5. Case Studies

5.1 King Alfred Way, Newton Poppleford

Between October 2014 and October 2017, AC Archaeology undertook archaeological investigations on agricultural land south of King Alfred Way, Newton Poppleford, east of Exeter, Devon (Fig. 31). The work, in response to a planning condition set by East Devon District Council for a residential development on the site, was funded by the developer Cavanna Homes (Rainbird and Lichtenstein 2018).

In 2014 archaeological trench evaluations identified several pits and postholes containing Middle Neolithic pottery, together with buried cultivation soils and large numbers of worked prehistoric flints.

Subsequent mitigation in 2017 resulted in the opening of two excavation areas positioned to investigate probable prehistoric features.

Following stripping within Area 1, an unsuspected ring ditch with an internal diameter of c. 8m was revealed (F1004; Fig. 32a). Sparse finds in the U-shaped ditch — 0.9m–1.5m wide; 0.4–0.6m deep (Fig. 32b) — consisted of 56 pieces of worked flint and a single sherd of later Iron Age pottery. Although no mound, outer or internal bank material, or other features survived in its interior, sections across the ditch showed that its primary fill probably derived from a barrow mound or internal bank that had originally existed.

After the ring ditch had infilled, two stratigraphically-related graves (F1034 and F1041) were dug cutting its inner lip.

The earliest of them (F1041) contained the calcined bones of an adult human, a single sherd of Peterborough Ware and a worked flint. The second grave (F1034) also contained cremated adult human remains, along with three more conjoining ceramic sherds from the same vessel as the sherd from F1041, worked flint, a piece of iron slag, fired clay and two small sherds of later ceramics.

Given that cremation deposits associated with Middle Neolithic ceramics are unexpected and unusual, and the fact that ring ditches of this form would commonly be assigned a Bronze Age date, a radiocarbon determination was obtained from a single fragment of calcined bone from F1041 to determine its age (Table 5).^[Footnote]

The result was calibrated with Bchron (Haslett and Parnell 2008) using IntCal20 (Reimer et al. 2020). The calibrated radiocarbon date, 3335–3030 cal BC (95% probability; Fig. 33) provides a terminus ante quem for the construction of the ring ditch and is the first example of this monument class in south-west England to have scientific dating.

Although the ring-ditch is a significant archaeological discovery made as a result of the planning process in its own right, radiocarbon dating has situated the monument as belonging to a diverse group of circular earthworks dating the 34–31st centuries cal BC in Britain.

Previously the south-west extent of Neolithic round mounds and ring-ditches in England was believed to be Dorset and Wiltshire (Kinnes 1979; Leary et al. 2010), and so dating of the monument at Newton Poppleford has now extended this distribution into Devon.

Footnote

The reported result is a conventional radiocarbon age (Stuiver and Polach 1977). The laboratory maintains a continual programme of quality assurance procedures, in addition to participation in international inter-comparisons (Scott et al. 2017). These tests indicate no laboratory offsets and demonstrate the validity of the precision quoted. The sample dated in SUERC was processed and measured by Accelerator Mass Spectrometry, according to the procedures described in Dunbar et al. (2016).↩

5.2 The skeleton in the car park

Excavation in 2012 by the University of Leicester on the site of Grey Friars church in Leicester (Fig. 34), demolished after the Reformation and subsequently built over, revealed the remains of the friary church with a grave in a high-status position beneath the choir (Buckley et al. 2013; Fig. 35).

The project, to search for the lost grave of King Richard III, the last English king to die in battle, had uncovered a battle-scarred skeleton with spinal curvature.

How old are the bones found under the Grey Friars church? Clearly, they cannot be any more recent than the Dissolution of AD 1538. But if they are earlier than AD 1485, then they cannot be the remains of Richard III.

Four samples of rib bone from the individual interred in the grave beneath the choir were dated at the Oxford Radiocarbon Accelerator Unit (ORAU) and Scottish Universities Environmental Research Centre (SUERC).

At SUERC the samples were pretreated following a modified Longin (1971) method. They were then combusted to carbon dioxide (Vandeputte et al. 1996), graphitised (Slota et al. 1987), and measured by Accelerator Mass Spectrometry (AMS) (Xu et al. 2004).

The samples at ORAU were pretreated and combusted as described in Brock et al. (2010b), graphitised (Dee and Bronk Ramsey 2000) and dated by AMS (Bronk Ramsey et al. 2004).

The reported conventional radiocarbon ages (Stuiver and Polach 1977) and stable isotope results are shown in Table 6.

Both laboratories maintain continual programmes of quality assurance procedures, in addition to participation in international inter-comparisons (Scott et al. 2017). These tests indicate no laboratory offsets and demonstrate the validity of the precision quoted.

