Chelsea and Ballintober swords: Typology, chronology and use

Bronze Age Warfare: Manufacture and Use of Weaponry Edited by Marion Uckelmann Marianne Mödlinger BAR International Series 2255 2011 Published by Archaeopress Publishers of British Archaeological Reports Gordon House 276 Banbury Road Oxford OX2 7ED England bar@archaeopress.com www.archaeopress.com BAR S2255 Bronze Age Warfare: Manufacture and Use of Weaponry © Archaeopress and the individual authors 2011 ISBN 978 1 4073 0822 7 Printed in England by Blenheim Colour Ltd All BAR titles are available from: Hadrian Books Ltd 122 Banbury Road Oxford OX2 7BP England www.hadrianbooks.co.uk The current BAR catalogue with details of all titles in print, prices and means of payment is available free from Hadrian Books or may be downloaded from www.archaeopress.com CHELSEA AND BALLINTOBER SWORDS: TYPOLOGY, CHRONOLOGY AND USE Steven Matthews historically and culturally situated artefacts, and integral to wider discussions of Bronze Age societies. Typology and chronology should have an enormous bearing on any discussion of usewear, and in particular edge damage on weapons. Where once existed reasonably neat series of successive weapon types, with only minimal degrees of overlap (cf. Colquhoun and Burgess 1988), reassessment of these has created a far more subtle and complex landscape of rapier and sword types during the later Bronze Age.1 Such a complex theatre must, whenever possible, be given serious critical attention when looking at the use of different weapon types. In this paper issues of typology, chronology and use, will be discussed in relation to Chelsea and Ballintober swords, as a means of demonstrating aspects of the above agenda. ABSTRACT This paper lays out an agenda for the detailed recording and dissemination of edge damage on Bronze Age weapons, combined with detailed typology and chronology. This approach is applied to swords of Type Chelsea and Ballintober in southern England. Their relationship to other swords of related types in northwest Europe is discussed, where their origin is critically reviewed in light of an earlier emergence under the influence of rod-tanged swords of Type Arco/Terontola. The identification of a number of chronologically significant variants highlights the later modification of Chelsea and Ballintober swords under the influence of early Urnfield and Atlantic swords types, particularly in Ireland and France. KEYWORDS HISTORY AND TYPOLOGY OF CHELSEA AND BALLINTOBER SWORDS Chelsea – Ballintober – Arco/Terontola – Twickenham – Mixnam – Battersea – typology – chronology – edge damage – use wear English series These early tanged swords were first identified by H. W. M. Hodges (1956), and named for the find from Ballintober, Co. Mayo, which he regarded as intermediary between rapiers and the earliest indigenous flange-hilted swords. It was B. Trump (1962, 93) who first discussed their origins, which she saw occurring amongst the multitude of different weapon types in the Seine valley, before being introduced to England. G. Eogan (1965, 7 f.) similarly regarded rapier and sword development on the near Continent as being important but argued that Ballintober swords developed instead in England, in the Thames valley. C. Burgess (1968b, 15 f.) recognised that this series incorporated a number of different types, predicated on the form of the blade, of which there were three: Rosnoën rapiers, with parallel-sided blades and the same flat midsection as Group IV Atlantic rapiers, swiftly followed by Chelsea swords, retaining the same flat blade section but incorporating a leaf-shaped blade, and finally the Ballintober sword, which INTRODUCTION This paper presents tentative steps toward exploring the development of two important aspects of usewear analysis: 1) The detailed recording of individual edge damage and, most importantly, its dissemination so that comparisons with other corpora can be made; 2) The critical integration of both typological and chronological frameworks so that more detailed comparisons can be made spatially and temporally. Thus far, usewear analysis has largely failed in both of these respects. These aspects are important if we are to achieve a greater degree of integration between metalwork studies and other aspects of Bronze Age research, particularly when addressing regional material manifestations and their interpretation. Without these we risk our studies remaining at the level of mere anecdotal observation, at best testing the limits of the mechanical properties of these objects rather than presenting them as 1 Cf. Needham 1982; Burgess 1991; Bridgford 2001; Brandherm and Burgess 2008. 85 Chelsea and Ballintober swords: Typology, chronology and use Figure 1: Distribution of swords of Types Chelsea and Ballintober and their variants in Southern Britain saw the adoption of the flat-lozenge blade section seen on the earliest leaf-shaped Urnfield swords. Burgess,2 disagreed that its origins were to be found in relation to the earliest Urnfield swords, specifically Type Hemigkofen, arguing instead that it was a combination of influences involving Rosnoën rapiers and leaf-shaped rod-tanged swords of Type Arco/Terontola. Of the latter, there is a noticeable concentration in the Seine valley,3 and a single example from the Thames valley (Colquhoun and Burgess 1988, no. 3). M. Rowlands (1976, 79 f.) and S. Needham (1982, 28) lend credence to this argument, as both saw Chelsea as the earliest form and Ballintober undergoing constant later modification, with the arrival of these early Urnfield swords, though the emphasis here was upon Rosnoën rather than Arco/Terontola origins. With the publication by I. Colquhoun and C. Burgess (1988) of ‘The Swords of Britain’, in the Prähistorische Bronzefunde series, the mixed Rosnoën-Arco/ Terontola origins for Chelsea and Ballintober swords was maintained, only now Ballintober swords were early and Chelsea relegated to being a derivative local variant. The primary distribution of these swords is found in northwest Europe, with the greatest concentration being in southeast England 2 (fig. 1), with notable outliers in the southwest and in southern Wales (Colquhoun and Burgess 1988, pl. 118. 119), provenanced examples demonstrate a liberal spread across the north of Ireland (Eogan 1965, fig. 85), and on the Continent concentrations in northern and northwest France (Gomez 1987, fig. 1). For Britain Colquhoun and Burgess (1988, 19–24) list a total of 39 swords, consisting of 23 Type Ballintober swords (no. 22–43. 776), 12 variant Chelsea swords (no. 44–54. 775), and four believed to be Irish imports (no. 55–58). However, given the significance of the Chelsea sword, particularly as a restricted regional series, we believe it should be termed a type rather than a variant, as it has been latterly assigned. These two types (fig. 2), Chelsea with its flat-sectioned blade and Ballintober with its lozenge-sectioned blade, form the basis of the typological scheme used in this paper, which will be further subdivided into variants, which are discussed below. Colquhoun and Burgess (1988, 19) established that Irish Ballintober swords differ from these English swords in having shoulders that are far more widely splayed, and identify four swords as being probable Irish imports. These are all Ballintober swords and are included amongst this group in the discussion here, being regarded as merely late forms. The sword from 3 Cf. McArdle 1969, 81–91 fig. 9; partially illustrated by O’Connor 1980, map 30. Burgess 1968a, 44, postscript; 1974, 205, 318 no. 270. 86 Steven Matthews 5 2 3 6 1 4 Figure 2: Rapier and sword types discussed in the text. – 1. Type Rosnoën rapier from Kingston on Thames, R. Thames (no. 4). – 2. Type Arco/Terontola sword from Corbeil, Essonne; early Type Hemigkofen from the R. Thames (no. 64). – 3. Type Ballintober, variant Twickenham sword from Twickenham, R. Thames (no. 23). – 4. Type Chelsea, variant Mixnam sword from Kingston on Thames, R. Thames (no. 48). – 5. Type Ballintober, variant Battersea sword from Battersea, R. Thames (1. 3–5 after Colquhoun and Burgess 1988; 2 after Mohen 1977, no. 234) Sandford, Oxford (Colquhoun and Burgess 1988, no. 77) has been included amongst those Chelsea swords. Two unpublished Ballintober swords have come to the author’s attention, both from the River Thames at London, and are listed in the figure 10 as swords A and B.4 As well as a further Ballintober sword recovered from the gravel pits at Mixnam, Egham, Surrey, and listed here as sword C.5 An unusual northern example of a Ballintober sword from Barnhills Farm, Corsewall, Wigtownshire in Scotland, has recently been identified (Cowie and O’Connor 2007, 316–318 fig. 28. 