The Antiquaries Journal, 84, 84 2004, pp 1– 22 THE WINCHESTER HOARD: A FIND OF UNIQUE IRON AGE GOLD JEWELLERY FROM SOUTHERN ENGLAND J D Hill, FSA, Anthony J Spence, Susan La Niece, FSA, and Sally Worrell* An unusual group of gold jewellery was discovered by a metal detectorist near Winchester in southern England in 2000. The hoard included two possibly unique massive necklaces made in a clearly classical style, but different from typical classical necklaces and from the torcs and collars of Iron Age Europe. The hoard also contained extremely rare gold versions of types of brooches commonly made in bronze and iron in north-west Europe during the first century BC, the end of the pre-Roman Iron Age. This paper describes these unique objects and the results of an archaeological investigation of their find spot. Detailed scientific analysis of the objects’ technology has proven crucial for interpreting their origins and broader significance. Finally, the broader consequences of the find for interpreting the significant changes that took place in southern Britain in the century before the Roman conquest are discussed. INTRODUCTION This paper presents a recently discovered hoard of unusual pre-Roman Iron Age jewellery from southern England. The find includes two unique gold necklaces or torcs together with rare gold versions of Iron Age brooches (fig 1). The technology used to make the torcs points to jewellers trained in the techniques of classical goldsmiths, yet the heavy gold neck ornaments are still recognizably traditional temperate European Iron Age ‘barbarian’ status symbols. The find is also important because close liaison between the finder, the landowner and archaeologists enabled the find spot to be investigated. This has allowed archaeologists to gain a better context for the hoard. This close co-operation between metal detectorists and archaeologists is one of the highest profile successes of the Portable Antiquities Scheme for England and Wales to date. This paper briefly describes the hoard and its discovery, before outlining some of the wider implications of the find.1 DISCOVERY In September 2000 Mr Kevan Halls was metal detecting in a field east of Winchester in Hampshire, southern England. Detecting with the permission of the landowner, he found a * J D Hill, Anthony J Spence and Susan La Niece, The British Museum, Great Russell Street, London WC1B 3DG, UK. E-mail: <prehistoryandeurope@thebritishmuseum.ac.uk> (JDH); <science@thebritishmuseum.ac.uk> (SLN). Sally Worrell, Portable Antiquities Scheme Research Officer, Institute of Archaeology, University College London, 31 – 34 Gordon Square, London WC1H 0PY, UK. E-mail: <s.worrell@ucl.ac.uk>. 2 THE ANTIQUARIES JOURNAL Fig 1. The Winchester Hoard. Photograph: Sandra Kemp gold brooch with a fine gold chain still attached. A few minutes later, working in the direction of the recent ploughing, he found the twin of the first brooch. All finds of ancient gold or silver objects in England and Wales, other than single coins, must be promptly reported to the appropriate authorities under the terms of the Treasure Act 1996.2 Mr Halls reported his find to Sally Worrell, the Finds Liaison Officer for Hampshire, based at the Winchester Museums Service. Sally contacted the British Museum, which identified the items as gold versions of late pre-Roman Iron Age brooches. Gold pre-Roman Iron Age brooches are extremely rare finds in temperate Europe. The Portable Antiquities Scheme is a project designed to encourage metal detectorists and members of the public to report and record the archaeological finds they make.3 Funded from central government and the Heritage Lottery Fund, the project began as a pilot scheme in 1997 and in 2003 was extended to cover the whole of England and Wales. The scheme’s Recording Officers also act as the initial point of contact for reporting finds of potential items of Treasure (ie, Treasure as defined under the 1996 Act and its subsequent amendments). Because of the importance of the find, Mr Halls was asked to mark the locations of any further finds he made in the field. A week after his first finds, he found the larger of two necklace torcs with its securing pin, another gold brooch along with one and half gold bracelets. THE WINCHESTER HOARD 3 The complete bracelet was found next to the third brooch. The first necklace torc was a completely unparalleled object, classical in inspiration but found with Iron Age brooches. It was rapidly decided that an immediate archaeological investigation of the find spot was needed. This was aimed at discovering more about the specific context of the find spot, and to establish whether the artefacts came from a larger archaeological site that could be seriously damaged if pillaged by irresponsible metal detectorists. With the good relationship built up between the Finds Liaison Officer and Mr Halls, and the close co-operation of the landowner, archaeologists from the British Museum and Winchester Museums Service were able to investigate the site in the autumn of 2000 and spring 2001. At the same time Mr Halls continued to detect in the field, finding a second necklace torc at the end of October and, finally, a fourth brooch and other bracelet half in December. All the objects show evidence of having been damaged by farm machinery. This damage includes fresh scratches, roughened surfaces and bending of three of the brooches. Mr Halls found the objects at various depths in the plough soil, and most were found within a 25m by 12.5m area on top of a small hillcrest. The only objects not found in this area appear to have been moved up and down the field in the direction of the ploughing. Brooches 1 and 2 with the chain were found about 125m to the south, and the smaller necklace torc was found 200m to the north. The damage to the smaller necklace torc suggests that it was caught on the end of a plough or harrow and fell off when