The four radiocarbon determinations are statistically consistent at the 5% significance level (T'=3.8; n=3; T'(5%)=7.8; Ward and Wilson 1978) and thus a weighted mean (Grey Friars 2012: 451±11 BP) has been calculated as providing the best estimate for the age of the individual.

Calibration of the weighted mean using the probability method (Stuiver and Reimer 1993), and IntCal20 (Reimer et al. 2020) provides a calibrated date for the death of this person of cal AD 1432–1454 (95% probability; Fig. 36). Clearly taken at face value this would indicate that the remains cannot be those of Richard III, but the stable isotope measurements indicate that this individual had a protein-rich diet that included a significant amount of non-terrestrial food sources.

Diet-induced radiocarbon offsets when an individual has taken up carbon from a reservoir not in equilibrium with the terrestrial biosphere can have implications for accurately estimating when a person died.

In order to correct for the dietary information that the stable isotopes provide we need to estimate the contribution of non-terrestrial sources in the diet of the dated individual.

A proportional diet profile for the Grey Friars 2012 individual was established using the proportional mixing model FRUITS (Food Reconstruction Using Isotopic Transferred Signals) v 2.1.1 (Fernandes et al. 2014).

Baseline isotopic values for food sources used in the FRUITS modelling (Table 7) were drawn from medieval faunal isotope values (Müldner and Richards 2005). As isotopic values for terrestrial vegetation were unavailable, a proxy was derived from the average cattle and sheep isotopic values (Müldner and Richards 2005) less trophic enrichments of 1‰ (δ¹³C) and 3‰ (δ¹⁵N).

Human diet-to-consumer enrichment values were set at 4.0±0.5‰ (δ¹³C; Fernandes 2016) and 5.0±0.5‰ (δ¹⁵N; O’Connell et al. 2012), with the weight and concentration of the diet sources comprising 100%, following Fernandes et al. (2014) for unrouted diet models.

The FRUITS source-proportional mixing model (Table 8; Fig. 37) indicates that animal-derived protein contributed 77.6% of the individual’s diet. Such protein-rich diets are notable in medieval populations, where meat and fish made up a significant proportion of the food intake for aristocrats, clergy and wealthy merchants (Müldner and Richards 2005, 40).

Eel and freshwater fish (56.7±6.9%) account for the greatest proportion of the diet of the individual from Grey Friars, with that from terrestrial protein being considerably smaller (18.3±14%). The contribution from marine fish is estimated to be just 2.7±2.3% of diet, suggesting that diet-derived offsets from marine radiocarbon reservoirs will be minimal; however, the diet derived offsets from eel and freshwater fish could be significant.

We can account for the different sources of carbon in the measured radiocarbon age of the individual from Grey Friars by calibrating the result in OxCal 4.4 (Bronk Ramsey 2009a), using a mixture of the terrestrial calibration curve (Reimer et al. 2020), the marine calibration curve (Heaton et al. 2020) with a ΔR correction of −169±56 BP (Harkness 1983) and the terrestrial calibration curve offset for an appropriate freshwater reservoir in the proportions suggested by the dietary analysis (Fig. 37).

Given the lack of all-embracing research on freshwater offsets in England, we have implemented a uniform ΔR distribution from 750–0 BP based on the range of offset in archaeological fish bone from Coppergate, Earith and Flixborough (2±54 BP to 703±32 BP) identified by Keaveney and Reimer (2012, table 1). The calibration is constrained by including the prior information that the individual must have been interred before AD 1538.

This suggests that the individual excavated from Grey Friars died in cal AD 1459–1539 (95% probability; Fig. 38a). Furthermore, constraining the death of the individual to be before AD 1486 shows good agreement (Amodel: 65; Fig. 38b).

The very wide calibrated date shown in outline in these graphs reflects the great uncertainty on the freshwater reservoir correction, which in this case can be constrained by the historical evidence.

A perfect mitochondrial DNA match was found between the sequence obtained from the Grey Friars skeleton and one living relative of Richard III, and a single-base substitution was found when compared with a second relative. However, Y-chromosome haplotypes from male-line relatives and the Grey Friars skeleton do not match, which could be attributed to a false paternity event occurring in any of the intervening generations (King et al. 2014).

Combining all the non-genetic data (radiocarbon, estimated age at death, sex, presence of scoliosis and presence of perimortem wounds) together with the genetic data (mtDNA and Y-chromosome) in a Bayesian framework, however, does provide extremely strong support for the probability that the skeleton in the car park is that of King Richard III (King et al. 2014; Fig. 39).

