9 Material, style and identity in first century AD metalwork, with particular reference to the Seven Sisters Hoard Mary Davis and Adam Gwilt Introduction (Mary Davis and Adam Gwilt) ‘Celtic’ (or native) and Roman artefacts have usually been categorised by art historians on the basis of their style, or alternatively by scientists according to their material composition. Surprisingly, few have attempted to interpret the combined complexities of material and style. During the first century AD, there was an enormous influx of new technologies and materials acquired through trade and by the invasion of the Roman army. At this time of acculturation and change, La Tène art was made and used in Britain, and its style continued to develop (or devolve). When both the style and material of decorated metalwork are examined, a complex picture emerges, highlighting a degree of ‘hybridisation’. Supposedly ‘native’ styles and supposedly ‘Roman’ materials (e.g. brass) are used simultaneously and deposited within the same hoards, for example the Seven Sisters and Melsonby (Stanwick) hoards. As a consequence, questions arise relating to how much materials and objects were linked with native or tribal identities. The specific historical narrative of this period must inevitably come into play: the confrontation of native Iron Age peoples, for example the Iceni, Silures and Brigantes with the Roman army are well known. In these regions, and others, there was strong and prolonged resistance and the clash of cultures became established and entrenched. Traditional yet evolving native styles and technologies were developing in response to a massive invading force, which itself brought an influx of materials, technologies and artisans. How much were these new technologies and materials being appropriated and fitted into ‘old’ ideas of wealth, status, prestige, ideology and artefact use? First indications suggest a distinct homogeneity and quality in the type of metal used for native metalwork, in contrast with early Romano-British artefacts, where scrap metal was almost certainly reused to produce a wide range of less specific alloys. This paper looks specifically at the Seven Sisters hoard, deposited during the period of Campaigning of the Roman army with the Silures tribe in south Wales on the western frontier of Britain. This is one of a similar group of metalwork hoards buried across Britain at around this time, for example those from Polden Hills, Saham Toney, Melsonby 146 Mary Davis and Adam Gwilt and Middlebie. Also from Wales, it is argued that the Pentyrch and Tal-y-Llyn hoards containing decorated metalwork (though here containing ironwork and shields respectively, and here with less emphasis on horse equipment) were deposited at around the same time. An important question therefore, is how much unity or regional divergence there was across Britain at this time, demonstrated through changing style, art and technology. Were ‘hybridised’ artefacts being produced for regional leaders and local vying elites? Here, the Seven Sisters hoard illustrates how a range of material and stylistic changes were occurring and how a similar pattern is discernible in western Britain, northern Britain and East Anglia between AD 50–75. These changes, particularly apparent upon horse equipment and vessels, are also witnessed upon weapons and larger items of personal adornment. Therefore, this hoard has broad significance for the characterisation and chronology of the late La Tène art style in Britain. In considering the apparent motivation for creation and the geography of use, the name Native Campaigning Art is proposed for this material, until a more accurate or satisfactory one may be found. The impetus behind this metalwork is seen (initially at least) as a selective native or Iron Age appropriation of Roman materials and technologies, rather than a controlling Roman dilution of an Iron Age style. The first half of this paper focuses on the combined material and stylistic attributes of the Seven Sisters hoard and the potential for wider interpretation and further analytical research. The second part outlines the chronological context for the Seven Sisters hoard, while also illustrating some emerging stylistic trends in La Tène decorated metalwork between AD 40–100. The Seven Sisters hoard: relationships between technology, style and function (Mary Davis) Introduction The Seven Sisters hoard consists of 37 pieces of metal work that were discovered in 1875. The artefacts were initially extensively reported on and illustrated by Romilly Allen in 1905, and then re-evaluated by Davies and Spratling (1976) who reinvestigated the hoard’s discovery as well as the artefacts themselves. Within the hoard there is both Roman and native British material, plus several ingots, casting jets and pieces of ‘scrap’ metal which are less easily categorised by style or period (Figure 9.1). The first account of the discovery and content of the hoard was published by Romilly Allen in 1905. It was found by children in a streambed in 1875, following a storm and flood, then kept in the possession of a farming family in the Dulais Valley. The find-spot is situated near Seven Sisters, northeast of Neath. It was not until 1902 that the finds were purchased (by a local antiquarian Dr W Bickerton Edwards) and donated to the Welsh Museum of Natural History, Arts and Antiquities, Cardiff, the immediate predecessor of the National Museum of Wales. In 1927, a relative of Edwards, offered a previously unknown strap-union from this hoard to the British Museum (Brailsford 1953, 62 and plate XI.2). It remains possible that further items from the hoard have remained unfound, lost or not reported. Davies and Spratling reviewed the circumstances of the hoard’s deposition but despite their best efforts to re-establish the find-spot (Davies and Spratling 1976, 123–5 and Fig. 2), there is still some doubt as to its precise location. 9. Material, style and identity in first century AD metalwork 147 Davies and Spratling (1976) divided the hoard into three categories: Roman, native British and Roman or native (and considered the last group to be predominantly native). The Roman material consists mainly of military equipment of which the strap union (Appendix 9.1: 04.130), and the tinned disc (Appendix 9.1: 04.143), which is part of a phalera, are the most diagnostic pieces that date to between c. AD 50–75. None of the other objects would be out of place in this time period. The style of the native material (except for the stylistically distinct and ‘restrained style’ of the pendant hooks), “point to the middle years of the first century A.D. for their manufacture and deposition, which agrees well with the dating of the Roman pieces in the hoard” (Davies and Spratling 1976, 137). For a detailed review of the chronology of the hoard see the second part of this paper. Functional categories and style When assessing function, the objects from the hoard have been grouped into four categories: • Horse equipment consists of items used for horses, chariots and carts, whether these are terrets, strap unions or horse pendants and whether in ‘curvilinear’, ‘geometric’ or Roman style. • Personal military ornament distinguishes this group of material from other personal decorative ornaments such as torcs, bracelets and brooches – of which there are none in the hoard, and other military material such as that used for horses. This group includes the pendant hooks and helmet crest. Figure 9.1. The Seven Sisters hoard. Reproduced in colour on page *** 148 Mary Davis and Adam Gwilt • Feasting and drinking items form a further group; there are five tankard handles from the Seven Sisters hoard, the largest group to have been found together in Britain. Tankard handles often exhibit the highest quality in terms of design and manufacture. It is also likely that the tightly folded pieces of bronze sheet metal are the remains of either drinking vessels or cauldrons. • Finally there is a group of material that seems to be associated with metalworking: ingots, weights, scrap metal and two casting jets. It is the inclusion of these items from the hoard that have led to discussions as to whether this was a founders hoard, and whether it was deposited by Romans or native Britons (Romilly Allen 1905; Davies and Spratling 1976). Stylistically, the hoard has been divided into three distinct groups: • Roman military material, which is mostly military horse equipment, but also contains a buckle. • ‘Geometric’ native/Iron Age style material, which is all related to horse gear. • ‘Curvilinear’ La Tène style material, which is composed of objects and scrap related to feasting and drinking and personal military equipment, but not horse equipment. Both the ‘geometric’ and the ‘curvilinear’ style objects equate to Davies and Spratling’s ‘native British’ material (1976), and are considered Iron Age as opposed to Roman. Metal alloy composition Davies and Spratling when assessing the artefacts in their paper state: ‘the limited space available has meant that many of the problematic aspects of the bronzes, which only extensive treatment could hope to resolve, have been merely touched upon, or altogether omitted from the discussion’ (Davies and Spratling 1976, 121). They go on to say: ‘We use the term ‘bronze’ throughout this paper in place of the more accurate but more clumsy ‘copper-base alloy’, although we recognise that several of the pieces may actually be of brass’ (Davies and Spratling 1976, 144). Peter Northover (unpublished) undertook metallurgical analysis on 16 out of the 37 pieces, and this data, generously provided by Northover, contributes significantly to the present study. Further quantitative and qualitative analysis has been done at the National Museum of Wales (NMW) by the author. Recent assessment and analysis of the composition of the artefacts has been undertaken to see how Roman and native materials compare, but also to look at the compositional differences in stylistically distinct native style artefacts (i.e. ‘geometric’ and ‘curvilinear’). It is now also possible to examine whether function as well as style has a bearing on what metals and decorative techniques were used. Analysis has shown a range of copper alloys present: • Bronze: copper and tin, where tin is 5% or more • Brass: copper and zinc, where zinc is 5% or more • Impure bronze: mainly tin/copper alloy (over 5% tin) where there is also a small amount of zinc present (approx. 1–3%) and sometimes lead as well. • Impure brass or ‘tin brass’ (Northover in Beswick 1990, 22), (over 5% zinc) where there is a small but significant amount of tin (1–3%) and sometimes also lead. • Pure copper: only trace amounts of other elements present. (Results will be reported in full elsewhere) Figure 9.2 shows the proportions in which these various alloys occur. Although the 9. Material, style and identity in first century AD metalwork 149 SEVEN SISTERS: METALS AND ALLOYS 20 18 16 NUMBER OF ITEMS 14 12 10 8 6 4 2 0 BRASS BRONZE IMPURE BRASS IMPURE BRONZE COPPER Figure 9.2. Graph showing the metal alloys present in the Seven Sisters hoard. majority of the artefacts are bronze or brass, the overall elemental composition of the hoard is as mixed metallurgically as it is stylistically: bronze, brass and gunmetal, both leaded and unleaded, are all present. Composition, function and style Metallurgical analysis can be considered in conjunction with a number of aspects of the objects found within the hoard such as the cultural style of the objects, their function, and the variety of applied decoration. When the alloy composition is plotted against the type of object found, some patterns are evident; most types of object are predominantly made of a specific metal type (Figure 9.3). For example much of the horse harness equipment is brass, whereas the tankard handles and the sheet scrap (vessels) are bronze. In part, the apparent relationship between alloy type and function may reflect stylistic/ cultural preferences. Thus, when the Roman and native styles are considered in conjunction with material analysis, a strong correlation is discernible. Iron Age ‘curvilinear’ material is bronze; ‘geometric’ material is brass, and the Roman material is mixed: brass, bronze, gunmetal and leaded bronze (Figure 9.4). The stylistically native artefacts fit into at least two distinct categories: all the horse equipment is brass, with a consistent zinc content of about seventeen percent, and all is decorated with polychrome enamel. The second group of artefacts are either of personal armour (helmet crest, pendant hooks) or related to feasting and drinking in the form of tankard handles or scrap sheet bronze; none are horse equipment. This latter group is all tin bronze with a consistent tin content averaging about twelve per cent; where this material has additional decoration, ‘sealing wax’ red glass is used. The grouping of the Seven Sisters objects by style, function and composition is striking. Brass is not used for objects of ‘curvilinear’ style; and functionally similar types of object are made from the same alloy (e.g. objects used for feasting and drinking are all bronze). However, the general trend for the manufacture of horse equipment from 150 Mary Davis and Adam Gwilt SEVEN SISTERS: METALLURGICAL COMPOSITION AND FUNCTION 16 14 NUMBER OF ITEMS 12 10 8 6 4 2 0 HORSE EQUIPMENT METAL WORKING BRASS BRONZE MILITARY PERSONAL ORNAMENT IMPURE BRASS IMPURE BRONZE VESSEL COPPER Figure 9.3. Graph showing the metallurgical composition and function of the artefacts in the Seven Sisters hoard. SEVEN SISTERS: METALLURGICAL COMPOSITION AND STYLE 14 12 NUMBER OF ITEMS 10 8 6 4 2 0 CURVILINEAR GEOMETRIC BRASS BRONZE IMPURE BRASS ROMAN IMPURE BRONZE UNKNOWN COPPER Figure 9.4. Graph showing the metallurgical composition and the style of the artefacts in the Seven Sisters hoard. Wales and elsewhere is broader. In addition to the Seven Sisters ‘geometric’ style pieces, much native horse gear from Wales is made in the ‘curvilinear’ style using bronze, as is one of the horse harness sets from Melsonby (Stanwick) (see below). Further study and analysis of trends are needed. Both ‘curvilinear’ La Tène and ‘geometric’ native styled objects in this hoard could be interpreted as elite, high status artefacts, made from specific and controlled metallurgical compositions. There is no mixing of materials resulting in gunmetal, and no addition of 151 9. Material, style and identity in first century AD metalwork SEVEN SISTERS: METALLURGICAL COMPOSITION 16 BRONZE 14 12 % TIN 10 8 6 4 2 BRASS 0 50 55 60 65 70 75 80 85 90 95 100 % COPPER CURVILINEAR GEOMETRIC ROMAN Figure 9.5. Scatter diagram showing the copper and tin content of items in the Seven Sisters