Marine shells

Marine shells

Marine shell beads from three inland Later Mesolithic sites in western Britain

A substantial number have no obvious funerary associations and derive from middensshell middens (køkkenmøddings) and in human graves close to Holocene marine shorelines. Some of the most celebrated examples come from Mesolithic burials in Brittany (Péquart & Péquart 1929, 1954; Péquart et al. 1937), but a substantial number have no obvious funerary associations and derive from middens along the Atlantic façade of western Europe.

little thought has been given to the spatial distribution of shell beads in the landscape or at an inter-site level. For example it has frequently been assumed that they invariably occur at coastal sites, close to the point of source, with the result that finds from other types of location have often been ignored or neglected.

In this paper, we draw attention to three inland cave localities in western Britain where recent excavations have uncovered evidence of marine shell artefacts.

We look at the relationship of the three sites to other bead findspots in Britain and Europe and will argue that the predominantly western bias in their distribution reflects genuine patterning in the archaeological record rather than simply arising from taphonomic factors of preservation.

The three inland Mesolithic sites

Madawg Rockshelter, Herefordshire

Location and stratigraphic contexts

Madawg Rockshelter (NGR SO 5474 1527) is located about 1 km southwest of Symonds Yat on the right bank of the River Wye in Herefordshire and about 40 km inland from the nearest coast (Fig. 1). The site is at an elevation of c. 60 m OD, and occurs in limestone bluffs of Lower Crease Limestone known locally as the Seven Sisters rocks. The rockshelter opens high above the river and is crescentic in shape with a shallow overhang covering a floor area of c. 270 m2 (inside the drip line). It faces almost due west and, under conditions of low vegetation, would have provided good views downstream of the River Wye. Access from the bottom of the valley is possible today via a steep scramble up scree slopes; the plateau above the site is more easily accessible.

The more recent programme focused on an area of c. 30m2 in the northern half of the rockshelter (Barton et al. 1997).

Later Mesolithic artefacts including an assemblage of 11 cowrie (Trivia sp.) and two flat periwinkle (Littorina obtusata) shells were recovered by excavation and fine sieving of the BTE deposits (Fig. 2). The shells were distributed in an area of 2.25 m2 with a concentration occurring in a zone 50 cm in diameter (in square L6) and within a discrete 10 cm thick band. The only ‘feature’ in these sediments was an elongate depression (centred on squares K11-K12) which contained well-preserved charcoal and a few charred fruit stones. The upper part of the 30 cm deep depression consisted of charcoal of mixed oak woodland species (Barton 1997, 107). Charred wood was much sparser in the lower part of the depression and consisted of pine (Pinus) and yew (Taxus). Also identified in the lower levels (Wendy Carruthers pers. comm.) were the seeds and stones of hawthorn (Crataegus sp.) and sloe/blackthorn (Prunus spinosa). Small mammal remains from this context and the equivalent stratigraphic unit comprised wood mouse (Apodemus sylvaticus) and bank vole (Clethrionomys glareolus), both of which are compatible with the charcoal evidence and confirm the local presence of deciduous woodland.

Microlith associations

Later Mesolithic artefacts including narrow geometric microliths were recovered from the BTE and top of the underlying BSCE sediments.

Dating evidence

Two AMS radiocarbon dates were obtained on charred specimens from the lower levels of the depression:

OxA-6082 charred hazelnut shell (MDG 528) 6655 ± 65 BP

OxA-6081 charred sloe stone (MDG 527) 8710 ± 70 BP

The sloe stone was recovered in a spit some 7 cm below the hazelnut sample. These dates are slightly difficult to interpret as they do not overlap at one standard deviation and it is likely that they represent different phases of site use and accumulation. On comparative typological grounds, we would suggest that the younger of these two dates is likely to provide an age for the geometric microliths and shell beads.