5.3 A slice across 'Cataractonium' Roman town

The Leeming to Barton A1 road scheme comprised 19km of road improvements to upgrade the existing dual carriageway to motorway status. Upgrading of the A1 in this part of North Yorkshire passed through an area of known prehistoric and historic significance, including the scheduled Roman town of Cataractonium.

The aim of the archaeological investigations, undertaken by Northern Archaeological Associates and funded by Highways England, was to mitigate for the impact of road construction works on the extant archaeological remains.

The quantity and quality of the evidence for Roman period activity was exceptional and, in recognition of this, a number of research themes were formulated as part of post-excavation programme, including one focussing on death, burial and identity (Speed and Holst 2018).

At the Roman settlement of Cataractonium (Fig. 40), excavations at Brough Park (Field 172), Fort Bridge (Field 176FB), Brompton West (Fields 177/178) and Brompton East (Field 179) recovered 26 inhumations (Fig. 41) and nine cremation burials, including a bustum burial that formed part of a small cremation cemetery located at Brough Park (Field 172).

At Brompton West (Fields 177/178) a larger number of burials was recorded to the rear of the northern suburb of Cataractonium.

In order to understand the chronology of burials at Roman Cataractonium and how burials here related to other burials excavated during the road scheme and to Roman burial practices elsewhere, particularly in the north of England, an extensive radiocarbon dating programme was undertaken (Table 9).

The samples dated at SUERC were processed and measured by Accelerator Mass Spectrometry, according to the procedures described in Dunbar et al. (2016). The reported results are conventional radiocarbon ages (Stuiver and Polach 1977).

The laboratory maintains a continual programme of quality assurance procedures, in addition to participation in international inter-comparisons (Scott 2017). These tests indicate no laboratory offsets and demonstrate the validity of the precision quoted.

When we calibrate a radiocarbon measurement (Fig. 42 – outline distribution), we assume that the calendar date of the sample is equally likely to fall at any point on the calibration curve. For one sample, this is a reasonable assumption; but as soon as we wish to calibrate a second measurement from a site, this assumption is no longer valid. The radiocarbon measurements on burials from Cataractonium Roman Town are therefore related.

What we need is a way to account for the ‘relatedness’ of sets of radiocarbon dates. Bayesian statistics enable us to do this. Given that the burials were all found in association with the Roman settlement, we can postulate that when burial started at Cataractonium, it continued at a relatively constantly rate for some period of time, and it then ended.

By using the archaeological information that the dates relate to burial activity that continued for a certain period (and that burial started before it ended!), our model can assess how much of the scatter on the radiocarbon dates comes from statistics and how much is real, historical duration. Furthermore, the model formally estimates when burial began and when it ended (see section 2.1 and section 2.2).

The model, implemented in OxCal 4.4 (Bronk Ramsey 2009a; 2017) using IntCal20 (Reimer et al. 2020),is based on the assumption of a uniform rate of burial within and around Cataractonium during its occupation (Zeidler et al. 1998).

Given that the stable isotope results (Table 9) indicate that the dated individuals consumed a diet predominantly based on terrestrial C₃ foods (Fig. 22), the radiocarbon results are unlikely to be affected by any significant reservoir effects, so a fully terrestrial calibration curve can be employed.

The model provides estimates for the beginning of burial in cal AD 160–230 (95% probability; StartBurial; Fig. 42), probably in cal AD 185–220 (68% probability); and its demise in cal AD 275–395 (95% probability; EndBurial; Fig. 42), probably in cal AD 315–375 (68% probability).

Figure 43 shows the estimated length of the phase of burial activity around Cataractonium to be between 50–210 years (95% probability; DurationBurial), probably between 95–180 years (68% probability).

The dates of the first and last dated burials at Brough Park, Fort Bridge, Brompton West, together with the single dated inhumation from Brompton East are summarised in Figure 44.

By comparing the posterior density estimates, it is possible to calculate the probable order of pairs of different events (Table 10). For example, it is 68.5% probable that the first dated burial at Fort Bridge (FirstFortBridge; Fig. 44) began before the first dated burial at Brough Park (FirstBroughPark; Fig. 44).

A greater understanding of the chronology of burial practices at Roman Cataractonium has provided further insights into how these individual lived their lives and identified themselves within society (Speed and Holst 2018, 599). When integrated with evidence for funerary rites, grave form and accompanying grave goods the chronology has been able to provide glimpses into the spatial variations of society with Cataractonium.

