Land North of Culmstock Road, Hemyock, Devon

Post-Excavation Assessment and Updated Project Design

for Waddeton Park Ltd

CA Project: EX0032 CA Report: EX0032_1

July, 2019

Land North of Culmstock Road, Hemyock Devon

Post-Excavation Assessment and Updated Project Design

CA Project: EX0032 CA Report: EX0032_1

Document Control Grid Revision Date Author Checked by Status Reasons for Approved revision by A 22 May Jonathan Internal 2019 Orellana review B 24.07.19 Richard Internal QA KEW Massey review

This report is confidential to the client. Cotswold Archaeology accepts no responsibility or liability to any third party to whom this report, or any part of it, is made known. Any such party relies upon this report entirely at their own risk. No part of this report may be reproduced by any means without permission.

© Cotswold Archaeology

© Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

CONTENTS

SUMMARY 1

1. INTRODUCTION 2

2. ARCHAEOLOGICAL BACKGROUND 3

3. AIMS AND OBJECTIVES 6

4. METHODOLOGY 7

5. STRATIGRAPHIC RECORD: FACTUAL DATA AND STATEMENT OF

POTENTIAL 9

6. FINDS: FACTUAL DATA AND STATEMENTS OF POTENTIAL 17

7. BIOLOGICAL EVIDENCE: FACTUAL DATA AND STATEMENTS OF POTENTIAL

20

8. DISCUSSION 23

9. UPDATED AIMS AND OBJECTIVES ...... 25 10. PUBLICATION ...... 296

11. PROJECT TEAM ...... 297

12. TIMETABLE ...... 29

13. ACKNOWLEDGEMENTS ...... 29

14. STORAGE AND CURATION ...... 29

15. REFERENCES ...... 30

APPENDIX A:STRATIGRAPHIC SUMMARY ...... 34 APPENDIX B: MIXED FINDS ...... 35 APPENDIX C: IRONWORKING RESIDUES ...... 36 APPENDIX D: ENVIRONMENTAL MATERIAL ...... 56 APPENDIX E: GEOARCHAEOLOGICAL ASSESSMENT ...... 60 APPENDIX F:OASIS REPORT FORM ...... 70

1 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

2 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

LIST OF ILLUSTRATIONS

Fig. i South-facing section of the natural feature recorded in Area 14 Fig. ii North-east facing section of the natural feature recorded in Area 15 Fig. 1 Site location plan (1:25,000) Fig. 2 The site, showing excavation areas, archaeological features, previous evaluation trenches and geophysical survey results (1:1250) Fig. 3 Excavation areas, showing archaeological features (1:250) Fig. 4 Photographs: general views of Areas 13 and 15 Fig. 5 Pit 13018: section and photograph (1:20) Fig. 6 Furnace 14006: section and photograph (1:20) Fig. 7 Furnaces 15004 and 15009: sections and photographs (1:20)

LIST OF TABLES

Table 1: Radiocarbon dating results from the evaluation Table 2: Summary of undated pits and postholes in Area 13 Table 3: Quantification of finds Table 4: Task List Table 5: Distribution of worked and burnt flint/chert Table 6: Summary catalogue of material submitted as archaeometallurgical residues Table 7: Summary of material in collection by area and context Table 8: Radiocarbon dates, with those of comparative sites showing similar technology Table 9: Assessment of Environmental Remains Table 10: Monolith Sample 6a Table 11: Monolith Sample 6b Table 12: Monolith Sample 26a Table 13: Monolith Sample 26b

3 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

SUMMARY

Project Name: Land North of Culmstock Road Location: Hemyock, Devon NGR: 313380 113490 Type: Excavation Date: 1−16 May, 2018 Planning Reference: 16/01772/MOUT Location of Archive: To be deposited with the Royal Albert Memorial Museum (RAMM), Exeter, and the Archaeology Data Service (ADS) RAMM Reference Number: RAMM: 42/2017 Site Code: CRDH 18

A programme of archaeological excavation was undertaken by Cotswold Archaeology in May 2018, at the request of Waddeton Park Ltd, on land off Culmstock Road, Hemyock, Devon. Three excavation areas, collectively covering an area of 1,890m², were excavated across the development site. Archaeological features were identified within all three excavation areas, and comprised the remains of four confirmed iron smelting pit furnaces, two deposits of metalworking debris, a large rectangular pit, two ditches and a number of pits and postholes. A non-slag tapping iron smelting furnace, which contained an in situ iron slag puddle representing its last smelt, was identified in Area 14. Deposits of metallurgical residues associated with this furnace had been recorded during a previous evaluation, and were radiocarbon-dated to the early medieval period. In Area 15, three further non-slag-tapping pit furnaces were revealed, together with a slag deposit and three further pits. One of the furnaces retained the in situ remains of a clay lining and adhering slag. Two furnaces contained fragments of fired clay in their upper fills, representing the collapse of furnace superstructures. Two closely-associated pits may represent hollows for the operation of bellows. A large rectangular pit recorded in Area 13 was interpreted as a possible ore-roasting pit. Monolith sampling revealed the presence of an undated palaeochannel crossing Areas 14 and 15, probably representing a former tributary of the nearby River Culm. This document presents a quantification and assessment of the evidence recovered from the excavation. It considers the evidence collectively, in its local, regional and national context, and presents an updated project design for a programme of post-excavation analysis to bring the results to appropriate publication.

1 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

1. INTRODUCTION

1.1 This document provides a detailed report on a programme of archaeological investigation undertaken by Cotswold Archaeology (CA) in May, 2018, on land north of Culmstock Road, Hemyock, Devon (centred at NGR: 313380 113490; Fig. 1). This excavation was commissioned by Waddeton Park Ltd, prior to an application for residential development on the site.

1.2 An outline planning application (ref: 16/01772/MOUT) was submitted to Mid Devon Council, the local planning authority, for the development of up to 40 dwellings on the site, together with public open space and associated infrastructure. Stephen Reed (Senior Historic Environment Officer, Devon County Council Historic Environment Team; DCCHET), the archaeological advisor to Mid Devon District Council, recommended the following condition, to be attached to any planning permission granted for the scheme:

No development of any phase shall take place until the applicant has secured the implementation of a programme of archaeological work in accordance with a written scheme of investigation which has been submitted by the applicant and approved by the Planning Authority.

1.3 The excavation followed a geophysical survey (PCG 2016), and trial trench evaluation (CA 2017; Fig. 2). The evaluation identified deposits of metallurgical residues in the western part of the site, which were interpreted as dumps of waste associated with a non-slag tapping iron-smelting furnace. Two of these deposits were radiocarbon-dated to the early medieval period. Two undated ditches, and a possible gravel surface, were also recorded during the evaluation.

1.4 Three excavation areas (Areas 13, 14 and 15), totalling 1,890m² in extent, were excavated in the western part of the site (Figs. 2 and 3). These areas targeted archaeological features identified by evaluation and geophysical survey. The excavation was carried out in accordance with a detailed Written Scheme of Investigation (WSI) produced by CA (2018) and approved by Stephen Reed. The scope and methodology of excavation were also in accordance with the Specification for Archaeological Excavation (Devon County Council 2016), Standard and Guidance for Archaeological Excavation (CIfA 2014), Management of Research

2 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Projects in the Historic Environment (MoRPHE) PPN 3: Archaeological Excavation (Historic England 2015a).

1.5 The excavation fieldwork was monitored by Stephen Reed, including site visits made on May 2, 9 and 14, 2018.

The site 1.6 The overall area within the ‘red-line’ defining the wider site is 6.6ha. However, the proposed development is limited to a footprint covering an area of only 2.1ha, and much of the known archaeological interest of the site is located outside this. At the time of excavation comprised four agricultural fields located on the north-western margins of Hemyock village (Figs. 1, 2 and 4). Culmstock Road (the B3391) runs along part of the southern site boundary. Houses front the road beyond the south- eastern and south-western corners of the site, and a cemetery borders the road beyond the south-central boundary. A small stream defines the eastern site boundary, with residential estates located beyond. Further farmland lies to the north and west of the site (Fig. 4).

1.7 The site occupies a north-facing slope, which falls from an elevation of c.140m above Ordnance Datum (aOD), at its southern boundary, to less than 130m aOD at its northern.

1.8 The underlying bedrock geology of the area is mapped as Mercia Mudstone Group, of the Triassic Period. This is overlain by recent colluvial deposits in the northern and western parts of the site, although no superficial deposits are recorded elsewhere (BGS 2018).

2. ARCHAEOLOGICAL BACKGROUND

2.1 The site has been the subject of a desk-based heritage assessment (CA 2015a), a geophysical survey (PCG 2016) and an archaeological evaluation (CA 2017). The following section is drawn from these sources, which have been augmented by data from the Devon and Dartmoor Historic Environment Record (DDHER) and the published archaeological record.

2.2 Recorded evidence for prehistoric or Roman activity in the vicinity of the proposed development site is very limited, although it is possible that a number of field boundaries and earthwork remains of former and divisions recorded by the Historic

3 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Environment Record are of prehistoric origin. A number of these have been mapped from Lidar and aerial photographic data, including those at Brownings Farm, Hemyock (DDHER MDV 68839), approximately 1km south-west of the site.

2.3 A Bronze Age hoard from the vicinity of Hemyock, apparently comprising palstave axes, is the subject of an antiquarian record, although the precise location of this find was not recorded.

2.4 An iron bloomery site was recorded to the south of Tedburrow Farm, c. 570m east of the site (DDHER MDV114775; NGR ST 136131). This was assigned a Roman date, although no justification for this is apparent; large quantities of undated tap-slag were recorded on this site. Finds of slag and remains of ironworking activity are elsewhere widespread around Hemyock. At Churchill’s Farm (DDHER MDV 114773; NGR ST136131; Young 2013) c. 350m south-east of the site, excavation by Exeter Archaeology revealed remains of furnace bases and tap-channels, associated with large quantities of slag. To the north, an area of pits and ditches was associated with pottery of twelfth and thirteenth-century date (Smart (ed), 2018).

2.5 Smelting waste, slag, furnace lining and charcoal were recorded following ploughing, at a location east of Castle Dene, Hemyock (DDHER MDV 76408; NGR ST 135133; Tabor 2010), and an undated deposit of slag was recorded 350m north of Bodham’s Farm, Hemyock (DHER MDV 114617; NGR ST 1421290). Numerous extraction pits have been recorded from Lidar and aerial photographic surveys around Hemyock, especially along Culmstock Road, and while undated, it is possible that many of these features are associated with ironworking activity of medieval and post-medieval date (Environment Agency 1998-2002).

Early medieval period AD 410−1066

2.6 More confidently-dated evidence of ironworking was recorded within the centre of Hemyock village. Iron-working slag and furnace fragments were recorded during trial trench evaluation at Halls Engineering Works, with an early medieval date suggested (Exeter Archaeology 2015). Medieval ironworking debris, dated through associated finds, was also recorded during trial trench evaluation, at No. 3 Broadway (Allum 2009).

2.7 Archaeological excavation undertaken immediately south of Culmstock Road, in 2012, and located only c.100m to the south-west of the current site, revealed remains of slag-pit furnaces, pits and gullies, which were sealed beneath a deposit of slag and

4 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

fuel waste (Rainbird and Young 2018). This excavation followed a geophysical survey which identified extensive magnetic anomalies suggesting industrial activity, and a trial trench evaluation which revealed gullies sealed by deep slag deposits (Stead and Payne 2012). Two radiocarbon dates indicated a probable late eighth- century date for this activity. The proximity of this activity to the current site strongly suggests that these comprise contiguous parts of a more extensive area of early medieval ironworking activity. This concentration of recorded sites and evidence demonstrates that Hemyock was a significant centre of iron production in the early medieval (Saxon) and later medieval periods (Blaylock 1989; Hughes and Jones 2013; Tabor 2010).

Medieval period (1066 – 1549) 2.8 Settlement at Hemyock is thought to have originated in the early medieval period, and the village is recorded in the Domesday Survey. The Parish Church of St Mary includes 11th/12th-century elements, and buildings are thought to have been present on the site of Hemyock Castle in the 13th-century. Hemyock Castle is a roughly rectangular-plan structure, with a defined moat, of largely fourteenth-century date. Upstanding remains include corner towers, and circular towers flanking a main entrance, together with the remains of a curtain wall (Blaylock 1989).

2.9 Archaeological investigation c.100m south-east of the development site identified a linear bank potentially associated with the defences of Hemyock Castle. These works also recorded building remains, with associated features and deposits of 12th to 14th-century date, as well as an underlying ‘yard surface’ of earlier origin (Tabor 2010).

2.10 There is no evidence to suggest that medieval settlement extended as far as the present site, which most probably comprised part of the agricultural hinterland of Hemyock at this time. A narrow north/south linear bank is visible at the eastern edge of the north-eastern field on mid-20th century aerial photographs. This is broadly aligned with the eastern field boundary, and may represent a surviving medieval land division. Wider evidence suggests a settlement pattern of dispersed farmsteads in the medieval and post-medieval periods, with earthworks associated with minor settlements visible on a number of aerial photographs and Lidar images, including at Regency House, Hemyock (DDHER MDV 115865; NGR ST132127).

Post-medieval (1540 – 1800) and modern periods (1810 – present)

5 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

2.11 The site appears to have continued in agricultural use throughout the post-medieval period. The 1843 Hemyock Parish Tithe map depicts the development site located within four agricultural fields, to the north-west of Hemyock village. The accompanying Tithe Apportionment lists these fields under the sub-heading ‘Castle’, suggesting an historical association with Hemyock Castle.

Geophysical Survey and Evaluation

2.11 A gradiometer survey, in March 2016, identified a number of zones of stronger magnetic variation on the site, which were suggested to be related to medieval ironworking (Pre-Construct Geophysics 2016; Fig. 2). Results also indicated a number of potential pits, a group of which were detected towards the north-eastern margins of the site. A magnetically-depleted linear anomaly in the south-east was tentatively interpreted as a former trackway.

2.12 The evaluation (CA 2017; Fig. 2) recorded a feature interpreted at that stage as an ironworking platform. Deposits associated with this feature were radiocarbon-dated to the early medieval period (Table 1, below). Retrieved metallurgical residues were consistent with the operation of non-slag tapping iron-smelting furnaces. The evaluation also recorded two ditches and a gravel surface, all of which were undated, although one ditch contained a single piece of slag and a dump of burnt waste. The radiocarbon dates obtained are detailed in Table 1, below, and broadly indicate an eighth-century pattern of activity:

Table 1: Radiocarbon dating results from the evaluation

Feature Lab No. Material δ 13C Radiocarbon Calibrated radiocarbon Calibrated radiocarbon age age 95.4% probability age 68.2% probability

Context 906 SUERC Charcoal -26.8‰ 1170 ± 29yr BP 771–903 cal AD (91.8%) 777–793 cal AD (11.4%) Ironworking -73284 Corylus avellana 919–963 cal AD (13.6%) 801–893 cal AD (56.8%) platform 904 (hazel) Context 907 SUERC Charcoal -26.2‰ 1277 ± 29 yr BP 663–775 cal AD (95.4%) 684–721 cal AD (39.8%) Ironworking -73283 Alnus glutinosa 741–767 cal AD (28.4%) platform 904 (Alder)

3. AIMS AND OBJECTIVES

3.1 The aims of the excavation were to establish the character, quality, date, significance and extent of any archaeological remains surviving within the development area. The objectives of the excavation were laid out in the WSI (CA 2018), as follows:

6 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

• to identify, investigate and record any significant buried archaeological deposits/features at the site, prior to their destruction by the proposed development;

• recover and analyse any artefactual evidence;

• sample and analyse environmental remains to enhance understanding of past land use and economy; and

• archive, report on and publish the archaeological results at a level appropriate to their significance.

3.2 The aims and objectives of fieldwork were consistent with the period research strategies outlined in the South West Archaeological Research Framework (Grove and Croft (eds) 2012), and in particular:

Research Aim 47: To assess the archaeological potential for studying medieval economy, trade, technology and production; and

Research Aim 44: To develop understanding and identification of early medieval technologies

4. METHODOLOGY

4.1 The programme of archaeological work comprised the excavation of three excavation areas (Areas 13, 14 and 15) near the site’s western boundary (Figs. 2, 3 and 4). The three areas totalled 1,890m² in extent, and were targeted on features identified by the evaluation (Fig. 2):

• Area 13 measured 875m², and was centred on a ditch containing iron slag and burnt material, recorded in Trench 8 of the previous evaluation (Fig. 4);

• Area 14 measured 275m², and was centred on a suspected ironworking surface recorded in Trench 9 of the evaluation; and

• Area 15 measured 740m², and was centred on a gravel layer recorded in Trench 10 of the evaluation, and on anomalies identified by geophysical survey, but was not found to relate to the archaeological features recorded by evaluation (Fig. 4).

