194 : A VIKING-AGE SETTLEMENT IN CO.

7· 5 SILVER

John Sheehan

Introduction

Forty-two items of Viking-Age silver have been found at Woodstown, Co. Waterford, mostly in hack-silver form. Of these, forty were recovered from the site's removed topsoil layers, either through metal detecting or, to a much lesser extent, by dry sieving. Preliminary analysis of the very large number of topsoil finds indicated an extensive disturbance of this material, resulting from ploughing within the site's soil mantle since its use, with little evidence to relate the artefacts to one another or to underlying features (O'Brien et al. 2005, 66-y). The remaining two finds were conventionally excavated, from stratified deposits, but were not associated with each other. The scattered nature of the silver on the site indicates that it was randomly lost, piece by piece, probably during the circulation of currency in the course of trade. The Woodstown collection, therefore, must be regarded as an 'assemblage' rather than a hoard, and as such it represents the first find of a significant number of individual silver items from a Scandinavian settlement site in . Given that only a small area of the site was archaeologically investigated, it is likely that this material represents only a moderate proportion of the original amount of silver that circulated and came to be lost there. The obvious comparanda for the items that make up the Woodstown material, which are predominantly ingots and ingot-derived hack-silver, lie in Ireland's silver hoards. However, the question of whether there actually is a meaningful and direct correspondence between the Woodstown assemblage and the phenomena represented by Ireland's hoards is debatable. Many of the hoards represent wealth, including the display and storage of it in the form of complete ornaments, while the assemblage of hack-silver from this site represents the use of wealth as currency. Discoveries made over the past decade or so in Scandinavia, Britain and, now, Ireland, suggest that, apart from single finds, there are two main distinct classes of contexts for Viking-Age silver - hoards and assemblages from settlement finds. Unfortunately, there is, as yet, no equivalent for the Woodstown silver assemblage from Viking-Age Dublin or, indeed, from elsewhere in Ireland (though the results from the recent commencement of work on the Linn Duachaill longphort, Co. Louth, seem promising in this regard). However, there are comparable assemblages from elsewhere in the Viking world, most notably from the central places at Kaupang, Vestfold, Norway (Hardh 2008), and Uppakra, Skane, Sweden (Hardh 2ooo), as well as from the winter camp established by the Danish 'great army' at Torksey, Lincolnshire, in 872-3 (Blackburn 20 I I). In each of these cases, it appears that the finds represent random losses from the pools of silver that circulated in these settlements. As was the case at Woodstown, most of the silver at these sites occurs in the form of hack-silver and is often highly fragmented, and, significantly, all of the sites have also produced impressive quantities of balance-scale weights. The Woodstown silver collection comprises two complete ingots, eleven hack-silver ingot terminals, thirteen further hack-silver ingot fragments, six hack-silver fragments of arm-rings, with two different forms represented, a hack-silver fragment of sheet, a brooch fragment, possibly hack-silver, a weight, a rod fragment and six pieces of casting waste (fig. 7. 54). In addition, the site has produced an iron rod fragment, striated and with silver inlay. Although this may not be classified as a silver/hack-silver item, it is included here for METAL 195

Figure 7.54 (Above) Selection of silver assemblage recovered from Woodstown. L. of longest complete ingot 60.9mm.

Figure 7.55 (Right) Ingot from metalworking furnace 02E0441 :2367:3. L. 31.9mm (photograph: Studio lab).

convenience. With two exceptions, all of the above were retrieved from topsoil and were therefore out of their primary archaeological contexts. Of the two finds from stratified deposits, the first derived from a post-medieval field bank (O'Brien et al. 2005, 63), clearly not its original context (o2E0441:1999:3 1) , while the second, a complete ingot (fig. 7· 55), derives from a fill within an apparent furnace (o2E0441:2367=3). This furnace, which was predominantly used for iron production, is of interest as preliminary analysis indicated that it also produced evidence of small scale, non-ferrous metalworking, including both of copper alloy and of silver (ch. 6). The finds from this feature include a lead balance-scale weight as well as the silver ingot (ch. 4.1), an association that would not be unexpected in an ingot-production context, as well as sherds of cupels/'heating-trays'. Cupel sherds were also recovered from other areas of the site (ch. 8.rc). It has been suggested that cupels, which were used in the assaying and refining of silver in several other contexts in Viking-Age Scandinavia, as well as in Hiberno- WOODSTOWN: A VIKING-AGE SETTLEMENT IN CO. WATERFORD

Scandinavian Dublin, functioned as an element in the control of the production and use of silver as a means of payment, in the same way as balance-scales and weights did (Soderberg 2004, 122-3). Indeed, Young's analyses of several of the Woodstown cupels demonstrate that they were used in silver assaying, and that some crucibles were used for melting silver. To date, consideration of the Woodstown furnace and its associated material indicates the possibility of on-site silver-smelting, and the possible production of ingots. However, while the silver finds from the site also include a number of pieces of casting waste, this material does not in itself necessarily represent on-site silverworking. This is so because casting waste also occurs in a number of hoards from Britain and Ireland (below), indicating that it formed part of the broad pool of silver that was circulating in the Irish Sea area during the Viking Age. Nevertheless, despite this reservation, silver smithing and artefact production at Woodstown may seem a plausible possibility within the overall archaeological context of the site. If this is the case, then the first archaeologically attested evidence for Scandinavian silverworking in Ireland potentially emerges from Woodstown, even though it is accepted that Viking-Age Dublin was a more important silverworking centre. Each item of silver from the Woodstown assemblage is identified and described below. A discussion of each object-type is presented in terms of its cultural attribution and date. An overview of the material is presented within the context of related assemblages from the Scandinavian world and, finally, the potential connections of the Woodstown silver with Viking-Age hoards from early medieval Munster are explored.

The Woodstown assemblage

Ingots and ingot hack-silver Ingots and ingot-derived hack-silver dominate the Woodstown silver, accounting for twenty­ six of the forty-two silver items from the site. The ingots and related hack-silver are of the standard Viking-Age type- oblong bar-shaped, usually of plano-convex, trapezoidal or sub­ rectangular cross-section, most often with rounded terminals - and, as such, they do not exhibit any diagnostic features that could be of use in defining their period of currency. Ingots form a significant element of Viking-Age hoards throughout the Viking world, and functioned primarily as a simple means of storing bullion. They occur, in either complete or hack-silver form, in at least half of the eighty recorded pre-ADrooo silver hoards from Ireland that contain non-numismatic material. Sixteen of these finds also contain coins and thus their deposition dates may be determined, generally to within a short timespan. All of these are of tenth-century date, though the large number of ingots and ingot-derived hack­ silver in the Cuerdale hoard, Lancashire, deposited c.905-10 and possibly originating in large part in Hiberno-Scandinavian Dublin (Graham-Campbell 1987, 332- 6),9 serve to illustrate that ingots must also have been in circulation in Ireland during the last decades of the ninth century. Other evidence, including the unusual nature of ingots in a small number of hoards from Ireland that are ornamented with cruciform motifs that only find parallel in the Cuerdale find, suggests that ingots were more common in later ninth-century Ireland than the evidence of the coin-dated hoards alone indicates. It is also relevant to note in this regard that ingots or ingot-derived hack-silver occur in association with ornaments in fifteen non-coin dated hoards from Ireland, in most cases with Hiberno-Scandinavian broad-band arm-rings, a type that developed during the later ninth century, perhaps c.88o, and continued in general circulation until c.930 (Sheehan 201 r, 99-100).

9 For an alternative view, see Williams 201 r, 70-r. METAL 197

In short, therefore, the overall evidence indicates that ingots were more common in later ninth-century Ireland than the evidence of the coin-dated hoards alone indicates, and that they formed a regular element of Ireland's tenth-century hoards. On the basis of this evidence, the ingots and ingot-derived hack-silver, which forms the bulk of the Woodstown silver material, could date to anywhere between the mid-ninth and the end of the tenth century. One of the two silver finds from a stratified context at Woodstown- an ingot (fig. 7. 55) -was recovered from a fill within what was identified as a smithing hearth within the ditch of Enclosure One. A radiocarbon date of AD420-620 was obtained from a sample of oak charcoal from one of the hearth's fills, which was stratigraphically lower than the fill from which the ingot was derived (ch. 4.1, app. 1.2). This is an unusually early date for a silver ingot of this form in Ireland, and it may be that the charcoal from which the date was derived was from old wood (app. 1.2). That this was the case is inferred when the artefacts from this ditch are considered, for they include amber and ivory as well as the silver ingot. These are materials that one would not normally expect to find together in Ireland during the fifth to seventh centuries, while they are rather more familiar from Scandinavian and Hiberno-Scandinavian contexts of the Viking Age (Sheehan 2oo8a, 287-8). Some ingot moulds of pre-Viking date or contexts are on record from early medieval Ireland; for instance, from and crannogs such as Lagore, Co. Meath, Garranes, Co. , and Moynagh Lough, Co. Meath (Craddock 1989, 174), but these are not known to be associated with silverworking and it appears more likely that they were used to cast copper­ alloy ingots. Indeed, one of the copper ingots from Garranes fits perfectly into a matrix in one of the ingot moulds from the site (Comber 2004, 3 5). The Woodstown ingot is a distinctive hammered example of almost rectangular section that is closely paralleled in a Viking-Age hoard from Tiree, , the deposition of which is coin-dated to c.970-8o (Graham-Campbell 1995, 97-8, pl. Ja). It would seem prudent, therefore, to set aside the Woodstown radiocarbon date associated with this ingot unless it can be substantiated by further evidence. With regards to the cultural attribution of the Woodstown ingots, it is interesting to note that ingots appear to have been more common in Ireland and Denmark than they were in other parts of the Viking world (Kruse 1993 , 188). This tendency emphasises the important links that existed between the Viking-Age silverworking traditions oflreland and southern Scandinavia (Sheehan 200 I, 5 8-9); the recent discovery of significantly large assemblages of ingots and ingot-derived hack-silver from Kaupang and Uppikra, in present day Norway and Sweden respectively, adds to this impression as both of these sites occur in regions that lay within the area of Danish territorial control during parts of the ninth century. In this context, it is also of interest to find fragments of two Kufic coins among the Woodstown material (ch. 7.8b), as coins of this type are of much more common occurrence in southern Scandinavia than in Norway. The majority of the hoards from Britain and Ireland that contain such coins are from the Irish Sea region, specifically north-west England and the northern part of the modern province ofLeinster, with two-thirds of the total being deposited between c.900 and c.930 (Sheehan 2ooob, 53, fig. 5). Coins of this type may have generally formed part of the same circulation as non-numismatic silver forms, and XRF analyses of ingots from Britain and Ireland have indicated that some were made from recycled Kufic coins as well as from a mixture of other available silver types, both numismatic and otherwise (Kruse and Tate 1992, 323). Given that Viking-Age ingots are generally neither culturally nor regionally diagnostic, it is not possible to state unequivocally that all of the ingots from Viking-Age Ireland, WOODSTOWN: A VIKING-AGE SETTLEMENT IN CO. WATERFORD