2) but has not been included here. Though it is worth noting that the blade and hilt shape bear more resemblance to the Irish than English Ballintober series. 4 A: British Museum, London, acc. no. 1957.5-3.1. B: Salisbury and South Wiltshire Museum, Salisbury, acc. no. 1c4A.12. 5 Tomalin 1982, fig. 2. 1; not included in Colquhoun and Burgess 1988. 87 Chelsea and Ballintober swords: Typology, chronology and use Figure 3: Chronology and sequence of the sword development Irish series Gomez (1987). Coffyn (1985, carte 13) lists 16 examples, which he divides into two groups. Type classique being the characteristic Ballintober sword with lozenge sectioned blade and subrectangular tang with four rivet-holes, and Type évolvé having notched tangs and blade outlines. Gomez (1987) similarly divides the Continental series in two groups. Type classique is again the Ballintober sword, of which there are seven examples, and a continentale variant, again based on notched tangs and the presence of blade outlines, of which there are 11 examples. The typologies of Coffyn (1985) and Gomez (1987) do not, however, correspond with that of Colquhoun and Burgess (1988). There are both Ballintober (e.g. Gomez 1987, no. 5. 8) and Chelsea (Gomez 1987, no. 1. 3) swords in Gomez’s Type classique, and the variant continentale consists of both riveted and notched (Gomez 1987, no. 6. 7. 13. 14) tanged swords, plain and grooved blades (Gomez 1987, no. 15. 16. 18), and examples both with (Gomez 1987, no. 9. 16) and without ricassi. A review of the entire series from northwest Europe suggests that the difference in blade section, largely ignored in discussions of these swords in Ireland and on the Continent, is indeed the major typological distinction. It is possible to identify further variants on the basis of the shape For Ireland, Eogan (1965) listed 24 weapons and they remain undifferentiated within his Class 1 swords. Trump (1962, 93) was correct in her assessment that the Irish series was derivative of and later than the English series. The blades of the Irish swords are generally thinner, having a more sinuous outline, the tang is often more trapezoidal than any found in England, and the ends of the shoulders are noticeably extended. Tentatively, it would appear that the Irish series comprises 21 Ballintober6 and three Chelsea swords (Eogan 1965, no. 3. 10. 20). Colquhoun and Burgess (1988, 19, n. 2) suggest that nine of these Irish swords (Eogan 1965, no. 1–7. 12. 23) might be English imports, with all but one (no. 3) being of Type Ballintober. However, this author is inclined to view Eogan’s no. 4 and 5 as being of Irish form. Equally, all could simply be early examples of Class 1 swords. French series The similarity of the French series to those of England and Ireland was first discussed by J. Briard (1965, 164–166 fig. 55, 1–4), and more recently by A. Coffyn (1985, carte 13) and J. 6 Eogan 1965, no. 1. 2. 4–9. 11–19. 21–23. 618. 88 Steven Matthews of the tang and butt plate, which will be described below. and Needham (1982, 28) in emphasising an earlier primary role being played by Rosnoën rapiers in the development of Chelsea and Ballintober swords, with early Urnfield swords impacting only on later Ballintober forms, also leave open the possibility for influence from rod-tanged swords of Type Arco/Terontola. Based on the then long Penard chronology, Needham suggested a two stage development for Chelsea and Ballintober swords, with the earliest examples emerging during Ha A1, and therefore contemporary with Arco/Terontola swords and Rosnoën rapiers, followed by secondary influence from early Urnfield swords of Types Hemigkofen and Erbenheim during Ha A2, with the development of a fuller leaf-shaped blade and more pronounced drooping shoulders, mirroring those of the flange-hilted swords. Rowlands (1976, 79 f.) similarly envisioned a two stage development, with Chelsea swords emerging first followed by Ballintober swords, which exhibit greater degrees of experimentation in hilt form, as a consequence of influence from early Urnfield swords. The two stage development suggested by Needham has considerable merit but misses the crucial characteristic that might be used to identify late examples, and that is presence of ricasso notches, absent on swords of the English series but common on Irish Class I weapons,10 where the bevelled edges have been drawn out to form primitive ricasso notches, and examples of which are all lozenge sectioned Ballintober swords. If the illustrations are reliable in this respect,11 their occurrence is significant, as the earliest ricasso notches only begin to appear regularly in southern England on Atlantic swords of Type Limehouse.12 This suggests that Irish Class 1 swords were still current when the earliest Atlantic swords began to develop there during Ha A2, suggesting a later date for these Irish Chelsea and Ballintober swords than in England. A late date is also confirmed by the occurrence of a Ballintober hilt fragment with ricasso notches from the hoard at Kerguerou en Rédené, Finistère (Briard 1965, fig, 57, B), which was found associated with an Ha A2 hilt fragment of an Atlantic sword of Type Essonne, the northern French equivalent of Type Limehouse (Burgess and O’Connor 2004, 191). We have so far suggested that there were three stages of development in Chelsea and Ballintober CHRONOLOGY There is much debate concerning the dating of Chelsea and Ballintober swords (e.g. Gerloff 2007, tab. 13. 2; 2010, tab. 3), and whether they should be positioned earlier, in the mature stages of the Penard phase at the transition between Bz D and Ha A1, or later in Ha A2 during the early stages of Late Bronze Age 2, developing under the influence of early leaf-shaped Urnfield swords. The introduction of a short Penard phase, c. 1300–1150 cal. BC, (cf. Needham et al. 1996, 87. 90 ill. 15), means that the earlier divisions of Penard I and II, and their associated metalwork types (cf. Burgess 1974, 205; 1980, 266), are no longer valid, and has wider implications for the relative position of early sword types. The early date for Chelsea and Ballintober swords is based on their emerging in association with Rosnoën rapiers under influence from leaf-shaped rodtanged swords of Type Arco/Terontola,7 of which there are a number in the Seine valley, and a single English example from the Thames.8 An early date was also suggested by Briard (1965, 298), who saw them emerging prior to the arrival of the early Urnfield swords, and similarly in conjunction with Rosnoën rapiers during the later stages of his Atlantic Bronze Final I, our Penard phase. As our chronologies have altered significantly since Colquhoun and Burgess (1988) established the chronology of these different types, the relative position of the weapons discussed here have been compared and updated in figure 3. There remains scepticism as to the role that these rod-tanged swords played in the development of Chelsea and Ballintober swords, given the general paucity of examples in southern England (cf. O’Connor 1989, 126), with the early Urnfield swords still being seen as the more likely origin for the adoption of the leaf-shaped blade.9 Similar to Burgess, a number of early detailed discussions (e.g. McArdle 1969, 52; Coombs 1972, 173) also saw a significant role for these leaf-shaped rodtanged swords in the development of Chelsea and Ballintober swords. Unfortunately these remain unpublished. However, Rowlands (1976, 79 f.) 7 Burgess 1974, 318, n. 270; 1980, 266; Colquhoun and Burgess 1988, 22 f. 8 Burgess 1968a, 44, postscript; Colquhoun and Burgess 1988, no. 9. 9 Lawson 1982, 282; Gerloff 2007, tab. 13. 2; 2010, tab. 3. 10 Eogan 1965, 7, no. 5. 8. 9. 11. 13. 14. 20. 21. See also Burgess 1968a, fig. 3. 6; 1968b, fig. 3. 6. 12 E.g. Colquhoun and Burgess 1988, no. 97. 100. 101. 103. 104. 