Beyond Stonehenge: Carn Menyn Quarry and the Origin and Date of Bluestone Extraction in the Preseli Hills of South-West Wales

Home , Carn Menyn, Preseli Hills

Beyond Stonehenge: Carn Menyn Quarry and the origin and date of bluestone extraction in the Preseli Hills of south-west Wales

Timothy Darvill1 & Geoff Wainwright2

Recent investigations at Stonehenge have been accompanied by new research on the origin o f the famous ‘bluestones’, a mixed assemblage o f rhyolites and dolerites that stand among the much taller sarsens. Some o f the rhyolite debitage has been traced to a quarry site at Craig Rhosyfelin near the Pembrokeshire coast; but fieldwork on the upland outcrops o f Carn Menyn has also provided evidence for dolerite extraction in the later third millennium BC, and for the production o f pillar-like blocks that resemble the Stonehenge bluestones in shape and size. Quarrying at Carn Menyn began much earlier, however, during the seventh millennium BC, suggesting that Mesolithic communities were the first to exploit the geology o f this remote upland location.

Keywords: Britain, Preseli Hills, Stonehenge, Mesolithic, Neolithic, megalithic, bluestone, dolerite, meta-mudstone, prehistoric quarrying Introduction For nearly a century it has been known that the 80 or so pillars of dolerite, rhyolite and tuff, collectively known as ‘bluestones’, that were incorporated into the structure of Stonehenge, Wiltshire, originated over 220km away to the north-west in the Preseli Hills of south west Wales (Thomas 1923). However, despite the publication of numerous excavations and surveys within and around Stonehenge (RCHME 1979; Richards 1990; Cleal et al. 1995; Darvill & Wainwright 2009; Parker Pearson 2012), it is only relatively recently that systematic archaeological fieldwork has been undertaken in west Wales to investigate the bluestone sources and their original context (Darvill & Wainwright 2002; Darvill et al.

1 Department o f Archaeology and Anthropology, Faculty o f Science and Technology, Bournemouth University, Fern Barrow, Poole BH 12 5BB, U K (Email: [email protected]) 2 March Pres, Ponfaen, Fishguard SA65 9TT, U K (Email: [email protected]) © Antiquity Publications Ltd. ANTIQUITY 88 (2014): 1099-1114 http://antiquity.ac.uk/ant/088/ant0881099.htm 1099 Beyond Stonehenge

2012b with earlier references; Nash et al. 2012; Parker Pearson 2012: 261-91; Cummings & Richards 2014). Careful geological detective work has established that outcrops at Craig Rhosyfelin, Pembrokeshire, match the main types of rhyolite found as waste debris at Stonehenge (Ixer & Bevins 201 la & b), and archaeological investigations around the outcrops are now ongoing (Parker Pearson 2012: 261—91). Attention has also been given to the dolerites at Stonehenge that have long been sourced to a series of exposed outcrops, locally known as earns, at the eastern end of the Preseli Ridge (Thomas 1923; Thorpe et al. 1991). A recent re-analysis of published geochemical data suggests at least three main kinds of dolerite are represented at Stonehenge, with Carn Goedog on the northern side of the ridge providing the closest match for just over half of the samples examined. Other outcrops, possibly including Cerrig Marchogion to the south-west and Carn Gyfrwy/Carn Menyn to the south-east, collectively contributed the remainder (Bevins etal. 2014; see below for further discussion). In July 2012 the authors excavated a trench on the southern flanks of Carn Menyn, Mynachlog Ddu, to investigate evidence of stone quarrying and a dolerite-working area. The investigation revealed a well-preserved stratigraphic sequence spanning the period from before 3000 BC through to 1000 BC that provides secure evidence for pre-Neolithic stone quarrying in the region and absolute dates for the extraction of dolerite pillars from a source outcrop high in the Preseli Hills. Following a brief overview of the background surveys, this report describes the stratigraphic sequence revealed at the Carn Menyn Quarry and assesses its wider implications.

Landscape surveys Between 2002 and 2011 the Strumble-Preseli Ancient Communities and Environment Study (SPACES) undertook extensive walk-over surveys combined with geophysical surveys, aerial photography and limited excavation within a study area of 450km2 extending from Mynydd Dinas in the west to Crymych in the east (Figure 1). The overarching aim was to provide an archaeological context for the source of the Stonehenge bluestones. More than 500 sites and monuments of archaeological interest were recorded, many of them for the first time. Two key findings rapidly became clear. First was that throughout the study area prehistoric communities had a close relationship with local stone, variously selecting and manipulating blocks for the construction of monuments including portal dolmens, chambered tombs, circles, standing stones, and sometimes just lifting slabs out of the ground as ‘propped rocks’. Second, within the eastern part of the study area our fieldwork supports a suggestion by Richard Bradley (2000: 92—96) that the arrangement of various bluestone lithologies used in the later stages of Stonehenge broadly replicates in microcosm the actual arrangement of stone types across the landscape of the Preseli Hills and surrounding areas. Thus, the dolerites of the Bluestone Horseshoe in the centre of Stonehenge derived from the central Preseli Ridge, while the various rhyolites and tuffs present in the Outer Bluestone Circle originated at outcrops within a wider catchment (Darvill 2006: 136-39). Geological investigations in the early 1990s by a team from the Open University used petrological and chemical analysis to show that two or three types of dolerite were represented amongst the pillar-stones and debris sampled from Stonehenge, and that these derived from © Antiquity Publications Ltd.