including those from Woodstown, are of Hiberno-Scandinavian manufacture. Nevertheless, there is a strong probability that they are, especially given that broad-band arm-rings - the quintessential Hiberno-Scandinavian object-type - were simply made by hammering out ingots. Indeed, diagnostic Hiberno-Scandinavian material forms part of fifteen out of the twenty hoards from Ireland in which ingots are associated with ornaments and/or ornament­ derived hack-silver. It remains a distinct possibility, nevertheless, that some ingots from Ireland were imported. For instance, an unlocalised hoard from Co. Dublin, deposited c.93 5 and consisting of an ingot and two hack-silver fragments of'Permian' rings (Graham-Campbell 1976, 49, pl. 2), may have come direct from southern Scandinavia, through England. Since ingots primarily functioned as a means of storing bullion, it is probable that they, alongside other objects such as broad-band arm-rings, had a role to play in Ireland's Viking­ Age metal-weight economy. The fact that many ingots in hoards from Ireland are represented in hack-silver form may be taken as evidence to support this, as may the occurrence of minor nicks or pecks on them, as on 46 per cent of the twenty-four examples of ingot-derived hack-silver from Woodstown. Silver objects acquired nicks of this type during circulation in commercial transactions, and these represent a characteristic Scandinavian method of assessing silver quality as well as testing for plated forgeries.

02E0441:6oo:337 (fig. 7.56). Ingot. Topsoil, Trench 34· Small oblong bar, sub­ rectangular section, with rounded ends. Uppermost face features several voids and a long scrape, the other face and sides are pitted from the mould. Nicks: 2. L. 6o.9mm. W 7.9mm. Th. 7.7mm. Wt 22.22g.

o2E0441:236T3 (fig. 7· 55). Ingot. Deposit sealing smelting furnace/smithing hearth (F 2 3 3o), Trench 34· Small oblong bar, rectangular section, tapering towards rounded ends. Both faces and sides slightly rough. Nicks: o. L. 3 1.9mm. W 7mm. Th. 4· 7mm. Wt 5.6rg.

o2E0441:r999:3 r. Ingot, terminal fragment. Layer within post-medieval earthen bank (F2468), Trench 34· Cut and broken from bar, sub-oval section, one rounded end. Uppermost somewhat uneven. Nicks: o. L. 14.rmm. W 13.9mm. Th. 6.8mm. Wt 7.rg.

o2E0441:6oo:2200. Ingot, terminal fragment. Topsoil, Trench 34· Broken from bar, plano-convex section, one rounded end. Nicks: 2. L. r6.8mm. W 8.3mm. Th. 5.6mm. Wt 4.68g.

o2E0441:6oo:2565. Ingot, terminal fragment. Topsoil, Trench 28. Broken from bar, trapezoidal section, one rounded end. Nicks: o. L 12.5mm. W 9.6mm. Th. o.8mm. Wt 4.89g.

o2E0441 :6oo: 8 3 3. Ingot, terminal fragment. Topsoil, Trench r 8h. Cut and broken from bar, plano-convex section, one rounded end. Nicks: r. L. 6.4mm. W 9.3mm. Th. 4.9mm. Wt r. 5 3g.

02£0441:600:467 5. Ingot, terminal fragment. Topsoil, Trench rk. Cut in two blows, broken from bar flat, rectangular section. One rounded end, hammered. Nicks: o. L. 13.2mm. W ro.9mm. Th. 3.5mm. Wt 3.92g.

02E0441:6oo:r684. Ingot, terminal fragment. Topsoil, Trench 34· Cut and broken from bar, trapezoidal section, tapering in both width and thickness to a rounded end. The uppermost face features evidence of hammering, other face and sides rough from the mould. Nicks: o. L. 20.4mm. W 8.7mm. Th. 8.2mm. Wt 9.r6g. METAL 199

Figure 7.56 Sil ver ingot 02E044 1 600:337. L. 60.9mm (photograph: Studiolab).

02E0441:6oo:88I. Ingot, terminal fragment. Topsoil, Trench 3 5· Cut and broken from bar, sub-trapezoidal section, tapering in both width and thickness to a rounded end. Each face and side features evidence of hammering. Nicks: 6. L. 2.42mm. W 8.8mm. Th. 7.rmm. Wt 9.36g.

o2E0441:6oo:507. Ingot, terminal fragment. Topsoil, Trench rb. Cut and broken from a bar of plano-convex section, with a rounded end. The uppermost face is pitted and the curved face is rough from the mould. Nicks: o. L. 3.32mm. W 1.2mm. Th. o.62mm. Wt r6.89g.

02E0441:6oo:840. Ingot, terminal fragment. Topsoil, Trench r8g. Cut and broken from bar, sub-oval section, one rounded end hammered. Nicks: 2. L. 6.9mm. W 9.6mm. Th. 6.4mm. Wt 2.38g.

o2E0441 :6oo:21 9 5. Ingot, terminal fragment. Topsoil, Trench 3 5. Cut and broken from bar, sub-oval section, one rounded end. One face, lowermost in mould, hammered. Nicks: o. L. 8.6mm. W 7.6mm. Th. 5.8mm. Wt 2.25g. \. o2E0441:6oo:863. Ingot, terminal fragment. Topsoil, Trench 3 5· Cut and broken from the sub-angular end, and along one side, of a bar that was probably originally of trapezoidal section. Cuts made from opposite faces. Nicks: o. L. 17.rmm. Th. 11.5mm. Wt 8.o7g.

02E0441:6oo:774 (fig. 7.57). Ingot fragment. Topsoil, Trench 35· Cut and broken at both ends, from a bar, sub-oval section, thicker at one side and tapering in both width and thickness from one end to the other; both cuts made from same face. Evidence of hammering on both faces . Transverse cut on one face. Nicks: 7· L. 2o.rmm. W 12.9mm. Th. 8.3mm. Wt 13.3g.

o2E0441 :6oo:41 3 5. Ingot fragment. Topsoil, wetland. Cut and broken at both ends from bar, plano-convex section; both cuts made from the same face. Uppermost face evidence of hammering. Nicks: o. L. 7mm. W ro.5mm. Th. 5. 7mm. Wt 2.25g.

o2E0441 :600:780. Ingot fragment. Topsoil, Trench 3 5. Cut and broken at both ends from bar, plano-convex section; cuts made from opposite faces and both ends hammered following cutting. Small transverse cut on lower face. Nicks: 4· L. 9.3mm. W 14.4mm. Th. 6-4mm. Wt 5.8g.

02E0441 :6oo: 500. Ingot fragment. Topsoil, Trench r 5. Cut and broken at both ends, and along one side, from bar probably originally of sub-oval section. Nicks: o. L. 6. 5mm. Th. 5.6mm. Wt 1.94g.

02E0441:6oo:I978. Ingot fragment. Topsoil, Trench 3 5. Cut and broken at both ends from bar sub-triangular section, thicker along one side; cuts made from same face. Nicks: r. L. 96mm. W ro.3mm. Th. 5.6mm. Wt 2.69g. 200 WOODSTOWN: A VIKING-AGE SETTLEMENT IN CO . WAT ERFORD

Figure 7.57 Silver ingot fragment 02E044 1:600:774 . L. 20.1mm (photograph: Studio lab).

o2E044I:6oo:I442. Ingot fragment. Topsoil, Trench 25. Cut and broken at both ends, in one case with two angled blows from opposite faces, from bar of plano-convex section. Uppermost face slightly pitted. Nicks: o. L. 9.8mm. W 9-4mm. Th. 5.9mm. Wt 3·35g.

o2E044I:6oo:25 59 · Ingot fragment. Topsoil, Trench 3 5. Cut and broken at both ends, in one case with two blows, from bar of sub-trapezoidal section. Ends hammered following the cutting and the lowermost face hammered flat. Nicks: 1. L. Io.6mm. W I1.5mm. Th. 9.2mm. Wt 6.35g.

o2E044I :600:2220. Ingot fragment. Topsoil, Trench 3 5. Broken at both ends, from a bar, sub-square section. Both faces and sides are very rough. Nicks: o. L. I 3 .2mm. W 8.2mm. Th. 6.8mm. Wt 3.27g.

o2E044I:6oo:7 53· Ingot fragment. Topsoil, Trench 35 · Cut and broken at both ends from opposite faces, in one case with two blows, from bar of sub-triangular section. One end hammered following the cutting. Nicks: o. L. I5.2mm. W 8.2mm. Th. 9.3mm. Wt 6.66g.

o2E044I:6oo:821. Ingot fragment. Topsoil, Trench Ii. Cut and broken at both ends from a tapering bar of plano-convex section; cuts made from the same face. Nicks: r . L. IO.Imm. W ro.6mm. Th. 7.3mm. Wt 4.44g.

o2E044I:6oo:506. Ingot fragment. Topsoil, Trench I b. Cut and broken at both ends from a bar of plano-convex section; cuts made from opposite faces, one angled. Evidence of hammering on convex surface. Nicks: o. L. 7.5mm. W I5.3mm. Th. 7·5mm. Wt 4.67g.

o2E044I :6oo: 39 3. Ingot fragment. Topsoil, Trench I 1. Cut and broken at both ends, and along one side, from bar that was probably originally of sub-trapezoidal section. Cuts made from opposite directions. Nicks: o. L. I3.5mm. Th. 8-4mm. Wt 5.92g.

E3434:5:2733. Ingot fragment. Topsoil, Field 21. Cut and broken at both ends, from a bar of sub-trapezoidal section; cuts made from opposite faces. Nicks: 1. L. I9.5mm. T. 7·9mm. Wt J3.Ig.

Broad-band arm-ring fragments Silver arm-rings were by far the most common products of Ireland's Hiberno-Scandinavian silverworking tradition, which was at its height between c.850 and c.950. Several classes have been identified, but the most important of these in numerical terms is the Hiberno- M ETAL 201

Figure 7.58 Arm-ring fragment 02E0441 :600: 347. L. 8.2m m (photograph: Studiol ab).