7 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

4.2 Excavation Areas 13 and 15 were extended beyond their initially-intended boundaries, due to the presence of archaeological remains which extended beyond the excavation areas (CA 2019; Figs. 2 and 3). These extensions were agreed and reviewed with Stephen Reed.

4.3 The excavation areas were set out on OS National Grid (NGR) co-ordinates, using Leica GPS, and were scanned for live services by trained CA staff, using CAT and Genny equipment, in accordance with the CA Safe System of Work for Avoiding Underground Services.

4.4 Topsoil and subsoil layers were stripped from the excavation areas using a mechanical excavator equipped with a toothless grading bucket. All machining was conducted under archaeological supervision, and ceased when the first significant archaeological horizon or natural substrate was revealed, whichever was encountered first.

4.5 Generated spoil was monitored in order to recover artefacts. Where necessary, the hand-cleaning of stripped surfaces was employed to better define any identified archaeological deposits/features. The investigation of archaeological features aimed to recover their plan and any structural or depositional sequences. All industrial features and deposits were 100% excavated. All discrete features (postholes and pits) were 50% hand-excavated. All linear features (ditches) were sampled to a maximum of 10%, including all terminals and intersections. All archaeological features were recorded in plan, using Leica GPS.

Environmental remains 4.6 Care was taken to identify deposits exhibiting environmental potential, and 26 bulk samples and four monolith samples were taken, in accordance with methodologies outlined in Cotswold Archaeology guidelines (CA 2012). Secure deposits, especially those related to industrial activity, were considered for the sampling of charred plant remains, charcoal and mineralised remains.

Metallurgical residues 4.7 Slag and other metallurgical residues were hand-collected and recovered from samples. All slag and metalworking debris identified during the excavation was recovered and retained in accordance with current guidance (Historic England 2015b).

8 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

5. THE STRATIGRAPHIC RECORD: FACTUAL DATA AND STATEMENT OF POTENTIAL (FIGS 3–7)

The Stratigraphic Record: Factual Data

5.1 This section provides an overview of the excavation results; detailed assessments of the stratigraphy, finds and environmental samples (biological evidence) are to be found in Appendices A–E of this report.

Fieldwork summary 5.2 Archaeological features were identified in all three excavation areas (Areas 13, 14 and 15), and included pits, postholes, two ditches, two deposits of slag and metalworking debris, a large rectangular pit and the remains of four non-slag tapping iron smelting furnaces together with associated features, including undated pits and postholes (Table 2; Fig.3). These features were provisionally assigned to the early medieval period, on the basis of the radiocarbon dates obtained from metalworking debris from Trench 9 of the evaluation, together with the spatial relationships between features and the presence, or otherwise, of ironworking residues within their fills.

Soils and Subsoils 5.3 The natural substrate identified in all excavation areas comprised a mix of grey/yellow and red clays, with outcrops of small and medium-sized stones and gravel. The natural was generally overlain by a modern topsoil of 0.2−0.45m thickness, although a sequence of peaty, alluvial and colluvial layers, measuring up to 1.2m in thickness, infilled a natural depression in the north-western part of all three excavation areas. Modern stone field-drains were also encountered in all areas (Fig. 3). All archaeological features were cut into the natural substrate, and were sealed either by topsoil, or by colluvial deposits, where present (near the natural depression). The visibility of features across the site was relatively good, and stratigraphic sequences, where present, were established with a high degree of confidence.

9 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Phasing 5.4 Stratigraphic analysis of finds and features has indicated a single phase of archaeological activity, comprising the early medieval period (Period 1), of AD 410- 1066. Small assemblages of prehistoric flint/chert and post-medieval pottery were recovered as redeposited or intrusive material within early medieval or undated features, and no features of confirmed features of these dates were recorded. A number of features could not be assigned to period on the basis of finds or spatial or stratigraphic relationships, and therefore remain undated.

Period 1: early medieval (AD 410−1066) 5.3 Two early medieval radiocarbon dates were obtained from metallurgical deposits associated with the feature interpreted as an ironworking platform found in evaluation Trench 9 (CA 2017; Table 1). During the current excavation, no dateable artefacts of early medieval date were recovered from identified features. However, the metallurgical activity identified across the three excavation areas has been assigned to the early medieval period on the basis of these radiocarbon dates, and the similarity of features and slag deposits to those recorded immediately south of the site (Rainbird and Young, 2015; see archaeological background above).

5.4 Iron smelting furnace 14006 was identified in Area 14, immediately to the south of evaluation Trench 9 (Fig. 3). This excavation area was targeted on the possible ironworking platform recorded in Trench 9, although it was evident that the slag- dumps and metallurgical residues recorded during the evaluation were not located within an ironworking platform. These instead represented the accumulated ironworking residues which had been dumped down-slope from adjacent pit furnace 14006.

Furnace 14006 5.5 In Area 14, the construction of furnace 14006 was revealed by a sub-circular and steep-sided cut with a concave base (Fig. 6, section BB). This measured 0.7m in length, 0.6m in width and 0.54m in depth, and was situated on a level surface, some 2m to the south-east of a natural depression. Furnace 14004 contained a sequence of six fills (14007, 14008, 14012, 14009, 14010 and 14011). The basal fill comprised a 0.07m-thick, dark-grey silty clay, containing abundant charcoal inclusions, which represented the remnants of fuel used during the last smelting. Fill 14009, a slag- puddle or cake, was identified near the base of the furnace, covering basal fill

10 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

14007. This comprised a substantial lump of iron slag measuring 0.35m in length, 0.25m in width and 0.16m in thickness. The slag-puddle probably represents a deposit from the last smelting, which had simply been left in situ. Slag-puddle 14009 was overlain by a sequence of two fills (14010 and 14011), the uppermost of which (14011) contained common fragments of orange-coloured fired clay. This deposit may represent the collapsed lining of the cone-shaped superstructure of the smelting furnace.

5.6 Layer 14013 was located 3m to the north-west of furnace 14004, and at the bottom of the adjacent natural depression (Fig. 3). This layer measured 4.2m in length, 3.5m in width and 0.2m in thickness, and comprised dark-black/grey silty clay, with frequent inclusions of iron slag and charcoal. Layer 14013 probably represents a sequence of slag and residue deposits which had been dumped down the natural slope, and which originated from successive smelting episodes of the adjacent furnace.

5.7 Three pit furnaces (15004, 15009 and 15017), three pits and a slag deposit were identified in Area 15, 35m to the north-east of furnace 14004 (Figs. 3 and 7). The features in Area 15 were revealed to the immediate south-east of previous evaluation Trench 11. The suggested gravel surface recorded in this trench was observed during excavation to correspond to a natural gravel deposit.

5.8 Furnace pits 15004, 15009 and 15007 were clustered on a level terrace on the natural slope, close to the edge of the natural depression exposed in the three excavation areas (Fig. 3). Furnace 15004 was sub-circular in plan, and measured 0.72m in length, 0.6m in width and 0.5m in depth, with steep sides to a concave base (Fig. 7, section CC). It contained two fills (15005 and 15006), the lowest of which (15005) contained frequent charcoal inclusions. An upper fill, 15006, of mid- grey/brown clay silt, produced several fragments of burnt fired clay, which most probably represented the collapse of the furnace superstructure. Furnace 15004 retained part of its clay lining (15023) in situ; this was 0.12m thick, and was attached to the furnace cut on its western part. A large fragment of iron slag was found adhering to this clay lining.

5.9 To the immediate north-west of furnace pit 15004, was located sub-oval furnace pit 15007 (Fig. 3). This comprised a shallow cut measuring 1.15m in length, 0.8m in width and 0.18m in depth, which featured moderately-sloping sides and a flat base.

11 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Iron slag, fired clay and charcoal were retrieved from its single fill, 15005, of dark- brown clay silt. In view of the flat base of this cut feature and its proximity to furnace 15004, it has been suggested that pit 15007 could represent a hollow associated with the operation of a bellows mechanism, which would have enabled the furnace to reach the required smelting temperature.

5.10 Furnace 15009 was recorded some 0.5m to the north of furnace 15004, and was evident within an oval cut measuring 1.15m in length, 0.8m in width and 0.32m in depth. It featured steep sides with a flat base (Fig. 7, section DD). A dark, grey/brown/red concentration of large lumps of iron slag (15010) was found at the base of furnace 15009, although none appeared to comprise an in situ slag-puddle comparable to that from furnace 14004, described above. Slag deposit 15010 was covered by fill 15011, of dark-grey/yellow/brown silty clay, from which slag and several large charcoal fragments were recovered. These charcoal fragments appear to represent the remnants of fuel used during the smelting process. Fill 15011 was covered by upper fill 15012, which contained common fragments of burnt clay, probably representing the collapse of the clay superstructure of the pit furnace.

5.11 Further evidence for the use of furnaces 15004 and 15009 was observed in slag deposit 15013, of mid-brown/grey silty clay, which was identified 4m to the west, and within the slope of the natural depression (Fig. 3 inset). This layer measured 2.4m in length, 1.2m in width and 0.1m thick, and contained abundant slag and metallurgical residues, which had most probably originated from furnaces 15004 and 15009, and been dumped downslope.

5.12 The remains of another probable pit furnace (15017) and an associated pit, 15020, were located on a level terrace by the edge of the natural slope, and 5m to the north of the group of features represented by furnaces 15004 and 15009, and pit 15007 (Fig. 3). A probable furnace, 15017, was evident as a sub-circular cut with moderately-sloping sides and a concave base, which measured 1.1m in length, 1m in width and 0.28m in depth. It contained two fills (15018 and 15019), the uppermost of which (15019), of dark-grey/brown clay silt, contained large fragments of iron slag. A number of redeposited items of struck flint or chert were recovered from both fills (Table 5). Although probable furnace 15017 contained no evidence of in situ slag- puddles or attached clay-lining material, the presence of substantial slag fragments within its upper fill suggested a function as a furnace.

12 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

5.13 To the immediate north-east, pit 15021 measured 0.85m in length, 0.6m in width and 0.15m in depth, and featured a single, charcoal-rich fill of brown/grey silt. Pit 15021 appeared similar to pit 15007, and suggested the accommodation of a bellows mechanism servicing adjacent furnace 15017.

5.14 Within Area 13, the most significant feature comprised a large pit (13018), which had been identified as a geophysical anomaly and was encountered in Trench 8 during the evaluation, where it was interpreted as a ditch (Figs. 3 and 5, section AA). During the excavation it was shown that instead of a ditch it was a large pit of rectangular shape in plan. Pit 13018 measured 3.1m in length, 1.25m in width and 0.25m in depth, with moderately sloping sides and a flat base (Figs. 3 and 5, section AA). It contained two undated fills (13019 and 13020), the lower of which (13019), of dark-grey clay silt, contained a substantial quantity of charcoal and burnt flint (see tables 9 and 5). The upper fill, 13020, of dark, brown-grey clay silt, yielded iron-slag fragments. Pit 13018 was truncated by a modern stone drain, and could be related to the iron smelting activity identified in Areas 14 and 15, possibly as an ore-roasting pit. Alternatively, it might represent a feature relating to charcoal production for the nearby furnaces.

5.15 In the north-western part of Area 13, a 0.1m-thick charcoal-rich layer (13082), of light-grey silty clay, was identified (Fig. 3). No finds were retrieved from this layer, which most probably represents a downslope deposit related to the industrial activity associated with rectangular pit 13018.

Undated 5.16 Pit 15014, located in the south-eastern part of Area 15, was circular in plan, with moderately sloping sides and a concave base, and measured 1.5m in width and 0.25m in depth (Fig. 3 inset). This pit contained two fills, the lower of which, 15015, comprised light/mid-grey silty clay. Upper fill 15016, of dark-grey silty clay, yielded charcoal but no finds. The function this pit was unclear.

5.17 Numerous undated features in Area 13 included thirty-seven individual pits and postholes, and two ditches (Fig. 3; Table 2). These features lacked dateable finds and clear spatial or stratigraphic relationships with other features. Their exact functions remain unclear at present. A small number produced small quantities of ironworking slag, fired clay and charcoal indicating possible association with early medieval ironworking activity.

13 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Table 2: Summary of undated pits and postholes in Area 13

Pit Fill Description of fill Finds

L (m) W (m) D (m) Sides Base Assoc’n Sample

13002 0.3 0.3 0.06 13003 Mod Flat Grey/brown silty clay 13004 - - 13004 0.35 0.36 0.12 13005 Mod Flat Grey/brown silty clay 13002/6 - - 13006 0.31 0.31 0.08 13007 Mod Conc Dark-grey silty clay 13002/4 - <1> 13008 0.48 0.48 0.09 13009 Gent Flat Brown/Grey silty clay - - - 13010 0.51 0.51 0.2 13011 Conc Conc Dark-grey silty clay - B. Stone <2> 13012 0.30 0.37 0.14 13013 Mod Conc Grey/brown silty clay - Charcoal <3> 13014 0.5 0.64 0.09 13015 Gent Flat Grey/brown silty clay 13018 - - 13016 1.5 0.82 0.32 13017 Mod Conc Grey/brown silty clay 13021 - - 13024 0.7 0.5 0.29 13023 Vert Irreg Grey/brown silty clay 13059 Slag & char - 13026 0.2 0.2 0.16 13025 Vert Flat Grey/brown silty clay 13028 Charcoal - 13028 0.35 0.3 0.12 13027 Mod Conc Grey/yellow silty clay 13026 Charcoal - 13029 0.15 0.15 0.11 13030 Steep Flat Grey/brown silty clay 13031 Charcoal - 13031 0.11 0.11 0.06 13032 Mod Conc Grey/brown silty clay 13029 - - 13033 0.21 0.29 0.11 13034 Steep Flat Grey/brown silty clay 13035 Fired clay - 13035 0.5 0.62 0.27 13036 Steep Flat Brown/Grey silty clay 13033 Fired Clay <5> 13040 0.45 0.35 0.12 13039/8 Mod Conc Grey/brown silty clay - Charcoal - 13042 0.37 0.35 0.12 13041 Mod Conc Grey/brown silty clay 13028 Charcoal - 13044 0.25 0.2 0.03 13043 Gent Conc Light-grey silty clay 13042 - - 13045 0.32 0.41 0.18 13046 Mod Conc Grey/brown silty clay - Charcoal - 13047 0.32 0.44 0.2 13048 Mod Flat Grey/brown silty clay - Charcoal - 13049 0.17 0.32 0.15 13050 Vert Conc Grey/brown silty clay 13065 Charcoal - 13052 0.6 0.43 0.14 13051 Mod Conc Grey/brown silty clay 13054 Charcoal - 13054 0.3 0.2 0.13 13053 Mod Conc Grey/brown silty clay 13052 Charcoal - 13056 0.2 0.22 0.06 13055 Mod Conc Grey/brown silty clay 13058 Charcoal - 13058 0.17 0.15 0.04 13057 Mod Conc Light-grey silty clay 13056 - - 13059 0.43 0.68 0.25 13060 Mod Conc Yell/brown silty clay 13024 Charcoal - 13065 0.17 0.17 0.09 13066 Steep Conc Grey/brown silty clay 13024 Charcoal - 13067 1.25 1.1 0.4 13068/9 Mod Irreg Grey/brown silty clay 13027 Slag & char - 13070 0.2 0.2 0.1 13071 Steep Conc Grey/brown silty clay 13083 - - 13072 1.06 0.8 0.1 13073 Conc Irreg Black/grey silty clay 13067 Charcoal <15 13074 0.27 0.27 0.13 13075 Mod Conc Grey/brown silty clay - - > 13076 0.23 0.23 0.11 13077 Steep Conc Grey/brown silty clay 13078 - - 13078 0.18 0.18 0.09 13079 Mod Conc Grey/brown silty clay 13076 - - 13080 0.49 0.66 0.19 13081 Mod Flat Grey/brown silty clay - - - 13083 0.58 0.4 0.16 13084 Conc Conc Light-brown silty clay 13070 - - 13090 1.1 2.9 0.37 13091 Mod Flat Red/Brown silty clay 13092 Charcoal - 13092 0.58 1.05 0.15 13093 Gent Conc Grey/brown silty clay 13090 Charcoal - -

14 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

5.18 The notably concentration of undated small pits and postholes in Area 13 may be significant, although it is not always possible to differentiate between postholes and small pits per se, and nowhere do their arrangements in plan suggest coherent evidence of post-built structures. However, distinct concentrations of these features are evident towards the south-western and south-eastern corners of Area 13, where they may represent simple, possibly open-sided structures (Fig. 3). Silty clay fills appeared to be fairly consistent throughout and a large proportion of the fills were associated with varying inclusions of charcoal but surprisingly little slag. This may suggest an area of activity distinct from the operation of furnaces. A number of smaller features, which could be more confidently interpreted as post or stake-holes displayed shallow, asymmetrical profiles, often with evidence of disturbed or silted fills, suggesting that posts or stakes had been deliberately removed rather than decayed in situ. It is probable that many of these represent evidence for simple, very transient structures, presumably associated with ironworking activity.