the tractor turned at the end of the field. The find spot of the main group of objects is probably a good indication of where the hoard was deposited. This was on the false crest of a low hilltop capped with clay, on one side of a small valley. The investigation of the find spot, led by Tony Spence, has involved small-scale excavations, field survey and an intensive metal-detector survey. The excavations over each find spot revealed no evidence for archaeological features below the plough soil. Ploughing over the capping of clay had disturbed the soil down to a depth of 200 – 300mm and would have destroyed any shallow feature or features in which the hoard might originally have been buried. There are no other artefacts in the plough soil other than a few struck flints of a much earlier date than the hoard and modern farming detritus. The lack of any other archaeological material in the plough soil suggests the hoard did not come from a settlement, cemetery or Late Iron Age=Roman shrine or temple. Further fieldwork in 2001 was suspended because of the foot and mouth epidemic. The hoard was declared to be treasure at a coroner’s inquest at Winchester on 22 March 2001. Because of its international significance, the hoard has been acquired by the British Museum with substantial help from the National Heritage Memorial Fund, the National Art Collections Fund and the British Museum Friends. The hoard is now on display in the British Museum and has been displayed in Winchester. THE HOARD The hoard contains two gold necklace torcs, two pairs of gold brooches, one pair still linked by a chain, and a pair of gold bracelets (weights and dimensions are summarized in table 1). The two necklace torcs and brooch pairs suggest two sets of personal ornaments and, as the necklace torcs are of different sizes, perhaps a male and female set (as illustrated in fig 14) or senior and junior. Since the hoard was acquired, one of the authors (SLaN) of the British Museum’s Department of Scientific Research has undertaken a technological study of the objects. The full results of this research will be published elsewhere.4 4 THE ANTIQUARIES JOURNAL Table 1. Metal composition, dimensions and weights of the objects in the hoard Metal composition (%) Necklace torc 1 Weight (gm) Dimensions (mm) 516.7 480 l. chain terminal Gold Silver Copper wire sheet gold beaded wire 93.9 95.7 94.8 93.3 5.2 3.7 4.5 4.9 0.9 0.7 0.7 1.8 wire coiled wire granule s-filigree head solder area 96.8 97.0 96.6 97.7 96.5 94.9 2.4 2.1 2.6 1.6 2.6 1.5 0.8 1.0 0.8 0.7 0.9 3.6 pin Necklace torc 2 332.1 440 l. chain terminal pin Brooch 1 22.2 60 l. brooch collar 92.7 92.3 5.1 5.0 2.2 2.7 Brooch 2 22.5 60 l. brooch collar 93.5 93.6 4.9 4.4 1.6 2.0 Brooch chain 23.6 170 l. chain collar 93.9 94.3 4.8 4.0 1.3 1.7 Brooch 3 20.7 80 l. 90.4 6.7 3.0 Brooch 4 20.5 80 l. 90.3 6.9 2.8 Complete bracelet 94.1 90 diam. 95.2 3.1 1.7 Broken bracelet 53.3; 53.1 98.9 1.0 0.2 The analyses have a precision (measure of reproducibility of the result) of c  1 2 per cent of the concentration of gold and 5 20 per cent of the concentration of silver and copper, and the accuracy of the analyses are similar. Gold Composition The metal composition was determined by X-ray fluorescence analysis on small cleaned areas and the results are given in table 1. Only gold, silver and copper were detected. The analyses have a precision (measure of reproducibility of the result) of c  1 2 per cent of the concentration of gold and 5 20 per cent of the concentration of silver and copper, and the accuracy of the analyses are similar. There is little difference in composition between the uncleaned original surface and clean metal several micrometers below, indicating that virtually no surface enrichment has occurred either during the manufacturing process or by corrosion during burial. The gold content of all the items in the hoard is above 90 per cent. The percentage of copper is low – so low in the necklace torcs that it might be regarded as an accidental impurity rather than a deliberate addition to the alloy – but it is significantly higher in areas of solder. The silver content, below 7 per cent, is unusually low for Iron Age Britain. Gold and silver THE WINCHESTER HOARD 5 Fig 2. Ternary plot of the gold-silver-copper alloy composition of the objects in the hoard (table 1). The larger triangle shows the detail enlarged from the full diagram, above left: all the alloy compositions are higher than 90 per cent gold and less than 7 per cent silver. The outlying grey triangular point is an analysis of a solder area on the small torc; the solder is copper-rich objects, other than coins, are rare from Iron Age Britain and the sources of the metal are difficult to establish. The gold content of all the pieces in this hoard is significantly higher than that of Iron Age coins from late pre-Roman Iron Age Britain and France, so local coinage as a direct source of metal can be excluded.5 Gold with less than 10 per cent silver does occur naturally in Britain, but gold of this composition does not seem to have been exploited in prehistoric times.6 Where the composition of pre-Roman British objects is known, it almost always has more than 10 per cent of silver in the gold alloy, for example the Ipswich torcs7 and the objects from Snettisham.8 Even amongst much earlier Bronze Age gold work of the British Isles, silver contents as low as these are rare.9 If this gold came from Britain it was either from a source apparently not otherwise exploited, or was refined gold. There is no evidence for gold refining in Britain or France and we have to look to the classical world where coinage was of refined gold. Analyses of Continental gold torcs have identified very few with high purity gold, one exception being a group from south-western France with 1 – 2 per cent silver or less.10 Whether the gold used to make the Winchester hoard included refined metal or not, it is indisputably very unusual for Britain, and another source or sources of gold for all the items in the hoard must be looked for, probably within the Roman and Hellenistic world. 