In September 1993 John Bridgewater, Gordon Jones, and Graham Skidmore, metal-detectorists, discovered an Iron Age coin hoard (7 gold and 976 silver or silver-plated coins), which was promptly reported. Archaeological excavation of a small area was carried out by the Hereford and Worcester County Archaeological Service (now Worcestershire Archive & Archaeology Service – WAAS) in late September 1993, in order to establish the immediate archaeological context of the hoard (HER ref WSM 20060; Hurst et al. Reference Hurst, Napthan and Pearson1993), and to provide information for the coroner for use at an inquest to determine the status of the coins with regard to Treasure Trove legislation. A second concentration of coins (comprising up to 5 gold and 290 silver or silver-plated coins) was discovered during this excavation. Although similar in date to those from the first hoard, the composition of this group was markedly different. At the time of discovery the combined assemblage represented one of the largest caches of Iron Age coins ever found in Britain (Williams Reference Williams1993a, 13). The finds of gold and silver from the site were declared to be Treasure Trove in May 1994 and were subsequently purchased by the British Museum. A limited campaign of fieldwork continued after 1994 with the purpose of establishing the extents of the archaeological site and of clearing it of any further hoards, and this was completed by 1999 (Hurst Reference Hurst2000).
The site archive is held by the Worcestershire County Museum, apart from the finds which were declared as Treasure Trove and are now in the British Museum.
Location, Topography and Geology
The site is located to the west of the farmhouse of Allesborough Farm, just north-west of Pershore (SO 93230 46437; Fig. 1) in Worcestershire, and the original hoard(s) find-spot was situated on the north side of a gentle south-west facing hill-slope at just over 35 m od where the slope is slightly indented, and below the crest of the hill (at c. 55 m od) to the east. From here there is a sweeping view westwards across the Severn valley to the Malvern Hills ridge 16 km away. The solid geology is Jurassic (Lias clay) with localised river terrace deposits (5th River Avon terrace) of gravel and gravelly sand in the vicinity (see below).
Fig. 1 Location of site with areas of 1993–1999 fieldwork indicated
Archaeological Investigation
by Derek Hurst
No previous archaeological work had been undertaken in the more immediate vicinity of the site. A late Iron Age to mid-Roman settlement presence has recently been identified in nearby Pershore itself (Hughes & Vaughan Reference Hughes and Vaughan2009). Faint traces of ridge and furrow (aligned NNW to SSE; Payne Reference Payne1999, fig. 5) are indicative of arable agriculture in the medieval period, and the area of the site remains in arable production today.
Iron Age coin finds from Worcestershire are mainly individual finds from about 30 sites, and data from the Portable Antiquities Scheme (PAS) indicate a number of Iron Age coins from within c. 1.5 km of the Allesborough site, with at least some accompanied by Roman finds. This suggests a Late Iron Age to Roman settlement sequence (A. Bolton, pers. comm.; see below, and also Appendix 3). The nearest substantial caches of Iron Age artefacts to Pershore are both hoards of iron ‘currency’ bars: from the Littletons (WSM 2835) 15 km to the east; and from the Malvern Hills a similar distance to the west (WSM 3744).
A 12 m2 trench was excavated by hand in September 1993 with funding from Hereford & Worcester County Council, and this was centred on the original coin hoard find-spot, all metal finds being located three-dimensionally (Figs 1–2). All deposits, including the ploughsoil, were carefully excavated by hand with a view to maximising finds recovery. The clayey nature of the soil prevented sieving of soil and, instead, the spoil was checked for metal artefacts by metal-detector.
Following on from discovery of the first hoard by the original finders a systematic, intensive metal-detector scan was carried out in September 1993 as part of an English Heritage-funded project to facilitate management of the site. This covered an area of 1 ha around the excavated trench (Fig. 1). Only non-ferrous objects were retrieved and this resulted in the recovery of further coins especially to the south of the hoard, and of other metalwork (eg, Fig. 2). It was considered important to assess whether further hoards were present as these could have been placed at risk in the circumstances.
Fig. 2 Results of fieldwalking (see Fig. 1 for explanation of fieldwork areas)
Specialist metal-detector survey by A. Pacitto was also carried out in 1997–8 using a Garrett deep-seeking ‘Bloodhound’ kit designed to detect larger targets. As this also detected ferrous targets these had to be constantly eliminated by either digging (within the ploughsoil), or using a fluxgate gradiometer (for deeper readings). Selected areas were also searched using a standard 12.5-inch (31.75 cm) coil. An area of about 6 ha centred on the site of the original hoard was searched in this way (WSM 29125; Fig. 1), and no further hoards were found.
Fieldwalking was carried out during 1994 using 20 × 20 m grid squares over an area of 16 ha (WSM 29124; Fig. 2). There was a wide scatter of mainly prehistoric and Roman finds across the survey area, with Roman material being the most common (Buteux Reference Buteux2000).
Geophysical survey was carried out by the Ancient Monuments Laboratory (English Heritage) in 1994 and in 1998 (WSM 29123; Payne Reference Payne1999), and overall this covered an area of about 7 ha stretching mainly northwards form the site of the original hoard (Fig. 3). Both magnetometer and magnetic susceptibility techniques were used, the latter increasingly, as it was possible to scan more quickly with this. Magnetometer survey was carried out over an area of about 6 ha, and magnetic susceptibility survey over an area of about 4.5 ha, the latter mainly overlapping with the former.