hoard. lead – despite the fact that this was an option and the materials would have been available (e.g. Mattingly 2006, 139), and could have made the technological processes of casting the metal and applying the glass/enamel easier (Bateson and Hedges 1975, 185– 6; Maryon 1971, 174; Bayley and Butcher 2004, Appendix 9.1; Bateson 1981, 79–81; Northover 1999, 142–3). The control and purity of the copper alloys used for the Iron Age/native alloys is in contrast to some of the Roman material from the hoard. Most of the recognisable military gear is, like the native material, either bronze or brass – but of a less consistent composition. However, some pieces from the hoard, as with the Roman domestic material from Camerton (Cowell 1990), as well as many contemporary brooches (Bayley and Butcher 2004), and other Roman/Romano-British artefacts (Dungworth 1996; 1997, 5), are of more mixed alloys such as impure brass, gun metal and impure bronze. The metalworking material is also interesting; the pelta shaped brass ingots are native rather than Roman in form (see Joy this volume) and, as with other native ‘geometric’ objects from the hoard show a very consistent and unadulterated brass composition. Trace element analysis is inconclusive as to whether there are characteristic compositions for native as opposed to Roman brass artefacts, though native material in this hoard does show a tight compositional group (see Figures 9.5, 9.6 and 9.7). The percentage of zinc used for these artefacts is lower than would be produced by an efficient cementation process (Bayley 1990), and implies the metal could have been melted previously (there is a decrease in the zinc content of brass of approximately ten percent with each remelting (Bayley 1990; Dungworth 1997, 8)). However, as stated above, the ‘geometric’ style objects maintain a high degree of purity in comparison with other objects made from remelted brass, such as brooches and domestic Roman items. The native character of some of the metal working material is illustrated by the presence of a ‘Celtic’ weight in tin bronze, identified by Spratling (Wainwright and Spratling 1973). The two casting jets – the most evident pieces of ‘scrap’ in the hoard are probably Roman and show the most mixed and inconsistent composition of all the objects. 152 Mary Davis and Adam Gwilt SEVEN SISTERS AND CAMERTON 16 BRONZE 14 12 % TIN 10 8 6 4 2 BRASS 0 50 55 60 65 70 75 80 85 90 95 100 % COPPER SEVEN SISTERS CURVILINEAR CAMERTON IRON AGE SEVEN SISTERS GEOMETRIC CAMERTON ROMAN SEVEN SISTERS ROMAN Figure 9.6. Scatter diagram showing the copper and tin content of items in the Seven Sisters hoard and the Camerton assemblage. SEVEN SISTERS, CAMERTON & MELSONBY (STANWICK) 30 25 BRASS % ZINC 20 15 10 5 BRONZE 0 50 55 60 65 70 75 80 85 90 95 100 % COPPER SEVEN SISTERS CURVILINEAR CAMERTON IRON AGE SEVEN SISTERS GEOMETRIC CAMERTON ROMAN SEVEN SISTERS ROMAN MELSONBY HORSE HARNESS Figure 9.7. Scatter diagram showing the copper and zinc content of items in the Seven Sisters hoard, the Camerton assemblage and the Melsonby (Stanwick) hoard. Seven Sisters, Camerton and Melsonby (Stanwick) The elemental composition of objects from the Seven Sisters hoard may be compared with existing data on the metalwork assemblage from Camerton in Somerset (Cowell 1990) and with metalwork from the Melsonby, North Yorkshire, hoard (Dungworth 1996; Fitts et al. 1999): these show both chronological and stylistic parallels with the material from Wales. 9. Material, style and identity in first century AD metalwork 153 Figure 9.5 shows a graph of copper plotted against tin for the Seven Sisters hoard. This illustrates that the ‘curvilinear’ La Tène style material from Seven Sisters, forming a discrete group towards the right-hand top corner is bronze, and the ‘geometric’ style objects forming a different group, along the lower axis, are manufactured from brass. The Roman material includes both bronze and brass, but has a more heterogeneous distribution in terms of composition due to more variable concentrations of tin and zinc, and by the addition of lead. This pattern of composition is reinforced when the Seven Sisters metal work is compared directly to artefacts from Camerton (Figure 9.6). The Camerton Iron Age material predominantly consists of horse equipment, button and loop fasteners and tankard handles (equating with the ‘curvilinear’ La Tène style material from Seven Sisters). These objects are mostly of unleaded tin bronze (top right-hand corner), whereas the Roman objects are of both bronze and brass and have a much broader scatter (Figure 9.6). Again there is a strong difference between the small range of alloys used for the native artefacts compared with the wide range used for the stylistically Roman material. Cowell (1990) has observed that in the Camerton assemblage, Roman military equipment is mostly of unleaded brass or bronze (i.e. they are of relatively controlled compositions as with the Seven Sisters material) but leaded copper alloys and gunmetal are commonly used for the manufacture of personal and domestic Romano-British items. There appears to be less close control over the manufacturing processes relating to personal or domestic items. The Seven Sisters hoard can also be compared directly with the horse harness sets from Melsonby analysed by Dungworth (1996). Both Leeds (1933) and Macgregor (1962) observed, on stylistic grounds, four separate sets of horse harness equipment within the Melsonby hoard. These observations were later confirmed, and slightly amended by analytical work carried out by Dungworth (1997, 5; in Fitts et al. 1999). One harness set is of bronze, and three are of brass. In Figure 9.7, copper is plotted against zinc; the brass ‘geometric’ style artefacts from the Seven Sisters hoard sit directly with the three brass harness sets from Melsonby (top right-hand corner). The bronze horse harness set from Melsonby, the ‘curvilinear’ style objects from Seven Sisters hoard and the Iron Age objects from Camerton all sit along the horizontal axis. The relatively tight grouping of the native pieces is in contrast to the larger variation in the composition of the Roman copper alloys. Although regionality of styles has been alluded to for the native ‘geometric’ objects (Macgregor 1962), there does seem to be some broader pattern in the technological style. The use of relatively pure brass as a metal, continued employment of the lost wax method of casting and the use of new alloys for the production of horse harness equipment all seems consistent between south Wales and North Yorkshire. Further analysis of material such as the hoard from Middlebie (Dumfriesshire), and the Santon and Saham Toney hoards from Norfolk would make interesting comparisons. Decorative Techniques Within the Seven Sisters hoard there is a wide range of decorative techniques applied to the metalwork including the addition of glass, enamel and niello to recesses, plus the use of inscribed and punched decoration directly to the metal surface. The style of decoration, the form of application of additional materials and the colour of both supplementary decoration and the metal substrate allow for numerous possible 154 Mary Davis and Adam Gwilt SEVEN SISTERS: DECORATION AND STYLE 12 NUMBER OF ITEMS 10 8 6 4 2 0 RED GLASS POLYCHROME ENAMEL TIN/NIELLO NATIVE CURVILINEAR ENGRAVED LINES NATIVE GEOMETRIC PUNCHED ROMAN Figure 9.8. Graph showing the decoration and style of the artefacts in the Seven Sisters hoard. combinations of effects; but again there are clear discernible patterns as to what is used in what context. Figure 9.8 shows, for example, that red glass is only present on native ‘curvilinear’ style material, polychrome enamel is used exclusively on ‘geometric’ style objects and tinning and niello only on Roman material. Heated glass inlays versus enamels As stated above, Davies and Spratling’s native style artefacts, which they refer to as either ‘restrained’ or ‘polychromed jewelled’, have been separated into two stylistically distinct groups. The ‘restrained’ (‘curvilinear’) group represents recognisably late La Tène style and has flowing curvilinear patterns; shapes and voids are filled with cross-hatching or inlaid glass to emphasise the motifs. ‘This technique was used to integrate recessed areas of copper alloy with inlaid opaque red glass to form complex curvilinear patterns’ (Rigby in Stead 2005, 120). The native, but technically and stylistically very distinct ‘polychromed jewelled’ group (‘geometric’), incorporates the use of imported Roman technology, but uses recognisably Iron Age or ‘Celtic’ form and style. Either ‘red glass’ or ‘coloured enamels’ are used to decorate the native style artefacts from the hoard. Although ‘glass’ and ‘enamel’ are similar materials, the two terms are used here to distinguish between the better-preserved, larger areas of red inlay, and the smaller areas of degraded polychrome decoration. The appearance of the inlays is very different and they were manufactured in technologically distinct ways. The term ‘glass’ is used here to denote a heat-softened inlay of red glass, and ‘enamel’ as an inlay applied as ground glass within the cells, which was fused in situ by heating to form a cohesive block of colour. The Iron Age red ‘sealing wax’ glass, heavily identified with late La Tène material, has an intense colour and opacity due to manufacturing the glass in reducing conditions in which cuprite dendrites are formed. In Wales, so far this glass has only ever been found applied to a bronze substrate, and in this hoard it is present on the pendant hooks and two 9. Material, style and identity in first century AD metalwork 155 of the tankard handles (‘curvilinear’ style). The decoration was almost certainly applied by cutting the glass to shape and then inserting it into shallow cut or cast-in recesses (Haseloff 1991, 639; Rigby unpublished). The glass pieces could then be heat-softened so as to fit accurately to the shapes, and to help fuse them to the metal substrate; the surface was then polished. Excessive heat would be avoided; the inlays were not melted and it was important to avoid oxidising the red coloured cuprite to green copper oxide. These recesses are carefully shaped, often have curved edges and a relatively large surface area; they interplay with areas of metal to form the main element of the design. In the ‘geometric’ style material, the designs used are executed in a different manner “ with (their) greater emphasis on rectilinear work which appears to have been a new development at about the time of the Claudian invasion’ (Davies and Spratling 1976, 137). These objects have small, shallow cast cells of regular shape (rectangular, petal etc.) filled with varying colours of enamel. The inset decoration is surrounded by narrow metal borders; several cells together define ‘geometric and complex curvilinear designs’ (Rigby unpublished). The enamel was probably applied as crushed or ground glass (Henderson 1991) and then heated in situ to melt the glass and fuse it to the metal. The small cells would help reduce cracking caused by the differential expansion and contraction occurring as the object is heated and cooled (Maryon 1971, 175). The enamels are much more variable in composition than the red glass and much more degraded, which makes it harder to establish their original colours (probably a combination of red, yellow, white and blue). This is a style again seen in south east Wales in the strap union from Chepstow (Taylor and Brailsford 1985; Savory 1976) and the Boverton collar (Gwilt – see below) and which Macgregor refers to as “the Silurian predilection for enamel in rectangular cells” (Macgregor 1962, 34). Glass compositions in the hoard Henderson concluded that the red glass present on the pendant hooks fitted into a tradition of inlaid red glass dating to the late Iron Age in Britain (Henderson 1989a) and especially Wales (1989b). Analysis also showed distinct compositions for the polychrome enamel inlaid into the ‘geometric’ style artefacts; these are relatively degraded but analysis has helped determine their originals colours. • Red glass: high copper (average 9%), high lead glass (average 31%); coloured by copper oxide dendrites • Red enamel: low copper (average 1.9%) high lead glass (average 28%); coloured by small discrete particles of copper and copper oxide • Yellow enamel: high lead (average 17%); coloured by lead antimonate • White enamel: low lead (average 0.3%); coloured by calcium antimonate. • Blue enamel: copper (1–6%); calcium antimonate. Figure 9.9 shows how these two forms of enamel/glass are compositionally distinct. The opaque red glass has higher copper and lead oxide content than most of the polychrome enamel and correlates closely to red glass used on many different Iron Age artefacts from England and Wales. The polychrome enamel is more diverse compositionally, but is much more comparable to Romano-British enamels than to Iron Age glass. 