11 89 Chelsea and Ballintober swords: Typology, chronology and use swords: 1) their emergence in relation to Rosnoën rapiers and rod-tanged Arco/Terontola swords, providing the characteristic tang shape and leafshaped blade, followed by 2) secondary influence from early Urnfield swords, notably Hemigkofen swords, contributing further changes to the shape of the blade, tang and shoulders, and finally 3) the adoption of traits otherwise found only on early indigenous Atlantic swords, such as those of Type Limehouse/Essonne, in the form of ricasso notches. It is worth briefly considering the conditions under which the earliest development of Chelsea and Ballintober swords took place, particularly in respect of their relationship to Arco/Terontola swords, as this remains the greatest point of contention in both typological and chronological terms. Although mistaken in seeing the earliest origins of Ballintober swords occurring in the Seine valley, and then being introduced into England and then Ireland, Trump (1962, 93) was correct in alluding to the importance of this region in terms of early sword development. This region has not only the greatest variety of weapon types during this period (Needham 1982) but also a noticeable concentration of leaf-shaped rod-tanged Arco/Terontola swords.13 N. K. Sandars (1976, 237) is alone in observing that there was real, credible alternatives to the flange-hilted swords of Urnfield origin during this period, in the form of these rod- and tanged weapons, which are to be found liberally distributed across France and northern Italy. Indeed, the earlier straight-bladed rod-tanged swords had a significant impact on local Atlantic rapier traditions, resulting in the Ambleside/Bardouville series of swords (Needham 1982). In his description of this eclectic series of swords, Needham provided much of the necessary groundwork as to the possible origins of Chelsea and Ballintober swords. The tanged straight-bladed Ambleside/ Bardouville series of swords belong to early Penard, contemporary with Bz D, and emerged alongside Rosnoën rapiers, with leaf-shaped rodtanged Arco/Terontola swords belong to the same period, but emerging in Ha A1, late in the Penard phase. All three weapons types are prominent in the Seine valley, and in the case of Arco/Terontola swords, represent the earliest occurrence of leafshaped blades in the Atlantic region. The impact of these weapons on contemporary rapier traditions should not be underestimated, as contemporary Group IV riveted and notched Atlantic rapiers of Type Appleby are, unlike the rest of northwest Europe, largely absent from France, despite ha13 ving previously maintained an Atlantic rapier tradition of comparable scale to that of Britain and Ireland (Matthews 2010, 84). The Seine and Thames valleys invariably kept pace with each other, and likely represent a significant interaction zone, what Rowlands (1998, 162–166) termed the Channel Core Area. Therefore, whilst only one Arco/Terontola sword has been found in the Thames valley, it is not unreasonable to assume that they once had a comparable presence to that of the Seine valley. D. Coombs (1972, 172 f.) and I. Colquhoun and C. Burgess (1988, 21) have suggested the means by which the development of the tang on these rod-tanged swords could have progressed to that found on contemporary Chelsea and Ballintober swords. This would begin with the removal of the rod itself, and relocating the rivet-holes from the top of the blade onto the tang itself, which would have similarly been accompanied by a change in blade shape to accommodate changes in balance (Colquhoun and Burgess 1988, 21). These changes would presumably have occurred in the Thames valley rather than in France. Were this not the case, the French series would otherwise constitute an apparent chronological problem in terms of the typological development of Chelsea and Ballintober swords, as this region is populated by swords with primarily later rather than earlier typological traits. The absence of typologically early Chelsea and Ballintober swords was likely due to the overwhelming dominance of Rosnoën rapiers in this region, and also to continuing experimentation with swords of Ambleside/Bardouville types. Production in the Thames valley, however, has always orientated toward increased standardisation rather than experimentation and difference, a regional distinction also found amongst subsequent Atlantic swords. With the Rosnoën tradition largely absent from southern England, it is less surprising then that it was here, in the Thames valley rather than the Seine valley, that the earliest Chelsea and Ballintober swords emerged, took root and flourished. In terms of form there are noticeable differences between Arco/Terontola and Chelsea and Ballintober swords. The necessity of these changes has already been remarked upon above. However, it is worth noting that a number of Chelsea swords have a less than sinuous blade outline, one that errs close to parallel sided, and is highly reminiscent of some Arco/Terontola swords. It is true that on some, but not in fact Cf. McArdle 1969, Class B1 swords, 81–91 fig. 9; partially illustrated by O’Connor 1980, map 30. 90 Steven Matthews all, Arco/Terontola swords, the greatest width of the blade is located closer to the tip than on any Chelsea or Ballintober sword. However, as Colquhoun and Burgess (1988, 21) have suggested, the absence of the rod would have required that the greatest width of the blade be moved further back. The significance of this difference in blade shape between Arco/Terontola and Chelsea swords may in fact be more apparent than real, and one that should be considered in terms of the overall blade proportions or ratios, as we shall see below. Variant Mixnam These swords have noticeably elongated tangs with subtle trapezoidal or almost parallel sides. The spacing of the rivets can be of two forms, being either close together or wide apart but generally they are not placed close to the terminal of the tang. This terminal can be either rounded or flat. It is likely that variant Mixnam represents a later rather than earlier form, the tang being the longest and the blade the most sinuous. Weapons similar to those of variant Mixnam are common amongst Irish Class, and this is not surprising given that the Irish series likely represents a later development, subsequent to the development of Ballintober swords in the Thames Valley. These Irish weapons differ in a number of respects from their English counterparts. First, the top pair of rivet holes can be set far closer to the terminal of the tang than is usually found in England, and the tang is often more noticeably squared than in England. The Irish series likely requires the identification of subtly different variants than those in England but at present this is not possible given the difficulty in using the poor quality of the illustrations provided by Eogan (1965). CHELSEA AND BALLINTOBER VARIANTS Having identified probable early and later traits amongst Chelsea and Ballintober swords throughout northwest Europe, we can use these to establish a limited number of variants (fig. 2), largely based on the shape of the tang, that appear to have a greater degree of chronological validity than does the basic distinction in blade form between the Chelsea and Ballintober types. It should be noted that the below descriptions are based on ideal forms and hybrids, as is so often the case, are common, particularly amongst such a short lived series. Uniquely, the same variants also appear to be valid for both types. Variant Battersea Variant Twickenham Variant Battersea are those swords with a broad tang, usually with the appearance of being square. These swords clearly represent a different approach to strengthening the hilt attachment to that found on swords of variant Mixnam. It is possible that they are slightly earlier than Mixnam but not significantly so and both likely ran contemporaneously until Ballintober swords were no longer fashionable or viable. Battersea swords are also prevalent amongst the Irish series but here they often appear combined with the lengthened tang of variant Mixnam, demonstrating their contemporaneity. That a number of Battersea swords in Ireland have primitive ricasso notches again confirms their late date. Whether the variants recognized for England have any relevance to the local development of Irish Ballintober swords remains to be seen. However, as the greater part of the Irish series is here considered to be later than the majority of the English series, it is surely no coincidence that the majority of the Irish series conform to our Mixnam and Battersea variants, with a great number appearing to be a hybrid of both, but few of variant Twickenham. Variant Twickenham swords stand out readily from the more developed Chelsea and Ballintober variants. The tangs are usually short with an oval terminal and subtle trapezoidal sides that, on presumably early swords, sweep outwards in a gently sloping diagonal line about the end of much restricted shoulders where they join the butt plate or upper part of the blade. Already mentioned, the sword from Sandford, Oxford (Colquhoun and Burgess 1988, no. 7), has been included amongst our Type Chelsea, variant Twickenham. Although not a reliable association, this sword is also reputed to have been found with an Atlantic rapier (Burgess and Gerloff 1981, 106 f. no. 624) of Group IV Type Appleby, variant Weybridge. This type of rapier is tentatively assigned to very early in the Penard phase, as it retains the more archaic trapezoidal shaped butt, the rivet-holes placed towards the upper corners of the hilt plate rather than at the sides, the latter being more characteristic of later Type Appleby rapiers. This suggests an earlier rather than later date for variant Twickenham swords. 