5.4 An enclosed landscape: Iron Age settlement on the Northumberland plain

Archaeological excavations conducted in advance of housing development at White Hall Farm, Cramlington, Northumberland (Fig. 45) were commissioned by Persimmon Homes and Belway, and undertaken by Archaeological Services Durham University, following an evaluation that identified features together with material culture indicating the presence of an Iron Age settlement and a small ditch of unknown date (ASDU 2019).

Two areas were opened for excavation (Areas 1 and 2; Fig. 45), given the potential the site had to contribute to a number of themes in the North East regional research framework (NERRRHE 2022)— in particular: La2 ‘How can we improve our understanding of the chronology of late Bronze Age and Iron Age north-east England?’; and La1 ‘How can we improve our understanding of late prehistoric settlement and settlement patterns?’.

Area 1 contained part of a rectilinear settlement enclosed by a palisade within which evidence survived for a central roundhouse (RH1) and two smaller roundhouses (RH2 and RH3; Fig. 46). These structures contained elements of internal wall construction slots together with eaves-drip gullies, and all three appeared to not have been rebuilt.

Outside the entrance to the palisaded enclosure, two pits, one probably an open hearth and the other a covered earth oven, represent evidence for earlier activity.

Area 2 included part of a ditch.

A total of 14 radiocarbon measurements were made on 14 samples, all but one from Area 1 (Table 11).

The samples were processed and measured by Accelerator Mass Spectrometry at SUERC, according to the procedures described in Dunbar et al. (2016). All results are conventional radiocarbon ages (Stuiver and Polach 1977).

The laboratory maintains a continual programme of quality assurance procedures, in addition to participation in international inter-comparisons (Scott et al. 2017). These tests indicate no laboratory offsets and demonstrate the validity of the precision quoted.

Eleven samples were single fragments of charcoal, and the remaining three were single fragments of hazelnut shell. All the charcoal fragments were identified as from relatively short-lived species, bar two that were from oak of unknown maturity (SUERC-87916 and SUERC-87925).

From Area 1 the samples from F12 (SUERC-87917) and F14 (SUERC-87918) derive from primary fuel debris deposits associated with the use of the pits dug before the construction of the palisaded enclosure settlement (section 3.2.2).

The remaining 11 samples from Area 1 derived from postholes, gullies and trenches related to the construction and use of the roundhouses, and from the primary fill of the palisade trench. These 11 samples therefore most likely derived from activity associated with the construction and use of the settlement, although SUERC-87925, from the fill of the construction trench F34 for Round House 1, is clearly residual; and the fragment of charcoal (SUERC-87919) from the fill of F8 appears to be associated with the use of the external pits.

The samples from the two external pits have been interpreted as freshly deposited in their contexts, and those from the palisaded enclosure settlement as deriving from its use, apart from SUERC-87919 and SUERC-87925, which have been modelled as termini post quem using the AFTER function in OxCal (shown in Fig. 47).

The model for this case study has been calculated in OxCal v4.4 (Bronk Ramsey 2009a) using IntCal20 (Reimer et al. 2020), includes the archaeological interpretation that the dated material from the Area 1 palisaded enclosure settlement derives from a single continuous phase of activity (Buck et al. 1992) and that its relationship to the other dated material from the external pits outside its entrance and from Area 2 is unknown.

The model has good overall agreement (Amodel: 93; Fig. 47) and suggests that the enclosed rectilinear settlement was established in 430–220 cal BC (95% probability; startRectilinearSettlement; Fig. 47) probably 405–300 cal BC (68% probability) and ended in 345–120 cal BC (95% probability; endRectilinearSettlement; Fig. 47) probably 280–160 cal BC (68% probability).

By comparing the estimated dates for start and end of activity associated with the enclosed rectilinear settlement, we can suggest that it was in use for 1–200 years (95% probability; RectilinearSettlement Fig. 48), probably for 30–135 years (68% probability).

The food preparation activity represented by the two pits outside the enclosure entrance predates the beginning of activity associated with the enclosed rectilinear settlement by 1–370 years (95% probability; Gap; Fig. 49) probably 40–215 years (68% probability). This interval has been calculated as the difference between the estimated last dated material associated with the external pits (LastExternalPits) and the estimated start of activity associated with the enclosed rectilinear settlement (startRectilinearSettlement).

The secondary fill of the ditch in Area 2 was deposited sometime in the late Iron Age and it is therefore plausible that it relates to wider use of the landscape by the inhabitants of the enclosed rectilinear settlement.

This dating of the enclosed rectilinear settlement at White Hall Farm is further evidence for the start of more permanent settlement in the area and a landscape that became increasingly ‘enclosed’ on the Northumberland coastal plain in the third quarter of the 1st millennium cal BC.