156 Mary Davis and Adam Gwilt COLOURED GLASS AND ENAMEL 50 45 40 % LEAD OXIDE 35 30 25 20 15 10 5 0 0 2 4 6 8 10 12 14 % COPPER OXIDE IRON AGE RED GLASS ROMANO-BRITISH COLOURED ENAMEL SEVEN SISTERS COLOURED ENAMEL SEVEN SISTERS RED GLASS Figure 9.9. Scatter diagram showing the copper oxide and lead oxide content in the Seven Sisters glass and enamel, plus other Iron Age and Romano-British examples. COLOURED GLASS AND ENAMEL 16 14 % ZINC OXIDE 12 10 8 6 4 2 0 0 2 4 6 8 10 12 14 % COPPER OXIDE IRON AGE RED GLASS ROMANO-BRITISH COLOURED ENAMELS SEVEN SISTERS COLOURED ENAMELS SEVEN SISTERS RED GLASS Figure 9.10. Scatter diagram showing the copper oxide and zinc oxide content in the Seven Sisters glass and enamel, plus other Iron Age and Romano-British examples. A further distinctive characteristic of the enamel used in the native ‘geometric’ artefacts from the Seven Sisters hoard, is its zinc content (Figure 9.10). Iron Age red glass and Romano-British enamels from sites such as Prestatyn (Henderson 1989b) and Dinorben, as well as those from England analysed by Bateson and Hedges (1975) show virtually no zinc content, whereas the Seven Sisters enamels show a significant zinc presence. However, Bateson and Hedges have noted idiosyncratic quantities of other elements within enamels such as tin, lead and antimony, which are present irrespective of their use 9. Material, style and identity in first century AD metalwork 157 as colourants, and suggest the recipes of individual craftsmen or workshops might have been used. This again is different to the red glass, which seems to have been made and traded in blocks (Hughes 1972, 99). The addition of zinc to the enamels may be due to special properties associated with zinc ores. Their seemingly alchemical nature in transforming copper to gold-coloured brass could have influenced its addition to the enamels, where it apparently serves no practical purpose. The contrast between these materials would seem to be technological as well as stylistic. Red glass was essentially Iron Age and polychrome enamel was originally Roman. However, in Britain polychrome enamelling seems to have been adopted and used culturally in a native/Iron Age way during the first century AD following the Claudian invasion. Other applied decoration The application of niello and tinning were both Roman decorative techniques, and are found only on the Roman military horse gear within the Seven Sisters hoard (Figure 9.8). Engraved or punched decoration applied directly to the metal surface occurs on many artefacts decorated in the late (curvilinear) La Tène style, including the pendant hooks and some of the tankard handles in the Seven Sisters hoard. However both techniques are absent on the ‘geometric’ objects within the Seven Sisters hoard (although not on the closely paralleled strap union from Chepstow (Savory 1976)). The use of engraved lines was common on many objects from Britain inlaid with red glass, where the inset decoration has a line around its edge (as seen on the pendant hooks). Within this hoard there also seems to be an increasing occurrence of more ornate lines including zigzags on the pendant hooks and linear areas of hatching on the tankard handles. Punched decoration is used on some of the Seven Sisters tankard handles and on other late La Tène artefacts from south Wales. Both engraved and punched decoration are present on several of the pieces of Roman military equipment and are widely used techniques during the Early Romano-British period (see the second part of this paper for a fuller discussion of decoratio and style). Discussion: Colour and style Both chemical analysis and stylistic interpretations suggest there are two distinct Late Iron Age traditions occurring: ‘curvilinear’ and ‘geometric’. Both styles are close chronologically and both are different to contemporary Roman material. The first of these, the ‘curvilinear’ La Tène style, consists of artefacts made from bronze using the lost wax technique. They are often decorated with inlays of red and sometimes yellow glass. The large filled recesses or voids in the metal are integral to the ‘curvilinear’ style designs, (as with the earlier use of basket hatching to enhance the design). The high lead content of both the red and yellow glass almost certainly made it easier to soften and inlay the relatively large areas; it would be technologically far more taxing to fill these voids or recesses with ground glass heated in situ. The largely restricted use of red and yellow coloured glass, but not blue or white, for more traditionally designed ‘curvilinear’ La Tène artefacts, seems significant to a greater extent in Wales in this period than in the much more rapidly Romanised south east of England. There is frequent use of red (and occasionally yellow) glass on horse equipment, 158 Mary Davis and Adam Gwilt and later on similarly ‘curvilinear’ styled artefacts such as tankard handles, mounts, bowls, and even figurines. Yellow glass seems to be acceptable in these contexts to some degree as with the Hambledon, Buckinghamshire strap-union (Haseloff 1991 642), the recent find of a massive strap union from Maendy in south Wales, and the massive armlets from north-east Scotland). In south east England, the correlations shown between type of metal, colour of decoration and style of ornament are not so clear-cut. For example, the ‘curvilinear’ style ‘Suffolk’/Lakenheath terret (Foster 2002) has blue glass used with red enamel, as does the harness brooch from Folly Lane (Foster in Niblett 1999); this object also uses a brass substrate for a ‘curvilinear’ style artefact. Other colours of inlay were also used; for example the Westhall quadrelobed harness mounts have a pale coloured inlay (Bateson 1981, 18). It therefore appears that the coloured inlay and the metal alloy used for these artefacts from south-east England were not selected to such rigid formulae. This region of England was using a combination of native and imported Roman technology soon after the invasion (e.g. the bridle bit from Folly Lane (Foster in Niblett 1999). However, there appears to be a more chaotic and less structured approach to the use of these different styles and technologies. Perhaps this reflects the more immediate upheaval produced by rapid Romanisation compared to the lengthy campaigning and formalised cultural resistance reflected in artefact technology further north and west. It is also possible that the relatively rigid use of colour is a reference to the past where the restricted and symbolic use of colours was practised more rigorously (e.g. symbolism equated with the martial/bloody and sometimes masculine nature of red as well as possibly its dynamic properties and magical powers (Jones and Macgregor 2002; Young 2006). This practice then retained more significance in frontier zones. Of the four main colours of glass used in the Iron Age in Britain (red, yellow, blue and white), blue and white are those used least for inlays, and are notable for their absence on most martial and feasting gear decorated in the ‘curvilinear’ La Tène style. Blue has possible gender connotations: colour symbolism is discussed in more detail by Mel Giles in this volume, but it is interesting to note that in the Yorkshire burials blue glass beads were exclusively buried with women (Fitzpatrick 2007; Giles this volume). It is also worth noting the extensive use of white, yellow, colourless and blue glass, but hardly ever red, for the beads from Meare and Glastonbury (Henderson 1987; 1995). There is also a technologically limiting factor for the application of different coloured inlays: both red and yellow glasses are heavily leaded, and are therefore more easily softened, cut and applied into large irregular shaped areas. Neither the blue nor white Iron Age glass would have these properties; they were more likely to be applied as coloured dots. It is not until enamelling techniques using powdered glass were introduced that these colours could be applied in more controlled ways to shaped champlevé recesses. Even then, larger cells appear to have been relatively difficult to ‘enamel’. The importance of the colour of the metal substrate should also be considered. The ‘geometric’ style objects in the Seven Sisters hoard are made from brass, which involved a very different technique of production (i.e. the cementation process (Bayley 1990)). Brass looks quite distinct from bronze and possesses a colour and sheen only previously seen on gold metal itself. The adoption of this gold coloured metal is notable when gold artefacts (non-coins) were uncommon in the west of Britain in the Iron Age, and in particular in Wales – apart from the odd coin imported from England and the border 9. Material, style and identity in first century AD metalwork 159 area (for possible Iron Age gold artefacts from Wales see Gwilt 2007). It is likely that imports of gold from Ireland to Wales had ceased by the end of the Bronze Age (Northover 1995, 529), and most gold entering Britain from the continent was to the south east of Britain, becoming relatively scarce further north and west of this region (Beswick et al. 1990, 27). The metal composition of prestigious Late Iron Age ‘curvilinear’ style artefacts from south Wales was bronze, without zinc or lead. This was closely monitored, as was the type of artefact manufactured and the type of design used; for example with decorated terrets, harness brooches, and vessels (see appendix). This implies a large degree of control of metal and metal manufacturing by an elite. The radical change in colour and design used for the ‘geometric’ style objects is striking, especially as the artefact types remain the same (horse harness equipment) and there is a similar use of a relatively pure and consistent metallurgical composition, though in this case using brass. The implication is that those commissioning/manufacturing both these styles of artefacts were innovative and well connected – and able to access both the technology and the materials needed. The near contemporary juxtaposition of these two Iron Age styles sends out two messages – one of change the other of continuity. There are several possibilities to consider for their apparent contemporaneity: are these ‘geometric’ style objects being commissioned by competing Iron Age elites, tapping into new resources and technologies and literally showing their colours? Alternatively are we viewing rapid changes in production at this time? Producers could have been responsible for accessing and using different materials, with consumers, a new elite, not minding the difference as long as the right messages regarding wealth and status were explicit. Both the old and the new styles were circulating and deposited together during the mid to late first century AD. It is likely that Late La Tène ‘curvilinear’ style material continued to be made in Wales during the first century AD, at a similar time to the ‘geometric’ style. The Seven Sisters hoard is a particularly good illustration of this: Davies and Spratling (1976) note that the pendant hooks were broken but not used, implying the contemporary manufacture of both ‘native’ styles of object. Despite the colour of the metal, artefacts manufactured from pure brass are not used for high status (non-brooch) personal ornament in Wales, such as torcs and collars (for example Dinnington (Northover in Beswick 1990, 22), Boverton (work in progress) and Tre’r Ceiri (Savory 1971, 67)); these are brass or bronze, but the metal is not so pure. They tend to contain tin and zinc, as well as some lead, and in this respect appear more like brooches. It is necessary to reference two tiers of personal ornament in the first century AD: first torcs and native armlets, rarely found in domestic or Romano-British military settings, and second, brooches, which represent a different and much more democratised style of personal ornament/decoration. The (relative) mass introduction of brooches in the first century BC into southeast England (Hill’s fibula event horizon 1997; Haselgrove 1997) and their subsequent spread, form a separate group of artefacts which is not the subject of this paper or project. Many brooches from the first century AD were manufactured using two-piece moulds and their metal content often incorporated the use of brass; this alloy was not used in Britain before the influence of Roman trade (Bayley 1990). The proliferation of brass occurred in Augustus’ reign during the controlled manufacture of brass coinage. Some of the earliest of the Romano-British 160 Mary Davis and Adam Gwilt brooches seem to be closely derived from La Tène III style brooches (i.e. Nauheim derivative and Colchester A brooches) and interestingly, the composition of these, like the elite metalwork from the Seven Sisters hoard is either relatively pure brass or bronze, not gunmetal, and with no added lead (Dungworth 1995). However, both the bronze and brass content within the majority of first century AD brooches appears debased compared to both coinage and Roman and native military ornament. The melting of state controlled Roman brass, with the subsequent addition of bronze and lead (Dungworth 1995) was probably happening amongst the entourage following in the wake of the Roman army. A large influx of artisans brought on an enhanced role for new forms of artefact by the sheer quantity of metal now accessible, and due to a cultural shift, accepting its availability. This allowed those from outside a very limited elite to use adornment made from material previously unavailable, and pushed in train the development of what are now perceived as Romano-British style brooches. The elite and functional quality of decorated Late Iron Age artefacts for horses, feasting and drinking from Wales and the skill used for their manufacture and design imply these bronze and brass objects were valued for their practical attributes, over and above the importance and worth attached to the display of personal ornament or precious metals; the relative strength and usefulness of brass may have appealed more than gold – which can really only be used for ornamentation or tokens of exchange. In many societies, this is why gold is perceived as valuable (Herbert 1984) – it is purely a luxury rather than a functional material; it is also a material that is best displayed as jewellery to promote individual social standing and wealth. So far only the Seven Sisters and Melsonby hoards containing Iron Age horse equipment have undergone detailed analysis, and both show a similar pattern in the use of either bronze or brass for certain types and styles of artefact. Some single objects such as the Folly Lane harness brooch and bridle bit (Northover in Niblett 1999, 142–3), and the Saham Toney strap union (Northover in Beswick 1990, 22) have also been examined. It is not yet possible to tell how unusual the use of pure brass is for Iron Age horse harness equipment and more analysis is needed. However, there is a discernible pattern emerging for objects discussed in this paper. Conclusion As highlighted, the artefacts of the ‘geometric’ Iron Age style in the Seven Sisters hoard involved the selective uptake of Roman materials and technologies, but incorporated them into objects, which were of recognisably native/Iron Age style and form. In this regard, they clearly contrast with the Roman military styles. The brass used appears to be of a finely defined composition, with the craftsmen maintaining the production of a high and consistent calibre of the alloy. The zinc content closely matches what Northover (1999) suggests is the maximum quantity that can be used in a brass while still producing a good cast (average seventeen percent). The metal chosen was certainly not for ease of use in producing a cast or in attaching enamel – so could imply the maintenance of skilled and elite metalworkers. The high zinc component of the enamels, and the pelta shaped brass ingots could suggest native manufacture of the materials used; or at least the acquisition of relatively newly manufactured brass, rather than the collecting and remelting of Roman scrap. Zinc ores are reasonably abundant compared to tin ores and occur in areas of relative proximity to south Wales. 