6 THE ANTIQUARIES JOURNAL What is not clear is whether all these objects were made at the same time by the same workshop. The initial impression when the hoard was discovered was that it was all made together as two sets of jewellery. This was because of the extreme rarity of gold brooches and the unique nature of the necklace torcs. The unusual purity of the gold in all the objects might support this interpretation. However, there are differences in metal composition between individual items that are greater than the differences between components of each item (fig 2). The exception to this is the copper-rich solder area on the small necklace torc (the outlier in fig 2 with the most copper). In particular, the brooches of the larger pair (3 and 4) are virtually identical in composition to each other, but distinct from Brooches 1 and 2. Furthermore, there are differences in the manufacture between the pairs of brooches that might point to different makers. The Necklace Torcs The most unusual objects are the necklace torcs (figs 3 and 4). Although not identical in size or detail, they were made in the same manner. No other objects of this type have so far been found from Iron Age temperate Europe. The term ‘necklace torc’ is used because they are related to metal torcs, collars and neck-rings that were a common ornament in pre-Roman Iron Age Europe, yet these look very different. They are also made in a very different manner and are far more flexible than other contemporary torcs or collars, such as those from Snettisham. Other British and Irish Iron Age torcs are made either as tubes of thin gold sheet wrapped around an iron or organic core (eg, the Broighter torc) or by twisting thick bars or strands of fine wires together (eg, examples from Snettisham). At best, the finer wire rope torcs have some flexibility to allow them to be put on, but all were rigid collars when worn. The Winchester necklace torcs were made with loop-in-loop chains that would hang when worn and move as their wearers moved. The loop-in-loop chains were constructed by threading and bending presoldered wire rings through each other (fig 5).11 The technology was commonplace in the Roman and Hellenistic world at this time. However, Greek and Roman chains made in this way were usually constructed of fine wire of c 0.5mm diameter and the chains are rarely more than 4mm thick. The brooch chain found with this hoard is typical of the thickness of these chains, but the smaller necklace torc is c 10mm thick and made of wire 1.2 – 1.4mm in diameter, while the larger is c 12mm thick, constructed from twisted wire 1.5 – 1.7mm in diameter. Examination of the wire of both chains under a microscope shows extensive wear, for the wire is compressed where the chain flexed with movement, especially near the clasp. Both necklace torcs are closed with a clasp, the construction and decoration of which is Roman or Hellenistic in inspiration and technology (figs 6 and 7). The terminals consist of decorated sheet gold cylinders soldered around one end of the chain. The necklace torc is closed by inserting the other end of the chain into the opening of the cylindrical clasp and passing a split pin through the whole assemblage. The cylindrical clasp of the larger necklace torc is 22mm long and 12mm in diameter and is made up of several components soldered together. The end of the chain to which the terminal is soldered was roughly cut through. The dominant decorative feature is the pair of openwork tubes of coiled gold wire soldered around the clasp, framed by strips of beaded gold wire (fig 6). The clasp of the smaller necklace torc is 19.8mm long and 20.7mm in maximum diameter. Again, the dominant decorative features are openwork tubes of coiled gold wire, but these are widely spaced apart (figs 7a and b). It is more elaborately decorated, with fine granulation and filigree. On one side of the clasp are three pyramids of gold granules, each made up of five gold THE WINCHESTER HOARD Fig 3. The larger necklace torc. Photograph: Sandra Kemp Fig 4. The smaller necklace torc. Photograph: Sandra Kemp 7 8 THE ANTIQUARIES JOURNAL Fig 5. The loop-in-loop technique for making necklace torc chains. Drawing: Karen Hughes Fig 6. The terminal of the larger necklace torc, decorated with a pair of spirals of coiled wire, framed by strips of beaded wire. Photograph: A Milton THE WINCHESTER HOARD a b Figs 7a and b. The terminal of the smaller necklace torc: (a) showing the granulation on one side and decorative pin head at the top; (b) showing the filigree on the other side. Photograph: A Milton 9 10 THE ANTIQUARIES JOURNAL a b Figs 8a and b. Scanning electron micrographs of granulation and beaded wire on the clasp of the smaller torc. The diameter of each granule is c 2mm. Photograph: Susan La Niece THE WINCHESTER HOARD 11 Fig 9. Scanning electron micrograph showing the bonding between the granules on the smaller torc, indicative of the classical technique of diffusion soldering. Photograph: Susan La Niece spheres just over 2mm in diameter. On the other side are two figure-of-eight shapes of beaded wire (figs 8 and 9). There is no visible solder, suggesting that the filigree and granulation were soldered by the classical technique of diffusion soldering. Instead of using a lower meltingpoint metal as a solder, the method uses a glue and copper compound mixture. When heated the burning glue reduces the copper compound to copper, which alloys with the gold to form a bond between the gold components.12 In contrast, the flow of metal where the larger components of the clasps are joined suggests the use of metallic gold solder. On the smaller necklace torc the pin passes through the centre of the cylindrical