91 Chelsea and Ballintober swords: Typology, chronology and use Figure 4: L/W ratio of the discussed sword types 92 Steven Matthews Figure 5: The blade sections and units of the used terminology Variant Nantes some of the typological and chronological claims made above. The types and variants recognised in southern England also appear to be relevant to the French series, demonstrating a mixture of Twickenham, Mixnam and Battersea swords. A long, thin and almost universally notched variant is also found (Briard 1965, fig. 55, 1. 2) and differs from any English example, and is therefore perhaps deserving of a local designation. Although blade outlines have no typological significance they may at least be chronologically late, given their almost exclusive association with the Mixnam variant in England and with similar swords in France. This unusual French variant forms a fairly homogenous grouping in lower northwest France and has therefore been named variant Nantes, again irrespective of type. Confirmation of a late date for this form is provided by the hilt fragment of an undecorated example from the hoard at Kerguerou en Rédené, Finistère (Briard 1965, fig. 57, B), which also bears primitive ricasso notches. Blade comparisons M. A. Brown (1982) and S. Needham (1982) have demonstrated the validity of testing individual blade proportions against typological groups. The blade proportion is calculated on the basis of the blade length to maximum width ratio (blade width ÷ blade length). The blade length was measured from the tip to the bottom of any ricasso that might be present but does not include the ricasso itself. This allows comparison between a wider range of weapon types regardless of different hilting methods or the presence of a ricasso, which is particularly important for the Penard phase where a large number of different weapon forms co-existed contemporaneously. The individual blade length/width (L/W) ratios of a number of types relevant to the above typological and chronological discussion are plotted in figure 4, where we have compared Chelsea swords against Ballintober, Arco/Terontola, Hemigkofen and Rosnoën weapons. A clear but subtle difference between Chelsea and Ballintober swords in terms of their blade proportions is evident. Chelsea swords have a narrow L/W distribution of 8–11, whereas Ballintober swords have a much broader L/W distribution of 9–14. Moreover, Chelsea swords generally range much lower in their L/W measurements than Ballintober swords. The degree of overlap, however, justifies the identification of the same variants amongst both types. With Chelsea and Ballintober swords having identifiably distinct value ranges in the distribution of their L/W blade ratios, and Chelsea swords having been argued to be chronologically earliest, TYPOLOGICAL COMPARISON OF WEAPONS Having established the significance of a series of typo-chronological traits amongst Chelsea and Ballintober swords, represented by these different variants, we might further explore the possible chronologically successive influence of developed leaf-shaped rod-tanged Arco/Terontola swords and the early Urnfield Hemigkofen swords. Much of what has been discussed above can be substantiated by comparing a limited number of quantitative morphological characteristics between Chelsea and Ballintober swords and those of other contemporary weapon types. Two aspects of form will be compared, the blade and hilt, which substantiate 93 Chelsea and Ballintober swords: Typology, chronology and use Class Description Class 1 Cut: takes the form of a V-shaped indentation, with tapering sides that end in a narrow point rather than a U-shaped terminal, usually with no loss of material, forming a lip either at the terminal or on either side of the cut. Class 2 Notch: takes the form of a U-shaped indentation that generally has parallel sides, ending in a U-shaped terminal that is usually the same width as the beginning of the notch. The excess material from a notch is always missing. Class 3 Dent: takes the same form as Class 2 edge damage – a U-shaped indentation that generally has parallel sides, ending in a U-shaped terminal that is usually the same width as the beginning of the notch. There is no loss of material and a lip the terminal of the dent, or a bulb running the entire length of the damage, occurs on one side of the flat face of the blade. Class 4 Broad shallow notch: takes the same form as Class 2 edge damage but wherein the length is always greater than its depth and is bracketed by ends that are usually parallel sided. Excess material, in the form of a lip, is not present. Class 5 Broad shallow dent: takes the same form as Class 3 edge damage but wherein the length is always greater than its depth and is bracketed by gently sloping edges. There is no loss of material and a lip the terminal of the dent, or a bulb running the entire length of the damage, occurs on one side of the flat face of the blade. Figure 6: Description of the different use wear classes it is this group that we have compared with the remaining weapon types. As only a single Arco/Terontola sword is known from England, examples from the Seine valley were also included. The L/W ratio of these swords has the same restricted distribution to that of Chelsea swords, between 8–11/12, with some examples similarly ranging low in L/W measurements. Their overall distribution, however, entirely encapsulate the distribution of L/W ratios of Chelsea swords. The fact that, despite some differences in blade form, they share the same overall blade proportions lends greater credence to the view that Chelsea swords were indeed influenced by Arco/Terontola swords. The distribution of Hemigkofen swords is more complicated, and appears to fall into two discrete groups. The first have L/W rations between 8–9, and are represented by those swords with short broad blades. The second group have ratios between 10–14, and have long slender blades. Whilst the first group has a distribution similar to that of Chelsea swords their range is so high they fall almost entirely outside of their distribution, unlike the Arco/ Terontola swords. Both groups of Hemigkofen swords, however, have a distribution similar to that of Ballintober swords. The differences between Chelsea and Hemigkofen swords, and the similarity between the latter and Ballintober swords, appears to substantiate Needham (1982, 28) and Rowlands (1976, 79) view of a continual process of development amongst Chelsea and Ballintober swords in relation to new influences, with Hemigkofen representing secondary rather than primary influence. The two distinct distributions also lend considerable weight to the argument for earlier and later Hemigkofen forms (cf. Cowen 1951). Finally, Chelsea swords were compared with five English Rosnoën rapiers. These were found to have quite different distributions suggesting that, whilst they may share a common link in terms of their hilt arrangement and blade section, any further similarity ends there. These comparisons substantiate the primacy of Arco/Terontola swords in the development of Chelsea swords. The significant difference between the L/W ratios of Rosnoën rapiers and Chelsea, Hemigkofen and Arco/Terontola swords (fig. 4) demonstrate that they represent entirely different classes of weapon, as has always been suggested. Whilst the different L/W ratios suggest subtle but important differences between sword types, they indicate a real qualitative difference between rapiers and swords, substantiating the argument that they represent distinct suites of gestures and techniques (Burgess and Gerloff 1981, 113), and that the traditional terminological distinction of ‘rapier’ and ‘sword’ should