Scandinavian broad-band type (Graham-Campbell 1976, 51-3; Sheehan 201 r). Rings of this type are normally penannular and were made from a flat band of silver that tapered in width from the mid-point towards the terminals; though plain examples occur, these rings were usually decorated with punched ornament and frequently feature rows of transversely disposed grooves, often with a diagonal-cross motif on the expanded central area of the arm­ ring. The small arm-ring fragment with stamped decoration from Woodstown (o2E0441: 6oo: 34 7; fig. 7. 58) is derived from an ornamented example of this Hiberno-Scandinavian broad-band type. In terms of its decoration, it is somewhat atypical in that it is ornamented with a bar-stamp bearing a lattice-type pattern, a form that is by no means as common as the stamps with serrated edges. On the Woodstown fragment, the lattice-type design is formed from a row of diagonal crosses, positioned adjacent to each other so that they form lozenge shapes between the crosses. Examples of this type of punching are also present on arm-rings in the Emy hoard, Co. Monaghan, and in the Hiberno-Scandinavian hoards from Cuerdale and Huxley, north-west England (Sheehan 2009, 6 5, fig. 7 -4). Both of the latter finds were deposited during the early decades of the tenth century, and the Cuerdale hoard indicates the broad types of silver objects that circulated in Ireland during the closing decades of the ninth century. It is worth noting that broadly similar stamps were used to ornament arm-rings in the hoards from Athgarret, Co. Kildare, Ballyaghan (Cave Hill), Co. Antrim, and north-west lnishowen, Co. Donegal. The three plain broad-band fragments from Woodstown are probably also derived from rings of this type (02E0441:6oo-46I, o2E0441:6o0.2900, 02E3434=5=2973). Plain broad­ band arm-rings are not uncommon in Ireland's Viking-Age silver hoards, given that they are represented in at least six of the twenty-nine hoards in which broad-band arm-rings are present. Although none is on record from coin-dated deposits in Ireland, the presence of a substantial number in the Cuerdale hoard indicates that plain arm-rings of this type were most probably already familiar in Ireland during the closing decades of the ninth century. Recently, two hack-silver fragments have been excavated from the longphort at Linn

Duachaill, Co. Louth. ro Although it may seem tempting to dismiss plain broad-band arm­ rings as no more than simple rings within the Hiberno-Scandinavian tradition, it is interesting to note that similar plain examples occur in ninth-century hoards from southern Scandinavia. Broad-band arm-rings may be dated using a number of approaches, the most useful of which is their occurrence in coin-dated hoards. Arm-rings of this type occur as part of sixteen coin-dated hoards, of which only three are from Ireland. Some finds provide clear evidence for their occurrence within culturally Danish contexts of the later ninth century (Brooks and Graham-Campbell 1986, 97-8), and it is evident that the southern Scandinavian prototypes for the classic form had developed by the 87os (Sheehan 2011,99- 100). It seems likely, therefore, that the development began in the 85os or 86os and that the type, in its Scandinavian manifestation, was a short-lived development confined to the

IO I am grateful-to Eamonn P. Kelly, NMI, fo r showing me these items. 202 WOODSTOWN: A VIKING-AGE SETTLEMENT IN CO. WAT ERFORD second half of the ninth century. Broad-band arm-rings do not appear in coin-dated deposits again until the first decade of the tenth century, when they form part of four hoards. It is of note that three of these finds are from Ireland, while the fourth, from Cuerdale, displays strong Hiberno-Scandinavian connections. The deposition dates of the hoards, all within a few years of the apparent collapse of the Hiberno-Scandinavian longphort settlement of Dublin in 902, suggest that these finds represent some of the dispersed wealth of later nimh-century Dublin. Most of the arm-rings in these finds belong to the developed Hiberno-Scandinavian series, indicating that its formulation had begun there during the closing decades of that century, possibly c.88o.

Table 7.7 Coin-dated hoards containing the developed form of Hiberno-Scandinavian broad-band arm-ring in Ireland and Britain.

PROVENANCE DATE

Silverdale, Lancashire C.900-IO Cuerdale, Lancashire c.905-10 ----,------Dysart Island no. 4, Co. Westmeath C.9IO - Co. Antrim Magheralagan, Co. Down c.910? ------Goldsborough, Yorkshire C.925 - - - Bangor, Gwynedd C.925 - - --- Warton, Lancashire C.925? ------Vale ofYork, Yorkshire C.927-9

While all coin-dated depositions of the Hiberno-Scandinavian series of broad-band arm­ rings from Britain and Ireland fall within the period 900-30 (Table 7.7), the development of the type probably began in Ireland c.88o. There is little doubt that it continued to be produced beyond, and possibly during, the apparent hiatus ofViking Dublin from 902 to 91 7, as reflected in its coin-dated occurrences. However, it does not occur in any of the post- 930 coin-dated hoards containing non-numismatic silver from Britain or Ireland, indicating that- by then- such arm-rings had gone out of fashion. The majority of broad-band arm­ rings appear, therefore, to have been produced during the fifty-year period between c.88o and c.930, though some prototypes for the classic form had developed during thepreceding decades (Sheehan 20II, 99-100). Therefore, the Woodstown rings should be assigned to within the broad date-range of c.85o-c.930. Broad-band arm-rings, although they could serve as ornaments and status objects, appear to have been manufactured primarily for the storage and circulation of silver as a form of currency in Ireland's metal-weight economy (Sheehan 2004, I 8 3). An examination of the weights of a sample of complete rings suggested a target weight-range for them of c.25.9±0.4g to c.27.3±0.9g, and quantum analysis on this material suggested a target of 26.1 5±0.9g, which lies close to Bmgger's postulated Scandinavian eyrir of 26-4g (1921, 95). Kruse has, however, pointed our that problems may arise from the use of quantum analysis if multiple-weight standards are present in the material (1988, 293). Nevertheless, the c. 26.1 5g result was recently tested again using the data from an increased total of complete arm-rings of this type, and it was found that the weights of almost two-thirds of the rings conformed generally to the proposed unit and its multiples, with the greatest proportion conforming to the range of the 2-unit value (Sheehan 2009, 67, fig. 7.6). This result is of METAL 203 particular interest given that the proposed silver weight-unit differs only marginally, by less than half a gram, from the dominant unit identified by Wallace among the lead scale­ weights from tenth-century Dublin (1987, 206-7). Some broad-band rings, however, amounting to about a third of the sample tested, were clearly not produced to conform to this target value. It is possible that these examples were not primarily intended to function in the Hiberno-Scandinavian metal-weight economy and may have been produced specifically to serve as status gifts for recipients outside this milieu. I I Broad-band arm-rings are generally underrepresented in terms of their occurrence in Munster, and right across the southern third oflreland, where they are more likely to occur in hack-silver form than as complete rings (Sheehan 1998b, 153). It is perhaps not surprising, therefore, that relatively few examples of this type of object are present in the Woodstown assemblage. As with ingots, the nature of a metal-weight economy necessitated the reduction of ornaments to hack-silver, and examples of broad-band arm-rings in hack­ silver form occur in twenty hoards from Ireland, as well as in the Woodstown assemblage. Silver arm-rings might acquire minor nicks and pecks during commercial transactions, and such nicks occurs on two of the Woodstown broad-band arm-ring fragments. In overall terms, the occurrence of the broad-band hack-silver at Woodstown is of importance for two reasons. Firstly, it indicates, independently of the evidence of the ingots, another date-range derived from the silver for the occupation of the site, c.8 50-c.930. This does not, of course, necessarily imply that Woodstown was not also occupied before and/or after this period. Secondly, it indicates that there is a small but recognisable Hiberno­ Scandinavian element in the Viking-Age silver from the site.

o2E0441:6oo:347 (fig. 7.58). Arm-ring. Topsoil, Trench IJa. Fragment, ofHiberno­ Scandinavian broad-band type, with punched ornament; cut at both ends. One face bears a single vertical transverse bar-stamp, the punch having a somewhat haphazard lattice pattern. Nicks: o. L. 8.2mm. H. 19.3mm Th. 2.6mm. Wt 2.96g.

o2E044I:6oo:46r. Arm-ring. Topsoil, Trench 4· Fragment, plain, probably of a Hiberno­ Scandinavian broad-band arm-ring; cut at both of its ends and sides. Nicks: o. L. 13.7mm. H. r2.3mm. Th. 4.rmm. Wt 5·53g.

o2E0441:6oo:2900. Arm-ring. Topsoil, Trench 3 5. Terminal fragment, plain, cut and broken from a probable Hiberno-Scandinavian broad-band arm-ring; the terminal is of angular form. Nicks: 2. L. 23-4mm. H. ro.rmm. Th. 3.7mm. Wt 7-4g.

£3434=5=2973. Arm-ring. Topsoil, Field 21. Fragment, plain, probably of a Hiberno­ Scandinavian broad-band arm-ring; cut at both of its ends and sides. Nicks: o. L. 9.7mm. H. 5.6mm. Th. r.6mm. Wt 0.7rg.

Lozenge-sectioned rod fragments The two rod fragments of lozenge-shaped section from Woodstown are almost certainly derived from arm-rings, with one of them actually retaining its curvature (fig. 7. 59). The most characteristic arm-ring type formed from such rods in Viking-Age Britain and Ireland is 'ring-money'. This is the term used to describe a specific type of silver ring that is particularly characteristic of Scotland's Viking-Age hoards, within which over ninety complete examples have been found (Graham-Campbell 1995, 57-9). Developed rings of this type are penannular, with the hoops usually being made from single rods of lozenge-

