Radiocarbon Dating and Bayesian Modelling of the Late Iron Age Cremation Burial Cemetery at Westhampnett (West Sussex / GB)
The later iron age in northWest europe is characterised by major changes such as the appearance of forti fied oppida and new burial rites. This study examines the absolute chronology of the adoption of cremation burial in southern england through radiocarbon dating of human bone from the late iron age cremation cemetery at Westhampnett (West Sussex / GB). The earliest and second largest of its type in Britain, the cemetery occupied a low but prominent hill in the coastal plain, some 9 km from
... n, some 9 km from the modern coast and 4 km east of the modern city of Chichester ( fig. 1) . Some 161 graves, four shrines and numerous pyre sites were found when the cemetery was excavated in 1992 before a new road was built 1 . In the Early Bronze Age, a ring ditch had been constructed on the hill and this monument may have provided the focus for the iron age cemetery. The use of space was strongly defined: the shrines and pyres lay in discrete areas, and the graves were arranged around a circular space with later burials added to the periphery. Within this distribution, a minority of wellfurnished burials appeared to have been the foci for small clusters of graves. as the graves rarely intercut it is likely that they were marked in some way. The burials were unurned, but most of the graves contained pottery, including some early wheelmade vessels, which were placed as grave goods, and about a quarter contained brooches or other metal objects that had accompanied the deceased on the pyre ( fig. 2 ). The number of burials suggested that the cemetery was used fig. 1 The location of the Westhampnett cremation burial cemetery (West Sussex / GB). -(Map P. Lowther after Fitzpatrick 1997, fig. 1 ). 360 a. p. fitzpatrick et al. · radiocarbon Dating and Bayesian Modelling fig. 2 Westhampnett (West Sussex / GB). Two representative graves from the cemetery selected for dating. The modelled radiocarbon probabilities are also shown. -(After Fitzpatrick 1997, figs 70. 98). -Scale: metalwork 1:2; pottery 1:4. 361 archäologisches Korrespondenzblatt 47 · 2017 by a number of settlements. all the graves, pyres and pyrerelated features (i. e. features that contained pyre debris, but were not pyres) were wholeearth sampled in order to retrieve as much evidence as possible. In addition to the Bronze Age ring ditch and the Iron Age cemetery, the low hill was also the site of a small romanoBritish cremation cemetery (1 st 2 nd century) and a small angloSaxon inhumation cemetery (5 th 7 th centuries). Both later cemeteries were also fully excavated but are not considered here 2 . The excava tions around the hill yielded evidence for occupation for all the periods between the late upper Palaeolithic and the Anglo-Saxon, and some flints of Neolithic date were found on the hill itself 3 . changing chronologies: The 1997 reporT and subsEquEnt changEs to latE iron agE chronologiEs The monograph on the Westhampnett cemeteries was published in 1997. at that time it was not possible to radiocarbon date cremated human bone and the modest budget available for dating was applied to the Late Upper Palaeolithic to Early Bronze Age evidence from the road scheme, as it was considered that the scientific dates would be more helpful for these periods. The dating of the iron age cemetery relied on the pyre and grave goods, especially the brooches. This was not without its difficulties, whilst widely distributed in Continental Europe, where they are typical of the Lt D1 and D2 horizons, the brooches were relatively rare types in Britain. Three main categories are repre sented ( fig. 3 ). One-piece filiform brooches with external chords and short 2-or 4-coil springs predominate (Feugère Type 2a-b) 4 , but there were three brooches with internal chords belonging to the nauheim family (Feugère Type 5a-b) and seven boss-on-bow brooches of the Almgren 65 family (Feugère Type 8b) 5 . no brooches definitely of Middle La Tène construction were present. The cemetery dating consequently rested on the continental typo-chronologies for these brooches, which had very recently undergone a major revision, resulting in a significantly earlier start for Lt D1 6 . The frag mentation caused by the process of cremation, the transfer of the brittle objects to the grave, and sub sequent post-depositional disturbance (mainly by ploughing) made it difficult to distinguish between earlier or later varieties of iron filiform brooches. However, the relatively curved bows of the more complete brooches were considered closer to examples attributed to Lt D2a by A. Miron 7 . While acknowledging the earlier appearance of Nauheim and filiform types on the continent, the Westhampnett brooches were attributed to a period straddling Lt D1b-D2a and a date range for the cemetery of 100-40 BC was sug gested, with a preferred range within that of 90-50 BC 8 . Few other metal objects from the cemetery were from well-defined groups with relatively secure typologies and chronologies. These were a winged belt-hook, an iron razor, and a British gold coin. The pottery assem blage was considered to occupy a transitional date between the local Middle Iron Age Saucepan pottery tra dition (the St Catherine's Hill / Worthy Down style) and the Late Iron Age pottery of the »Aylesford-Swarling« tradition best known from cremation burials in south-east England. However, the typological affinities of the assemblage were considered to be as much with lower normandy, and to a lesser extent armorica, as within Britain. as the spatial organisation of the cemetery seemed to have been established early on, the site was regarded as only having a single phase. While cautious in relation to A. Miron's chronology for the Saar-Moselle region, the preferred date range of 90-50 BC for Westhampnett represented a significantly earlier dating for the adoption of cremation burial in England, which had previously been dated to after c. 50 BC. With its fine-grained and seemingly well-dated evidence, Westhampnett has become a reference site for late iron age mortuary rituals and chronology in southern Britain 9 , but 20 years on, the time has come to revisit the dating. 