King Arthur’s Cave, Herefordshire

Location and stratigraphic contexts

King Arthur’s Cave (NGR SO 5458 1558) is one of a series of small caves and rockshelters that open on the west side of Great Doward Hill, about 3 km south of the village of Whitchurch, Herefordshire (Fig. 1). It lies not far from Madawg Rockshelter (350 m to the north-northwest) but at a slightly higher altitude of c. 110 m OD. The cave overlooks a small dry valley that leads down to the River Wye via a sharply dipping slope. The cave itself is formed of Lower Crease Limestone (Welch & Trotter 1961), its roof occurring at the geological junction with the oolitic Upper Crease Limestone. The cave mouth consists of two interconnected entrances that lead back to a small network of shallow chambers and passages little more than 15 m deep. In front of the cave is a broad platform covered by spoil from earlier excavations of the cave.

Isolated examples of Mesolithic flints, three cowrie (Trivia sp.) and two flat periwinkle (Littorina obtusata) shell beads were recovered in a small alcove on the north-west side of the Second Chamber (Figs 2 and 3). Originally it was believed that this area had been undisturbed by mining activities or the 19th century excavations, but it soon became clear from our small test pit of 70 x 50 x 50 cm in the alcove that there was a mixture of Bronze Age pottery, Mesolithic and Later Upper Palaeolithic finds (Barton 1996). Nonetheless, the fact that the shells all came from reddish stony cave earth deposits and were found within a few centimetres depth of one another suggests that they derived from within the alcove itself or as localised backfill from the Second Chamber. The discovery of these artefacts provided the first tangible links between human activities at this cave and Madawg Rockshelter. It is also interesting to note that there is a reference to a drilled shell of Neritoides obtusatus (Littorina obtusatus) amongst finds recorded from the nearby site of Merlin’s Cave (Hewer 1926, 220).

Three Holes Cave, Devon

Location and stratigraphic contexts

Three Holes Cave (NGR SX 8154 6747) is situated in Dyer's Wood on the south-west side of the Torbryan Valley in South Devon, about halfway between the granite uplands of Dartmoor and the coast (Fig. 1). This deeply incised valley (now largely infilled by sediment) is located on the Am Brook, which is linked to the River Dart via the River Hems. The cave is set in a line of low cliffs of Devonian limestone. The main entrance opens to the north-east and lies about 4 m above the level of the present valley floor (68.5 m OD). A short curving passage leads back from the entrance to meet a single ‘chamber’ partly formed by a cross-fissure in the bedrock that continues upwards until it eventually connects with the surface. The roof is choked with large limestone blocks or bridged by rock and stalagmite, so as to give the impression of three main 'holes' in the chamber roof. Archaeological material has been found at the cave dating from the Lower Palaeolithic, Middle Palaeolithic, Late Upper Palaeolithic (Creswellian and Final Palaeolithic), Later Mesolithic, Neolithic, Bronze Age and Romano-British periods (Roberts 1996). Three Holes is the only one of the 12 known caves in the Torbryan valley to have produced definite evidence of Later Mesolithic artefacts. However, the Torquay Natural History Society collections contain a single-perforated periwinkle shell (A2553) found on 14th September 1936 during the Society’s excavations at 'Torbryan 2’ (Tornewton Cave), which suggests that Later Mesolithic activities in the valley may have been more extensive.

Palaeoenvironmental analyses show that the contemporary environment in the Later Mesolithic was one of mature woodland. The small mammal assemblage, dominated by bank vole (Clethrionomys glareolus), with field vole (Microtus agrestis) and some wood mouse (Apodemus sylvaticus), is indicative of deciduous woodland with no major clearances in the area (Price 1993). The terrestrial mollusca from the same context included species such as Oxychilus cellarius, Aegopinella nitidula and Discus rotundatus, and shows a local presence of shaded habitats, with no open-county preferring species (Seddon 1996, 194).

A total of 30 perforated marine shells can be attributed to the Later Mesolithic horizon at Three Holes Cave (Figs 2 and 3). In the collection from the London Institute of Archaeology excavations, now held by Torquay Museum, we recorded 16 periwinkle (Littorina obtusata) shells with a single perforation and nine cowrie (Trivia sp.) shells either with a double perforation (6) or broken in the area where perforations might have been located (3). This is the same number (25) listed by Masson-Phillips (1981) although his description of 11 periwinkle and 14 cowrie shells does not match our findings. A further five modified marine shells were recorded during the British Museum work, four periwinkles and a single modified Dentalium shell. These were found at the top of the LBST deposit and adjacent to the 1961 excavation area. Several fragments of unmodified mussel shell (possibly freshwater mussel) were also found in this deposit. Most of the modified shells are complete, and the generally good condition of these delicate objects is noteworthy given the otherwise generally trampled nature of the Later Mesolithic archaeological deposit. One possible explanation might be that they were deposited at the end of this phase of use of the cave.