Fig. 3 Results of geophysical survey
Site Sequence
The following account of site sequence is based on combining the results of the various stages of fieldwork. The limited nature of the trenching has particularly hampered the construction of a site narrative as no clear stratigraphic inter-relationships were noted, and the dating and nature of features hinted at by the geophysical survey results remain largely unknown. One general observation, that may be significant ultimately, was that small quantities of hand-collected cremated bone were consistently noted in most excavated features (and in overlying soils), and, though none of this was positively identified as human, the possibility remained (S. Pinter-Bellows, pers. comm.).
Phase 1 Lower Palaeolithic
A stone hand-axe (Fig. 4) was found during the specialist metal-detector survey. Its likely association with the 5th River Avon terrace (dated c. 290–230 kya) would make it one of the earliest artefacts ever discovered in the region. It was fashioned in a homogeneous crypto-crystalline siliceous rock, probably black chert (R. Ixer, pers. comm.; cf, Lang & Keen Reference Lang and Keen2005, 71), probably originally derived from local gravels deposited in the Pleistocene period (D. Roe, pers. comm.).
Fig. 4 Hand-axe
Phase 2 Earlier prehistoric
During the fieldwalking a thin scatter of worked flint was recovered, dating to the Mesolithic period and also, based on the use of worked local gravel flint (Saville Reference Saville1990), to the Bronze Age (R. Jackson, pers. comm.).
Phase 3 Iron Age
The geophysical survey revealed that the coin hoard site was at the southern periphery of an extensive complex of enclosures and other features centred about 200 m to the north-west (Figs 2–3). The main concentration of anomalies coincided broadly with the greater density of later prehistoric finds recovered by fieldwalking (Fig. 2). The ditches of an irregular pattern of enclosures of various shapes and sizes were present in large numbers, and there were also pits, hearths or ovens. In several cases there were small ring-ditches of 12–18 m diameter which were tentatively identified as round-houses. All of the latter seemed to lie outside the enclosures, which may, therefore, be animal compounds. These results clearly showed that the hoards were originally associated with a contemporary settlement complex of at least 2.5 ha.
In the excavated trench (Fig. 5) none of the features was positively identified as of Iron Age date and Iron Age ceramics were mainly retrieved from the subsoil or topsoil. The pottery comprised: Malvernian ware (Peacock Group A; WAAS fabric 3); palaeozoic limestone-tempered ware (Peacock Group B1; WAAS fabric 4.1); and a more common (local) fabric tempered with fine quartz and occasional plate shell. Droitwich briquetage (mainly organic tempered type WAAS fabric 2) was also present in the fieldwalked ploughsoil.
Coin hoard locations
Hoard 1: The archaeological evidence relating to the deposition of hoard 1 was not well preserved, as this hoard had already been removed before archaeological fieldwork commenced. It is possible that it had been originally contained in a small pit cut slightly into natural (E104; Fig. 5), as the coins were largely found in a single tight clump; on its north side, a (?subsoiling) furrow was identified, which was the most likely cause of some coin dispersal so facilitating the discovery. No other finds directly related to the deposition of this hoard were discovered during the subsequent fieldwork.
Hoard 2: It was initially considered during excavation that a second hoard had been lying in the fill of a flue (F103; Fig. 5) to an oven (K107), a view based on the burnt material attached to both sides of the main clump of coins, and its apparent stratigraphic position – indeed that it comprised part of the fill itself rather than being deposited in a discrete event that could be detected archaeologically (in any case there was also no clear evidence that they had been placed in a container, though the clumping might suggest this). Certainly the distribution pattern of the coins from the excavation trench seemed to indicate the presence of a ‘second hoard’ (Fig. 5) and, subsequently, further evidence came to light to corroborate this (see below). However, the pottery associated with the rest of the oven fill indicated that it had been infilled at a much later date than the coins, in the mid-Roman period (3rd–4th century ad; Hurst et al. Reference Hurst, Napthan and Pearson1993). (However, for an alternative interpretation see below.)
Fig. 5 Trench plan showing distribution of coins (grid positions of individual coins are listed in the catalogue – see inset top right for key to grid locations)
The different compositions of the excavated group of coins and the first hoard make any direct association between the two very unlikely (Williams Reference Williams1993b, and more below). Thus, compositional difference appears to support the excavators’ theory that there was a second hoard and rule out the possibility that the excavated material was scattered content from the first hoard. The location of the second hoard, however, requires some explanation (see also above). More in-depth review of the coin location data during subsequent analysis may now suggest another possible interpretation of the archaeological evidence for the excavated (ie second) hoard. The reasoning for this is as follows:
a) the coins attributed to this second hoard are mainly located at a depth of c. 0.28–0.30 m (especially the small clumps of coins) which is just above the top of the natural and where a possible buried soil was identified across much of the trench. It was observed that the top of the oven fill was disturbed, as it was mixed with this buried soil. It is suggested, therefore, that the coins had originally been mainly associated with this buried soil rather than with the fill of the oven, which would potentially have a significant impact on their context dating. In which case;
b) it would be clear that the coins were most likely very disturbed in this second hoard location (ie, the occurrence of several small clumps fairly close together). The scar of a possible (?subsoiling) furrow immediately adjacent (Fig. 5) may account for any such major disturbance.