clasp, guided into position by two small tubes inside the cylinder that allow the pin to line up with the loop at the end of the chain. The pin has a decorative head and a safety catch mechanism to prevent it coming right out of the terminal. Unlike the large torc, the end of the chain to which the clasp is attached is neatly finished rather than cut. The quality of the work in all aspects of the small torc is noticeably higher than the larger torc. This might suggest they were made by different craftspeople. On the other hand, the use of soldered wire components, granulation and filigree, along with the loop-in-loop technique, are features that are not characteristic of British and north European Iron Age goldwork. The unique style of the two necklace torcs, the similarity in metal composition and the shared decorative features links them closely. They may have been made in the same workshop, but if not, the goldsmith who made the second torc (whichever that might have been) had examined the other closely enough to model the work on it. 12 THE ANTIQUARIES JOURNAL a b Fig 10. Brooches 1 (a) and 3 (b). Drawing: Karen Hughes Brooches The brooches form two chained pairs, although one chain is missing. Both pairs of brooches belong to types of fibula (safety-pin-style brooches) commonly made in bronze and iron across central and western Europe in the first century BC (fig 10). All the brooches have four coil springs with internal cords and simple openwork catch plates. Each brooch was cast in one piece, then worked to extend and shape the coil spring and pin. Brooches 1 and 2 are gold versions of Knotenfibeln, so called because of the collar on the bow.13 Bronze versions of similar brooches are known in south-east England. Brooches 3 and 4 are versions of a Gallic La Tène III type found mostly in France.14 They are virtually identical in composition to each other, and slightly baser than the other pieces in the group. The decoration around the head is unusual. Exact parallels for Brooches 3 and 4 are harder to find. Each end of the fine loop-in-loop chain linking the first pair of brooches has a gold collar soldered to a ring that hooks through a wire loop with identical decorative collar on the head of each brooch. The very close similarity in gold composition between Brooches 1 and 2, their collared links and the chain suggests that they were made together. All four brooches show little obvious wear on their pins, which is unusual given the soft nature of the gold. This is in contrast with the very noticeable wear on their collared links where they were attached to their chains and on the wire of the surviving brooch chain (fig 11). This wear is consistent with THE WINCHESTER HOARD 13 Fig 11. Detail of the compression wear (arrowed) on the links of Brooch 1, indicating significant use. Photograph: Susan S La Niece the pull that would be exerted by the weight of clothing, and if the brooches were worn with their spring end uppermost. The lack of wear on the pins might suggest they were not often unfastened. Perhaps the cloak was taken off over the wearer’s head without unpinning the brooches. Alternatively, could the lack of wear on the pins suggest the wearers did not dress themselves and had other people pinning and unpinning the brooches for them? Gold and silver brooches are extremely unusual in temperate Iron Age Europe, and most date to the first century BC. Silver brooches are more common than gold, usually in chained pairs, and many are versions of Knotenfibeln.15 There are perhaps fewer than twenty Iron Age gold brooches known from temperate Europe. In Britain, there were only two or three previous examples, none with exact provenances. They include a gold Knotenfibeln reputed to have been in Hampshire.16 There is reason to assume many of these brooches were made in temperate Europe and some were probably made in southern Britain or northern France.17 Bracelets or Ingots The bracelets are undecorated and less well finished than the other objects. Gold or silver bracelets are also unusual in pre-Roman Iron Age temperate Europe and these have no immediate parallel. Several features suggest that the purpose of the Winchester bracelets might 14 THE ANTIQUARIES JOURNAL be as ingots rather than ornament. No attempt was made to smooth the casting shrinkage at one end of the broken bracelet. The fracture surfaces have a patina indicating that the break is not modern and the weight of the two halves is so similar as to suggest that the bracelet was deliberately halved. Furthermore, the purity of the gold is very high for wearable jewellery and is likely to have been refined. WHEN WERE THE OBJECTS MADE AND DEPOSITED? Had the necklace torcs been found on their own they would not have been recognized as dating to the pre-Roman Iron Age. Instead they would have been interpreted as unusual Roman-period objects dating to after the Roman conquest of southern Britain in AD 43. The only datable objects in the hoard are the brooches. These are versions of types common in north-west Europe broadly between c 80=70 and 40=30 BC (the typological period known as La Tène D2a), but might start earlier and carry on in use later.18 The necklace torcs and the brooches all have evidence of considerable wear. How long such wear patterns might take to form is difficult to estimate, and depends on whether these objects were worn regularly, or only on special occasions. Assuming that there was not a very long time period between the manufacture of the necklaces and the brooches, then all were probably made some time between c 80 and 30 BC. This is the date of their manufacture, not necessarily the date when they were deposited. THE LOCAL CONTEXT The hoard was found on a low hilltop overlooking a small east – west valley. While not a prominent feature in the landscape, there are good views in many directions from this bluff. Fieldwork so far has found no evidence