Dating evidence

Two modified bones of red deer were AMS dated and the results indicate that the site was occupied during a relatively restricted period of the early Atlantic, or climatic optimum. This dating is in keeping with the palaeoenvironmental assemblages, including the herpetofaunas, which indicate warmer conditions than the present day (Gleed-Owen 1996).

OxA-4491 C. elaphus, proximal end of radius (THRFA 1181) 6330 ± 75 BP

OxA-4492 C. elaphus, cubo-navicular (THRFA 896) 6120 ± 75 BP

The Marine Shells

The shell types represented are typical Atlantic genera (Fig. 5). Mapped information available from the Conchological Society of Great Britain and Ireland indicates that today they are widely distributed around the western shorelines of Britain and Ireland (Wilkinson 2011; Conchological Society 2012; Smith 2013a, 2013b), as well as along much of the Atlantic coastal shelf as far south as Spain and Portugal.

Cowrie (Trivia sp.)

There are two species of cowrie in British waters: Trivia arctica (Pulteney 1799) or Northern Cowrie and Trivia monacha (da Costa 1788) or Spotted Cowrie, both of which prefer the sublittoral zone of rocky coastline habitats. As fresh shells these can be easily told apart by the appearance of three brown dorsal spots on the shell of T. monacha which are absent on T. arctica. The shells of both species are pinkish white, with pink to dark pink or buff intervening grooves. In fossil examples coloration is rarely preserved and so it is uncertain to which species the Mesolithic specimens described here belong. Moreover, since they occupy similar environments they cannot be differentiated by habitat except perhaps at extreme ends of their geographic range. Some success has been achieved in separating the species using Fournier analysis on the apical outlines of shells (Dommergues et al. 2008) but here the sample sizes are very small and this was not attempted.

Marine shell beads from three inland Later Mesolithic sites in Western Britain

Flat periwinkle (Littorina obtusata)

Only one species of periwinkle has been recorded at the Mesolithic sites. Today the flat periwinkle can be found on exposed and semi-exposed rocky shore locations often with a strongly sloping profile, and subject to considerable wave action (Wilkinson 2011). The colour of fresh shells varies from olive-green to yellow or orange and can include surface banded and chequered patterns. They are common around western British coasts and this includes the upper reaches of the Bristol Channel. Perhaps surprisingly, given the similarities in coastal morphology, they appear to be less common in Ireland except in the extreme north-east.

Dentalium (Dentalium sp.)

These molluscs are sometimes referred to as tooth shells or tusk shells on account of their elongate, conical and curved form. They are usually whitish in colour in their fresh state and when found as dead shells they are hollow and open at both ends. Dentalium shells are widely distributed in western Britain but are less common in Ireland except on the north-east coast. Only one record exists, from Three Holes Cave (THR 302), and it is unclear to which species it belongs. As far as we are aware there are no other examples of modified Dentalium shells yet known from the British Later Mesolithic.

Shell beads from Later Mesolithic coastal sites in Britain, Ireland and northern France

Identical examples to the shell beads described here have been found at other Later Mesolithic sites in western Europe (Fig. 1, Tables 3 and 4). Most of them come from sites along the Atlantic coast or from related estuarine situations. The potential for recovery of Mesolithic sites is particularly high in these locations where shell accumulations occur and Holocene coastal sediments are actively being eroded by rising sea level (Bell et al. 2006). A brief overview of some of the main sites to have produced beads is presented below.

Although shell middens are known from many locations in south-west England, marine shell beads have only certainly been recorded at Culverwell on the Isle of Portland, Dorset.