In the light of these observations a direct association with the Roman oven seemed to be called into question, since the possibility of disturbance may mean that the coins had been dragged into their final positions by later ploughing, and so redeposited out of their original context, in which case a different date for their deposition could then be possible. Such an interpretation would presume that original burial level had potentially been on top of the natural and within what subsequently, with cultivation, became virtually indistinguishable from the modern subsoil horizon. Another possibility that the coins had been discovered in the Roman period and then reburied in a defunct feature was thought unlikely, as the main clumps of coins did not take advantage of the full depth of the oven flue – however, such an explanation could not be entirely ruled out. Whichever account is correct, this cache had been greatly disturbed, most probably in quite recent times. Indeed such finds are likely to have come to light only through agricultural disturbance, and, therefore, being relatively shallowly buried, great care must be taken in their excavation in order to have the best chance of understanding the original burial context.
Iron Age metalwork finds other than the coins – which are discussed in detail below – were rare, and these were all from the 1 ha metal-detecting survey. Two fragments of gold objects were possibly of Iron Age date, including notably a small fragment of possible torc (Fig. 6), and fragments of copper alloy shield binding (found just north-east of, and c. 12 m north of the excavation trench respectively).
Fig. 6 Gold wire (cut at both ends) possibly from a twisted wire torc (cm scale). Found c. 20 m to north-east of excavation trench during metal-detecting survey
A notable later metal-detecting find from the same site has been a gold pellet (6.64 g), cast in a mould and hammered around the edge, and probably produced originally as a blank for a coin (possibly related to the Gallo-Belgic C stater or an early British derivative copy of the late 2nd–early 1st centuries bc). Such an object is very rare, but understood not to directly represent evidence for production, and so does not create a case for suggesting that Allesborough is another mint site (J. Williams, pers. comm.).
Phase 4 Roman
The correlation of geophysical anomalies with the pattern of Roman finds from fieldwalking indicated that many of these features were likely to have a Roman phase, especially to the east where the Iron Age finds were less evident. As mentioned above the shallow base of the oven was infilled in the mid- Roman period and contained pottery datable to the 3rd century ad. It seemed most likely to have been domestic, and post-holes in the vicinity could be contemporaneous. The apparent association of the oven with a possible second Iron Age coin hoard has been discussed above. Roman finds were mainly ceramic, and a typical range of material of mid-1st–3rd/4th century ad date was present.
Phase 5 Post-Roman
A buried cultivation soil sealed all features, and, though it retained no structural integrity, closely reflected the underlying stratified deposits. This had localised disturbance from ridge and furrow cultivation and probably also from subsoiling.
The Coins
by Ian Leins
At the time of its discovery, the Pershore assemblage represented the largest group of Iron Age coins recovered under controlled archaeological conditions in the UK and, even today, only Hallaton, Leicestershire (Leins Reference Leins2007; Score Reference Score2011) is larger. The significance of the Pershore coins stems from the fact that so many were recovered during archaeological fieldwork, providing an insight into depositional practices (as was the case at Hallaton). The preservation of archaeological context also allows different deposits at the site to be compared and analysed with reference to finds from other locations, revealing the multi-regional nature of the ‘Western’ coinage series that has been traditionally and collectively ascribed to the Dobunni.
The vast majority of the Iron Age coins found at Pershore belong to the Western regional stylistic series, identified by Evans (Reference Evans1864) and the subject of a detailed study by Allen (Reference Allen1961). The series appears to have circulated throughout an area centred on the modern county of Gloucestershire, but with finds coming from parts of all of the adjacent counties. Both the gold and silver of this series are recognised to have derived from coins issued in the neighbouring ‘Southern’ region, which was centred on Hampshire, Berkshire, and West Sussex (see below). Allen (Reference Allen1961, 81–2) classified the silver coinage using the letters A–K (with the rarer variants L–O and MX). The most recent catalogue, by Cottam et al. (Reference Cottam, de Jersey, Rudd and Sills2010: Ancient British Coins (ABC)), essentially preserves this arrangement; although two types, I and J, have been amalgamated, as they cannot be distinguished from one another. The adapted nomenclature of Allen, used in ABC, and ABC numbers, are used throughout this paper.
In total 1494 Iron Age coins (23 gold or gold-plated and 1471 silver or silver-plated) were recovered (additional coin finds included ten Roman coins (one silver and nine copper alloy) and one modern penny). The coins show little evidence of wear to indicate that they had circulated much prior to deposition, although this is true of the majority of Iron Age coin hoards. All these coins are included in the coin catalogue (Appendix 1). A small group of 17 Iron Age coins, recorded as coming from the parish of Pershore on the databases of the Celtic Coin Index and the Portable Antiquities Scheme, are also listed (Appendix 3). It has been noted already that two hoards can be identified within the Pershore coin assemblage: ‘Hoard 1’ comprises 983 metal-detector finds (7 gold and 976 silver coins); and a second hoard, ‘Hoard 2’, can be identified amongst the coins recovered during excavation and appears to consist of 295 coins (5 gold and 290 silver). In the coin catalogue these are arranged by type and archaeological context, with an indication of the degree of certainty with which each context can be ascribed to Hoard 2. The remaining coin finds include 227 stratified and unstratified finds that cannot be associated conclusively with either hoard. Excluding the Roman and post-Roman coins this group comprises ten gold coins, a gold coin blank, and 205 silver or silver-plated coins.
The Pershore assemblage is dominated by uninscribed Western silver coinages, including substantial numbers of the early uninscribed ‘Type B’ (ABC 2015), ‘Type C’ (ABC 2018) and ‘Type D’ (ABC 2021) coins and the later, stylistically distinct, ‘Type IJ’ (ABC 2036; see Fig. 7) coins. Basic numismatic analysis confirms that Hoards 1 and 2 were discrete deposits, with Type IJ accounting for a significantly larger proportion of Hoard 1 than Hoard 2 (Table 1). The composition of the ‘non-hoard’ material suggests that it could include disturbed content from Hoard 2 or from both hoards. However, the possibility that there were originally more than two hoards is also plausible (see Leins Reference Leins2007, on Hallaton, for a possible parallel).