11 I am grateful ro Dr Garerh Williams, BM, for discussion on rhis poinr. 204 WOODSTOWN: A VIKING-AGE SETTLEMENT IN CO. WATERFORD shaped cross-section; the terminals are frequently spatulate and the rings are usually plain. The angles of the hoop are frequently modified by flattening or rounding, as on both of the Woodstown fragments. The occurrence of developed 'ring-money' in coin-dated hoards from Scandinavian Scotland indicates that it was being manufactured and used there during the period between the mid-tenth and the mid-eleventh century (Graham-Campbell and Sheehan 2007, 53 6-7). These rings are sometimes interpreted as a form of currency, and there is some metrological evidence to support the notion that they were produced to an approximate weight standard (Warner I976, 136-43). While the Woodstown rod fragments appear to derive from developed 'ring-money', 12 there is a small number of other arm-ring types from Britain and Ireland that are also formed from lozenge-sectioned rods. However, these are frequently of heavier form, sometimes feature stamped decoration, occasionally are of annular form, and, on the basis of their occasional hoard-associations with broad-band arm-rings, appear to be earlier than developed 'ring-money'. An example of this type occurs, for instance, in an unlocalised hoard from Co. Galway (B0e I940, II2, fig. 74), accompanied by Hiberno-Scandinavian broad-band arm-rings, and several examples also occur in the Cuerdale hoard (Graham­ Campbell 20I I, I03-4). A fragment of a lighter arm-ring, formed with a rod of the same dimensions as the Woodstown fragments, bur ornamented by stamping, forms part of the hoard from Carraig Aille II, Co. (6 Riordiin I949, fig. 7), which was probably deposited during the late ninth or early tenth century. These types of rings appear to have formed the background for the development of 'ring-money' and, given the overlaps between the types, there is potential for uncertainty in assigning cultural attribution and date to fragments of plain lozenge-sectioned rods. The classification here of the Woodstown rod fragments as derived from 'ring-money' must be regarded as provisional. Finds of this developed Scotto-Scandinavian type occur in relatively small numbers in Ireland, where they are known from only four hoards: those from the , Co. Clare; Corran Island crannog (Lough Sewdy), Co. --' Westmeath; Knockmaon, Co. Waterford; and a recently discovered find from Lurgabrack, Co. Donegal. ' 3 These are all hoards that contain no other types of non-numismatic artefacts, but with the addition of coins in two cases. The deposition of the Knockmaon rings hoard has been dated to c.rooo (Dolley I966, 57; Blackburn and Pagan 1986, 297), while the Corrao Island rings were reportedly found with a coin of Eadmund (British Numismatic journal 58 (I988), I 54, no. r 59). Knockmaon is near Dungarvan, an area known to have been controlled by the Hiberno-Scandinavians ofWaterford (Bradley I988, 65) and, given that the River Shannon hoard is known to have been found somewhere between Limerick and Killaloe, it may also be regarded as deriving from an area of Hiberno-Scandinavian influence (Sheehan 2oooa, 37-8). The Scandinavian material recently discovered in Dunmore Cave, Co. Kilkenny, which was apparently deposited c.965 or soon afterwards (Bornholdt Collins 2010, 19), includes one example of possible 'ring-money'. The possible presence of 'ring-money' among the Woodstown silver assemblage potentially extends the period of use of the site into the first half of the eleventh century.

o2E0441:6oo:4390 (fig. 7·59, top). Arm-ring. Topsoil, Trench rg. Rod fragment, cut with two blows at one end and three at the other, from opposite sides, from a rod of lozenge-shaped section with dimensions of 4· 3 by 5.6mm. The fragment retains a slight curvature and its lower angle has been pared. Nicks: 2. L. c. 22mm. Wt 3. 53 g.

12 The author is grateful to James Graham-Campbell for his views on these Woodstown fragments. 13 The author is grateful to Eamonn P. Kelly and Maeve Sikora, NMI, for informing him of this discovery. METAL 205

Figure 7.59 Arm-ring fragments 02E044 1: 600:4390, 2576 (photograph: Hugh Kavanagh) .

0 1cm

o2E0441:6oo:2576 (fig. 7·59, bottom). Arm-ring. Topsoil, Trench 3 5· Rod fragment, cut at one end and cut-and-broken at the other, from a rod of lozenge-shaped section with dimensions of 5. 3 by 5.4mm. The fragment is straight and two of its opposite angles have been pared. Nicks: o. L. 30-4mm. Wt 5.02g.

Scale weight Among the material from Woodstown is a small scale weight (600=41 16), of polyhedral form (fig. 7.6o). This appears to be of silver, though this has not been verified by XRF analysis, and it is possible that, like a weight of broadly similar silvery appearance from Kaupang (Pedersen 2008, fig. 6.21), it is of copper alloy. Weights of this form and size ultimately originated in the Islamic world and are of common occurrence in Scandinavia during the second half of the ninth century, particularly in its Baltic region (Pedersen 2008, 136). The Woodstown example weighs 2. 5g, and it is interesting to note that this is one of the target weights identified by Kyhlberg in her analysis of a series of small polygonal weights from Birka, Sweden (1980, 261). Figure 7.60 Scale weight of polyhedral form Weights of polygonal type are absent from the large assemblage of Viking-Age weights 02E044 1: 600:4 1 16. from Dublin (Wallace 1987, 212), though a number are on record from England. For Max. L. 8.6m m. example, at least sixty copper-alloy polyhedral weights, some with ornamentation, are among the finds from the Viking winter camp atTorksey, Lincolnshire, associated with the 872-3 campaign by the Danish 'great army' (Blackburn 2011, 236-8, fig. 9a-c). What is unusual about the Woodstown weight is that it may be of silver, and as such it is without parallel among the polyhedral weights in the Viking world. It is interesting to note, however, that the only other possible silver weights on record are two conical examples from the immediate vicinity of the probable longphort at Athlunkard, Co. Limerick (Kelly and O'Donovan 1998, 14). The Woodstown weight is diagnostically Scandinavian and most probably dates to the late ninth century.

o2E0441:6oo:41I6 (fig. 7.60). Scale weight. Topsoil, wetland. Polyhedral form, with fourteen faces, apparently unornamented. Nicks: o. Dims 8.6 by 8.3 by 7.7mm. Wt2.5Ig.

Sheet E3434:5:2133 . Sheet. Topsoil, Field 21. Small fragment of a thin, flat sheet, almost square. Unornamented. Nicks: o. L. 9.9mm. W 9-4mm. Th. 0.5mm. Wt 0.38g. 206 WOODSTOWN: A VIKING-AGE SETTLEMENT IN CO. WATERFORD

Rod fragment Small fragments of thin silver rods, such as the example from Woodstown, occur occasionally in Ireland's Viking-Age hoards. They frequently show evidence of torsion, indicating that they have been cut from plaited-rod or twisted-rod ornaments, most likely arm-rings and neck-rings. Rod ornaments of these types become a Scandinavian fashion from the mid-ninth century onwards (Graham-Campbell 2006, 74-8), and many twisted rod fragments are present in the Cuerdale hoard (Graham-Campbell2oii, 212-13, pl. 37). The only examples of such fragments that occur in coin-dated hoard contexts in Ireland are those from Dysart Island no. 4, Co. Westmeath, and Magheralagan, Co. Down, the depositions of which are both dated to c. 9 ro. A single example also occurs in the coinless hoard from Kilmacomma, Co. Waterford (Sheehan 2oo8b, 286-7).

o2Eoo44I:6oo:4I05. Rod. Topsoil, wetland. Fragment of a thin rod, of sub-rectangular section (I. 7 by I. 2mm), showing evidence of torsion; cut at an angle at one end, the other end being flattened. Nicks: o. L. c.1 4-4mm. Wt o.r7g.

Casting waste Three amorphous lumps or melts (fig. 7.6ra), two minute nodules/droplets and a 'bean-shaped ingot' combine to represent silver casting waste from Woodstown. This interpretation is partly based on the results of a series of research experimental castings of silver ingots ofViking-Age type undertaken by Kruse, Smith and Starling some years ago. These experiments indicate that the small round droplets and 'bean-shaped ingots' were most probably formed unintentionally as by-products during the casting of ingots (Kruse et al. 1988, 90-r). They should not, therefore, be viewed as ingots in the formal sense, a~ d they are without metrological or classificatory significance. Likewise, the three amorphous lumps or melts of silver from Woodstown must be interpreted simply as casting waste, produced when molten silver cooled. The presence of these six items could be interpreted as evidence for silver-smelting at the site, an activity otherwise evidenced there by a furnace that produced evidence of small-scale, non-ferrous metalworking, including silver, and which, interestingly, also produced a silver ingot and a lead scale weight (see chs 4.1, 7.6).' 4 Casting waste in the form of melts or lumps are on record from the assemblages from Kaupang and Torksey, but they also occur in a small number of hoards from Britain and Ireland, such as those from Cuerdale (Kruse and Graham-Campbell2orr, 84-5, pl. 17), deposited c.905-IO, and Rivory, Co. Cavan, probably deposited during the first half of the tenth century. Small droplets and 'bean-shaped ingots' are also known from the Cuerdale find, as well as from three coin-dated hoards in Ireland: Dysart Island no. 4, Co. Westmeath, deposited c.910, Mohil no. 2 (Dunmore Cave, Co. Kilkenny), deposited c.970, and Ladestown, Co. Westmeath, which has a terminus post quem deposition date of 9 8 5. The faces of the Mohil droplet were flattened by hammering, in the same manner as an example from Kaupang (Hirdh 2008, fig. 5.9). Two 'bean-shaped ingots' also form part of the coinless hoard from Ballywillin Crannog, Lough Kinale, Co. Longford, probably of early tenth-century date. It is clear from this range of dates that this type of material, though rather unusual, was circulating in Ireland and the Irish Sea area throughout the later ninth and tenth centuries. Its presence in hoards indicates that some waste silver, at least, formed part of this broad pool of silver and that they were presumably used to top-up weights in economic transactions. On the other hand, the fact that none of the six Woodstown pieces

14 This ingot is ozE0441.2367·3, while the registration number of the weight is not noted in the published references to this find (O'Brien et al. 2005, 45 , 63 , 81 ). M ETAL 207

Figur·e 7.6 1a Th ree pieces of casting waste 02E044 1: 600:593. L. 25mm (/eft) , 862. L. 25.3mm (centre), 2768. L. I8.4 mm (right) (photogra ph: Hugh Kavanagh).

bear nicks or pecks, while almost half of its ingot-derived hack-silver does, suggests that they may have functioned within different spheres. Nonetheless, the Woodstown casting waste material, given that it is derived from topsoil and is without an archaeological context, cannot be definitively interpreted as evidence for on-site silver smelting (contra O'Brien et al. 2005 , 77), even though this may seem a plausible possibility in the overall archaeological context of the site.

o2E0441:6oo:4264. Nodule/droplet or 'bun-shaped ingot', probably casting waste. Topsoil, Trench 35- Hemispherical section. Nicks: o. Dims ro.7mm by 9.1mm. Wt 3-43g-

o2E0441:6oo:4309. Nodule/droplet or 'bun-shaped ingot', probably casting waste. Topsoil, Trench 34- Sub-spheroid form. Nicks: o. Dims 7-9 by 5.9mm. Wt r.38g.

02£0441:600:43 89 . 'Bean-shaped ingot', probably casting waste. Topsoil, Trench 3 5. Sub­ ovoid section, rounded ends. Nicks: o. L. 8.9mm. W 4.6mm. Th. 5.1mm. Wt r.o5g.

o2E0441:6oo:593 (fig. 7.61 , left). Casting waste. Topsoil, Trench rg. Irregular shape. Rough and uneven on one surface. Nicks: o. Dims 25mm x 16.6mm. Th. 4.9mm. Wt 5-72g.

02£0441 :6oo:862 (fig. 7.61, centre). Casting waste. Topsoil, Trench 3 5. Irregular shape. Rough and uneven on one surface. Nicks: o. Dims 25.3 by 24.4mm. Th. 6. 7mm. Wt 12.3g.

o2E0441:6oo:2768 (fig. 7.61, right). Casting waste. Topsoil, Trench 34- Irregular shape. Rough and uneven on both surfaces. Nicks: o. Dims 18.4 by q .1mm. Th. 6.8mm. Wt 5.21g.