A lack of published detail makes it difficult to interpret the age of the beads but AMS radiocarbon determinations on a stratified sequence of unmodified Monodonta (sea snail) shells from trench 41 exhibit ages ranging from 6410 ± 55 BP to 6800 ± 60 BP within the artefactually richest part of the midden and a date of 7285 ± 60 BP from beneath the midden accumulation (Thomas & Mannino 1999, 94; Mannino & Thomas 2001, 1105; Table 2). These indicate the period of midden accumulation and provide likely ages for at least some of the ornaments (Thomas & Mannino 1999). According to these authors, there are rare examples of Trivia in the midden but, to our knowledge, no reference to any beads of this kind exists.

Various shell middens are known from Wales. Unfortunately, early work at Nanna’s Cave, Caldey Island, Dyfed, did not lead to the systematic recovery of finds. A reference exists to a shell midden containing human bones encrusted with stalagmite (Leach 1916). The latter contained cockle, limpet and oyster shells (David 2007, 34) and may have been the source of a perforated cowrie shell identified with other material from Nanna’s Cave among the collections of the National Museums and Galleries of Wales (David & Walker 2004, 328). Also reportedly from the same cave are perforated examples of a netted dog whelk (Hinia reticulata) and a flat periwinkle (Littorina obtusata) (David & Walker 2004, 328). In north Wales, two sites have yielded evidence of modified shell ornaments. The first is at Bryn Newydd, Prestatyn, Flintshire where a perforated oyster shell disc is recorded in association with a microlithic industry in tufa deposits (Clark 1938, 1939, 201). The absence of geometric forms other than narrow scalene triangles in the assemblage would suggest that it occupies a chronologically early position in the Later Mesolithic. This is seemingly confirmed by two AMS radiocarbon dates from this site on associated hazelnut shell which yielded ages of 8700 ± 100 BP and 8730 ± 90 BP (David 2007 and Table 2). The second is a new discovery from a shallow rockshelter known as Snail Cave, Great Orme, Llandudno (G. Smith pers. comm.). The find consists of a double-perforated cowrie retrieved from an eroded surface in amongst shell midden debris.

Finds of perforated shell beads have long been associated with Obanian coastal locations in western Scotland (Lacaille 1954), and are extensively reviewed in recent literature (Hardy & Wickham-Jones 2002, Hardy 2010; Saville 2004). Most of the reported examples are restricted to cowrie and pecten shells (Saville 2004), and, except for Carding Mill, Oban (Connock et al. 1992) and Sand, Wester Ross (Hardy & Wickham-Jones 2009), originate from off-shore islands. The greatest concentrations of double-perforated cowries come from the middens on Oronsay (Bishop 1914; Mellars 1987), smaller numbers are known from Sand (seven cowrie beads) and Carding Mill Bay 1 (no figure available), while a solitary example has been reported from Ulva Cave (Bonsall et al. 1994 and Bonsall pers. comm.). Interestingly, despite the prevalence of Littorina shells on Oronsay and even a midden made up almost entirely of Littorina littorea (Mellars 1987, 105), these shells seem only to have been rarely used in Mesolithic bead-making.

The dating of the shell beads is relatively well documented on Oronsay where broad ages can be assigned for individual middens in which they were found (Table 2). These suggest that none is likely to be older than the basal age of 6190 ± 80 BP for the midden at Caisteal nan Gillean I, while representative ages for this and other middens on Oronsay appear to cluster between 5700 and 5400 BP. This dating overlaps at 2 sigma with the early shell midden at Carding Mill Bay I in which double-perforated cowrie beads were recovered from contexts VII, XIV, XV and XVII (Connock et al. 1992; Table 2). At Ulva Cave, the solitary example of a double-perforated cowrie bead was recovered in Area C, grid square N5, layer 1 (Russell et al. 1995). Bulked Patella shells from a column in the adjacent grid square P5, though not in direct association, may provide a minimum age of 5685 ± 65 BP for the cowrie bead (Table 2). The only site that seems significantly older is that of Sand where the sole dated context with a cowrie bead was Spit 4 of the midden with an age of 7855 ± 60 BP (Wickham-Jones pers. comm.). It should be noted however that the excavation was in spit units and significantly younger ages have been obtained from the overlying spit 3 (Hardy & Wickham-Jones 2009), so the date should be treated with caution.