1100 Timothy Darvill & Geoff Wainwright

A tla n tic N o rth Sea O cean Research

SPACES C h3p s tu d y a re a

200

Land over 244m ASL

Figure 1. Location o f the SPACES study area, Pembrokeshire, U K (drawing by Vanessa Constant).

a number of source outcrops in the landscape of the eastern Preselis including Carn Menyn, Carn Goedog and Carn Breseb (Thorpe et al. 1991). A recendy published re-analysis of the original data, supplemented by a sample from Stone 34 recovered during our 2008 excavations at Stonehenge (Darvill & Wainwright 2009: 17) and additional samples from outcrops in the eastern Preselis, provides further resolution (Bevins et al. 2014). Based on concentrations of four selected elements (MgO, Ni, Cr and Fe2 0 3 ) this work confirmed and characterised three distinct types of dolerite amongst the Stonehenge samples (designated Groups 1—3). Bivariate analysis of the same elements in samples from the eastern Preselis (Bevins et al. 2014: figs. 13—16 with sample source areas listed on the keys but selectively quoted in tab. 3 and in the text) suggests that, with appropriate margins of uncertainty to © Antiquity Publications Ltd.

1101 Beyond Stonehenge accommodate sample variability, Group 1 correlates with samples from Carn Goedog on the northern side of the ridge (see Bevins et al. 2014: fig. 1); Group 2 with samples from Carn Bica, Carn Arthur, Carn Sian, Craig Talfynyndd and Cerrig Machogion/Mynydd-bach to the south-west; and Group 3 with samples from the adjacent outcrops of Carn Gyfrwy, Carn Menyn, Carn Breseb and Carn Gwr to the south-east. A Principle Components Analysis shows much the same picture, although the clusters are far from discrete. What becomes very clear is that the chemical composition of dolerites in the Preselis varies considerably from place to place. Uneven sampling during the original fieldwork means that some outcrops are very poorly represented. This is especially problematic in the case of Carn Menyn, which, although widely considered to be a major source, is represented by just one sample (CM1) that cannot possibly reflect the variability expected across the outcrop. Clearly, further informed sampling is needed, and Bevin et alls recent publication can be taken as an alternative rather than a revised interpretation of the Open University’s datasets (Bevins et al. 2014: 181, 192). Moreover, as only around 55 per cent of the 21 Stonehenge samples in the study (including unattributed debris as well as identified extant pillar-stones) can be attributed to a source at Carn Goedog (Bevins et al. 2014: 189), any claim that this was the main source for the Stonehenge bluestones must be treated with extreme caution. Notwithstanding the implications of these recent debates, Carn Menyn remains a probable source for some of the dolerite pillar-stones used at Stonehenge. Archaeologically, our surveys show that Carn Menyn was the focus of a great deal of activity in the later Neolithic and early Bronze Age. Burial cairns have been recorded at either end, standing stones on the outcrops, a walled enclosure around the highest point, and natural springheads elaborated through the creation of pools and the occasional application of rock art on the southern side. The surveys found that on the southern flanks of Carn Menyn there was a scatter of broken or abandoned dolerite pillar-stones of the same size and proportion as the stones present in the two visible structures at Stonehenge today (see Figure 2), as well as occasional hammer stones. Shallow hollows suggestive of quarry pits were also recognised and, on the basis of surface evidence, seemed to be for the extraction of a fine, light-grey coloured meta-mudstone (Darvill et al. 2008). Surface evidence including a broken pillar-stone, intercutting quarry pits and indications of dolerite extraction on a terrace on the southern slopes of Carn Menyn at an altitude of 310m asl was especially noteworthy (NGR: SN 143324). Samples of spotted dolerite taken from this area were submitted for analysis and described by Ixer and Bevins (2011a: 13) as petrologically indistinguishable from Stone 35a and very close to Stones SH34 and SH61 at Stonehenge. Accordingly, in July 2012 a single trench covering an area of 30m2 was excavated across the terrace in order to further examine the features visible on the surface (Figure 3). The re-analysis of geochemical data discussed above, published after the fieldwork described here was completed, goes some way to support the initial petrological work. No geochemical data is available for SFI35a, but Stones SH34 and SH61, which were compared favourably with the Carn Menyn rock samples, both fall within dolerite Group 3 (Bevins et al. 2014: tab. 3). They are thus not from Carn Goedog but most likely from outcrops at the eastern end of the Preseli Ridge which, as already discussed, includes outcrops in the area of Carn Menyn/Carn Gyfrwy (cf. Bevins et al. 2014: figs. 1, 9, 13-16). Further geochemical analyses using samples from © Antiquity Publications Ltd.