be maintained. Hilt comparisons A single variable – the width of the shoulders – also provides a reliable basis on which to compare both variants and types within the Chelsea and Ballintober series, and again with other contemporary weapon types, particularly as changes 94 Steven Matthews in this aspect of their form has also been argued to have a degree of typo-chronological significance. Rowlands (1976, 79 f.) and Needham (1982, 28) discussed changes in the shape of the shoulders of Ballintober swords in relation to the form of shoulders found on early Urnfield swords. However, unlike flange-hilted swords, where the shoulders would be visible through the attached hilt-plates, the shoulders of Chelsea and Ballintober swords would not, being obscured entirely by the hilt in the same way they would have been on Atlantic rapiers and AmblesideBardouville swords. The angle of the upper face of the shoulders, whether high and straight, or short and drooping, would therefore have been stylistically inconsequential. The conjectural reconstruction of hilts for a number of weapon types found during Bz D–Ha A1 by Needham (1982, 28–36 fig. 12), including Ambleside/ Bardouville swords, Rosnoën rapiers and a number of rod-tanged swords similarly provide a likely model for the hilting arrangement of Chelsea and Ballintober swords. Whilst Needham suggested that these drooping shoulders might be a late trait, we believe it is instead the width of the shoulders. Attempts to widen the width of the butt as whole, most likely to mimic the greater width of shoulders on early Urnfield swords, had the unintended consequence of rendering almost horizontal the upper face of the shoulders on some later Ballintober swords. That this is a later trait is confirmed by the presence of such shoulders on those Ballintober swords with ricasso notches, as well as amongst the later Mixnam and Battersea variants, and on some French notched swords (e.g. Briard 1965, fig. 55, 1. 2), and is all but ubiquitous on the Irish series. We can see this trend in more detail if we compare the shoulder width of Chelsea and Ballintober swords with other contemporary weapon types. If we look at the shoulder width of Chelsea and Ballintober swords by type and variant they and contemporary swords have a mean average of: Ballintober 4.26 cm (19 examples); Chelsea 4.23 cm (12 ex.); Twickenham 4.11 cm (15 ex.); Battersea 4.57 cm (seven ex.); Mixnam 4.35 cm (12 ex.); Irish Class 1 swords 4.95 cm (22 ex.); French Ballintober 4,08 cm (17 ex.);14 Type Hemigkofen 5.15 cm (15 ex.); Type Erbenheim 5,5 cm (3 ex.); Atlantic swords: Type Clewer 5.78 cm (9 ex.); Type Limehouse 6.36 cm (13 ex.); Type Limehouse, variant Mugdrum swords of 6.66 cm (5 ex.); Type Taplow 5.81 cm (8 ex.); Type Limehouse/Essonne 5.45 cm (4 ex.). In summary, whilst the subtle difference in the English types and variants may not be wholly significant alone their chronological arrangement is confirmed by the difference in shoulder width found on the Irish Class I and French variant Nantes swords, both of which have similar shoulder widths to Type Hemigkofen swords and the earliest Atlantic swords of Type Limehouse/ Essonne. EDGE DAMAGE ON CHELSEA AND BALLINTOBER SWORDS IN ENGLAND AND WALES Having substantiated our typo-chronological framework for Chelsea and Ballintober swords we now turn our attention to certain aspects of their use, through a discussion of edge damage. The socio-functional aspects of Ballintober swords, in particular their depositional circumstances, have been briefly discussed by R. Bradley (1998, 125 f. fig. 28). We are concerned, however, with those aspects of their use life prior to deposition and after production, a part of the life-biography (Gosden and Marshall 1999) that is difficult to ascertain and often neglected. This section will introduce limited aspects of the recording, categorisation, dissemination and analysis of edge damage currently being prepared by the author (Matthews, in prep.) of the Atlantic rapiers and swords from northwest Europe, from the Taunton/Portrieux/ Mont-Saint-Aignan/Boix-Saint-Croix phase until the end of the late Wilburton/Blackmoor/St. Nazaire phase (c. 1400–950 cal. BC; fig. 3). Two aspects of edge damage will be discussed in this study. The first is the type of edge damage, and the second is the spatial distribution of each of these different types of damage (fig. 5. 6). These are presented according to the different typological groups in figures 7 and 8, and in tabular form by each individual sword in figures 9–12. Of the 43 Chelsea and Ballintober swords from southwest and southeast England and southern Wales (fig. 1), 21 have been examined and traces of edge damage recorded and analysed. This is presented in figures 9–12 by the catalogue numbers used by Colquhoun and Burgess (1988). Two Ballintober swords not included in that work have also been examined and are listed in figure 10 as A and B. Excluding 14 However, as an eclectic group this last number increases dramatically amongst the French series if we consider only those of variant Nantes. 95 Chelsea and Ballintober swords: Typology, chronology and use Figure 7: The compared edge wear damage on the swords of Types Chelsea and Ballintober 96 Steven Matthews Figure 8: The compared edge wear damage on the swords of Types Chelsea, Rosnoën and Arco/Terontola 97 Chelsea and Ballintober swords: Typology, chronology and use Type Chelsea (8 examples) No. Var. 44 M. 48 M. Name Upper 1 near Battersea, R. Thames Kingston on Thames, R. Th. Kingston on Thames, R. Th. Wandsworth, R. Thames Chelsea, R. Thames 49 T. 51 B. 52 B. 54 T. Thorpe Hall 77 T. Sandford, R. Thames 775 T. Oystermouth 2 3 Middle 3 1 2 Lower 3 2 1 2 3 2 2 2.2 2 3 3 3.2 2 2 3.2.2 2.2 R L R L R L R L R L R L R L R L Figure 9: Edge damage on Type Chelsea swords (after Colquhoun and Burgess 1988). – Variants: T. = Twickenham; M. = Mixnam; B. = Battersea Type Ballintober (13 examples) No. Var. Name Brentford, R. Thames Twickenham, R. Thames 22 M. 23 T. 24 M. Egham 25 M. Peterborough 36 37 M. M. Barking, R. Thames Kingston Bridge, R. Th. Battersea, R. Thames 38 T. 41 M. Worth 43 M. London 55 B. Unknown 56 B. Shepperton, R. Thames A T. R. Thames B T. R. Thames Upper 1 2 3.3 5 Middle 3 3 3.1 3 3.5 5 1 2 1 3.5.5 3.4 2 3 3 Lower 3 1 2.4 1 2 2.3 2.3.3 3 2.2 2.2.5 3.3 4 3.5 3 2 2 1 4 1.3.3.3 3 3 2 4.4 3 3 3 3 3.5 2.2.2.2 3 2 3 2.2.2 3 2.2.2 2(x4).3 2 2 1 3 1.2 3 4.4 3.4 2.2 5 3 3 2.2 3 3 3 4 2 2.2 4.2 2 4.2 2.2 2.4 2 2 3 R L R L R L R L R L R L R L R L R L R L R L R L R L Figure 10: Edge damage on Type Ballintober swords: (no. all after Colquhoun and Burgess 1988; A. B after author, see footnote 4). – Variants: T. = Twickenham; M. = Mixnam; B. = Battersea 98 Steven Matthews Type Arco/Terontola (1 example) No. Name 3 London, R. Thames Upper 1 2 2 Middle 3 2 1 2 2.2.2 Lower 3 2 1 2.4 2 2 2.2 3 2 R L Figure 11: Edge damage on a Type Arco/Terontola sword (after Colquhoun and Burgess 1988) Type Rosnoën (4 examples) No. 4 5 8 14 Name Upper 1 2 Kingston on Thames, R. Th. Lambeth, R. Thames Putney, R. Thames Methwold 2 2.2.2.2 3 Middle 3 3.3 3 1 1 5 1.3 3 1.3.3.3 Lower 2 3.3 3 3 2.2 2 3 1 1 1.3 2.3 3.3.3 2 3 3 2 R L R L R L R L Figure 12: Edge damage on Type Rosnoën rapiers (after Colquhoun and Burgess 1988) fragments, this represents just below 50 % of all Chelsea and Ballintober swords in England and Wales. The single English Arco/Terontola sword and four English Rosnoën rapiers have also been examined for edge damage and are included in this study.15 Figure 7 presents the edge damage from both Chelsea and Ballintober swords, and is arranged in a series of separate tables according to the five different classes of edge damage discussed below. The edge damage from these Chelsea swords is also presented in figure 8 alongside the small sample of English Rosnoën and Arco/Terontola weapons. on this basis. Similar to the approach adopted by R. D. A. Savage (1979, pl. 1) in his analysis of two Ewart Park swords from the Carp’s-Tongue hoard at Watford, Kent, this cutting blade is divided into three equal sections, which are then further subdivided into three. These divisions and the terminology employed are illustrated in figure 5. These divisions allow the frequency and distribution of different types of edge damage to be analysed and compared relatively, irrespective of the actual length differences exhibited not only by individual weapons within a series but also between different weapon series and types. Five different classes of prehistoric edge damage have