Of the 200 or more shell middens known from the coast of Ireland (Milner & Woodman 2007) relatively few can be assigned with certainty to the Later Mesolithic. Exceptions include Ferriter’s Cove, Co. Kerry (Woodman et al. 1999) and those around Lough Swilly, Co. Donegal, a large fjord on the north coast of Ireland (Harte 1866). Recent surveys along Lough Swilly have located at least 11 shell middens around Inch Island, which lies about 20 km from the sea coast (Milner & Woodman 2007, 105), of which Baylet (Harte site 3) has produced material of Mesolithic age. Specimens of pig bone and charred hazelnut shell from beneath the midden have yielded radiocarbon ages of 6450 ± 50 and 6065 ± 40 BP respectively (Milner & Woodman 2007, 105). A double-perforated cowrie bead was also recovered from a ‘dark layer’ presumed to be the buried land-surface in sample 74, square F9, trench 3 extension collected in 2002 (McCaffrey 2012). This is so far the only known Irish record of a perforated cowrie shell from the equivalent of a Later Mesolithic context.

In Brittany (France), the occurrence of shell bead ornaments (parure) has long been recognised in the Late Mesolithic (Péquart & Péquart 1929; Rozoy 1970; Taborin 1974; Marchand 2003). The most famous sites are those of Hoëdic and Téviec in coastal Morbihan (Péquart et al. 1937; Péquart & Péquart 1954), where richly adorned burials have yielded prolific numbers of perforated beads in a variety of shell species (Fig. 1 and Table 3). By far the most common forms are double-perforated Trivia and single perforated Littorina obtusata beads (Dupont 2006). An example of the enormous numbers involved is illustrated by Sepulchre C at Hoëdic, where two infant burials were described as being literally stuffed (‘farcis’) with 2900 Littorina beads (Péquart & Péquart 1954, 35). Direct AMS dating of a young male adult associated with this grave (C1(2)) provided an age of 6280 ± 60 BP (OxA-6706) (Schulting & Richards 2001, 320; see Marchand et al. 2009 for calibration discussion). At Téviec, finds of double-perforated Trivia beads (3839) slightly outnumber those of Littorina obtusata (2934) (Dupont 2006, 175). One of the richest burials at Téviec is that of grave H3 which contains beads of diverse shell types including Nassarius reticulatus, Antalis sp., Cerastoderma edule and Littorina obtusata (Rigaud 2011). Direct dating of this burial of a young adult female gave an age of 6530 ± 60 BP (OxA-6702) (Schulting & Richards 2001, 320; see Marchand et al. 2009 as above). The currently available dating evidence suggests that graves with personal ornaments potentially span a period of around 1400 years at Hoëdic, longer than the nearby site of Téviec (Dupont 2006, but for further discussion see Schulting 2005a).

Inland Later Mesolithic sites with shell ornaments in Britain and northern France

Besides the three sites mentioned above examples of shell beads from inland contexts in Britain are extraordinarily rare. An exception appears to be the small rift cave of Aveline’s Hole, Burrington Coombe (Somerset), which lies some 16 km from the present coast but would have been considerably further inland during its use in the Mesolithic. Excavations in 1919 by the University of Bristol Spelaeological Society recovered ‘two or three shell beads’ from stalagmite while other examples of ‘drilled shells of Neritoides (Littorina) obtusatus occurred throughout the (underlying) layer’(Davies 1921, 69). Unfortunately much of the evidence was destroyed or badly damaged in wartime bombing and it is uncertain whether periwinkles were included amongst the grave goods, although a single broken shell bead reportedly came from the general area of the double burial (see Schulting 2005b, 183).

Early Mesolithic but the association of shell beads either with the early burials or in relation to Later Mesolithic activity remains enigmatic.