9. Material, style and identity in first century AD metalwork 161 For example, there are extensive deposits of smithsonite (zinc carbonate) in the Mendips in Somerset (British Geological Survey 1998), an area already exploited for lead soon after the Roman invasion (Mattingly 2006, 139); many zinc ore sources also exist further west and north within Wales itself (Bevins 1994). Once the cementation technique of production had been mastered, materials needed for the manufacture of brass were probably easier to obtain than those needed for bronze. Late La Tène ‘curvilinear’ style material from Wales continues to be made during the first century AD at a similar time to the ‘geometric’ style. In Wales, the Snowdon bowl, the Seven Sisters pendant hooks and the Pentyrch terret are all first century objects with ‘curvilinear’ La Tène designs: all these are made from bronze with heat-softened red glass inlays, rather than enamelling sensu stricto. It is hoped further analysis can be undertaken, especially of enamelled artefacts from areas in conflict with Rome, for example Norfolk and south Wales, and other artefacts from the south west and north of Britain. It would be interesting to see whether style, technology and geographical locations correlate, and to assess the significance of colour in the Late Iron Age. This paper has shown the potential of using material analysis applied to technological and stylistic studies. Some trends are becoming apparent for native decorated metalwork in the first century AD, not just within the Seven Sisters hoard and south Wales, but possibly also in other regions during this period of resistance to, and assimilation with, Rome. The Chronological and stylistic context of the Seven Sisters Hoard (Adam Gwilt) Chronological Context Since its original publication, the Seven Sisters hoard has figured prominently within the wider Celtic Art literature for Britain (e.g. Wheeler 1925, 209–10 and Fig. 84; Leeds 1933, 101–5 and Pl. 2; Fox 1958, 127–9; Alcock 1963, 28–9; Megaw 1970, 173, Cat. 301; Spratling 1972; Savory 1976, 43 and 62–3, Cat. 34; Macgregor 1976, 25–6 and 178– 9; Jope 2000, 155 and Pls. 230, 276–7, 295). It has also contributed significantly to specific artefact studies of tankards (Corcoran 1952) and horse equipment (Spratling 1972; Palk 1988). The significance of Roman cavalry pieces, directly associated with metalwork exhibiting native Iron Age style has ensured its continuing place within narratives of the invasion and campaigning of the Roman army in Wales (e.g. Dudley and Webster 1965, 194; Webster 1981, 135; 1984, 282; Davies 2000, 11–12; Chapman 2005, 188; Howell 2006, 59–62). Here, it is necessary to restate and re-affirm the view that the Seven Sisters hoard was deposited during the third quarter of the first century AD, as Davies and Spratling rightly asserted (1976, 139). In other words, it is pre-Flavian or, at latest, very early Flavian in date. In this regard, it is contemporary with the decades of campaigning of the Roman army on this western frontier and situated within a context of military clash and immediate native responses. As the precise dating of the Seven Sisters hoard continues to have wider significance for the appearance and currency of metalwork styles and decorative techniques, in Wales and across Britain, a review of the basis for dating the 162 Mary Davis and Adam Gwilt hoard is merited, especially in the light of the last three decades of research and new discoveries. The most persuasive dating evidence within the hoard remains the Roman cavalry equipment (Appendix 9.1; Davies and Spratling 1976, 124–5, Cats. 1–5). The junction loop and ring (Appendix 9.1, 04.130) has a Type 4f junction loop and a cast ring of a type and form, largely of pre-Flavian use (Bishop 1988, 100 and 134, Table 4; Bishop and Coulston 2006, 120–1; Chapman 2005, 134, Cat. Td01, and for parallels see Bishop 1988, 160, Table 7). A similar junction ring was found in the chiefly cremation burial at Folly Lane, Verulamium, dated to around AD 50–55 (Rigby 1999, 182–92; Foster 1999a, 143–5, Cat. 8 and fig. 55; 1999b, 175–6). The strap union or junction ring (Appendix 9.1, 04.135) remains unparalleled (Bishop 1988, 134, Table 4), but has a junction loop of Type 3e attached (Bishop 1988, 160, Table 7; Chapman 2005, 133, Cat. Tc01, and for parallels see Bishop 1988, 159–60, Table 7; Webster 1993, 207–8, Cats. 19– 21). Two of three ‘double spectacle’ examples from the Orchard Site, Abergavenny are securely dated to the first construction phase of the Roman fort in AD 55–60 (Webster 1993, 207–8). Moreover this triangular strap union from Seven Sisters would seem to be of transitional form between earlier junction rings and phalerae, therefore confidently pre-Flavian. The phalera in this hoard is of Type 1g and could be Claudio-Neronian in date (Appendix 9.1, 04.143; Bishop 1988, 140, Table 5; Chapman 2005, 133, Cat. Tc02; Bishop and Coulston 2006, 121) This example finds close parallels with examples in the Fremington Hagg hoard (Bishop 1988, 140, Table 5; Webster 1971, 109 and Cats 3,16 and 17). This has been dated as pre-Flavian (Webster 1971, 108), or early Flavian at latest (Bishop 1988, 178, note 87). A similar example of Type 1e has been discovered in a pre-Flavian context at Usk (Webster 1995a, 38–9, Cat. 2). The ‘trifid’ pendant is of Type 1n, and may date from the Claudian period onwards (Appendix 9.1, 04.134; Bishop 1988, 96 and for parallels see Davies and Spratling 1976, 125; Bishop 1988, 146, Table 6; Chapman 2005, 150, Cat. Wb01; Unz and Deschler Erb 1997, 40, Cat. Nos. 1393–6). The female strap-slide (Appendix 9.1, 04.133) is not included in Bishop’s corpus, though a possible female and double ‘spectacled’ parallel is found in a pre-Flavian context at Usk (Chapman 2005; 136, Cat. Te03; Webster 1995a, 39–40, Cat. 6). To summarise, all five Roman cavalry pieces could date to as early as the Claudian period and both the junction loop and ring (04.130), and strap union (04.135), are probably pre-Flavian. Moreover, the range of parallels cited from other, mainly southern British and Continental Roman military contexts, is consistent with a Claudio-Neronian and possibly early Flavian currency. On this evidence, a date of deposition before AD 75 seems likely. The decorated turned circular object and buckle within the hoard (Appendix 9.1, 04.148 and 04.123; Davies and Spratling 1976, Cats. 6 and 10) would also not be out of place within Claudio-Neronian contexts. The former is paralleled by a similar curved fitment in the Folly Lane, Verulamium chiefly burial, securely dated to AD 50–55, which has been interpreted as a cart pole end (Foster 1999a, 148–50, Cat. 12). Here, it is also associated with a nave band, sharing similar Romanised ivy leaf motifs, as on the Seven Sisters piece (Foster 1999a, 146–8, Cat. 11 and Fig. 57). This cart or chariot presence is also paralleled with the axle mount found in the Melsonby (Stanwick) hoard, dated to AD 43–70 (Macgregor 1962, 36 and 52, Cat. 136; Fitts et al. 1999, 48). The two horse 9. Material, style and identity in first century AD metalwork 163 bells in the Seven Sisters hoard are of a form that span the first and early second centuries AD and are often found on Roman military sites (e.g. Webster 1995b, 55; Lloyd–Morgan 1997, 270). However, pre-Flavian examples are known, for example at Fishbourne and Usk (Cunliffe 1971, 112–3, Cat. 107; Webster 1995b, 55–8). In south Wales, a similar example has recently been found near Sennybridge, in association with a Late Iron Age toggle decorated with curvilinear La Tène decoration in-filled with red glass (Gwilt in press a). Dating evidence for the curvilinear and geometric La Tène or native style horse pieces (the bridle bits, terrets and strap unions), in this hoard currently relies upon detailed stylistic arguments and parallel hoard associations. All agree a general first century AD currency for Group II ‘Derivative-three link’ bits, Type 2 strap unions and knobbed terrets, continuing into the early second century AD in northern Britain (e.g. Leeds 1933, 113–26; Spratling 1972, 35–7, 94–7; Macgregor 1976, 25–30; Taylor and Brailsford 1985). For early associations, the Saham Toney (Ovington), Melsonby and Polden Hills hoards are particularly relevant. These have generally been dated to between the middle and the end of the third quarter of the first century AD, though there are conflicting views and emphases within this. The Boudican revolt of AD 60/61 has divided some, with regard the dating of the Saham Toney hoard (e.g. Macgregor 1976, 26; Hutcheson 2004, 33–4), however it was probably deposited between AD 50–75 (e.g. Spratling 1972, 311–2). The Melsonby and Polden Hills hoards are both dated to AD 43–70 (Macgregor 1962, 36; 1976, 26; Spratling 1972, 309; Brailsford 1975, 234; Fitts et al. 1999, 48). The Middlebie hoard is also relevant stylistically, though in this instance a less precise dating to the later first to early second century AD is posited (Macgregor 1976, 28; Hunter pers comm). Key for this paper, is the earliest appearance of polychrome enamels on geometric La Tène or native style metalwork. Its use within the Saham Toney and Fremington Hagg hoards is relevant, each deposited between AD 50–75 (see above). Most researchers have tended to see these developments as essentially post-Invasion (i.e. after AD 43), but they are difficult to place with any degree of precision. An important recent marker is the complete brass bridle bit, with geometric and polychrome glass and enamel decoration, found within the chiefly burial at Folly Lane Verulamium (Foster 1999a, 134–7, Cat. 1; Northover 1999, 136–7 and see above for dating). In south Wales, a substantial strap hook recalling the form of a dress fastener (Gillam 1958, 79–85; Wild 1970, 137–55), with La Tène motifs and double bosses and decorated with enamelled rosettes, has been found within the Roman fort at Abergavenny and dated to AD 50–75 (Savory 1993, 211–4). In form, it closely matches the similarly dated strap terminals in the Melsonby hoard (Magregor 1962, 38–43, Cats. 23–36; Savory 1993, 211). The geometric rosette finds good parallel with the Seven Sisters terrets and bridle bit. This piece was found within a levelling layer associated with the construction of the pre-Flavian fort, which is dated to AD 55–60. However, slightly later intrusive material also seems to have been added to this layer (Blockley 1993, 171 and 178). Nevertheless, when taken together, these examples currently provide reasonable evidence for the manufacture of the geometric style and polychrome enamels between AD 50–60. This early date has bearing on the, possibly broadly synchronous, development of the ‘Boss Style’ in northern Britain, whose currency has generally been seen as slightly later and spanning the later first and early second centuries AD (Leeds 1933, 110–2; Macgregor 1976, 29 and 100; Savory 1993, 213–4; Hunter pers comm). 164 Mary Davis and Adam Gwilt Four of the tankard handles within the hoard have a degree of stylistic coherence (Jackson 1990, 44–5), finding closest parallel with the Hod Hill, Waddon Hill (Stoke Abbot) and Camerton tankard handles (Brailsford 1962, 15, Cats. I5 and 6 and Fig. 14; Corcoran 1952, 99–100, Cats. 12–14 and Pl. X; Jackson 1990, 44–5, Cats. 119–20 and Pl. 12). These have been dated between the late 1st century BC and the mid 1st century AD, while both Hod Hill and Camerton, as sites, date to the early to mid decades of the first century AD, with activity ceasing here at around AD 50 and AD 60 respectively (Jackson 1990, 18–25; Richmond 1968, 117–23). The recent discovery of a tankard at Biddlesden, Buckinghamshire, with three surviving copper alloy bands and a handle is another close parallel stylistically. It is securely associated with Claudio-Neronian pottery (Jody Joy pers comm), providing important supporting evidence for a pre-Flavian date for the Seven Sisters tankard handles. One of the Seven Sisters tankard handles, with vertical zig-zag lines scorped up each groove (Appendix 9.1: 04.142), finds close stylistic parallel with an example from Newstead fort (Curle 1911, Pl. 54.7; Corcoran 1952, 101, Cat. 25 and Pl. IX.3; Davies and Spratling 1976, 133; Macgregor 1976, Cat. 290) and seems a reasonable candidate for being the latest manufactured piece within the hoard. The Newstead parallel is from an early site context, probably dating to the first phase of fort building during the 80s AD, immediately after the first Agricolan campaign. In addition, the Seven Sisters terret with scalloped flange and attachment loop (Appendix 9.1: 04.129;) also appears to be a Roman form and there are flanged and looped examples from Corbridge, Newstead, Poltross Burn and Melandra Castle (Curle 1911, Pl. 75.12; Forster and Knowles 1911, Pl. 4.2; Spratling 1972, 420, Cat. 98; Macgregor 1976, Cat. 99) from seemingly late 1st and early 2nd AD contexts. Nevertheless, it is suggested that the Seven Sisters tankard handle and terret date the early currency of these forms, while the northern British examples fix the later currency of long-lived types. In other words, it is not necessary, nor convincing, to force a post-AD 75 dating of the Seven Sisters hoard, on the basis of these two less chronologically diagnostic forms. To see Seven Sisters as dated to AD 50–75 is to situate it within the 30 years, or two generations of campaigning and guerrilla warfare between the Roman Army and the largely resistant tribes in Wales between AD 47–78 (Jackson 1951; Webster 1981; Davies 2000, 3–24; Manning 2002; 2004; Howell 2006, 65–74). This context arguably created the conditions for martial identities to flourish and for rapid technological innovations to occur. While it is dangerous to seek to match the dating of metalwork to specific historical events, such as the Boudican revolt or Caratacus’s last stand, nevertheless the broad and attested intensity of campaigning, marching camps and fort construction by the Roman army in Wales at this time (e.g. Nash-Williams 1969; Manning 1981, 40–4 and Fig. 9; 2002; Jones and Mattingly 1990, 64–140; Davies 2000, Figs. 1.1 A and B and 2.2C; Jarrett 2002, Figs. 3.5 and 3.6) is likely to have created contexts in which the use and burial of decorated metalwork was magnified. In other words, we may be observing a general ‘tide-mark’ of visible metalwork use and deposition at this time (Macgregor 1976, 178). Items of La Tène or native inspiration are certainly found in well dated early fort assemblages of pre-Flavian and early Flavian date here (Manning, Price and Webster 1995; Savory 1993; Gwilt 2007) and these can reasonably be seen as part of a wider general phenomenon of deposition. 