Previous dating of nearby sites of East and West Brunton (Hamilton 2010; Fig. 50) demonstrates that it is 87% probable that the beginning of activity associated with the enclosed rectilinear settlement at White Hall Farm post-dates the start of unenclosed settlement at East and West Brunton, but predates their ‘enclosure’ (Fig. 51).

5.5 Geoarchaeological investigations at 2 Pier Road, North Woolwich, London

Increasing redevelopment of the former industrial parts of east London, and the infringement of urban sprawl into parts of Essex and Kent driven by the growing requirements of the city, has resulted in a major expansion in developer-funded archaeological work over the last decade (Fig. 52).

Geoarchaeological investigations in advance of residential and commercial development at 2 Pier Road, North Woolwich, London Borough of Newham (Fig. 53), undertaken by Museum of London Archaeology (MoLA) and funded by Higgins Construction Plc, led to the recovery of alluvial sequences that contained an archive of Holocene environmental change, and in particular of changes in relative sea level and regional vegetation cover (Stastney et al. 2021).

In order to provide a chronological framework for the multi-proxy environmental work undertaken on core BH03 from 2 Pier Road, seven AMS radiocarbon dates were obtained (Table 12).

The three samples dated at SUERC were processed and measured by Accelerator Mass Spectrometry, according to the procedures described in Dunbar et al. (2016) and those at Beta Analytic were dated by AMS following the methods outlined at https://www.radiocarbon.com/. The reported results are conventional radiocarbon ages (Stuiver and Polach 1977).

Both laboratories maintain continual programmes of quality assurance procedures, in addition to participation in international inter-comparisons (Scott et al. 2017). These tests indicate no laboratory offsets and demonstrate the validity of the precision quoted.

These dates have been included in the age-depth model shown in Figure 54, constructed using rBacon and the IntCal20 terrestrial dataset for the northern hemisphere (Reimer et al. 2020).

Given the evidence that different litho-facies formed under different depositional environments (Stastney et al. 2021, table 1), we have included boundaries at 7.55m b.g.l. (channel margin/semi-terrestrial alder carr) and 5.88m b.g.l. (semi-terrestrial alder carr/intertidal floodplain) to take account for potential changes in accumulation rates.

Age-depth modelling implemented with rBacon is similar to that outlined in Blaauw and Christen (2005), but more numerous and shorter sections are used to generate a more flexible chronology (Blaauw and Christen 2011).

Radiocarbon age distributions are derived from the Student-t distribution, which produces calibrated distributions with longer tails than the Normal model (Christen and Pérez 2009). The longer tails on radiocarbon dates, and a prior assumption of unidirectional sediment accumulation, mean that in most cases excluding outliers is not necessary when using rBacon.

The memory or coherence in accumulation rates through a sequence is a parameter based on the degree to which the accumulation rate at each interval depends on the previous interval. Thus, the memory for modelling accumulation in organic-rich (peat) sediments is higher than for lacustrine sediments because accumulation of peat in peat bogs is less dynamic over time than the accumulation of sediments in a lake. We used the default memory properties given in Blaauw and Christen 2011; mem.strength = 4 and mem.mean = 0.7).

The treatment of outliers in rBacon is analogous to the OxCal General Outlier Model (Bronk Ramsey 2009b) in that both draw from a long-tailed Student-t distribution. The number of parameters employed in the process by rBacon is significantly different from OxCal, generally resulting in more flexibility towards potential outliers than the approach implemented in OxCal, and also enabling the model to account for possible unknown or underestimated errors associated with the ¹⁴C determinations (Christen and Pérez 2009).

The resulting age-depth model is shown in Figure 54 along with plots that describe: (top left panel) the stability of the model (log objective vs iteration); (top middle panel) the prior (entered by the user) and posterior (resulting) accumulation rate, and; (top right panel) the prior and posterior memory properties.

The model has excluded two dates, SUERC-88163 and Beta-515736. The Alnus catkin, SUERC-88163, appears to be residual and probably represents reworked material from the river channel margins. The carbonised spelt grain, Beta-515736, would appear to be intrusive (cf Pelling et al. 2015), as it was an isolated find among a plant macrofossil assemblage dominated by species associated with wetlands and marshy ground (Stastney et al. 2021, 5).

The wider tails of the rBacon calibration model reduce the need for detecting and removing outliers (Blaauw and Christen 2011, 476). The model is very stable (Fig. 54, top left panel) with the posteriors for the accumulation rate and its variability showing excellent comparability to their priors (Fig. 54, top middle/right panels).