About 5 km away from Aveline’s and also about 16 km from the coast, a perforated periwinkle (Littorina obtusata) shell was found during work in 1927-28 at Gough’s Cave on the left bank of Cheddar Gorge (Somerset) (Parry 1928; Jacobi 1986). Sadly the exact find location and context is unclear, and no link can be proven with specific deposits, including the important Later Upper Palaeolithic deposits, or the ‘Cheddar Man’ Early Mesolithic skeleton (BM-525 9080 ± 150 BP; OxA-814 9100 ± 100 BP). However, Roger Jacobi believed that the bead was probably related to microliths from Gough’s Cave found at the same time and most likely of Mesolithic age (Jacobi 1985, 108 and pers. comm.). In discussion with the authors Roger also mentioned a bead find from a cave in the British Midlands. He recalled that a perforated Littorina obtusata shell had been recorded from the upper layers of Pin Hole Cave on the north side of Creswell Crags (Derbyshire) during work by A.L. Armstrong in 1925-32, and was possibly of Mesolithic age. Pin Hole Cave is about 80 km from the coast and if this perforated shell can be shown to be of Mesolithic age then it would be the furthest inland find of such an artefact. However, the age and association of the perforated Littorina beads found during the early excavations at all three of these inland cave sites are likely to remain uncertain.

One inland site with possibly better potential in this respect is that of Blashenwell Farm, Corfe Castle, Dorset which occupies a position a few kilometres from the coast. The site was first mentioned by Clement Reid in 1895 and includes Littorina shells and geometric Mesolithic microliths from ‘kitchen midden’ deposits in a tufa (Reid 1896; Clark 1938). The biostratigraphy and dating of this site were re-assessed in 1980 by Richard Preece who obtained dates of 5750 ± 140 BP and 5425 ± 150 BP (BM-1257, 1258) on mammal bone from Reid's midden which were attributed by Preece to the upper levels of the tufa on the basis of mollusca recovered from the sediment inside the bone cavities. Preece proposed that tufa deposition began at the site shortly before 9000 BP and ceased about 5000-4000 BP, providing a Mesolithic age for the whole tufa deposit. A date of 6450 ± 150 BP (BM-89), also on mammal bone from Reid’s midden, had previously been attributed to the 'middle zone of the tufa', but is unrelated to the molluscan sequence. The three dates together suggest that the midden perhaps accumulated over about a 1000 year period during the later part of the Mesolithic. Some years ago Roger Jacobi in conversation with the authors suggested that one of the Littorina shells from the kitchen midden had been modified into a bead. It has not been possible to confirm this occurrence as yet.

A similar paucity of shell beads has been noted from inland Mesolithic sites in northern France. One important exception is that of ‘Petit Marais’ at Chaussée Tirancourt which lies in a small side valley of the Somme River in Picardie (Fig. 1). It is located approximately 55 km from the present coast. It consists of ‘Middle’ Mesolithic (in a French sense) material rather than the Final Mesolithic assemblages of coastal Brittany. At Chaussée Tirancourt the occupation horizons are intersected by a number of large pits (fosses) which unusually contain cremated human remains (Ducrocq & Ketterer 1995). Littorina beads with a single perforation are widely scattered throughout the site while about 20 perforated Cardium shells and a fragment of Dentalium have been recorded from Fosse 2 which is dated to 7840 ± 90 BP (Gif-8913) on oak charcoal found in the pit fill (Ducrocq & Ketterer 1995; Ducroq pers comm. and Table 4). The retouched tool assemblage from the site comprises narrow backed geometric microliths that can occur in the British Later Mesolithic but the presence of small points with surface retouch or feuilles de gui, is a distinctive feature of such assemblages that date to around 8000 BP (Gob 1985), and is not found in Britain. A second pit, Fosse 1, is partly truncated by Fosse 2 and contains a slightly older assemblage according to carbonised hazelnut and animal bone that have been dated to 8460 ± 70 BP (Gif-9329) and 8360 ± 90 BP (GifA-95471), respectively (Ducrocq & Ketterer 1995). The contents of the older pit include burnt specimens of Tertiary fossil Bayania (sea snail) shell beads. A point of potential interest is the use of fossil shell at Mesolithic sites further to east in the Oise valley at Warluis II (around 8800 BP) and at Loschbourg in Luxembourg where Bayania shells have been reported from Mesolithic deposits (Ducroq pers. comm.). Examples of fossil Ampullina (deep-water sea snail) shells are also known from older Mesolithic contexts in the Île-de-France, implying a gradual change in Mesolithic traditions marked by an increase in use of fresh shell in the later sites, possibly reflecting a greater orientation of activities along coastal zones at a time of rising Holocene sea levels. However, the presence of marine shell ornaments in burials at the early Mesolithic site of La Verne (Charente-Maritime), that would have been about 60-80 km from the contemporary coastline, indicates that the subject is complex (Schulting et al. 2008).