Ditches 13061 and 13063 5.19 These features appeared to represent a single, short length of ditch, located towards the south-eastern corner of Area 13, which had been separated by a modern north- west/south-east aligned land drain (Fig. 3). Ditch 13061 was aligned broadly north/south, but south of the intervening land drain, as Ditch 13063, turned to the south-east. Collectively, the two features comprised a short, irregular ditch segment, of c. 7.5m length, and unknown function.

5.20 Ditch 13061 measured 1.5m in width and 0.23m in depth, with gentle, concave sides and a slightly irregular base. Fill 13062, of light-grey silty clay, contained inclusions of sub-angular stones, but no dateable material. Ditch 13063 was of more variable width, and measured only 0.76m wide and 0.45m deep at the excavated intervention. Fill 13064, of grey/brown silty clay, also contained sub-angular stones which suggested a backfilling deposit. The excavator speculated whether ditches 13061/13063 might represent a drainage feature or sump, designed to keep adjacent working areas dry.

Stratigraphic Record: Statement of Potential 5.21 Following the completion of the fieldwork, an ordered, indexed and internally consistent site archive was compiled in accordance with specifications presented in the Management of Research Projects in the Historic Environment (MORPHE):

15 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Project Manager’s Guide (Historic England 2015a). The fieldwork generated the following records:

Context sheets 132 Sections (1:10, 1:20) 26 Sample sheets 28 Digital photographs 234 Matrices 1

5.22 The survival and intelligibility of the site stratigraphy was relatively good, with archaeological remains surviving largely as negative features and deposits. Despite a paucity of stratigraphic relationships, most features have been assigned to a period on the basis of previous radiocarbon dates and spatial relationships.

5.23 Cut features were present at relatively low density throughout the site as a whole, but were notably more dense in Area 13, and were largely discrete. With the exception of those features positively identified as iron furnaces, most contained single fills with no stratigraphic depth or complexity, and few stratigraphic relationships between features were evident. An absence of Period 1 cultural material in all features required that dating be inferred from the radiocarbon dates obtained at evaluation stage, and from spatial relationships and the presence of ironworking residues. Only those features representing confirmed iron furnaces, including 14004, 15004, 15009 and 15017, contained any depth of stratified deposits (Figs. 6 and 7), and these were of some value in interpreting metallurgical process and furnace structure. These remains have been assessed in detail, in Appendix C of this report. A number of features, including individual pits and associated features, remain unattributed to industrial function, and further interrogation of the fieldwork record and comparative analysis with similar features recorded on the Culmstock Road South site (Rainbird and Young 2015) has been identified in the objectives of the Updated Project Design outlined below. The stratigraphy of all three excavation areas therefore retains some potential to contribute some additional information regarding the industrial function and chronological development of the site.

16 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

6. THE FINDS: FACTUAL DATA AND STATEMENTS OF POTENTIAL

6.1 Finds recovered are listed in Table 3, below. Detailed assessments are to be found in Appendices B and C of this report.

Table 3: Quantification of finds Type Category Count Weight (g)

Pottery Medieval/post-medieval 4 50 Worked flint/chert 31 61 Burnt flint/chert 38 91 Fired/burnt clay 304 2801 Ironworking residues Slags 350 151,000

6.2 The range of finds from this site was very limited, and overwhelmingly comprised metalworking residues and burnt clay fragments of early medieval date. None of the other finds recovered related to this phase of activity.

The Finds: Factual Data

6.3 A small group of mixed finds included both worked and burnt items of flint or chert, of which only one appeared to represent a fully worked flake rather than debitage. This material was not diagnostic of period, and appears to represent transient prehistoric activity on or around the site. A small pottery group, comprising four sherds of predominantly post-medieval date, may represent a general pattern of dispersal around the margins of Hemyock village.

6.4 The finds assemblage was overwhelmingly dominated by fired clay fragments and ironworking residues of assumed early medieval date. The latter are described in detail in Appendix C of this report. The former, while generally retaining no features diagnostic of function, appear to represent the fragmented remains of furnace structures and linings. The assemblage of ironworking residues included two largely intact furnace bases (FBs), one found in situ in Furnace 14006, and one dumped into pit 15009, as a part of fill.

6.5 Furnace 14006 contained a large slag block (a ‘furnace base’) and some smaller fragments (total weight 25.4kg) in adjacent contexts, probably representing most of the slag from its last smelt. Fragmented slag material was present in two distinct facies, including puddled dense slag and plano-convex masses of porous slag. The assemblage included moderate amount of furnace ceramic and associated

17 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

materials, with a total weight of almost 16kg. The majority of the vitrified material displayed almost planar faces of vitrified material, backed by dominantly grey furnace ceramic. Small quantities of iron ore (30g) and iron-rich concretions were also recovered. Most of this material was closely associated with furnaces 14006 and 15004, although quantities of residues were also recovered from pit fills and dumps across Areas 14 and 15.

The Finds: Statements of potential Mixed Finds 6.6 The assemblage of mixed finds is small, and with the exception of fired clay, and metalworking debris bears no relationship to the principal early medieval phase of activity on this site. The pottery and worked and burnt flint/chert has no potential to add to knowledge of the development of the site, or of excavated features or deposits, and does not warrant further analysis. The assessment of the lithic and pottery assemblages presented in this report is therefore sufficient for the purposes of the archive.

6.7 Some 304 fragments of fired clay (2801g) were assessed as part of the mixed finds assemblage, most of which retained no features or surfaces indicative of function, although in a few cases the adherence of slag deposits suggested an origin as furnace lining. Most of the fired clay assessed will have no potential to enhance knowledge of furnace structure or metallurgical process, and therefore does not warrant further analysis. However, an analysis of the small number of fragments associated with adhering slag or related residues may have potential to add to knowledge of metallurgical process, particularly in conjunction with the wider assemblage of ironworking residues. The mixed finds are assessed in further detail in Appendix B of this report.

The Ironworking Residues

6.8 This assemblage includes excellent examples of intact FBs, including the in-situ example from furnace 14006. This resource is supplemented by isolated pieces of other FBs, together with furnace ceramic and probably-related ore fragments. The smelting sites are compact, with a volume of residues, suggesting that they are likely to have been relatively short-lived, and hence the materials within them have a higher chance than typical of being directly related. They are therefore likely to

18 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

provide evidence that is much more straightforward to interpret than that from larger sites.

6.9 These residues are representative of a smelting technique that appears widespread across Wessex in the late-middle Saxon period. Regionally, the most complete description of such materials and furnaces relates to the site of Burlescombe (Reed et al. 2006), although no detailed analysis was undertaken there. Some analysis was undertaken for the site at Culmstock Road South (Young 2014; Rainbird & Young 2015), but the analysed materials were all ex-situ from the slag dump (estimated as a total of 22 tonnes), and the site included a wide variety of furnace forms of uncertain age-range. The sites at Ramsbury (Haslam 1980) and Millbook (Tebbutt 1982) were investigated before the advent of modern analytical techniques and their theoretical frameworks. In conjunction with contemporary excavated sites at Churchills Farm and Culmstock Road South, Hemyock, the present material is thus ideally suited to provide detailed evidence for the nature of this particular style of iron-making, and should be capable of shedding important light, not only on the operation of this particular site, but also of the other, larger, sites in Hemyock, and of similar sites further afield, across southern England.

6.10 Accordingly, detailed laboratory analysis is recommended. If possible, this should be undertaken through detailed dissection and analysis of sub-samples from a single FB (to ensure coherency of raw materials between sub-samples), supplemented where necessary by additional isolated material from the secondary deposits (including furnace ceramic and iron ore). Interpretation of bulk elemental analyses may be made through the mass-balance modelling approaches of Thomas and Young (1999a,1999b), as modified by the graphical approach of Young (2016). This should enable understanding of the local mass balance (i.e. how the various inputs of iron ore, fuel ash and furnace clay combine to produce slag and iron) for the various components of the FB (to understand how the FB forms), as well as the overall balance that enables calculation of output and efficiency. Microstructural investigations by SEM on a smaller suite of materials will enable characterisation of the mineralogy of the slag (allowing investigation of such features as the ability of the slag to capture phosphorus and prevent it entering the iron), and the crystallisation history of the slag (providing further evidence for the evolution of the FB).

19 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

6.11 By following similar methodology to that employed on comparative material at Churchills Farm (Young 2018a), the functioning of the deep, wood-packed furnaces may be directly compared with the later Saxon cereal-packed non-slag tapping and slag tapping technologies employed in later operations at other Hemyock sites.

6.12 The post-excavation laboratory analysis of the materials should be accompanied by a thorough review of the field evidence for the nature of the various pits, particularly within Area 13 (Fig. 3), and by a comparison of the nature of those pits with the multiple pits forming the various ‘groups’ at the adjacent site of Culmstock Road S (Rainbird and Young 2015), with the aim of differentiating the various possibilities of clay extraction, charcoal burning, ore roasting, additional furnaces, working hollows outside a furnace arch, and hollows associated with the blowing apparatus.

6.13 Irrespective of the commissioning of further analysis, this assemblage is important, and as much as possible of the assemblage should be deposited with the institution receiving the site archive. As an absolute minimum, this should include one of the two intact FBs and a representative selection of other components of the assemblage.

7. THE BIOLOGICAL EVIDENCE

7.1 The biological evidence recovered is confined to the 15 bulk samples assessed from furnaces, pit fills and deposits across the site (of the 26 originally recovered), and the four monolith samples taken along the course of a suspected palaeochannel, across Areas 14 and 15. Detailed assessments of the biological evidence are presented in Appendices D and E of this report

Biological Evidence: Factual Data

Charred Plant Remains 7.2 A total of 15 environmental samples were processed from a range of features across the site, mostly of confirmed, or suspected, early medieval date (Appendix D). Samples contained moderate quantities of charred plant remains, mostly of wild species. Samples 4 and 16 contained large quantities of charcoal fragments, a proportion of which were silt and iron-impregnated. This was interpreted as waste material from hearths or furnaces.

20 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

7.3 Four samples (11, 12, 13, and 14) from the fills of furnace 14006 contained no cereal grains, but seeds of wild species and mostly moderate quantities of roundwood charcoal. Similarly, samples from pit furnaces 15004, 15017 and 15009 principally contained oak charcoal fragments, including both roundwood and twigwood, together with sparse seeds of wild plant species. Peaty layer 14003 contained no charred plant remains, and only sparse charcoal fragments.

Monolith Samples 7.4 Four monolith samples were taken from the course of a suspected palaeochannel running across Areas 14 and 15 (Appendix E). These recorded a sequence of sediments and peat deposits, which collectively suggested slower-moving water higher up the palaeochannel, indicating progressive separation of the tributary channel from the River Culm. Lower units comprised silty sediments and peat deposits characteristic of sedimentation in shallow, slow-moving water, with associated anaerobic conditions and correspondingly well-preserved plant remains. Unit 4 in the sequence was more distinctive in texture and colouration, and appeared to represent a deposit of more alluvial character, with evidence of a change in the depositional regime. Unit 2 produced evidence of colluvial accumulation, possibly resulting from historical tillage upslope.

Biological Evidence: Statement of Potential

Charred Plant Remains 7.5 There is no potential for the charred plant remains to provide further information on the nature of surrounding settlement and landscape, the range of crops and the crop processing techniques during the early medieval period, due to the paucity of remains recovered.

Charcoal 7.6 There is some potential for a more detailed analysis of a selection of the charcoal assemblages from a number of features from the early medieval period, to provide further information regarding the range of taxa used and the exploitation and management of local woodland resources. The potential for such analysis is limited, due to the condition and silt/iron impregnation of much of the charcoal recovered, as this will make it difficult/impossible to identify this to species. Further analysis may assist in determining whether there was any species selection process employed for use in iron smelting. The assemblages could be compared with charcoal

21 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

assemblages from other industrial features of this date in the area, such as that from previous work at Culmstock Road South (Rainbird and Young 2015). This information has the potential to augment the data for iron-working in this area during this period.

7.7 It is recommended that a selection of charcoal assemblages be considered for more detailed analysis from the early medieval period. The specific samples that should be considered for detailed charcoal analysis are samples 4 (Pit 13018), 15 (Pit 15007), 11 (Furnace 14006), 13 (Furnace 14006), 18 (Pit Furnace 15004), 22 (Pit Furnace 15017) and 23 (Pit Furnace 15009).

7.8 No further work is required for the charred plant remains. This is due to the general lack of environmental evidence within the assemblages, and such limited environmental evidence that is present has little or no potential to add to wider knowledge or understanding of the site.

Radiocarbon Dating 7.9 A selection of charcoal assemblages are suitable for submission for further radiocarbon dating. This recommendation has been itemised in the Updated Project Design, below. The specific samples with good potential for radiocarbon dating are samples 4 (pit 13018), 15 (pit 15007), 11 (Furnace 14006), 13 (Furnace 14006), 18 (Pit Furnace 15004), 22 (Pit Furnace 15017), 25 (Pit Furnace 15017) and 23 (Pit Furnace 15009).

The Monolith Samples

7.10 The peat deposit, Unit 5, has high paleoenvironmental potential. The waterlogged, anaerobic conditions which have resulted in this deposit have provided an excellent environment for the preservation of archaeological and palaeoenvironmental remains. Units 6 and 4 also exhibit anaerobic condition favourable for pollen preservation. Therefore, if dating evidence is available, and peat formation was contemporary with ironworking activity, the sampling of Units 4, 5 and 6 for pollen and waterlogged remains could be considered. However, it should be emphasised that the palaeochannel is an otherwise undated, natural feature and even if contemporary is unlikely to have had a functional relationship with ironworking activity on this site. Pollen analysis would therefore have very little potential for enhancing knowledge of early medieval ironworking activity.

22 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

7.11 Units 3 and 2 only have a moderate/low palaeoenvironmental potential. Any possible archaeological and palaeoenvironmental remains are likely to be derived and redeposited from the area upslope. Therefore, sampling for pollen and waterlogged remains from these units is not recommended. Unit 1, a recent topsoil, has no palaeoenvironmental potential, and no sampling of this deposit is recommended.

8. DISCUSSION

8.1 The excavation confirmed the results of the geophysical survey and field evaluation, indicating that the remains of early medieval ironworking furnaces and associated features were present on the site. Although the finds assemblage associated with these remains was largely confined to metalworking residues, an assessment of these was able to confirm the metallurgical processes involved. Activity on this site appears to be contiguous, and probably contemporary with that recorded to the south of Culmstock Road in 2010 (Rainbird and Young 2015), and thus part of a significant concentration of early and later medieval ironworking sites around Hemyock and across this part of the Blackdown Hills (Hughes and Jones 2013; Blaylock 1989).

8.2 Hemyock is suggested as a royal manor at the time of the Domesday Survey, and the history of ironworking on this, and a number of contemporary sites, including Pucklechurch, Glos and Ramsbury, Wilts, appears to reflect later Saxon royal policy rather than the evolution of technology (Young 2018, 214). A suggested orientation of ninth-century economic development towards the south-west under Alfred (ibid.) is supported by the radiocarbon dates from Churchills Farm, Hemyock, where ironworking commenced in cal AD 870-910 (68% probability), and ended in cal AD 905-950 (68% probability). This suggests a relatively short period of industrial activity, which broadly conforms to the reigns of Alfred (r. 871-899) and Edward the Elder (r. 899-924). Dates obtained from nearby Burlescombe, Devon suggest a comparable ironworking chronology on that site, of cal AD 840-945 (68% probability) (Reed et al. 2006). The technical development of smelting, including the ninth- century introduction of slag-tapping furnaces, suggests that influence of itinerant metalworkers, possibly from those parts of Mercia which had fallen under the control of Wessex.