Striated rod fragment, with silver inlay T he surface of this fragment of a straight iron rod, of circular section, features circumferential striations that are inlaid with silver. 15 It derives from a larger object, being cut at both ends, though the nature of this item is uncertain. While no clear parallels for the style and form of this decorative rod are immediately evident within the corpora of Viking­ Age metalwork, the technique of inlaying silver wires on iron is represented on a small number of Viking-Age artefacts. These tend to be high-quality pieces, such as the sword hilt from Steinsvik, L0dingen, Nordland, Norway, and the Mammen axehead, from Jutland, Denmark (Graham-Campbell 1980, 69, nos 245, 493), and silver-inlaid sword hilts are also

15 T he silver was identified as inlaid by Susannah Kelly, archaeological conservator, School of Archaeology, UCD . 208 WOODSTOWN: A VIKING-AGE SETTLEMENT IN CO. WATERFORD

on record from early Viking-Age burials in Dublin (Walsh I988, 229). A small number of iron pins and rods with spirally twisted shanks are on record from Anglo-Scandinavian York, at Coppergate (Ottaway I992, 634, fig. 265, 69 3-6, fig . 300), from where a pin with a grooved shank is plated with a tin- lead alloy (ibid., 696, pl. xlvi:a). In terms of its striation, the Woodstown rod is somewhat reminiscent of the form of rods used to manufacture Permian rings, silver neck-rings that were produced in the Permian region I in east during the early Viking Age. From there, examples were imported into southern ~ Scandinavia, particularly the Baltic islands and Denmark, where they tended to be wound into the form of spiral arm-rings, and local copies, of lighter form, were produced (Hardh 2008, 108-I3). In Britain and Ireland, where such spiral-rings are of rare occurrence, they occur in hack-silver form (Graham-Campbell 20I I, 88-90). The Woodstown piece, however, differs from these in that it is of silver-inlaid iron, as well as having striations of straight rather than spiralling form. It is straight, and shows no sign of formerly having curvature. It could be interpreted as deriving from a decorative iron pin or, perhaps, from another type of object that was influenced in form by the silver Permian rings. Copies ofViking-Age silver arm-rings in other metals are of unusual occurrence, and are usually in gold or copper alloy, though a penannular broad-band arm-ring in iron, ornamented with transversely disposed, punched, plain grooves, is on record from S0rhus, Vats, Skjold, Rogaland (Petersen I928, I 56, fig. I90). Unfortunately, however, on present evidence, it is not possible to identifY what type of object the Woodstown fragment is derived from. '6

Figure 7.6 1b Iron rod fragment with inlaid silver striations 02E0441 :600:519. L. 29.9mm (photograph: 0 2cm John Su nderland).

o2E044I:6oo:5I9 (fig. 7. 6Ib). Rod fragment. Topsoil, Trench 6o. Fragment, cut from an iron rod of circular section with dimensions of 7.I by 7. 6mm. Cleanly cut at one end and obliquely cut-and-broken at other, with both cuts from same side. Fragment is straight, and its surface features straight circumferential striations inlaid with silver. Nicks: 3· L. 29.9mm. Wt 7.9g.

Brooch fragment Among the finds from Woodstown is a terminal from a cast and gilt silver penannular brooch (2E004I :6oo: I66I) oflnsular type (see ch. 7-4 for catalogue and discussion). It is circular and featured a central stud set in a band of chip-carved interlace. It may have been intended to utilise this as a cap for a lead weight, though silver was not normally used for this purpose. It is also possible that it was perceived as a piece of hack-silver; certainly it fits comfortably within the weight-range of hack-silver from the site, and brooch fragments are occasionally represented in hoards containing hack-silver from Ireland. Due to a lack of certainty concerning its purpose, it is not included as hack-silver in this section. Discussion The Woodstown silver assemblage of forty-two items comprises two complete ingots, thirty­ two pieces of hack-silver, predominantly derived from ingots, a brooch fragment, six pieces of casting waste and one weight. The vast majority of items were recovered from the site's

I6 I am grareful ro James Graham-Campbell for discussing this obj ecr with me. METAL 209 topsoil mantle and there is little evidence to relate them to one another or to underlying archaeological features. Consequently, the results of the archaeological investigations conducted at the site are of limited assistance in dating and interpreting this material. One find was excavated from a contemporary stratified context- a furnace- which indicates the possibility of silver-smelting on the site. There is nothing to indicate that the silver derives from a hoard or hoards. Rather, the scattered distribution of the individual finds on the site appears to indicate that they represent random losses from a pool of silver that circulated there. In this respect, the assemblage is similar to those from the Scandinavian settlements at Birka, Kaupang and Uppikra, as well as the camp of the Danish 'great army' at Torksey, Lincolnshire, occupied in 872-3. The silver assemblages from these sites are also analogous to the Woodstown one in terms of the preponderance within them of highly fragmented hack-silver. On the basis of the combined dating evidence for the various components of the Woodstown silver assemblage, as outlined above, which is largely based on parallel material that occurs in hoards from Britain and Ireland, it appears that it contains elements dating from as early as the ninth century to, potentially, as late as the mid-eleventh century. However, comparing this type of silver with material from hoards is problematic and must be approached with caution, essentially because of the differing nature of the two bodies of evidence (Hardh zooS, I I 5). The Woodstown hack-silver presumably represents lost and mislaid single items, some possibly manufactured on site, that were in use as currency and in active circulation on a regular basis over the period of use of the settlement. The hoards differ from this type of manifestation in that they are 'sealed' collections, being deposited at particular points in time. Their components were not necessarily drawn solely from the same pool of silver that circulated for economic purposes, though hoards of hack-silver should be viewed in this manner, following Graham-Campbell who noted that 'silver bullion will have been so rendered for commercial purposes and not for reasons of status' (I989, 55). Many hoards served as accumulated stores of wealth and/or currency to which the holder, over a period of time, may have added to and/or withdrawn from, while some may even have been deposited for non-economic reasons, without the intention of recovery (Graham-Campbell and Sheehan 2009). In some instances, they may have been deposited for the long term, while in others they may have served as shorter-term caches of current currency. Despite these reservations, arising from the potentially distinct character of the two types of data, it appears reasonable to consider that there may be some degree of correlation between the Woodstown silver and the hoard evidence from its broad geographic locality in Ireland. Given that the assemblage is dominated by hack-silver, those hoards that feature a significant proportion of hack-silver, particularly if it is highly fragmented, deserve special consideration in this respect. Hack-silver hoards are relatively rare in Ireland. Of the eighty recorded hoards that contain non-numismatic silver, only a dozen consist exclusively of hack-silver. Most of these contain ingot-derived material, sometimes in association with ornament-derived hack-silver, and half of them also feature coins in their composition; each of the latter hoards was deposited during the tenth century (Table 7.8). It is noteworthy that not all hoards that contain hack-silver may be classified as 'true' hack-silver hoards in the sense in which this phenomenon is understood in Scandinavia. Hardh, for instance, has defined the latter as finds 'where half or more of the objects are fragments, and where most of the objects weigh less than five grams' (I 9 9 6, 3 3), a phenomenon that she feels 'is best explained as a collection of currency' (zooS, 99). Under these terms, many of the hoards with hack-silver from Ireland fall outside of her definition, mainly due to the limitation imposed by the weight stipulation. However, if the Woodstown assemblage constituted a hoard - which is 210 WOODSTOWN: A VIKING-AGE SETTLEMENT IN CO. WAT ERFORD

evidently not the case - it would easily qualify as a Scandinavian-type hack-silver deposit given the high degree of silver fragmentation represented in the assemblage. In this respect, the Woodstown silver is much more closely related to the assemblages from the Scandinavian trading settlements at Kaupang, Birka and Uppakra, as well as the 'camp' at Torksey, than it is to the majority of the hoards from Ireland. Thus, among the more important aspects of the Woodstown assemblage is that, for the first time in Ireland, a significantly large collection of hack-silver has been found in a demonstrably Scandinavian or Hiberno-Scandinavian settlement site. The presence of the hack-silver at Woodstown, alongside a large number of scale weights (see ch. 7.6), clearly indicates that this longphort had an important and developed economic function.

Table 7.8 Hoards consisting exclusively of hack-sil ver; w ith or w ithout coin s, from Ireland.

PROVENANCE INGOT HACK-SILVER ORNAMENT COIN-DATED H ACK-SILVER C~ r, Co:_D ~negal _ -_···_· ~~~~------•--· ···------· ---- Ireland_no. 3, une :. ovenan ~~?__ _ • Allardstown, Co. Louth • ------··· Knockmaon, Co. Waterford • c. 1000 Nure(~~liput ,_S:.~ :.. Westt?~at_h ______• Mohil, Co. Kilkenny :....___ • Kilmacomma, Co. Waterford ·-- - --·-· . • • Dysart Glebe, Co. Kilkenny • • --~---·--·-- ·-····· --- ·····------Millockstown, Co. Louth • TPQ9o5/6 ------· - ---·- Ath~~~~et, Co. !9lda~e --- ... -• - - --··· ---· Co. Antrim C.910 - •-· --····------•------· Nr Dublin c.88o-973 ------·------• - ·· - --

Much of the hack-silver from Kaupang, Birka and Uppakra, as from Woodstown, derives from plough-zones and, thus, is only datable to the general periods of use of these sites. At Kaupang, however, there are some pieces from secure stratified contexts that are dated to the decades leading up to c.85o (Pedersen and Pil0 2007, 186; Hardh 2008, IJ4). 17 This is of importance in that it provides evidence for the use of silver as currency in this particular Scandinavian context from the mid-ninth century onwards, indicating an earlier development of the hack-silver phenomenon than is evidenced by many of the coin-dated Scandinavian silver hoards alone. In Ireland and the Irish Sea region, on the sole basis of the coin-dated hoard evidence, the phenomenon of 'true' hack-silver appears to date from the early tenth century onwards. However, on the basis of the evidence of two of the coin-dated hoards that contain large amounts of hack-silver, it seems likely that this may also have been a feature of at least the later ninth century in Ireland. These finds - from Dysart Island no. 4, Co. Westmeath (Ryan et al. 1984, 339-56, pls 6-12), and Cuerdale (Graham­ Campbell 201 1), a Lancashire hoard with strong Hiberno-Scandinavian connections- were both deposited during the first decade of the tenth century but may have been assembled, at least in part, earlier than this. It is of interest to note that the high degree of fragmentation in the Dysart Island no. 4 hoard is best paralleled in Ireland by the Woodstown assemblage. The apparent absence from Ireland of ninth-century hoards containing hack-silver may simply be a reflection of the fact that there are no ninth-century 'mixed' hoards - coin

I7 These contexts belong to SP II sub-phase 2 of the site. METAL 2II hoards that also contain a non-numismatic element- on record from there. This is perhaps not surprising when it is considered that Ireland has produced only four coin hoards of this date, while there are over fifty tenth-century coin and mixed hoards on record. The conclusion is that coinless hack-silver hoards from Ireland, which, on the basis of the parallel coin-dated hoard evidence alone, might appear likely to be of tenth-century date, could equally have been deposited during the second half of the ninth century. Indeed, on the basis of recent developments in the study and dating of silver usage in southern Scandinavian, specifically at Kaupang and Uppakra, it may be the case that the potential date of the first appearance of hack-silver in Viking-Age Ireland should be extended backwards to as early as the mid-ninth century. Even though the silver from Woodstown cannot be independently dated to this period, having a broad collective date-range from the ninth century to the mid-eleventh century, it seems quite likely, for two reasons, that it dates to the earlier part of this timespan. Firstly, on the basis of the totality of the archaeological evidence from Woodstown, it seems that most of the occupation and related activities on the site took place during the ninth century. Secondly, it has been previously suggested on the basis of hoard evidence that the origins of the diagnostic components of the Hiberno­ Scandinavian silverworking tradition are closely related to ninth-century southern Scandinavia (Sheehan 200I, 58-9), and, as has been noted above, it now emerges that the closest parallels for the nature of the Woodstown assemblage also lie within this region, specifically at Kaupang and Uppakra, both of which developed in the ninth century.