1102 Timothy Darvill & Geoff Wainwright

• Bluestone Oval Outer Bluestone Circle ♦ Carn Menyn Pillar-stones

600

500 E Research

in 400 in a> c -s' 300 u y

200 (Vc +->o LT> 100

0 200 400 600 800 1000 1200 1400 Stone width (mm)

Figure 2. Plot showing the width and thickness o f measurable dolerite pillar-stones in the Outer Bluestone Circle and Bluestone Oval at Stonehenge in relation to abandoned pillar-stones (some broken) on the slopes o f Carn Menyn, Pembrokeshire, recorded during the SPACES research 2002—2013 (chart by Timothy Darvill). our excavations will be undertaken during post-excavation analysis to redress the balance of samples included in the wider study.

The Carn Menyn Quarry sequence The 2012 excavation revealed a complicated and well-preserved stratigraphic sequence representing three main periods of activity (Figure 4) dated by a series of 12 radiocarbon determinations (Table 1). These were variously associated with the working of spotted dolerite, outcropping as an igneous intrusion at the northern end of the trench, and meta mudstone formed through the metamorphosis of mudstones of the Aber Mawr Formation along the edge of the dolerite dyke in the southern part of the trench. The earliest period of activity takes the form of a quarry pit for the exploitation of meta mudstone (Figure 5). It is one of a series of at least a dozen such pits that follow the edge of the dolerite dyke around the southern shoulder of the hill. Fire-setting appears to have been part of the extraction process (Figure 6). Four radiocarbon dates were determined on the abundant charcoal from oak stick-wood in the primary fills, CIO and C9. These fall into two groups. Two dates, 7060-6700 cal BC (UBA-22341: 7987±50 BP) and 6640- 6460 cal BC (UBA-21360: 7711±47 BP) from CIO, clearly belong to the first half of the sixth millennium BC, although they are not statistically consistent when compared using a standard x 2 test (T' = 16.2; T'(5%) = 3.8; v = 1). Two later dates, 5300-4980 cal BC (UBA-23207: 6170T54 BP) from CIO and 5480-5310 cal BC (UBA-21359: 6396T42) © Antiquity Publications Ltd.

1103 Beyond Stonehenge

Dolerite Pillar- Top of dolerite \ I Stone outcrop '

Dolerite Pillar- Stone Meta-mudstone quarry 2

□ M Contours in metres ASL Dolerite intrusion i____ | Meta-mudstone General area of clitter A

Figure 3. Plan o f the 2012 excavation at the Cam M enyn Quarry site, Pembrokeshire, U K (drawing by Vanessa Constant based on field plans by Timothy Darvill and Hubert Wilson). from C9, are clearly later fifth millennium BC, although again they are not statistically consistent (T7 = 10.8; T 7(5%) = 3.8; v = 1). Together these dates suggest at least two successive episodes of working separated by a full millennium. © Antiquity Publications Ltd.

1104 Timothy Darvill & Geoff Wainwright

Section 1 Section 2 SE NW

2270-2020-BC 2350-2060 BC s 2990-2640 BC 1380-1110 BC Section 3

1190-920 BC

2030-1770 BC 2200-1930 BC 2200-1960 BC

0 2 M D - Dolerite M - Meta-mudstone

Figure 4. Trench plan (above) and sections (below) ofthe C am Menyn Quarry site, Pembrokeshire, UK. The source positions o f samples used fo r radiocarbon dating are shown on the plan fo r visible layers and projected onto the stratigraphy represented on the sections (drawing by Vanessa Constant based on field plans and sections by Timothy Darvill and Hubert Wilson).

This modest quarry, no more than 5m wide, up to 6m long and 1.2m deep (Figure 3, meta-mudstone quarry 2), relates to the extraction of fine-textured, hard, light-coloured meta-mudstone. It is possible that some of the other pits recorded through fieldwork in the area are of similar date, but some could be earlier and there is also evidence for later working (see below). Hammer stones and numerous flakes were recovered during the excavation and are currently undergoing detailed study. Research is also under way to identify raw material and worked pieces in contemporary assemblages beyond Preseli, some of which may be reported in the literature as adinole’ or misidentified in hand-specimens as ‘rhyolite’. The early date for quarrying on the upper slopes of Carn Menyn accords with palaeo-environmental evidence from the Gors Fawr Bog close by to the south that shows © Antiquity Publications Ltd.

1105 Beyond Stonehenge

Table 1. Radiocarbon dates from the Cam Menyn Quarry.

Calibrated age BC Laboratory number Age BP (95.4% confidence)1 Phase and context

UBA-22341 7987+50 7060-6700 Early meta-mudstone working. FI. CIO UBA-21360 7711+47 6640-6460 Early meta-mudstone working. FI. CIO UBA-23207 6170+54 5300-4980 Early meta-mudstone working. FI. CIO UBA-21359 6396+42 5480-5310 Early meta-mudstone working. F 1. C9 UBA-22345 4188+34 2900-2640 Standing stone socket. F2. C l 1 UBA-23208 3795+34 2350-2060 Standing stone socket. F2. C l 1 UBA-22342 3724+32 2270-2020 Standing stone socket. F2. C l 1 UBA-22344 3685+32 2200-1960 Dolerite working. C7 UBA-21357 3672+45 2200-1930 Dolerite working. C7 UBA-22343 3567+32 2030-1770 Dolerite working. C7 UBA-21358 2979+30 1380-1110 Late meta-mudstone working. C6 UBA-21361 2871+34 1190-920 Late meta-mudstone working. ClOa

'Dates calibrated and rounded to 10 years using OxCal v4.2 (Bronk Ramsey 2013) with the IntCall3 calibration curve (Reimer et al. 2013).