23 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

8.3 The source of iron ores in the Hemyock area remains uncertain. At Churchills Farm, recovered ore bore a strong similarity to that from Dunkeswell Airfield, some 5km to the south (CA 2015; Nichol 2016). This area is associated with extensive spreads of ore-extraction pits, which are readily visible on Lidar survey plots, but seemingly without the characteristic signatures of iron furnaces (Smart 2018b, 217, fig. 130; Environment Agency 1998-2002). Other possible sources around Hemyock include Hackpen Hill, 1.6km to the south-west, which is also associated with extensive areas of extraction pits (Griffith and Weddell 196, 32), and North Hill Common, Broadhembury, with ore pits covering an area of approximately 30ha (Smart 2018b, 216-7, fig. 130). North Hill Common and Dunkeswell effectively represent a single zone of extraction, which extended for some five kilometres along the crest of the Upper Greensand ridge. Smart (2018b, 218) has pointed out that slag dumps and smelting sites are principally located along valley sides, within areas characterised by those irregular and semi-irregular field patterns which represent the earliest elements in the scheme of enclosure (Smart 2018b, 219, fig. 131; Rippon et al. 2015). Numerous extraction pits have been recorded by Lidar survey in this part of the Blackdown Hills (Smart 2018b, 221, fig. 133; Reed 1997). Most of these were either located on the margins of fields or cut across by field boundaries, thus suggesting a relatively early date. Such relationships suggest that the iron industry was integral to the early medieval rural landscape, and not a liminal activity.

8.4 Most slag dumps in the surrounding area remain undated. That at Tedburrow Farm, some 1.1km to the west of Churchills Farm, was dated by charcoal to 72-395 cal AD, while that at Bywood Farm was dated to 664-889 cal AD, indicating a long-term interest in ironworking in this area. Ongoing requirements for clay for the maintenance and rebuilding of furnaces are estimated at approximately 1 tonne per furnace per annum. Suitable clay may have been dug on, or very close to, the site, as Triassic mudstones underlay Hemyock village, although at Churchills Farm, there was no confirmation that the pits associated with furnaces had originally been dug for clay extraction.

8.5 Wood charcoal taxa associated with the Churchill’s Farm site were surprisingly diverse, predominantly comprising oak (83%), hazel and alder, but also including willow, pear and apple, blackthorn, cherry and hawthorn. Young and Rainbird (2015, 311) have suggested that the apparent dominance of oak in furnace assemblages may in fact represent pit packing, rather than furnace fuel, which may, as here, have been altogether more varied in composition. It is estimated that 0.8ha of woodland

24 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

was required to produce sufficient charcoal to smelt one tonne of iron, suggesting that considerable demands were made on surrounding woodland resources, probably involving highly-managed coppicing. No evidence of charcoal-burning platforms has been identified, although these are likely to be located within the more heavily wooded higher valley slopes. The presence of willow and poplar charcoal suggests that trees adjacent to local watercourses, including the River Culm, were cut or managed in the same way. The presence of scrub species, including blackthorn, hawthorn and hazel, suggests that surrounding hedged field boundaries were established prior to the operation of the furnaces, particularly as there is no evidence of large-scale reorganisation of the surrounding countryside.

9. UPDATED AIMS AND OBJECTIVES

9.1 Assessment of site stratigraphy, finds and biological evidence indicates that these have little potential for further analysis and that, with the exception of the ironworking residues, the scope for updated aims and objectives is limited. However, to optimise the potential of site data, the following updated objectives may provide a basis for further analysis:

Objective 1: to refine the dating framework for ironworking 9.2 While two early medieval radiocarbon dates were obtained at evaluation stage (Table 1), the assessments of both ironworking residues and charred plant remains contained in this report have identified the desirability and feasibility of further scientific dating to provide a broader chronological framework for the site. Specific charcoal samples with good potential for radiocarbon dating have been identified from samples 15 (furnace pit 15007), 11 and 13 (furnace pit 14006) and 18 (Pit Furnace 15004).

Objective 2: to analyse ironworking processes 9.3 Young (Appendix C) considers this assemblage of ironworking residues to be ideally suited to provide detailed evidence for the style of iron-making practiced here, and to be capable of providing important information, not only on the operation of this particular site, but of the other sites in Hemyock and the surrounding area. Accordingly, detailed laboratory analysis should be undertaken through detailed dissection and analysis of sub-samples from a single FB, to be supplemented where necessary by additional isolated material from secondary deposits (including furnace ceramic and iron ore). Microstructural investigations by SEM on a smaller suite of

25 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

materials characterise the mineralogy of the slag (allowing investigation of such features as the ability of the slag to capture phosphorus and prevent it entering the iron) and the crystallisation history of the slag.

Objective 3: comparative analysis of ironworking and ancillary features 9.4 Post-excavation laboratory analysis of ironworking residues should be accompanied by a thorough review of the field evidence for the nature of the various pits and associated features, and by a comparison of these with the multiple pits forming the various ‘groups’ at the adjacent site of Culmstock Road South (Rainbird and Young 2015). Such comparison would have the aim of differentiating various industrial functions, including clay extraction, charcoal burning and ore roasting. Such comparative analysis would also be of value in identifying additional furnaces, working hollows outside a furnace arch, and the hollows associated with blowing apparatus.

10. PUBLICATION

10.1 The results from the investigations north of Culmstock Road, Hemyock are of at least regional significance, and merit publication. The evidence of early medieval ironworking complements that from a number of contemporary sites around Hemyock, and across the Blackdown Hills more generally. The significance of this site may be enhanced by the fact that the smelting areas are compact, with a volume of residues, which suggests that they may have been relatively short-lived, and consequently the materials within them have a high chance of being directly related. They are therefore likely to provide evidence which is much more straightforward to interpret than that of larger sites. The analysis of ironworking residues and associated features will have implications for the development of ironworking technologies during this period, and will place Culmstock Road amongst a number of significant contemporary comparator sites across southern England. It is proposed that an appropriate report on the investigation and subsequent analysis is published in a future volume of the Proceedings of the Devon Archaeological Society.

Synopsis of Proposed Report 10.2 An Early Medieval Ironworking Site north of Culmstock Road, Hemyock by Jonathan Orellana and Tim Young

26 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Words

Introduction Location, topography and geology 200 Archaeological background 300 Project background 150 Excavation Results Chronological discussion of the major phases and features of the site Site narrative 1500 Finds 300 Ironworking Residues 2500 Plant macrofossils and charcoal 500 Monolith samples 300

Discussion 1500

Bibliography 400 Appendices

Total words 5500 Approximate pages @ 800 words/page 7

Pages

Illustrations 4 Tables 4

Total publication estimate 15 pages

11. PROJECT TEAM

11.1 The analysis and publication programme will be quality assured by Karen Walker MCIfA (Principal Post-Excavation Manager), and managed by Richard Massey MCIfA (Post-Excavation Manager), who will contribute to the discussion as senior author, and co-ordinate the work of the following personnel:

• Jonathan Orellana (Project Officer: PO): Post-excavation phasing, draft report preparation, research and archive

• Eleanor Cox (Illustrator: ILL): Production of report illustrations

27 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

• Emma Aitken and Sarah Wyles (Senior Environmental Officer, SEO): Charred Plant Remains and charcoal

• Katie Marsden (Finds Officer FO): Mixed Finds

Contributions by the following external consultant will, where appropriate, be managed by the Post-Excavation Manager:

Tim Young, Archaeometallurgist (GeoArch): Ironworking residues

11.2 The final publication report will be edited and refereed internally by CA senior project management, and will be subject to an external review by a referee appointed by the county journal.

Task List

Table 4: Task List

TASK PERSONNEL DURATION/ COST Project Management PXM 0.5 day Research, comparanda PXM 0.5 day Mixed Finds Analysis and report FO 0.25 day Ironworking Residues Analysis and report fee Biological material Analysis & report of charcoal external fee Selection and submission of radiocarbon samples SEO Radiocarbon dating fee Excavation results PO 0.5 day Illustration I 2 days Discussion, conclusions PXM/TY 0.5 day/fee Acknowledgements, bibliography PXM 0.25 day QA PPXM 0.5 day Submission to external referees Editing & revisions PXM 0.5 day SUBMISSION OF PUBLICATION TEXT Archive Research archive completion FS 0.5 day Deposition FS 0.5 Museum and ADS Costs fees Publication Printing Devon Proc fee

28 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

12. TIMETABLE

12.1 For a journal publication project, CA would normally aim to have completed a publication draft within six to nine months following approval of the updated publication project design. Such an estimated timescale would be subject to the requirements of specialist analysis and reporting of ironworking residues.

13. ACKNOWLEDGEMENTS

13.1 The excavation and post-excavation projects were funded and supported by Waddeton Park Ltd. Fieldwork was undertaken by Jonathan Orellana, assisted by George Gandham, Parris Stubbings, Jake Godfrey, Christina Tapply and Jerry Austin. The excavation report was written by Jonathan Orellana. The mixed finds reports were written by Katie Marsden, and the charred plant remains and charcoal reports by Emma Aitken and Sarah Wyles. The assessment and reporting of the monolith samples was undertaken by Agata Kowalska, and the illustrations were prepared by Eleanor Cox. The archive has been compiled and prepared for deposition by Hazel O'Neill. The fieldwork was managed for CA by Derek Evans, and the post-excavation was managed by Richard Massey, who also contributed to this report.

14. STORAGE AND CURATION

14.1 The archive is currently held at CA offices in Andover, while post-excavation work proceeds. Upon completion of the project, and with the agreement of the legal landowners, the site archive and artefactual collection will be deposited with Royal Albert Memorial Museum (RAMM), Exeter which has agreed to accept the complete archive upon completion of the project (Accession number RAMM:42/2017). A summary of information from this project, set out within Appendix F, will be entered onto the OASIS online database of archaeological projects in Britain.

29 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

15. REFERENCES

Allum, C. 2009 Land to the east of Castle Dene, Culmstock Road, Hemyock, Devon. An archaeological excavation and watching brief report, Unpublished Context One report.

BGS (British Geological Survey) 2014 Geology of Britain Viewer http://mapapps.bgs.ac.uk/geologyofbritain/home.html. Accessed 9 December, 2014.

Blaylock, S.R. 1989 An Archaeological Survey of Hemyock Castle, Unpublished Exeter Museums Archaeological Field Unit Report 89.03.

Canti, M., G. 2003 ‘Earthworm Activity and Archaeological Stratigraphy: A Review of Products and Processes’, Journ. Archaeol.Science 30, 135–148.

Cotswold Archaeology (CA) 2009 Land North of Culmstock Road, Hemyock: Archaeological Evaluation, Unpublished CA Report No. 17743.

Cotswold Archaeology (CA) 2012 The taking and processing of environmental and other samples from archaeological sites, CA Technical Manual No. 2.

Cotswold Archaeology (CA) 2015 Land at Dunkeswell Airfield, Dunkeswell, Devon. Archaeological Strip Map and Sample Excavation Report, Unpublished Cotswold Archaeology Report No. 14418. Cotswold Archaeology (CA) 2015 Land North West of Hemyock, Devon: Heritage Desk- Based Assessment, Unpublished CA Report No. 15098

Cotswold Archaeology (CA) 2017a Land north of Culmstock Road, Hemyock, Devon: Archaeological Evaluation, Unpublished CA Report No. 16643

Cotswold Archaeology CA 2017b Fieldwork Recording Manual, CA Technical Manual No.1.

Cotswold Archaeology CA 2019 Land North of Culmstock Road, Hemyock, Devon. Archaeological Excavation. Fieldwork Summary, Unpublished CA report.

Chartered Institute for Archaeologists (CIfA) 2014 Standard and Guidance for Archaeological Excavation.

Crew, P. 1993 Blackdown Hills, Devon: preliminary report on the slags from Sweetlands, Liverhays, Moorlands, Hemyock and Woodbury Great Close, Unpublished report.

English Heritage 2007 Geoarchaeology: Using Earth Sciences to Understand the Archaeological Record, English Heritage.

30 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Environment Agency, 1998-2012, LiDAR DSM data JPEG image (1m resolution), LIDAR ST0906 Environment Agency JPEG DSM 05-MAR-2010 (Cartographic).

Exeter Museum Archaeological Field Unit 1992-1993 Blackdown Hills, Survey Archive.

Exeter Archaeology 2005 Archaeological evaluation at the former Halls Engineering Works, Market Square, Hemyock, Devon, Unpublished Exeter Archaeology Report No. 05.04.

Griffith, F.M. and Weddell, P.J. 1996 ‘Ironworking on the Blackdown Hills: results of recent survey’, in Mining History: The Bulletin of the Peak District Mines Historical Society 13(2), Historical Metallurgy Society Special Publication: The Archaeology of Mining and Metallurgy in South-West Britain, 27-34.

Grove, J and Croft, R. (eds) 2012 The Archaeology of South West England. South West Archaeological Research Framework: Research Strategy, 2012-2017, Taunton, Somerset County Council.

Haslam, J. 1980 ‘A Middle Saxon iron smelting site at Ramsbury, Wiltshire’, Medieval Archaeology, 24, 1-68.

Historic England 2015a Management of Research Projects in the Historic Environment: The MORPHE Project Managers’ Guide.

Historic England 2015b Archaeometallurgy: Guidelines for Best Practice.

Hughes, S. and Jones, P. 2013 Hemyock Motors Workshop Site, Culmstock Road, Hemyock: Results of an archaeological watching brief and trench evaluation, Unpublished AC Archaeology report ACD207/2/0.

Jones A.P., Trucker M.E. and Hart J.K. 1999 The Description and Analysis of Quaternary Stratigraphic Field Sections, Quaternary Technical Association, Technical Guide No. 7, 27-76.

Lindbo, D.L., Stolt, M.H., Vepraskas, M.J. 2010 Redoximorphic Features, in: Stoops G., Marcelino, V. and Mees, F. (eds.) Interpretation of Micromorphological Features of Soils and Regoliths, Amsterdam, Elsevier, 129-147.

Marshall, P., Bronk Ramsey, C., Dunbar, E. and Reimer, P. 2018 ‘Radiocarbon Dating’, in Smart, C. (ed.) 2018, 29-49.

Munsell Color 2018 Munsell Soil Colour Chart (2009 rev.), Michigan.

Nichol, M. 2016 ‘A summary report on the excavation of iron ore extraction pits at Dunkeswell Airfield, Devon’, Proc. Devon Archaeol. Soc. 74, 207-16.

31 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Pre-Construct Geophysics (PCG) 2016 Land at Culmstock Road, Hemyock, Unpublished PCG geophysical survey report.

Rainbird, P. and Young, T. 2015 ‘Saxon Iron Smelting Furnaces at Culmstock Road, Hemyock, Devon’, Proc. Devon Archaeol. Soc. 73, 279-313.

Rapp, R.G. and Hill, L.Ch. 1998 Geoarchaeology: The Earth-Science Approach to Archaeological Interpretation (1st ed.), Yale University Press.

Reed, S.J. 1997 Blackdown Hills Ironworking Project, Archaeological recording of an iron extraction pit, Broadhembury, Devon, Unpublished Exeter Archaeology Report 97.38.

Reed, S.J., Juleff, G. and Bayer, O.J. 2006 ‘Three late Saxon iron-smelting furnaces at Burlescombe, Devon’, Proc. Devon Archaeol. Soc., 64, 71-122.

Rippon, S., Smart, C. and Pears, B. 2015 The Fields of Britannia. Continuity and Change in the Late Roman and early Medieval Landscape., Oxford, Oxford University Press.

Smart, C. (ed) 2018a Industry and the Making of a Rural Landscape: Iron and Pottery Production at Churchills Farm, Hemyock, Devon, Oxford, BAR Br. Ser. 636.

Smart, C. 2018b ‘Sourcing of raw materials: iron ore and clay’, in Smart C. (ed) 2018, 216- 222.

South West Archaeology 2005 The Site to the East of ‘Castle Dene’, Culmstock Road, Hemyock, Devon: Results of Evaluation Trenches, Unpublished South West Archaeology Report No. 050702.

Stace, C. 1997 New Flora of the British Isles, Cambridge, Cambridge University Press.

Stead, P. and Payne, N. 2012 Land to the south of Culmstock Road, Hemyock, Unpublished AC Archaeology report A/D476/2/0.

Tabor, R. 2010 ‘A medieval building and metalworking debris on land to the east of Castle Dene, Hemyock’, Proc. Devon Archaeol. Soc. 68, 199-208.