for burials, settlement or any building where the hoard was discovered. In common with many finds of Iron Age fine metalwork from Britain, this hoard appears to have been deposited ‘in the landscape’, away from domestic settlements. The hilltop on which the hoard clustered has a capping of heavy clay. Most of the surrounding slopes and hilltops have thin chalk soils. It is unlikely that this patch of heavy clay was cultivated and it might instead have been grassland, scrub or woodland when the hoard was placed there. The evidence from aerial photography, field survey, metal detecting and excavation suggests the wider area east of Winchester in the second and first centuries BC had a relatively high density of small settlements, but it is not clear if the area where the hoard was found had a similar density of settlement. Other finds by Kevan Halls help fill in the picture. These include a possible settlement focus with first-century AD brooches, Republican and early Imperial Roman silver coins and other material 750 – 1,500m from the hoard find spot. In the valley bottom less than 300m from the gold find came an unusual Roman bronze handle fitting (fig 12). In the shape of a duck or swan’s head with niello eyes, this held one of two handles on a shallow bronze bowl made in Italy probably between c AD 1 and AD 70.19 The valley is within 10km of the excavated contemporary settlement complex at Owslebury.20 This settlement was receiving imported Roman wine by the late second or early first century BC. It also contained a very atypical midfirst-century BC burial that included weapons and a probable Continental high-tinned bronze belt hook. This burial became the focus of a cremation cemetery, which contained imported THE WINCHESTER HOARD 15 Fig 12. A handle fitting in the shape of a duck or swan’s head from an Italian metal vessel. Photograph: Sandra Kemp north-eastern French pottery. It is not known if this is the only settlement complex of its type in the area, or if others remain to be discovered. The clear links between this settlement’s inhabitants with other communities in parts of northern France and potentially further afield, before and after the Roman conquest of Gaul in the 50s BC, provide one context in which to set the extraordinary gold objects in this hoard. Winchester itself in the second to first centuries BC was probably an important settlement, taking the form of a large enclosed valley-side settlement at Oram’s Arbour.21 TREASURE OR RITUAL? What were these gold objects doing on the hilltop? The interpretation of finds of British Iron Age metal objects and coins, both on settlements and in the landscape, has been a controversial one in recent years. Many finds of Iron Age gold torcs and coins have traditionally been seen as wealth buried for safekeeping. However, recent approaches increasingly argue that deposits of torcs and gold coins were ritual events that specifically had to take place ‘in the landscape’ and away from domestic occupation.22 This pattern of ritual deposition of metal objects in specific parts of the landscape that were not usually marked by built structures such as shrines was common across later prehistoric Britain and northern Europe.23 Within these broad ritual traditions, it was appropriate to place particular types of objects in certain locations and not others. Torcs are common in dry-land contexts, weaponry and cauldrons are common finds from rivers and other wet places.24 This suggests that the Winchester hoard was not buried or hidden for safekeeping, but was probably an offering made away from domestic settlement on a particular hilltop. 16 THE ANTIQUARIES JOURNAL DISCUSSION The Roman=Hellenistic look and technology of these necklace torc set them apart from other torcs and collars from Iron Age Europe. No other torcs from temperate Europe are made in this way, although there is at least one other earlier massive loop-in-loop chain from the Scythian world.25 The unique features of the Winchester necklace torcs pose the questions of where, by whom and for whom were they made? Especially, were they made by native British or French craftspeople, or by craftspeople from the Roman and Hellenistic Mediterranean world? A small number of thin silver standard-size loop-in-loop chains are known across firstcentury BC temperate Europe. These include the brooch chain from Great Chesterford in Essex, and a chain from the La Câtillon hoard, Jersey, dating to the middle of the first century BC. A large gold chain made from single loops was excavated at Croft Ambrey, Herefordshire, and is dated to the ‘Late Iron Age’.26 Two gold necklaces from the Broighter hoard from Ireland are also relevant here.27 Seen as Hellenistic or Roman products, one is a single fine chain, the other three fine chains held with a terminal decorated with granulation. The pin holding this necklace together is similar to that of the smaller Winchester necklace torc. The Broighter necklaces are probably associated with a decorated first-century BC gold tubular torc and a gold model boat. However, the circumstances surrounding the discovery of the hoard are suspect. This has raised questions about whether these two unparalleled Hellenistic=Roman necklaces are genuine first-century BC=AD imports to Ireland. The Winchester hoard makes this more likely. Some fine loop-in-loop chains might be imports from Roman or Hellenistic workshops, but others were probably made in temperate Europe. Given the skill of Iron Age gold- and silversmiths, replicating the loop-in-loop technique would not be difficult. However, the techniques of granulation and filigree are virtually unknown on temperate European objects before the Roman conquest other than at Winchester and Broighter. These techniques would be very hard to copy from seeing a finished product and would require the processes to be studied at first hand. This suggests that the Winchester