Table 7.9 Viking-Age silver hoards ofthe ninth- or tenth-century date from the province - Kingdom of Munster:

PROVENANCE HOARD TYPE COIN- DATED DEPOSITION ---·------Cullen, Co. Tipperary coinless --- -'---- Rathbarry, Co. Cork coin --- 'Co. Cork' no. I coin less ·-· ---:----:- Scattery Island, Co. Clare coin less ·-·-- ···----· 'Co. Kilkenny' coinless -----,--- , Co. Cork com C.95 3 Mungret, Co. Limerick mixed 'Co. Tipperary' com -- ···--~------Kilbarry, Co. Cork coinless ·········--·-- ·······---- - 'N r Limerick', Co. Limerick coinless ---- Castlelohort, Co. Cork coinless --- -,----- Lackadufi, Co. Cork coinless ------·······-·-··--·- 'Co. Cork' no. 2 coinless ··-- - Derrynahinch, Co. Kilkenny coinless ------Shannon bank, Co. Clare coinless - ·······------·······----- ········--- Brennan's Glen, Co. Kerry coinless Knockmaon, Co. Waterford mixed c.Iooo ···------······------Rathmooley, Co. Tipperary coinless - - ·-·-· _.:.____ Carraig Aille II, Co. Limerick coinless --- Mohil no. I, Co. Kilkenny mixed ------Kilmacomma, Co. Waterford coinless - .... ······-··--- -··-·-·---- ·-·--- - · Dysart Glebe, Co. Kilkenny coinless _____::.___ Mohil no. 2, Co. Kilkenny mixed Cloghermore, Co. Kerry coinless -'---- 212 WOODSTOWN: A VIKING-AGE SETTLEMENT IN CO. WATERFORD

The question of how the Woodstown silver assemblage may relate to other finds of Viking­ Age silver from its broad geographical and political region is of some significance. Longphoirt and potentiallongphoirt tend to be positioned on political boundaries, suggesting that this is not a coincidental phenomenon and that the Scandinavians developed a strategy of taking advantage of economic potentials as well as political rivalries (Kelly and Maas I999, I40; Sheehan 2oo8a, 286). Woodstown is positioned on the , which formed the border between the kingdoms of Deisi Murnan and , and on the periphery of the main province-kingdom of Munster, known as Mumu. A general comparison between the silver from Woodstown and that represented by the Viking-Age hoards from the Munster province-kingdom is worth undertaking, as it may serve to highlight both potential connections and differences between the two bodies of evidence, and thereby better indicate the distinctive nature of the Woodstown assemblage. The province-kingdom of Munster was, at times, a region of shifting boundaries ruled by the E6ganacht from its royal centre at Cashel. For the purposes of this discussion, however, it is accepted that the broad extent and limits of Munster around AD900 are sufficiently well reflected in MacCotter's work on Ireland's early medieval territorial, political and economic divisions (2008, I25-254, app. 2), which shows that it then included the kingdoms and peoples of Osraige, L6igis and Ui Failge, now in the modern province of , in addition to the area covered by the modern province of Munster. No fewer than 7 twenty-four Viking-Age silver hoards of ninth- or tenth-century date are on record from this province-kingdom (Table 7.9, fig. 7.62), which hereinafter for convenience is simply referred to as Munster. The greatest numbers of these are coinless hoards, comprising seventeen examples, four are 'mixed', in which coins occur with non-numismatic silver, and three are coin hoards. ' 8 Each of the mixed and coin hoards was deposited during the tenth century, with dates ranging from c.930 to c.rooo. Five of these appear to have been composed of fewer than twenty coins, and only three contained coins in addition to Anglo-Saxon issues. One of these exceptions, the small mixed hoard from Mohil no. I (Dunmore Cave), Co. Kilkenny, deposited c.930 (Bornholdt Collins 20IO, I9), is of interest in the Woodstown context given the discovery of two Kufic coin fragments there (ch. 7.8b), in that this hoard represents the only other occurrence of Kufic coins in Munster. In purely bullion terms, however, it is the non-numismatic material in the Munster hoards, amounting to a collective minimum total of I I 9 items, which accounts for the great bulk of the silver wealth represented in this province-kingdom during the ninth and tenth centuries. This impressive amount of Viking-Age silver from Munster pales somewhat in significance when it is compared with the quantity on record from the province-kingdom of Southern Ui Neill (Purcell and Sheehan 20I3, 37-9, fig. 3.2), which occupies much of the northern part of the modern province of Leinster, but it nonetheless remains considerable when compared to the appreciably lesser amounts known from Ireland's other province-kingdoms. Therefore, Munster, as the immediate political context for Woodstown, is of interest in that it was a province-kingdom of some wealth in the ninth and tenth centuries. Its hoards do not form a coherent group, varying as they do in size, composition and date and in the presence or absence of coins and hack-silver, as well as the degree of fragmentation, when present. Some of these hoards appear to be socially motivated, representing alliances or gift-exchange, some may signifY tribute, while others are clearly economically inspired and represent the use of silver as currency (Sheehan 2004, I 8 I -8). In general terms,

I8 It has been demonstrated recently that the mixed hoard, formerly regarded as being from the west of Kilkenny (Dolley 19 66, 32- 3), should in fact be provenanced ro Kilkenny West, near Athlone, Co. Westmeath (Sheehan 20 10). It is therefore excluded from consideration here. M ETAL 213

Coinless hoard •t1l Mixed hoard Coin hoard \ 0 Dysart Glet~ 0 Single find Mohll2)

Rathmooley .Cullen • Derrynahinch• ,0~~ ~Kllmacomma \O~' WOODSTOWN

SOkm Figure 7.62 D istribution of hoards and single finds of

Hoards provenanced co county only ore omitted Viking-Age si lver in early (1 xc::JTipperary, 1 • • Kilkenny, 2 xe cork) medieval Munster: however, few, if any, of the Munster finds may be related in any way to the longphort at Woodstown -with the possible exception of a small number of hoards, to be discussed below - and the likelihood should be borne in mind that many of these hoards may relate to other temporal, social and economic contexts in the region, including its tenth-century Hiberno-Scandinavian towns, and their hinterlands, as well as its Irish centres of power. Despite these provisos, it is a worthwhile exercise to recognise the potential of the twenty-one hoards from Munster that contain non-numismatic silver, to reflect on the extent to which they share the general trends and features apparent in the Woodstown silver assemblage, and to consider their distribution and date-range. ' 9 The assemblage is rather distinctive, in a number of respects, from the extant and otherwise identifiable non­ numismatic items represented in the hoards. For instance, when the general weight-ranges of all non-numismatic items from the hoards and from Woodstown are plotted together (fig. 7.63),20 it emerges that 36 per cent of the total is 10g or less. However, while only 17 per cent of the hoard material falls into this category, some 90 per cent of the Woodstown silver does. The hoards tend to contain complete ornaments or ornament-derived hack-silver, which occurs in nineteen of the twenty-one finds, representing no less than 79 per cent of the extant or identifiable items in the hoards. Ornament-derived hack-silver, on the other hand, accounts for only I 8 per cent of the Woodstown material, which features no complete ornaments, while within the hoards slightly over half of the ornament material is represented by complete examples, for the most part arm-rings of various types. Ingots or ingot derived hack-silver, on the other hand, dominate the Woodstown material, accounting for some 6 5 per cent of it, while this amounts to only 21 per cent of the components of the Munster

19 In this section, the brooch fragment and silver-inlaid striated iron rod from Woodstown are omitted. Likewise, the scale weight is omitted from any calculations related to the nature or features of the silver from Woodstown, on the grounds that it differs from them in that it is unlikely to have served a bullion function. The casting waste is included, however, as it may have served such a function. 20 In cases where objects are lost or non-extant, the figures used to compile this table are estimates based on information contai ned in the relevant antiquarian sources and/or on the average weights of specific object types represented in Ireland's hoards. 214 WOODSTOWN: A VIKING-AGE SETTLEMENT IN CO. WATERFORD

Figure 7.63 40 Comparative • Munster distribution by • Woodstown weight of all non­ 35 numismatic items from Woodstown 30 and hoards in Munster: -~ 25 ""'0 iii 20 ..c ~ I5

IO

0

"'":" "'":" "' "'":" "' "' "' "' "' "'":" "' "'":" "' "'":" "'0-, "'0-, 0-,"' 0-,"' "'0-, "'0-, "'0-, "'0-, "'0-, "' N '>a ['.. 00 0 N 6 "' "'I "'I "' "' "' 0 6 "'0 "' "' 6 r 6 6['.. 00 ~ N "' .... £ '>a b b b 6 "' 8 ~ ~ S' "' Weight range (g)

hoards, indicating again a difference in composition between the two bodies of data and the pervasiveness of hack-silver at Woodstown. Casting waste, whether it be regarded as evidence for on-site silverworking or for its potential use as a form of bullion in a weight

economy, accounts for I 5 per cent of the Woodstown silver, while it accounts for less than