Tebbutt, C.F. 1982 ‘A Middle-Saxon iron-smelting site at Millbrook, Ashdown Forest, Sussex’, Sussex Archaeol.Coll., 120, 19–36.

Thomas, G.R. & Young, T.P. 1999a ‘Bloomery furnace mass balance and efficiency’, in Pollard, A.M. (ed) Geoarchaeology: exploration, environments, resources, Geological Society of London Special Publication 165, 155-164.

Thomas, G.R. & Young, T.P. 1999b ‘A graphical method to determine furnace efficiency and lining contribution to Romano-British bloomery iron-making slags (Bristol Channel Orefield, UK)’, in Young, S.M.M. et al. (eds), Metals in Antiquity, Oxford, British Archaeological Reports Int Ser 792, 223-226.

32 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Tucker, M. E. 2011 Sedimentary Rocks in the Field: A Practical Guide (4th ed), Wiley- Blackwell.

Wilkinson, K. 2009 Southern Regional Review of Geoarchaeology Colluvium, Research Department Report Series No. 3-2009.

Young, T.P. 2005 Evaluation of archaeometallurgical residues from the Heath-Mayfield N7 development (03E0151, 03E0966, 03E0461, 03E0603, 03E0633, 03E0679, 03E0602, 03E0635). GeoArch Report 2005/12.

Young, T.P., 2011 Archaeometallurgical residues from the N7 Castletown to Nenagh scheme, Camlin 3 (E3580), Co. Tipperary, Unpublished GeoArch Report 2011/23.

Young, T. 2013 Evaluation of Archaeometallic Residues from Churchill’s Farm, Hemyock, Devon, Unpublished GeoArch report 2013/06.

Young, T.P. 2014 Archaeometallurgical residues from Culmstock Road, Hemyock, Devon (ACD591), Unpublished GeoArch Report 2014-11.

Young, T.P. 2015a Analyses of ores and residues from Dunkeswell, Devon, Unpublished GeoArch Report 2014-31.

Young, T.P. 2015b Archaeometallurgical residues from Churchills Farm, Hemyock, Devon, Unpublished GeoArch Report 2015-31.

Young, T.P. 2016 Archaeometallurgical residues from Fleet Hill Farm, Finchampstead, Berkshire, Unpublished GeoArch Report 2016/35.

Young, T.P. 2017 Assessment of archaeometallurgical residues from north of Culmstock Road, Hemyock (880160), Unpublished GeoArch Report 2017/03.

Young, T. 2018a ‘The Iron Industry’, in Smart, C. (ed) 2018, 51-98.

Young, T. 2018b ‘Iron Production in Early Medieval Britain: possible cultural influences on technological change’, in Smart, C. (ed.) 2018, 216-21.

Young, T.P. 2018c Assessment of archaeometallurgical residues from Burlescombe (ACD1753 & ACD1767), Unpublished GeoArch Report 2018/11.

33 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

APPENDIX A: STRATIGRAPHIC SUMMARY

Archaeological features were identified at relatively low density throughout excavation Areas 14 and 15, and comprised the confirmed remains of four iron- smelting pit furnaces, two deposits of metalworking debris, a large rectangular pit, two ditches. A number of uncharacterised and undated pits and postholes were recorded in Area 13 (Fig. 3). The recorded features were located closely to the north- west of a group of similar contemporary features revealed to the south of Culmstock Road (Rainbird and Young 2015), and appear to comprise part of a wider area of ironworking of middle/late Saxon date. A non-slag tapping iron-smelting furnace, containing an in situ iron slag puddle from its last smelt, was identified in Area 14. Metallurgical residues associated with this furnace had been recorded during a previous evaluation, and were radiocarbon- dated by associated charcoal to the early medieval period. In Area 15, three further non-slag-tapping pit furnaces were recorded, together with a slag deposit and three further pits. One of the furnaces retained the remains of in situ clay lining and adhering slag. Two furnaces contained fragments of fired clay in their upper fills, representing the collapse of furnace superstructures. Two closely- associated pits may represent hollows for the operation of bellows. A large rectangular pit recorded in Area 13 was interpreted as a possible ore-roasting pit, but was otherwise uncharacterised. Cut features were largely discrete, with few stratigraphic relationships evident. With the exception of those features positively identified as iron furnaces, most contained single fills of no depth or complexity, and some more extensive spreads of ironworking debris appeared to represent discard around furnaces. The notable absence of early medieval cultural material from all features required that dating was inferred from the radiocarbon dates obtained at evaluation stage, and from the presence of ironworking residues. Only those features representing confirmed iron furnaces, including 14004, 15004, 15009 and 15017, contained any depth or complexity of stratified deposits, and these were of some value in interpreting metallurgical process and furnace structures.

34

© Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

APPENDIX B: MIXED FINDS

Mixed Finds by Katie Marsden Pottery A small assemblage of pottery, comprising four sherds (50g), was recovered from two deposits. Purple ware, a possible Cistercian type, of 15th to mid 17th-century date was recovered from topsoil 13000, with the remainder of sherds (two from 13000 and one handle sherd from deposit 14000) comprising glazed earthenwares of mid-16th to 18th- century date. The latter may comprise South Somerset wares. These finds may indicate a low level of post-medieval dispersal on or around the site.

Lithics A small group of prehistoric worked lithic items, totalling 31 items (61g), was recovered from four deposits (Table 5). The group comprises one worked flint, a flake with blade-like proportions, recovered from deposit 13000, and the remainder comprises chert flakes, five of which were also burnt. An additional 33 fragments of of unworked, burnt chert were recovered from deposit 13019 (91g). None of the recovered lithics can be closely dated. The chert may be derived from local greensand source.

Table 5: Distribution of worked and burnt flint/chert

Context Feature Class Desc of fill Other

inclusions Sample Sample No. Count Wt (g)

13000 Topsoil Flint n/a 1 11 - - 15008 Pit 15007 Flint 16 5 10 Brown/Grey clay silt Slag, charcoal, b. clay 15018 Pit 15017 Flint 25 12 17 Grey/brown clay silt 15019 Pit 15017 Flint 22 13 23 Grey/brown clay silt Slag, charcoal 13019 Feature 13018 B. flint 4 38 91 Dark-grey clay silt Abundant charcoal

With the exception of a single item from topsoil 13000, all worked and burnt flint was recovered from fills or features which were also associated with ironworking residues or waste material of assumed early medieval date. This material therefore appears to be entirely redeposited, and to represent transient activity of broad prehistoric date on or around the site.

35

© Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Fired Clay A large group of fired clay fragments, totalling 304 items (2801g) was recovered by hand-excavation and by bulk soil-sampling of eight deposits. The group is entirely amorphous, and retains no features to indicate form or function, with the exception of the fragments from deposit 15022/15023 which retain slag-like traces on one surface, suggesting use as a furnace lining.

APPENDIX C: IRONWORKING RESIDUES

Assessment of archaeometallurgical residues by Dr T.P. Young

Introduction The submitted assemblage comprised approximately 151kg of material (approximately 350 pieces) derived from the smelting of iron in non-tapping furnaces. In addition to the present assemblage, 14kg of residues were recovered during an earlier archaeological evaluation. The evaluation produced two contrasting 14C dates from the slag dump (15013), suggesting that smelting activity probably dated between the late 7th and 9th centuries.

Excavation revealed the basal pits of two non-slag tapping furnaces, 14006 and 15004, in Areas 14 and 15, respectively. These were of rather similar dimensions, being shaped like an inverted cone, and measuring 0.7m by 0.6m in plan and 0.5m in depth.

Furnace 14006 contained a large slag-block (a ‘furnace base’), and some smaller fragments (total weight 25.4kg), which probably represent most of the slag from the last smelt. Furnace 15004 contained very little slag, suggesting that it had been cleared after its last smelt. Both furnace pits contained significant quantities of ceramic material from the furnace superstructure, indicating that the furnaces had a planar blowing-wall, with a hollow which eroded close to the blow-hole during the smelt, which required repeated repair. The blow-hole (or a blow-hole; it is unclear whether there would have been more than one, but the width of the burr suggests this may possibly have been the case) was 18mm in diameter. This evidence closely parallels that from a previously investigated site to the south of Culmstock Road, from Burlescombe, and probably from an early phase at Ramsbury, Wilts.

Close to Furnace 15004 lay pit 15009. During the course of fieldwork this was described as a furnace but, if so, its elongated shape must indicate a very different

36

© Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design type of structure to Furnaces 14006 and 15009. This seems more likely to be a pit of uncertain purpose, which was used to dispose of 58kg of slag and furnace material, including one complete, fresh furnace base weighing 42.7 kg, and at least three, very worn substantial fragments of furnace bases. A furnace base of fresh appearance (weight 23kg) was of remarkably similar structure to the example from 14006, and must have been produced in an almost identical manner, perhaps in Furnace 15004.

Further north, pit 15017 was similar to pit 15009, and was also interpreted as a furnace; its upper fill yielded 25kg of residues.

Areas downslope of furnaces 14006 and 15004 contained small slag-dumps (contexts 14013 and 15013), yielding 5kg and 40kg of residues respectively. These were broadly of similar character to those closer to the furnaces, but more highly comminuted and weathered.

Features in Area 13 produced only a small quantity of slag of the most dense and robust types seen in the other areas. They also produced very small quantities of a box-stone ore, compatible with that of the ‘clay with flints’ deposits previously postulated to have been the ore employed in the Hemyock area. The site thus provides important new evidence for the smelting of iron at Hemyock.

Methodology

This assessment was conducted in June, 2019. All materials were examined visually, using a low-powered binocular microscope, where required. As an assessment, the materials were not subjected to any high-magnification optical inspection, nor to any form of instrumental analysis. The identifications of materials made in this report are therefore necessarily limited, and must be regarded as provisional.

This project follows a phase of field evaluation, for which an archaeometallurgical assessment was also undertaken (Young 2017). The results of that earlier work are included in this account and discussion. The site is referred to as Culmstock Road North in this report, to distinguish it from the early medieval iron smelting site, located just to the south of Culmstock Road (henceforth Culmstock Road South; Young 2014; Rainbird & Young 2015).

37

© Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Description of the assemblage

Iron smelting slags (Table 6) The assemblage included two, largely intact furnace bases (FBs), one found in situ in Furnace [14006], designated as context (14009), and one dumped into pit [15009] as a part of fill (15010).

Furnace [14006] contained a large slag block (a ‘furnace base’), and some smaller fragments (total weight 25.4kg) in adjacent contexts, probably representing most of the slag from its last smelt. The FB was 420mm wide, 250mm long and 250mm deep, and showed the development of a prominent ‘burr’ (the zone of interaction of the slag and wall) extending across most of the width of the cake. The right side of the cake (when viewed in the direction of blowing; the southern side in plan) showed a particularly tight angle between the blowing wall and the adjacent side. The FB formed within a pit which had been pre-packed with split wood. This wood appears to have been packed steeply (being preserved as moulds in the flow-slags of the lower part of the FB), following the shape of the sides of the pit. Although most of the upper part of the FB was formed of slag with finer porosity associated with the moulds of charcoal fuel, the wood moulds extended through to the upper surface of the slag cake within the tight southern corner. The wood moulds were surrounded by particularly dense slag, often coarsely lobate which, close to the pit wall, formed a continuous, dense sheet.

As far as could be determined from the outside, the upper-central parts of the FB comprised porous slags formed of amalgamated fine prill, which surrounded the fuel moulds. Away from the dense burr, the contact between the FB and the pit wall was mainly a non-wetted surface, showing a strongly lobate structure. The surface layer of this material was conspicuously dark, shiny and often with a rich lustre. Similar material formed the margins of some of the wood moulds.

The FB from context (15010) weighed 23kg. It measured 280mm front-back, 400mm wide and 240mm from flat top to base, with its burr approximately 240mm wide and 150mm deep. The top of the FB was approximately triangular. The right side of the cake turns through a right-angle, and has a foliated structure to the top, suggesting the presence of a collapsed wood-mould. Large wood-moulds are also apparent near the base. By contrast, the left side of the cake displays a prominent, flat-topped slag- flow extending obliquely at 130mm below the top. The base of the FB appears to

38

© Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design mould the base of furnace pit (or a layer of deposit within it), rather than being suspended above further wood.

Almost of all the fragmental slag recorded from the site can be matched with the components observed within these two FBs (in so far as the internal structure of the two can be judged from the external examination). Two slag facies found in the fragmental material are less certainly identifiable with the blocks:

1. Dense slag in the form of a puddle with a free surface (locally showing tapslag-like flow lobes) was an element of several collections within the assemblage (540g from (15019), 1775g from (15006), 2890g from (14011)). A small flow, although probably not of such dense slag, extends from the left-side of the 23kg FB from context 15010.

2. Secondly, some tabular to plano-convex masses of porous slag cannot be matched with observed components. This facies might be interpretable as either a friable upper layer from with a FB (the reserved top of both relatively complete FBs was somewhat fragmental and did not extended far from the wall, suggesting substantial damage during bloom removal), or as an unusual poorly-consolidated smithing-hearth cake (SHC). These materials were unfortunately rather weathered (perhaps indicating that they had contained some metallic iron), and their origin cannot be confirmed on visual inspection alone.

Iron smelting furnace ceramics The assemblage contained a moderate amount of furnace ceramic and associated materials, with a total weight of almost 16kg (over 10% of the assemblage). This component has been assessed separately from the assemblage of amorphous fragments of fired clay described under Mixed Finds, in Appendix B. The majority of the vitrified materials displayed an almost planar face of vitrified material, backed by dominantly grey furnace ceramic. A few pieces showed a more oxidised vitrified ceramic, which was interpreted as deriving from close to the air-supply. One fragment, from context 14011 preserved part of the margin of an 18mm- diameter blow-hole. The axis of the blow-hole was at approximately 60° to the plane of the wall-face within this fragment (though the orientation of the wall-face itself is unknown).

39

© Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Although the majority of vitrified hearth wall fragments showed a planar face, a few pieces showed evidence for the erosion (and relining) of large hollows which were worn back into the wall (probably above the blow-hole). Some pieces had complex textures, apparently showing blocks of the vitrified surface layer sloughing off the wall and sliding down over the face below. However, an alternative interpretation might be that these represent heavily heat-eroded surfaces cutting back into a relined wall area.

Several specimens (both in situ and dumped) showed dense smelting slag penetrating into the pit/superstructure wall along narrow fissures, sometimes in a probably reticulate pattern.

Iron ore A very small quantity of ore (total 30g) was recovered from two fills of features in Area 13 (contexts (13023) and (13071)). These fragments were from box-stones, in one case showing secondary iron oxides (probably goethite) in a fibrous habit between the dense, fine shells of the box-stone. The fragments show some superficial reddening, probably, but not certainly, resulting from roasting.

Concretions and iron Various examples of slag, particularly from the dumps into the lower-lying wetter areas, showed concretionary material adhering to their surfaces. This was developed to any significant extent in only a few examples, and was interpreted as the result of those slag pieces containing metallic iron. A single concretion weighing 126g, from context (15013), appeared to have been formed around a discrete piece of iron.

Distribution of the assemblage

The submitted assemblage comprised approximately 151kg of material (350 pieces), derived from the smelting of iron in non-tapping furnaces. In addition to the present assemblage, 14kg of residues were recovered during an earlier archaeological evaluation.

Furnace 14006 contained a large slag-block (a ‘furnace bottom’) and some smaller fragments (total weight 25.4kg), probably representing most of the slag from the last smelt. The upper part of the fill of the pit above the FB contained much material derived from the superstructure of the furnace, including pieces of the planar blowing wall, an eroded hollow in the wall, and part of a blow-hole of 18mm in diameter, along

40

© Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design with a large amount of red clay from unfired outer parts of the shaft, which were visible in photographs but not present in the collection.

Furnace 15004 contained very little slag, suggesting that it had been cleared after its last smelt. It did contain a moderate quantity of detritus from the superstructure, as well as several blocks of very dense slag, suggesting that fluid slag had descended towards the bottom of the pit, to form a ‘puddle’ below the main FB.

Close to Furnace 15004 lay pit 15009. During fieldwork, this was described as a furnace but, if so, its elongated shape must indicate a very different type of structure to Furnaces 14006 and 15009. It seems more likely that it was a pit of uncertain purpose, possibly used to dispose of 58kg of slag and furnace material (comprising 42.7kg, and including one complete fresh FB and at least three, very worn substantial fragments of FBs). The FB of fresh appearance (weight 23kg) was of remarkably similar structure to the example from 14006, and must have been produced in an almost identical manner. In the context of Area 15, it seems likely that this FB may have been produced in Furnace 15004. Further to the north, pit 15017 was similar to pit 15009, and was also interpreted as a furnace; its upper fill yielded 25kg of residues.