Figure 5. General view o f the Carn M enyn Quarry fully excavated. Scale measures 2m (photograph by Timothy Darvill). signs of an early elm decline and pre-Neolithic disturbance of the landscape caused by fire management (Darvill et al. 2005: 20-22). The second period of activity at the Carn Menyn Quarry related to the working of dolerite in the later third millennium BC. Quarrying per se is not necessary for extracting spotted dolerite, as natural columnar blocks spall from the exposed earns to create a spread of clitter and scree. At the northern end of the 2012 trench was a shallow socket (F2) for a standing stone. The stone lies fallen to the south; the packing stones that secured it were still largely in place (Figure 7). Oak stick-wood charcoal from the fill of the socket provided three radiocarbon dates of 2900-2640 cal BC (UBA-22345: 4188+34 BP), © Antiquity Publications Ltd.

1106 Timothy Darvill & Geoff Wainwright Research

Figure 6. Detail o f natural rock outcrop subject to fire-setting in the base o f the quarry. Scale measures lm (photograph by Timothy Darvill).

2350-2060 cal BC (UBA-23208: 3795±34 BP) and 2270-2020 cal BC (UBA-22342: 3724±32 BP). The earliest date is not statistically consistent with the two later dates (T' = 113.3; T'(5%) = 6.0; v = 2) and must be considered residual. The two later dates are statistically consistent (T' = 2.3; T'(5%) = 3.8; v = 1) and give a combined date of 2280-2040 cal BC. Broken spotted dolerite pillar-stones lie to the north and south, with the example to the south showing clear traces of flaking to roughly shape the original block; it was presumably abandoned because it broke during this process (Figure 8). An area of preserved old ground surface towards the southern end of the trench (see Figure 4 for position) contained a scatter of spotted dolerite flakes and hammer stones with evidence of burning directly associated with it. Oak charcoal from this area provided three radiocarbon dates: 2200—1960 cal BC (UBA-22344: 3685T32 BP), 2200— 1930 cal BC (UBA-21357: 3672T45 BP) and 2030-1770 cal BC (UBA-22343: 3567T32 BP). Together these form an overlapping range, but they are not statistically consistent (T' = 7.6; T'(5%) = 6.0; v = 2). However, the two earlier dates are entirely consistent (T' = 0.1; T'(5%) = 3.8; v = 1) and give a combined date of 2190-1970 cal BC. These are the first secure dates for prehistoric dolerite working in the Preseli Hills and clearly indicate that Carn Menyn was being actively exploited in the late third millennium BC. The third and final period of activity represented in the Carn Menyn Quarry sequence again relates to the extraction of meta-mudstone. The earlier quarry was more or less silted- up by the end of the second millennium BC, but a new pit appears to have been dug to the south, partly overlapping its ancient predecessor. Blocks of stone and waste from the new quarry were deposited on top of the earlier fills and covered parts of the contemporary ground surface as a fairly uniform layer C6. This deposit is important as it sealed the evidence of dolerite working on the old ground surface C7. Indeed, blocks of meta-mudstone from this late quarry seem to have built up into what appears to be a small mound or cairn © Antiquity Publications Ltd.

1107 Beyond Stonehenge

Figure 7. Fallen standing stone with half-sectioned socket and evidence o f packing stones. Scale measures 0.3m (photograph by Timothy Darvill). See Figure 4 for location o f socket (F2, upper left o f figure).

Figure 8. Broken fallen pillar-stone with traces o f working; excavation trench behind. Scale measures 2m (photograph by Timothy Darvill).

just outside the trench to the north-east. Charcoal from within the upcast layer C6 was dated to 1380—1110 cal BC (UBA-21358: 2979±30 BP), while charcoal from the primary fill of this phase of the quarry (ClOa) yielded a slightly later date of 1190-920 cal BC (UBA-21361: 2871i34 BP). It is notable that a large cairn was built after 1420—1250 cal BC (UBA-18773: 3073±31 BP) at the western end of Carn Menyn; two fresh flakes © Antiquity Publications Ltd.

1108 Timothy Darvill & Geoff Wainwright of meta-mudstone had been inserted upright as a special deposit in the ground surface under the mound before its construction (Darvill et al. 2012b: 31). A foundation deposit below the multi-phase round barrow at Croesmihangel at the eastern end of the outcrops has been dated to 1930-1740 cal BC (UBA-18774: 3509T32 BP) (Darvill et al. 2012b: 34-38).