Producing an age-depth model is often the first step towards determining the age of ‘events’ in proxy records at specific depths in a sequence that are not directly dated. For example, the Ulmus decline recorded in the BH03 pollen diagram (Stastney et al. 2021, fig. 5) at 8.64m bgl is estimated to have taken place in 4210–3980 cal BC (95% probability; Fig. 55a; Table 13) and the Tilia decline (6.40m bgl) in 1835–1450 cal BC (95% probability; Fig. 55d).

Age-depth models have the potential to significantly increase our ability to accurately date past events and thus to better understand past environmental changes and human impact on the landscape, but good prior information is essential for reliable age-depth models, particularly in cases where sequences have low numbers of radiocarbon dates.

5.6 ¹⁴C wiggle-matching at 4 Walesker Lane, Harthill with Woodall, South Yorkshire

The four-bay house at 4 Walseker Lane, Harthill with Woodall, near Rotherham (Fig. 56), is believed to be one of the earliest domestic buildings so far identified in South Yorkshire (Ryder 1987).

The medieval house apparently consisted of a central two-bay hall flanked by end bays. The shorter eastern bay of the hall perhaps housed the dais, with the bay beyond containing the solar, its status suggested by the collar purlin and braces over the bay being neatly chamfered. At the west end of the hall a substantial stone wall and details of the carpentry in the roof above, may suggest the position of the original hearth. The impressive crown-post roof survives virtually intact (Fig. 57).

Tree-ring analysis was commissioned to inform renovations of the building in 2019 (Arnold et al. 2020a). All timbers were from very fast-grown trees and were of clearly marginal suitability for dating by ring-width dendrochronology. A hybrid approach was therefore adopted using dendrochronology and radiocarbon wiggle-matching in partnership.

Core samples were obtained on two separate occasions (Table 14). Eight timbers were sampled in April 2019 and, following initial tree-ring analysis, which failed to produce grouping between any of the ring-width series, four single-ring samples from different timbers were submitted for radiocarbon dating to confirm the extent of surviving early fabric in the hall roof.

A further eight core samples were obtained in July 2019, which enabled the grouping and tentative dating of the ring-width series. Seven more samples were then submitted for radiocarbon dating to confirm the tentative dating suggested by the ring-width dendrochronology, and to test further tentative statistical and visual cross-matching between the ring-width series.

The annual growth ring-widths of all but one sample were measured. Allowing for the short lengths of the sample series, these measured data were then compared with each other by the Litton/Zainodin grouping procedure (Litton and Zainodin 1991; Laxton et al. 1988). This resulted in the production of a single cross-matching group of ten samples, which formed at a minimum t-value of 3.7.

The ring-width series were combined to form site chronology WLSKSQ01A (Fig. 58), which was compared to the reference chronologies for oak. This indicated that WLSKSQ01A cross-matched at two different possible positions with similar t-value levels (Table 15a–b), and so cannot be dated by ring-width dendrochronology.

Site chronology WLSKSQ01A was then compared with the five remaining measured but ungrouped samples. This indicated tentative statistical cross-matching with a further three samples, this 13-sample group forming at a minimum t-value of 3.1. These 13 ring-width series were also combined at the offset positions to form site chronology WLSKSQ01B (Fig. 58), which was similarly compared to the reference chronologies with inconclusive results (Table 15a–b).

The two measured samples that remain ungrouped both have less than 30 rings, which is insufficient for even tentative statistical cross-matching. However, an attempt was made to cross-match the ring-width series from these two samples visually with the other measured series from this building (Fig. 59). These additional two ring-width series were then combined with the 13 ring-width series included in WLSKSQ01B to form site chronology WLSKSQ01C (Fig. 58), which was again compared to the reference chronologies with inconclusive results (Table 15a–b).

The radiocarbon wiggle-matching was thus needed to confirm the inconclusive dating suggested by ring-width dendrochronology for site master sequences WLSKSQ01A–C, and to validate the tentative cross-matching of additional samples suggested both by weak statistical correlation (WLSKSQ01B) and by visual matching (WLSKSQ01C) (Fig. 59).

A Bayesian approach has been adopted for the radiocarbon wiggle-matching (Christen and Litton 1995), which incorporates the gaps between each dated annual ring in site master sequence WLSKSQ01C (Fig. 60), along with the radiocarbon measurements from all five cores that have been sampled for radiocarbon dating.^[Footnote] Two of these are securely linked to this sequence by statistics (WLS-K01 and WLS-K02A), one is tentatively linked to it by statistics (WLS-K04), and two are tentatively linked by visual matching (WLS-K06 and WLS-K08).

The model has been calculated using OxCal v4.4 (Bronk Ramsey 2009a) and IntCal20 (Reimer et al. 2020), and has good overall agreement (Acomb: 145.5, An: 21.3, n: 11; Fig. 61).