Areas downslope of Furnaces 14006 and 15004 contained small slag-dumps (contexts 14013 and 15013), yielding 5kg and 40kg of residues respectively. These were broadly of similar character to those closer in and near the furnaces, but more highly comminuted.

In addition to the furnaces and associated features/dumps in Areas 14 and 15, several features in Area 13 also produced ironworking residues. Prominent amongst these was the upper fill, 13020, of pit 130 (containing approximately 4kg from the evaluation and excavation), context 13069, which produced 1285g of material. In both cases, the slags indicated a very dense flow surrounding wood moulds. A small fragment (8g) of a dense-flow slag with fuel moulds also occurred in 13023, and was accompanied a small quantity of iron ore (26g). Further iron ore (4g) occurred in context 13071.

Interpretation Identification and location of furnaces

The excavation revealed the basal pits of two non-slag tapping furnaces, 14006 and 15004. These were of rather similar dimensions, ie. shaped like an inverted cone,

41

© Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

0.7m by 0.6m in plan and 0.5m in depth. The slag from the last use of Furnace 14006 remained in situ, inside the pit. The similarity of the 23kg FB from fill 15010 of pit 15009 to the in situ FB in Furnace 14006 provides strong evidence that these were produced by a very similar smelting technique in similar furnaces. Although interpreted as a furnace during fieldwork, pit 15009 is of a completely different form to 14006 and 15004. Its elongated shape, steep sides and flat floor indicate that any interpretation as a furnace would probably require the furnace shaft to be located over the northern end of the pit, where the upper fills contained much furnace superstructure material, together with the 23kg FB. Such an arrangement would allow the southern half of the pit to function as an external working-hollow for accessing the base of the furnace via a furnace arch.

Images of the pit appear to show the FB the right way up, but resting within the centre of the pit and not against the wall, as would be required for the formation of the burr and wall contact. The flow of slag from the mid-part of the FB can also be seen to be directed into the northern end of the pit, not towards the putative arch, as might be expected. The evidence suggests that pit [15009] does not represent a furnace, but had furnace waste dumped into it after its disuse.

Photographs appear to show pit 15009 with a burnt floor, although an alternative interpretation might be that this was employed as an ore-roasting structure (for which there is no evidence in terms of ore finds), or that it was a charcoal production pit. The dimensions of 15009 were 1.15m in length, 0.8m in width and 0.32m in depth (i.e. with a length:width ratio of 1.44). These dimensions are comparable with smaller examples of broadly contemporary charcoal-making pits recorded in central Ireland (Young 2005, 9-10).

Closely associated with furnace 15004 was a second pit, 15007, which was 1.15m in length, 0.8m in width and 0.18m in depth. This pit was of similar dimensions and shape to 15009, but 0.14m shallower. It was suggested during excavation that pit 15007 might represent a hollow associated with the use of bellows for Furnace 15004. That suggestion may be strengthened by the observation that the structure of the FB shows that the furnace appears to have been blown from the north-west.

Pit 15021 similarly lay close to pit 15019, and measured 0.85m in length, 0.6m in width and 0.15m in depth, with a single, charcoal-rich fill. It was likewise suggested that 15021 might have been the location of a bellows for a furnace over pit 15019. That remains a possibility, although the interpretation of pit 15019 as a furnace is not

42

© Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design conclusive. Pit 15017 measured 1.1m in length, 1m in width and 0.28m in depth; such dimensions are very wide and shallow compared with the pits of Furnaces 14006 and 15004.

In summary, the evidence for two furnaces (14006 and 15004) is strong, but the interpretation of the other features of Areas 14 and 15 require further interpretation after detailed interrogation of the field records.

Smelting technology The evidence indicates iron production in non-slag tapping furnaces. The two largely intact FBs display a wide burr, with the surviving furnace ceramic indicating a planar blowing-wall above an eroded embayment around the 18mm-diameter blow-hole. The slag pit was deep (0.5m) and, as dug, conical in profile and measuring 0.7m by 0.6m in plan. The width of the FB suggests the internal working width of the furnace at FB-top level was perhaps 450mm. The above-ground superstructure of the furnace appears to have been carried down into a placed clay pit-lining, although this may not have been entire.

If the two other pits claimed as furnace pits during fieldwork can be substantiated, then they must be rather different in form. In both cases these were shallower, and in the case of 15009 of an elongated shape, which would probably indicate the presence of a furnace arch. The pits of Furnaces 14006 and 15004 were packed with split wood prior to smelting, with the wood extending approximately to ground level. The superstructure was of uncertain form. Although there is no surviving evidence from the sides away from the blowing wall, an enclosed furnace is much more likely than a hypothetical type with a wall solely on the blowing side, as suggested for Burlescombe (Reed et al. 2006). Indeed, the evidence for both fired and unfired clay entering the tops of the abandoned furnace pits suggests a substantial superstructure.

The evidence forms a tight group, with the evidence from Culmstock Road south (Young 2014; Rainbird & Young 2015) and at Burlescombe (Reed et al. 2006), as well possibly as those at Ramsbury (Wiltshire) (Haslam 1980) and Millbook (Sussex) (Tebbutt 1982). This group of furnaces includes a few examples measuring around 0.5m diameter, but most are larger, ranging up to 1m diameter, and have a broadly splayed prolife and a planar blowing-wall.

43

© Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

The sole difference between the present assemblage and the comparative example at Burlescombe is the rare occurrence at Culmstock Road North of dense slag puddles from low in the pit. These were also recorded amongst the assemblage from dump (C4) at Culmstock Road North. They are a minor component of the slag assemblage at both Culmstock Road North and Culmstock Road South, and may have been overlooked at Burlescombe. Substantial slag puddles are a common feature of non-tapping furnaces with a cereal packing (e.g. at Churchills Farm in Hemyock (Young 2015b; Smart et al. 2018), and at Camlin, Ireland (Young 2011). It may be that puddles are able to form in voids formed within the pit packing of whichever composition.

Where dated, these comparative examples also have provided 7th-9th-century dates (Table 8). It would therefore appear that this technology may have been in wide use in late Middle Saxon Wessex, before becoming supplanted by other technologies in the Late Saxon period (Young 2015b; Smart et al. 2018).

Ore source The very limited evidence of ore fragments from Area 13 is significant. These pieces are of very similar morphology to the ironstones associated with the ‘clay-with-flints’ deposits on the top of the Blackdown Hills (Reed 1997; Young 2015a). This is presumed to have been the major ore source for the Hemyock iron smelting during the early medieval and medieval periods, but direct evidence has proved elusive on the other excavated sites. The evidence of ore provenance from other sites in Hemyock was discussed by Young (in Smart et al. 2018, 94-96). Although the evidence points to a general ‘clay-with-flints’ source, other different ore sources are known in the Hemyock area, with that at Churchills Farm most closely resembling an analysed ore from Dunkeswell (Young 2015a).

The finds of box-stone fragments on the present site present a rather more convincing potential model for ore sources than the poor ores recovered from Culmstock Road South and Churchills Farm.

Topographical considerations The location of the site adds to understanding of the topography of early ironmaking in Hemyock. The smelting in Areas 14 and 15 was carried out on ground on the eastern side of a low, wet area. The lower ground provided opportunities for waste disposal, and such a location may have been marginal to any agricultural land. The variety of cut features in Area 13 suggests that this area may have been treated

44

© Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design differently, with the ore and ash dump perhaps suggesting that materials were prepared here. The smelting site at Culmstock Road South was probably also on the margins of a damp area used as a waste dump, but here groundwater appears to have been such a problem that drainage was employed for some of the slag-pit furnaces.

The site has produced no evidence for bloom-smithing, as neither did the Churchills Farm nor the Culmstock Road South sites. The raw iron produced must have been worked into usable iron elsewhere. A model of dispersed iron smelting (perhaps dependent on woodland resources), with centralised bloom-smithing, seems likely. However, no centralised smithing facility has yet been discovered in Hemyock. A possible example of this site-type has recently been discovered near the nearby smelting area at Burlescombe (Young 2018).

45

© Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Table 6: Summary catalogue of materials submitted as archaeometallurgical residues. Weights in g, assm = assemblage of pieces, FB = furnace bottom.

context Item Item notes

Area 13

13020 452 1 dense lobate slags apparently intruding into furnace wall, inner face of block is fragmental and accretionary 1565 9 very dense slags with very large wood moulds 1315 2 very dense slags with large wood moulds, but massive on one side with non-wetted contact, one appears to grade into possible burr area too 306 1 very dense slags with large moulds showing intense surface reddening and smoothing, suggestive of secondary blast influence 70 4 FB fragments 4 1 heated and vitrified pebble fragment

13023 8 1 small fragment of dense slag with fuel moulds, secondarily reddened 26 3 fragments of a small box-stone nodule - shows fibrous goethite within the cracks between the concentric fine dense shell layers, some superficial reddening, so possibly roasted

13069 1285 8 dense flow slag with wood moulds

13071 4 1 tiny fragment of possibly roasted box-stone

Area 14

14007 374 5 dense flow slag with wood moulds

46 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

674 1 very dense lobed slag with non-wetted wall contact, approximately 35mm thick, thickening to 50mm, inner face covered with brecciated accretion, possibly with included discrete slag droplets - could be bloom removal surface?

14009 1 of 2 18000 1 large vertical FB, 200x420x280mm, burr most of width (1 side fractured away) and 120mm deep, rest of wall contact non-wetted, one end gently curved, the other side bends through almost right angle, possibly deformed but probably original, from edge extends to 200mm below top with top surface angling down , if burr section is vertical then 250mm front-back and 250mm top-bottom with flat top. Large wood impressions on base but moulds elsewhere dirty so detail not seen

14009 846 1 block of prilly charcoal-rich slag with wood and a non-wetted surface near one side context Item Item notes

144 1 lobate slags around wood moulds with some very shiny fragmented material 44 1 FB fragment with wood impression 126 6 fragmented FB piece

14010 454 7 fragments of vitrified surface of lining, some pieces appear to show layered structure within rear clay - built with clay roll, all reduced grey to rear 290 1 dense slightly irregular slag with rough surface attached to flat face of reduced-fired wall 58 1 sheet of dense slag with one bright shiny dark non-wetting face, other side irregular 12 1 slightly lobate slag scrap 394 1 porous dark FB material in mostly finely prilly structure, one face shows shiny dark non-wetted surface 274 1 dense irregular porous FB material, locally accreted and brightly reddened, possibly just rust but may be localised reheating?

14011 2890 1 very dense slag block, obscured in part by accretion, 80mm thick, slightly reddened top with localised flow lobes near margin, also trace of wood impressions on top, base convex with descending prills, probably a basal slag puddle, just very possibly a dense SHC 44 1 margin of blowhole at approximately 60 degrees to preserved face, 18mm diameter, preservation of vitrified layer identical to that of the planar fragments from same context 470 11 fragments of planar variously oxidised and reduced fired vitrified wall, some with very slightly oblique lamination within wall 122 1 part of margin of hollow (presumably obliquely above b/h) , reduced fired descending stalactites, resurfaced apparently

47 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

346 1 sloughed wall with trapped iron now forming rusted concretions 82 1 breccia of slag, oxidised ceramic and rust with lining slag, contains iron 494 5 fragments of porous hard rough FB fragments

14012 5050 19 masses of grey clay intruded by dense fine slag, faced by vitrified surface in slight hollow, then resurfaced with planar surface 114 1 dense angular slag fragment, resembles slag attached to face in large blocks 202 1 dense slag piece, probably a fragment of burr 398 1 burr fragment

14013 1090 1 very rusted and slightly exploding mass of lumpy greenish weathering slag, plano convex shape suggests a highly weathered SHC, 150x120x70mm (100mm with 'spike' on base), charcoal-rich 582 1 dense flow lobed slag around fuel moulds, very weathered 288 1 wall slag contact on margin of burr with wood stick running along wall 418 1 extremely dense burr fragment 2050 16 weathered ochreous fragments of FB context Item Item notes

170 assm bits and dust

Area 15

15006 474 1 vitrified surface to oxidised fired planar wall 206 5 small pieces of planar reduced lining 108 1 grey vitrified lining with approximately 100 degree angle - presumably an overhang 676 1 large block of sloughed furnace wall, potentially with embayment to one side originally? 340 16 grey vitrified wall fragments 134 5 grey wall debris with dense dark slag adhering and/or intruding 46 1 dense massive slag fragment 86 1 finely porous FB fragment 2 1 isolated prill

48 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

132 2 rusty fine FB type material 1775 3 blocks of very dense slag with flowed tops and wood moulds, largest indicates slag puddle more than 90mm deep with wood moulds on top and on base and non-wetted wall contact

15008 72 2 concretionary material, probably accreted piece of charcoal-rich slag 28 1 vesicular slag with large mould moulds, accreted 46 1 slightly lobate flow slag with accretion 104 1 unusual flow lobed slag apparently confined to subcircular tunnel - 25-30mm diameter - could be from mole hole/root or even a secondary fill in wood mould

15010 1 of 12000 1 rather worn deep FB. Top surface very well blown, RHS shows strong foliation through to upper surface - vertical wood or deformation, 210mm front-back, 300mm wide, 210mm from blown top to base, slag rises 30mm above flat top either side of central notch. No good wall contact preserved and no certain wood moulds - just lots of charcoal inclusions - though foliated side might indicate wood

15010 2 of 7700 1 200mm x 270mm x 160mm (front-back, width, top-bottom) rounded worn FB, right side angled and fragmented, worn down so no large wood moulds 634 assm fresh FB material with slags penetrating between large wood moulds 246 2 FB fragments, rounded, larger one contains some iron context Item Item notes

15010 3 of 3380 2 large blocks of porous FB material, larger block as basal ridge filled with pieces of brecciated non-wetted dense shiny surface, otherwise all finely prilly 50 assm bits from above

15010 4 of 23000 1 large FB, 280mm front-back, 400mm wide, 240mm from flat top to base, burr approximately 240mm wide and 150mm deep, flat triangular top to mass, RHS bends through right angle and has foliated structure to top - partly altered wall but also wood? Large wood showing near base, LHS shows prominent basal flat-topped flow 130mm below top; base appears to mould base of furnace pit

15011 1 of 288 14 small FB fragments and debris

49 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

824 1 slag with split wood- face of largest 80x95mm, other side of slab is rough, concave and perhaps fragmental, slag dense where in contact with various wood pieces 394 1 finely prilly FB fragment 2210 1 charcoal rich FB fragment, with some mineralised wood, one face is fine dark shiny non-wetted contact 642 2 round rusty lumps of FB material, just very small chance these could be charcoal-rich SHCs 1415 1 wedge-sectioned rounded slab, suggesting origin on lateral part of FB, porous charcoal-rich slag, as with above, just very slight chance this is porous SHC 50 1 twisted slag sheet from slag wall interaction 156 1 oxidised fired wall with slag lobes sliding down surface 1583 1 wall with gently curved face with broad 80mm deep embayment in it, repeatedly relined?