necklace torcs were probably made by craftspeople trained in a Mediterranean Roman or Hellenistic workshop. This interpretation is supported by the high purity of the gold, unparalleled in northern European gold at this period and best explained by the use of refined Roman gold. The lack of other examples of filigree and granulation in north-west Europe before about AD 1 – 50 suggests that few, if any, French and British goldworkers of the early to middle first century BC were using these Roman or Hellenistic techniques. If made in northern France or Britain, then Mediterranean-trained goldworkers did not stay or pass on their skills. Perhaps the necklace torcs were made further south, commissioned either by native elites or by Roman=Hellenistic elites as gifts for native rulers. The people who deposited them on the hilltop in Hampshire need not have been the original owners. Gold objects of this quality could have had long and complicated biographies. Who finally deposited the hoard is unknown. Historical evidence for southern England between 100 BC and AD 50 from coinage and Roman writings is very sparse, but there is often a strong temptation to link major finds, such as this hoard, to one of the poorly known figures or events recorded from these sources. For example, a generation or so after the Roman conquest of Gaul in the 50s BC a polity emerged that ‘controlled’ some 300 – 400 square kilometres of southern England, including the Winchester area. This polity is often called the ‘southern kingdom’ and was ruled by individuals who called themselves ‘Rex’ (King). These rulers had close interactions with Rome and Gaul, and they may have been political clients of Rome.28 It 17 THE WINCHESTER HOARD KEY Archaeological sites Rive r Modern cities ame Th s Reading Silchester Danebury Winchester River Avo 0 R Chichester Hayling Island n 0 Itchen Te st Southampton St Catherine’s Hill Owslebury iver River Salisbury 50 kilometres 50 miles Fig 13. A map of central southern England, showing the location of places mentioned in the text. Drawing: Karen Hughes is (too) easy to link the Winchester hoard to one of the leaders of this kingdom, perhaps Tincomarus who from around 20 BC issued the first coins of this polity to regularly carry a name – in Latin. Alternatively, he and his successors often call themselves on their coins ‘son of Commius’. This name is also found on a small number of coins that probably pre-date or may coincide with the first coins of Tincomarus.29 The Commius of the coins is often taken to be the same Commius known from Julius Caesar’s account of his conquest of Gaul. This Commius was a native aristocrat who initially sided with Rome but then changed sides and subsequently escaped or was sent to England in the late 50s or early 40s BC. Although the equation of the Commius of Caesar and the Commius on the coins is attractive, it is not without its problems.30 The temptation to link finds to the few poorly known figures of this time in order to write ‘history’ is very problematic and potentially dangerous. The Winchester hoard might have been associated with Caesar’s Commius, the other Commius (if not the same person) or Tincomarus. But it is just as likely that other, totally unknown, individuals deposited the hoard. The 18 THE ANTIQUARIES JOURNAL Fig 14. A reconstruction of how the jewellery might have looked when worn. Painting: Karen Hughes archaeological evidence for southern England in the first century BC (fig 13) shows that the Winchester hoard was probably made and deposited at a time of considerable social and cultural transformation.31 Contacts between northern France and southern England became extremely important for some individuals and groups from the second century BC onwards with the adoption of similar coinage in both areas, along with exchange, kinship and political links. At the same time, well-known Wessex hillforts such as Danebury were abandoned between 100 and 50 BC.32 The Winchester region seems to be typified by a different settlement and social pattern from the hillfort-dominated areas to the west. At Owslebury and to the east a new burial rite, similar to that in northern France, was adopted.33 Other links to northern France seen in this area include the construction of a temple or shrine complex at Hayling Island in the early to middle first century BC, at which were deposited large numbers of northern French coins.34 In north Hampshire a major planned agglomerated settlement at Silchester was founded in about THE WINCHESTER HOARD 19 50 BC, although the main occupation did not begin until around 25 BC, and again shows links to northern France.35 Coinage from Hampshire in the early–middle first century BC suggests the existence of several competing individuals and polities in the area, some of which are the forerunners of parts of the southern kingdom. Others have no later successors, such as a group centred north of the Solent and immediately south of Winchester.36 The archaeological, numismatic and limited written evidence suggests a complex and fluid social situation in this region throughout the first century BC, with a number of nameless and named groups and individuals establishing and restructuring their identities and sources of social power, some with close links to parts of northern France and Roman authorities. There is increasing evidence that some groups and individuals consciously adapted and adopted foreign objects and social practices to create and mark identifications and distinctions with others at this time.37 It is in this context that the strikingly different Winchester hoard should be set (fig 14). CONCLUSION Unique objects, especially when they are discovered by chance, pose difficult questions of interpretation. Had the massive chain torcs from this find not been found with associated brooches, they might not have even been recognized as pre-Roman objects. Investigations