9. Material, style and identity in first century AD metalwork 165 Seven Sisters and emerging La Tène stylistic trends of the first century AD (AD40–100) In this section, the appearance or heightened frequency of four stylistic traits, apparent in the decorated metalwork corpus for south Wales, are outlined. On current evidence, these are developments of the middle to later first century AD; that is the Campaigning period and for a generation or two afterward. These are related to the wider British corpus, the suggestion being that these were aspects of wider stylistic changes and trajectories, relating to a range of artefact classes and across Britain. The matter of whether these should properly be termed Stage VI art, or whether this term is discarded in favour of cultural-ethnic or purely descriptive names is later considered. However, chronological progression does seem apparent from the earlier ‘mirror-style’ (Stead’s Stage V) to these emerging and stylistically distinctive traits, albeit with overlap. This seems apparent, despite the current serious doubts over the chronology and chronological progression of Stage IV to Stage V art (e.g. Macdonald 2007). Borders, boundaries and new incised decorative techniques Two of the tankard handles in the Seven Sisters hoard (Figure 9.11a, 9.11b) illustrate the use of new decorative techniques: ring-punch work and incised cabling (Appendix 9.1: 04.138; 04.140). Similar ring-stamping is viewed on the bronze rim of the Brecon mirror (Figure 9.12), the Pentyrch large enamelled terret (Savory 1966a, fig. 1; Spratling 1972, Cat. 84) and the Chepstow strap-union (Nash-Williams 1932; Savory 1976, Cat 28, Pl. V; Jope 2000, Pl. 295f). Dot-punch work is also present on two terrets in the Polden Hills hoard (Brailsford 1975, 224, Pls. XVIIh and XIX f; Spratling 1972, Cats. 45–6) and one of the Hod Hill tankards (Corcoran 1952, 99, Cat. 13 and Pl. X.5). The use of ring stamping and dot punching is also prominent on the swords from Bardney, Lincolnshire and Congham in Norfolk, both regarded as having late Stage VI decoration (Stead 2006, 16 table 3 and Cats. 102–3). That very similar ring stamping occurs on the Roman disc phalera in the Seven Sisters hoard (Appendix 9.1; 04.143), and other early Roman military pieces, suggests the possible adoption of a Roman technique on Iron Age or native pieces. A possible corollary of this is that the Pentyrch terret on stylistic grounds, could now plausibly sit in the third quarter of the first century AD, rather than in the first half, as Savory asserted (1966a, 44). We might also view pointil work, as on the Pentyrch terret and the newly discovered bracelet pair from Boverton (Gwilt in press b), and on for example, horseequipment in the Melsonby and Polden Hills hoards, as another similar technique in the ascendancy. Incised line and cable decoration is used on the handle of the Brecon mirror (Figure 9.13a and 9.13b). This was buried with a cremation and associated with a Roman lamp (singly dated to between AD 50–100), two ‘mini-terrets’, a toilet set and a wheel-made carinated bowl and lid. This burial group was discovered during the 1990s by a metaldetectorist, beside the Roman road about a quarter of a mile from the Roman fort at Y Gaer, Brecon. It was subsequently acquired by Amgueddfa Cymru – National Museum Wales (Accession Number 97.9H). Nothing else is known about the precise burial context, though an early extra-mural cemetery is a possibility. The ceramic vessel is of Flavian form and has been used to date the burial from AD 70–100 (Sealey 2006, 16– 17). Given the early Flavian establishment of the nearby fort (Wheeler 1926; Jones and 166 Mary Davis and Adam Gwilt Figure 9.11a. (above left) Detail of punched decoration on tankard handle 04.140 in the Seven Sisters hoard, Neath-Port Talbot. Figure 9.11b. (above right) Detail of incised cabling on tankard handle 04.138 in the Seven Sisters hoard. Mattingly 1990, Map 4:32; Manning 2004, fig 5.2), a date of AD 75–90 for the burial seems plausible (a radiocarbon date is currently awaited from the cremated bone associated with this grave group). However, equally likely is that the mirror, a distinctive personal and high-status item, remained in circulation for a generation or two before burial. Therefore, it could feasibly have been made around AD 50–75. A very similar mirror handle is known from Ballybogey Bog, Ballymoney in County Antrim (Jope 1954; Raftery 1984, 208–10, fig. 104.1 and Pl. 66). Looking at these mirrors and the Seven Sisters tankard handles, there seems to be an increasing pre-occupation with the definition and elaboration of borders, zones and boundaries, and a new emphasis on symmetry. This is a departure from the freer earlier curvilinear tradition. A similar effect is also apparent on the handle of the Snowdon bowl (Savory 1976, Cat 32, fig. 78; Jope 2000, Pl. 295a–d). The central knobs and cabling on these mirror handles also recall the beaded torcs and collars of the later first century AD (e.g. Megaw 1971; Macgregor 1976, Cats. 199–208; Beswick et al. 1990; Jope 2000, Pls. 258–61), while the moulding, lips and ‘beaks’ are also commonly found on brooches, escutcheons and horse equipment, generally dated to the mid to later first century AD. Similar boundary definition is also apparent on the sword hilts of the late swords from Caerleon, Newport and Warton, Lancashire (Boon 1974; Stead 2006, Cats. 208 and 239). The legionary fortress burial association of the former and the use of brass in the latter (Dungworth 1996, 421) are consistent with their date at, or slightly after, the mid 1st century AD (Jope 2000, 128–9). Geometric design A synchronous development is the appearance of geometric designs on metalwork of native style and form. These are often inset cells to take polychrome enamels creating the jewelled effect discussed by Leeds (1933, 103–5). However they also appear as incised designs and borders. Prominent examples of the former are the bridle bits, strap unions and terrets in the Seven Sisters hoard, with their square, rectangular and petalled rosette enamelled designs (Figure 9.14a; Appendix 9.1: 04.125–8, 04.131). These are decorated with red, blue and white enamels of Roman technology. The Chepstow strap union complements this with its triangular red and yellow enamelled cells, forming zig-zags 9. Material, style and identity in first century AD metalwork 167 Figure 9.12. (above) Detail of punched decoration on the outer rim of the Brecon mirror, Powys. Figure 9.13a (below) The mirror within the Brecon grave group. Figure 9.13b (right) Handle of the Brecon mirror. confined within rectangular zones on each side plate (Savory 1976, Cat. 28, Pl. V; Taylor and Brailsford 1985, Cat. 26; Jope 2000, Pl. 295f). That this trend was not restricted here to horse equipment is now convincingly illustrated by the recent discovery of a decorated bronze collar with a burial at Boverton in the Vale of Glamorgan (Figures 9.14a, 9.14b and 9.14c). Along each collar side are sixteen rectangular cells. On the front, a narrow rectangular border hiding a slot terminal 168 Mary Davis and Adam Gwilt Figure 9.14a (above left) Detail of bridle bit 04.126 in the Seven Sisters hoard, showing the enamelled geometric and petal cells. Figure 9.14b (left) Detail of the Boverton, Vale of Glamorgan neck collar, showing the enamelled geometric cells. Figure 9.14c (above right) The neck collar and bracelet pair from the Boverton grave group. is decorated with small circular recesses, into which polychrome enamels were set. Four flat and highly polished plates were also riveted to the front and back surfaces of the piece. Overall, the impression gained is that the maker was experimenting with new technologies and surface decorative effects. The collar is a distinctive south Wales variant of the wider class of neck collars of western and northern Britain. These have been dated, largely on stylistic grounds, to between the mid 1st and early 2nd centuries AD (Megaw 1971, 153; Macgregor 1976, 99–101; Beswick et al. 1990, 29; Jope 2000, 292). Given the 1st century AD form of the bracelet pair associated with this burial (Crummy 1983, 37–8, Fig. 40; Stead 1986, 125 and Fig. 52.163–6; Johns 1996, 338 and Fig. 107.15) and the close stylistic parallels between this collar and the Seven Sisters pieces, a date of AD 50–75 is provisionally argued for this collar. (A radiocarbon date is currently awaited on the associated human bone in this disturbed burial, providing the opportunity to assess this assertion and wider dating assumptions for neck-collars in Britain). Similar polychrome enamelling technique is witnessed upon the decorated collar from Dorset (Megaw 1971, 147, Fig. 2; Jope 2000, Pl. 258b). It has a band of infilled triangles either side of the front terminals and enamel spots inset into the curvilinear design. While the polychrome enamels on the Trenoweth, Cornwall collar are inset spots, the tubular collar is edged with an incised repeated triangle pattern, the inward facing triangles infilled with stippling (Megaw 1967; 9. Material, style and identity in first century AD metalwork 169 Jope 2000, Pl. 259a). The use of brass in each case, rather than bronze, is consistent with a date broadly contemporary with the south Wales pieces. This geometric polychrome enamelling technique is a regional expression of a widespread trend in Britain. For example, it is paralleled on the late northern British swords from Asby Scar and Embleton in Cumbria and Thorpe Hall, East Yorkshire (Stead 2006, Cats. 203, 205 and 214; Rigby 2006, 118–21). The Asby Scar sword has enamelled triangles defining bronze rectangles on its hilt end, also with extensive use of incised decoration and it is regarded as decorated with Stage VI motifs (Stead 2006, 17, Table 4). The bordered enamelled cells on the Embleton sword hilt and scabbard, combined with the use of cable bordering decoration and an incised chequer-board pattern on the scabbard closely echoes the trends evident in south Wales. The enamelled triangles on the slightly later scabbard chapes from Housesteads, South Shields and Chesters are also relevant (Macgregor 1976, Cats. 166, 168 and 170). This style is also found on a range of horse pieces – terrets, strap unions, bridle bits and toggles – from across Britain, but especially northern Britain and East Anglia (e.g. Spratling 1972; Macgregor 1976; Hutcheson 2004). Some prominent associations and examples have already been cited (see above), but a number of further parallels can be found (e.g. Spratling 1972, Cats. 77, 80, 180, 232 and 237; Macgregor 1976, Cat. 10; Taylor and Bradford 1985, Cat. 25). In northern Britain and East Anglia, there are also belt fittings, mounts, fasteners and massive armlets, which are ascribed to the later first and second centuries AD. It has also long been apparent that many decorated brooches of the later 1st and 2nd centuries AD across Britain have similar jewelled enamel designs (Leeds 1933, 105; Bayley and Butcher 2004, 159–68). Perhaps less commented upon to date, is the allied increase in the frequency of repeated triangle recesses either infilled with red glass of Iron Age tradition or as incised borders, sometimes with in-filled line-work decoration. Though perhaps less formally geometric, this repetition of the same motif or element seems to be another manifestation of stylistic change. There are repeated red glass filled triangles on a number of terrets, a toggle and the quadrilobed strap union in the Polden Hills hoard (Brailsford 1975, Pls. XVIIa and g, XVIIIc, XXIc and XXIIIa). Similar technique is witnessed on a flat ringed terret from Swanton Morley, which also has a punched dot border (Palk 1988, Cat. 314; Hutcheson 2004, Cat. 137). Careful observation of the pendant hooks in the Seven Sisters hoard has revealed a zig-zag border surviving on one, effectively defining a border of interlinked triangles around the crescent ends (Appendix 9.1: 04.137). Identical borders are found around the enamelled zones on the side plates of the Chepstow strap union (Taylor and Brailsford 1985, Cat. 26; Jope 2000, Pl. 295f), while the same is viewed on the rectangular shaped tankard handle from Greenhill, Dorset (Corcoran 1952, 99 and Pl. X.6). This latter artefact has two zones of interlinked and incised triangles, carefully confined within long incised vertical rectangles themselves placed laterally either side of a mid-line groove and the handle has been tin plated (Spratling 1972, Cat. 366). One of the Seven Sisters tankard handles and the Newstead handle also have zig-zag scorper and rocked tracing lines on them (Appendix 9.1: 04.142; Macgregor 1976, Cat. 290). While not formally geometric, general parallel can be seen with the Greenhill tankard handle. This trend is also witnessed on three of the enamelled terrets in the Westhall, Suffolk hoard, where a series of incised triangles is located between the ridge on the inner edge 170 Mary Davis and Adam Gwilt of the ring and the ornamental crescent (Clarke 1940; Spratling 1972, Cats. 72–4). Alternate triangles are filled with punched dots, another of the newly popular decorative techniques (see above). This hoard has generally been dated to the third quarter of the 1st century AD (Clarke 1940, 68–9; Spratling 1972, 308–9; Macgregor 1976, 43). Further examples from Norfolk with this technique are known (Hutcheson 2004, Cats. 131 and 133; Palk 1988, Cat. 189). The same effect is also present on the side plate and boss of one of the shields in the Tal-y-Llyn, Gwynedd hoard (Savory 1964; 1966b; Jope 2000, 71–4; Pls. 96–8). The better preserved of the pelta shaped side plates has a triskele motif, surrounded by a repeated triangular border. The outer row of triangles is infilled with linear hatchwork, complementing the ‘mirror-style’ basket-work hatching of the negative spaces around the central triskele motif. A chequer board pattern of hatched rectangles on the outer edge of this shield side plate also closely recalls the decoration on the late Embleton scabbard (Macgregor 1976, Cat. 145; Stead 2006, Cat. 205, Fig. 101). In addition, one of the shield bosses has a central triskele motif, which is bordered by an outward facing circle of triangles. Rocked tracer work on the second shield boss and the trapezoid face plaques, and a trellis pattern border on one of the openwork discs (Savory 1964, Fig. 7), highlights the contemporary popularity of incised techniques, as discussed above. Taken in conjunction with the use of brass, the tinned plates and the Roman lock plate in this hoard, this adds confirmation to the now established view that the hoard belongs to the 1st century AD. If contemporary with the wider stylistic trend, a date of manufacture and burial between AD 50–75 can be suggested. To take these parallels one step further, the decorated panel above the handle on the Carrickfergus tankard, Co. Antrim is another late and unusual piece, whose incised infilling derives from the ‘mirror-style’ ornament (Raftery 1984, 223–5; Jope 2000, 228– 9; O’Neill 2002). It has