15011 2 of 3215 1 rusty rounded large FB block, mostly charcoal bearing and prilly, a few areas of lobes and possible wood

850 28 variable FB debris 108 4 indurated grey slag wall intersection 10 1 isolated slag prill with fuel/wood contacts

15013 1 of 1220 1 rounded mass of porous weathered slag, probably a worn FB fragment but low density SHC just possible

2030 1 large fragment of coarse porous FB, abundant moulds after charcoal 2070 1 part of bowl-shaped FB block with fine charcoal - forms shallow bowl up to about 100mm thick 20 1 vitrified and glazed grey lining fragment 126 1 rounded lump containing iron 168 4 fragments of porous slag with charcoal 484 5 rounded worn rusted blocks of FB 956 5 dense slags with large wood moulds context Item Item notes

136 2 vitrified indurated grey wall 234 1 rounded fragment of slag with wood mould 88 2 fragments of porous FB with areas of probable blown surface, very irregular and spikey 46 2 dense slag fragments

50 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

15013 2 of 2325 1 block with arcuate side of FB and probable top, some strange laminated Fe-oxides on likely margin, slag mainly rich in quite coarse charcoal fragments, top weathered and slag green and powdery where crushed by contact in transport 6265 1 rounded mass of worn and slightly accreted FB, subconical block but not original top, green weathered slag lobes on one face (top?) and wood moulds down a perpendicular face, otherwise no detail observable 220 5 irregular FB fragments, mostly from around wood moulds 260 1 fragment from burr with right angled cross section, appears to have worn into gravelly substrate as has lots of pale vitrified sandstone flakes included

15019 604 2 joining fragments of charcoal bearing slag with sheet of slag with tubular vesicles 98 2 two small fresher pieces of FB containing slag sheet with tubular vesicles 186 2 dense slags with wood impressions 408 7 fragments of charcoal-rich FB with a few wood impressions 48 1 complex slag sheet in dense shiny slag, lower face is full of smooth fine prills between charcoal moulds, opposite face is smooth and irregularly hollowed with possible fuel impressions 298 5 rounded rusty slag lumps without clear surface features 220 1 dense rounded fragment of burr 300 1 complex very dense block with veins of slag intruding into wall 80 1 vitrified wall fragment, probably resurfaced in silty clay over previous glaze 88 1 thick runnels of slag detached from planar wall face 22 1 dark lining slag detached from wall 66 1 dense rounded fragment probably from a burr

15019 2 of 6685 1 rounded block of fine FB, dense, but few helpful surface features

1585 1 rounded block of porous FB material, like similar sized blocks from (15010) 1015 1 broken in 2, base of burr, dense shiny sheet on wall below burr - shows some fragmentation, finely prilly with small charcoal elsewhere 626 1 very similar to 1015g block - might be from same burr but does not join 644 1 block with horizontal flows extending between large wood moulds 232 1 smaller fragment with dense flows between large wood moulds

51 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design context Item Item notes

244 1 dense prilly flows with resinous lustre on ?base overlain by dense slag with wood mould 262 1 dense 30mm sheet with tubular vesicles overlain by finely prilly charcoal mould bearing slag, probably from a burr but just possibly SHC 26 1 small rusty weathered nub, probably detached from another piece

15019 3 of 3255 1 large block of fine prilly FB

1792 5 porous prilly charcoal-rich FB with few if any wood moulds 1570 3 blocks of FB with abundant large moulds of wood, mostly round up to 40mm 1070 5 FB fragments, all rather rusty, 1 just possibly a very weathered SHC, 1 a curved sheet from wall? 540 1 margin of very dense slag in tapslag-like block, from margin of cake, flow lobed and slightly reddened on top, internally massive with wood, base non-wetted prills - clearly a puddle, 65mm in thickness 80mm from margin 1160 1 moderately dense FB fragment with two large moulds of sheet wood 12mm thick amongst other voids 1840 1 dense worn block of FB with coarse charcoal, rusty

15022/15023 5520 10 reduced-fired lining in a bowl shape, with deeply vitrified surface, bowl form hard to reconcile with known furnace shape

52 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Table 7: Summary of materials in collection by area and context (weights in g). context notes FB FB wall ore iron m a o t c e r s i a l

Area 13 13020 upper fill of pit 13018 3708 = 13020 (806) of evaluation 344 13023 8 26 13069 1285 13071 4

Area 14 14007 basal fill of furnace 14006 1048 14009 FB in furnace 14006 18000 1160 14010 upper fill of furnace 14006 1028 14011 uppermost fill of furnace 14006 (rich in red 3466 982 clay) 14012 fill below tip of FB in furnace 14006 714 5050 14013 tip from furnace 4598

Area 15 15006 upper fill of furnace 15004 2041 1938 15022/ part of furnace lining of furnace 15004 in 5520 15 situ 02 3 15008 fill of pit 15007?? 250 15010 base of pit/furnace 15009 42700 4310 15011 main fill of 'furnace' 15009 9898 1847 15013 slag dump NW of furnaces 15004 and 16366 156 126

53 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

15009 = 15013 (906) of evaluation 13983 58 15019 upper fill of furnace 15017 24644 300

totals 60700 88851 15851 30 126

54 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Table 8: Radiocarbon dates, with those of comparative sites showing similar technology.

Site Determination Radiocarbon Age substrate Context notes 2 sigma 2 sigma age d d a a t t e e h l i o g w h

Burlescombe AA-34524 (GU-8391) 1145 50 ivy/elder c553 in situ slag in furnace 525 720 1000 Burlescombe AA-34523 (GU-8392) 1195 45 oak sapwood c553 in situ slag in furnace 526 690 970

SUERC-51250 Culmstock Road S (GU33038) 1271 35 Euonymus c66 660 870 SUERC-51251 Culmstock Road S (GU33039) 1197 35 Alnus glutinosa c77 690 950 Millbrook 1220 70 oak (mature) 660 970

Ramsbury HAR-1606 1170 70 oak, hazel slag layer from early smelting 680 1000 Ramsbury HAR-1704 1070 70 soil with charcoal charcoal layer from early smelting 770 1160 Ramsbury HAR-1609 1180 70 oak, hazel furnace 3 680 990 Ramsbury HAR-1626 1130 70 charcoal hearth fill assoc. furnace 3 710 1030 Culmstock Road N SUERC-73284 1170 29 hazel c906 (main waste dump) 770 970 c907 (comminuted waste at base of Culmstock Road N SUERC-73283 1277 29 alder dump) 660 780

55 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

APPENDIX D: ENVIRONMENTAL MATERIAL

Palaeoenvironmental Assessment by Emma Aitken and Sarah F. Wyles

A series of 15 environmental samples (191 litres of soil) were processed from a range of feature types, mainly of early medieval date, from across the site. This was done with the intention of recovering environmental evidence of industrial activity, in particular that associated with a number of furnaces, on the site. The samples were processed by standard flotation procedures (CA Technical Manual No.2). Preliminary identifications of plant macrofossils are summarised in Table 9, below, following the nomenclature of Stace (1997).

The flots varied in size from small to large, with low to high incidence of rooty material and uncharred seeds. The charred material indicated poor levels of preservation, and it is difficult to firmly identify any of the charred plant remains, or to undertake further wood species identification for the charcoal. Much of the charcoal was poorly preserved, due to silt impregnation within its pores, which in some cases precluded full identification.

Early Medieval Pits Fill 13019 (sample 4) of Pit 13018 contained no charred cereal grains, and only a single possible culm node fragment. A moderate quantity of charred plant remains was recorded, and included seeds of dock (Rumex sp.), persicaria (Persicaria sp.) and turnip (Brassica rapa), alongside hazelnut (Corylus avellana) shell fragments and a small number of buds from heath/heather (Erica/Calluna type). Fill 15008 (sample 16) of Pit 15007 contained no charred plant remains at all. Both sample 4 and sample 16 contained high quantities of charcoal fragments larger than 2mm in size. Much of the charcoal from both samples was silt and iron- impregnated, with vitrification occurring in some places. Twigwood and roundwood charcoal fragments were noted within both samples, and sample 16 (Pit 15007) contained fragments of oak (Quercus) wood charcoal. These two environmental assemblages are likely to represent a dump of waste material from a furnace, hence of industrial rather than domestic character, in view of the large quantity of charcoal fragments present.

56 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Furnace 14006 Four samples (samples 11, 12, 13, and 14) were taken from four different fills (14007, 14010, 14011, and 14012 respectively), from within Furnace 14006. No charred cereal grains were recovered from any of the four samples, although sample 12 did contain a single possible rachis fragment which was silt/iron-encrusted. A small number of charred plant remains were recorded from all four samples, and included seeds of bedstraw (Galium sp.), buttercup (Ranunculus sp.), plantain (Plantago sp.), meadow grass/cat’s-tails (Poa/Phleum sp.), docks and the Lamiaceae family. A single thorn was recorded from within sample 11, but due to the poor preservation levels species identification was not possible. Sample 14 contained low quantities of charcoal fragments, whereas samples 11, 12 and 13 contained moderate quantities of charcoal fragments. Much of the charcoal was silt and iron-impregnated, with roundwood fragments noted within sample 11, and fragments of twigwood in sample 14. Sample 13 contained fragments of oak wood charcoal. These assemblages are likely to be representative of waste material from the furnace.

Pit Furnaces Pit Furnace 15004 (samples 18 and 19) contained no charred cereal grains, and only a small number of charred plant remains, which included monocot stem fragments and bedstraw seeds. Sample 18 contained a large volume of charcoal fragments larger than 2mm, which contained fragments of roundwood, oak and non-oak charcoal. Sample 19 contained moderate quantities of charcoal fragments, including fragments of twigwood charcoal.

Pit Furnace 15017 (samples 22 and 25) contained no charred cereal grains, and only a very small number of charred seeds, including indeterminate seed fragments, oat/brome grass (Avena/Bromus sp.) seeds and a bud. Moderately large quantities of charcoal fragments larger than 2mm were recorded from within both samples, and include fragments of oak roundwood and twigwood charcoal.

Fill 15011 (sample 23), of pit furnace 15009, contained no charred cereal grains or plant remains. Large quantities of charcoal fragments were recovered from within the assemblage, and include fragments of oak roundwood and twigwood charcoal.

All of the environmental assemblages discussed above are likely to represent waste material derived from industrial activity.

57 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Undated Colluvial Layers Colluvial Layers 14001 and 14002 contained no charred plant remains, and only a very small number of charcoal fragments larger than 2mm in size. These assemblages are likely to be representative of dispersed material.

Peaty Layers Peaty Layer 14003 contained no charred plant remains, and only a small number of charcoal fragments larger than 2mm. Peaty Layer 14004 contained no charred cereal grains, and only a single possible cabbage (Brassica sp.) seed. Very small quantities of charcoal fragments larger than 2mm in size were also noted during assessment. These environmental assemblages are likely to be indicative of dispersed material.

58 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Table 9: Assessment of Environmental Remains Proce Unproc Flot ssed essed size Root Charred Charcoal Feature Context Sample vol (L) vol (L) (ml) s % Grain Chaff Cereal Notes Other Notes for Table > 4/2mm Other Analysis Early Medieval Pits hazelnut shell frag; bud (Erica/Calluna type); 13018 13019 4 20 20 480 1 - * c.f. culm node *** Rumex; Persicaria; Brassica rapa; indet seed *****/***** - C 15007 15008 16 20 20 695 5 - - - - - *****/***** - C Furnace ind 14007 11 7 0 23 5 - - - * thorn ***/*** wste** C 14006 14010 12 21 0 28 15 - * c.f. rachis * Galium (iron encrusted); Ranunculus **/*** - 14011 13 20 0 50 60 - - - * Plantago; Galium; Lamiaceae family **/*** - C 14012 14 20 0 20 80 - - - * Plantago; Poa/Phleum; Rumex -/** - Pit Furnaces 15005 18 20 20 450 <1 - - - * Galium *****/***** - C 15004 15006 19 20 20 110 2 - - - * monocot stem frag **/*** - 15019 22 20 20 205 1 - - - * indet seed frag; Avena/Bromus ***/**** - C 15017 15018 25 20 20 45 2 - - - * bud **/*** - 15009 15011 23 1 0 240 <1 - - - - - *****/***** - C Undated Colluvial Layers 14001 7 0.5 0 1 2 ------/* - 14002 8 0.5 0 1 60 ------/* - Peaty Layers 14003 9 0.5 0 100 90 - - - - - */* - 14004 10 0.5 0 80 90 - - - * c.f. Brassica -/* -

Key: * = 1–4 items; ** = 5–19 items; *** = 20–49 items; **** = 50–99 items; ***** = >100 items

59 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

APPENDIX E: GEOARCHAEOLOGICAL ASSESSMENT

Geoarchaeological examination of monolith samples by Agata Kowalska

Introduction

In May 2018, Cotswold Archaeology carried out an archaeological excavation on land off Culmstock Road, Hemyock, Devon (henceforth the site). A floodplain of the River Culm is located c. 500m to the north of the site, and a small stream runs along the eastern site boundary (CA 2017). A possible paleochannel, running south-west to north-east across Areas 14 and 15, with a well-preserved thick peat deposit, was recorded by both evaluation and excavation (CA 2017a; CA 2019). A colluvium deposit overlying the mudstone bedrock geology is mapped in the northern and western parts of the site, accumulating downslope within the Culm Valley. The colluvium formed up to 2 million years ago in the Quaternary Period, and comprises unconsolidated clays and silts with intercalated peats and sands (BGS 2019).

Four monolith samples were taken from the possible palaeochannel running north-east to south-west across Areas 14 and Area 15. The samples were taken to assess the character of this feature and the mode of deposition of encountered sediments; and to provide paleoenvironmental evidence for an environmental reconstruction of the site.

This report presents the results of a geoarchaeological assessment carried out on the four monolith samples. The main objectives of the report are to:

• To describe and interpret sediments in order to characterize the depositional processes and nature of the feature;

• To assess the palaeoenvironmental potential of the sediments for reconstruction of changes in vegetation history.

Methodology

A total of four monolith samples were retained in steel tins, measuring 100 x 100 x 500mm, and were then wrapped and labelled following standard sampling procedures (CA 2017b). These were taken, in two sequences, from two sections of a possible palaeochannel recorded on the site. Two monolith samples were taken from Area 14, samples 6a and 6b; and two monolith samples from Area 15 , samples 26a and 26b (Figs. i and ii).

60 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

The monoliths were opened, and the deposits cleaned, photographed and recorded. The lithostratigraphy of the samples was described according to the standard geological criteria provided by Jones et al. 1999, Munsell Color 2018 and Tucker 2011.

All sedimentary Units were distinguished, based on the lithological characteristics of the sediments recorded in each monolith sample. The geoarchaeological observations were supplied by context sheets, photographs of samples locations and a summary report of the archaeological work (CA 2019). All observations are summarised in Tables 10-13, below.

Figure i: South-facing section of the natural feature recorded in Area 14. The red arrow points to a stony line within context 14001, Unit 2.

61 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Figure ii: North-east facing section of the natural feature recorded in Area 15. The red arrow points to a gravelly riverbed deposit.

Results

The lithological descriptions of four monolith samples are presented in Tables 10 to 13, below. The text description is presented in stratigraphic order, with the earliest unit described first.

The lowermost Unit 6, context 1404, comprises an homogenous, very dark, grey clayey silt, with very fine to fine sand mineral grains. An angular medium sandstone, derived from local geology, was recorded. Well-preserved, waterlogged wood/root fragments and rare charcoal granules were noted throughout the unit, and could be translocated from the unit above by bioturbation. The Unit was onl y recorded in monolith 6b from Area 14, and was 0.11m thick.

According to context sheet 14004, the context was described as a lower peat deposit. The grey sediment is characteristic of depleted groundmass and anaerobic conditions, and the homogeneity and fine silty/clay texture of the unit suggest sedimentation in shallow and slow-moving water (Rapp and Hill 1998). Therefore, it is interpreted as a deposit created by the natural deposition of sediments in slow-moving water within a natural feature.

Unit 5 was separated from Unit 6 by a sharp horizontal boundary. Unit 5 (contexts 14003 and 15002) was c. 0.15m to 0.30m thick, and comprised an homogenous, soft humic black to very dark-grey/brown clay silt. Amorphous organic material dominated this unit, although

62 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

rare roots and plant fragments were visible. In monolith sample 26b, of between 0.35m- 0.38m in thickness, a band of dark-grey silty clay was recorded. This band was irregular and mixed with humic material, possibly due to bioturbation. The clay sediments are not as organic as Unit 5, and possibly represent an in-wash of clay material which had settled at the bottom of the feature. Rare charcoal granules were also recorded.

According to context sheets, 14003 and 15002, Unit 5 was interpreted as a peat deposit and the assessment of the unit confirmed this. The deposit appears to have developed within the natural feature, and that would suggest local, stable, possibly slow-moving or standing water, which allowed a freshwater floral community to develop. The peat, which had formed as accumulated plant material, did not fully decay, due to acidic and anaerobic conditions (English Heritage 2007).

Overlying Unit 4, the lower part of context 14002 and top part of context 15002, was separated by a sharp (monolith 6b) to relatively diffuse (monolith 26b) horizontal boundary with Unit 5. The thickness of this unit varied from 0.07m to 013m, and consisted of a dark- grey/brown to dark-grey silty clay, with very fine to fine sand. The unit is homogenous, firm and stoneless, with concentrations of rare yellow/red iron oxides. The unit appears grey in colour, which suggests an iron oxide-depleted groundmass, due to a long (weeks to months) duration of water saturation and reduction (Lindbo et al. 2010, 138).