of the find spot, combined with detailed scientific analysis, have enabled us to provide a fuller context for understanding this hoard. The necklace torcs are probably the products of classically trained goldsmiths and may have been made in the Mediterranean world as a special commission to create a traditional north European status symbol in a distinctively different way. Who commissioned them and how many hands they might have passed through before arriving in the Winchester area is not known. There is a strong case, however, for these gold items having been made for one of the new rulers who held sway over the region from the late first century BC. When they were finally deposited on the hillside near Winchester cannot be precisely known, but in keeping with recent interpretations of other finds of British pre-Roman Iron Age fine metalwork, their deposition was probably a ritual act. The discovery of the Winchester hoard and its subsequent investigations is also a great success for the Portable Antiquities Scheme. A major problem with many previous finds of spectacular metalwork and coins from England and Wales has been a systematic lack of contextual information.38 Many previous finds do not even have a general location, and when the location is known, there is usually little other information about the context; was it from a settlement, intentionally buried or a casual loss; if buried, what with and how? Such information greatly increases the archaeological and historical significance of such finds. More detailed investigations of the find spots of major metalwork discoveries have previously happened largely by chance. This is an area that the Portable Antiquities Scheme can help systematize, as its staff are working closely with detectorists and will often be the first archaeologists to learn of a find. Through the personal contact already established between the Hampshire Finds Liaison Officer and Kevan Halls this find was reported promptly, allowing more detail about the circumstances of the find to be recorded than would otherwise have been possible. Advice given by the Liaison Officer led Kevan Halls to mark his further find spots and this in turn allowed the subsequent archaeological investigations to take place. Without the Portable Antiquities Scheme, it is doubtful that so much could have been learnt about the context for this spectacular find. Here the considerable help from the landowner also played a major role. If this find becomes a model for dealing with future discoveries, then many more 20 THE ANTIQUARIES JOURNAL ‘stray’ finds of often spectacular objects will cease to be just ‘treasures’ and become more valuable social ‘facts’ for understanding past English and Welsh societies. ACKNOWLEDGEMENTS The authors wish to express their thanks to Robert Anderson, Richard Whinney and Ken Qualmann, whose help was essential for the prompt investigation of the site; to Pete Makey and Penny Copeland for digging in often appalling weather; to Karen Hughes for her illustrations and Sandra Kemp for her photography; to Beverley Hirschel and Susan Walker for help with comparisons, and to John Collis, Michael Cowell, Paul Craddock, Roger Bland, Andrew Fitzpatrick, Colin Haselgrove, Richard Hobbs, Catherine Johns, Ian Stead and Jonathan Williams for reading and commenting on earlier drafts of this paper. Especial thanks must be given to the landowner, without whose help, interest and enthusiasm much of what we now know about the hoard would not be known. NOTES 1. Technical and scientific aspects of the study of the hoard will be covered in La Niece, S and Hill, J D forthcoming. 2. For further information about the Treasure Act, see <http://www.finds.org.uk/background/ treasure.asp>. 3. For further information about the Portable Antiquities Scheme, see <http://www.finds. org.uk>. 4. La Niece, S and Hill, J D forthcoming. 5. Cowell 1992; Northover 1992. 6. Hartmann 1980; Leake et al 1993. 7. Brailsford and Stapley 1972. 8. Stone 1987; Hook unpublished. 9. Hartmann 1980; Hook and Needham 1989. 10. Eluère 1987. 11. Ogden 1982. 12. Components: Ogden 1982. 13. The first two brooches found are Almgren 65s and Feugère type 8bs in these standard Iron Age brooch typologies. 14. These two brooches equate to Feugère’s (1985) types 2, 4, 5b, 6b, 9b in this standard brooch typology of western European Iron Age brooches. 15. Krämer 1971. 16. Hattatt 1987, no. 749. 17. Fitzpatrick and Megaw 1987. 18. Collin 1998; Feugère 1985; Gebhard 1991. 19. Compare this terminal to those on bronze vessels from Pompeii such as vessels 2866 and 3375 in Tassinari 1993. 20. Collis 1968; 1970. 21. Whinney 1994. 22. See opposing arguments in Stead 1991 and Fitzpatrick 1992 for the interpretation of why the Snettisham torcs were originally buried. 23. See Bradley 1990; 2000; Fitzpatrick 1984; 1992; Haselgrove 1999; Hill 1995; Hunter 1997. 24. Fitzpatrick 1984; Bradley 1990; Stead 1991. 25. Artamonov 1969. 26. Stanford 1974. 27. Raftery 1984. 28. Creighton 2000. 29. Some, such as Creighton 2000, argue that the Commius coins well pre-date those of Tincomarus. However, the Commius coins may only pre-date Tincomarus by a few years or even coincide with the first issues of Tincomarus coins – Jonathan Williams, pers comm. 30. Creighton 2000 gives the most recent full interpretation that identifies the Commius of the southern kingdom’s coins with the Commius of Caesar. This identification, however, should not be taken as proven. 31. Cunliffe 1987; 1995; 2000; Creighton 2000; Haselgrove 1982; 1989; 1999; Hill 1995; Fitzpatrick 2001; Sharples 1990. 32. Cunliffe 2000. 33. Fitzpatrick 1997. 34. King and Soffe 1994; 2001; Briggs et al 1992. 35. Fulford and Timby 2000. 36. Haselgrove 1993; 1994. 37. Haselgrove 1994; 1999; Hill 2002; Sharples 1990. 