362 a. p. fitzpatrick et al. · radiocarbon Dating and Bayesian Modelling fig. 3 Westhampnett (West Sussex / GB). Principal types of brooches from the cemetery. The brooches illustrated are all of iron apart from those from graves 20622 (silver) and 20484, 20609 and 20675 (all copper alloy). -(After Fitzpatrick 1997, fig. 47). 363 archäologisches Korrespondenzblatt 47 · 2017 Firstly, a series of studies using Bayesian modelling to develop independent radiocarbon chronologies for later iron age artefact typologies have yielded earlier than expected dates for insular metalwork previously dated by reference to the continent 10 . Some of the dates obtained in these programmes are not without their difficulties and will be discussed elsewhere, but others raise questions about the accepted chronolo gies. To give one example, modelled dates for two Arras-culture inhumation burials in East yorkshire in northern England imply that one-piece filiform brooches of Late Iron Age form appeared before the mid-2 nd century BC 11 . Although the brooches in question have solid rather than framed catch-plates like the Westhampnett examples, and would generally be considered to be typologically later, they pose questions about the accepted dating of the series that need to be addressed. Secondly, the chronology of the presumed continental brooch prototypes is also open to question. The start of Lt D1b west of the Rhine, for which Nauheim and one-piece filiform brooches are one of the principal markers, is usually set around 120 BC, but a date of 130 BC has been advocated in some areas 12 . independent evidence that would resolve the matter is not only in short supply but also, particularly where it takes the form of den drochronological dates from loosely associated pieces of wood, it is sometimes open to question. Moreover, dating the start of Lt D1b establishes when these brooch types became widespread, not when they first appeared. In Lorraine, some contexts containing early varieties of Nauheim brooches are now thought to belong to an earlier phase of lt D1 13 . One-piece iron filiform brooches with external chords and long springs also occur in lt D1a graves in eastern France 14 and in the Bern region (CH) the type is sug gested to have appeared between 160 and 125 BC 15 . These brooches have longer springs than the exam ples from Westhampnett but show that filiform brooches developed from Lt C2 types rather than from the nauheim 16 . Lastly, boss-on-bow brooches related to the Almgren 65 -the latest form at Westhampnettare recorded in lt D1 contexts 17 , which may also have implications for the end date of the cemetery. Thirdly, radiocarbon dating of calcined bone is now widely applied to prehistoric burials of all periods across northWest europe 18 , including the iron age 19 . allied to Bayesian modelling, which allows archaeologists to date events to margins of decades rather than centuries 20 , it should be possible to develop robust chro nologies for cremation burial cemeteries that provide a similar level of precision to artefact typochronolo gies. Westhampnett provides a good test case; quite apart from the merits of revisiting any chronology that is now 20 years old in the light of current understanding, the presence of brooches in many of the graves allows us directly to confront the two forms of dating. any addition to the limited corpus of independent dates for the brooch types that form a mainstay of Late Iron Age chronologies on both sides of the Channel has to be of value. More widely, the study could open the way to the systematic dating of cremation burials across Central and Western Europe, which are widely seen as a diagnostic trait of the Late La Tène. it was clear, therefore, that successful radiocarbon dating of the Westhampnett cemetery could have a sig nificance reaching far beyond southern England. The radiocarbon dating programme reported on here was undertaken with the aid of a grant from the UK Natural Environment Research Council (NERC) Radiocarbon Facility for archaeology. radiocarbon dating crEmation burials The method of pretreating calcined bone to enable radiocarbon dating was first published by J. N. Lanting et al. 21 Before this, it was impossible to date calcined bone. an interlaboratory study by P. naysmith et al. showed good reproducibility of radiocarbon dates on cremated bone by six laboratories, demonstrating the technical method to be reliable 22 . unlike the radiocarbon age produced from extracted bone collagen (i. e. from an inhumation burial), which dates the death of the individual, the processes of cremation are 364 a. p. fitzpatrick et al. · radiocarbon Dating and Bayesian Modelling such that the radiocarbon measurement on this material directly dates the cremation. a potential problem for interpretation is the possibility for carbon exchange between the bioapatite, the datable fraction of the cremated