Fig. 7 Western silver units from Pershore, of Type C (left) and IJ (centre and right). The average diameter of these coins is around 12–13 mm
Table 1 SILVER COINAGE AT PERSHORE BY TYPE
Hoard 1, Hoard 2, and other finds are shown together with a site total; the composition of the 1860 Nunney hoard and the surviving sample of provenanced Western silver coins recorded in the CCI (2008) are included for comparison
Perhaps the most striking feature of the Pershore coin assemblage is the complete absence of the Western silver types E (ABC 2024) and F (ABC 2027). Coupled with the high proportions of IJ coinage, this distinguishes the site's numismatic profile from the overall composition of Western silver coinage as exhibited by the surviving corpus of provenanced finds recorded on the Celtic Coin Index (see Table 1). Explanations for these differences could lie in the spatial or temporal limits of the hoarding event(s) represented by the coin deposits at the site. Before attempting to determine the significance of the coin profile it is necessary to trace the basic development of Western coinage.
The chronology of Western Iron Age coinage
The Iron Age coinage of west-central England comprises a number of uninscribed and inscribed issues in gold and silver which can be linked together by their stylistic continuities. The series was catalogued by Evans (Reference Evans1864) and attributed to the Dobunni by Allen (Reference Allen1944, 2–3; Reference Allen1961, 75–102). Allen compared the distribution of the find-spots of Western coinage with the territory of the later Roman civitas Dobunnorum as described by the Greek geographer Ptolemy in the mid-2nd century ad (Geography II, ii). This attribution has been repeated by van Arsdell (Reference Arsdell1989, 266 ff.) and Cottam et al. (Reference Cottam, de Jersey, Rudd and Sills2010, 103 ff.) but cannot be corroborated. While the Dobunni are likely to have had roots in the pre-Conquest period (see Dio 60.20 for a reference to ‘part of the Boduni’ during the Roman Conquest), it does not follow that all users of Western coinage were necessarily ‘Dobunnic’. It is entirely possible that a number of independent groups employed Western coinage. Indeed, it is of some interest that Aquae Calidae (Bath), which Ptolemy associates with the Belgae, lies well within the distribution of Western coinage.
Western coinage cannot be studied in isolation from Southern coinages or seen as a single coherent group. The earliest Western gold, British RA (ABC 2003), is derived directly from the Southern British QB (ABC 488), exhibiting an identical reverse but with a branch emblem appearing on the obverse. Despite their different numismatic attributions (to the Atrebates and Dobunni), it is likely that QB and RA coinages drew on the same sources of production expertise. As QB was struck during phases 5–6 (Haselgrove Reference Haselgrove1993, 35; 42), specifically between about 50 and 30 bc, it is unlikely that RA was struck much before c. 40 bc. British RB (ABC 2009), the earliest gold quarter stater, was of similar date and again exhibits stylistic continuities with Southern coinages. ‘Type A’ (ABC 2012) is usually seen as the earliest Western silver coinage, although it exhibits strong links to Southern coinages (Leins Reference Leins2012). The increasingly abstract descendants of this coinage are the Type B, C, and D coins that are common in the Pershore assemblage. Together, these developments can be ascribed to an early phase of coin production, here called WE1, running from c. 40–10 bc (Fig. 8).
Fig. 8 Chronology of Western coinages (pale shading = gold only; dark shading = silver only; mid-tone = coinages in gold and silver)
The development of Western coinage after c. 10 bc (WE2–WE3) is complex. At least two streams of uninscribed silver production can be detected and these appear to have overlapped with the earliest inscribed coin production. Types E–F and IJ both developed directly from D; E–F feature more abstract versions of the obverse and reverses designs on D, while the obverse of IJ is derived from type D and coupled with a new reverse, perhaps influenced by contemporaneous coinages of the Eastern series. The unique silver unit of INAM (ABC 2063; see Rudd Reference Rudd2006) is also based on Type D.
Of the remaining inscribed Western coinages, CORIO and BODVOC can be identified as the earliest on the basis of metrological and metallurgical analysis (van Arsdell Reference Arsdell1989, 266–8; 1994, 5; Haselgrove Reference Haselgrove1993, 45). BODVOC silver units (ABC 2042/45) include portraits copied from the portrait coinages of Tasciovanos (ABC 2655 ff.), which have been dated to the middle part of Tasciovanos’ reign, c. 15–10 bc (Kretz Reference Kretz2006, 187 ff.). It is probable, therefore, that BODVOC coinages were introduced around the turn of the 1st millennium bc/ad.