been identified and these are described in figure 6. Although similar, there are some differences between the categories used here and those established by Bridgford (2000, 105) on the basis of her experimental damage types. Those described by Bridgford occur in significant quantities on swords of Type Ewart Park but rarely earlier. Violent and excessive edge damage is not common even on Type Wilburton swords, except where purposefully enacted (Matthews 2011). Moreover, the long term effects of depositional circumstances on individual swords, in terms of corrosion and bronze disease, have had a significant effect on the nature of edge damage. The simplification of the classes of damage used Edge damage: definitions and terminology The basic unit of analysis is obviously the individual weapon. In this study, these have been divided into a number of different analytical parts and grouped accordingly. As the research this paper is drawn from is concerned with both rapiers and swords (Matthews, in prep.), where ricassi are entirely absent from the former group, this element has been differentiated from the cutting blade and grouped with the hilt, so that comparisons between different weapon forms could be achieved. Measurements given in the tables below for the length of blade are therefore made 15 Colquhoun and Burgess 1988, no. 3–5. 8. 14; Needham 1982, fig. 6 no. 7. 99 Chelsea and Ballintober swords: Typology, chronology and use Balance point Ballintober Chelsea Sword no. Weight cm/gms Corresponding blade Length from Blade unit tip Top Bottom 44 591 g 8,924.1 15.1 28 Middle 1 28.8 24 48 309 g 3677.1 11.9 22 Middle 1 22.6 18.9 51 299 g 3,348.8 11.2 27.4 Middle 1 28.2 23.5 49 417 g 6,171.6 14.8 20.1 Middle 1 20.9 17.4 52 385 g 4,966.5 12.9 23.1 Middle 1 24 20 54 – – – – – – – 775 353 g 5,506.8 15.6 23.7 Middle 1 26.2 21.9 77 386 g 5,944.4 15.4 25.8 Middle 1 27.6 23 22 663 g 9,613.5 14.5 25.6 Middle 1 26.8 22.4 23 683 g 11611 17 29.3 Middle 1 30.8 25.6 24 587 g 8,687.6 14.8 26.8 Middle 1 27.8 23.1 25 517 g 7,444.8 14.4 26.5 Middle 1 27.2 22.6 36 393 g 7,427.7 18.9 29.2 Middle 1 32 26.6 37 583 g 8,628.4 14.8 25.3 Middle 1 26.8 22.4 38 332 g 5,046.4 15.2 22.9 Middle 1 25.4 21.1 41 261 g 3,053.7 11.7 19.2 Middle 1 20.6 17.1 43 605 g 8,772.5 14.5 27.8 Middle 1 28.2 23.5 55 560 g 8,568 15.3 29.6 Middle 1 30 25 56 338 g 5,577 16.5 24.3 Middle 1 27.2 22.6 A 387 g 7,314.3 18.9 27.8 Middle 1 31.2 26 B 492 g 6,199.2 12.6 23.5 Middle 1 27.2 22.6 Figure 13: The length/weight ratios (cm/gms) relating to the effective balance of individual Types Ballintober and Chelsea swords, as calculated from their weight x the distance of the balance point to the hilt; measurements in cm here have had to reflect these circumstances, as has the necessity of devising categories that have universal applicability to both the rapier and sword series, which themselves reflect a very real difference in both technical craftsmanship and use in combat. Edge damage: data It is clear from the tables in figure 7 that the overall volume of edge damage on Ballintober swords both as a whole, and as individual objects (fig. 10), is far greater than that found on Chelsea swords (fig. 9). Amongst Ballintober swords, every blade unit has some evidence of damage but usually only one or two instances, meaning that the edge damage is unequally distributed between the swords examined. For example, only one sword 100 (fig. 10 no. A) was found to have only one blade unit undamaged, whilst three swords (fig. 10 no. 38. 43. 56) had damage only on one blade unit. All Ballintober swords have at least some edge damage on half or more of their blade units, with the majority of this damage occurring between blade units Middle 3 and Lower 1, primarily the lower half of the cutting blade, and constituted by edge damage of Classes 1 to 3, with the greatest volume being Classes 2 and 3. Half of the Chelsea swords exhibited no edge damage (fig. 9 no. 51. 52. 54. 77), and of the remaining three had edge damage on only three blade units (fig. 9 no. 44. 49. 775) and the remaining one had damage on only two blade units (fig. 9 no. 48). Whilst there is a small concentration of edge damage of Class 2 on blade unit Lower 2 amongst Chelsea swords, there were otherwise no significant concentrations. What small amount of Steven Matthews Corresponding blade unit Sword Centre of no. percussion Blade Top Bottom unit 14,000 13,000 12,000 11,000 Medium 10,000 9,000 7,000 8,000 Good 6,000 5,000 4,000 3,000 Light 44 Middle 3 19.2 14.4 48 12.2 Middle 3 15.1 11.3 49 49 13.3 Middle 3 13.9 10.4 51 16.5 Middle 3 18.8 14.1 52 15.2 Middle 3 16 12 54 – – – – 775 14.7 Middle 3 17.5 13.1 77 15.2 Middle 3 18.4 13.8 22 16.6 Middle 3 17.9 13.4 23 20.3 Middle 3 20.5 15.4 24 16.6 Middle 3 18.5 13.9 25 16.3 Middle 3 18.1 13.6 36 17.9 Middle 3 21.3 16 37 17.5 Middle 3 17.9 13.4 38 15.3 Middle 3 16.9 12.7 41 12.6 Middle 3 13.7 10.3 43 18.3 Middle 3 18.8 14.1 55 16.5 Middle 3 56 15.9 Middle 3 18.1 A 23.2 Middle 2 B 17.5 Middle 3 18.1 51 52 775 77 22 23 24 25 36 Ballintober Ballintober 18.9 48 Chelsea Chelsea 44 37 38 41 43 A B 55 56 14 20 26 15 13.6 20.8 13.6 15 Figure 14: The distribution of Types Chelsea and Ballintober swords according to their balance category, calculated according to their ‘turning moment’ (weight x distance of the hilt from the centre of gravity = cm/gms), with the categories of Good (< 10,000 cm/gms), Medium (10,000–15,000 cm/gms) according to Bridgford’s experiments (1997, 104 f.) Figure 15: Length from tip to the centre of percussion, as represented by the widest part of the blade, and the upper and lower measurements for the corresponding blade unit; measurements in cm damage there was, restricted to edge damage of Class 2 and 3, was spread across the remaining blade units, except for Upper 1 for which there was no damage found on any individual Chelsea sword. The greatest number of instances across all blade units of a single type of edge damage was of Class 3, with 41 instances of damage on Ballintober swords and only three instances on Chelsea swords, with Class 2 damage being the next most significant with 39 instances across all blade units amongst Ballintober swords and 14 instances amongst Chelsea swords. The most significant volume of any one type of edge 101 damage was found to be Class 2 of which there 16 instances on Ballintober swords but only one instance amongst Chelsea and occurred on blade unit Middle 3. The next significant volume was also Class 2 of which there were nine instances on Ballintober swords and again only one instance on Chelsea swords, which occurred on blade unit Lower 1. The sample of Rosnoën and Arco/Terontola weapons (fig. 8. 11. 12) is not significant enough to draw any firm conclusions regarding their edge damage. Comparatively, however, it is noticeable that there is a more even spread of the number of instances of edge damage along the length of the Chelsea and Ballintober swords: Typology, chronology and use cutting blade, similar to that on Chelsea swords, compared to the noticeable bias in the distribution of edge damage toward the lower half of the cutting blade on Ballintober swords. Of the four Rosnoën rapiers examined one was found to have no edge damage at all (fig. 12 no. 5), whilst the remaining rapiers had damage on at least six of their blade units (fig. 12 no. 8), five of their blade units (fig. 12 no. 14) and four of their blade units (fig. 12 no. 14). The Arco/Terontola sword (fig. 11 no. 3) was found to have edge damage on all but 2 of its blade units. Like Chelsea swords, the majority of its edge damage was of Class 2 and 3, however, Rosnoën rapiers also had five instances of Class 1 edge damage, of which Chelsea swords exhibited no evidence of. More significantly, although having only half the number of examples, Rosnoën rapiers exhibited 35 instances of different classes of edge wear compared to only 17 instances amongst Chelsea swords. There is clearly a marked difference in both the quantity and the distribution of edge damage between Chelsea and Ballintober swords (fig. 7). Edge damage of all types show their densest concentration on the bottom half of the blade on Ballintober swords, particularly in respect of Class 2 and 3 damage, with notable concentrations in blade units Middle 3, and Lower 1 and 2. With the exception of blade unit Middle 3, all edge damage on Chelsea swords is concentrated in the centre and higher parts of the blade, with the only noticeable concentration of edge damage being of Class 2 in blade unit Middle 1. The damage on Chelsea swords is confined entirely to an even spread from Upper 2 to Middle 3 and only of Class 2 and 3. On Ballintober swords the distribution of edge damage is much wider, with at least every blade unit being affected across the entire group, although the Upper part of the blade demonstrates significantly less damage than the Middle Lower areas of the blade. The sort of heavy damage represented by edge damage of Class 1 is infrequent on Ballintober swords when compared to the far more numerous instances of Class 2 and 3, whilst damage of Class 1 is entirely absent on Chelsea swords. In order to assess the significance of these patterns concerning the spatial distribution of the edge damage on the two sword types, we must first establish a number of technical characteristics, in particular the position of the balance point and the centre of percussion, in relation to individual blade units, as both are likely to have been a contributing factor in the weapons effective capability, and consequently the occurrence of edge damage. 