38. Stead 1998. THE WINCHESTER HOARD 21 BIBLIOGRAPHY Artamonov, M I 1969. Treasures from the Scythian Tombs, London Bradley, R 1990. The Passage of Arms: An Archaeological Analysis of Prehistoric Hoards and Votive Deposits, Cambridge Bradley, R 2000. The Archaeology of Natural Places, London Brailsford, J and Stapley, J 1972. ‘The Ipswich torcs’, Proc Prehist Soc, 38, 38 219 – 34 Briggs, D, Haselgrove, C and King, C E 1992. ‘Iron Age and Roman coins from Hayling Island temple’, Brit Numis J, 62, 62 1– 62 Collin, A 1998. Chronologie des oppida de la Gaul non me´diterrane´enne, Paris Collis, J R 1968. ‘Excavations at Owslebury, Hants: an interim report’, Antiq J, 48, 48 18 – 31 Collis, J R 1970. ‘Excavations at Owslebury, Hants: a second interim report’, Antiq J, 50, 50 246– 61 Collis, J R (ed) 2001. Society and Settlement in Iron Age Europe, Sheffield Cowell, M 1992. ‘An analytical survey of the British Celtic gold coinage’, in Mays (ed) 1992, 207– 33 Creighton, J 2000. Coins and Power in Late Iron Age Britain, Cambridge Cunliffe, B W 1987. Greeks, Romans and Barbarians: Spheres of Interaction, London Cunliffe, B W 1995. Iron Age Britain, London Cunliffe, B W 2000. The Danebury Environs Programme: The Prehistory of a Wessex Landscape, Oxford Eluère, C 1987. ‘Celtic gold torcs’, Gold Bulletin, 20, 20 1ii, 22– 37 Feugère, M 1985. Les Fibules en Gaule Me´ridonale de la Conqueˆte à la Fin du Cinquie`me Sie`cle apre´s J-C, Paris Fitzpatrick, A P 1984. ‘The deposition of La Tène Iron Age metalwork in watery contexts in southern England’, in Aspects of the Iron Age in Central Southern England (eds B W Cunliffe and D Miles), 178 – 90, Oxford Fitzpatrick, A P 1992. ‘The Snettisham, Norfolk, hoards of Iron Age torques: sacred or profane?’, Antiquity, 66, 66 395 – 8 Fitzpatrick, A P 1997. Archaeological Excavations on the Route of the A27 Westhampnett Bypass, West Sussex. Vol 2: The Cemeteries, Salisbury Fitzpatrick, A P 2001. ‘Cross-channel exchange, Hengistbury Head, and the end of hillforts’, in Collis (ed) 2001, 82– 97 Fitzpatrick, A P and Megaw, J V S 1987. ‘Further finds from Le Câtillon Hoard’, Proc Prehist Soc, 53, 53 433 – 44 Fitzpatrick, A P and Morris, E (eds) 1994. The Iron Age in Wessex: Recent Work, Salisbury Fulford, M and Timby, J 2000. Late Iron Age and Roman Silchester: Excavations on the Site of the Forum-Basilica 1977, 1980 – 86, London Gebhard, M 1991. Die Fibeln aus dem Oppidum von Manching, Frankfurt Hartmann, A 1980. ‘Analyses of British prehistoric gold and some British ores’, in Bronze Age Gold Work of the British Isles (ed J J Taylor), 138 –41, Cambridge Haselgrove, C C 1982. ‘Wealth, prestige and power: the dynamics of late Iron Age centralization in south eastern England’, in Ranking, Resource and Exchange (eds A C R Renfrew and S Shennan), 79 – 88, Cambridge Haselgrove, C C 1989. ‘The later Iron Age in southern Britain and beyond’, in Research on Roman Britain 1960 – 89 (ed M Todd), 1– 18, London Haselgrove, C C 1993. ‘The development of British Iron Age coinage’, Numis Chron, 153, 153 31– 64 Haselgrove, C C 1994. ‘Coinage and currency in Wessex’, in Fitzpatrick and Morris (eds) 1994, 22– 5 Haselgrove, C C 1999. ‘The Iron Age’, in The Archaeology of Britain (eds J R Hunter and I B M Ralston), 113 – 34, London Hattatt, R 1987. Brooches of Antiquity, Oxford Hill, J D 1995. ‘The pre-Roman Iron Age in Britain and Ireland: an overview’, J World Prehist, 9=1, 47 – 98 Hill, J D 2002. ‘Not just about the potters wheel: making, using and depositing Middle and Late Iron Age pottery in south-east England’, in Prehistoric Britain: The Ceramic Basis (eds A Woodward and J D Hill), 143– 60, Oxford Hook, D R. ‘Unpublished semi-quantitative analyses of Snettisham hoards G to L for the Treasure process’, The British Museum Hook, D R and Needham, S P 1989. ‘A comparison of recent analyses of British Late Bronze Age gold work with Irish parallels’, Jewellery Studies, 3, 15 – 24 Hunter, F 1997. ‘Iron Age hoarding in Scotland and northern England’, in Reconstructing Iron Age Societies (eds A Gwilt and C C Haselgrove), 108 –33, Oxford King, A C and Soffe, G 1994. ‘The Iron Age and Roman temple on Hayling Island, Hampshire’, in Fitzpatrick and Morris (eds) 1994, 114–16 King, A C and Soffe, G 2001. ‘Internal organization and deposition at the Iron Age temple on 22 THE ANTIQUARIES JOURNAL Hayling Island, Hampshire’, in Collis (ed) 2001, 111– 24 Krämer, W 1971. ‘Silberne Fibelpaare aus dem Letzten vorchristlischen Jahrundert’, Germania, 49, 49 111 –32 La Niece, S and Hill, J D forthcoming. ‘A unique 1st century BC hoard of gold from southern Britain: an interface between the Celtic and classical worlds’, in Actas del I Symposium Internacional sobre Tecnologia del oro Antiguo: Europa y America. Madrid 2002 (eds A Perea, I Montero and O Gracÿ́a-Vuelta) Leake, R C, Bland, D J and Cooper, C 1993. ‘Source characterization of alluvial gold from mineral inclusions and internal compositional variation’, Trans Inst Mining Metallurgy, 102, 102 B65 – B82 Mays, M (ed) 1992. Celtic Coinage: Britain and Beyond, Oxford Northover, J P 1992. ‘Materials issues in the Celtic coinage’, in Mays (ed) 1992, 235 – 99 Ogden, J 1982. Jewellery of the Ancient World, London Perry, B 1982. ‘Excavations at Bramdean, Hampshire 1973 – 1977’, Proc Hampshire Fld Club Archaeol Soc, 38, 38 57– 74 Raftery, B 1984. La Te`ne in Ireland: Problems of Origin and Chronology, Marburg Sharples, N M 1990. ‘Late Iron Age society and continental trade in Dorset’, in Les Gaulois d’Armorique (eds A Duval et al), 299 – 304, Rennes Stanford, S C 1974. Croft Ambrey, Hereford Stead, I M 1991. ‘The Snettisham treasure: excavations in 1990’, Antiquity, 65, 65 447 – 65 Stead, I M 1998. The Salisbury Hoard, Stroud Stone, E 1987. ‘Aspects of later Iron Age metalcraft in southern Britain’, unpublished BA dissertation, University of Oxford Tassinari, S 1993. Il Vasellami Bronzeo di Pompei, Rome Whinney, R 1994. ‘Oram’s Arbour: the Middle Iron Age enclosure at Winchester, Hampshire’, in Fitzpatrick and Morris (eds) 1994, 86– 90