I per cent of the material in the Munster hoards. Clearly, there are important distinctions between the silver represented in the Woodstown assemblage and the Munster hoards. Hack-silver is composed of pieces of silver that are deliberately cut and broken to be used as a means of payment in a weight-based economic system. The amount of the Woodstown silver that occurs in this form, derived from both ingots and ornaments, is 8o per cent of the full assemblage, while hack-silver represents 52 per cent of the items recorded in the Munster hoards. If, however, casting waste pieces are designated as an equivalent of hack-silver, in the sense that they could be used alongside it as small-scale components in bullion transactions, then the hack-silver figures for Woodstown and the Munster hoards rise to 9 5 per cent and 53 per cent respectively. Either way, a clear and pronounced distinction emerges between the relative prevalence of hack-silver in the Woodstown assemblage and in the hoards. The differentiation between the two bodies of evidence extends well beyond the value of these figures, however, for much of the hack-silver from the hoards is composed of rather large individual pieces while the Woodstown material is highly fragmented (fig. 7.63); the average weight of the Woodstown hack-silver is 4.8g, while the weight of the extant and identifiable hack-silver items in the Munster hoards averages over twice this. Clearly, the nature of hack-silver represented at Woodstown is different from that evident in the overall body of Munster hoards. The importance of weight arises from the fact that the Viking Age had a silver metal­ weight economy for much of its duration, with economic and commercial transactions being based on, or related to, this medium. In Ireland, this form of economy appears to have functioned until at least the mid-tenth century, when imported coins began to be retained. The formal transformation to a coined-money economy took place at the end of that century, when Dublin commenced minting. The actual weight systems underlying the metal-weight economies varied somewhat from one area of the Viking world to another, and may be identified, with varying degrees of accuracy, from the study of balance-scale weights (ch. 7.6). Unfortunately, no systematic programme of metrological analysis has been METAL 215

Figure 7.64 Distribut ion by weight of hack-silver from Woodstown. 4 ~------4----wr--~------~------4 e .~ 3 ...... 0 :3 ~ z~2 r-~----~~,.--~~--~r-.------.------~

0 "' "' "' "'a:- "' "' "' "' "' "'a:- "'a:- "' "'ci "' "'... "' '

2 Figure 7.64. ' They vary between o. 17g and 16.89g, with a significant proportion of them, almost Go per cent, being less than 5g. There is, however, little evident correlation between the values present in the weight distribution and the conventional sub-divisions of the Viking-Age ore. Values equating to the ortugar, for instance, one-third of the ore, lie between 8g and 9g, a range where there is only one, rather marginal, hit. It should be noted, however, that this is a rather small collection of data and it would be surprising if statistically significant tendencies became evident within it. Several interesting trends emerge, however, when the Woodstown weights are plotted against those from the Kaupang and Uppakra assemblages, which, it must be noted, are significantly larger (fig. 7.65). By far the greater proportion of silver from each assemblage is composed of pieces less than 1og. However, in the cases of the two assemblages from Scandinavia, the proportion of pieces that weigh under 5g is overwhelming, between So per cent and 90 per cent in both cases, while the equivalent figure from the Woodstown assemblage is 6o per cent. A more significant

2I The two complete ingots, the scale weight the brooch fragment and the six items of casting waste are excluded from this exercise. 216 WOODSTOWN: A VIKING-AGE SETTLEMENT IN CO . WATERFORD

Figu re 7.65 Comparative distribution by weight of 40 ,------~------, hack-silver from • Woodstown Woodstown, Kaupang 35 ~------~ • Uppiikra and Uppakra. • Kaupang 30 ~~~------L------~

s 2) ~~~~------1 -~ '"""...0 20 .cOJ 8 I) i !0

0 "' "'":' "' "'":' "' "' "' "' "' "' "' "' "' "' "' "'":' "' "' "'":' "' "'ci "' "'.,;. "'.;., \ci"' "' oO"' "'d., "'ci 5' "' "' "'.,;. \ci"' "' 00 "'d., .)., I I I 6 I ,!,"' r: .!r J, " 00 "' "" I I " I I " "' ~ ~ ~ b :r & '8 b ~ 8:' Weight range (g)

difference emerges when the proportion of the very small pieces, less than 2g in weight, is considered, for at both Kaupang and Uppakra this accounts for the majority of the assemblages, in excess of 50 per cent, while only r 5 per cent of the Woodstown hack-silver falls into this category. Is it possible that such highly fragmented hack-silver, which dominates the Kaupang and Uppakra assemblages, was of such lesser significance at Woodstown? This seems unlikely, especially given that the proportion of pieces within this weight-range in Ireland's greatest hack-silver find, Dysart Island no. 4, which one might expect to be low or even non-existent given that it is a hoard rather than an assemblage of random losses, is actually double the figure evidenced at Woodstown. It is possible that this important difference between the dominant weight-ranges evident at Woodstown and the two Scandinavian central-places - and indeed in the Dysart Island no. 4 hoard, which presumably derives from Dublin and is therefore indicative of the type of hack-silver in use there- is simply a product of the different types of archaeological investigations conducted at these sites. It is possible, if not likely, that the most highly fragmented hack-silver, which characterises the Kaupang and Uppakra assemblages, and which is also present at Woodstown, but in much smaller proportions, is actually underrepresented in the silver assemblage recovered from the site. Nicking is represented in at least fifteen of the twenty-one Munster hoards that contain non-numismatic silver, where it occurs on 58 per cent of the extant items. In the Woodstown assemblage, it occurs on 3 3 per cent of the finds, predominantly on ingot­ derived hack-silver. In the hoard material, it also occurs much more often on ingots/ingot­ derived hack-silver than on the ornaments/ornament-derived hack-silver that dominate these finds, and it is more likely to feature on hack-silver than on complete objects. The number of nicks present on an item, whether it derives from the hoard material or the Woodstown assemblage, varies from one to eight, but in over 75 per cent of cases in both instances the number of nicks present is either only one or two. On the basis that the dispersed nature and the high degree of fragmentation of silver found on trading places and related sites indicate a well-established economy in which it was regularly used as currency, perhaps even in small transactions of quotidian character, this type of silver should only be compared to that from the silver hoards with a degree of caution, as has been noted above. Bearing these reservations in mind, it is remarkable that METAL 217

Figure 7.66 Distribution of hoards and single fi nds of Viking-Age silver in Osraige and Deisi Muman .

..l.· ··.. \ ·."· ······!osrai ··: ... :····-...... ,.... :' ,_:·e. ··. :•••M ·..: ..············ :. .. . ) ·:... .!

• Coinless hoard

(%) Mixed hoard

N 0 Singlefind 0 SOkm each of the three hoards from early Munster that are undoubtedly of highly fragmented character and, additionally, meet Hardh's stipulations for 'true' hack-silver finds, are from the eastern side of the province-kingdom, where Woodstown is also situated, and that all of them derive from within the river valleys of the 'Three Sisters' - the Suir, the Nore and the Barrow- that flow into , the longphort's geographical setting. These are the finds from Mohil (no. r), Dysart Glebe and Kilmacomma. In addition, the lost find from Derrynahinch, Co. Kilkenny, also appears likely to have been of this type, while the second hoard from Mohil (no. 2) comes close to meeting Hardh's stipulations. Members of the first group of three hoards have a consistent pattern and a close correspondence with the Woodstown assemblage in terms of their composition, as well as in the fragmentation and weight of their hack-silver. This is suggestive of a relationship between both phenomena and it seems reasonable to suppose, therefore, that the hoards may ultimately derive from Woodstown. The two hoards in the second group, though less definitive in this respect, should not be omitted from consideration. Four of these five hoards are located in the kingdom of Osraige, while the fifth lies in Deisi Murnan (fig. 7.66). The Kilmacomma hoard, from Deisi Murnan, is a most significant find. It was found on a partially levelled that immediately overlooks the River Suir, at a location some 30km upriver from Woodstown. Directly across the river is the ecclesiastical site oflnishlounaght, from where, interestingly, a single fragment of hack-silver has also been recovered (Sheehan 1992, 211, fig. 3) . The hoard comprises twelve items, all in hack-silver form, with seven derived from ingots, two from rods of circular section, two from broad-band arm-rings and one from a thin rod with evidence of torsion (fig. 7.67) . Interestingly, three of these four hack-silver types, the exception being the circular-sectioned 218 WOODSTOWN: A VIKING-AGE SETTLEMENT IN CO. WATERFORD

Figure 7.67 Si lver hoard recovered at Ki lmacomma, Co. Waterford (photograph: John Sheehan).

0---= ===2c m rod fragments, are also represented among the silver finds from Woodstown, where together they form the bulk of the assemblage. Eight of the Kilmacomma pieces weigh less than 5g, with most of these weighing less than rg, and three of them feature nicks. Clearly, in every respect, this hoard is very closely paralleled in the Woodstown assemblage, from where it may have ultimately derived in its entirety (Sheehan 2oo8b, 245-8). The Kilmacomma find belongs to a small group of Ireland's silver hoards in which the components consist exclusively of hack-silver derived from both ingots and ornaments (Sheehan 20oob, 61-2). There are only three other hoards of this type on record, one of which is from Dysart Glebe (Hall 1993; fig. 7.68). This is an ecclesiastical site of early medieval date located on the banks of the Nore, about 6okm north of Woodstown. The small hoard comprises only two items, one derived from an ingot and the other from a plain broad-band arm-ring and, as is the case with Kilmacomma's components, both items are also represented among the Woodstown silver. Both weigh less than 5g, and each bears a nick, potentially linking them further with the Woodstown assemblage. The Mohil no. r find, from close to Dysart Glebe, was found when geologists conducted a small excavation in the floor levels of Dunmore Cave (Drew and Huddart 1980, 17). It is of a type related to the Kilmacomma and Dysart Glebe finds, though it is a 'mixed' hoard, consisting of an ingot fragment, nicked and less than 5g, accompanied by ten coins or coin fragments. These comprised Anglo-Saxon issues, coins of the of East Anglia and Northumbria, two Kufic coins (one fragmentary) and one unidentifiable coin fragment. In his study of the numismatic element of the hoard, Dolley placed its deposition at c.928 (1975, 78), though Bornholdt Collins has recently proposed that this be adjusted to c.930 (2010, 19).lt is of particular interest given it has a Kufic content, as two Kufic coin fragments also form part of the Woodstown assemblage. One of the Mohil coins, an Abbasid dirham of Caliph Al-Mu'tamid, of the Arminiyah mint, was issued between 879 and 892, and thus it was potentially in circulation for up to fifty years before it was deposited. It is tempting to postulate that the hoard's ingot hack-silver item and this coin derived from the same pool of silver, perhaps ultimately ofWoodstown origin, before being joined by the other coin types, neither of which is represented at Woodstown, before deposition. Mohil no. 2 was discovered in 1999 in a rock crevice in Dunmore Cave's Town Hall Chamber, a separate chamber from the location of the first hoard. The hoard, along with METAL 219

Figure 7.68 Silver hoard recovered from D ysart Glebe, Co. Kilkenny (photograph: John Sheehan).

oo....--===:i1cm

Figure 7.69 The silver hoard recovered at Mohil 2, Co. Kilkenny (photograph: John Sheehan).