an identical repeated triangle border, the exterior inward facing triangles infilled with linear hatchwork. This tankard has been seen as first century AD in date and now a date around AD 50–75 can be suggested. This is in accord with the stylistic parallels cited by Raftery, the Greenhill tankard handle and the Snowdon Bowl, who saw its possible manufacture in north Wales as a late example of the ‘mirror-style’ (Raftery 1984, 224–5 and Fig. 112). Together, both Carrickfergus and Tal-y-Llyn may plausibly be viewed as transitional pieces between the ‘mirror-style’ and the geometric style that followed it. Fragmented motifs and the breakdown of curvilinear flow The breakdown of curvilinear flow epitomised upon the metalwork of the true ‘mirrorstyle’ (Stead’s Stage V) seems to be a further aspect of stylistic change during the first century AD. Motifs once connected and integral to wider designs now appear increasingly fragmented as isolated islands surrounded by metal margins. In south Wales, the trend is best illustrated by the sequential changes observed on the large strap union from Alltwen, Neath-Port Talbot (Figure 9.15a), the Pentyrch terret, Cardiff (Figure 9.15b) and a recently reported large strap-union from Maendy hillfort, near Treorchy, Rhondda Cynon Taf (Gwilt 2007, 310; Figure 9.15c). On the Alltwen piece, individual motifs are unified within two large trumpet and circle designs, arranged as a reversed opposed pair. While the feeling of flow is maintained on the Pentyrch terret, motifs are now increasingly physically separated both on the flange and the bosses. The end stage of this process is 9. Material, style and identity in first century AD metalwork 171 viewed on the Maendy piece. Now, familiar motifs are simplified into triangles, halfcircles and circles. The overall design is separated into five enamelled zones, and while there are elements of reflection and symmetry, all aspect of flow and interlinkedness is gone. The effect is heightened by the use of yellow glass insets within bronze borders, themselves captured within red glass zones. This piece is distinctively ‘chunky’ in manufacture, hinting that there may also have been a later trend towards the manufacture of heavy pieces, also echoed on the Snowdon Bowl, the Boverton collar (Figure 9.14c), the Brecon mirror handle (Figure 9.13b) and certain heavy brooch forms from north Wales (Chapman pers. comm.). One could posit an overlapping chronological sequence of AD 1–70 for Alltwen, AD 50–80 for Pentyrch and AD 60–120 for Maendy, though equally, there may have been little to separate these pieces between AD 50–75. Nevertheless, the direction of travel seems plausible. While the Alltwen strap union is a single find and therefore its date relies upon parallels (Spratling 1972, 116–8; Taylor and Brailsford 1985, Cats. 43–54; Palk 1988, Cats. 442–5; Feachem 1991) and stylistic considerations, the Maendy piece was associated with a large bronze horse bell. These are repeatedly found on Roman military sites in Britain from AD 43 onwards (e.g. Wheeler 1926, 116–7, fig 58.16; Cunliffe 1971, 112–3, Cat. 107; Webster 1995a, 55–8; Lloyd-Morgan 1997, 269–71); none can confidently be attributed as pre-Invasion. On balance, the use of yellow glass Figure 9.15a. (above left) The Alltwen strap union, Neath-Port Talbot. Figure 9.15b. (left) The Pentyrch terret, Lesser Garth hoard, Cardiff. Figure 9.15c. (above right) The strap-union from Maendy hillfort, Rhondda Cynon Taf 172 Mary Davis and Adam Gwilt on this piece also suggests a date of manufacture no earlier than AD 50. The artefacts associated with the Pentyrch terret are not chronologically diagnostic (Savory 1966a, though see previous discussion). In the light of this, the two pendant hooks with red enamel decoration in the Seven Sisters hoard (Appendix 9.1: 04.136–7), suggest a similar trend, with captured circles, isolated biconcave sided triangles and the absence of flow, although it must be admitted that here, the crescent spaces only provided for limited options. Further illustration of this trend is provided by the enamelled and decorated vessel escutcheon found in Carmarthen, within early occupation deposits belonging to Moridunum, the civitas capital of the Demetae (Webster 2003, 317, Cat. 56 and fig. 8.4). Circles and biconcave sided triangles infilled with Roman enamel ‘float’ across the lateral plates of this piece, in an isolated and disjointed manner. A central rectangular panel of four oblong shaped cells, also with enamel inlays also illustrates the contemporary trend towards geometric design (see above). Found in a mid to late 2nd AD context, it was repaired and probably remained in circulation for at least 50 years before burial. Some confirmation of this trend, captured at mid-way stage, is provided through reference to the enamelled motifs upon quadrilobed strap unions (Feachem 1991, fig. 2; Jope 2000, Pl. 296–7; Hutcheson 2004, Cat. 67). These large plates provide ample space for decoration, but the overall designs, though showing symmetry and complexity, do not flow in curvilinear style. Enamelling tends to be confined to tightly spaced but discrete zones (see for example the Polden Hills, Somerset and Norton, Suffolk mounts), and isolated captured circles abound. The use of blue and yellow, in addition to red glass on many pieces is also chronologically significant, as is the range of incised zones with dot, punched and pecked decoration. Located across south western and southern England, East Anglia, and now south Wales, the associations in the Polden Hills, Westhall and Santon hoards suggest they were made and used between AD 50–75 (Clarke 1940; Spratling 1972, 304–12; Brailsford 1975, 232–4; Hutcheson 2004, 28). The zig-zag border on the example from Ober-Olm, Germany also captures a chronological convergence of stylistic developments, exhibiting both the geometric trend and an incised triangle border (Jope 2000, Pl. 297f). Many decorated terrets also exhibit a similar breakdown of decorative motifs. In the Polden Hills hoard, many of the winged terrets are enamelled and have infilled incised motifs, which are dispersed and isolated (Brailsford 1975, Pls. XVIIb,c,g and XVIIIa,b,d; Spratling 1972, Cats. 57–8, 62–3). Further examples of this trend can be found (e.g. Spratling 1972, Cats. 44, 48–50, 52; Hutcheson 2004, Cat. 111). A number of flat ringed terrets have many captured circles and a ‘slackness’ of definition of the curvilinear decoration, which suggests a breaking down process on the later pieces (e.g. Spratling 1972, Cats. 65, 69–70, 78–9; Macgregor 1976, Cat. 62; Hutcheson 2004, Cat. 93). The incised and enamelled decoration on some linchpins also shows similar trends (Spratling 1972, Cats. 116–8; Hutcheson 2004, Cats. 47, 51 and 55). At present, it is difficult to identify with confidence the breakdown of curvilinear flow and isolation of motifs on swords and scabbards, although candidates might include decoration on the Bardney and Congham scabbards, the suspension loops from Icklingham and Towcester, the base of the scabbard from Mortonhall, the openwork on the Asby Scar scabbard, the scabbard from Flag Fen and the crown-hilt guards from Hod Hill and Waddon Hill (Stead 2006, Cats. 102–3, 107–8, 203, 206, 222, 234, 248, 9. Material, style and identity in first century AD metalwork 173 251). In the mirror series, the Nijmegen example shows a degree of isolation of motifs, which may be a chronological development rather than purely a cultural aberration (Jope 2000, Pl. 254–5), while in the collar series, the curvilinear motifs on the Trenoweth collar appears to be similarly breaking down (Jope 2000, Pl. 259). Discrete petal motifs – single and composite The fourth and final aspect of stylistic change observed is the increasing frequency of the use of discrete petal motifs. These are either repeated singly or incorporated into clusters of two, three and four. This may be no more than a particular aspect of two trends already discussed: the increasingly geometric design and the fragmentation of motifs and breakdown of curvilinear flow on La Tène decorated metalwork. Petal shaped elements were known and used on earlier metalwork, for example, often incorporated as lobes within larger trumpet motifs on Stage V or ‘mirror-style’ art (e.g. Fox 1946, 48, fig. 24; 1958, 147–8, figs. 82–3). However here, they are usually interconnected in flowing design, rather than separated and reconfigured into isolated or geometric design. Four petal geometric ‘stars’ are also present on a number of pieces with an apparent earlier ancestry (e.g. Piggott 1950, 22–3; Spratling 1972, Cat. 298; Stead 2006, Cats. 33, 87 and 89; Jope 2000 Pl. 207f and g). Nevertheless, there does seem to be an additional and distinctive late resurgence, worthy of specific comment. Three and four-petal rosettes are incorporated in the enamelled designs upon the bridle bit, terrets and strap unions in the Seven Sisters hoard (Appendix 9.1: 04.125–8; 04.130; Figures 9.14a; 9.16a). The large strap-hook from Abergavenny (Figure 9.16b) also incorporates four-petal rosettes within its enamelled design (Savory 1993, 221–4, Illus. 14.43). Similar rosettes are seen on the terrets in the Saham Toney, Norfolk hoard (Macgregor 1976, 26, fig. 1; Spratling 1972, Cats. 91–3), while the strap-unions in both Saham Toney and Middlebie, Dumfriesshire hoards incorporate petals in both motif element and larger design forms (Macgregor 1976, 26, fig. 1.2 and Cat. 22; Jope 2000, Pl. 278b). Petals are also recognisable on three of the terrets in the Polden Hills hoard (Brailsford 1975, Pl. XVII b and g, Pl. XVIIIa; Spratling 1972, Cats. 52, 61 and 63). Figure 9.16a. (above) Illustration of terret 04.128 and strap union 04.131 in the Seven Sisters hoard, showing enamelled petal motifs. Figure 9.16b. (right) Illustration of the strap hook from the Roman fort at Abergavenny, Monmouthshire, with petal motifs. The drawing was reproduced courtesy of the Royal Archaeological Institute from The Archaeological Journal 150, page 212. Drawn by Howard Mason. 174 Mary Davis and Adam Gwilt The use of twin petal motifs is also seen on a sword belt ring from High Nash, Coleford, Gloucestershire (Webster 1990; Stead 2006, Cat. 128, fig. 82; Rigby 2006, 121, fig. 34). The petals, in reserved bronze have surrounding recesses inlaid with red enamel. A recently discovered tankard handle from Castell Henllys hillfort, Pembrokeshire has a four petal ‘star’ in repoussé, on its single surviving attachment plate. A late first or early second century AD date seems likely, on stylistic grounds. In north Wales, two face plaques in the Tal-y-Llyn, Gwynedd hoard have petals in repoussé (Savory 1964; 1966b; Jope 2000, Pl. 98a–c) and this hoard is now suggested as dating to the third quarter of the first century AD. To this may be added the isolated petal on the handle attachment of the late mirror from Llechwedd-du Bach, Gwynedd (Fox 1925, fig. 1; Jope 2000, Pl. 256a and b). Probably once associated with a burial, it was found with a lathe-turned, tinned bronze platter of early Roman form, which on stylistic grounds dates to the later first century AD (Fox 1925, 257; Boon 1980, 744; Jope 2000, 292, Cat. 256a–b). The mirror, with an undecorated plate and plain handle with three hoops, was therefore probably manufactured between AD 50–75. To expand these observations both geographically and in terms of artefact classes, petals and four petal ‘stars’ are found on a number of scabbards (Macgregor 1976, Cat. 143; Stead 2006, Cats 101, 204, 245 and 246). The enamelled hilt end from Hod Hill also has repeated petals, infilled with blue enamel within a red enamelled surrounding band (Rigby 2006, 120–1, fig. 34). These are all late, with first century AD currency (e.g. Stead 2006; Macgregor 1976, 80–5) and it is worth comment that Stead sees a transition from Stage V to Stage VI decoration on his Group D swords and scabbards (Stead 2006, 16, Table 3 and 50). These include the High Nash, Coleford and Isleham swords, scabbards and belt ring. Petal rosettes and stars are also seen on strap unions (e.g. Spratling 1972, Cats. 185, 195, 197 and 214), toggles (Spratling 1972, Cat. 239; Palk 1988, Cat. 353; Foster 1999, Cat. 2, fig. 52) and terrets (e.g. Bushe-Fox 1926, Pl. XIII.16; Palk 1988, Cat. 178). They are particularly prominent on the Rise, East Yorkshire bridle bit (Macgregor 1976, Cat. 10; Jope 2000, Pl. 276g), while petal strips and pairs are associated with geometric enamel cells on the Birrenswark, Dumfriesshire bridle bit (Macgregor 1976, Cat. 2; Jope 2000, Pl. 279a and b). Petals are also visible on mounts and strap junctions from northern Britain (Macgregor 1976, Cats. 15, 29, 30, 36 and 38). Bossed petals are abundant decorative motifs on the massive armlets of northern Scotland (Macgregor 1976, Cats. 231–50). These are typically dated to the late first and early second centuries AD. Some also have geometric enamelled cells within roundels, including the Pitkelloney, Perthshire example with its four petal rosettes (Macgregor 1976, Cat. 243). Petals, isolated, repeated in strips or incorporated into flowing design are also common attributes on the heavy bronze collars of western Britain, including the Llandysul, Ceredigion collar (Megaw 1971; Beswick et al. 1991; Jope 2000, Pls. 258 and 261d). These date from the mid 1st to early 2nd centuries AD. Finally and importantly, in the Santon (Norfolk) hoard, a pan lid with bird mounts is associated with six petal incised rosettes. In addition a bucket handle from this hoard is ornately decorated with an incised and repeated petal strip, the inner borders of which are infilled with pointillé ornament (Spratling 1972, Cat. 425; Jope 2000, Pl. 166k). This hoard and its associated horse pieces have been dated to the second half of the 1st century AD (Spratling 1972, 304–8; Hutcheson 2004, 28). 