The two radiocarbon dates from core WLS-K04, which is only tentatively linked to the site master on statistical grounds, have good individual agreement in the model (ETH-104565, A: 76 and ETH-104566, A:132; Fig. 61), as do the four radiocarbon dates from cores WLS-K06 and WLS-K08, both of which are only tentatively linked by visual matching to the site master chronology (ETH-104567, A: 154, ETH-99778, A: 133, ETH-104568, A: 93, and ETH-99779, A: 133; Fig. 61).

These statistics suggest that the offset positions tentatively suggested by the statistical and visual cross-matching of the ring-width data are valid.

The model suggests that the final ring of WLSKSQ01C formed in cal AD 1428–1436 (95% probability; WLSKSQ01C felling; Fig. 61), probably in cal AD 1430–1434 (68% probability). Furthermore, when the last ring of the wiggle-match is constrained to be AD 1432, the model again has good overall agreement (Acomb: 153.3, An: 20.4, n: 12), and all the radiocarbon dates have good individual agreement (A > 60).

The results from the radiocarbon wiggle-matching allow one of the two tentative matches provided by the ring-width dendrochronology to be considered as a radiocarbon-supported dendrochronological date, that spanning AD 1376–1432 (Table 15a), with the trees represented felled in the winter of AD 1432/33_DR. The subscript _DR indicates that this is not a date determined independently by ring-width dendrochronology, and that the master sequence, WLSKSQ01A–C, should not be utilised as a ring-width master sequence for dating other sites.

The alternative tentative cross-dating for this sequence suggested by the ring-width dendrochronology, as spanning AD 1407–1463 (Table 15b) is clearly spurious, as it is incompatible with the radiocarbon wiggle-matching.

Footnote

Radiocarbon dating was undertaken by the Laboratory of Ion Beam Physics, ETH Zürich, Switzerland in 2019–20. Cellulose was extracted from each ring using the base-acid-base-acid-bleaching (BABAB) method described by Němec et al. (2010), combusted and graphitised as outlined in Wacker et al. (2010a), and dated by Accelerator Mass Spectrometry (Synal et al. 2007; Wacker et al. 2010b). Data reduction was undertaken as described by Wacker et al. (2010c).
The facility maintains a continual programme of quality assurance procedures (Sookdeo et al. 2020), in addition to participation in international inter-comparison exercises (Scott et al. 2017; Wacker et al. 2020).
These tests demonstrate the reproducibility and accuracy of these measurements. The results are conventional radiocarbon ages, corrected for fractionation using δ¹³C values measured by Accelerator Mass Spectrometry (Stuiver and Polach 1977).↩

5.7 Stratigraphy at Chalk Hill, Ramsgate

Three ditches of a causewayed enclosure, with a maximum diameter of 150m, were revealed during excavations by the Canterbury Archaeological Trust in advance of road building for Kent County Council in 1997–8 (Fig. 62; Clark et al. 2019).

Thirteen segments of the inner arc, seven of the middle arc and three of the outer arc were investigated (Fig. 63).

Dating of the causewayed enclosure was undertaken in partnership with the Gathering Time project, funded by English Heritage and the Arts and Humanities Research Board, and based in Cardiff University (Whittle et al. 2011).

A total of 23 radiocarbon measurements were made on 21 samples, all but one from the outer ditch (Table 17).^[Footnote] Thirteen of the samples were either of articulating groups of animal bone or carbonised residues on groups of sherds from a single vessel, and so it is unlikely that this dated material was residual (see section 3.2.2).

Two other samples are less certainly considered to be freshly deposited because of their fragility: fragments of a cattle skull (UBA-14307) and a single sherd of Neolithic Bowl preserving a carbonised residue (OxA-15391). These interpretations are critical, because they mean that the stratigraphic sequence of fills through the outer ditch should be the same as the sequence of dated samples, and so the stratigraphy can be used as prior information for the model (Fig. 64).

The status of the results on the four samples of disarticulated animal bone, and two samples of disarticulated human bone is unclear. All could be residual, although they all are short-life, single-entity samples. Although the bone dated by UBA-14305 was not identified to species, its stable isotope values are compatible with those of a terrestrial herbivore (Table 17).

All the models in this case study have been calculated in OxCal v4.4 (Bronk Ramsey 2009a) using IntCal20 (Reimer et al. 2020). They interpret all the samples as freshly deposited in their contexts, bar the six disarticulated bones, which are modelled as termini post quos using the AFTER function in OxCal.