The relative order of BODVOC and CORIO has been the subject of debate (cf, van Arsdell Reference Arsdell1989; Reference Arsdell1994; Sills Reference Sills2003). The most salient point, however, is that CORIO staters and quarters reveal direct stylistic links to the uninscribed British RA/RB coinage, while the staters also reveal links to the inscribed staters of COMVX, CATTI, ANTED, and EISV. As BODVOC cannot be slotted within this sequence, either between the uninscribed coinage and CORIO, or between CORIO and the later inscribed issues, his coinage is likely to represent a separate stream of production. Given the metrological and metallurgical similarities noted above, it is most likely that the BODVOC coinage was contemporary with CORIO, with both being struck close to the turn of the 1st millennium bc/ad. This conclusion appears consistent with ongoing work on the distribution of these two coinages, which continue to show very distinct distribution patterns (Sellwood Reference Sellwood1984, 196–200; Haselgrove Reference Haselgrove1993, 57; Leins Reference Leins2008, 106–7). The remaining inscribed Western coinages COMVX, CATTI, ANTED, and EISV are best grouped in the period c. ad 20–45 (WE3), but could be contemporaneous or successive issues. It should be noted that the rare ‘Pershore type’ stater (ABC 2006), which is named after the present assemblage (as three of the four known examples come from this site), has been included in period WE2. The dating of this type is unclear, but its manufacture clearly post-dates that of uninscribed British RA (ABC 2003). It is best viewed as a late uninscribed issue of WE2 on stylistic grounds, although the designs above the horse may represent a blundered inscription, meaning the possibility of a later production date in WE3 should not be discounted; see Cottam et al. (Reference Cottam, de Jersey, Rudd and Sills2010, 103) for a similar conclusion.
Pershore hoards: dating and significance
The presence of a small number of ANTED and EISV coins in Pershore Hoard 2 suggests that this hoard was probably assembled during, or after, the last phase of production (WE3, c. ad 20–45). It could have been deposited in the late pre- or post-Conquest period. Although there are no inscribed silver coins in Hoard 1, the fact that it includes a significantly lower proportion of the earlier Type A–D coins than Hoard 2, suggests that Hoard 1 was probably the later of the two. Furthermore, it should also be noted that Hoard 1 includes gold coins of BODVOC and CATTI.
At this point it is useful to compare the Pershore assemblage to the second largest hoard of Western silver found at Nunney in Somerset in 1860 (see Table 1). The post-Conquest assembly and deposition dates of Nunney appear secure. In addition to featuring higher proportions of ANTED and EISV silver (and gold coins of ANTED and CATTI), Nunney seems to have included a number of Roman silver and bronze coins; the latest an issue of Claudius for Antonia struck in c. ad 41–50. While Nunney may be slightly later than Pershore, both assemblages are clearly dominated by late uninscribed coinage. Pershore includes Type IJ but no E–F; Nunney includes Types E–F but no IJ.
This difference raises the possibility that the composition of Pershore reveals geographical differences in the circulation of Western coinage. Pershore can be seen to lie within the northern part of the overall circulation area of Western coinage; Nunney on the south-western margin of this circulation zone. Although Type E–F coins are found in the Pershore area, they appear less common here than further to the south and south-west. Type IJ, by contrast, seem not to have circulated extensively in southerly parts and did not reach the area around Nunney (Fig. 9). The stylistic divergence of Western coinage after Type D is, thus, mirrored by distributional differences. Analysing the surviving corpus of Western silver by type and modern county (Table 2) confirms this pattern, suggesting that those counties central to the distribution of Western coinage (those with the largest number of finds – Gloucestershire, Oxfordshire, and Worcestershire) exhibit the most even balance of E–F and IJ types. The counties with the highest proportion of Type IJ silver units lie to the north and north-east (Herefordshire, Warwickshire, Shropshire, and Northamptonshire); those with the highest proportion of Types E–F are Wiltshire, Avon, and Somerset to the south.
Fig. 9 Distribution of Western E–F and IJ types overlaid on a density map of all Western coin finds
Table 2 WESTERN SILVER COINAGE (HOARDS EXCLUDED) BY MODERN COUNTY (PERCENTAGE OF BCD, EF AND IJ IN EACH)
The compositions of the Pershore hoards expose significant divisions within the Western coin-using zone, particularly when considered together with the Nunney hoard. The above analysis suggests that the high proportions of Type IJ at Pershore are symptomatic of the general circulation pool in northern parts of the overall circulation area of Western coinage. The absence of Type E–F is unsurprising as this coinage was most dominant in the south-west. Although the split between E–F and IJ is similar to that between BODVOC and CORIO coinages (Fig. 10), the patterns are not identical, with BODVOC and CORIO appearing to split along more of an east/west axis. While it is possible that these differences in the circulation of later coinages, highlighted by the Pershore hoards, provide evidence of splits within a previously socio-politically unified Dobunni, it is perhaps better to see the wider distribution of some of the earliest coins made and used in the Western region as masking the existence of a fluid socio-political structure characterised by a number of independent and autonomous groups within the later Roman civitas Dobunnorum. Differences in the apparent circulation areas of late uninscribed silver (which also exhibits greater stylistic diversity), and inscribed coinage, may reveal something of the complex social, political and economic interaction between different communities in this region in the Late Iron Age.
Fig. 10 Distribution of coins of CORIO and BODVOC
Non-Western coins
It has been noted above that almost all of the Iron Age coins belong to the Western series. It is, however, worth commenting briefly on the exceptions to this. Leaving aside the single North Eastern silver half unit in Hoard 1, all of the ‘imported’ coins are gold. This is unsurprising given the propensity for gold coins to travel further from their source than silver (see, for example, Leins Reference Leins2012, 201–2). Three of the gold coins are North Eastern ‘South Ferriby’ types (two of which are plated). Although the distribution of these coins is centred on Lincolnshire, finds are not uncommon in the Midlands and further afield. Three are Southern coins; two being British QB staters, of the type discussed above, one is a Southern QC quarter stater. There is also a single coin of Cunobelin and a Belgic half stater. The latter is an unusual find for Britain that finds a parallel in the assemblage from the religious site of Waltham St Lawrence in Berkshire (CCI 83.0189; see Burnett Reference Burnett1990).