9. Material, style and identity in first century AD metalwork 175 Discussion In assessing the chronological attributions suggested, some may accuse the author of chronological reductionism, succumbing to the tyranny of late Roman associations. Perhaps. However this ‘Celtic’ Art, which is visible to us, seems increasingly to be a first century AD art. The combination of associated evidences in Wales: the Roman horse bells, the Roman cavalry fittings, the Roman lock-plate, the Roman lamp, the Roman fort and fortress burial contexts and the use of brass, is difficult to downplay. They indicate a range of stylistic changes and developments were happening here, at least between AD 50–75. While immediate pre-invasion beginnings for these developments remain a possibility, yet this part of western Britain saw few of the Gallicising and Romanising influences occurring within southern and eastern England between 200 BC–AD 43. Moreover, taking a very broad overview, it is remarkable how few early La Tène decorated artefacts have been found across Wales to date, with the prominent exceptions of the Cerrigydrudion Crown and elements within the Llyn Cerrig Bach lake assemblage. Here, the small number of decorated artefacts with probable currencies spanning the 2nd or 1st centuries BC or earlier are increasingly being swamped by significant recent discoveries of probable 1st century AD date. This contrasts with southern and eastern England, which does have established corpora of earlier decorated weapons, mirrors, horse equipment and torcs, which probably belong to the 2nd and 1st centuries BC. This is not to deny the abundance here, of decorated metalwork also dating to the 1st century AD. The selective native uptake of Roman technologies, such as brass and enamels seems, on current evidence, most plausible during the Campaigning period of the Roman Army, rather than much earlier. Admittedly, it remains a possibility that Late Iron Age communities in south Wales were accessing Roman materials and technologies indirectly, and even a little before the invasion of the Roman Army, by means of long-distance communications and exchanges. Here, enduring maritime contacts with the south west of England and beyond, seem most plausible. That these native stylistic trends also continued later, between AD 75–125, is also illustrated by the late manufacture of native bridle bits, weapons and mounts at Prestatyn, Flintshire (Blockley and Day 1989). Nevertheless, the overall weight of the evidence still suggests vigour of expression and development during the third quarter of the 1st century AD. To return to the contentious issue of naming these artistic developments, is this 1st century AD art, best considered as British Art? In short and on current evidence, the answer is no, if the label is to be given cultural-ethnic connotations, in addition to a merely loose geographical definition. The concept of Britannia and British identity was arguably more in the minds of the invader than the invaded, at this early and unresolved juncture in the conquest of Britain. Wide ranging stylistic changes, communicated and effected between Iron Age societies, do not require or imply the cultural-political unification of Britain, as the term British conjures. An overview of the kingdoms and tribes of Britain on the eve of the Roman invasion, far better indicates a continuing and complex mosaic of regional and sometimes super-regional identities, yet still far from being a nascent pan-British identity. The most plausible and powerful unifying strand, at this particular juncture, was a convergence of opinion amongst many Iron Age societies towards defining themselves as ‘not Roman’ in their metalwork styles. This fortuitous shared feeling amongst native warrior elites, either against a common enemy, or at least 176 Mary Davis and Adam Gwilt signalling difference, is not to deny that tribal or dynastic identities could have been expressed in other areas of social and political life. If not British, and not wishing to re-ignite the Celtic Art debate, what are the possible alternatives? Stage VI La Tène art, as a term, is only strictly valid for as long as art Stages I–V work in chronological sequence, and this now seems increasingly questionable. However, it clearly had a La Tène ancestry, while these late stylistic developments of the 1st century AD do seem chronologically to succeed the ‘mirror-style’, yet with some overlapping apparent. Neutral descriptive terms while less interpretive, can however be regarded as ‘opt outs,’ failing to capture the social and historical significance of the material in question. Until a more apt or accurate name is forthcoming, Native Campaigning Art is here offered. This term gives rise to a chronologically specific ‘horizon’, bridging Iron Age to Romano-British worlds, dating perhaps between AD 40–120 in Wales and England, but ending later in Scotland. It also serves to focus attention on the phenomenon of the increased deposition of decorated metalwork at this time. Horse and chariot related display, feasting and personal ornamentation, it is argued, was an intensive expression inextricably linked with native warrior elites and their clashes with and responses to the campaigning Roman army. The name therefore aims to encapsulate something of the spirit, impulses and motivations behind the generation of these new styles. This is not to deny that it also came to take on a range of additional meanings and values, for example, how it was also perceived and used by Roman soldiers; secondly, how meanings changed for the first and second generation of native people after campaigning had ceased and living within the cultural milieu of responding to prolonged military and frontier occupation. Conclusion (Adam Gwilt and Mary Davis) The Seven Sisters hoard has been argued as pivotal in illustrating the range of stylistic changes that were occurring in Britain at this time. As such, it provides a benchmark and chronological corroboration for the dating of many other single artefacts and less securely dated associations across Wales and Britain. While other regions may have experienced slightly different trajectories of change and lengths of stylistic currency, the evidence from Wales and this western frontier shows changes were underway early, certainly by comparison with the traditional dating of the northern British material. It is therefore an important bridge, informing the early development and currency of these styles. This first century AD art appears to have been super-regional and not the confined expressions of regional tribal identities (though some regional stylistic variations are also tentatively emerging). In other words, this paper has attempted to demonstrate that these south Wales trends are also witnessed in south west England, East Anglia and parts of northern Britain in particular, but also less prominently across southern England. That individual artefacts and associated groups are frequently and repeatedly seen to exhibit two or more of the new trends posited, provides additional chronological consistency and validation. Perhaps those tribes and frontier regions, which offered prolonged resistance to the Roman army or experienced prolonged military occupation, were for longer able, or more motivated to modify and develop their metalwork and art styles to suite native 9. Material, style and identity in first century AD metalwork 177 tastes. In southern England, these developments had begun, but seem to have been snuffed out at an early stage by rapid advances of the Roman army. At the same time, this art and decoration seems to be interlinked with the consolidation and expression of the power of native political and warrior elites, through the continued preoccupation with horse and chariot display, with drinking and personal ornamentation. The extent to which elements encountered in these decorative trends: boundary definition, the breakdown of free flow, isolation of elements and increasingly controlled symmetry and geometry, may have also echoed some of the wider social concerns at this time of cultural conflict and change, is an intriguing proposition. Agency and chronological context are key to this interpretation. ‘Who’ made and ‘who’ employed this art, and ‘when’, seem to be the crux. Here, further analytical studies will help to clarify broader diagnostic patterns of cultural selectivity in the use of materials and techniques, while new radiocarbon dates for decorated objects will begin to enable some independent evaluation of our problematic typological schemes. However, on current evidence, it is argued that between AD 50–75, many native societies in Wales, East Anglia, south west England and northern Britain, retained the autonomy and wherewithal to make and wear their own distinctive styles of metalwork. In this art, it is argued that one can still see the agency of native or Iron Age workers exerting control and preference for native and Iron Age people, rather than dilution or decline, through the controlling hand of Roman patronage for novel products from native metalworkers. However, tipping points may have been later reached, when Iron Age styles became incorporated into the inclusive Romano-British repertoire. The evidence of native metalworking at Prestatyn in Flintshire, north east Wales from AD 100–120, yet associated with rectangular timber buildings and succeeded by a Roman bath-house, is tantalising with respect to the identity of the maker and the destined recipients (Blockley 1989, 23–46). In conclusion, we have attempted to show how these emerging observations and trends help us to grapple afresh with broad issues of identity, cultural resistance and assimilations with Rome, in Britain during the 1st century AD. Acknowledgements Many people have helped with discussions and ideas, and by making material and data available; for this we wish to thank Richard Brewer, Evan Chapman, Simon James, Jody Joy, J. D. Hill, Fraser Hunter, Mark Lodwick and Mansell Spratling. Duncan Garrow encouraged us to contribute the paper, and has pursued, with determination, the much needed radiocarbon dates for artefacts under discussion here, as part of the ‘Technologies of Enchantment’ project. Mary Davis would like to extend particular thanks to Peter Northover for allowing us to use his data, and Ian Freestone and Niall Sharples for detailed discussions and comments. Appendix 1: Table of objects from the Seven Sisters Hoard, their style, metal alloy and applied decoration 178 Acc. No. Mary Davis and Adam Gwilt D&S Object No. Object type ‘Style' Alloy Inlay NMW 04.125 16A bridle-bit ring horse equipment geometric brass enamel NMW 04.126 16B bridle-bit ring horse equipment geometric brass enamel colour Other decoration Red, white or blue Red, white or blue Red, yellow Red, blue, yellow NMW 04.127 19 Terret horse equipment geometric brass enamel NMW 04.128 20 Terret horse equipment geometric brass enamel NMW 04.129 9 Terret horse equipment Roman bronze NMW 04.130 1 strap union horse equipment Roman brass (+ tin) niello NMW 04.131 17 strap union horse equipment geometric brass enamel NMW 04.132 10 buckle horse equipment Roman bronze NMW 04.133 5 strap slide horse equipment Roman leaded bronze NMW 04.134 3 pendant horse equipment Roman leaded brass NMW 04.135 4 strap union brass curvilinear bronze curvilinear bronze engraved NMW 04.138 21 horse equipment personal military ornament personal military pendant hook ornament tankard handle Vessel Roman pendant hook curvilinear bronze engraved NMW 04.139 22 tankard handle Vessel curvilinear bronze NMW 04.140 23 tankard handle Vessel curvilinear bronze NMW 04.136 14 NMW 04.137 15 NMW 04.141 24 tankard handle Vessel curvilinear bronze NMW 04.142 25 tankard handle Vessel curvilinear bronze NMW 04.143 2 disc (phalera) horse equipment Roman bronze Roman bronze curvilinear bronze niello, tinning heated glass heated glass heated glass engraved red red red engraved punched, engraved engraved engraved punched, engraved NMW 04.144 13 ring NMW 04.145 12 helmet crest NMW 04.146 8 bell horse equipment Roman NMW 04.147 7 horse equipment Roman horse equipment Roman brass (+ tin) punched NMW 04.149 11 bell fragment (hubb?) weight bronze (+ lead) bronze metal working curvilinear bronze engraved NMW 04.15 26 ingot metal working geometric brass NMW 04.151 27 ingot metal working geometric NMW 04.152 29 casting jet metal working Roman NMW 04.153 30 NMW 04.154 32 NMW 04.155 31 NMW 04.156a 33 NMW 04.156b 34 NMW 04.156c 35 NMW 04.157 28 BM 1928 1-16 18 casting jet tool ingot folded sheet folded sheet folded sheet lump strap union metal working metal working metal working Vessel Vessel Vessel metal working horse equipment Roman curvilinear Roman curvilinear curvilinear curvilinear ? geometric brass leaded bronze (+ zinc) leaded bronze bronze brass bronze bronze bronze copper brass enamel NMW 04.148 6 horse equipment personal military ornament Red, blue (D&S No. refers to Davies and Spratling’s catalogue numbers (1976)) ? missing (? red) Red, blue punched, engraved 9. Material, style and identity in first century AD metalwork 179 Bibliography Alcock, L. 1963. Celtic Archaeology and Art, In E. Davies (ed.), Celtic Studies in Wales; A Survey, 3–46. Cardiff: University of Wales Press. Arnold, C. J. and Davies, J. L. 2000. Roman and Early Medieval Wales, Stroud: Sutton Publishing Ltd. Bateson, J. D. 1981. Enamel-working in Iron Age, Roman and Sub-Roman Britain. The Products and techniques. BAR British Series 93. Bateson, J. D. and Hedges, R. E. M. 1975. The scientific analysis of a group of Roman-Age enamelled brooches, Archaeometry 17, 177–190. Bayley, J and Butcher, S. 1989. Romano-British plate brooches: composition and decoration, Jewellery Studies 3, 25–32. Bayley, J. 1990. The Production of Brass in Antiquity with Particular reference to Roman Britain, In P. T. Craddock (ed.), 2000 Years of Zinc and Brass, 7–27. London: British Museum Occasional Paper No. 50. Bayley, J. and Butcher, S. 2004. Roman Brooches in Britain; A Technological and Typological Study based on the Richborough Collection, London: The Society of Antiquaries of London. Beswick, P., Megaw, R, Megaw, J. V. S. and Northover, P. 1990. A Decorated Late Iron Age Torc from Dinnington, South Yorkshire, Antiquaries Journal 70, 16–33. Bevins, R. E. 1994. A Mineralogy of Wales. Cardiff: National Museum of Wales. Bishop, M. C. 1988. Cavalry Equipment of the Roman Army in the First Century A.D., In J. C. Coulston (ed.), Military Equipment and the Identity of Roman Soldiers; Proceedings of the Fourth Roman Military Equipment Conference, 67–195. Oxford: British Archaeological Reports International Series 394. Bishop, M. C. and Coulston, J. C. N. 2006. Roman Military Equipment; From the Punic Wars to the Fall of Rome, Oxford: Oxbow Books (2nd edition). Blockley, K. 1989. Prestatyn 1984–5; An Iron Age Farmstead and Romano-British Settlement in North Wales, Oxford: British Archaeological Reports British Series 210. Blockley, K. 1993. Excavations on the Roman Fort at Abergavenny, Orchard Site, 1972–73, Archaeological Journal 150, 169–242. Blockley, K. and Day, M. 1989. Moulds, In K. Blockley, 1989, 183–92. Boon, G. C. 1974. A ‘Worton’-Type Bronze Sword-Hilt at Caerleon, Proceedings of the Prehistoric Society 40, 205– 6 and Pl. XX. Boon, G. C. 1980. A Neglected Late-Celtic Mirror-handle from Llanwnda near Fishguard, Bulletin of the Board of Celtic Studies 28(IV), 743–4. Brailsford, J. W. 1953. Later Prehistoric Antiquities of The British Isles, London: British Museum. Brailsford, J. W. 1962. Hod Hill; Volume I: Antiquities from Hod Hill