Discussion and conclusions

The three inland caves presented here all provide examples of locations where perforated shell beads have been recovered from in situ Later Mesolithic contexts. In terms of chronological evidence, the most securely dated is that of Three Holes Cave where the finds can be related to cut-marked animal bone with overlapping ages at 2 sigma of 5476–5078 and 5291–4844 cal BC (Table 2). These dates together with the younger of two radiocarbon ages from Madawg Rockshelter (5673–5481 cal BC) suggest that the manufacture and use of shell beads at the inland sites was broadly contemporary with coastal shell middens such as Culverwell in Dorset, and at sites in Brittany and Ireland (Baylet) that have provided shell beads of similar species to those from the

Marine shell beads from three inland Later Mesolithic sites in Western Britain

203

English sites. The same may not be true for the majority of Scottish coastal middens for which the shell beads appear to be of somewhat younger age (Table 2). However, we would suggest some caution in this interpretation since many of these ages are based on conventionally dated charcoals (and see, for example, dates on bone from Sand).

Whether any chronological significance should be read into changes in the types (and styles) of beads is hard to say. In Britain, in the Early Mesolithic there is certainly a preponderance of perforated beads made of stone and other fossil material (Clark 1954; Barton et al. 1995; David 2007), whereas in the Later Mesolithic marine shell seems to have been a preferred medium for beads. In this context, it is interesting to reflect on the finds from Aveline’s Hole which would appear to suggest mixed use of fossil ammonites and perforated periwinkle shell in an individual burial. In this case it may be no coincidence that the dated human remains appear to occupy a chronologically young position in the Early Mesolithic (of around 8460–8140 cal BC). The same could also apply to the site of Prestatyn which has a date of 8200–7575 cal BC for the lithic assemblage although the oyster shell disc is probably too large to be considered a ‘bead’. A more convincing argument for a gradual transition in the use of stone to shell beads can be put forward to explain the intermediate age of pits with fossil Bayania and fresh marine shell at the French Middle Mesolithic site of Chaussée Tirancourt. Here the younger of the two pits (Fosse 2, 7031–6501 cal BC) contained perforated examples of fresh shells, including Littorina, while the older pit (Fosse 1: 7597-7355 and 7581–7179 cal BC) also contained fossil Bayania shell beads (Ducrocq & Ketterer 1995). In summary, although the dating evidence is sparse, it seems at present that sites with Trivia and Littorina bead assemblages (Table 2) appear to be younger than sites containing stone or fossil beads. Similarly, there are hints that the use of shell beads only becomes more common towards the end of the Early Mesolithic period, whereas stone beads are known from much older Mesolithic contexts (e.g. Star Carr, North Yorkshire).

Earlier in this paper we noted that marine shell ornaments were incorporated as grave goods in Later Mesolithic burials in Brittany and in association with fragmentary human remains scattered in middens in some of the Scottish coastal Later Mesolithic sites. This is, however, clearly not the case for the three inland locations described here. Three Holes Cave and Madawg Rockshelter both provide examples of beads found in association with domestic occupation debris and there is no evidence of associated human bone. The same may be true for King Arthur’s Cave, where no Mesolithic human remains have been recorded either inside or outside the cave. Until proven otherwise the presumption is therefore that they too derived from a domestic setting. If the beads from the three inland sites were not incorporated within burials, then what other reasons might be put forward to explain their presence? One interesting possibility is proposed here for Three Holes Cave where the shells survive in remarkably fresh and undamaged condition, in contrast to the faunal remains in the contemporary layer which are broken and heavily trampled. In particular, the shells from the 1990s collection came from the very top of the Later Mesolithic occupation level and their good condition cannot be explained by any special circumstances such as the proximity of large stones that may have afforded protection from trampling. We would suggest this is plausible evidence that the beads were deposited at the end of the occupation phase. A corollary of this, of course, is the question of whether they were accidentally dropped and lost, or whether they represent a deliberate act perhaps signalling metaphorical closure of the site on leaving, and/or as a marker for future visitors. It is hoped that further work on the spatial analysis of the 1950s assemblage will be able to provide additional data regarding their deposition.