Deposition of Later Iron Age Coins and Other Artefacts
In general the rather limited investigation of the Pershore hoard location has made the archaeological interpretation of the site quite difficult. However, this is not particularly unusual, since where excavation has followed other similar discoveries, it has often been intended to increase the numbers of coins rather than to establish the context of the discovery (Haselgrove Reference Haselgrove1987, 133). Assessing the contexts of such finds is inevitably made more difficult by so few discoveries being from excavation.
Purpose of deposition
The consensus of opinion about the purposes of hoarding tends to be variable according to period (Casey Reference Casey1986). Though there is a shortage of data about the contexts of most hoards, this has not prevented speculation about the circumstances giving rise to their presence, and broadly the reasons for deliberate hoarding can be summarised as follows:
a) secretion of wealth as a temporary measure, for instance during times of trouble, or;
b) burial of objects as votive offerings at a religious site.
Late Iron Age Coin Hoards
by Derek Hurst and Ian Leins
The following short discussion of British hoard sites concentrates on those associated with Late Iron Age coin hoards where excavation has been possible following initial discovery, and where this has not been compromised by illicit digging, as at Wanborough, Surrey, which produced very large numbers of Iron Age coins (Cheesman Reference Cheesman1994; Hobbs Reference Hobbs2003, 8; 136–7 and 142–4).
Precious metal hoards, whether comprising objects or coins, are relatively rare in the Late Iron Age compared to later periods. Hoards deposited in southern and eastern Britain at approximately the time of Julius Caesar's invasion in the mid-1st century bc have sometimes been interpreted as crisis hoarding of wealth (Cunliffe Reference Cunliffe1978, 73). Although Iron Age hoards have been seen as votive offerings (Fitzpatrick Reference Fitzpatrick1984, 182), such explanations have been most frequently applied to deposits of ironwork (cf. Hingley Reference Hingley1990) and bronze objects (for example at Llyn Cerrig Bach, Wales; Manning Reference Manning1972). Iron Age coin assemblages have also been used to tentatively identify shrines (as at Bath; see Cunliffe Reference Cunliffe1980). The clearest examples are Hayling Island (Hampshire) and Harlow (Essex), where large quantities of coin finds have been seen to identify Iron Age shrines at sites that eventually developed Romano-British temples (Haselgrove Reference Haselgrove1987; Reference Haselgrove2005).
In many cases there is no evidence of contemporaneous occupation at the sites where Iron Age hoards were deposited (Haselgrove Reference Haselgrove1987) and the function of the site remains ambiguous. Extensive archaeological work at Snettisham (Norfolk) in the early 1990s demonstrated that the original hoards discovered in 1948–50 formed part of a complex history of hoard deposition. An impressive collection of worked flint (of Neolithic and later date) has come to light there, as well as a number of torc and coin hoards of various dates. Although the reason for their deposition is disputed – being first identified as a metalworking hoard site (Clarke Reference Clarke1954) and as part of a treasury (Stead Reference Stead1991) – it is unlikely that the coin groups can be interpreted in simple terms as emergency or savings hoards. The coin and torc hoards deposited at the Ken Hill site in Snettisham clearly represent one element of a wider tradition of precious metal and artefact deposition over an extended period of time, whereas the cultural purpose is less clear (eg, Fitzpatrick Reference Fitzpatrick1992).
A hoard of 153 East Anglian silver coins was found nearby at Fring, Norfolk. Most of the coins were recovered as a concreted lump and scatter from the ploughsoil, but excavation of a subsoiling furrow through the natural produced 27 of the coins, textiles, and fragments of a pottery container (Chadburn & Gurney Reference Chadburn and Gurney1991, 219). Subsequent metal-detecting at the site has increased the hoard total to 192, and also resulted in the discovery of a second hoard of 173 earlier Gallo-Belgic gold staters in an adjacent field. Although Chadburn and Gurney (Reference Chadburn and Gurney1991, 223–4) considered a number of explanations for the initial hoard, the presence of a second hoard in close proximity to the first appears to support their interpretation as votive offerings. Though the later (ie, cache of silver coins) may have been a Boudiccan hoard, it is not necessary to see this as differently motivated, if all the deposits are essentially components of a ritual landscape, where deposition of wealth favoured key locations. Another excavated hoard from the same vicinity is the Sedgeford hoard (Dennis & Faulkner Reference Dennis and Faulkner2005), which included 39 Gallo-Belgic gold staters, 20 of which were buried inside a cow bone, and the ritual function of these coins appears more conclusive. The number and proximity of these sites in north-west Norfolk perhaps supports the idea that these coin hoards formed part of an extended ritual landscape operative over at least several generations.
Excavation of the find-spot of the 2008 Wickham Market, Suffolk, hoard (Talbot & Leins Reference Talbot and Leins2010), which comprised 840 gold staters and a ceramic container, failed to uncover contemporaneous structural/occupational remains. While the hoard could be seen as an emergency or savings hoard, the fact that it was buried adjacent to a contemporary ditch suggests that it was not necessarily buried for safekeeping at an isolated location away from human activity and could be another example of a votive deposit.
