fragmentation. However, if we consider the admittedly rare examples where broken spearheads can be associated with function—namely the contemporary Middle Bronze Age spearhead from Tormarton embedded in a vertebra—we can conclude that this spearhead may have broken through use. Overall it is likely that the fragmentation across the body of the spearhead was unintentional.
Figure 5. The Sandy's Farm spearhead, Devon. By permission of the Royal Albert Memorial Museum and Art Gallery, Exeter.
Table 4. Damage rankings for Sandy's Farm spearhead and Abdie axehead.
A flat axehead from Abdie
In 1889, two complete Early Bronze Age flat axeheads were purchased by the (then) National Museum of Antiquities of Scotland, said to have been found in the parish of Abdie in Fife (Mitchell, Reference Mitchell1889–90: 13). One of these axeheads is damaged (Figure 6). It is bent by some 8° in the centre of the body, and there is a fracture associated with this bend. The axehead is 10.8 mm thick at this point. Analysis of the corrosion under an optical microscope suggests that the break occurred before deposition, and earth remains in the fracture. The bend and the break are related but it is difficult to identify associated tool marks. This is due to post-recovery cleaning and small hammer marks covering the cracked face of the axehead. One tip of the cutting edge has also broken off and the cutting edge is notched and scratched. Patina preserving the original surface survives on the opposite face.
Figure 6. The Abdie flat axehead, Fife. © National Museums Scotland.
This object shows a combination of damage rankings (Table 4). Some damage, such as the scratched patina on the cracked surface, should be ranked 0, since it occurred post-recovery. Likewise, the fragmented blade tip shows fresh corrosion build-up, indicating recent material loss. The hammer marks that cover the cracked surface possibly represent prehistoric action; but it is more likely that they represent post-recovery actions, perhaps linked with cleaning, and should, thus, be damage-ranked 1 on this basis. The post-recovery damage means the cracking and bending cannot be absolutely classed as deliberate damage (i.e. damage-ranked 3), but the survival of original corrosion and earth in the fracture coupled with the thickness of this piece strengthens this hypothesis. Furthermore, experimental research indicates that it is difficult to break tin-bronze flat axeheads (Moyler, Reference Moyler2007: 144–49), and a comparison of this axehead with other broken flat axeheads in Scotland reveals a similar pattern. Overall, the bending and breaking can be damaged-ranked 2.
A clearer assessment and contextualization of the damage observed on this axehead allows us to approach this object from a new, informed perspective. Drawing on a range of factors, one can assess the biography of this object while clearly articulating the reasons for the assessment, something that is often missing from interpretations of such material.
Discussion
The steps illustrated in these case studies are usually quick processes of judgment, and there is inevitably flexibility within the rankings presented here. For instance, an object that has broken across a structurally weak point (Damage Ranking 1), but also displays deliberate impact marks associated with the breakage (Damage Ranking 2), might be more appropriately considered within Damage Ranking 2 than 1. A sword fragment evidencing a multi-break (Damage Ranking 3), but which has been cleaned of any patina and has a contested post-recovery history, must be considered ‘Uncertain’, as the post-recovery processes hinder accurate determination. There is unfortunately no overarching rule for assessing objects showing multiple forms of damage and what should take precedent. Moreover, each ranking only applies to individual forms of damage, which contributes to the overall narrative of that object, encompassing aspects of manufacture, use, post-depositional treatment, and so on. The important aspect I wish to stress is that, when making these assessments, we should follow approaches grounded in informed and ongoing research. The individual case studies show how the DRS can be applied to single finds and how this can in turn contribute to our understanding of the objects, in isolation as well as in relation to similar or associated material. By recognizing the likelihood that damage seen on metalwork is intentional, we can develop narratives that accurately portray an object's history immediately prior to deposition, without resorting to ‘obvious’ features. The damage on the spearhead from Sandy's Farm, for instance, is difficult to interpret without drawing on the parameters set out here. However, by recognizing that at least some of the damage may be accidental due to structural weaknesses, and drawing on parallel fragmentation of other contemporary spearheads, one can develop ideas with greater confidence about how this object became damaged in different ways and how these relate to each other. Similar analyses have been conducted on the Bronze Age sword from Werkhoven in the Netherlands (Fontijn et al., Reference Fontijn, Theunissen, van Os, Amkreutz, Bakels and Kamermans2012) and hot-short Early Bronze Age axeheads from Central Europe (Kuijpers, Reference Kuijpers2018).
A natural progression is to apply the DRS more broadly (e.g. to a group of objects or a region), which would allow an investigation into trends of intentional or unintentional damage. While this is not presented here, a preliminary study applying the DRS to Late Bronze Age socketed axeheads from Cornwall allowed me to identify patterns in the treatment of axeheads prior to deposition; most notably twenty-three of the thirty-two axeheads studied (c. 72 per cent) were probably or definitely deliberately damaged (Knight, Reference Knight2017: 213–18). Moreover, single finds and hoarded objects showed similarities in treatment, raising the possibility that distinguishing between the two depositional practices may not necessarily be helpful. Current shifts in theoretical perspectives are embracing both symbolic and functionalist approaches to single finds and hoards of damaged objects (Fontijn, Reference Fontijn2019; Wiseman, Reference Wiseman2018); the DRS offers the chance to assess the variation and diversity of damage within single groups or across wider areas to interpret trends in damage, as well as relationships between complete and incomplete objects and use-related versus intentional damage. This can change and strengthen how we reflect on pre-depositional processes, such as selection, accumulation, and pre-depositional treatment, and ultimately gain insights into the social role of Bronze Age metalwork.
No attempt has been made to incorporate specific parameters of quantifiable properties, such as hardness or metallurgical composition, into the DRS. This is because of the variability in these features. One can generalize that a high tin-bronze is harder than a low tin-bronze, but to set parameters around the effect this might have would require additional analysis outside the scope of this research. An appreciation of these factors is, of course, essential and should form part of the discussion wherever possible, in much the same way as context should be considered for determining the nature of the damage. However, the generality of the DRS means it has the potential to be utilized even when no scientific analysis has been undertaken or detailed context is known, such as the volume of material recovered through metal-detecting.
Overall, it should be stressed that the DRS is a working methodology, subject to alteration and refinement as new research and analyses are conducted on Bronze Age metalwork. One can envisage incorporating the use of high-powered microscopy and 3D modelling, for example, to enhance our understanding of observed damage (see Horn & Karck, Reference Horn and Karck2019). The DRS offers a starting point for the classification and interpretation of damage that does not rely on assumptions about the material based from its form or presumed social role, but on knowledge of material properties, trends in the archaeological data, and integrating the increasing body of experimental archaeology involving bronze objects. From this we can build interpretations of deliberately destroyed metalwork with greater confidence.
Acknowledgements
This article stems from PhD research undertaken at the University of Exeter (2014–2018), funded by the AHRC (Grant no. AH/L503939/1). Numerous conversations with colleagues over the past few years have shaped it, and I am grateful to all who engaged in methodological chats. I am most thankful to the many museum curators who accommodated my research visits and allowed me to use photographs from their collections. I would like to thank particularly my three supervisors, Profs Anthony Harding, Joanna Brück, and Linda Hurcombe, for helping me hone my ideas. I am grateful to Valerio Gentile, Tobias Mörtz, and Fraser Hunter for offering their thoughts on earlier drafts of this article, and to the anonymous reviewers and editors for their critical insights. This article is much stronger as a result. All remaining imperfections are my own.