Discussion Research The Carn Menyn Quarry sequence has a number of important implications for understanding the special significance of the site itself, the early history of stone quarrying in Britain, and the interpretation of Stonehenge in its wider cultural setting. It has underlined, once again, the likelihood that Carn Menyn was one of the sources of the Stonehenge bluestones, with evidence for third-millennium quarrying of dolerite blocks and abandoned bluestones in situ on the hillslope. The main stratigraphically determined periods of activity described above appear to be more or less discrete episodes spatially connected by the power of place. Figure 9 shows a provisional Bayesian model of the radiocarbon dates based on the stratigraphic sequence. Five chronological phases are identified for the purposes of estimating the start and end date of key events in the overall sequence. The dates shown in italics are posterior density estimates derived from mathematical modelling and are quoted at 95% probability (Bronk Ramsey 2009). The extraction of meta-mudstone is probably represented as two stages in the sixth and fifth millennia BC, with Phase 1 modelled as starting at 7720-6660 cal BC and ending between 6640 and 5710 cal BC, and Phase 2 starting at 6290-5310 cal BC and ending at 5300—3480 cal BC. These early dates make the Carn Menyn Quarry the earliest-recorded securely dated stone extraction site in Britain. It may not be alone, however. A close parallel for the exploitation process found at Carn Menyn can be seen on the far side of Cardigan Bay at Mynydd Rhiw on the Lleyn Peninsula of north Wales. Tufaceous sediment (altered shale) adjacent to a dolerite intrusion was removed for the manufacture of axes, knives, points and scrapers (Floulder 1961; Burrow 2011). The date at which quarrying at Mynydd Rhiw started is not known, but pieces of stone from this source have been identified in a scatter dated to the period 7500-4000 BC on Bardsey Island, some 13km to the south west (Edmonds et al. 2004); radiocarbon dates from quarry pits at Mynydd Rhiw show exploitation during the middle and late fourth millennium BC, perhaps with some re-use of the site in the late second millennium BC, when half a dozen round barrows were built on the hill (Burrow 2011: 255). Further north, a recently discovered chert quarry at Burnetland Hill in the Scottish Borders has been dated to 4045-3975 cal BC (Bjarke Ballin & Ward 2013). On a wider front, the discovery of formal quarrying in the British late Mesolithic adds significantly to the growing list of monuments and structures from this period. Particular note may be made of the monumental alignment of 12 possible post-sockets set within larger pits at Warren Field, Crathes, Aberdeenshire, first constructed c. 7800 BC and re-cut c. 4000 BC. These have been variously interpreted as a ceremonial or calendrical structure of some kind (Murray et al. 2009; Gaffney et al. 2013). A large pit or shaft 3m across and 2m deep was found at Little Dartmouth Farm, Devon, with a primary fill dated to © Antiquity Publications Ltd.

1109 Beyond Stonehenge

P»Ctf ?Bry Rnmm<20m axw ilW* X1 20i3y

E nd 5 ± = — - U8A-21361 -£r UBA-21358 -5r Phase 4 - Late meta-mudstone extraction Start 5 — •*- E nd 4 ± r ~ UBA-22343 -A- ' UBA-21357 Hi- U8A-22344 H

Phase 3 - Dolerite working Start 4 jk

E nd 3 UBA-22342 -ir UBA-23208 _Jk_

UBA-22345 JIL.

Phase 2 - Standing stone

Start 3

E nd 2 UBA-23207

UBA-21359

Phase 2 - Early meta-mudstone extraction

Start 2 — E nd 1

UBA-21360 - ± UBA-22341 Phase 1 - Early meta-mudstone extraction

Start 1

_L» a _ j 12000 10000 8000 “6000 4000 2060' 1BC/1AD

Modelled date (BC/AD)

Figure 9. Model o f the available radiocarbon dates for activity at the Cam Menyn Quarry site (plot compiled by Timothy Darvill. Bayesian modelling is based on Bronk Ramsey 2009).

5640—5510 cal BC (SUERC-24615: 6645±35 BP), a re-cut dated to 4950—4740 cal BC (SUERC-24617: 5970±35 BP), and a small assemblage of worked flint and chert (Tingle 2013). Best-known of all are the three certain postholes, a posthole later re-cut as a pit, and a possible tree-throw hollow found around 200m north-west of Stonehenge during the construction and expansion of the visitor car park in 1966 and 1988 (Cleal et al. 1995: 43—47). Not all of these features were contemporary; episodic activity rather than continuous use is most likely. O f the dated features, Pit A is the earliest at 8840-7740 cal BC (HAR-455: 9130±180 BP) with Pit 9580 cut soon after or contemporary with it. This complicated feature was initially lm across and 1.3m deep, and perhaps served as a posthole before being re-profiled as a broad shallow pit 1.9m across and 0.7m deep. This secondary feature is dated by three radiocarbon determinations from bottom to top of 8260—7750 cal © Antiquity Publications Ltd.