2cm

two copper-alloy ingots, appears to have been contained within a folded garment of high quality. It comprised a complete arm-ring, a looped rod and five pieces of hack-silver (fig. 7.69). One of these is a brooch pin fragment, while three pieces derived from ingots. The fifth piece is a small 'droplet', usually interpreted as casting waste, the faces of which were flattened by hammering in the same manner as an example from Kaupang (Hardh 2008, fig. 5.9). It is worth noting that 'droplets' are not on record from any other Munster hoard, their only other occurrence in the province being in the Woodstown assemblage. The coins from the find comprise over a dozen Anglo-Saxon examples, on the basis of which Bornholdt Collins dates its deposition 'to c.965 or soon after' (2010, 19). The Derrynahinch hoard, only one component of which is known to survive, was discovered in 1851 in the valley between the Nore and the Suir, about 25km north of Woodstown (Graves 1852, 355-6; Graham-Campbell 1976, 68). It comprised 'about a quart full of rings and pieces of silver', including twelve or fourteen rings, some of which were of'bullion-ring' type (Sheehan 2012, 395-6), and others, possibly of twisted-rod form, as well as a large quantity of hack-silver described by Graves as 'many flat pieces of silver, some square, about the size of a shilling'. Its bullion weight is unknown, but it was clearly a substantial find and it was undoubtedly one oflreland's largest hack-silver hoards. It seems probable, based on the limited antiquarian evidence available regarding its contents and on the occurrence of 'bullion-rings' within it, that this hoard dates to somewhere between the mid-ninth and the mid-tenth century. 220 WOODSTOWN: A VIKING-AGE SETTLEMENT IN CO. WATERFORD

These five hoards are somewhat atypical in the Munster context, where over half of the other sixteen finds that feature non-numismatic silver are composed of complete ornaments and are thus more likely to reflect the social rather than the economic dimensions of silver. These include two finds from the same political contexts - Osraige and Deisi Murnan - as the five hoards being considered here, those from an unlocalised spot in Co. Kilkenny (Graham-Campbell I976, 67) and from Rathmooley, Co. Tipperary (Sheehan 1992, 2I1- I 5), respectively, both of which contain only rings of various types. Most of the remaining Munster hoards tend to feature a combination of complete objects and hack-silver, but in nearly all cases the latter is without high fragmentation and, consequently, they lack the character of both 'true' hack-silver hoards and the Woodstown silver. Nevertheless, as hoards containing hack-silver, these should be viewed as economic, as silver is reduced to this form only for commercial purposes. It is remarkable that each of the five hoards discussed above is from one of the river valleys that flow into Waterford Harbour, suggesting a link with the Scandinavian presence there. Three of them, those from Mohil (no. I), Dysart Glebe and Kilmacomma, have a consistent pattern and their hack-silver shares a close correspondence in terms of composition, fragmentation, nicking and weight with the silver from Woodstown, from where they may have ultimately derived. The relationship between the Mohil no. 2 droplet and Woodstown is of particular interest but, even though the degree of fragmentation of hack-silver in this Mohil find lies within the weight-range of the Woodstown silver, it is not as closely comparable to it as the Kilmacomma material is. Not enough is on record about the Derrynahinch find to arrive at any firm conclusions about it, but the fact that it was the biggest of these five finds, as well as geographically the closest to Woodstown, may be of significance. It is a matter of some interest that the most obvious parallels for the Woodstown material from throughout Munster lie in these five silver hoards. All are based in the broad geographic and political locality of the longphort, with four of them in the kingdom of Osraige and the fifth in the adjacent polity. While it is the case that together these hoards appear to form a coherent group within Munster, linked in the main by the fact that their components seem to have been largely drawn from the same type of silver that circulated for economic purposes, including at Woodstown, the question of whether there actually is a meaningful correspondence between each of them and Woodstown, or subsequent Scandinavian settlement in Waterford, remains debatable. To some extent, this relates to the dates of the hoards, two of which, both from Mohil, have approximate deposition determinations of c. 9 3 o and c. 96 5. While it is accepted that these determinations are appropriate, they do not necessarily date the periods of currency of all the materials represented in the finds. For instance, as has already been noted, one of the Kufic coins in the Mohil no. I hoard was in circulation, somewhere between Armenia and Kilkenny, for a considerable time, potentially for up to fifty years, before it was deposited in Dunmore Cave. Likewise, it is likely that the Kufic coin fragments from Woodstown circulated for at least a century prior to their loss (see ch. 7.8b). It is very difficult to assess, therefore, how long non-numismatic silver objects and hack-silver existed, and what form(s) they took, prior to their final deposition, and thus the material in the two coin-dated Mohil hoards, apart from the particular coins that provide the determinations, may be somewhat earlier than the termini post quem of the hoards themselves. The issue of the non-numismatic hoards from this part of Munster is even more problematic, as no clear termini post quem for their depositions are evidenced by the finds themselves. Nonetheless, it seems reasonable to suggest that the materials in these five hoards circulated and were accumulated during an METAL 221 elastic century that commenced around 8 5o, or perhaps slightly earlier, and that, therefore, the hoards were deposited at some points during this period. It has been argued above that the five hoards may be linked, with varying degrees of assurance, with the longphort at Woodstown and/or the later Scandinavian presence at Waterford Harbour. The fact that four of these hoards were deposited in the kingdom of Osraige suggests that the economic hinterland of Waterford focused here, and that the Osraige developed a particular relationship with the Scandinavian settlement. Moreover, the character of the hoards, distinguished by their highly fragmented hack-silver, indicates that this relationship was more economic in nature than that which prevailed in some other kingdoms, where hoards are more often of a type that seems best interpreted in social terms (Sheehan 2004, r8 1-4; Purcell and Sheehan 2013, 39-41). A similar trend is evident from the hoard evidence from the province-kingdom of Southern Ui Neill, where the Clann Cholmiin developed a strong relationship with Scandinavian Dublin in the ninth and tenth centuries. This relationship seems to have excluded neighbouring kingdoms, and the paucity of silver hoards in south Leinster, adjoining Osraige, suggests that the province-kingdom of was consciously excluded from sharing in the economic benefits of Scandinavian Waterford. This leads to the question, given that the hoards evidence exchange: what do the historical sources inform us concerning the relationship between the Osraige and the Scandinavians? The kingdom of Osraige formed the easternmost part of Munster until 8 59, when it submitted to Southern Ui Neill and achieved a degree of independence from Mumu. Its leader at this time was Cearbhall mac Dunlainge, a king of some ambition who ruled for over forty years, from 842 until 8 8 8 (Downham 2004). He was doubly linked to the Clann Cholmiin of Southern Ui Neill, through his sister's marriage to Miel Sechnaill mac Miel Ruanaid, high king from 846 to 862, and through his own marriage to Miel Sechnaill's daughter. Miel Sechnaill's successor in the high kingship from 879 to 916 was , Cearbhall's nephew. Thus, throughout his long reign as king of Osraige, Cearbhall was politically well connected through both marriage and blood to the Southern Ui Neill, a relationship that undoubtedly benefited Osraige. In Byrne's view, by the time of his later years he had achieved the status of the second most powerful king in Ireland (1973, 266), gaining a measure of independence from the Munster province-kingdom. Possibly influenced by his Clann Cholmiin connections, he may have adopted intricate political and economic strategies in terms of his interaction with the Scandinavians. He both fought against and allied with them, depending on circumstances, skilfully playing off one group of Scandinavians against another. He became a significant player in Irish and Scandinavian affairs during the middle and later years of the ninth century and, although he started his career as a subordinate to the over-kings of Munster, he developed his role into a position of some influence. In the late 8 50s, with help from Scandinavian allies, including Olifr and fvarr, he inflicted a defeat on other Scandinavian forces in Munster. In 862, he defeated Rauoulfr and destroyed his longphort on the Barrow at Dunrally (Kelly and Maas, r 3 2-4 3). The Scandinavian activities in Osraige during this period reflected, to an extent, the rivalries between Scandinavian Dublin, Limerick and Waterford, as well as the strategic importance of the area of Waterford Harbour within this conflict, which inevitably drew the Osraige into the situation. His descendant, Donnchad mac Gilla Patriic, became king in 1003 and commissioned a work, partly preserved in the Fragmentary Annals of Ireland, in which Cearbhall's dealings with the Scandinavians are eulogised and glorified (Radner 1978). 6 Corriin has suggested that Cearbhall's appearance in Icelandic literary sources stems from 222 WOODSTOWN: A VIKING-AGE SETTLEMENT IN CO. WATERFORD the popularity of the Fragmentary Annals among the Hiberno-Scandinavians, from whom knowledge of him was transmitted to Iceland, granting him, under the name Kjarvalr frakonungr, the status of an important ancestral figure in Landndmabok (1998, 444). Clearly, Cearbhall mac Dunlainge was a king who would have appreciated the potential benefits of having Woodstown on the doorstep of his kingdom. Conclusions The Woodstown assemblage represents the first find of a significant number of silver items from a non-hoard context in a demonstrably Scandinavian settlement site in Ireland. It appears to represent many random losses from the pool of silver that circulated in the settlement, which clearly had important and developed economic functions. The best parallels from comparable assemblages include those from the central places at Kaupang and Uppakra, as well as the winter camp of the 'great army' at Torksey. As is the case at Woodstown, much of the relevant material at these sites occurs in highly fragmented hack­ silver form within the same general contexts from which quantities of balance-scale weights also derive. The results of the archaeological investigations conducted at Woodstown are of only limited assistance in dating its silver assemblage. However, while elements of it could conceivably date from as early as the ninth century to the mid-eleventh century, a case has been made that it is more likely to date early within this timespan, quite possibly around the mid-ninth century. This argument is largely based on recent developments in the study and dating of silver usage in southern Scandinavia, as well as on the totality of the archaeological evidence from Woodstown itself The Woodstown silver appears to be related to a number of other finds ofViking-Age silver from its broad geographical and political region in early medieval Munster, particularly from within the kingdom of Osraige. It is reasonable to suppose that these hoards ultimately derive, in some way, from Woodstown, suggesting that the economic hinterland of Waterford may have been largely focused on this region. It is suggested that this is related to the ambitions of Cearbhall mac Dunlainge, king of Osraige, who may have been influenced by the achievements of the Clann Cholmain of Southern Ui Neill in terms of its interaction with the Scandinavians of Dublin.