The largest assemblage of Iron Age coins ever recovered under controlled archaeological conditions in Britain was discovered at Hallaton, Leicestershire, in 2000. In total 5296 Iron Age and early Roman coins were recovered by University of Leicester Archaeological Services (ULAS). Excavation around the find-spot uncovered a concentration of 14 discrete hoards, clustered in an area of 4 m2 (Priest Reference Priest2004, fig. 10; Priest et al. Reference Priest, Hill and Leins2006; Score Reference Score2011), which were buried in the entranceway of a ditch that appeared to divide off a section of the hilltop. The coin hoards lay on one side of the entranceway and were mirrored by animal bone (mainly pig bone) deposits on the other side. Another large coin hoard was buried together with a Roman auxiliary cavalry helmet against the side of the ditch itself. All of these hoards can be dated to the early post-Conquest period on the basis of their coin compositions which include Iron Age and, in most cases, Roman Republican and/or Julio-Claudian coins. Depositional activity on a more limited scale seems to have taken place in the decades immediately preceding the Roman invasion of ad 43. The coin groups lay just under the ploughsoil. The shallow archaeology meant that the context of the hoards’ deposition was only identified during the excavations and was not apparent from the geophysical survey. Most significantly this hoard suggests that other scattered coin hoards could, in fact, have represented multiple votive deposits at the point of burial (Leins Reference Leins2007, 39; Haselgrove Reference Haselgrove1987, 119–20).
Quantities of Iron Age coins also occur significantly in Roman contexts as multiple discrete finds of gold and silver coins, most often associated with temple sites, where they were once interpreted as casual or residual finds (Haselgrove Reference Haselgrove1987), and especially where an earlier religious site continues into the Roman period (Curteis Reference Curteis2006). At Harlow, Essex, for instance, excavation has produced a large number of Iron Age coins found scattered across the site of a later Roman temple (Allen Reference Allen1964, Reference Allen1967; Reference Allen1968), where most of the coins seem to be derived from an early Roman horizon. The reason for the deposition of coins in this way at Harlow is uncertain, though there is a likelihood that such coins were votive offerings associated with an earlier, so far unidentified, religious site, and so it might be posited that this may represent the adoption of a new custom before the Roman coin supply was sufficient (Haselgrove Reference Haselgrove1989). In the south-east the most common metal for coins in this type of context is silver (Haselgrove Reference Haselgrove1987).
Hoards are also known, however, in the vicinity of later Roman religious sites, such as where Iron Age coins were deposited in a post-2nd century ad context at Castle Rings Camp, Donhead St Mary in Wiltshire (Cowell et al. Reference Cowell, Oddy and Burnett1987); and on sites adjacent to a Roman temple (ie, Weycock Hill, Berkshire).
In contrast, where stratified Iron Age coins have been found in the wider south-east, they have been largely copper-alloy types, and often have Claudian (ad 41–54) associations (eg, Sheepen in Colchester, and Hayling Island, Hampshire (Haselgrove Reference Haselgrove1987, 362–74 and 402–6 respectively); and whilst gold and silver types were common at Hayling temple, the coins were mostly plated).
Conclusions
by Derek Hurst
It is particularly striking that there are many similarities between the largest recent Iron Age coin hoard sites at Pershore, Worcestershire, and Hallaton, Leicestershire, both in terms of the hoard composition (ie, mainly coins), and of the general natural and archaeological landscapes represented by the site locations. Putting aside the earlier prehistoric activity at both sites, though the possibility of Early Bronze Age burial sites seems to exist in either case, both sites were previously unknown and located on (as once thought) unpromising geologies for settlement, and more specifically occupy exposed positions on hills, but not on the very brow. Both sites also enjoy sweeping views across a river valley. Also, in both cases, possible round-houses have been identified nearby and to the north, and the hoards seem to be located on the edges of the main Iron Age finds scatters (cf, Priest Reference Priest2004, 22), but in areas where Roman finds are later to feature (Fig. 2; Score Reference Score2011, fig. 3). Therefore, though the limited area of excavation at Pershore prevents any detailed comparison, there are apparent similarities in terms of general topographic location, while it remains feasible that the Pershore site has a similar chronology as well. Whereas extensive geophysical survey was undertaken at both sites, the observation at Hallaton that key features associated with the ritual phase did not appear in the geophysical plot (Score Reference Score2011, 11–12) suggests caution with this type of data when comparing such sites.
For the ultimate interpretation of any hoard it is clear that the date of the deposition is crucial, particularly for understanding the motives of the depositors, and, since any precision about this is presently lacking for the Pershore find, this means that any conclusions about the hoard(s) here will retain some uncertainty. Whereas at Hallaton, based on the more extensive investigation, it could be concluded that the depositions of coins and other finds were mainly carried out in the early post-conquest period (c. ad 43–50; Leins Reference Leins2011, 42) and at an open-air ritual site (Score Reference Score2011, 163), the less detailed evidence in some respects at Pershore seemed broadly similar. Taking into consideration especially the site topography (its relatively elevated position), and the likely fact that the Pershore hoard(s) were probably the source of all the coins/finds rather than there being a combination of hoards and many individual coins (the latter in particular characterising religious sites continuing into the Roman period), then it seems perhaps most likely that the Pershore site also represents a pre-Roman religious shrine, with an adjacent contemporary settlement which continued into the Roman period. The hoards would, therefore, have been for votive purposes, which is also perhaps more credible, given that, if the objects had been buried with the intention of recovery, then a more remote location might have been expected, as, apparently, at Fring, for instance. Whether such Iron Age religious spaces were unusual is, of course, unknown, and it may well just be unusual that such a site is ever recognised archaeologically, and so may well be in this respect that the Pershore coin hoard discovery is the more significant. Further archaeological investigation specifically focused on the character of the later prehistoric activity in the Pershore area generally would, of course, be key to providing a wider and more detailed context for any further discussion of the hoard(s) site and the remarkably rich finds that form the subject of this report.