102 Balance Establishing the relative position of both the balance point and the centre of percussion requires that the swords studied were largely complete, and therefore broken swords are excluded. Of the twenty one swords examined in this study only twenty were therefore viable for further examination. The balance point of each weapon was recorded and measured from the top of the weapon. When calculated according to the relative position of individual blade units on each Chelsea and Ballintober sword, the balance point was found to correspond consistently to area Middle 1 (fig. 13). In her study of the Irish Bronze Age swords, S. Bridgford (1997, 104) assessed the effective balance of each sword by calculating the turning moment, being the weight x the distance of the hilt from the centre of gravity or balance point. These measurements presented here are approximate as no allowance has been made for the missing hilt plates, though this difference is likely to be negligible (cf. Bridgford 1997, 105), and unlikely to affect its assignation to a particular blade unit. The location of the balance point, as measured from the bottom of the hilt, the weight, and the approximate length/weight ratio for these swords is given in figure 13 by group. Bridgford arranged these cm/gms results into three balance categories based on her own subjective handling of these weapons. The position of these measurements in relation to their balance categories are shown in figure 14. As Bridgford did not present a category below that of Good we have termed this group Light. The majority of our Ballintober swords fell into the Good category. No. 38, a particularly short example, ranged below this, and no. 23, an especially large broad bladed example, ranged above in the Medium category. Only one Chelsea sword placed in the Medium category, whilst the remaining swords ranged throughout the Light category. Bridgford (1997, fig. 6) lists six Irish Class 1 swords in her study and all fall within the Good category. No corpus was provided by Bridgford but as only three Class 1 weapons are possibly Type Chelsea it is likely that the swords studied were Ballintober, and therefore match our distribution within the different balance categories. Finally, much in the same way that we have with the balance point, it is necessary to establish the relative position of the centre of percussion. This has been identified as that part of the sword which, when swung, would yield the greatest velocity and weight, and is therefore that part of the Steven Matthews blade most likely intended to come into contact with another object (cf. Bridgford 1997, 103; Karasulas 2004, 510). Different from the balance point, the centre of percussion is located forward of or rather below this point, approximately a third of the length of the blade back from the tip and ideally corresponds with the widest part of the blade (Brewis 1923, 255, n. 2). Figure 15 shows that for every individual sword examined the centre of percussion was consistently found to correspond with blade unit Middle 3, with only one exception (no. A). by the presence of twice as much edge damage on Rosnoën as Chelsea weapons, despite an even greater difference in blade form and the sample examined being half the size again. The greater distribution of edge damage on the latter weapons was more similar, however, compared to the more concentrated damage found on Ballintober swords. Although only one example was included in this study, it is of note that instances of damage on the Arco/Terontola sword displayed only edge damage of Classes 2 and 4 and Chelsea those of Classes 2 and 3. Discussion of edge damage CONCLUSION It is hard to account for the notable differences between the occurrence and spatial distribution of edge damage on Chelsea and Ballintober swords. There is no appreciable difference in the length of the two different sword groupings, with the median length of all British Ballintober swords being c. 47.8 cm and the median length of Chelsea swords being c. 46.1 cm. However, as already discussed, the overall blade proportions between the two groups differ (fig. 4), and clearly relate to their differing positioning within the balance categories in relation to their cm/gms ratio. Although slightly more than half the number of Chelsea swords were examined compared to Ballintober swords, it is clear that the occurrence of less edge damage on this group cannot be explained simply as a consequence of sampling, especially as there is almost twice as much edge damage on the Rosnoën rapiers compared to Chelsea swords, despite the sample size again being only half as big. The greatest degree of edge damage on the Ballintober swords, whilst not corresponding exactly to blade unit Middle 3, the location of the centre of percussion, the highest number of instances of edge damage at least occur consistently on the lower half of the cutting blade, on blade units Middle 3, Lower 1 and Lower 3. The spatial distribution of edge damage found on Chelsea swords, however, does not correspond at all with that found on Ballintober swords. On the former group almost no edge damage occurs at all on the Lower part of the blade. Instead, the highest number of instances of edge damage corresponds with blade unit Middle 1, which relates not to the centre of percussion but rather the balance point of these weapons. Despite the blade proportions of Chelsea swords being only slightly different from Ballintober weapons, it is highly likely that this significant quantitative difference in edge damage was a product of technique, as indicated In some way, both British and Irish writers have been correct in their differing interpretations of the origins of Chelsea and Ballintober swords. The three step model suggested here accommodates the basic divergent views on the subject, with both rod-tanged Arco/Terontola and Hemigkofen swords playing some role in their development. It seems to this author that the swords of Chelsea type do in fact stand at the head of the Ballintober family proper, closely followed by those of Ballintober type with their lozenge section (cf. Burgess 1968b, 15 f). It is this later type that was then transmitted to Ireland, and which eventually incorporated ricasso notches. There are clearly notable regional differences in the three series of swords. In France notched tangs, although more numerous amongst this small group of Continental swords, are equally numerous amongst the Irish and southern English groups. These notched tangs, however, only occur on Ballintober swords and never on Chelsea weapons, whilst blade outlines and primitive ricasso notches can be considered a late rather than typological trait amongst Ballintober swords. Short, slim line, sinuous blades are characteristic of Irish Class 1 swords. A notably trapezoidal tang, widest closest to the blade and narrowing toward the terminal, is also common in Ireland and France but not in England. There is a noticeable regional difference in the positioning of the rivets on these trapezoidal tangs, being more traditionally positioned in Ireland but situated at opposite ends of the tang in France, with the lower set often being notched. This suggests that, as valid as we believe the above discussion of variants to be to the wider series, there is a need to develop local manifestations of these variants for each region. For example, there appears to be no obvious variant Twickenham swords in Ireland, and the