1110 Timothy Darvill & Geoff Wainwright

BC (GU-5109: 8880T80 BP), 7740-7370 cal BC (OxA-4219: 8520T80 BP) and 7600- 7180 cal BC (OxA-4220: 8 4 0 0 il0 0 BP). A piece o f‘rhyolite’ was reportedly found in the upper fill associated with the last of these dates, although its petrography has not yet been studied (Cleal etal. 1995: 47). The latest feature is Pit B at 7460-6650 cal BC (HAR-456: 8090T140 BP). The extension of the cultural sequence at Carn Menyn back into the sixth and fifth millennia BC brings the pattern of activity there into close accord with what is known Research of the Stonehenge area, and begins to flesh out the bare bones of parallel but connected developments in the two areas. Stonehenge was not constructed in a virgin landscape. Pine charcoal residual within a later feature in the centre of Stonehenge has been dated to 7310— 7070 cal BC (OxA-18655: 8183T36 BP) (Darvill & Wainwright 2009: 12). It suggests activity in the area later occupied by the stone circles that are broadly contemporary with the aforementioned posts set up 200m to the north-west. Just 3km east of Stonehenge, excavations at Blick Mead below Vespasian’s Camp, Amesbury, have brought to light what appears to be debris from a riverside settlement around a spring. Provisionally dated to the period 6250 BC through to 4000 BC, its occupation is contemporary with the quarries at Carn Menyn; one worked piece of slate/mudstone has been reported from Blick Mead, although it has not yet been accurately provenanced to source (Jacques et al. 2012). The second main period of activity at the Carn Menyn Quarry also divides into two phases: the erection of a standing stone and dolerite-working. Modelling suggests that the standing stone was set up before the dolerite-working took place, but in the absence of constraining dates the modelled start date for the standing stone (Figure 9, Phase 3) is wide at 4090—2640 cal BC. The end-date for Phase 3, 2210—2000 cal BC, is artificially constrained by the dates for the dolerite-working, which could in fact be contemporary. The modelled start date for the dolerite-working (Figure 9, Phase 4) is 2160—1960 cal BC, ending at 2030—1610 cal BC. Although geochemical data are not yet available, petrological studies support the proposition that rock outcropping at Carn Menyn matches some of the spotted dolerites found at Stonehenge, where they had been set up as pillar-stones in the late third millennium BC. The shape and size of the stones also match. Local uses of Carn Menyn dolerite pillars might also be considered, but it is notable that no local monuments incorporated shaped pillar-stones of the kind that were found abandoned during field survey at Carn Menyn and which were used as structural components at Stonehenge. These findings finally lay to rest the theory that the bluestones arrived on Salisbury Plain through glacial action (Kellaway 1971), an idea that has been remarkably persistent and periodically revived (e.g. John 2008) despite being comprehensively discredited by geologists, geomorphologists and glaciologists more than a decade ago (Green 1997: 264; Scourse 1997; Clark et al. 2004; Bowen 2005: 147—48). We cannot say whether dolerite pillars were present in earlier stages of Stonehenge, but they still stand as part of the decayed remains of the Bluestone Horseshoe and the Outer Bluestone Circle. It is also important to recognise that several other lithologies are represented within the assemblage of bluestones from Stonehenge and that these derive from other source outcrops that have yet to be identified (but see Ixer & Bevins 2011b for ongoing research). Continued activity within and around Stonehenge through the second millennium BC is well represented (Darvill 2006: 157-79), and ends with the digging of two rings of pits © Antiquity Publications Ltd.

1111 Beyond Stonehenge around the central setting (the Y and Z holes) and the construction of numerous round barrows on the ridges and hilltops overlooking the site. The same thing seems to happen at Carn Menyn with the re-working of the earlier quarry pits for the extraction of meta mudstone (Figure 9, Phase 5), modelled with a start date of 1800-1090 cal BC and an end date of 1210—420 cal BC. At much the same time the Carn Menyn Cairn and the Croesmihangel Round Barrow were constructed at the western and eastern ends of the outcrops respectively (Darvill et at. 2012b). Recognition of the early date of quarrying on Carn Menyn and the confirmation of bluestone working there contemporary with the movement of pillar-stones to Stonehenge means that the challenge now is to explain what was so special about the Preseli bluestones, and why and for what purpose they were transported to Salisbury Plain for exclusive use at Stonehenge. As discussed elsewhere, we believe the bluestones were powerful items within a structure whose framework was formed by the great sarsen trilithons and the Sarsen Circle (Darvill & Wainwright 2011). Together, the sarsen elements replicate in stone the kind of structure that in many cases elsewhere had been built in timber (Wainwright & Longworth 1971: 207-16; Wainwright 1989: 115—34). Once in place the framework remained static, but the bluestones were periodically moved and reorganised as the Double Bluestone Circle (Q and R holes) which later became the Outer Bluestone Circle, and the Inner Bluestone Circle which later became the Bluestone Oval and finally, with the removal of stones from the north-eastern sector, the Bluestone Horseshoe seen today. Many explanations as to why the bluestones were considered sufficiently important and meaningful to move from Wales to Wiltshire can be proposed, and there may be more than one reason. The demonstrable antiquity of stone extraction on Carn Menyn, long before the building of Stonehenge began, tells us something about the ancestral significance and power of the landscape from which the bluestones were taken. Perhaps Mynydd Preseli was the home of the gods: the Mount Olympus of Neolithic Britain. But we also believe that the association between bluestones and healing springs in the Preseli Hills was important (cf. Jones 1992), and something that resonates with long-standing oral traditions that were first written down in the thirteenth century AD (Piggott 1941). Springs were a significant and persistent feature of the Stonehenge landscape, as the recent work at Blick Mead shows (Jacques et al. 2012). Soon after the bluestones were installed at Stonehenge (Stage 2) the central structure was linked by an Avenue to Stonehenge Bottom and the River Avon (Stage 3), thereby fixing and formalising the relationship to water (Darvill et al. 2012a: 1035). The idea that powerful stones were moved from their source outcrops on a special, ancestral or sacred place to ‘franchise’ distant shrines and temples finds parallels in West African societies and elsewhere (Insoll 2006). We propose that, after the earthwork enclosure at Stonehenge ceased to be a major cremation cemetery sometime about 2500 BC, bluestones from Carn Menyn and other nearby outcrops in west Wales were brought to Stonehenge and set up within a temple whose structure had already been built from sarsen stones. From that time onwards, pilgrims and travellers were drawn to Stonehenge because of the special properties that had empowered Stonehenge to provide pastoral and medical care of both body and soul: tending the wounded, treating the sick, calming troubled minds, promoting fecundity, assisting and celebrating births and protecting people against malevolent forces in a dangerous and uncertain world. The bluestones hold the key to the meaning of Stonehenge, © Antiquity Publications Ltd.