Model 1 has poor overall agreement (Amodel: 56), with the dates on two samples having poor individual agreement (sherd group 265, A: 4 and UBA-14310, A: 26), both of which are slightly later than expected from their positions in the model. It is difficult to identify the cause of the problems with these dates. Sherd group 265 includes more than 15 sherds from the same Plain Bowl, and the carbonised residue was on the inside of the pot and well-preserved.

There are three statistically consistent measurements on this residue, made by two different laboratories who dated different chemical fractions of the material (the solid residue after an acid wash and multiple water rinses in Oxford, the alkali-soluble fraction from an acid-base-acid pretreatment in Groningen).

UBA-14310 is from a fragment of human skull from a sub-adult/adult, possibly female, individual in what was interpreted as a placed deposit. Clark et al. (2019, 83) interpret this sample as providing the most reliable date for this context, with UBA-14311 interpreted as residual. But another date, on an articulating animal bone group from the same feature (UBA-14309) is much closer to UBA-14311 and is unlikely to be residual, although the presence of enough of the skull for age and sex to be determined suggests that it was too big to be plausibly intrusive. So perhaps UBA-14310 is slightly too young?

Since we do not know why these particular dates are problematic, we have constructed two alternative models using different forms of outlier analysis.

Model 2 employs the s-type outlier model, which addresses potential underestimation of measurement uncertainty in the laboratory (Bronk Ramsey 2009b, 1037−8), and Model 3 employs the general outlier model, which is appropriate when the source of the outliers is unclear (Bronk Ramsey 2009b, 1028).

In both cases, the prior probability that any result is an outlier has been set to 5%. In Model 2, three dates have posterior outlier probabilities greater than 10% (UBA-14305, O: 10; UBA-14310, O: 10; and sherd group 265, O: 44); and in Model 3, two dates have posterior outlier probabilities greater than this (UBA-14310, O: 11 and sherd group 265, O: 13). These dates have been down-weighted proportionately in these models.

Model 2 is illustrated in Figure 65, and key parameters from all three models are shown in Figures 66 and 67. It is clear from this sensitivity analysis that the estimated chronology for the Chalk Hill enclosure is robust against the choice of modelling approach. The medians of the key parameters shown in Figures 66 and 67 vary by between 1 and 6 years for Models 1 and 2, and by between 3 and 16 years for all three models.

Our choice of a preferred model is therefore not critical, but, given the character of the sampled material discussed above, we consider that slight under-estimation of some of the laboratory errors is the most plausible interpretation of these data and report the results of Model 2.

This suggests that the causewayed enclosure at Chalk Hill was established in 3780–3675 cal BC (95% probability; start Chalk Hill; Fig. 65), probably in 3730–3690 cal BC (68% probability).

The outer arc was built in 3760–3675 cal BC (95% probability; build outer Chalk Hill; Fig. 65), probably in 3715–3685 cal BC (68% probability).

Any estimate for the inner arc is extremely tentative because it is based on a single measurement, although the model suggests that this arc was built in 3740–3645 cal BC (95% probability; build inner Chalk Hill; Fig. 65), probably in 3710–3700 cal BC (5% probability) or 3690–3650 cal BC (63% probability).

In segments 3 and 5 of the outer arc, the stratigraphically latest dated samples were from close to the end of sequences of numerous recuts. The estimate for the end of use of the enclosure based on these samples is therefore close to the age of the final deposits in these segments.

The model suggests that the Chalk Hill enclosure was abandoned in 3625–3510 cal BC (95% probability; end Chalk Hill; Fig. 65), probably in 3620–3565 cal BC (68% probability). By comparing the modelled date estimates for the initial construction of the enclosure and its abandonment, we can suggest that the enclosure was used for 55–200 years (95% probability; use Chalk Hill; Fig. 67 (black)), probably for 70–135 years (68% probability).

Footnote

The samples dated in Groningen were processed and measured by Accelerator Mass Spectrometry, according to the procedures set out in Aerts-Bijma et al. (1997; 2001) and van der Plicht et al. (2000). Samples processed in Oxford were dated according to the procedures described by Hedges et al. (1989) and by Bronk Ramsey et al. (2004a-b). Collagen from the bone samples dated at Belfast was extracted as described by Longin (1971), graphitised as described by Slota et al. (1987), and dated by AMS.
The results reported there are conventional radiocarbon ages (Stuiver and Polach 1977).
All three laboratories maintain continual programmes of quality assurance procedures, in addition to participation in international inter-comparisons (Scott 2003). These tests indicate no laboratory offsets and demonstrate the validity of the precision quoted.
The group of replicate measurements on the carbonised residue on sherd group 265 are not statistically significantly different at the 5% significance level (Ward and Wilson 1978; Table 17).↩

Back to contents | 4. Project Organisation and Planning | Where to get advice