Unit 4 was distinguished in monolith sample 26b by a different coloration and textual composition. In monolith sample 6b, the unit was more distinctive, and in the field was recorded as context 14002. According to context sheet 14002, the sediment was described as a potential alluvial deposit. A sharp contact boundary separated Unit 5 and Unit 4, which implies an erosional boundary and change in depositional regime. The Unit is homogenous, without coarse inclusions, and is relatively more clayey in texture than the overlying unit. This texture is characteristic of very low-energy regimes such as flood (alluvial) deposits and standing water (Rapp and Hill 1998, 59). Possibly the mineral material entered the channel from upslope, and finer particles settled at the bottom of the negative feature. The grey coloration of the unit suggests the presence of anaerobic conditions.

Unit 4 is overlain by Unit 3, with contexts represented by the upper part of 14002 and lower part of 15001. The thickness of the unit varies between 0.17m and 0.25m. The horizontal boundary between the units is diffuse, and suggests a gradual transition. The unit consists of a dark-yellow/brown to grey/brown, very fine to fine sandy silt/clay. Randomly distributed patches of fine, light-yellow/brown sand within the deposit give the Unit a heterogenous appearance. The unit is slightly stony, and the angular to sub-rounded clasts are randomly distributed through it. The presence of rare concentrations of yellow/red iron oxides implies

63 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

changing oxidation conditions. Very few charcoal granules were recorded within the unit. According to context sheet 15001, the context was described as possible colluvium, and context 14002 was described as alluvium. In course of the geoarchaeological assessment, context 15001 was divided into Unit 3 and Unit 2, due to changes in texture. The unit is heterogeneous and coarser, with higher very fine to fine sand particles and silt. The coarse texture, and presence of angular clasts derived from local bedrock, suggests relatively higher depositional energy. These sediments seem to have derived from the inwashing of the material from the slope above.

Unit 2 (context 15001 and 14001), is separated by a diffuse horizontal boundary with Unit 3. The Unit is 0.10m to 0.13m thick, and consists of homogenous, brown, very fine to fine sandy clayey silt, with rare, small angular sandstones derived from local geology. The unit is oxidized, due to changing oxidation conditions. Rare and fine macropores and roots indicate the penetration of the unit by plant growth. The unit appears to have a higher content of clay than Unit 1 (topsoil), as a result of possible clay illuviation.

According to context sheets 14001 and 15001, the sediments were interpreted as a possible colluvium. The very slightly stony and bioturbated texture exhibits a similar composition and stone lithology to the sediment and soil further upslope (Figs. i and ii). Most probably, this unit was deposited due to the downslope movement and accumulation of material. The processes could be triggered by weathering/erosion of relatively poorly vegetated soil/sediments by water movement (rain), and tillage (English Heritage 2001; Wilkinson 2009). A line of medium-sized clasts which had accumulated in lower parts of the unit, only visible in context 14001 (Fig. i), could represent evidence of soil-creep and higher-energy deposition.

Uppermost Unit 1 (contexts 14000 and 15000), is separated by a diffuse horizontal boundary with overlying Unit 2. The thickness of the Unit differs from 013m to 0.17m. Unit 1 consists of a brown, very fine to fine sandy silt/clay, with rare medium and angular clasts. The unit is oxidized, implying impeded drainage, and roots and fine rootlets are common. The unit represents a loamy ploughsoil, recorded elsewhere across the site.

Assessment

The natural feature recorded on the site could be interpreted as a paleochannel, and was probably associated with a relict tributary stream or oxbow lake of the River Culm. The channel appears to cut into the natural geology, and the riverbed was gravelly and stony (Fig ii.; pers. comm. J. Orellana). The coarse nature of this horizon demonstrates the presence of relatively fast water movement in an active river channel. Further up, the sequence recorded

64 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

in the monoliths began to decrease in size to more finer sediments. This indicates slower water-flow or standing water, suggesting that the palaeochannel became progressively separated from the river. By this stage, the channel had become vegetated, and a peat deposit had started to form.

The upper sediments are associated with sediment movement downslope, towards the possible inactive channel. The hill-wash deposit appears to have derived from the in- washing of neighbouring ploughsoils, due to agricultural practice.

65 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Table 10: Monolith sample 6a.

Monolith Unit Depth Cont’ Description [m] t 1 0-0.17 14000 10YR 4/3 brown very fine to fine sandy silt/clay. Homogenous and friable. Few (5%) 5YR 4/6 yellow/red iron oxides. One medium (c. 40mm) platy angular flint. Roots and fine rootles common. Diffuse boundary to:

2 0.17- 14001 10YR 3/4 brown, very fine to fine 0.33 sandy clayey silt. Homogenous and friable. Few (5%) 5YR 4/6 yellow/red iron oxides mottling. Very few (<2%) fine rootless. Diffuse boundary to:

10YR 5/2 grey/brown, very fine to fine 3 0.33- 14002 sandy silty clay. Patches of fine 10YR 0.50 6/4 light-yellow/brown loose sand. Very few (1%) 5YR 4/6 yellow/red iron oxide concentrations. Stoneless. Very few (1%) charcoal granules.

66 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Table 11: Monolith sample 6b.

Monolith Unit Depth Context Description [m] 3 0-0.12 14002 0.10m overlap with monolith 6a. 10YR 5/2 grey/brown, very fine to fine sandy silty clay. Homogenous and friable. Very few (1%) 5YR 4/6 yellow/red iron oxide concentrations. Stoneless. Very few (1%) charcoal granules. Diffuse boundary to:

4 0.12- 14002 10YR 4/2 dark grey/brown silty clay, 0.25 with very fine to fine sand. Homogenous and firm. Very few (1%) 5YR 4/6 yellow/red iron oxides mottling. Stoneless. Sharp boundary to:

5 0.25- 14003 0.39 10YR 2/1 humic black clay silt. Homogenous and soft. 1 angular medium flint (20mm). Sharp boundary to:

6 0.39- 14004 0.50 10YR 3/1 very dark grey clayey silt, with very fine to fine sand mineral grains. 1angular medium sandstone. Very few (<2%) waterlogged wood/root fragments and charcoal granules.

67 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Table 12: Monolith sample 26a.

Monolith Unit Depth Context Description [m] 1 15000 0-0.13 10YR 4/3 brown, very fine to fine sandy silt/clay. Homogenous and friable. Few (5%) 5YR 4/6 yellow/red iron oxides. 1 medium (c. 40mm) platey angular flint. Roots and fine rootles common. Diffuse boundary to:

2 15001 0.13- 10YR 5/3 brown, very fine to fine 0.25 sandy silty clay. Homogenous and firm. Few (5%) 5YR 4/6 yellow/red iron oxide mottling. Very fine macropores. Very slightly stony (2%), angular small (<10mm) sandstones. Very few (<1) rootless. Diffuse boundary to:

3 15001 0.25- 10YR 4/4 dark-yellow/brown, very fine 0.50 to fine sandy silt/clay. Firm. Very few (1%) 5YR 4/6 yellow/red iron oxide concentrations. Patches of 10YR 6/4 light-yellow/brown fine sand. Very few (2%) charcoal granules. Very slightly stony (1%), small to medium size clast of angular to sub-rounded sandstone and flint.

68 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

Table 13: Monolith sample 26b.

Monolith Unit Depth Context Description [m] 3 0-0.14 15001 10YR 4/4 dark-yellow/brown, very fine to fine sandy silt/clay. Firm. Very few (1%) 5YR 4/6 yellow/red iron oxide concentrations. Very few (2%) charcoal granules. Very slightly stony (1%), small to medium size clast of angular to sub-rounded sandstone and flint. Diffuse (gradual) boundary to:

4 0.13- 15002 0.20 10YR 4/1 dark-grey silty clay, with very fine to fine sand. Homogenous and firm. Stoneless. Diffuse (gradual) boundary to:

5 0.20- 15002 0.50 10YR 3/2 very dark grey/brown clayey silt, with very fine to fine sand mineral grains. Homogenous and soft. Very slightly stony (1%), small (<2mm) angular sandstones. Common, partially waterlogged roots and plant material. Between 0.35m-0.38m 10YR 4/1 dark-grey silty clay reworked band. Very few charcoal granules.

69 © Cotswold Archaeology Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design

APPENDIX F: OASIS REPORT FORM

PROJECT DETAILS

Project Name Land North of Culmstock Road, Hemyock, Devon Short description A programme of archaeological excavation was undertaken by Cotswold Archaeology in May 2018 at the request of Waddeton Park Ltd on land off Culmstock Road, Hemyock, Devon. Three excavation areas, together amounting to 1,890m², were excavated across the development area. Archaeological features were identified throughout the three excavation areas and comprised the remains of four iron smelting pit furnaces, two deposits of metalworking debris, a large rectangular pit, two ditches and a number of undated pits and postholes. A non-slag tapping iron smelting furnace which contained an in situ iron slag puddle from its last smelt was identified in Area 14. Metallurgical residue deposits associated with this furnace had been recorded during a previous evaluation and had been radio- carbon dated to the early medieval period. Monolith sampling revealed the presence of an undated palaeochannel crossing Areas 14 and 15, representing a former tributary of the nearby River Culm.

Project dates 1−16 May 2018 Project type Excavation

Previous work Geophysical survey (PCG 2016) Evaluation (CA 2017)

Future work Unknown PROJECT LOCATION Site Location Land North of Culmstock Road, Hemyock, Devon Study area (M2/ha) 1,890m² Site co-ordinates 313380 113490

PROJECT CREATORS Name of organisation Cotswold Archaeology Project Brief originator NA Project Design (WSI) originator Cotswold Archaeology

Project Manager Derek Evans Project Supervisor Jonathan Orellana MONUMENT TYPE Iron smelting furnaces SIGNIFICANT FINDS Ironworking residues PROJECT ARCHIVES Intended final location of archive Content

Physical Royal Albert Memorial Museum, Exeter to be decided RAMM: ref no. 42/2017 Paper Royal Albert Memorial Museum, Exeter Context sheets, section RAMM: 42/2017 NO THEY DON’T drawings, registers TAKE IT Digital NO = ADS Digital survey, digital photos, geomatic data etc BIBLIOGRAPHY

CA (Cotswold Archaeology) 2019 Land North of Culmstock Road, Hemyock, Devon: Post-Excavation Assessment and Updated Project Design. CA typescript report EX0032_1

70 313000 315000

115000

113000

111000

SOMERSET BATH AND Andover 01264 347630 NE SOMERSET N Cirencester 01285 771022 Cotswold Exeter 01392 573970 Milton Keynes 01908 564660 Archaeology Suffolk 01449 900120 w www.cotswoldarchaeology.co.uk SOMERSET e [email protected] PROJECT TITLE Land North of Culmstock Road, Hemyock, Devon DEVON DORSET FIGURE TITLE Site location plan 0 1km

DRAWN BY EC PROJECT NO. EX0032 FIGURE NO. © Crown copyright and database rights 2019 CHECKED BY DJB DATE 11/06/2019 TORBAY Ordnance Survey 0100031673 APPROVED BY JO SCALE@A4 1:25,000 1 313500 313300 N 113700

T1

Site boundary A2 Evaluation trench (CA 2017) Excavation area Archaeological feature A3 Deposit Field drain

T11 T10

2 Area 15 T12 3 Geophysics produced by Pre-Construct Geophysics Ltd (PCG 2016) T2 >Predominately

T9 1 Area 14 Predominately natural, although archaeological rema produce weak magnetic anomalies within this ran ditches/pits). Exceptions include fired material (e kilns, hearths and other sites subject to intense

< Predominately modern (rubble, metal objects/fenci T8 Area 13 Potentially archaeological T7 T3 Modern path Service EP EP Electricity pole 113500

A1 T5 A4

T4

0 1:1250 50m

© Crown copyright and database rights 2019 Ordnance Survey 0100031673 EP

T6 Andover 01264 347630 Cirencester 01285 771022 Exeter 01392 573970 Cotswold Milton Keynes 01908 564660 Archaeology Suffolk 01449 900120 w www.cotswoldarchaeology.co.uk e [email protected]

PROJECT TITLE EP Land North of Culmstock Road, Hemyock, 4 Devon

FIGURE TITLE The site showing excavation areas, archaeological features, previous evaluation B3391 trenches and geophysical survey results

DRAWN BY EC PROJECT NO. EX0032 FIGURE NO. CHECKED BY DJB DATE 12/06/19 APPROVED BY JO SCALE@A3 1:1250 2 AREA15

pit 15020 pit 15017

D furnace pit 15009 layer 15013

C D furnace pit 15004 See inset map furnace pit 15007 C

pit 15014

051:100 m

Area 15

Area 14

layer 14013

furnace pit B 14006

B

pit 13080 pit 13078 pit 13076

pit 13070 pit 13072

pit Area 13 13083 pit 13067 pit 13012

pit pit 13014 13018 . pit 13074 A A

pit pit 13016 13010 130 pit 13021

ditch pit 13061 13008 pit 13004 ditch pit 13063 pit 1302 pit 13006 pit pit 13094 13042 pit 13026 pit 94 13085 pit 13029 13044 pit 1 13028 pit pit 13031 pit 13040 13092 1309 pit pit pit 13047 pit 13033 13035 13090 pit 13058 pit pit pit pit 13045 13056 13059 13049 pit 13054 pit 13065 pit pit 13052 13024

Andover 01264 347630 Cirencester 01285 771022 N Exeter 01392 573970 Cotswold Milton Keynes 01908 564660 Archaeology Suffolk 01449 900120 w www.cotswoldarchaeology.co.uk e [email protected] Limit of excavation PROJECT TITLE Land North of Culmstock Road, Hemyock, Archaeological feature (excavated/unexcavated) Devon Deposit FIGURE TITLE Excavation areas, showing Field drain archaeological features

DRAWN BY EC PROJECT NO. EX0032 FIGURE NO. CHECKED BY DJB DATE 12/06/19 0 1:250 10m APPROVED BY JO SCALE@A3 1:250 3 PROJECT TITLE Land North of Culmstock Road, Hemyock, Devon

FIGURE TITLE Photographs

DRAWN BY EC PROJECT NO. EX0032 FIGURE NO. CHECKED BY DJB DATE 12/06/19 APPROVED BY JO SCALE@A4 N/A 4 Section AA 136.4m EWAOD 136.34 m AOD 13020 13020 Modern field drain 13019 13019 pit 13018

011:20 m

Pit 13018, looking north-east (1m scale)

Andover 01264 347630 Cirencester 01285 771022 Exeter 01392 573970 Cotswold Milton Keynes 01908 564660 Archaeology Suffolk 01449 900120 w www.cotswoldarchaeology.co.uk e [email protected]

PROJECT TITLE Land North of Culmstock Road, Hemyock, Devon

FIGURE TITLE Pit 13018 section and photograph

DRAWN BY EC PROJECT NO. EX0032 FIGURE NO. CHECKED BY DJB DATE 12/06/19 APPROVED BY JO SCALE@A4 1:20 5 Section BB

SE NW 133.6 m 133.5m AOD 14011 AOD 14005 natural

14008 14010 FE slag 1409 14012 14007

furnace 14006

011:20 m

Furnace 14006, looking south-west (0.4m scale)

Andover 01264 347630 Cirencester 01285 771022 Exeter 01392 573970 Cotswold Milton Keynes 01908 564660 Archaeology Suffolk 01449 900120 w www.cotswoldarchaeology.co.uk e [email protected]

PROJECT TITLE Land North of Culmstock Road, Hemyock, Devon

FIGURE TITLE Furnace 14006 section and photograph

DRAWN BY EC PROJECT NO. EX0032 FIGURE NO. CHECKED BY DJB DATE 12/06/19 APPROVED BY JO SCALE@A4 1:20 6 Section CC

SN 133.6 m 133.51m AOD AOD

15006

15003 natural 15005

furnace 15004

Furnace 15004, looking south-west (0.4m scale)

Section DD

SN 131.72m 131.8 m AOD AOD

15011 15012

furnace 15009

011:20 m

Andover 01264 347630 Cirencester 01285 771022 Exeter 01392 573970 Cotswold Milton Keynes 01908 564660 Archaeology Suffolk 01449 900120 w www.cotswoldarchaeology.co.uk e [email protected] PROJECT TITLE Furnace 15009, looking west (1m scale) Land North of Culmstock Road, Hemyock, Devon

FIGURE TITLE Furnace 15004 and 15009 sections and photographs

DRAWN BY EC PROJECT NO. EX0032 FIGURE NO. CHECKED BY DJB DATE 12/06/19 APPROVED BY JO SCALE@A3 1:20 7

71