1112 Timothy Darvill & Geoff Wainwright and Preseli was the special place from whence they came at a high cost to society in labour and time, as befitted such important talismans.

Acknowledgements -Cl u We thank Fran Murphy, Hubert Wilson, Judith Wainwright and Vanessa Constant for their assistance with the bn excavation at Carn Menyn in 2012; Rob Ixer and Richard Bevins for advice on geological sources; Ralph

References D arvill, T. & G. W ainw right. 2002. SPACES—exploring the Neolithic landscapes in Bevins, R.E., R.A. Ixer & N.J.G. Pearce. 2014. Cam the Strumble-Preseli area of southwest Wales. Goedog is the likely major source of Stonehenge Antiquity 76: 623—24. doleritic bluestones: evidence based on compatible element geochemistry and Principal Component — 2009. Stonehenge: excavations in 2008. The Analysis. Journal o f Archaeological Science 42: Antiquaries Journal 89: 1-19. 179-93. http://dx.doi.org/10.1016/ http://dx.doi.org/10.1017/ j.jas.2013.11.009 S000358150900002X — 2011. The stones of Stonehenge. Current Archaeology Bjarke Ballin, T. & T. WARD. 2013. Burnetland Hill chert quarry: a Mesolithic extraction site in the 252: 28-35. Scottish Borders. The Quarry 9: 3—23. D arvill, T., D .M . Evans, R. Fyfe & G. W ainw right. 2005. Strumble-Preseli Ancient Communities and Bow en, D .Q . 2005. South Wales, in C.A. Lewis & A.E. Richards (ed.) The glaciations o f Wales and Environment Study (SPACES): fourth report 2005. adjacent areas: 145—64. Logaston: Logaston Press. Archaeology in Wales 45: 17-23. BRADLEY, R. 2000. An archaeology of natural places. D arvill, T., G. W ainw right, K. Armstrong & R. London: Routledge. Ixer. 2008. Strumble-Preseli Ancient Communities and Environment Study (SPACES): sixth report Bronk Ramsey, C. 2009. Bayesian analysis of 2007-08. Archaeology in Wales 48: 47-56. radiocarbon dates. Radiocarbon 51: 337—60. D arvill, T., P. M arshall, M . Parker Pearson & G. -2013. OxCal radiocarbon calibration program v4.2. WAINWRIGHT. 2012a. Stonehenge remodelled. Oxford: Research Laboratory for Archaeology and Antiquity 86: 1021^0. the History of Art, Oxford University. D arvill, T., G. W ainw right, F. LOth & N. BURROW, S. 2011. The Mynydd Rhiw quarry site: MOLLER-SCHEESSEL. 2012b. Strumble-Preseli recent work and its implications, in V. Davis & M. Ancient Communities and Environment Study Edmonds (ed.) Stone axe studies III: 247-60. (SPACES): seventh report 2009—11 .Archaeology in Oxford: Oxbow. Wales 51: 27-44. C lark, C.D., P.L. G ibbard & J. Rose. 2004. E d m o n d s, M „ R. J o h n so n , E. La T robe-Bateman, J. Pleistocene glacial limits in England, Scotland and G riffith & G. W arren. 2004. Bardsey Island. Wales, inj. Ehlers & EL. Gibbard (ed.) Quaternary Archaeology in Wales 44: 146M7. glaciations extent and chronology. Part I: Europe: 47—82. London: Elsevier. Gaffney, V., S. Fit c h , E. Ramsey, R. Yorston, E. C h ’n g , E. Baldw in, R. Bates, C. Gaffney, C. C leal, R., K.E. W alker & R. M ontague. 1995. R uggles, T. Sparrow, A. M cM illan, D. C owley, Stonehenge in its landscape. Twentieth-century S. Fraser, C. M urray, H. M urray, E. H opla & excavations (English Heritage Archaeological A. H oward. 2013. Time and a place: a luni-solar Reports 10). London: English Heritage. ‘time-reckoner’ from 8th millennium BC Scotland. C um m ings, V. & C. Richards. 2014. Exploring Internet Archaeology 34. Neolithic construction at Garn Turne. Current http://dx.doi.org/10.11141 /ia.34.1 Archaeology 286: 32—35. G reen, C.P. 1997. The provenance of rocks used in the D arvill, T. 2006. Stonehenge: the biography o f a construction of Stonehenge. Proceedings o f the landscape. Stroud: Tempus. British Academy 92: 257-70.