in the Durden Collection, London: British Museum. Brailsford, J. W. 1975. The Polden Hill Hoard, Somerset, Proceedings of the Prehistoric Society 41, 222–34. Brewer, R. J. 2002. The Second Augustan Legion and the Roman Military Machine, Cardiff: National Museums and Galleries of Wales. British Geological Survey 1998. Minerals in Britain. Lead and Zinc www.mineralsuk.com/britmin/leadzinc.pdf. Bushe-Fox, J. P. 1926. First Report on the Excavation of the Roman Fort at Richborough, Kent, London: Reports of the Research Committee of the Society of Antiquaries of London, No. VI. Chapman, E. M. 2005. A Catalogue of Roman Military Equipment in the National Museum of Wales, Oxford: British Archaeological Reports British Series 388. Clarke, R. R. 1940. The Iron Age in Norfolk and Suffolk, Archaeological Journal XCVI, 1–113. Corcoran, J. X. W. P. 1952. Tankards and Tankard Handles of the British Early Iron Age, Proceedings of the Prehistoric Society 18, 85–102. Cowell M. R. 1990. Scientific report, in R. Jackson, Camerton: the late Iron Age and early Roman metalwork, 69–80. London: British Museum Publications. Crummy, N. 1983. Copper Alloy Armlets, In N. Crummy, Colchester Archaeological Report 2: The Roman small finds from excavations in Colchester 1971–9, 37–45. Colchester: Colchester Archaeological Trust Ltd. 180 Mary Davis and Adam Gwilt Cunliffe, B. 1971. Other objects of bronze and silver, In B. Cunliffe, Excavations at Fishbourne 1961–1969; Volume II: The Finds, 107–26. London: Reports of the Research Committee of the Society of Antiquaries of London No. 27. Curle, J. 1911. A Roman Frontier Post and its People; The Fort of Newstead in the Parish of Melrose, Glasgow: J. Maclehose and Sons. Davies, J. L. 2000. Roman Wales: An Introduction, In C. J. Arnold and J. L. Davies, 2000, 1–140. Davies, J. L. and Spratling, M. G. 1976. The Seven Sisters hoard: a centenary study, In G. C. Boon and J. M. Lewis (eds.), Welsh Antiquity; Essays mainly on Prehistoric Topics Presented to H.N. Savory upon his Retirement as Keeper of Archaeology, 121–47. Cardiff: National Museum of Wales. DCMS, in press. Treasure and Portable Antiquities Scheme Report 2005–06, London: Department for Culture, Media and Sport. Dudley, D. R. and Webster, G. 1965. The Roman Conquest of Britain, London: Batsford. Dungworth, D. 1995. Iron Age and Roman Copper alloys from Northern Britain, Unpublished PhD thesis, University of Durham. Dungworth, D .B. 1996. The Production of Copper Alloys in Iron Age Britain, Proceedings of the Prehistoric Society 62, 399–421. Dungworth, D. 1997. Iron Age and Roman Copper Alloys from Northern Britain, Internet Archaeology 2. http:/ /intarch.ac.uk/journal/issue2/dungworth_index.html Dungworth, D. 1997. Copper metallurgy in Iron Age Britain, In A. Gwilt and C. Haselgrove (eds.), 1997, 46–50. Feachem, R. 1991. Note: Two Quadrilobed Harness-Mounts from Hambleden, Buckinghamshire, Antiquaries Journal 71, 216–20. Fitts, R. L., Haselgrove, C. C., Lowther, P. C. and Willis, S. H. 1999. Melsonby Revisited: Survey and Excavation 1992–95 at the Site of Discovery of the ‘Stanwick’, North Yorkshire, Hoard of 1843, Durham Archaeological Journal 14–15, 1–52. Fitzpatrick, A. P. 2007. Dancing with Dragons: fantastic animals in the earlier Celtic art of Iron Age Britain, C. Haselgrove and T. Moore (eds.), 2007, 339–57. Forster, R. H. and Knowles, W. H. 1911. VII – Corstopitum: Report on the Excavations in 1910, Archaeologia Aeliana, 7(third series), 143–267. Foster, J. 1999a. The Metal Finds, In R. Niblett, 1999, 133–71. Foster, J. 1999b. The Funerary Finds – General Discussion, In R. Niblett, 1999, 175–6. Foster, J. 2002. Life and Death in the Iron Age. Oxford: Ashmolean Museum. Fox, C. 1925. A Late Celtic Bronze Mirror from Wales, Antiquaries Journal 5, 254–7. Fox, C. 1946. A Find of the Early Iron Age from Llyn Cerrig Bach, Anglesey, Cardiff: National Museum of Wales. Fox, C. 1958. Pattern and Purpose; A Survey of Early Celtic Art in Britain, Cardiff: National Museum of Wales. Gillam, J. P. 1958. Roman and Native, A.D. 122–97, In I. A. Richmond (ed), Roman and Native in Northern Britain, 60–90. Edinburgh: Nelson and Sons Ltd. Gwilt, A. 2007. Silent Silures?: Locating People and Places in the south Wales Iron Age, In C. Haselgrove and T. Moore (eds), 2007, 297–328. Gwilt, A. in press a. Maescar, Powys: toggle and bell (05.6), In DCMS (in press). Gwilt, A. in press b. Boverton, Vale of Glamorgan: Iron Age collar and bracelets (05.10), In DCMS (in press). Gwilt, A. and Haselgrove, C. (eds.) 1997. Reconstructing Iron Age Societies, Oxford: Oxbow Monograph 71. Haselgrove, C. 1997. Iron Age brooch deposition and chronology, In A. Gwilt and C. Haselgrove (eds.), 1997, 51–72. Haselgrove, C. and Moore, T. (eds) 2007. The Later Iron Age in Britain and Beyond, Oxford: Oxbow Books. Haseloff, G. 1991. Celtic Enamel, In The Celts (Palazzo Grassi, Venice exhibition catalogue), 639–42. Bompiani. Herbert, E. W. 1984. Red Gold of Africa: Copper in Precolonial History and Culture, 301. Madison, Wisconsin: University of Wisconsin Press. Henderson, J. 1987. The Archaeology and Technology of Glass from Meare Village East, In J. M. Coles, Meare Village East; The excavations of A Bulleid and H. St. George Gray 1932–1956, 170–82. Somerset Levels Papers No. 13. 9. Material, style and identity in first century AD metalwork 181 Henderson, J. 1989a. The Scientific Analysis of Ancient Glass and its Archaeological Interpretation, In J. Henderson (ed.), Scientific Analysis in Archaeology, 30–62. Oxford and Los Angeles: Oxford University Committee for Archaeology, Monograph 19 / UCLA Institute of Archaeology, Archaeological Research Tools 5. Henderson, J. 1989b. Analysis of enamels and glasses used in the decoration of brooches from Prestatyn and its archaeological interpretation, In K. Blockley, 1989, 181–3. Henderson, J. 1991. Technological characteristics of Roman Enamels, Jewellery Studies 5, 65–75 Henderson, J. 1995. A reconsideration of the glass: some archaeological and technological aspects, In J. Coles and S. Minnitt, ‘Industrious and fairly civilized’. The Glastonbury Lake Village, 155–60. Somerset Levels Project and Somerset County Museums Service, Exeter. Hill, J. D. 1997. ‘The end of one kind of body and the beginning of another kind of body?’ Toilet instruments and ‘Romanisation’, In A. Gwilt and C. Haselgrove (eds.), 1997, 96–107. Howell, R. 2006. Searching for the Silures: An Iron Age Tribe in South-East Wales, Stroud: Tempus Publishing Ltd. Hughes, M. J. 1972. A technical study of opaque red glass of the Iron Age in Britain, Proceedings of the Prehistoric Society 38, 98–107. Hutcheson, N. C. G. 2004. Later Iron Age Norfolk; Metalwork, landscape and society, Oxford: British Archaeological Rreports British Series 361. Jackson, J. 1951. Tacitus, The Annals IV–XII (English Translation), London: Heinemann. Jackson, R. 1990. Camerton; The Late Iron Age and Early Roman Metalwork, London: British Museum Publications. Jarrett, M. G. 1969. The Roman Frontier in Wales, Cardiff: University of Wales Press. Jarrett, M. G. 2002. Early Roman Campaigns in Wales, rev. P. Webster, In R. J. Brewer (ed), 2002, 41–66. Johns, C. 1996. Bracelets, In R. P. Jackson and T. W. Potter, Excavations at Stonea, Cambridgeshire 1980–85, 334–38. London: British Museum Press. Jones, A. and Macgregor, G. 2002. Colouring the Past; The Significance of Colour in Archaeological Research. Oxford: Berg. Jones, B. and Mattingly, D. 1990. An Atlas of Roman Britain, Oxford: Blackwell. Jope, E. M. 1954. The Keshcarrigan Bowl and a Bronze Mirror-Handle from Ballymoney, Ulster Journal of Archaeology 17, 92–6. Jope, E. M. 2000. Early Celtic Art in The British Isles, Oxford: Clarendon Press (2 vols). Leeds, E. T. 1933. Celtic Ornament in the British Isles Down to A.D. 700, Oxford: The Clarendon Press. Lloyd-Morgan, G. 1997. Objects of Copper Alloy and Silver, In A. G. Marvell and H. S. Owen-John, Leucarum: Excavations at the Roman Auxiliary Fort at Loughor, West Glamorgan 1982–84 and 1987–88, 234–73. Britannia Monograph Series No. 12, London: Society for the Promotion of Roman Studies. Macdonald, P. 2007. Perspectives on insular La Tène art, In C. Haselgrove and T. Moore (eds.), 2007, 329–38. Macgregor, M. 1962. The Early Iron Age Metalwork Hoard from Stanwick, N.R. Yorks., Proceedings of the Prehistoric Society 28, 17–57. Macgregor, M. 1976. Early Celtic Art in North Britain; A study of decorative metalwork from the third century B.C. to the third century A.D., Leicester: Leicester University Press (2 vols.). Manning, W. H. 1981. The Pre-Flavian Forts of South Wales, In W .H. Manning (ed.), Report on the Excavations at Usk 1965–1976; The Fortress Excavations 1968–1971; Cardiff: University of Wales Press. Manning, W. H. 2002. Early Roman Campaigns in the South-West of Britain, In R.J. Brewer (ed.) 2002, 27–44. Manning, W. H. 2004. The Conquest of Wales, In M. Todd (ed.), A Companion to Roman Britain, 60–74. Oxford: Blackwell. Manning, W. H., Price, J. and Webster, J. 1995. Report on the Excavations at Usk 1965–1976; The Roman Small Finds, Cardiff: University of Wales Press. Maryon, H. 1971. Metalwork and Enamelling, New York: Dover Publications. Mattingly, D. 2006. An Imperial Possession. Britain in the Roman Empire 54 BC–AD 409. London: Allen Lane/ Penguin. Megaw, J. V. S. 1967. The Trenoweth Collar, Cornish Archaeology 6, 5–8. Megaw, J. V. S. 1970. Art of the European Iron Age; A study of the elusive image, Bath: Adams and Dart. 182 Mary Davis and Adam Gwilt Megaw, J. V. S. 1971. A Group of Later Iron Age Collars or Neck-Rings from Western Britain, The British Museum Quarterly XXXV, 1–4, 145–56. Nash-Williams, V. E. 1932. Miscellanea: Harness-Trapping from Chepstow, Monmouthshire, Archaeologia Cambrensis 87, 393–4. Nash-Williams, V .E. 1969. The Roman Frontier in Wales, rev. M.G. Jarrett, Cardiff: University of Wales Press. Niblett, R. 1999. The Excavation of a Ceremonial Site at Folly Lane, Verulamium, Britannia Monograph Series 14, London: Society for the Promotion of Roman Studies. Northover, J. P (unpublished) Data on the metallurgical composition of various Iron Age artefacts from Wales. Northover J. P. 1995. Bronze Age Gold in Britain, In G. Morteani and J. P. Northover (eds.), Prehistoric Gold in Europe; Mines, Metallurgy and Manufacture, 515–31. London: Kluwer Academic Publishers. Northover, J. P. 1999. Metallurgical analysis of the bridle bit, In R. Niblett, 1999, 136–7. O’Neill, J. 2002. Reconstructing the Carrickfergus tankard, Ulster Journal of Archaeology 61 (third series), 8–16. Palk, N. A. 1988. Metal Horse Harness of the British and Irish Iron Ages, University of Oxford: unpublished Ph.D. thesis. Piggott, S. 1950. Swords and Scabbards of the British Early Iron Age, Proceedings of the Prehistoric Society 16, 1– 28. Raftery, B. 1984. La Tène in Ireland; Problems of Origin and Chronology, Veröffentlichung des Vorgeschichtlichen Seminars Marburg Sonderband 2, Marburg. Richmond, I. 1968. Hod Hill; Volume Two; Excavations Carried Out Between 1951 and 1958 for the Trustees of the British Museum, London: British Museum. Rigby, V. 1999. Gallo Belgic Wares and Local Imitations, In R. Niblett, 1999, 182–93. Rigby, V. 2005. Appendix 2; Weapons and fittings with enamelled decoration, In I. M. Stead, 2006, 115–21. Romilly Allen , J. 1905. Find of Late-Celtic Bronze Objects at Seven Sisters, near Neath, Glamorganshire, Archaeologia Cambrensis V(sixth series), 127–46. Savory, H. N. 1964. A New Hoard of La Tène Metalwork from Merionethshire, Bulletin of the Board of Celtic Studies XX (Part IV), 449–75, Pls. I–VI. Savory, H. N. 1966a. A Find of Early Iron Age Metalwork from the Lesser Garth, Pentyrch (Glam.), Archaeologia Cambrensis 115, 27–44. Savory, H. N. 1966b. Further Notes on the Tal-y-Llyn (Mer.) Hoard of La Tène Metalwork, Bulletin of the Board of Celtic Studies XXII (Part I), 88–103, Pls. I–V. Savory, H. N. 1971. Excavations at Dinorben, 1965–9, Cardiff: National Museum of Wales. Savory, H. N. 1976. Guide Catalogue of the Early Iron Age Collections, Cardiff: National Museum of Wales. Savory, H. N. 1993. Strap Hook, In K. Blockley, 1993, 211–4. Sealey, P. 2006. Two New Decorated Iron Age Mirror Finds from Essex, In P. Ottaway (ed.), A Victory Celebration: Papers on the Archaeology of Colchester and Late Iron Age-Roman Britain presented to Philip Crummy, 11– 18. Colchester: Colchester Archaeological Trust. Spratling, M. G. 1972. Southern British Decorated Bronzes of the Late Pre-Roman Iron Age, University of London: unpublished Ph.D. thesis(2 vols.). Stead, I. M. 1986. Other Bronze Objects, In I. M. Stead and V. Rigby, Baldock; The Excavation of a Roman and PreRoman Settlement, 1968–72, 125–40. Britannia Monograph Series No. 7. Stead, I. M. 2005. British Iron Age Swords and Scabbards, London: The British Museum Press. Taylor, R. J. and Brailsford, J. W. 1985. British Iron Age Strap-Unions, Proceedings of the Prehistoric Society 51, 247– 72. Unz, C. and Deschler-Erb, E. 1997. Katalog der Militaria aus Vindonissa, Veröffentlichungen der Gesellschaft pro Vindonissa, Band XIV. Wainwright, G. and Spratling, M. 1973. The Iron Age Settlement of Gussage All Saints, Antiquity XLVII, 109– 130. Webster, G. 1971. A Hoard of Roman military equipment from Fremington Hagg, In R. M. Butler (ed), Soldier and civilian in Roman Yorkshire, 107–25. Leicester: Leicester University Press. Webster, G. 1981. Rome Against Caratacus; The Roman Campaigns in Britain AD 48–58, London: Batsford. 9. Material, style and identity in first century AD metalwork 183 Webster, G. 1990. Note: A Late Celtic Sword-Belt with a Ring and Button Found at Coleford, Gloucestershire, Britannia 21, 294–5. Webster, J. 1993. Objects of Copper Alloy, In K. Blockley, 1993, 201–11. Webster, J. 1995a. Harness Fittings, In W. H. Manning, J. Price and J. Webster, 1995, 36–41. Webster, J. 1995b. Bells, In W. H. Manning, J. Price and J. Webster, 1995, 55–8. Webster, J. 2003. Bronze Objects, In H. James, Roman Carmarthen; Excavations 1978–1993, 289–325. Britannia Monograph Series No. 20, London: Society for the Promotion of Roman Studies. Webster, P. V. 1984. The Roman Period, In H. N. Savory (ed.) Glamorgan County History; Volume II Early Glamorgan: Pre-History and Early History, 277–313. Cardiff: Glamorgan County History Trust. Wheeler, R. E. M. 1925. Prehistoric and Roman Wales, Oxford: Clarendon Press. Wheeler, R .E. M. 1926. The Roman Fort near Brecon, Y Cymmrodor; The Magazine of the Honourable Society of Cymmrodorion 37, 1–260. Wild, J. P. 1970. Button-and-Loop Fasteners in the Roman Provinces, Britannia 1, 137–55. Young, D. 2006. The colour of Things, In C. Tilley, W. Keane, S Küchler, M. Rawlings and P. Spyer (eds.), Handbook of Material Culture 173–85. London: Sage Publications.