bone, and the carbon (CO and CO 2 ) in the pyre »atmosphere« that is derived from the fuel source. This possibility for exchange has been demonstrated using both controlled laboratory cremation 23 and small realworld experiments on joints of animal meat 24 . While the processes are not fully understood, for an offset to occur requires the pyre fuel to contain »old« carbon (i. e. coal, old growth wood, or even peat). although the analysis of the charcoals found in the pyres at Westhampnett indicated a preponderance of roundwood from coppices that were managed to provide fuel 25 , the occurrence of iron nails and structural fittings in almost every pyre site and pyre-related feature strongly suggests that some wooden objects and / or seasoned timbers were reused as fuel 26 . In practice, the need to sustain temperatures of over 700 °C for several hours in order to cremate the corpse 27 and for the pyre to maintain its shape in order to provide the heat to do this means that seasoned timbers, probably of oak, were almost certainly used 28 . The rarity of charcoals from heartwood amongst the excavated charcoals at Westhampnett may be due to these tim bers being reduced to ash (presumably because they were left to burn overnight and it was only possible to approach the pyre the following morning). The smaller, coppiced, timbers may have been placed both within and around the pyre and it may be that only the smaller timbers that were placed around the pyre and fell away from it, survived as identifiable charcoals. Therefore as a precaution, replicate measurements on charred seeds or roundwood charcoal from ten graves were sought as a crosscheck on the cremated bone results. This has been shown to demonstrate the reliability of the bone measurements for accurately dating cremation burials 29 . Simulation models were run to estimate the number of samples needed to establish the date and duration of the cemetery, and to investigate whether there was discernible spatial and chronological patterning between the graves. As well as dating all graves with identifiable brooches, a representative sample of graves without brooches was also dated because brooches may not have featured in the funerary rites throughout the lifetime of the cemetery. Graves containing other typical late iron age objects such as locally made copies of armorican wheelmade pottery were also dated, as was the one grave with a funer ary monument (grave 20566). To accommodate a longer timespan than the 40-60 years proposed in 1997, multiple simulations were created that allowed for the cemetery to have been used for up to 100150 years. These models estimated that 50-55 dates were required, including replicate measurements on ten graves. mEthodology A total of 54 samples from 44 of the 161 cremation graves (27 %) were submitted to the Oxford Radiocarbon Accelerator Unit (ORAU) for dating by accelerator mass spectrometry (AMS). All the samples were single entities 30 . They consisted of a single fragment of cremated bone from each of the 44 burials ( fig. 4 ) and ten samples of nonhuman material from nine burials to test for offsets in the dates of the cremated bone. Six of the replicate samples consisted of carbonized cereal remains, along with a charred hazelnut shell, fragments of hazel and ash charcoal and a piece of cremated sheep long bone. All of these samples were found amongst the cremated bone. only a small minority of graves contained material suitable for replicate samples. The samples were pretreated following methods detailed in F. Brock et al. 31 and have been calibrated using the internationally agreed IntCal13 calibration curve of P. J. Reimer et al. 32 There was insufficient charred material for dating in one replicate sample (20601b), but three auto-replicate dates were generated as part of the ORAU internal Quality Assurance procedures, giving a total of 56 determinations (tab. 1). archäologisches Korrespondenzblatt 47 · 2017 Five of the eight pairs of samples have radiocarbon measurements that fail a chi-square test (tab. 2) 33 . as the paired samples were of shortlived material, there was no reason to expect any offset in age. For three pairs, the result from the replicate is clearly too early. The hazel charcoal from grave 20095 dates from the Neolithic, whilst the charred cereal from graves 20089 and 20245 is of earlier Iron Age date. The charred hazelnut shell from grave 20170 dates to the Late Iron Age, but fails the chi-square test and so appears to be residual in this grave. on the other hand, the late iron age cereal in grave 20018 would seem to be intrusive, as it is younger than its pair. These discrepancies are not particularly surprising in view of the abundant evidence for earlier activity nearby. This includes Neolithic flints from the hill itself and in general, there was a marked increase in settle ment on the West Sussex Coastal Plain through the Late Bronze Age and Iron Age 34 . Mention should also be made of a grain of bread wheat, submitted as a second replicate from grave 20170, which proved to be modern (OxA-32645, not listed in tab. 2). although rare in Britain during the iron age, bread wheat was widely cultivated in France 35 , so its presence in a cemetery with strong continental links would have been of considerable interest if the seed proved to be of iron age date. Truncation by ploughing and the fact that the burials were unurned made it comparatively easy for small materials like seeds to be intrusive amongst the cremated bone, as evidently happened in this case and with grave 20018 (above).