1113 Beyond Stonehenge

HOULDER, C.H. 1961. The excavation of a Neolithic RCHME Royal Commission on the Historical stone implement factory on Mynydd Rhiw in Monuments of England. 1979. Stonehenge and its Caernarvonshire. Proceedings o f the Prehistoric environs. Edinburgh: Edinburgh University Press. Society 27: 108-43. Reimer, P.J., E. Bard, A. Bayliss, J.W. Beck, P.G. INSOLL, T. 2006. Shrine franchising and the Neolithic Blackwell, C. Bronk Ramsey, P.M. G rootes, in the British Isles: some observations based upon T.P. Guilderson, H. H aflidason, I. H ajdas, C. the Tallensi, northern Ghana. Cambridge Hatt£, T.J. H eaton, D.L. H offmann, A.G. ArchaeologicalJournal 16(2): 223—38. H ogg, K.A. H ughen, K.F. Kaiser, B. Kromer, http://dx.doi.org/10.1017/S0959774306000138 S.W. Manning, M. N iu, R.W. Reimer, D.A. IXER, R. & R. BEVINS. 2011a. The detailed petrology of Richards, E.M. Scott, J.R. Southon, R.A. six orthostats from the Bluestone Circle, Staff, C.S.M. T urney & J. van der Plicht. Stonehenge. Wiltshire Archaeological and Natural 2013. IntCall3 and Marine 13 radiocarbon age History Magazine 104: 1-14. calibration curves 0-50,000 years cal BP. Radiocarbon 55: 1869-87. http://dx.doi.org/ -2011b. Craig Rhos-y-Felin, Pont Saeson is the 10.2458/azu_js_rc.5 5.16947 dominant source of the Stonehenge rhyolitic ‘debitage’. Archaeology in Wales 50: 21—32. RICHARDS, J. 1990. The Stonehenge environs project (HBMCE Archaeological Reports 16). London: Jacques, D „ T. Phillips & T. Lyons. 2012. Vespasian’s English Heritage Camp: cradle of Stonehenge? Current Archaeology 271:28-33. SCOURSE, J.D. 1997. Transport of the Stonehenge bluestones: testing the glacial hypothesis. J ohn, B. 2008. The bluestone enigma. Stonehenge, Preseli Proceedings o f the British Academy 92: 271—314. and the Ice Age. Newport: Greencroft. THOMAS, H.H. 1923. The source of the stones of JONES, F. 1992. The holy wells o f Wales. Cardiff: Stonehenge. The Antiquaries Journal 3: 239-60. University of Wales Press. http://dx.doi.org/10.1017/S0003581500005096 KELLAWAY, G.A. 1971. Glaciation and the stones of T horpe, R.S., O. W illiams-T horpe, D.G. Jenkins & Stonehenge. Nature 233: 30-35. http://dx.doi.org/ J. WATSON. 1991. The geological sources and 10.1038/233030a0 transport of the bluestones of Stonehenge, M urray, H.K., J.C. M urray & S.M.A. Fraser. 2009. Wiltshire, UK. Proceedings of the Prehistoric Society A tale of the unknown unknowns: a Mesolithic pit 57: 103-57. http://dx.doi.org/10.1017/ alignment and a Neolithic timber hall at Warren S0079497X00004527 Field, Crathes, Aberdeenshire. Oxford: Oxbow. TINGLE, M. 2013. A new Mesolithic pit site and Beaker N ash, G. A. Stanford, I. T herriault & T. features at Little Dartmouth Farm, Devon. Past 74: WELLICOME. 2012. Transcending ritual boundaries 5-6. from dolmen to menhir: the excavation of the WAINWRIGHT, G. 1989. The henge monuments. Trefael Stone, south-west Wales. Archaeology in London: Thames & Hudson. Wales 51:51-62. W ainwright, G. & I.H. Longworth. 1971. Parker Pearson, M. 2012. Stonehenge: exploring the Durrington Walls: excavations 1966—1968 (Reports greatest Stone Age mystery. London: Simon & of the Research Committee of the Society of Schuster. Antiquaries of London 29). London: Society of PlGGOTT, S. 1941. The sources of Geoffrey of Antiquaries. Monmouth. II. The Stonehenge story. Antiquity 15: 305-19.

Received: 23 October 2013; Accepted: 10 December 2013; Revised: 12 February 2014

1114 Copyright of Antiquity is the property of Cambridge University Press and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.