Acknowledgments
Worcestershire Archive & Archaeology Service (WAAS) would like to thank John Bridgewater, Gordon Jones, and Graham Skidmore, the discoverers of the original hoard, for its prompt reporting; and John and Sarah Bomford for allowing access for fieldwork, and for their interest and assistance in the project. Much of the WAAS fieldwork was supervised by Mike Napthan, with the assistance of Martin Cook, Luke Fagan, Paul Godbehere, Robin Jackson, and Nigel Topping. During reporting thanks were owed to the following British Museum staff for their assistance: Dr Ian Stead, who kindly identified some of the objects and gave advice during the early stages of the project; Dr Jonathan Williams and Dr Andrew Burnett, who provided preliminary reporting on the coins; Dr Jody Joy and Dr Julia Farley for discussion of the gold torc fragment; and Dr Richard Hobbs, Dr Philippa Walton, and Dr Eleanor Ghey for general comments on other objects; and similarly thanks are also owed to Dr Derek Roe and Dr Rob Ixer, who commented on the stone hand-axe, Angie Bolton (Portable Antiquities Scheme), Patrick Clay (University of Leicester Archaeological Services), Tom Richards (Herefordshire & Worcestershire Earth Heritage Trust), Carolyn Hunt (illustrator, WAAS), and Emma Hancox and Oliver Russell (HER). Derek Hurst would also especially like to thank the following for enabling the project to progress in the right direction: Simon Woodiwiss (WAAS), who deftly led its early development; Andy Payne (Ancient Monuments Laboratory, English Heritage) for the extensive geophysical survey effort across several years; and Kate Clark, Tony Fleming, Helen Keeley, Barney Sloane, and Tony Wilmott (English Heritage) for funding post-excavation analysis and reporting. And, finally, the authors most gratefully acknowledge the insightful suggestions made by the anonymous referees towards enhancing the original report draft.
This paper is published with the support of a grant from English Heritage.
APPENDIX 1 : CATALOGUE OF THE IRON AGE COINS
by Ian Leins
The locally produced silver coinage, which dominates the Pershore assemblage, is generally referred to by a single letter, A–K. This nomenclature is based on the classification of Allen (Reference Allen1961), which has been maintained in most subsequent catalogues including ABC (Cottam et al. Reference Cottam, de Jersey, Rudd and Sills2010). Here, as in ABC, Allen's types I and J have been amalgamated, as the distinction between the two can no longer be maintained. The common IJ type has been further divided into a number of varieties identified on the basis of the Pershore coins:
IJ variety 1: Reverse: right-facing horse; V below tail. Flower pattern with X, dot and O shapes below horse.
IJ variety 2: Reverse: right-facing horse; X below tail. Flower pattern with X, dot and O shapes below horse.
IJ variety 3: Reverse: right-facing horse; ‘arrow’ below tail. Flower pattern with X, dot and O shapes below horse.
IJ variety 4: Reverse: right-facing horse; straight line or ‘arrow’ below tail. Crescent shape above horse.
IJ variety 5: Reverse: right-facing horse; curve and dot below tail. Flower pattern below.
IJ variety 6: Reverse: right-facing horse; flower or X below tail. Square with pelleted corners above.
IJ variety 7: Reverse: left-facing horse.
In addition to ABC, BMC (Hobbs Reference Hobbs1996) references are given for the gold and non-local Iron Age silver coins. The single continental coin has a DT reference (Delestrée & Tache Reference Delestrée and Tache2002). The Roman Republican coin is catalogued according to RRC (Crawford Reference Crawford1974); the two identifiable Imperial period coins using RIC 5 (Mattingly et al. Reference Mattingly, Sydenham and Webb1933) and RIC 7 (Sutherland et al. Reference Sutherland, Carson and Bruun1966). In Hoard 2, the degree of certainty to which particular coins and contexts can be ascribed to the hoard is indicated by a number (1 = definitely part of the second concentration; 2 = probably; 3 = possibly). Weights are provided, with broken coins indicated by the suffix (b), coin fragments by (f) and fused coins by (fsd). The weights of the most common varieties of type IJ in Hoard 1 are summarised in Appendix 2 in order to save space in the catalogue; Appendix 3 includes a list of other finds from the parish of Pershore recorded by the Portable Antiquities Scheme (PAS) and the Celtic Coin Index (CCI). For grid context location of excavated coins, where indicated below, see Figure 5.
HOARD 1 (983)
Gold Coinage (7)
Silver Coinage (932)
Plated Coinage (44)
HOARD 2 (295)
Key to context numbers: 100 = ploughsoil; 101 = subsoil (except in grid E, where it is 105); E103 = fill of E104; F102 = fill of F103; L102 and M102 are equivalent to 101. MD = metal-detecting find. Cut features listed here are indicated in text & on Fig. 5
Gold Coinage (5)
Silver Coinage (279)
Plated Coinage (11)
OTHER CONTEXT & SPOIL (227)
Gold Coinage (11)
Silver Coinage (195)
Plated Coinage (10)
Roman, modern, etc. (11)
APPENDIX 2: Weight Ranges and Average Weights of All IJ Varieties in Hoard 1
APPENDIX 3: Other Metal-Detector Finds From Pershore Environs
Silver Coinage (13)
Plated Coinage (4)