few probable Chelsea swords that are present 103 Chelsea and Ballintober swords: Typology, chronology and use are all of very late form given the shape of the tang. There is no reason to assume that any of the Irish Class 1 swords need be significantly earlier than Ha A2. Similarly, it seems likely that variant Nantes developed later rather than earlier. Early Urnfield swords of Type Hemigkofen will have arrived in northern France first and likely upset development of Chelsea and Ballintober series much earlier than in England, by which time the series had already taken hold. The Irish series appears to be entirely dominated by swords similar to variants Mixnam and Battersea, and various hybrids of both. Ricasso notches are few in France compared to Ireland and we can therefore assume, given that so few early flange-hilted swords have been found in Ireland, that they survived latest there. As for the edge damage, only the briefest and most tentative interpretation can at this point be offered. The devil, as they say, is in the details, and the subtle differences in both the numerical and spatial distribution of the classes of edge damage found on each of the weapon types discussed above suggests that the typological differences alluded to, whilst significant, provide only a partial picture of use. Instead, it is those techniques, those learned ways of doing things (cf. Mauss 1973), that will likely provide greater interpretative depth in this respect. This paper has sought to lay the foundations for such an approach, based on comparison, which although lies beyond the scope of the current paper, the author intends to explore elsewhere at greater length (Matthews, in prep.). BIBLIOGRAPHY Bradley, R. 1998. The Passage of Arms (2nd edition). Oxford, Oxbow. Brewis, W. P. 1923. The bronze sword in Great Britain. Archaeologia XXIII, 253–265. Bridgford, S. D. 1997. Mightier than the pen? (An edgewise look at Irish Bronze Age swords). In J. Carman (ed.), Material harm: Archaeological Studies of War and Violence, 95–115. Glasgow, Cruithne Press. Bridgford, S. D. 2000. Weapons, Warfare and Society in Britain 1250–750 BC. Unpublished PhD thesis, University of Sheffield. Bridgford, S. D. 2001. Late Bronze Age swords and spears from the Power Station and Flag Fen. In F. Pryor (ed.), The Flag Fen Basin: Archaeology and Environment of a Fenland Landscape, 309–317. London, English Heritage. 104 Briard, J. 1965. Les Dépôts Bretons et l’Age du Bronze Atlantique. Rennes, Travaux Rennes: Travaux du laboratoire ‘Anthropologie – Préhistoire – Protohistoire – Quaternaire Armoricains’, Université de Rennes. Brown, M. A. 1982. Swords and sequence in the British Bronze Age. Archaeologia 107, 1–42. Burgess, C. 1968a. The Later Bronze Age in the British Isles and North-Western France. Archaeological Journal 125, 1–45. Burgess, C. 1968b. Bronze Age dirks and rapiers as illustrated by examples from Durham and Northumberland. Transactions of the Architectural and Archaeological Society of Durham and Northumberland 1, 3–26. Burgess, C. 1968c. Bronze Age Metalwork in Northern England, c. 1000 to 700 BC. Newcastle, Oriel Press. Burgess, C. 1974. The Bronze Age. In C. Renfrew (ed.), British Prehistory: A New Outline, 165– 232, 291–329. London, Duckworth. Burgess, C. 1980. The Bronze Age in Wales. In J. A. Taylor (ed.), Culture and Environment in Prehistoric Wales. British Archaeological Reports, British series 76, 243–286. Oxford, BAR. Burgess, C. B. and Gerloff, S. 1981. The Dirks and Rapiers of Great Britain and Ireland. Prähistorische Bronzefunde, IV. 7. München, C. H. Beck. Coffyn, A. 1985. Le Bronze Final Atlantique dans la Peninsula Iberique. Paris, Publications du Centre Pierre Paris no. 11. Colquhoun, I. and Burgess, C. B. 1988. The Swords of Britain. Prähistorische Bronzefunde IV, 5. München, C. H. Beck. Coombs, D. G. 1972. Late Bronze Age metalwork in the south of England: Typology, associations, distribution, chronology and industrial traditions. Unpublished PhD thesis, University of Cambridge. Cowen, J. D. 1951. The earliest bronze swords in Britain and their origin on the Continent of Europe. Proceedings of the Prehistoric Society 17, 195–213. Cowie, T. and O’Connor, B. 2007. Late Bronze Age swords from Scotland: Some finds old and new. In C. Burgess, P. Topping and F. Lynch (eds), Beyond Stonehenge: Essays on the Bronze Age in Honour of Colin Burgess, 316–344. Oxford, Oxbow Books. Eogan, G. 1965. Catalogue of Irish Bronze Swords. Dublin, National Museum of Ireland. Gerloff, S. 2007. Reinecke’s ABC and the chronology of the British Bronze Age. In C. Burgess, P. Topping and F. Lynch (eds), Beyond Stonehenge: Steven Matthews Essays on the Bronze Age in Honour of Colin Burgess, 117–161. Oxford, Oxbow Books. Gerloff, S. 2010. Atlantic Cauldrons and Buckets of the Late Bronze Age and Early Iron Ages in Western Europe. Prähistorische Bronzefunde II, 18. Stuttgart, Steiner. Gomez, J. 1987. Les épées du Cognaçais (Charente) et la chronologie des épées du type de ChelseaBallintober en France. In J-C. Blanchet, G. Bailloud, J. Briard, C. Burgess, G. Gaucher, J-P. Mohen and C. Mordant (eds), Les relations entre le continent et les Iles Britanniques à l‘âge du Bronze. Congrès Actes du colloque tenu dans le cadre du 22ème Congrès Préhistorique de France (1984), 125–131. Amiens, Supplément a la Revue Archéologique de Picardie. Gosden, C. and Marshall, Y. 1999. The cultural biography of objects. World Archaeology 31 (2), 169–178. Hodges, H. W. M. 1956. Studies in the Late Bronze Age of Ireland: 2. the typology and distribution of bronze implements. Ulster Journal of Archaeology 19, 29–56. Karasluas, A. 2004. Zaimokuza reconsidered: The forensic evidence, and classical Japanese swordsmanship. World Archaeology 36 (4), 507–518. Lawson, A. J. 1985. An unusual Bronze Age sword from Highclere, Hampshire. Proceedings of the Hampshire Field Club and Archaeological Society 41, 281–284. McArdle, T. D. 1969. Personal Armament in Middle and Late Bronze Age France. Unpublished PhD thesis, Edinburgh University. Matthews, S. 2010. Notes on a typological scheme for Atlantic Rapiers in France. Association pour la promotion des researches sur l’âge du bronze (APRAB) bulletin 7, 82–85. Matthews, S. 2011. Wrapping bronzes: Pottery encased metalwork in southern England and northern France during the Late Bronze Age. Association pour la promotion des researches sur l’âge du bronze (APRAB) bulletin 8. Matthews, S. in prep. Techniques and Society: the Middle and Late Bronze Age weapons of northwest Europe. PhD thesis, University of Groningen. Mauss, M. 1973. Techniques of the body. Economy and Society, 2, 70–88. Needham, S. 1982. The Ambleside Hoard: A Discovery in the Royal Collections. British Museum Occasional Paper no. 39. London, British Museum. Needham, S. 1990. The Penard-Wilburton succession: new metalwork finds from Croxton (Norfolk) and Thirsk (Yorkshire). Antiquaries Journal 70 (2), 253–270. O’Connor, B. 1980. Cross-Channel Relations in the Later Bronze Age: Relations between Britain, North-Eastern France, and the Low Countries during the Later Bronze Age and the Early Iron Age, with Particular Reference to the Metalwork. British Archaeological Reports, International series 91. Oxford, BAR. O’Connor, B. 1989. The swords of Britain: Ian Colquhoun and Colin Burgess. Scottish Archaeological Review 6, 126–128. Rowlands, M. J. 1976. The Production and Distribution of Metalwork in the Middle Bronze Age in Southern Britain. British Archaeological Reports, British series 31. Oxford, BAR. Rowlands, M. 1998. Kinship, alliance and exchange in the European Bronze Age. In K. Kristiansen and M. Rowlands (eds), Social Transformations in Archaeology: Global and Local Perspectives, 142–177. London, Routledge. Sandars, N. K. 1976. Review: Hartmann Reim, Die spätbronzezeitlichen Griffplatten-, Griffdorn- und Griffangelschwerter in Ostfrankreich. Germania 54 (1), 237–239. Savage, R. D. A. 1979. Technical notes on the Watford sword fragments. In C. Burgess and D. Coombs (eds), Bronze Age Hoards: Some Finds Old and New. British Archaeological Reports, British series 67, 221–228. Oxford, BAR. Tomalin, D. 1982. Bronze Age swords of the Ballintober type found at Mixnams Pit, Thorpe, Surrey. Surrey Archaeological Collections 73, 163–167. Trump, B. 1962. The origin and development of Middle Bronze Age rapiers. Proceedings of the Prehistoric Society 28, 80–102. Steven Matthews Groningen Institute of Archaeology (GIA) University of Groningen Poststraat 6 9712 ER Groningen The Netherlands 105