A further clue as to the special significance of these three sites for bead finds may lie in the positions of each of the caves in relation to rivers (Roberts 1987; Barton & Roberts 2004, 352). Three Holes Cave, for instance, is linked to a major river catchment system (River Dart) and lies at an interesting sheltered mid-point between the coast and the upland massif of Dartmoor. Similarly, the Wye Valley sites are situated well inland, affording access to the uplands of the Black Mountains in the west and, looking eastwards, to the wide floodplain of the Severn Valley and the Gloucestershire lowlands. Each of these hinterland areas would have offered ample scope for gathering seasonally available plant foods and for hunting game, while the rivers themselves presented an obvious means by which people could transport food and resources in an upstream direction and towards the coast. In this respect it is also worth remarking that the distance from the shell sources may have been considerably greater than the distance to the nearest coastline. For example, King Arthur’s Cave and Madawg Rockshelter lie about 24-32 km upriver from the confluence of the rivers Wye and Severn, but about 104-112 km by the coastal/river route from the nearest sources of cowrie shells today (Ben Rowson pers. comm.). It is not unreasonable therefore to suggest that such cave sites may have acted as ‘staging posts’ for hunter-gatherers moving along the coast and river valleys. Under these circumstances, it would not be surprising if objects that were carefully ‘curated’ and carried around were sometimes lost or left behind. The presence of the same shell species at King Arthur’s Cave and Madawg Rockshelter also provides strong circumstantial evidence for the contemporary use of the sites by the same or related social groups. The slightly different combination of shells and range of cowrie sizes at Three Holes may imply subtly different selection processes exercised by other groups.

Whatever the explanation for the shell beads at these sites we believe that that such items must have been

No Stone Unturned: Papers in Honour of Roger Jacobi

204

invested with special meaning even without the presence of burials and irrespective of the low numbers recovered. It is therefore likely in our opinion that their presence signifies a deliberate act of disposal or deposition rather than random loss. We would guess that they were left as intentional markers, perhaps by peoples occupying overlapping territories and exchanging with groups with whom they shared similar symbolic and linguistic systems. The fact that the shells were of marine and not terrestrial origin, may also have been a deliberate reference to the coastal realm.

Finally, one intriguing aspect persistently raised by shell bead finds is whether their western distribution is a proxy for the spread of Later Mesolithic traditions along the Atlantic façade, or whether it simply reflects taphonomic bias and low archaeological visibility of such finds and sites, particularly in eastern Britain. We would suggest that the inland sites presented here may provide a pertinent test of such ideas. For instance, if shell beads were being moved by peoples inland, then this should be manifest elsewhere during the Later Mesolithic in areas proximate to coastlines where the shell species occur. According to present-day mapping, Trivia and Littorina distributions are not limited to western Britain, but extend around much of the southern and eastern coastlines (Wilkinson 2011; Smith 2013a, 2013b). The absence of shell beads at coastal locations where modern recovery techniques have been used, most notably at Goldcliff (Bell 2007), Prestatyn (Bell 2007), Westward Ho! (Balaam et al. 1987) and Ferriter’s Cove (Woodman et al. 1999), also implies that these items were by no means ubiquitous in the Later Mesolithic. However, we would expect that if shared social networks extended west-east as well as north-south, some evidence of marine shell beads ought to have emerged in eastern Britain by now. This is a situation that may change in the light of new research into the Mesolithic of these areas.

Three Holes Cave

During the excavation of 1955-56, a number of perforated shells were discovered. These included 14 cowrie shells (Trivia monacha) with a perforation at each end, and 11 Periwinkle shells (Littorina littoralis) each with a single perforation. The strongly parallels almost identical practices in the Obanian culture of Scotland. (8)